Title: | Spatially Explicit Capture-Recapture |
---|---|
Description: | Functions to estimate the density and size of a spatially distributed animal population sampled with an array of passive detectors, such as traps, or by searching polygons or transects. Models incorporating distance-dependent detection are fitted by maximizing the likelihood. Tools are included for data manipulation and model selection. |
Authors: | Murray Efford [aut, cre] , Philipp Jund [ctb] (faster transect search and spacing), David Fletcher [ctb] (overdispersion), Yan Ru Choo [ctb] (<https://orcid.org/0000-0002-8852-7178>, goodness-of-fit) |
Maintainer: | Murray Efford <[email protected]> |
License: | GPL (>= 2) |
Version: | 5.1.1 |
Built: | 2024-11-07 08:31:05 UTC |
Source: | https://github.com/murrayefford/secr |
Functions to estimate the density and size of a spatially distributed animal population sampled with an array of passive detectors, such as traps, or by searching polygons or transects.
Package: | secr |
Type: | Package |
Version: | 5.1.1 |
Date: | 2024-11-07 |
License: | GNU General Public License Version 2 or later |
Spatially explicit capture–recapture is a set of methods for studying marked animals distributed in space. Data comprise the locations of detectors (traps, searched areas, etc. described in an object of class ‘traps’), and the detection histories of individually marked animals. Individual histories are stored in an object of class ‘capthist’ that includes the relevant ‘traps’ object.
Models for population density (animals per hectare) and detection are defined in secr using symbolic formula notation. Density models may include spatial or temporal trend. Possible predictors for detection probability include both pre-defined variables (t, b, etc.) corresponding to ‘time’, ‘behaviour’ and other effects), and user-defined covariates of several kinds. Habitat is distinguished from nonhabitat with an object of class ‘mask’.
Models are fitted in secr by maximizing either the full likelihood
or the likelihood conditional on the number of individuals observed
(). Conditional likelihood models are limited to homogeneous
Poisson density, but allow continuous individual covariates for
detection. A model fitted with
secr.fit
is an object
of class secr
. Generic methods (plot, print, summary, etc.) are
provided for each object class.
A link at the bottom of each help page takes you to the help index. Several vignettes complement the help pages:
General interest | |
secr-overview.pdf | general introduction |
secr-datainput.pdf | data formats and input functions |
secr-version4.pdf | changes in secr 4.0 |
secr-manual.pdf | consolidated help pages |
secr-tutorial.pdf | introductory tutorial |
secr-habitatmasks.pdf | buffers and habitat masks |
secr-spatialdata.pdf | using spatial data |
secr-models.pdf | linear models in secr |
secr-troubleshooting.pdf | problems with secr.fit, including speed issues |
More specialised topics | |
secr-densitysurfaces.pdf | modelling density surfaces |
secr-finitemixtures.pdf | mixture models for individual heterogeneity |
secr-markresight.pdf | mark-resight data and models |
secr-multisession.pdf | multi-session capthist objects and models |
secr-noneuclidean.pdf | non-Euclidean distances |
secr-parameterisations.pdf | alternative parameterisations sigmak, a0 |
secr-polygondetectors.pdf | using polygon and transect detector types |
secr-sound.pdf | analysing data from microphone arrays |
secr-varyingeffort.pdf | variable effort in SECR models |
The datasets captdata
and ovenbird
include examples of fitted
models. For models fitted to other datasets see secr-version4.pdf Appendix 2.
Add-on packages extend the capability of secr and are documented separately. secrlinear enables the estimation of linear density (e.g., animals per km) for populations in linear habitats such as stream networks (secrlinear-vignette.pdf). secrdesign enables the assessment of alternative study designs by Monte Carlo simulation; scenarios may differ in detector (trap) layout, sampling intensity, and other characteristics (secrdesign-vignette.pdf). ipsecr fits some awkward models (e.g., for single-catch traps) by simulation and inverse prediction (ipsecr-vignette.pdf). openCR fits open population models, both non-spatial and spatial (openCR-vignette.pdf).
The analyses in secr extend those available in the software Density (see www.otago.ac.nz/density/ for the most recent version of Density). Help is available on the ‘DENSITY | secr’ forum at www.phidot.org and the Google group secrgroup. Feedback on the software is also welcome, including suggestions for additional documentation or new features consistent with the overall design.
Murray Efford [email protected]
Borchers, D. L. and Efford, M. G. (2008) Spatially explicit maximum likelihood methods for capture–recapture studies. Biometrics 64, 377–385.
Borchers, D. L. and Fewster, R. M. (2016) Spatial capture–recapture models. Statistical Science 31, 219–232.
Efford, M. G. (2004) Density estimation in live-trapping studies. Oikos 106, 598–610.
Efford, M. G. (2011) Estimation of population density by spatially explicit capture–recapture with area searches. Ecology 92, 2202–2207.
Efford, M. G., Borchers D. L. and Byrom, A. E. (2009) Density estimation by spatially explicit capture-recapture: likelihood-based methods. In: D. L. Thomson, E. G. Cooch and M. J. Conroy (eds) Modeling Demographic Processes in Marked Populations. Springer, New York. Pp. 255–269.
Efford, M. G., Borchers D. L. and Mowat, G. (2013) Varying effort in capture–recapture studies. Methods in Ecology and Evolution 4, 629–636.
Efford, M. G., Dawson, D. K. and Borchers, D. L. (2009) Population density estimated from locations of individuals on a passive detector array. Ecology 90, 2676–2682.
Efford, M. G., Dawson, D. K. and Robbins C. S. (2004) DENSITY: software for analysing capture-recapture data from passive detector arrays. Animal Biodiversity and Conservation 27, 217–228.
Efford, M. G. and Fewster, R. M. (2013) Estimating population size by spatially explicit capture–recapture. Oikos 122, 918–928.
Efford, M. G. and Hunter, C. M. (2017) Spatial capture–mark–resight estimation of animal population density. Biometrics 74, 411–420.
Efford, M. G. and Mowat, G. (2014) Compensatory heterogeneity in capture–recapture data.Ecology 95, 1341–1348.
Royle, J. A., Chandler, R. B., Sollmann, R. and Gardner, B. (2014) Spatial capture–recapture. Academic Press.
Royle, J. A. and Gardner, B. (2011) Hierarchical spatial capture–recapture models for estimating density from trapping arrays. In: A.F. O'Connell, J.D. Nichols and K.U. Karanth (eds) Camera Traps in Animal Ecology: Methods and Analyses. Springer, Tokyo. Pp. 163–190.
read.capthist
,
secr.fit
,
traps
,
capthist
,
mask
## Not run: ## generate some data & plot detectors <- make.grid (nx = 10, ny = 10, spacing = 20, detector = "multi") plot(detectors, label = TRUE, border = 0, gridspace = 20) detections <- sim.capthist (detectors, noccasions = 5, popn = list(D = 5, buffer = 100), detectpar = list(g0 = 0.2, sigma = 25)) session(detections) <- "Simulated data" plot(detections, border = 20, tracks = TRUE, varycol = TRUE) ## generate habitat mask mask <- make.mask (detectors, buffer = 100, nx = 48) ## fit model and display results secr.model <- secr.fit (detections, model = g0~b, mask = mask) secr.model ## End(Not run)
## Not run: ## generate some data & plot detectors <- make.grid (nx = 10, ny = 10, spacing = 20, detector = "multi") plot(detectors, label = TRUE, border = 0, gridspace = 20) detections <- sim.capthist (detectors, noccasions = 5, popn = list(D = 5, buffer = 100), detectpar = list(g0 = 0.2, sigma = 25)) session(detections) <- "Simulated data" plot(detections, border = 20, tracks = TRUE, varycol = TRUE) ## generate habitat mask mask <- make.mask (detectors, buffer = 100, nx = 48) ## fit model and display results secr.model <- secr.fit (detections, model = g0~b, mask = mask) secr.model ## End(Not run)
Tools to construct spatial covariates for existing mask or traps objects from a spatial data source.
addCovariates(object, spatialdata, columns = NULL, strict = FALSE, replace = FALSE)
addCovariates(object, spatialdata, columns = NULL, strict = FALSE, replace = FALSE)
object |
mask, traps or popn object |
spatialdata |
spatial data source (see Details) |
columns |
character vector naming columns to include (all by default) |
strict |
logical; if TRUE a check is performed for points in |
replace |
logical; if TRUE then covariates with duplicate names are replaced; otherwise a new column is added |
The goal is to obtain the value(s) of one or more spatial covariates
for each point (i.e. row) in object
. The procedure depends on
the data source spatialdata
, which may be either a spatial
coverage (raster or polygon) or an object with covariate values at
points (another mask or traps object). In the first case, an
overlay operation is performed to find the pixel or
polygon matching each point. In the second case, a search is conducted
for the closest point in spatialdata
.
If spatialdata
is a character value then it is interpreted as
the name of a polygon shape file (excluding ‘.shp’).
If spatialdata
is a SpatialPolygonsDataFrame,
SpatialGridDataFrame or 'sf' object from sf then it will be used
in an overlay operation as described.
If package terra has been installed then spatialdata
may also be a RasterLayer from package raster or SpatRaster from terra. If provided counts
should be a single name that will be used for the values (otherwise 'raster' will be used).
If spatialdata
is a mask
or traps
object then it
is searched for the closest point to each point in object
, and
covariates are drawn from the corresponding rows in
covariates(spatialdata)
. By default (strict = FALSE
),
values are returned even when the points lie outside any cell of the mask.
An object of the same class as object
with new or augmented
covariates
attribute. Column names and types are derived from the input.
Use of a SpatialGridDataFrame for spatialdata
is untested.
make.mask
, read.mask
, read.traps
## In the Lake Station skink study (see ?skink), habitat covariates were ## measured only at trap sites. Here we extrapolate to a mask, taking ## values for each mask point from the nearest trap. LSmask <- make.mask(LStraps, buffer = 30, type = "trapbuffer") tempmask <- addCovariates(LSmask, LStraps) ## show first few lines head(covariates(tempmask))
## In the Lake Station skink study (see ?skink), habitat covariates were ## measured only at trap sites. Here we extrapolate to a mask, taking ## values for each mask point from the nearest trap. LSmask <- make.mask(LStraps, buffer = 30, type = "trapbuffer") tempmask <- addCovariates(LSmask, LStraps) ## show first few lines head(covariates(tempmask))
Add sighting data on unmarked individuals and/or unidentified marked individuals to an existing capthist object.
addSightings(capthist, unmarked = NULL, nonID = NULL, uncertain = NULL, verify = TRUE, ...)
addSightings(capthist, unmarked = NULL, nonID = NULL, uncertain = NULL, verify = TRUE, ...)
capthist |
secr capthist object |
unmarked |
matrix or list of matrices of sightings of unmarked animals, Tu, or file name (see Details) |
nonID |
matrix or list of matrices of unidentified sightings of marked animals, Tm, or file name (see Details) |
uncertain |
matrix or list of matrices of uncertain sightings, Tn, or file name (see Details) |
verify |
logical; if TRUE then the resulting capthist object is
checked with |
... |
other arguments passed to |
The capthist object for mark-resight analysis comprises distinct marking and sighting occasions, defined in the markocc attribute of traps(capthist)
. Add this attribute to traps(capthist)
with markocc
before using 'addSightings'. See also read.traps
and read.capthist
.
Mark-resight data may be binary (detector type ‘proximity’) or counts (detector types ‘count’, 'polygon' or 'transect'). The detector type is an attribute of traps(capthist)
. Values in unmarked
and nonID
should be whole numbers, and may be greater than 1 even for binary proximity detectors because multiple animals may be detected simultaneously at one place.
Arguments unmarked
, nonID
, uncertain
provide data for attributes
‘Tu’, ‘Tm’, ‘Tn’ respectively. They may take several forms
a single integer, the sum of all counts*
a matrix of the count on each occasion at each detector (dimensions K x S, where K is the number of detectors and S is the total number of occasions). Columns corresponding to marking occasions should be all-zero.
for multi-session data, a list with components as above
a character value with the name of a text file containing the data; the file will be read with read.table
. The ... argument allows some control over how the file is read. The data format comprises at least S+1 columns. The first is a session identifier used to split the file when the data span multiple sessions; it should be constant for a single-session capthist. The remaining S columns contain the counts for occasions 1:S, one row per detector. Further columns may be present; they are ignored at present.
* although this is convenient, the full matrix of counts provides more flexibility (e.g., when you wish to subset by occasion), and enables modelling of variation across detectors and occasions.
A capthist object with the same structure as the input, but with new sighting-related attributes Tu (sightings of unmarked animals) and/or Tm (unidentified sightings of marked animals). Input values, including NULL, overwrite existing values.
** Mark-resight data formats and models are experimental and subject to change **
markocc
,
read.capthist
,
read.traps
,
sim.resight
,
Tm
,
Tu
,
Tn
,
secr-markresight.pdf
## construct capthist object MRCH from text files provided in ## 'extdata' folder, assigning attribute 'markocc' and add unmarked ## and marked sightings from respective textfiles datadir <- system.file("extdata", package = "secr") captfile <- paste0(datadir, '/MRCHcapt.txt') trapfile <- paste0(datadir, '/MRCHtrap.txt') Tufile <- paste0(datadir, '/Tu.txt') Tmfile <- paste0(datadir, '/Tm.txt') MRCH <- read.capthist(captfile, trapfile, detector = c("multi", rep("proximity",4)), markocc = c(1,0,0,0,0)) MRCH1 <- addSightings(MRCH, Tufile, Tmfile) ## alternatively (ignoring marked, not ID sightings) MRCH <- read.capthist(captfile, trapfile, detector = c("multi", rep("proximity",4)), markocc = c(1,0,0,0,0)) Tu <- read.table(Tufile)[,-1] # drop session column MRCH2 <- addSightings(MRCH, unmarked = Tu) summary(MRCH2)
## construct capthist object MRCH from text files provided in ## 'extdata' folder, assigning attribute 'markocc' and add unmarked ## and marked sightings from respective textfiles datadir <- system.file("extdata", package = "secr") captfile <- paste0(datadir, '/MRCHcapt.txt') trapfile <- paste0(datadir, '/MRCHtrap.txt') Tufile <- paste0(datadir, '/Tu.txt') Tmfile <- paste0(datadir, '/Tm.txt') MRCH <- read.capthist(captfile, trapfile, detector = c("multi", rep("proximity",4)), markocc = c(1,0,0,0,0)) MRCH1 <- addSightings(MRCH, Tufile, Tmfile) ## alternatively (ignoring marked, not ID sightings) MRCH <- read.capthist(captfile, trapfile, detector = c("multi", rep("proximity",4)), markocc = c(1,0,0,0,0)) Tu <- read.table(Tufile)[,-1] # drop session column MRCH2 <- addSightings(MRCH, unmarked = Tu) summary(MRCH2)
Animal locations determined by radiotelemetry can be used to augment
capture–recapture data. The procedure in secr is first to form a
capthist object containing the telemetry data and then to combine this
with true capture–recapture data (e.g. detections from hair-snag DNA)
in another capthist object. secr.fit
automatically detects the
telemetry data in the new object.
addTelemetry (detectionCH, telemetryCH, type = c('concurrent','dependent','independent'), collapsetelemetry = TRUE, verify = TRUE) xy2CH (CH, inflation = 1e-08) telemetrytype (object) <- value telemetrytype (object, ...)
addTelemetry (detectionCH, telemetryCH, type = c('concurrent','dependent','independent'), collapsetelemetry = TRUE, verify = TRUE) xy2CH (CH, inflation = 1e-08) telemetrytype (object) <- value telemetrytype (object, ...)
detectionCH |
single-session capthist object, detector type ‘single’, ‘multi’, ‘proximity’ or ‘count’ |
telemetryCH |
single-session capthist object, detector type ‘telemetryonly’ |
type |
character (see Details) |
collapsetelemetry |
logical; if TRUE then telemetry occasions are collapsed to one |
verify |
logical; if TRUE then |
CH |
capthist object with telemetryxy attribute |
inflation |
numeric tolerance for polygon |
object |
secr traps object |
value |
character telemetry type replacement value |
... |
other arguments |
It is assumed that a number of animals have been radiotagged, and their telemetry data
(xy-coordinates) have been input to telemetryCH
, perhaps using
read.capthist
with detector = "telemetryonly"
and fmt =
"XY"
, or with read.telemetry
.
A new capthist object is built comprising all the detection
histories in detectionCH
, plus empty (all-zero) histories for
every telemetered animal not in detectionCH
. Telemetry is associated with new sampling occasions and a new detector (nominally at the same point as the first in detectionCH
). The number of telemetry fixes of each animal is recorded in the relevant cell of the new capthist object (CH[i, s, K+1] for animal i and occasion s if there were K detectors in detectionCH).
The new sampling occasion(s) are assigned the detector type ‘telemetry’ in the traps attribute of the output capthist object, and the traps attribute telemetrytype
is set to the value provided. The telemetry type may be “independent” (no matching of individuals in captured and telemetered samples), “dependent” (telemetered animals are a subset of captured animals) or “concurrent” (histories may be capture-only, telemetry-only or both capture and telemetry).
The telemetry locations are carried over from telemetryCH as attribute ‘xylist’ (each
component of xylist holds the coordinates of one animal; use
telemetryxy
to extract).
The default behaviour of 'addTelemetry' is to automatically collapse all telemetry occasions into one. This is computationally more efficient than the alternative, but closes off some possible models.
xy2CH
partly reverses addTelemetry
: the location
information in the telemetryxy attribute is converted back to a capthist with
detector type ‘telemetry’.
A single-session capthist object with the same detector type as
detectionCH
, but possibly with empty rows and an ‘telemetryxy’ attribute.
Telemetry provides independent data on the location and presence of a sample of animals. These animals may be missed in the main sampling that gives rise to detectionCH i.e., they may have all-zero detection histories.
The ‘telemetry’ detector type is used for telemetry occasions in a combined dataset.
capthist
,
make.telemetry
,
read.telemetry
,
telemetryxy
telemetered
## Not run: # Generate some detection and telemetry data, combine them using # addTelemetry, and perform analyses # detectors te <- make.telemetry() tr <- make.grid(detector = "proximity") # simulated population and 50% telemetry sample totalpop <- sim.popn(tr, D = 20, buffer = 100) tepop <- subset(totalpop, runif(nrow(totalpop)) < 0.5) # simulated detection histories and telemetry # the original animalID (renumber = FALSE) are needed for matching trCH <- sim.capthist(tr, popn = totalpop, renumber = FALSE, detectfn = "HHN") teCH <- sim.capthist(te, popn = tepop, renumber=FALSE, detectfn = "HHN", detectpar = list(lambda0 = 3, sigma = 25)) combinedCH <- addTelemetry(trCH, teCH) # summarise and display summary(combinedCH) plot(combinedCH, border = 150) ncapt <- apply(combinedCH,1,sum) points(totalpop[row.names(combinedCH)[ncapt==0],], pch = 1) points(totalpop[row.names(combinedCH)[ncapt>0],], pch = 16) # for later comparison of precision we must fix the habitat mask mask <- make.mask(tr, buffer = 100) fit.tr <- secr.fit(trCH, mask = mask, CL = TRUE, detectfn = "HHN") ## trapping alone fit.te <- secr.fit(teCH, mask = mask, CL = TRUE, start = log(20), ## telemetry alone detectfn = "HHN") fit2 <- secr.fit(combinedCH, mask = mask, CL = TRUE, ## combined detectfn = "HHN") # improved precision when focus on realised population # (compare CVD) derived(fit.tr, distribution = "binomial") derived(fit2, distribution = "binomial") # may also use CL = FALSE secr.fit(combinedCH, CL = FALSE, detectfn = "HHN", trace = FALSE) ## End(Not run)
## Not run: # Generate some detection and telemetry data, combine them using # addTelemetry, and perform analyses # detectors te <- make.telemetry() tr <- make.grid(detector = "proximity") # simulated population and 50% telemetry sample totalpop <- sim.popn(tr, D = 20, buffer = 100) tepop <- subset(totalpop, runif(nrow(totalpop)) < 0.5) # simulated detection histories and telemetry # the original animalID (renumber = FALSE) are needed for matching trCH <- sim.capthist(tr, popn = totalpop, renumber = FALSE, detectfn = "HHN") teCH <- sim.capthist(te, popn = tepop, renumber=FALSE, detectfn = "HHN", detectpar = list(lambda0 = 3, sigma = 25)) combinedCH <- addTelemetry(trCH, teCH) # summarise and display summary(combinedCH) plot(combinedCH, border = 150) ncapt <- apply(combinedCH,1,sum) points(totalpop[row.names(combinedCH)[ncapt==0],], pch = 1) points(totalpop[row.names(combinedCH)[ncapt>0],], pch = 16) # for later comparison of precision we must fix the habitat mask mask <- make.mask(tr, buffer = 100) fit.tr <- secr.fit(trCH, mask = mask, CL = TRUE, detectfn = "HHN") ## trapping alone fit.te <- secr.fit(teCH, mask = mask, CL = TRUE, start = log(20), ## telemetry alone detectfn = "HHN") fit2 <- secr.fit(combinedCH, mask = mask, CL = TRUE, ## combined detectfn = "HHN") # improved precision when focus on realised population # (compare CVD) derived(fit.tr, distribution = "binomial") derived(fit2, distribution = "binomial") # may also use CL = FALSE secr.fit(combinedCH, CL = FALSE, detectfn = "HHN", trace = FALSE) ## End(Not run)
Terse report on the fit of one or more spatially explicit capture–recapture models. Models with smaller values of AIC (Akaike's Information Criterion) are preferred. Extraction ([) and logLik methods are included.
## S3 method for class 'secr' AIC(object, ..., sort = TRUE, k = 2, dmax = 10, criterion = c("AIC","AICc"), chat = NULL) ## S3 method for class 'secrlist' AIC(object, ..., sort = TRUE, k = 2, dmax = 10, criterion = c("AIC","AICc"), chat = NULL) ## S3 method for class 'secr' logLik(object, ...) secrlist(..., names = NULL) ## S3 method for class 'secrlist' x[i]
## S3 method for class 'secr' AIC(object, ..., sort = TRUE, k = 2, dmax = 10, criterion = c("AIC","AICc"), chat = NULL) ## S3 method for class 'secrlist' AIC(object, ..., sort = TRUE, k = 2, dmax = 10, criterion = c("AIC","AICc"), chat = NULL) ## S3 method for class 'secr' logLik(object, ...) secrlist(..., names = NULL) ## S3 method for class 'secrlist' x[i]
object |
|
... |
other |
sort |
logical for whether rows should be sorted by ascending AIC |
k |
numeric, penalty per parameter to be used; always k = 2 in this method |
dmax |
numeric, maximum AIC difference for inclusion in confidence set |
criterion |
character, criterion to use for model comparison and weights |
chat |
numeric optional variance inflation factor for quasi-AIC |
names |
character vector of names (optional) |
x |
secrlist |
i |
indices |
Models to be compared must have been fitted to the same data and use the
same likelihood method (full vs conditional). From version 4.1 a warning is
issued if AICcompatible
reveals a problem.
AIC is given by
where is the number of "beta" parameters estimated.
AIC with small sample adjustment is given by
The sample size is the number of individuals observed at least once (i.e. the
number of rows in
capthist
).
Model weights are calculated as
where refers to differences in AIC or AICc depending on the
argument ‘criterion’ (see Notes).
Models for which delta > dmax
are given a weight of zero and are
excluded from the summation. Model weights may be used to form
model-averaged estimates of real or beta parameters with
modelAverage
(see also Buckland et al. 1997, Burnham and
Anderson 2002).
The argument k
is included for consistency with the generic method AIC
.
secrlist
forms a list of fitted models (an object of class
‘secrlist’) from the fitted models in .... Arguments may include
secrlists. If secr components are named the model names will be retained unless ‘names’ is specified.
(see Examples).
If chat () is provided then quasi-AIC values are computed (secr >= 4.6.0):
A data frame with one row per model. By default, rows are sorted by ascending 'criterion' (default AIC see Notes).
model |
character string describing the fitted model |
detectfn |
shape of detection function fitted (halfnormal vs hazard-rate) |
npar |
number of parameters estimated |
logLik |
maximized log likelihood |
AIC |
Akaike's Information Criterion |
AICc |
AIC with small-sample adjustment of Hurvich & Tsai (1989) |
And depending on criterion
:
dAIC |
difference between AIC of this model and the one with smallest AIC |
AICwt |
AIC model weight |
or
dAICc |
difference between AICc of this model and the one with smallest AICc |
AICcwt |
AICc model weight |
logLik.secr
returns an object of class ‘logLik’ that has
attribute df
(degrees of freedom = number of estimated
parameters).
If the variance inflation factor 'chat' is provided then outputs AIC, AICc etc. are replaced by the corresponding quasi-AIC values labelled QAIC, QAICc etc.
It is not be meaningful to compare models by AIC if they relate to
different data (see AICcompatible
).
Specifically:
an ‘secrlist’ generated and saved to file by mask.check
may be supplied as the object argument of AIC.secrlist
, but the
results are not informative
models fitted by the conditional likelihood (CL = TRUE
) and
full likelihood (CL = FALSE
) methods cannot be compared
hybrid mixture models (using hcov argument of secr.fit) should not be compared with other models
grouped models (using groups argument of secr.fit) should not be compared with other models
multi-session models should not be compared with single-session models based on the same data.
A likelihood-ratio test (LR.test
) is a more direct way to
compare two models.
The issue of goodness-of-fit and possible adjustment of AIC for overdispersion has yet to be addressed (cf QAIC in MARK).
The user may select between AIC and AICc for comparing models. AICc is widely used, but AIC may be better for model averaging even when samples are small (Turek and Fletcher 2012; Fletcher 2019, p. 60). The default was changed to AIC in version 5.0.0).
Buckland S. T., Burnham K. P. and Augustin, N. H. (1997) Model selection: an integral part of inference. Biometrics 53, 603–618.
Burnham, K. P. and Anderson, D. R. (2002) Model Selection and Multimodel Inference: A Practical Information-Theoretic Approach. Second edition. New York: Springer-Verlag.
Fletcher, D. (2019) Model averaging. SpringerBriefs in Statistics. Berlin: Springer-Verlag.
Hurvich, C. M. and Tsai, C. L. (1989) Regression and time series model selection in small samples. Biometrika 76, 297–307.
Turek, D. and Fletcher, D. (2012) Model-averaged Wald confidence intervals. Computational statistics and data analysis 56, 2809–2815.
AICcompatible
,
modelAverage
,
AIC
,
secr.fit
,
print.secr
,
score.test
,
LR.test
,
deviance.secr
## Compare two models fitted previously ## secrdemo.0 is a null model ## secrdemo.b has a learned trap response AIC(secrdemo.0, secrdemo.b) ## Form secrlist and pass to AIC.secr temp <- secrlist(null = secrdemo.0, learnedresponse = secrdemo.b) AIC(temp)
## Compare two models fitted previously ## secrdemo.0 is a null model ## secrdemo.b has a learned trap response AIC(secrdemo.0, secrdemo.b) ## Form secrlist and pass to AIC.secr temp <- secrlist(null = secrdemo.0, learnedresponse = secrdemo.b) AIC(temp)
Determine whether models can be compared by AIC. Incompatibility may be due to difference in the data or the specifications of the groups, hcov or binomN arguments to secr.fit
,
## S3 method for class 'secr' AICcompatible(object, ...) ## S3 method for class 'secrlist' AICcompatible(object, ...)
## S3 method for class 'secr' AICcompatible(object, ...) ## S3 method for class 'secrlist' AICcompatible(object, ...)
object |
|
... |
other |
The capthist objects are checked for strict identity with the function identical
.
All elements in the output must be TRUE for valid AIC comparison or model averaging using AIC or AICc.
Named logical vector with elements ‘data’, ‘CL’, ‘groups’, ‘hcov’ and ‘binomN’.
AICcompatible(secrdemo.0, secrdemo.CL) ## Not run: ## A common application of AICcompatible() is to determine ## the compatibility of models fitted with and without the ## fastproximity option. ovenCHp1 <- reduce(ovenCHp, by = 'all', outputdetector = 'count') ob1 <- secr.fit(ovenCHp, buffer = 300, details = list(fastproximity = TRUE)) ob2 <- secr.fit(ovenCHp1, buffer = 300, details = list(fastproximity = FALSE)) ob3 <- secr.fit(ovenCHp1, buffer = 300, details = list(fastproximity = FALSE), binomN = 1) AICcompatible(ob1,ob2) AICcompatible(ob1,ob3) ## End(Not run)
AICcompatible(secrdemo.0, secrdemo.CL) ## Not run: ## A common application of AICcompatible() is to determine ## the compatibility of models fitted with and without the ## fastproximity option. ovenCHp1 <- reduce(ovenCHp, by = 'all', outputdetector = 'count') ob1 <- secr.fit(ovenCHp, buffer = 300, details = list(fastproximity = TRUE)) ob2 <- secr.fit(ovenCHp1, buffer = 300, details = list(fastproximity = FALSE)) ob3 <- secr.fit(ovenCHp1, buffer = 300, details = list(fastproximity = FALSE), binomN = 1) AICcompatible(ob1,ob2) AICcompatible(ob1,ob3) ## End(Not run)
Method for generic as.data.frame
function that partially reverses make.capthist
.
## S3 method for class 'capthist' as.data.frame(x, row.names = NULL, optional = FALSE, covariates = FALSE, fmt = c("trapID", "XY"), ...) ## S3 method for class 'traps' as.data.frame(x, row.names = NULL, optional = FALSE, usage = FALSE, covariates = FALSE, ...) ## S3 method for class 'capthist' as.array(x, ...)
## S3 method for class 'capthist' as.data.frame(x, row.names = NULL, optional = FALSE, covariates = FALSE, fmt = c("trapID", "XY"), ...) ## S3 method for class 'traps' as.data.frame(x, row.names = NULL, optional = FALSE, usage = FALSE, covariates = FALSE, ...) ## S3 method for class 'capthist' as.array(x, ...)
x |
|
row.names |
unused argument of generic function |
optional |
unused argument of generic function |
covariates |
logical or a character vector of covariates to export |
fmt |
character string for capture format |
usage |
logical; if TRUE then usage columns are appended if present |
... |
other arguments (not used) |
By default individual covariates are not exported. When exported they are repeated for each detection of an individual.
A data frame or list of data frames (in the case of a multisession input).
For capthist objects –
The core columns are (Session, ID, Occasion, TrapID) or (Session, ID, Occasion, x, y),
depending on the value of fmt
. Additional columns for covariates and signal
strength (detector ‘signal’) are appended to the right.
For traps objects –
The core columns are (x, y). Usage columns are named u1, u2, ..., uS where S is the number of occasions.
The as.array
method for capthist objects returns an object with the same dimensions and dimnames but different class, or a list of such objects in the case of multisession input.
as.data.frame (captdata) as.data.frame (traps(captdata))
as.data.frame (captdata) as.data.frame (traps(captdata))
This function is used primarily for plotting covariates, for which the plot.mask function has greater functionality than plot.traps
. It also generates pretty maps of grid cells.
as.mask(x)
as.mask(x)
x |
an object of class 'traps' |
A mask derived by coercion with as.mask
may behave
unpredictably e.g., in secr.fit
.
If x
is a single-session traps object –
an object of class c("mask", "data.frame")
If x
is a multi-session traps object –
an object of class c("mask", "list"), for which each component is a single-session mask.
make.mask
, plot.mask
,
mask, traps
plot(as.mask(traps(captdata)), dots = FALSE, meshcol = "black") plot(traps(captdata), add = TRUE)
plot(as.mask(traps(captdata)), dots = FALSE, meshcol = "black") plot(traps(captdata), add = TRUE)
This function converts a spatstat "ppp" object (Baddeley et al. 2015), making it easier to use the simulation capability of spatstat in secr.
as.popn(x)
as.popn(x)
x |
an object of class 'ppp' |
Not all attributes are carried over.
An object of class c("popn", "data.frame") with attribute "boundingbox". The attribute "Lambda" (spatstat class "im") is also carried over if present (used for the intensity surface of LGCP simulations).
Baddeley, A., Rubak, E., and Turner, R. 2015. Spatial Point Patterns: Methodology and Applications with R. Chapman and Hall/CRC Press, London. ISBN 9781482210200, https://www.routledge.com/Spatial-Point-Patterns-Methodology-and-Applications-with-R/Baddeley-Rubak-Turner/p/book/9781482210200/.
Find plausible initial parameter values for secr.fit
. A
simple SECR model is fitted by a fast ad hoc method.
autoini(capthist, mask, detectfn = 0, thin = 0.2, tol = 0.001, binomN = 1, adjustg0 = TRUE, adjustsigma = 1.2, ignoreusage = FALSE, ncores = NULL)
autoini(capthist, mask, detectfn = 0, thin = 0.2, tol = 0.001, binomN = 1, adjustg0 = TRUE, adjustsigma = 1.2, ignoreusage = FALSE, ncores = NULL)
capthist |
|
mask |
|
detectfn |
integer code or character string for shape of detection function 0 = halfnormal |
thin |
proportion of points to retain in mask |
tol |
numeric absolute tolerance for numerical root finding |
binomN |
integer code for distribution of counts (see |
adjustg0 |
logical for whether to adjust g0 for usage (effort) and binomN |
adjustsigma |
numeric scalar applied to RPSV(capthist, CC = TRUE) |
ignoreusage |
logical for whether to discard usage information from
|
ncores |
integer number of threads to be used for parallel processing |
Plausible starting values are needed to avoid numerical
problems when fitting SECR models. Actual models
to be fitted will usually have more than the three basic parameters
output by autoini
; other initial values can usually be set to
zero for secr.fit
. If the algorithm encounters problems obtaining
a value for g0, the default value of 0.1 is returned.
Only the halfnormal detection function is currently available in autoini
(cf
other options in e.g. detectfn and sim.capthist
).
autoini
implements a modified version of the algorithm proposed
by Efford et al. (2004). In outline, the algorithm is
Find value of sigma that predicts the 2-D dispersion of individual locations (see RPSV
).
Find value of g0 that, with sigma, predicts the observed mean number of captures per individual (by algorithm of Efford et al. (2009, Appendix 2))
Compute the effective sampling area from g0, sigma, using thinned mask (see esa
)
Compute D = /esa(g0, sigma), where
is the number of individuals detected
Here ‘find’ means solve numerically for zero difference between the observed and predicted values, using uniroot
.
Halfnormal sigma is estimated with RPSV(capthist, CC = TRUE)
. The factor adjustsigma
is applied as a crude correction for truncation of movements at the edge of the detector array.
If RPSV
cannot be computed the algorithm tries to use observed
mean recapture distance . Computation of
fails if there no recaptures, and all returned
values are NA.
If the mask has more than 100 points then a proportion 1–thin
of
points are discarded at random to speed execution.
The argument tol
is passed to uniroot
. It may be a
vector of two values, the first for g0 and the second for sigma.
If traps(capthist)
has a usage attribute (defining effort
on each occasion at each detector) then the value of g0 is divided by
the mean of the non-zero elements of usage. This adjustment is not
precise.
If adjustg0
is TRUE then an adjustment is made to g0 depending
on the value of binomN
. For Poisson counts (binomN = 0
)
the adjustment is linear on effort (adjusted.g0 = g0 /
usage). Otherwise, the adjustment is on the hazard scale (adjusted.g0 =
1 – (1 – g0) ^ (1 / (usage x binomN))). An arithmetic average is taken
over all non-zero usage values (i.e. over used detectors and times). If
usage is not specified it is taken to be 1.0.
Setting ncores = NULL
uses the existing value from the environment variable
RCPP_PARALLEL_NUM_THREADS (see setNumThreads
).
A list of parameter values :
D |
Density (animals per hectare) |
g0 |
Magnitude (intercept) of detection function |
sigma |
Spatial scale of detection function (m) |
autoini
always uses the Euclidean distance between detectors and
mask points.
You may get this message from secr.fit: “'autoini' failed to find g0; setting initial g0 = 0.1”. If the fitted model looks OK (reasonable estimates, non-missing SE) there is no reason to worry about the starting values. If you get this message and model fitting fails then supply your own values in the start argument of secr.fit.
Efford, M. G., Dawson, D. K. and Robbins C. S. (2004) DENSITY: software for analysing capture–recapture data from passive detector arrays. Animal Biodiversity and Conservation 27, 217–228.
Efford, M. G., Dawson, D. K. and Borchers, D. L. (2009) Population density estimated from locations of individuals on a passive detector array. Ecology 90, 2676–2682.
capthist
, mask
, secr.fit
, dbar
## Not run: demotraps <- make.grid() demomask <- make.mask(demotraps) demoCH <- sim.capthist (demotraps, popn = list(D = 5, buffer = 100), seed = 321) autoini (demoCH, demomask) ## End(Not run)
## Not run: demotraps <- make.grid() demomask <- make.mask(demotraps) demoCH <- sim.capthist (demotraps, popn = list(D = 5, buffer = 100), seed = 321) autoini (demoCH, demomask) ## End(Not run)
Forms a new covariate, replacing values of an old covariate by the central value of equal-width bins.
binCovariate(object, covname, width)
binCovariate(object, covname, width)
object |
secr object with covariates attribute (capthist, traps, mask) |
covname |
character name of covariate |
width |
numeric bin width |
The name of the new covariate is paste0(covname, width)
.
Fails if covariate not found or is not numeric or there is already a covariate with the new name.
Multi-session objects are handled appropriately.
Object of the same class as the input with new covariate.
# bin values of skink snout-vent length (mm) infraCH <- binCovariate (infraCH, 'SVL', 5) table(covariates(infraCH[[1]])$SVL5) # bin values of trap covariate 'HtBrack' (height of bracken, cm) traps(infraCH) <- binCovariate(traps(infraCH), "HtBrack", 20) table(covariates(traps(infraCH)[[1]])$HtBrack20)
# bin values of skink snout-vent length (mm) infraCH <- binCovariate (infraCH, 'SVL', 5) table(covariates(infraCH[[1]])$SVL5) # bin values of trap covariate 'HtBrack' (height of bracken, cm) traps(infraCH) <- binCovariate(traps(infraCH), "HtBrack", 20) table(covariates(traps(infraCH)[[1]])$HtBrack20)
American black bears Ursus americanus were surveyed with baited hair snags in the Great Smoky Mountains National Park, Tennessee, in the summer of 2003.
blackbearCH GSM blackbear.0 blackbear.h2bk
blackbearCH GSM blackbear.0 blackbear.h2bk
American black bears Ursus americanus were surveyed in the Tennessee sector of Great Smoky Mountains National Park over 9 June–15 August 2003. Baited hair snags (barbed wire enclosures) were operated for 10 weeks at 65 sites, about 1 km apart and mostly close to trails. Bait consisted of bakery products in a small waxed-paper bag. Raspberry extract was used as a scent lure.
Genotyping and non-spatial capture-recapture analysis of a data subset were described by Settlage et al. (2008). The sex of each genotyped bear was determined subsequently and some additional samples were included (J. Laufenberg pers. comm. 2012-05-09).
The dataset is a single-session capthist object with binary proximity detector type. Snags were visited weekly, so there were 10 occasions in the raw data. A single covariate ‘sex’ was recorded for each individual.
The dataset comprises 282 detections of 81 females and 58 males. Female 15 apparently made a long movement (17 km) between occasions 1 and 3.
The hair snag array sampled less than 20% of the area of the park. The
unforested area outside the park on the northwestern boundary of the study area
was not considered to be black bear habitat (F. van Manen pers. comm. 2012-05-18)
and should be excluded in analyses. The approximate boundary of the park is
included as a shapefile ‘GSMboundary.shp’ in the ‘extdata’ folder of the package
and as the sf sfc_POLYGON object GSM
. The latter may be used in
make.mask
(see Examples).
Two models (blackbear.0 and blackbear.h2bk) were fitted as shown in the Examples.
The data were provided by Jared Laufenberg, Frank van Manen and Joe Clark.
Settlage, K. E., Van Manen, F. T., Clark, J. D., and King, T. L. (2008) Challenges of DNA-based mark–recapture studies of American black bears. Journal of Wildlife Management 72, 1035–1042.
summary(blackbearCH) # GSM is the approximate boundary of Great Smoky Mountains National Park # Make a habitat mask restricted to the park tr <- traps(blackbearCH) msk <- make.mask(tr, buffer = 6000, type = 'trapbuffer', poly = GSM) # Plot plot(GSM) plot(msk, add = TRUE) plot(blackbearCH, tracks = TRUE, add = TRUE) plot(tr, add = TRUE) # Fit models # suppress fastproximity to allow learned response setNumThreads() # as appropriate # null model blackbear.0 <- secr.fit(blackbearCH, detectfn = 'EX', hcov = 'sex', mask = msk, details = list(fastproximity = FALSE), trace = FALSE) # sex differences and site-specific behavioural response blackbear.h2bk <- secr.fit(blackbearCH, detectfn = 'EX', hcov = 'sex', model = list(g0~bk+h2, sigma~h2), mask = msk, details = list(fastproximity = FALSE), trace = FALSE) AIC(blackbear.0, blackbear.h2bk) summary(blackbear.h2bk) # How many if we extrapolate to GSM NP? region.N(blackbear.h2bk, region = GSM)
summary(blackbearCH) # GSM is the approximate boundary of Great Smoky Mountains National Park # Make a habitat mask restricted to the park tr <- traps(blackbearCH) msk <- make.mask(tr, buffer = 6000, type = 'trapbuffer', poly = GSM) # Plot plot(GSM) plot(msk, add = TRUE) plot(blackbearCH, tracks = TRUE, add = TRUE) plot(tr, add = TRUE) # Fit models # suppress fastproximity to allow learned response setNumThreads() # as appropriate # null model blackbear.0 <- secr.fit(blackbearCH, detectfn = 'EX', hcov = 'sex', mask = msk, details = list(fastproximity = FALSE), trace = FALSE) # sex differences and site-specific behavioural response blackbear.h2bk <- secr.fit(blackbearCH, detectfn = 'EX', hcov = 'sex', model = list(g0~bk+h2, sigma~h2), mask = msk, details = list(fastproximity = FALSE), trace = FALSE) AIC(blackbear.0, blackbear.h2bk) summary(blackbear.h2bk) # How many if we extrapolate to GSM NP? region.N(blackbear.h2bk, region = GSM)
Convert data between ‘capthist’ and BUGS input format.
read.DA(DAlist, detector = "polygonX", units = 1, session = 1, Y = "Y", xcoord = "U1", ycoord = "U2", xmin = "Xl", xmax = "Xu", ymin = "Yl", ymax = "Yu", buffer = "delta", verify = TRUE) write.DA(capthist, buffer, nzeros = 200, units = 1)
read.DA(DAlist, detector = "polygonX", units = 1, session = 1, Y = "Y", xcoord = "U1", ycoord = "U2", xmin = "Xl", xmax = "Xu", ymin = "Yl", ymax = "Yu", buffer = "delta", verify = TRUE) write.DA(capthist, buffer, nzeros = 200, units = 1)
DAlist |
list containing data in BUGS format |
detector |
character value for detector type: ‘polygon’ or ‘polygonX’ |
units |
numeric for scaling output coordinates |
session |
numeric or character label used in output |
Y |
character, name of binary detection history matrix (animals x occasions) |
xcoord |
character, name of matrix of x-coordinates for each detection in |
ycoord |
character, name of matrix of y-coordinates for each detection in |
xmin |
character, name of coordinate of state space boundary |
xmax |
character, name of coordinate of state space boundary |
ymin |
character, name of coordinate of state space boundary |
ymax |
character, name of coordinate of state space boundary |
buffer |
see Details |
verify |
logical if TRUE then the resulting capthist object is
checked with |
capthist |
|
nzeros |
level of data augmentation (all-zero detection histories) |
Data for OpenBUGS or WinBUGS called from R using the package R2WinBUGS (Sturtz et al. 2005) take the form of an R list.
These functions are limited at present to binary data from a square
quadrat such as used by Royle and Young (2008). Marques et al. (2011)
provide an R function create.data()
for generating simulated
datasets of this sort (see sim.capthist
for equivalent
functionality).
When reading BUGS data –
The character values Y
, xcoord
, ycoord
,
xmin
etc. are used to locate the data within DAlist
,
allowing for variation in the input names.
The number of sampling occasions is taken from the number of columns
in Y
. Each value in Y
should be 0 or 1. Coordinates may
be missing
A numeric value for buffer
is the distance (in the original
units) by which the limits Xl, Xu etc. should be shrunk to give the
actual plot limits. If buffer
is character then a component of
DAlist
contains the required numeric value.
Coordinates in the output will be multiplied by the scalar
units
.
Augmentation rows corresponding to ‘all-zero’ detection histories in
Y
, xcoord
, and ycoord
are discarded.
When writing BUGS data –
Null (all-zero) detection histories are added to the matrix of
detection histories Y
, and missing (NA) rows are added to the
coordinate matrices xcoord
and ycoord
.
Coordinates in the output will be divided by the scalar
units
.
For read.DA
, an object of class ‘capthist’.
For write.DA
, a list with the components
Xl | left edge of state space |
Xu | right edge of state space |
Yl | bottom edge of state space |
Yu | top edge of state space |
delta | buffer between edge of state space and quadrat |
nind | number of animals observed |
nzeros | number of added all-zero detection histories |
T | number of sampling occasions |
Y | binary matrix of detection histories (dim = c(nind+nzeros, T)) |
U1 | matrix of x-coordinates, dimensioned as Y |
U2 | matrix of y-coordinates, dimensioned as Y |
U1 and U2 are ‘NA’ where animal was not detected.
Marques, T. A., Thomas, L. and Royle, J. A. (2011) A hierarchical model for spatial capture–recapture data: Comment. Ecology 92, 526–528.
Royle, J. A. and Young, K. V. (2008) A hierarchical model for spatial capture–recapture data. Ecology 89, 2281–2289.
Sturtz, S., Ligges, U. and Gelman, A. (2005) R2WinBUGS: a package for running WinBUGS from R. Journal of Statistical Software 12, 1–16.
hornedlizardCH
, verify
, capthist
write.DA (hornedlizardCH, buffer = 100, units = 100) ## In this example, the input uses Xl, Xu etc. ## for the limits of the plot itself, so buffer = 0. ## Input is in hundreds of metres. ## First, obtain the list lzdata olddir <- setwd (system.file("extdata", package="secr")) source ("lizarddata.R") setwd(olddir) str(lzdata) ## Now convert to capthist tempcapt <- read.DA(lzdata, Y = "H", xcoord = "X", ycoord = "Y", buffer = 0, units = 100) ## Not run: plot(tempcapt) secr.fit(tempcapt, trace = FALSE) ## etc. ## End(Not run)
write.DA (hornedlizardCH, buffer = 100, units = 100) ## In this example, the input uses Xl, Xu etc. ## for the limits of the plot itself, so buffer = 0. ## Input is in hundreds of metres. ## First, obtain the list lzdata olddir <- setwd (system.file("extdata", package="secr")) source ("lizarddata.R") setwd(olddir) str(lzdata) ## Now convert to capthist tempcapt <- read.DA(lzdata, Y = "H", xcoord = "X", ycoord = "Y", buffer = 0, units = 100) ## Not run: plot(tempcapt) secr.fit(tempcapt, trace = FALSE) ## etc. ## End(Not run)
A capthist
object encapsulates all data needed by
secr.fit
, except for the optional habitat mask.
An object of class capthist
holds spatial capture histories,
detector (trap) locations, individual covariates and other data needed
for a spatially explicit capture-recapture analysis with
secr.fit
.
A capthist
is primarily an array of values with dim(capthist) = c(nc,
noccasions, ntraps) where nc is the number of detected individuals.
Values maybe binary ({–1, 0, 1}) or integer depending on the detector type.
Deaths during the experiment are represented as negative values.
Ancillary data are retained as attributes of a capthist
object as follows:
traps — object of class traps
(required)
session — session identifier (required)
covariates — dataframe of individual covariates (optional)
cutval — threshold of signal strength for detection (‘signal’ only)
signalframe — signal strength values etc., one row per detection (‘signal’ only)
detectedXY — dataframe of coordinates for location within polygon (‘polygon’-like detectors only)
xylist — coordinates of telemetered animals
Tu — detectors x occasions matrix of sightings of unmarked animals
Tm — detectors x occasions matrix of sightings of marked but unidentified animals
Tn — detectors x occasions matrix of sightings with unknown mark status
read.capthist
is adequate for most data input. Alternatively, the parts of a
capthist object can be assembled with the function make.capthist
.
Use sim.capthist
for Monte Carlo simulation
(simple models only). Methods are provided to display and manipulate
capthist
objects (print, summary, plot, rbind, subset, reduce)
and to extract and replace attributes (covariates, traps, xy).
A multi-session capthist
object is a list in which each component
is a capthist
for a single session. The list maybe derived
directly from multi-session input in Density format, or by combining
existing capthist
objects with MS.capthist
.
Early versions of secr (before 3.0) used an individual x occasion matrix
for data from single-catch and multi-catch traps, instead of a 3-D array.
Entries in the matrix corresponded to trap numbers. The function
updateCH
converts the old format.
Borchers, D. L. and Efford, M. G. (2008) Spatially explicit maximum likelihood methods for capture–recapture studies. Biometrics 64, 377–385.
Efford, M. G., Borchers D. L. and Byrom, A. E. (2009) Density estimation by spatially explicit capture-recapture: likelihood-based methods. In: D. L. Thomson, E. G. Cooch and M. J. Conroy (eds) Modeling Demographic Processes in Marked Populations. Springer, New York. Pp. 255–269.
traps
, secr.fit
,
read.capthist
, make.capthist
,
sim.capthist
,
subset.capthist
, rbind.capthist
,
MS.capthist
,
reduce.capthist
, mask
Extract parts of an object of class ‘capthist’.
animalID(object, names = TRUE, sortorder = c("snk", "ksn")) occasion(object, sortorder = c("snk", "ksn")) trap(object, names = TRUE, sortorder = c("snk", "ksn")) alive(object, sortorder = c("snk", "ksn")) alongtransect(object, tol = 0.01) xy(object) xy(object) <- value telemetryxy(object, includeNULL = FALSE) telemetryxy(object) <- value telemetered(object)
animalID(object, names = TRUE, sortorder = c("snk", "ksn")) occasion(object, sortorder = c("snk", "ksn")) trap(object, names = TRUE, sortorder = c("snk", "ksn")) alive(object, sortorder = c("snk", "ksn")) alongtransect(object, tol = 0.01) xy(object) xy(object) <- value telemetryxy(object, includeNULL = FALSE) telemetryxy(object) <- value telemetered(object)
object |
a ‘capthist’ object |
names |
if FALSE the values returned are numeric indices rather than names |
sortorder |
character code for sort order (see Details) |
tol |
tolerance for snapping to transect line (m) |
value |
replacement value (see Details) |
includeNULL |
logical; if TRUE a NULL component is included for untelemetered animals |
These functions extract data on detections, ignoring occasions when an
animal was not detected. By default, detections are ordered by occasion, animalID
and trap (sortorder = "snk"
). The alternative is to order by
trap, occasion and animalID (sortorder = "ksn"
). (‘n’, ‘s’ and ‘k’ are the
indices used internally for animals, occasions and traps respectively).
For historical reasons, "ksn" is used for locations within polygons and similar
(xy
).
trap
returns polygon or transect numbers if traps(object)
has detector type ‘polygon’ or ‘transect’.
alongtransect
returns the distance of each detection from the
start of the transect with which it is associated.
Replacement values must precisely match object
in number of
detections and in their order. xy<-
expects a dataframe of x and y
coordinates for points of detection within a ‘polygon’ or ‘transect’
detector. telemetryxy<-
expects a list of dataframes, one per telemetered animal.
For animalID
and trap
a vector of numeric or character values, one per detection.
For alive
a vector of logical values, one per detection.
For occasion
, a vector of numeric values, one per detection.
For xy
, a dataframe with one row per detection and columns ‘x’ and ‘y’.
If object
has multiple sessions, the result is a list with one
component per session.
capthist
, polyID
, signalmatrix
## `captdata' is a demonstration dataset animalID(captdata) temp <- sim.capthist(popn = list(D = 1), make.grid(detector = "count")) cbind(ID = as.numeric(animalID(temp)), occ = occasion(temp), trap = trap(temp))
## `captdata' is a demonstration dataset animalID(captdata) temp <- sim.capthist(popn = list(D = 1), make.grid(detector = "count")) cbind(ID = as.numeric(animalID(temp)), occ = occasion(temp), trap = trap(temp))
Activity centres may be clumped (overdispersed) relative to a Poisson distribution,
the model used in secr.fit
(Borchers and Efford 2008). This can cause
the sampling variance of density estimates to be understated. One solution currently under investigation is to apply a variance inflation factor, a measure of overdispersion, based on the number of individuals detected at each detector (Bischof et al. 2020).
Functions described here compute the observed (nk) or expected (Enk) number of individuals detected at each detector and use that to compute Fletcher's estimate of overdispersion
for use as a variance inflation factor.
Enk
uses exact formulae for 'multi', 'proximity' and 'count' detector types. Other types may be simulated by setting a positive value for 'nrepl', which should be large (e.g., nrepl = 10000).
adjustVarD
adjusts the SE and confidence limits of density estimates
using Fletcher's . The implementation is limited to simple detection models
(see Warnings).
See Cooch and White (2022) for an introduction to measurement of overdispersion in capture–recapture. The focus here is on overdispersion of activity centres relative to a Poisson distribution, rather than on non-independence in the spatial detection process.
nk(capthist) Enk(D, mask, traps, detectfn = 0, detectpar = list(g0 = 0.2, sigma = 25, z = 1), noccasions = NULL, binomN = NULL, userdist = NULL, ncores = NULL, nrepl = NULL) chat.nk(object, nsim = NULL, ...) adjustVarD(object, chatmin = 1, alpha = 0.05, chat = NULL)
nk(capthist) Enk(D, mask, traps, detectfn = 0, detectpar = list(g0 = 0.2, sigma = 25, z = 1), noccasions = NULL, binomN = NULL, userdist = NULL, ncores = NULL, nrepl = NULL) chat.nk(object, nsim = NULL, ...) adjustVarD(object, chatmin = 1, alpha = 0.05, chat = NULL)
capthist |
secr |
D |
numeric density, either scalar or vector of length nrow(mask) |
mask |
single-session habitat mask |
traps |
|
detectfn |
integer code for detection function q.v. |
detectpar |
a named list giving a value for each parameter of detection function |
noccasions |
number of sampling intervals (occasions) |
binomN |
integer code for discrete distribution (see
|
userdist |
user-defined distance function or matrix (see userdist) |
ncores |
integer number of threads |
nrepl |
integer number of replicates for E(nk) by simulation (optional) |
object |
fitted secr model or dataframe (see Warnings for restrictions) |
nsim |
integer number of c-hat values to simulate (optional) |
... |
other arguments passed to |
chatmin |
minimum value of Fletcher's |
alpha |
alpha level for confidence intervals |
chat |
numeric chat (optional) |
If traps
has a usage attribute then noccasions
is
set accordingly; otherwise it must be provided.
The environment variable RCPP_PARALLEL_NUM_THREADS determines the number of
parallel threads. It is set to the value of ncores
, unless that is NULL
(see setNumThreads
).
A conventional variance inflation factor due to Wedderburn (1974) is
where
is the number of detectors,
is the number of estimated parameters, and
Fletcher's is an improvement on
that is less affected
by small expected counts. It is defined by
where and
.
chat.nk
may be used to simulate values under the given model (set nsim > 0). The ... argument may include 'ncores = x' (x>1) to specify parallel processing of simulations - the speed up is large on unix-like machines for which the cluster type of
makeCluster
is "FORK" rather than "PSOCK". If 'ncores' is not provided then the value returned by setNumThreads()
is used.
No adjustment is made by adjustVarD
when is less than the
minimum.
adjustVarD
by default computes Fletcher's ‘chat’ using chat.nk
,
but a value may be provided.
If chat has been computed separately and provided in the argument of that name, adjustVarD
also accepts a single dataframe as the argument ‘object’; the dataframe should have row ‘D’ and columns ‘link’, ‘estimate’, ‘SE.estimate’ as in the output from predict.secr
.
For nk
, a vector of observed counts, one for each detector in traps(capthist)
.
For Enk
, a vector of expected counts, one for each detector in traps
.
For chat.nk
, usually a list comprising –
expected.nk |
expected number at each detector |
nk |
observed number at each detector |
stats |
vector of summary statistics: mean(expected.nk), var(expected.nk), mean(nk), var(nk), nu (=df), X2/nu |
chat |
|
There are two variations –
If ‘verbose = FALSE’ then only the numeric value of is returned (a vector of 2 values if ‘type = "both"’).
If chat.nk
is called with ‘nsim > 0’ then the output is a list comprising –
type |
from input |
nsim |
from input |
sim.chat |
vector of simulated |
chat |
|
p |
probability of observing |
For adjustVarD
, a dataframe with one row for each session, based on predict.secr
or derived.secr
, with extra column ‘c-hat’.
These functions are experimental in secr 4.6, and do not work with polygon-like and single-catch detectors. No allowance is made for modelled variation in detection parameters with respect to occasion, detector or animal; this includes mixture models (e.g., g0~h2).
Versions before 4.5.11 did not correctly compute expected counts for multi-catch detectors.
Furthermore, we doubt that the adjustment actually solves the problem of overdispersion (Efford and Fletcher unpubl.).
Bischof, R., P. Dupont, C. Milleret, J. Chipperfield, and J. A. Royle. 2020. Consequences of ignoring group association in spatial capture–recapture analysis. Wildlife Biology wlb.00649. DOI 10.2981/wlb.00649
Cooch, E. and White, G. (eds) (2022) Program MARK: A Gentle Introduction. 22nd edition. Most recent edition available online at www.phidot.org/software/mark/docs/book/.
Fletcher, D. (2012) Estimating overdispersion when fitting a generalized linear model to sparse data. Biometrika 99, 230–237.
Wedderburn, R. W. M. (1974) Quasi-likelihood functions, generalized linear models, and the Gauss-Newton method. Biometrika 61, 439–47.
secr
,
make.mask
,
Detection functions
,
Fletcher.chat
temptrap <- make.grid() msk <- make.mask(temptrap) ## expected number of individuals per detector (multi-catch) Enk (D = 5, msk, temptrap, detectpar = list(g0 = 0.2, sigma = 25), noccasions = 5) # useful plotting function for simulated chat (nsim>0) plotchat <- function(chat, head = '', breaks = seq(0.5,2.5,0.05)) { hist(chat$sim.chat, xlim = range(breaks), main = head, xlab = 'c-hat', breaks = breaks, cex.main = 1, yaxs = 'i') abline(v = chat$chat, lwd = 1.5, col = 'blue') }
temptrap <- make.grid() msk <- make.mask(temptrap) ## expected number of individuals per detector (multi-catch) Enk (D = 5, msk, temptrap, detectpar = list(g0 = 0.2, sigma = 25), noccasions = 5) # useful plotting function for simulated chat (nsim>0) plotchat <- function(chat, head = '', breaks = seq(0.5,2.5,0.05)) { hist(chat$sim.chat, xlim = range(breaks), main = head, xlab = 'c-hat', breaks = breaks, cex.main = 1, yaxs = 'i') abline(v = chat$chat, lwd = 1.5, col = 'blue') }
Functions to answer the question "what radius is expected to include
proportion p of points from a circular bivariate distribution
corresponding to a given detection function", and the reverse. These
functions may be used to relate the scale parameter(s) of a detection
function (e.g., ) to home-range area (specifically, the area
within an activity contour for the corresponding simple home-range
model) (see Note).
WARNING: the default behaviour of these functions changed in version
2.6.0. Integration is now performed on the cumulative hazard (exposure)
scale for all functions unless hazard = FALSE
. Results will
differ.
circular.r (p = 0.95, detectfn = 0, sigma = 1, detectpar = NULL, hazard = TRUE, upper = Inf, ...) circular.p (r = 1, detectfn = 0, sigma = 1, detectpar = NULL, hazard = TRUE, upper = Inf, ...)
circular.r (p = 0.95, detectfn = 0, sigma = 1, detectpar = NULL, hazard = TRUE, upper = Inf, ...) circular.p (r = 1, detectfn = 0, sigma = 1, detectpar = NULL, hazard = TRUE, upper = Inf, ...)
p |
vector of probability levels for which radius is required |
r |
vector of radii for which probability level is required |
detectfn |
integer code or character string for shape of detection function 0 = halfnormal, 2 = exponential etc. – see detectfn for other codes |
sigma |
spatial scale parameter of detection function |
detectpar |
named list of detection function parameters |
hazard |
logical; if TRUE the transformation |
upper |
numeric upper limit of integration |
... |
other arguments passed to |
circular.r
is the quantile function of the specified circular
bivariate distribution (analogous to qnorm
, for example). The
quantity calculated by circular.r
is sometimes called 'circular
error probable' (see Note).
For detection functions with two parameters (intercept and scale) it is
enough to provide sigma
. Otherwise, detectpar
should be a
named list including parameter values for the requested detection
function (g0 may be omitted, and order does not matter).
Detection functions in secr are expressed in terms of the decline
in probability of detection with distance , and both
circular.r
and circular.p
integrate this function by
default. Rather than integrating itself, it may be more
appropriate to integrate
transformed to a hazard i.e.
. This is selected with
hazard = TRUE
.
Integration may also fail with the message “roundoff error is detected in the extrapolation table”.
Setting upper
to a large number less than infinity sometimes corrects this.
Vector of values for the required radii or probabilities.
The term ‘circular error probable’ has a military origin. It is
commonly used for GPS accuracy with the default probability level set to
0.5 (i.e. half of locations are further than CEP from the true
location). A circular bivariate normal distriubution is commonly assumed
for the circular error probable; this is equivalent to setting
detectfn = "halfnormal"
.
Closed-form expressions are used for the normal and uniform cases; in
the circular bivariate normal case, the relationship is . Otherwise,
the probability is computed numerically by integrating the radial
distribution. Numerical integration is not foolproof, so check
suspicious or extreme values.
When circular.r
is used with the default sigma = 1
, the result
may be interpreted as the factor by which sigma needs to be inflated to
include the desired proportion of activity (e.g., 2.45 sigma for 95%
of points from a circular bivariate normal distribution fitted on the hazard
scale (detectfn = 14) OR 2.24 sigma on the probability scale (detectfn = 0)).
Calhoun, J. B. and Casby, J. U. (1958) Calculation of home range and density of small mammals. Public Health Monograph No. 55. United States Government Printing Office.
Johnson, R. A. and Wichern, D. W. (1982) Applied multivariate statistical analysis. Prentice-Hall, Englewood Cliffs, New Jersey, USA.
## Calhoun and Casby (1958) p 3. ## give p = 0.3940, 0.8645, 0.9888 circular.p(1:3, hazard = FALSE) ## halfnormal, hazard-rate and exponential circular.r () circular.r (detectfn = "HR", detectpar = list(sigma = 1, z = 4)) circular.r (detectfn = "EX") circular.r (detectfn = "HHN") circular.r (detectfn = "HHR", detectpar = list(sigma = 1, z = 4)) circular.r (detectfn = "HEX") plot(seq(0, 5, 0.05), circular.p(r = seq(0, 5, 0.05)), type = "l", xlab = "Radius (multiples of sigma)", ylab = "Probability") lines(seq(0, 5, 0.05), circular.p(r = seq(0, 5, 0.05), detectfn = 2), type = "l", col = "red") lines(seq(0, 5, 0.05), circular.p(r = seq(0, 5, 0.05), detectfn = 1, detectpar = list(sigma = 1,z = 4)), type = "l", col = "blue") abline (h = 0.95, lty = 2) legend (2.8, 0.3, legend = c("halfnormal","hazard-rate, z = 4", "exponential"), col = c("black","blue","red"), lty = rep(1,3)) ## in this example, a more interesting comparison would use ## sigma = 0.58 for the exponential curve.
## Calhoun and Casby (1958) p 3. ## give p = 0.3940, 0.8645, 0.9888 circular.p(1:3, hazard = FALSE) ## halfnormal, hazard-rate and exponential circular.r () circular.r (detectfn = "HR", detectpar = list(sigma = 1, z = 4)) circular.r (detectfn = "EX") circular.r (detectfn = "HHN") circular.r (detectfn = "HHR", detectpar = list(sigma = 1, z = 4)) circular.r (detectfn = "HEX") plot(seq(0, 5, 0.05), circular.p(r = seq(0, 5, 0.05)), type = "l", xlab = "Radius (multiples of sigma)", ylab = "Probability") lines(seq(0, 5, 0.05), circular.p(r = seq(0, 5, 0.05), detectfn = 2), type = "l", col = "red") lines(seq(0, 5, 0.05), circular.p(r = seq(0, 5, 0.05), detectfn = 1, detectpar = list(sigma = 1,z = 4)), type = "l", col = "blue") abline (h = 0.95, lty = 2) legend (2.8, 0.3, legend = c("halfnormal","hazard-rate, z = 4", "exponential"), col = c("black","blue","red"), lty = rep(1,3)) ## in this example, a more interesting comparison would use ## sigma = 0.58 for the exponential curve.
Clone rows of an object a constant or random number of times
## Default S3 method: clone(object, type, ...) ## S3 method for class 'popn' clone(object, type, ...) ## S3 method for class 'capthist' clone(object, type, ...)
## Default S3 method: clone(object, type, ...) ## S3 method for class 'popn' clone(object, type, ...) ## S3 method for class 'capthist' clone(object, type, ...)
object |
any object |
type |
character ‘constant’, ‘poisson’, ‘truncatedpoisson’ or ‘nbinom’ |
... |
other arguments for distribution function |
The ... argument specifies the number of times each row should be
repeated. For random distributions (Poisson or negative binomial) ...
provides the required parameter values: lambda
for Poisson,
size, prob
or size, mu
for negative binomial.
One application is to derive a population of cues from a popn object, where each animal in the original popn generates a number of cues from the same point.
Cloning a capthist object replicates whole detection
histories. Individual covariates and detection-specific attributes
(e.g., signal strength or xy location in polygon) are also
replicated. Cloned data from single-catch traps will cause verify() to
fail, but a model may still be fitted in secr.fit
by overriding
the check with verify = FALSE
.
Object of same class as object
but with varying number of
rows. For clone.popn
and capthist
an attribute ‘freq’ is
set, a vector of length equal to the original number of rows giving the
number of repeats (including zeros).
If popn
or capthist
is a multi-session object the returned value will be
a multi-session object of the same length.
## population of animals at 1 / hectare generates random ## Poisson number of cues, lambda = 5 mics4 <- make.grid( nx = 2, ny = 2, spacing = 44, detector = "signal") pop <- sim.popn (D = 1, core = mics4, buffer = 300, nsessions = 6) pop <- clone (pop, "poisson", 5) attr(pop[[1]],"freq") clone(captdata, "poisson", 3) # To avoid losing any individuals use zero-truncated Poisson # First find lambda of truncated Poisson with given mean getlambda <- function (target) { fn <- function(x) x / (1-exp(-x)) - target uniroot(interval = c(1e-8, target), f = fn)$root } clone(captdata, "truncatedpoisson", getlambda(3))
## population of animals at 1 / hectare generates random ## Poisson number of cues, lambda = 5 mics4 <- make.grid( nx = 2, ny = 2, spacing = 44, detector = "signal") pop <- sim.popn (D = 1, core = mics4, buffer = 300, nsessions = 6) pop <- clone (pop, "poisson", 5) attr(pop[[1]],"freq") clone(captdata, "poisson", 3) # To avoid losing any individuals use zero-truncated Poisson # First find lambda of truncated Poisson with given mean getlambda <- function (target) { fn <- function(x) x / (1-exp(-x)) - target uniroot(interval = c(1e-8, target), f = fn)$root } clone(captdata, "truncatedpoisson", getlambda(3))
Estimate N, the size of a closed population, by several conventional non-spatial capture–recapture methods.
closedN(object, estimator = NULL, level = 0.95, maxN = 1e+07, dmax = 10 )
closedN(object, estimator = NULL, level = 0.95, maxN = 1e+07, dmax = 10 )
object |
|
estimator |
character; name of estimator (see Details) |
level |
confidence level (1 – alpha) |
maxN |
upper bound for population size |
dmax |
numeric, the maximum AIC difference for inclusion in confidence set |
Data are provided as spatial capture histories, but the spatial information (trapping locations) is ignored.
AIC-based model selection is available for the maximum-likelihood
estimators null
, zippin
, darroch
, h2
, and
betabinomial
.
Model weights are calculated as
Models for which dAICc > dmax
are given a weight of zero and are
excluded from the summation, as are non-likelihood models.
Computation of null
, zippin
and darroch
estimates
differs slightly from Otis et al. (1978) in that the likelihood is
maximized over real values of N between Mt1
and maxN
,
whereas Otis et al. considered only integer values.
Asymmetric confidence intervals are obtained in the same way for all
estimators, using a log transformation of
following Burnham et al. (1987), Chao (1987) and Rexstad and Burnham
(1991).
The available estimators are
Name | Model | Description | Reference |
null |
M0 | null | Otis et al. 1978 p.105 |
zippin |
Mb | removal | Otis et al. 1978 p.108 |
darroch |
Mt | Darroch | Otis et al. 1978 p.106-7 |
h2 |
Mh | 2-part finite mixture | Pledger 2000 |
betabinomial |
Mh | Beta-binomial continuous mixture | Dorazio and Royle 2003 |
jackknife |
Mh | jackknife | Burnham and Overton 1978 |
chao |
Mh | Chao's Mh estimator | Chao 1987 |
chaomod |
Mh | Chao's modified Mh estimator | Chao 1987 |
chao.th1 |
Mth | sample coverage estimator 1 | Lee and Chao 1994 |
chao.th2 |
Mth | sample coverage estimator 2 | Lee and Chao 1994 |
A dataframe with one row per estimator and columns
model |
model in the sense of Otis et al. 1978 |
npar |
number of parameters estimated |
loglik |
maximized log likelihood |
AIC |
Akaike's information criterion |
AICc |
AIC with small-sample adjustment of Hurvich & Tsai (1989) |
dAICc |
difference between AICc of this model and the one with smallest AICc |
Mt1 |
number of distinct individuals caught |
Nhat |
estimate of population size |
seNhat |
estimated standard error of Nhat |
lclNhat |
lower 100 x level % confidence limit |
uclNhat |
upper 100 x level % confidence limit |
If your data are from spatial sampling (e.g. grid trapping) it is
recommended that you do not use these methods to estimate
population size (see Efford and Fewster 2013). Instead, fit a spatial model
and estimate population size with region.N
.
Prof. Anne Chao generously allowed me to adapt her code for the variance of the ‘chao.th1’ and ‘chao.th2’ estimators.
Chao's estimators have been subject to various improvements not included here (e.g., Chao et al. 2016).
Burnham, K. P. and Overton, W. S. (1978) Estimating the size of a closed population when capture probabilities vary among animals. Biometrika 65, 625–633.
Chao, A. (1987) Estimating the population size for capture–recapture data with unequal catchability. Biometrics 43, 783–791.
Chao, A., Ma, K. H., Hsieh, T. C. and Chiu, Chun-Huo (2016) SpadeR: Species-Richness Prediction and Diversity Estimation with R. R package version 0.1.1. https://CRAN.R-project.org/package=SpadeR
Dorazio, R. M. and Royle, J. A. (2003) Mixture models for estimating the size of a closed population when capture rates vary among individuals. Biometrics 59, 351–364.
Efford, M. G. and Fewster, R. M. (2013) Estimating population size by spatially explicit capture–recapture. Oikos 122, 918–928.
Hurvich, C. M. and Tsai, C. L. (1989) Regression and time series model selection in small samples. Biometrika 76, 297–307.
Lee, S.-M. and Chao, A. (1994) Estimating population size via sample coverage for closed capture-recapture models. Biometrics 50, 88–97.
Otis, D. L., Burnham, K. P., White, G. C. and Anderson, D. R. (1978) Statistical inference from capture data on closed animal populations. Wildlife Monographs 62, 1–135.
Pledger, S. (2000) Unified maximum likelihood estimates for closed capture-recapture models using mixtures. Biometrics 56, 434–442.
Rexstad, E. and Burnham, K. (1991) User's guide for interactive program CAPTURE. Colorado Cooperative Fish and Wildlife Research Unit, Fort Collins, Colorado, USA.
capthist
,
closure.test
,
region.N
closedN(deermouse.ESG)
closedN(deermouse.ESG)
Perform tests to determine whether a population sampled by capture-recapture is closed to gains and losses over the period of sampling.
closure.test(object, SB = FALSE, min.expected = 2)
closure.test(object, SB = FALSE, min.expected = 2)
object |
|
SB |
logical, if TRUE then test of Stanley and Burnham 1999 is calculated in addition to that of Otis et al. 1978 |
min.expected |
integer for the minimum expected count in any cell of a component 2x2 table |
The test of Stanley and Burnham in part uses a sum over 2x2 contingency tables; any table with a cell whose expected count is less than min.expected is dropped from the sum. The default value of 2 is that used by CloseTest (Stanley and Richards 2005, T. Stanley pers. comm.; see also Stanley and Burnham 1999 p. 203).
In the case of a single-session capthist object, either a vector with the statistic (z-value) and p-value for the test of Otis et al. (1978 p. 120) or a list whose components are data frames with the statistics and p-values for various tests and test components as follows –
Otis |
Test of Otis et al. 1978 |
Xc |
Overall test of Stanley and Burnham 1999 |
NRvsJS |
Stanley and Burnham 1999 |
NMvsJS |
Stanley and Burnham 1999 |
MtvsNR |
Stanley and Burnham 1999 |
MtvsNM |
Stanley and Burnham 1999 |
compNRvsJS |
Occasion-specific components of NRvsJS |
compNMvsJS |
Occasion-specific components of NMvsJS |
Check the original papers for an explanation of the components of the Stanley and Burnham test.
In the case of a multi-session object, a list with one component (as above) for each session.
No omnibus test exists for closure: the existing tests may indicate nonclosure even when a population is closed if other effects such as trap response are present (see White et al. 1982 pp 96–97). The test of Stanley and Burnham is sensitive to individual heterogeneity which is inevitable in most spatial sampling, and it should not in general be used for this sort of data.
Otis, D. L., Burnham, K. P., White, G. C. and Anderson, D. R. (1978) Statistical inference from capture data on closed animal populations. Wildlife Monographs 62, 1–135.
Stanley, T. R. and Burnham, K. P. (1999) A closure test for time-specific capture–recapture data. Environmental and Ecological Statistics 6, 197–209.
Stanley, T. R. and Richards, J. D. (2005) A program for testing capture–recapture data for closure. Wildlife Society Bulletin 33, 782–785.
White, G. C., Anderson, D. R., Burnham, K. P. and Otis, D. L. (1982) Capture-recapture and removal methods for sampling closed populations. Los Alamos National Laboratory, Los Alamos, New Mexico.
closure.test(captdata)
closure.test(captdata)
Clusters are uniform groups of detectors. Use these functions to
extract or replace cluster information of a traps
object, or
extract cluster information for each detection in a capthist
object.
clusterID(object) clusterID(object) <- value clustertrap(object) clustertrap(object) <- value
clusterID(object) clusterID(object) <- value clustertrap(object) clustertrap(object) <- value
object |
|
value |
factor ( |
Easy access to attributes used to define compound designs, those in which a detector array comprises several similar subunits (‘clusters’). ‘clusterID’ identifies the detectors belonging to each cluster, and ‘clustertrap’ is a numeric index used to relate matching detectors in different clusters.
For replacement (‘traps’ only), the number of rows of value
must match exactly the number of detectors in object
.
‘clusterID’ and ‘clustertrap’ are assigned automatically by
trap.builder
.
Factor (clusterID
) or integer-valued vector
(clustertrap
).
clusterID(object)
may be NULL.
traps
, trap.builder
, mash
,
derivedCluster
, cluster.counts
,
cluster.centres
## 25 4-detector clusters mini <- make.grid(nx = 2, ny = 2) tempgrid <- trap.builder (cluster = mini , method = "all", frame = expand.grid(x = seq(100, 500, 100), y = seq(100, 500, 100))) clusterID(tempgrid) clustertrap(tempgrid) tempCH <- sim.capthist(tempgrid) table(clusterID(tempCH)) ## detections per cluster cluster.counts(tempCH) ## distinct individuals
## 25 4-detector clusters mini <- make.grid(nx = 2, ny = 2) tempgrid <- trap.builder (cluster = mini , method = "all", frame = expand.grid(x = seq(100, 500, 100), y = seq(100, 500, 100))) clusterID(tempgrid) clustertrap(tempgrid) tempCH <- sim.capthist(tempgrid) table(clusterID(tempCH)) ## detections per cluster cluster.counts(tempCH) ## distinct individuals
Extract coefficients (estimated beta parameters) from a spatially explicit capture–recapture model.
## S3 method for class 'secr' coef(object, alpha = 0.05, ...)
## S3 method for class 'secr' coef(object, alpha = 0.05, ...)
object |
|
alpha |
alpha level |
... |
other arguments (not used currently) |
A data frame with one row per beta parameter and columns for the coefficient, SE(coefficient), asymptotic lower and upper 100(1–alpha) confidence limits.
## load & extract coefficients of previously fitted null model coef(secrdemo.0)
## load & extract coefficients of previously fitted null model coef(secrdemo.0)
Estimates from one or more openCR models are formed into an array.
## S3 method for class 'secr' collate(object, ..., realnames = NULL, betanames = NULL, newdata = NULL, alpha = 0.05, perm = 1:4, fields = 1:4) ## S3 method for class 'ipsecr' collate(object, ..., realnames = NULL, betanames = NULL, newdata = NULL, alpha = 0.05, perm = 1:4, fields = 1:4) ## S3 method for class 'secrlist' collate(object, ..., realnames = NULL, betanames = NULL, newdata = NULL, alpha = 0.05, perm = 1:4, fields = 1:4)
## S3 method for class 'secr' collate(object, ..., realnames = NULL, betanames = NULL, newdata = NULL, alpha = 0.05, perm = 1:4, fields = 1:4) ## S3 method for class 'ipsecr' collate(object, ..., realnames = NULL, betanames = NULL, newdata = NULL, alpha = 0.05, perm = 1:4, fields = 1:4) ## S3 method for class 'secrlist' collate(object, ..., realnames = NULL, betanames = NULL, newdata = NULL, alpha = 0.05, perm = 1:4, fields = 1:4)
object |
secr or secrlist object |
... |
other secr objects |
realnames |
character vector of real parameter names |
betanames |
character vector of beta parameter names |
newdata |
optional dataframe of values at which to evaluate models |
alpha |
alpha level for confidence intervals |
perm |
permutation of dimensions in output from |
fields |
vector to restrict summary fields in output |
collate
extracts parameter estimates from a set of fitted secr
model objects.
fields
may be used to select a subset of summary
fields ("estimate","SE.estimate","lcl","ucl") by name or number.
A 4-dimensional array of model-specific parameter estimates. By default, the dimensions correspond respectively to
rows in newdata
(usually sessions),
models,
statistic fields (estimate, SE.estimate, lcl, ucl), and
parameters ("phi", "sigma" etc.).
It often helps to reorder the dimensions with the perm
argument.
collate (secrdemo.0, secrdemo.b, perm = c(4,2,3,1))[,,1,]
collate (secrdemo.0, secrdemo.b, perm = c(4,2,3,1))[,,1,]
Compute profile likelihood confidence intervals for ‘beta’ or ‘real’ parameters of a spatially explicit capture-recapture model,
## S3 method for class 'secr' confint(object, parm, level = 0.95, newdata = NULL, tracelevel = 1, tol = 0.0001, bounds = NULL, ncores = NULL, ...)
## S3 method for class 'secr' confint(object, parm, level = 0.95, newdata = NULL, tracelevel = 1, tol = 0.0001, bounds = NULL, ncores = NULL, ...)
object |
|
parm |
numeric or character vector of parameters |
level |
confidence level (1 – alpha) |
newdata |
optional dataframe of values at which to evaluate model |
tracelevel |
integer for level of detail in reporting (0,1,2) |
tol |
absolute tolerance (passed to uniroot) |
bounds |
numeric vector of outer starting values – optional |
ncores |
number of threads used for parallel processing |
... |
other arguments (not used) |
If parm
is numeric its elements are interpreted as the indices of
‘beta’ parameters; character values are interpreted as ‘real’
parameters. Different methods are used for beta parameters and real
parameters. Limits for the -th beta parameter are found by a
numerical search for the value satisfying
, where
is the maximized log
likelihood,
is the maximized profile log
likelihood with
fixed, and
is the
quantile of the
distribution with one degree of freedom. Limits
for real parameters use the method of Lagrange multipliers (Fletcher and
Faddy 2007), except that limits for constant real parameters are
backtransformed from the limits for the relevant beta parameter.
If bounds
is provided it should be a 2-vector or matrix of 2
columns and length(parm) rows.
Setting ncores = NULL
uses the existing value from the environment variable
RCPP_PARALLEL_NUM_THREADS (see setNumThreads
).
A matrix with one row for each parameter in parm
, and columns
giving the lower (lcl) and upper (ucl) 100*level
Calculation may take a long time, so probably you will do it only after selecting a final model.
The R function uniroot
is used to search for the roots of
within a
suitable interval. The interval is anchored at one end by the MLE, and
at the other end by the MLE inflated by a small multiple of the
asymptotic standard error (1, 2, 4 or 8 SE are tried in turn, using the
smallest for which the interval includes a valid solution).
A more efficient algorithm was proposed by Venzon and Moolgavkar (1988);
it has yet to be implemented in secr, but see plkhci
in
the package Bhat for another R implementation.
Evans, M. A., Kim, H.-M. and O'Brien, T. E. (1996) An application of profile-likelihood based confidence interval to capture–recapture estimators. Journal of Agricultural, Biological and Experimental Statistics 1, 131–140.
Fletcher, D. and Faddy, M. (2007) Confidence intervals for expected abundance of rare species. Journal of Agricultural, Biological and Experimental Statistics 12, 315–324.
Venzon, D. J. and Moolgavkar, S. H. (1988) A method for computing profile-likelihood-based confidence intervals. Applied Statistics 37, 87–94.
## Not run: ## Limits for the constant real parameter "D" confint(secrdemo.0, "D") ## End(Not run)
## Not run: ## Limits for the constant real parameter "D" confint(secrdemo.0, "D") ## End(Not run)
Display contours of the net probability of detection p.(X), or the
area within a specified distance of detectors. bufferContour
adds a conventional ‘boundary strip’ to a detector (trap) array, where
buffer
equals the strip width.
pdotContour(traps, border = NULL, nx = 64, detectfn = 0, detectpar = list(g0 = 0.2, sigma = 25, z = 1), noccasions = NULL, binomN = NULL, levels = seq(0.1, 0.9, 0.1), poly = NULL, poly.habitat = TRUE, plt = TRUE, add = FALSE, fill = NULL, ...) bufferContour(traps, buffer, nx = 64, convex = FALSE, ntheta = 100, plt = TRUE, add = FALSE, poly = NULL, poly.habitat = TRUE, fill = NULL, ...)
pdotContour(traps, border = NULL, nx = 64, detectfn = 0, detectpar = list(g0 = 0.2, sigma = 25, z = 1), noccasions = NULL, binomN = NULL, levels = seq(0.1, 0.9, 0.1), poly = NULL, poly.habitat = TRUE, plt = TRUE, add = FALSE, fill = NULL, ...) bufferContour(traps, buffer, nx = 64, convex = FALSE, ntheta = 100, plt = TRUE, add = FALSE, poly = NULL, poly.habitat = TRUE, fill = NULL, ...)
traps |
|
border |
width of blank margin around the outermost detectors |
nx |
dimension of interpolation grid in x-direction |
detectfn |
integer code or character string for shape of detection function 0 = halfnormal etc. – see detectfn |
detectpar |
list of values for named parameters of detection function |
noccasions |
number of sampling occasions |
binomN |
integer code for discrete distribution (see
|
levels |
vector of levels for p.(X) |
poly |
matrix of two columns, the x and y coordinates of a bounding polygon (optional) |
poly.habitat |
logical as in |
plt |
logical to plot contours |
add |
logical to add contour(s) to an existing plot |
fill |
vector of colours to fill contours (optional) |
... |
other arguments to pass to |
buffer |
vector of buffer widths |
convex |
logical, if TRUE the plotted contour(s) will be convex |
ntheta |
integer value for smoothness of convex contours |
pdotContour
constructs a rectangular mask and applies pdot
to
compute the p.(X) at each mask point.
If convex = FALSE
, bufferContour
constructs a mask and
contours the points on the basis of distance to the nearest detector at the
levels given in buffer
.
If convex = TRUE
, bufferContour
constructs a set of
potential vertices by adding points on a circle of radius =
buffer
to each detector location; the desired contour is the
convex hull of these points (this algorithm derives from Efford, 2012).
If traps
has a usage attribute then noccasions
is
set accordingly; otherwise it must be provided.
If traps
is for multiple sessions then detectpar should be a list
of the same length, one component per session, and noccasions may be a
numeric vector of the same length.
Increase nx
for smoother lines, at the expense of speed.
Coordinates of the plotted contours are returned as a list with one
component per polygon. The list is returned invisibly if plt =
TRUE
.
For multi-session input (traps
) the value is a list of such
lists, one per session.
The precision (smoothness) of the fitted line in bufferContour
is controlled by ntheta
rather than nx
when convex
= TRUE
.
To suppress contour labels, include the argument drawlabels =
FALSE
(this will be passed via ... to contour
). Other useful
arguments of contour
are col
(colour of contour lines)
and lwd
(line width).
You may wish to consider function st_buffer in package sf as an
alternative to bufferContour
.
bufferContour
failed with multi-session traps
before
secr 2.8.0.
Efford, M. G. (2012) DENSITY 5.0: software for spatially explicit capture–recapture. Department of Mathematics and Statistics, University of Otago, Dunedin, New Zealand https://www.otago.ac.nz/density/.
possumtraps <- traps(possumCH) ## convex and concave buffers plot(possumtraps, border = 270) bufferContour(possumtraps, buffer = 100, add = TRUE, col = "blue") bufferContour(possumtraps, buffer = 100, convex = TRUE, add = TRUE) ## areas buff.concave <- bufferContour(possumtraps, buffer = 100, plt = FALSE) buff.convex <- bufferContour(possumtraps, buffer = 100, plt = FALSE, convex = TRUE) sum (sapply(buff.concave, polyarea)) ## sum over parts sapply(buff.convex, polyarea) ## effect of nx on area buff.concave2 <- bufferContour(possumtraps, buffer = 100, nx = 128, plt = FALSE) sum (sapply(buff.concave2, polyarea)) ## Not run: plot(possumtraps, border = 270) pdotContour(possumtraps, detectfn = 0, nx = 128, detectpar = detectpar(possum.model.0), levels = c(0.1, 0.01, 0.001), noccasions = 5, add = TRUE) ## clipping to polygon olddir <- setwd(system.file("extdata", package = "secr")) possumtraps <- traps(possumCH) possumarea <- read.table("possumarea.txt", header = TRUE) par(xpd = TRUE, mar = c(1,6,6,6)) plot(possumtraps, border = 400, gridlines = FALSE) pdotContour(possumtraps, detectfn = 0, nx = 256, detectpar = detectpar(possum.model.0), levels = c(0.1, 0.01, 0.001), noccasions = 5, add = TRUE, poly = possumarea, col = "blue") lines(possumarea) setwd(olddir) par(xpd = FALSE, mar = c(5,4,4,2) + 0.1) ## reset to default ## End(Not run)
possumtraps <- traps(possumCH) ## convex and concave buffers plot(possumtraps, border = 270) bufferContour(possumtraps, buffer = 100, add = TRUE, col = "blue") bufferContour(possumtraps, buffer = 100, convex = TRUE, add = TRUE) ## areas buff.concave <- bufferContour(possumtraps, buffer = 100, plt = FALSE) buff.convex <- bufferContour(possumtraps, buffer = 100, plt = FALSE, convex = TRUE) sum (sapply(buff.concave, polyarea)) ## sum over parts sapply(buff.convex, polyarea) ## effect of nx on area buff.concave2 <- bufferContour(possumtraps, buffer = 100, nx = 128, plt = FALSE) sum (sapply(buff.concave2, polyarea)) ## Not run: plot(possumtraps, border = 270) pdotContour(possumtraps, detectfn = 0, nx = 128, detectpar = detectpar(possum.model.0), levels = c(0.1, 0.01, 0.001), noccasions = 5, add = TRUE) ## clipping to polygon olddir <- setwd(system.file("extdata", package = "secr")) possumtraps <- traps(possumCH) possumarea <- read.table("possumarea.txt", header = TRUE) par(xpd = TRUE, mar = c(1,6,6,6)) plot(possumtraps, border = 400, gridlines = FALSE) pdotContour(possumtraps, detectfn = 0, nx = 256, detectpar = detectpar(possum.model.0), levels = c(0.1, 0.01, 0.001), noccasions = 5, add = TRUE, poly = possumarea, col = "blue") lines(possumarea) setwd(olddir) par(xpd = FALSE, mar = c(5,4,4,2) + 0.1) ## reset to default ## End(Not run)
Extract or replace covariates
covariates(object, ...) covariates(object) <- value
covariates(object, ...) covariates(object) <- value
object |
an object of class |
value |
a dataframe of covariates |
... |
other arguments (not used) |
For replacement, the number of rows of value
must match exactly the number of rows in object
.
covariates(object) returns the dataframe of covariates associated with
object
. covariates(object)
may be NULL.
Individual covariates are stored in the ‘covariates’ attribute of a
capthist
object.
Covariates used for modelling density are stored in the ‘covariates’
attribute of a mask
object.
Detector covariates may vary between sampling occasions. In this case,
columns in the detector covariates data.frame are associated with
particular times; the matching is controlled by the
timevaryingcov
attribute.
## detector covariates temptrap <- make.grid(nx = 6, ny = 8) covariates (temptrap) <- data.frame(halfnhalf = factor(rep(c("left","right"),c(24,24))) ) summary(covariates(temptrap))
## detector covariates temptrap <- make.grid(nx = 6, ny = 8) covariates (temptrap) <- data.frame(halfnhalf = factor(rep(c("left","right"),c(24,24))) ) summary(covariates(temptrap))
The coefficient of variation of effective sampling area predicts the bias in estimated density (Efford and Mowat 2014). These functions assist its calculation from fitted finite mixture models.
CV(x, p, na.rm = FALSE) CVa0(object, ...) CVa(object, sessnum = 1, ...)
CV(x, p, na.rm = FALSE) CVa0(object, ...) CVa(object, sessnum = 1, ...)
x |
vector of numeric values |
p |
vector of class probabilities |
na.rm |
logical; if TRUE missing values are dropped from x |
object |
fitted secr finite mixture model |
sessnum |
integer sequence number of session to analyse |
... |
other arguments passed to predict.secr (e.g.,
|
CV
computes the coefficient of variation of x
. If
p
is provided then the distribution is assumed to be
discrete, with support x
and class membership probabilities
p
(scaled automatically to sum to 1.0).
CVa
computes CV() where
is the effective
sampling area of Borchers and Efford (2008).
CVa0
computes CV(a0) where a0 is the single-detector sampling
area defined as (Efford and Mowat 2014); a0 is a convenient
surrogate for a, the effective sampling area. CV(a0) uses
either the fitted MLE of a0 (if the a0 parameterization has been
used), or a0 computed from the estimates of lambda0 and sigma.
CVa
and CVa0
do not work for models with individual
covariates.
Numeric
Do not confuse the function CVa with the estimated relative standard
error of the estimate of a from derived
, also labelled CVa
in the output. The relative standard error RSE is often labelled CV
in the literature on capture–recapture, but this can cause unnecessary
confusion. See also RSE
.
Borchers, D. L. and Efford, M. G. (2008) Spatially explicit maximum likelihood methods for capture–recapture studies. Biometrics 64, 377–385.
Efford, M. G. and Mowat, G. (2014) Compensatory heterogeneity in capture–recapture data. Ecology 95, 1341–1348.
## Not run: ## housemouse model morning <- subset(housemouse, occ = c(1,3,5,7,9)) msk <- make.mask((traps(morning)), nx = 32) morning.h2 <- secr.fit(morning, buffer = 20, model = list(g0~h2), mask = msk, trace = FALSE) CVa0(morning.h2 ) ## End(Not run)
## Not run: ## housemouse model morning <- subset(housemouse, occ = c(1,3,5,7,9)) msk <- make.mask((traps(morning)), nx = 32) morning.h2 <- secr.fit(morning, buffer = 20, model = list(g0~h2), mask = msk, trace = FALSE) CVa0(morning.h2 ) ## End(Not run)
Internal function used by secr.fit
,
confint.secr
, and score.test
.
D.designdata (mask, Dmodel, grouplevels, sessionlevels, sessioncov = NULL, meanSD = NULL)
D.designdata (mask, Dmodel, grouplevels, sessionlevels, sessioncov = NULL, meanSD = NULL)
mask |
|
Dmodel |
formula for density model |
grouplevels |
vector of group names |
sessionlevels |
vector of character values for session names |
sessioncov |
optional dataframe of values of session-specific covariate(s). |
meanSD |
optional external values for scaling x- and y- coordinates |
This is an internal secr function that you are unlikely ever to
use. Unlike secr.design.MS
, this function does not
call model.matrix
.
Dataframe with one row for each combination of mask point, group and
session. Conceptually, we use a 3-D rectangular array with enough rows
to accommodate the largest mask, so some rows in the output may merely
hold space to enable easy indexing. The dataframe has an attribute
‘dimD’ that gives the relevant dimensions: attr(dframe, "dimD") =
c(nmask, ngrp, R)
, where nmask
is the number of mask points,
ngrp
is the number of groups, and R
is the number of
sessions. Columns correspond to predictor variables in Dmodel.
The number of valid rows (points in each session-specific mask) is stored in the attribute ‘validMaskRows’.
For a single-session mask, meanSD
is a 2 x 2 matrix of mean and
SD (rows) for x- and y-coordinates. For a multi-session mask, a list of
such objects. Ordinarily these values are from the meanSD attribute of
the mask, but they must be specified when applying a new mask to an
existing model.
Data of V. H. Reid from live trapping of deermice (Peromyscus maniculatus) at two sites in Colorado, USA.
deermouse.ESG deermouse.WSG
deermouse.ESG deermouse.WSG
Two datasets of V. H. Reid were described by Otis et al. (1978) and distributed with their CAPTURE software (now available from https://eesc.usgs.gov/mbr/software/capture.shtml). They have been used in several other papers on closed population methods (e.g., Huggins 1991, Stanley and Richards 2005). This description is based on pages 32 and 87–93 of Otis et al. (1978).
Both datasets are from studies in Rio Blanco County, Colorado, in the summer of 1975. Trapping was for 6 consecutive nights. Traps were arranged in a 9 x 11 grid and spaced 50 feet (15.2 m) apart.
The first dataset was described by Otis et al. (1978: 32) as from 'a drainage bottom of sagebrush, gambel oak, and serviceberry with pinyon pine and juniper on the uplands'. By matching with the ‘examples’ file of CAPTURE this was from East Stuart Gulch (ESG).
The second dataset (Otis et al. 1978: 87) was from Wet Swizer Creek or Gulch (WSG) in August 1975. No specific vegetation description is given for this site, but it is stated that Sherman traps were used and trapping was done twice daily.
Two minor inconsistencies should be noted. Although Otis et al. (1978) said they used data from morning trap clearances, the capture histories in ‘examples’ from CAPTURE include a ‘pm’ tag on each record. We assume the error was in the text description, as their numerical results can be reproduced from the data file. Huggins (1991) reproduced the East Stuart Gulch dataset (omitting spatial data that were not relevant to his method), but omitted two capture histories.
The data are provided as two single-session capthist
objects
‘deermouse.ESG’ and ‘deermouse.WSG’. Each has a dataframe of individual
covariates, but the fields differ between the two study areas. The
individual covariates of deermouse.ESG are sex (factor levels ‘f’, ‘m’),
age class (factor levels ‘y’, ‘sa’, ‘a’) and body weight in grams. The
individual covariates of deermouse.WSG are sex (factor levels ‘f’,‘m’)
and age class (factor levels ‘j’, ‘y’, ‘sa’, ‘a’) (no data on body
weight). The aging criteria used by Reid are not recorded.
The datasets were originally in the CAPTURE ‘xy complete’ format which for each detection gives the ‘column’ and ‘row’ numbers of the trap (e.g. ‘ 9 5’ for a capture in the trap at position (x=9, y=5) on the grid). Trap identifiers have been recoded as strings with no spaces by inserting zeros (e.g. ‘905’ in this example).
Sherman traps are designed to capture one animal at a time, but the data include double captures (1 at ESG and 8 at WSG – see Examples). The true detector type therefore falls between ‘single’ and ‘multi’. Detector type is set to ‘multi’ in the distributed data objects.
Object | Description |
deermouse.ESG | capthist object, East Stuart Gulch |
deermouse.WSG | capthist object, Wet Swizer Gulch |
File ‘examples’ distributed with program CAPTURE.
Huggins, R. M. (1991) Some practical aspects of a conditional likelihood approach to capture experiments. Biometrics 47, 725–732.
Otis, D. L., Burnham, K. P., White, G. C. and Anderson, D. R. (1978) Statistical inference from capture data on closed animal populations. Wildlife Monographs 62, 1–135.
Stanley, T. R. and Richards, J. D. (2005) A program for testing capture–recapture data for closure. Wildlife Society Bulletin 33, 782–785.
par(mfrow = c(1,2), mar = c(1,1,4,1)) plot(deermouse.ESG, title = "Peromyscus data from East Stuart Gulch", border = 10, gridlines = FALSE, tracks = TRUE) plot(deermouse.WSG, title = "Peromyscus data from Wet Swizer Gulch", border = 10, gridlines = FALSE, tracks = TRUE) closure.test(deermouse.ESG, SB = TRUE) ## reveal multiple captures table(trap(deermouse.ESG), occasion(deermouse.ESG)) table(trap(deermouse.WSG), occasion(deermouse.WSG))
par(mfrow = c(1,2), mar = c(1,1,4,1)) plot(deermouse.ESG, title = "Peromyscus data from East Stuart Gulch", border = 10, gridlines = FALSE, tracks = TRUE) plot(deermouse.WSG, title = "Peromyscus data from Wet Swizer Gulch", border = 10, gridlines = FALSE, tracks = TRUE) closure.test(deermouse.ESG, SB = TRUE) ## reveal multiple captures table(trap(deermouse.ESG), occasion(deermouse.ESG)) table(trap(deermouse.WSG), occasion(deermouse.WSG))
Mask points may be removed by one of three methods: clicking on points, clicking on vertices to define a polygon from which points will be removed, or specifying a polygon to which the mask will be clipped.
deleteMaskPoints(mask, onebyone = TRUE, add = FALSE, poly = NULL, poly.habitat = FALSE, ...)
deleteMaskPoints(mask, onebyone = TRUE, add = FALSE, poly = NULL, poly.habitat = FALSE, ...)
mask |
secr mask object |
onebyone |
logical; see Details |
add |
logical; if true then the initial mask plot will be added to an existing plot |
poly |
polygon defining habitat or non-habitat as described in
|
poly.habitat |
logical; if TRUE polygon represents habitat |
... |
other arguments to plot.mask |
The default method (onebyone = TRUE, poly = NULL) is to click on each point to be removed. The nearest mask point will be selected.
Setting onebyone = FALSE allows the user to click on the vertices of a
polygon within which all points are to be removed (the default) or
retained (poly.habitat = TRUE
). Vertices need not
coincide with mask points.
Defining poly
here is equivalent to calling make.mask
with poly
defined. poly
one of the several types described
in boundarytoSF
. Whether poly
represents habitat or
non-habitat is toggled with poly.habitat
– the default here
differs from make.mask
.
A mask object, usually with fewer points than the input mask.
if (interactive()) { mask0 <- make.mask (traps(captdata)) ## Method 1 - click on each point to remove mask1 <- deleteMaskPoints (mask0) ## Method 2 - click on vertices of removal polygon mask2 <- deleteMaskPoints (mask0, onebyone = FALSE) ## Method 3 - predefined removal polygon plot(captdata) poly1 <- locator(5) mask3 <- deleteMaskPoints (mask0, poly = poly1) }
if (interactive()) { mask0 <- make.mask (traps(captdata)) ## Method 1 - click on each point to remove mask1 <- deleteMaskPoints (mask0) ## Method 2 - click on vertices of removal polygon mask2 <- deleteMaskPoints (mask0, onebyone = FALSE) ## Method 3 - predefined removal polygon plot(captdata) poly1 <- locator(5) mask3 <- deleteMaskPoints (mask0, poly = poly1) }
Compute derived parameters of spatially explicit capture-recapture model. Density is a derived parameter when a model is fitted by maximizing the conditional likelihood. So also is the effective sampling area (in the sense of Borchers and Efford 2008).
derived(object, ...) ## S3 method for class 'secr' derived(object, sessnum = NULL, groups = NULL, alpha = 0.05, se.esa = FALSE, se.D = TRUE, loginterval = TRUE, distribution = NULL, ncores = NULL, bycluster = FALSE, ...) ## S3 method for class 'secrlist' derived(object, sessnum = NULL, groups = NULL, alpha = 0.05, se.esa = FALSE, se.D = TRUE, loginterval = TRUE, distribution = NULL, ncores = NULL, bycluster = FALSE, ...) ## S3 method for class 'secr' esa(object, sessnum = 1, beta = NULL, real = NULL, noccasions = NULL, ncores = NULL, ...)
derived(object, ...) ## S3 method for class 'secr' derived(object, sessnum = NULL, groups = NULL, alpha = 0.05, se.esa = FALSE, se.D = TRUE, loginterval = TRUE, distribution = NULL, ncores = NULL, bycluster = FALSE, ...) ## S3 method for class 'secrlist' derived(object, sessnum = NULL, groups = NULL, alpha = 0.05, se.esa = FALSE, se.D = TRUE, loginterval = TRUE, distribution = NULL, ncores = NULL, bycluster = FALSE, ...) ## S3 method for class 'secr' esa(object, sessnum = 1, beta = NULL, real = NULL, noccasions = NULL, ncores = NULL, ...)
object |
|
sessnum |
index of session in object$capthist for which output required |
groups |
vector of covariate names to define group(s) (see Details) |
alpha |
alpha level for confidence intervals |
se.esa |
logical for whether to calculate SE(mean(esa)) |
se.D |
logical for whether to calculate SE(D-hat) |
loginterval |
logical for whether to base interval on log(D) |
distribution |
character string for distribution of the number of individuals detected |
ncores |
integer number of threads used for parallel processing |
bycluster |
logical; if TRUE results are reported separately for each cluster of detectors |
beta |
vector of fitted parameters on transformed (link) scale |
real |
vector of ‘real’ parameters |
noccasions |
integer number of sampling occasions (see Details) |
... |
other arguments passed to |
The derived estimate of density is a Horvitz-Thompson-like estimate:
where is the estimate of effective sampling area for animal
with detection parameter vector
.
A non-null value of the argument distribution
overrides the value
in object$details
. The sampling variance of
from
secr.fit
by default is spatially unconditional
(distribution = "Poisson"
). For sampling variance conditional on the population of the
habitat mask (and therefore dependent on the mask area), specify
distribution = "binomial"
. The equation for the conditional
variance includes a factor that disappears in the
unconditional (Poisson) variance (Borchers and Efford 2007). Thus the
conditional variance is always less than the unconditional variance. The
unconditional variance may in turn be an overestimate or (more likely)
an underestimate if the true spatial variance is non-Poisson.
Derived parameters may be estimated for population subclasses (groups)
defined by the user with the groups
argument. Each named factor
in groups
should appear in the covariates dataframe of
object$capthist (or each of its components, in the case of a
multi-session dataset).
esa
is used by derived
to compute individual-specific
effective sampling areas:
where
is the probability an individual at X is
detected at least once and the
are optional
individual covariates. Integration is over the area
of the
habitat mask.
The argument noccasions
may be used to vary the number of
sampling occasions; it works only when detection parameters are constant
across individuals and across time.
Setting ncores = NULL
uses the existing value from the environment variable
RCPP_PARALLEL_NUM_THREADS (see setNumThreads
).
The effective sampling area ‘esa’ ()
reported by
derived
is equal to the harmonic mean of the
(arithmetic
mean prior to version 1.5). The sampling variance of
is estimated by
where is the asymptotic estimate of the
variance-covariance matrix of the beta detection parameters
(
) and
is a numerical estimate
of the gradient of
with respect to
, evaluated at
.
A 100(1–alpha)% asymptotic confidence interval is reported for density. By default, this is asymmetric about the estimate because the variance is computed by backtransforming from the log scale. You may also choose a symmetric interval (variance obtained on natural scale).
The vector of detection parameters for esa
may be specified via
beta
or real
, with the former taking precedence. If
neither is provided then the fitted values in object$fit$par
are
used. Specifying real
parameter values bypasses the various
linear predictors. Strictly, the ‘real’ parameters are for a naive
capture (animal not detected previously).
The computation of sampling variances is relatively slow and may be
suppressed with se.esa
and se.D
as desired.
For computing derived
across multiple models in parallel see
par.derived
.
From secr 5.0.0 the ... argument may be used to control the step size
(.relStep) used by fdHess
when estimating gradients for SE(D) and SE(esa).
Dataframe with one row for each derived parameter (‘esa’, ‘D’) and columns as below
estimate | estimate of derived parameter |
SE.estimate | standard error of the estimate |
lcl | lower 100(1--alpha)% confidence limit |
ucl | upper 100(1--alpha)% confidence limit |
CVn | relative SE of number observed (Poisson or binomial assumption) |
CVa | relative SE of effective sampling area |
CVD | relative SE of density estimate |
For a multi-session or multi-group analysis the value is a list with one component for each session and group.
The result will also be a list if object
is an ‘secrlist’.
derived()
may be applied to detection models fitted by maximizing the full likelihood (CL = FALSE
). However, models using g (groups) will not be handled correctly.
Before version 2.1, the output table had columns for ‘varcomp1’ (the variance in due to variation in
, i.e.,
Huggins'
), and ‘varcomp2’ (the variance in
due to uncertainty in estimates of detection parameters).
These quantities are related to CVn and CVa as follows:
Borchers, D. L. and Efford, M. G. (2007) Supplements to Biometrics paper. Available online at https://www.otago.ac.nz/density/.
Borchers, D. L. and Efford, M. G. (2008) Spatially explicit maximum likelihood methods for capture–recapture studies. Biometrics, 64, 377–385.
Huggins, R. M. (1989) On the statistical analysis of capture experiments. Biometrika 76, 133–140.
predict.secr
,
print.secr
,
secr.fit
,
empirical.varD
## Not run: ## extract derived parameters from a model fitted previously ## by maximizing the conditional likelihood derived (secrdemo.CL) ## what happens when sampling variance is conditional on mask N? derived(secrdemo.CL, distribution = "binomial") ## fitted g0, sigma esa(secrdemo.CL) ## force different g0, sigma esa(secrdemo.CL, real = c(0.2, 25)) ## End(Not run)
## Not run: ## extract derived parameters from a model fitted previously ## by maximizing the conditional likelihood derived (secrdemo.CL) ## what happens when sampling variance is conditional on mask N? derived(secrdemo.CL, distribution = "binomial") ## fitted g0, sigma esa(secrdemo.CL) ## force different g0, sigma esa(secrdemo.CL, real = c(0.2, 25)) ## End(Not run)
The function secr.fit
allows many options. Some of these are used
infrequently and have been bundled as a single argument details
to simplify the documentation. They are described here.
details$autoini
specifies the session number from which to compute starting
values (multi-session data only; default 1). From 4.0.0, the character value ‘all’
first forms a single-session capthist using join
(); this may be slow or not
work at all (especially with telemetry data).
details$centred
= TRUE causes coordinates of both traps and mask
to be centred on the centroid of the traps, computed separately for each
session in the case of multi-session data. This may be necessary to
overcome numerical problems when x- or y-coordinates are large
numbers. The default is not to centre coordinates.
details$chat
optionally specifies the overdispersion
of unmarked sightings Tu and unidentified marked sightings Tm. It is used only
for mark-resight models, and is usually computed within secr.fit
(details$nsim > 0
), but may be provided by the user. For a single session 'chat' is a vector of length 2; for multiple sessions it is a 2-column matrix.
details$chatonly
= TRUE used with details$nsim > 0
causes the
overdispersion statistics for sighting counts Tu and Tm to be estimated and
returned as a vector or 2-column matrix (multi-session models), with no further
model fitting.
details$contrasts
may be used to specify the coding of factor predictors. The value should be suitable for the 'contrasts.arg' argument of model.matrix
. See ‘Trend across sessions’ in secr-multisession.pdf for an example.
details$convexpolygon
may be set to FALSE for searches of non-convex polygons. This is slower than the default which requires poygons to be convex east-west (secr-polygondetectors.pdf).
details$debug
is an integer code used to control the printing of intermediate
values (1,2) and to switch on the R code browser (3). In ordinary use it should not be
changed from the default (0).
details$Dfn
is a function for reparameterizing density models; this is set internally when Dlambda = TRUE. Exotic variations may be specified directly by the user when Dlambda = FALSE. The defaults (Dfn = NULL, Dlambda = FALSE) leave the original density model unchanged. Note there is no connection to userDfn (except that the two are incompatible).
Dlambda
if TRUE causes reparameterization of density as the session-on-session finite rate of increase . Details at (secr-trend.pdf).
details$distribution
specifies the distribution of the number of
individuals detected ; this may be conditional on the number in the
masked area ("binomial") or unconditional ("poisson").
distribution
affects the sampling variance of the estimated
density. The default is "poisson". The component ‘distribution’ may also
take a numeric value larger than nrow(capthist), rather than "binomial"
or "poisson". The likelihood then treats n as a binomial draw from a
superpopulation of this size, with consequences for the variance of
density estimates. This can help to reconcile MLE with Bayesian
estimates using data augmentation.
details$externalpdot
names a mask covariate that is substituted for
when fitting a model for relative density (see
details$relativeD
).
This can be useful in a two-phase study when animals are tagged
in phase one and sampled in phase two, with no further tagging (Bottoms et al. in prep.).
The covariate may differ from by a constant factor.
details$fastproximity
controls special handling of data from binary proximity and count detectors. If TRUE and other conditions are met (no temporal variation or groups) then a multi-occasion capthist is automatically reduced to a count for a single occasion and further compressed by storing only non-zero counts, which can greatly speed up computation of the likelihood (default TRUE).
details$fixedbeta
may be used to fix values of beta
parameters. It should be a numeric vector of length equal to the total
number of beta parameters (coefficients) in the model. Parameters to be
estimated are indicated by NA. Other elements should be valid values on
the link scale and will be substituted during likelihood
maximisation. Check the order of beta parameters in a previously fitted
model.
details$grain
sets the grain argument for multithreading in RcppParallel parallelFor (default 1).
details$grain = 0
suppresses multithreading (equivalent to ncores = 1
).
details$hessian
is a character string controlling the computation
of the Hessian matrix from which variances and covariances are obtained.
Options are "none" (no variances), "auto" (the default) or "fdhess" (use
the function fdHess in nlme). If "auto" then the Hessian from the
optimisation function is used. See also method = "none" below.
details$ignoreusage
= TRUE causes the function to ignore
usage (varying effort) information in the traps component. The default
(details$ignoreusage
= FALSE) is to include usage in the model.
details$intwidth2
controls the half-width of the interval
searched by optimise() for the maximum likelihood when there is a single
parameter. Default 0.8 sets the search interval to where
is the ‘start’ value.
details$knownmarks
= FALSE causes secr.fit to fit a zero-truncated
sightings-only model that implicitly estimates the number of marked individuals,
rather than inferring it from the number of rows in the capthist object.
details$LLonly
= TRUE causes the function to returns a single
evaluation of the log likelihood at the ‘start’ values.
details$maxdistance
sets a limit to the centroid-to-mask distances considered. The centroid is the geometric mean of detection locations for each individual. If no limit is specified then summation is over all mask points. Specifying maxdistance
can speed up computation; it is up to the user to select a limit that is large enough not to affect the likelihood (?).
details$miscparm
(default NULL) is an optional numeric vector of
starting values for additional parameters used in a user-supplied
distance function (see ‘userdist’ below). If the vector has a names
attribute then the names will be used for the corresponding coefficients
(‘beta’ parameters) which will otherwise be named ‘miscparm1’,
miscparm2' etc. These parameters are constant across each model and do
not appear in the model formula, but are estimated along with other
coefficients when the likelihood is maximised. Any transformation (link
function) etc. is handled by the user in the userdist function. The
coefficients appear in the output from coef.secr
and
vcov.secr
, but not predict.secr
.
details$newdetector
specifies a detector type to use for this fit,
replacing the previous detector(traps(capthist))
. The value may be
a vector (one value per occasion) or for multi-session data, a list of vectors.
A scalar value (e.g. "proximity") is otherwise used for all occasions and sessions.
The true detector type is usually known and will be specified in the 'traps' attribute;
newdetector
is useful in simulation studies that examine the effect of misspecification. The capthist component of the output from secr.fit has the new type.
details$nsim
specifies the number of replicate simulations to
perform to estimate the overdispersion statistics for the sighting counts
Tu and Tm. See also details$chat
and details$chatonly
.
details$param
chooses between various parameterisations of the
SECR model. The default details$param = 0
is the formulation in
Borchers and Efford (2008) and later papers.
details$param = 1
was once used to select the Gardner & Royle parameterisation of
the detection model (p0, ; Gardner et al. 2009) when
the detector type is ‘multi’. This parameterisation was discontinued in 2.10.0.
details$param = 2
selects parameterisation in terms of
(,
) (Efford and Mowat 2014).
details$param = 3
selects parameterisation in terms of
(,
) (Efford and Mowat 2014). This
parameterization is used automatically if a0 appears in the model (e.g.,
a0 ~ 1).
details$param = 4
selects parameterisation of sigma in terms of
the coefficient sigmak and constant c (sigma = sigmak /
D^0.5 + c) (Efford et al. 2016). If c is not included explicitly in
the model (e.g., c ~ 1) then it is set to zero. This
parameterization is used automatically if sigmak appears in the model (e.g.,
sigmak ~ 1)
details$param = 5
combines parameterisations (3) and (4) (first
compute sigma from D, then compute lambda0 from sigma).
details$relativeD
fits a density model conditional on that describes
relative density instead of absolute density. This describes the distribution of
tagged animals. See also
details$externalpdot
details$savecall
determines whether the full call to secr.fit
is
saved in the output object. The default is TRUE except when called by
list.secr.fit
as names in the call are then evaluated, causing the
output to become unwieldy.
details$splitmarked
determines whether the home range centre of marked
animals is allowed to move between the marking and sighting phases of a spatial
capture–mark–resight study. The default is to assume a common home-range centre
(splitmarked = FALSE
).
details$telemetrytype
determines how telemetry data in the
attribute ‘xylist’ are treated. ‘none’ causes the xylist data to be
ignored. ‘dependent’ uses information on the sampling distribution of
each home-range centre in the SECR likelihood. ‘concurrent’ does that
and more: it splits capthist according to telemetry status and appends
all-zero histories to the telemetry part for any animals present in
xylist. The default is ‘concurrent’.
details$usecov
selects the mask covariate to be used for
normalization. NULL limits denominator for normalization to
distinguishing habitat from non-habitat.
details$userDfn
is a user-provided function for modelling a density
surface. See secr-densitysurfaces.pdf
details$userdist
is either a function to compute non-Euclidean
distances between detectors and mask points, or a pre-computed matrix of
such distances. The first two arguments of the function should be
2-column matrices of x-y coordinates (respectively detectors and
mask points). The third argument is a habitat mask that defines
a non-Euclidean habitat geometry (a linear geometry is described in
documentation for the package ‘secrlinear’). The matrix
returned by the function must have exactly
rows and
columns. When called with no arguments the function should return a
character vector of names for the required covariates of ‘mask’,
possibly including the dynamically computed density 'D' and a parameter
‘noneuc’ that will be fitted. A slightly expanded account is at
userdist, and full documentation is in the separate
document secr-noneuclidean.pdf.
**Do not use ‘userdist’ for polygon or transect detectors**
Efford, M. G., Dawson, D. K., Jhala, Y. V. and Qureshi, Q. (2016) Density-dependent home-range size revealed by spatially explicit capture–recapture. Ecography 39, 676–688.
Efford, M. G. and Mowat, G. (2014) Compensatory heterogeneity in capture–recapture data.Ecology 95, 1341–1348.
Gardner, B., Royle, J. A. and Wegan, M. T. (2009) Hierarchical models for estimating density from DNA mark-recapture studies. Ecology 90, 1106–1115.
Royle, J. A., Chandler, R. B., Sun, C. C. and Fuller, A. K. (2013) Integrating resource selection information with spatial capture–recapture. Methods in Ecology and Evolution 4, 520–530.
## Not run: ## Demo of miscparm and userdist ## We fix the usual 'sigma' parameter and estimate the same ## quantity as miscparm[1]. Differences in CI reflect the implied use ## of the identity link for miscparm[1]. mydistfn3 <- function (xy1,xy2, mask) { if (missing(xy1)) return(character(0)) xy1 <- as.matrix(xy1) xy2 <- as.matrix(xy2) miscparm <- attr(mask, 'miscparm') distmat <- edist(xy1,xy2) / miscparm[1] distmat } fit0 <- secr.fit (captdata) fit <- secr.fit (captdata, fixed = list(sigma=1), details = list(miscparm = c(sig = 20), userdist = mydistfn3)) predict(fit0) coef(fit) ## End(Not run)
## Not run: ## Demo of miscparm and userdist ## We fix the usual 'sigma' parameter and estimate the same ## quantity as miscparm[1]. Differences in CI reflect the implied use ## of the identity link for miscparm[1]. mydistfn3 <- function (xy1,xy2, mask) { if (missing(xy1)) return(character(0)) xy1 <- as.matrix(xy1) xy2 <- as.matrix(xy2) miscparm <- attr(mask, 'miscparm') distmat <- edist(xy1,xy2) / miscparm[1] distmat } fit0 <- secr.fit (captdata) fit <- secr.fit (captdata, fixed = list(sigma=1), details = list(miscparm = c(sig = 20), userdist = mydistfn3)) predict(fit0) coef(fit) ## End(Not run)
A detection function relates the probability of detection or the
expected number of detections
for an animal to the
distance of a detector from a point usually thought of as its home-range
centre. In secr only simple 2- or 3-parameter functions are
used. Each type of function is identified by its number or by a 2–3
letter code (version
2.6.0; see below). In most cases the name
may also be used (as a quoted string).
Choice of detection function is usually not critical, and the default ‘HN’ is usually adequate.
Functions (14)–(20) are parameterised in terms of the expected number
of detections , or cumulative hazard, rather than
probability. ‘Exposure’ (e.g. Royle and Gardner 2011) is another term
for cumulative hazard. This parameterisation is natural for the ‘count’
detector type or if the function is to be interpreted as a
distribution of activity (home range). When one of the functions
(14)–(19) is used to describe detection probability (i.e., for the binary
detectors ‘single’, ‘multi’,‘proximity’,‘polygonX’ or
‘transectX’), the expected number of detections is internally
transformed to a binomial probability using
.
The hazard halfnormal (14) is similar to the halfnormal exposure function
used by Royle and Gardner (2011) except they omit the factor of 2 on
, which leads to estimates of
that are larger
by a factor of sqrt(2). The hazard exponential (16) is identical to their
exponential function.
Code | Name | Parameters | Function |
0 HN | halfnormal | g0, sigma | |
1 HR | hazard rate | g0, sigma, z | |
2 EX | exponential | g0, sigma | |
3 CHN | compound halfnormal | g0, sigma, z | |
4 UN | uniform | g0, sigma | |
5 WEX | w exponential | g0, sigma, w |
|
6 ANN | annular normal | g0, sigma, w | |
7 CLN | cumulative lognormal | g0, sigma, z |
|
8 CG | cumulative gamma | g0, sigma, z |
|
9 BSS | binary signal strength | b0, b1 | |
10 SS | signal strength | beta0, beta1, sdS | |
11 SSS | signal strength spherical | beta0, beta1, sdS |
|
14 HHN | hazard halfnormal | lambda0, sigma | ;
|
15 HHR | hazard hazard rate | lambda0, sigma, z | ; |
16 HEX | hazard exponential | lambda0, sigma | ; |
17 HAN | hazard annular normal | lambda0, sigma, w | ; |
18 HCG | hazard cumulative gamma | lambda0, sigma, z | ;
|
19 HVP | hazard variable power | lambda0, sigma, z | ; |
Functions (1) and (15), the "hazard-rate" detection functions described by Hayes and Buckland (1983), are not recommended for SECR because of their long tail, and care is also needed with (2) and (16).
Function (3), the compound halfnormal, follows Efford and Dawson (2009).
Function (4) uniform is defined only for simulation as it poses problems for likelihood maximisation by gradient methods. Uniform probability implies uniform hazard, so there is no separate function ‘HUN’.
For function (7), ‘F’ is the standard normal distribution function and
and
are the mean and standard deviation on the
log scale of a latent variable representing a threshold of detection
distance. See Note for the relationship to the fitted parameters sigma
and z.
For functions (8) and (18), ‘G’ is the cumulative distribution function of the
gamma distribution with shape parameter k ( = z
) and scale
parameter ( =
sigma/z
). See R's
pgamma
.
For functions (9), (10) and (11), ‘F’ is the standard normal
distribution function and is an arbitrary signal threshold. The two
parameters of (9) are functions of the parameters of (10) and (11):
and
(see Efford et al. 2009). Note that (9) does
not require signal-strength data or
.
Function (11) includes an additional ‘hard-wired’ term for sound
attenuation due to spherical spreading. Detection probability at
distances less than 1 m is given by
Functions (12) and (13) are undocumented methods for sound attenuation.
Function (19) has been used in some published papers and is included for comparison (see e.g. Ergon and Gardner 2014).
Function (20) assigns positive probability of detection only to points within a square pixel (cell) with side 2 sigma that is centred on the detector. (Typically used with fixed sigma = detector spacing / 2).
The parameters of function (7) are potentially confusing. The fitted
parameters describe a latent threshold variable on the natural scale:
sigma (mean) = and z
(standard deviation) =
. As with other
detection functions, sigma is a spatial scale parameter, although in
this case it corresponds to the mean of the threshold variable; the
standard deviation of the threshold variable (z) determines the shape
(roughly 1/max(slope)) of the detection function.
Efford, M. G. and Dawson, D. K. (2009) Effect of distance-related heterogeneity on population size estimates from point counts. Auk 126, 100–111.
Efford, M. G., Dawson, D. K. and Borchers, D. L. (2009) Population density estimated from locations of individuals on a passive detector array. Ecology 90, 2676–2682.
Ergon, T. and Gardner, B. (2014) Separating mortality and emigration: modelling space use, dispersal and survival with robust-design spatial capture–recapture data. Methods in Ecology and Evolution 5, 1327–1336.
Hayes, R. J. and Buckland, S. T. (1983) Radial-distance models for the line-transect method. Biometrics 39, 29–42.
Royle, J. A. and Gardner, B. (2011) Hierarchical spatial capture–recapture models for estimating density from trapping arrays. In: A.F. O'Connell, J.D. Nichols & K.U. Karanth (eds) Camera Traps in Animal Ecology: Methods and Analyses. Springer, Tokyo. Pp. 163–190.
Extract or replace the detector type.
detector(object, ...) detector(object) <- value
detector(object, ...) detector(object) <- value
object |
object with ‘detector’ attribute e.g. |
value |
character string for detector type |
... |
other arguments (not used) |
Valid detector types are ‘single’, ‘multi’, ‘proximity’, ‘count’, ‘capped’,
‘signal’, ‘polygon’, ‘transect’, ‘polygonX’, and ‘transectX’. The
detector type is stored as an attribute of a traps
object.
Detector types are mostly described by Efford et al. (2009a,b; see also
secr-overview.pdf). Polygon and transect detector types are
for area and linear searches as described in
secr-polygondetectors.pdf and Efford (2011). The ‘signal’
detector type is used for acoustic data as described in
secr-sound.pdf.
The ‘capped’ detector type refers to binary proximity data in which no more than one individual may be detected at a detector on any occasion. The type is partially implemented in secr 3.1.1: data may be simulated and manipulated, but for model fitting these are treated as proximity data by secr.fit()
.
character string for detector type
Efford, M. G. (2011) Estimation of population density by spatially explicit capture–recapture with area searches. Ecology 92, 2202–2207.
Efford, M. G., Borchers D. L. and Byrom, A. E. (2009a) Density estimation by spatially explicit capture-recapture: likelihood-based methods. In: D. L. Thomson, E. G. Cooch and M. J. Conroy (eds) Modeling Demographic Processes in Marked Populations. Springer, New York. Pp. 255–269.
Efford, M. G., Dawson, D. K. and Borchers, D. L. (2009b) Population density estimated from locations of individuals on a passive detector array. Ecology 90, 2676–2682.
## Default detector type is "multi" temptrap <- make.grid(nx = 6, ny = 8) detector(temptrap) <- "proximity" summary(temptrap)
## Default detector type is "multi" temptrap <- make.grid(nx = 6, ny = 8) detector(temptrap) <- "proximity" summary(temptrap)
Compute the deviance or residual degrees of freedom of a fitted secr model, treating multiple sessions and groups as independent. The likelihood of the saturated model depends on whether the ‘conditional’ or ‘full’ form was used, and on the distribution chosen for the number of individuals observed (Poisson or binomial).
## S3 method for class 'secr' deviance(object, ...) ## S3 method for class 'secr' df.residual(object, ...)
## S3 method for class 'secr' deviance(object, ...) ## S3 method for class 'secr' df.residual(object, ...)
object |
secr object from secr.fit |
... |
other arguments (not used) |
The deviance is , where
is the value of the
log-likelihood evaluated at its maximum, and
is the
log-likelihood of the saturated model, calculated thus:
Likelihood conditional on -
Full likelihood, Poisson -
Full likelihood, binomial -
is the number of individuals observed at least once,
is the number of distinct histories, and
is the number in a chosen area
that we estimate by
.
The residual degrees of freedom is the number of distinct detection histories minus the number of parameters estimated. The detection histories of two animals are always considered distinct if they belong to different groups.
When samples are (very) large the deviance is expected to be distributed
as with
degrees of
freedom when
parameters are estimated. In reality, simulation is
needed to assess whether a given value of the deviance indicates a
satisfactory fit, or to estimate the overdispersion parameter
.
sim.secr
is a convenient tool.
The scalar numeric value of the deviance or the residual degress of freedom extracted from the fitted model.
Borchers, D. L. and Efford, M. G. (2008) Spatially explicit maximum likelihood methods for capture–recapture studies. Biometrics 64, 377–385.
deviance(secrdemo.0) df.residual(secrdemo.0)
deviance(secrdemo.0) df.residual(secrdemo.0)
It is sometimes useful to re-cast area-search (polygon or polygonX) data as if it was from a set of closely spaced point detectors, i.e. to rasterize the detection locations. This function makes that conversion. Each polygon detector in the input is replaced by a number of point detectors, each representing a square pixel. Detections are mapped to the new detectors on the basis of their x-y coordinates.
If object
contains transect data the problem is passed to snip
and reduce.capthist
.
discretize(object, spacing = 5, outputdetector = c("proximity", "count", "multi"), tol = 0.001, cell.overlap = FALSE, type = c("centre","any", "all"), ...)
discretize(object, spacing = 5, outputdetector = c("proximity", "count", "multi"), tol = 0.001, cell.overlap = FALSE, type = c("centre","any", "all"), ...)
object |
secr capthist or traps object |
spacing |
numeric spacing between point detectors in metres |
outputdetector |
character output detector type |
tol |
numeric fractional inflation of perimeter (see Details) |
cell.overlap |
logical; if TRUE the area of overlap is stored in usage attribute |
type |
character; see Details |
... |
other arguments passed to |
The input should have detector type ‘polygon’ or ‘polygonX’.
A new array of equally spaced detectors is generated within
each polygon of the input, inflated radially by 1 + tol to avoid some
inclusion problems. The origin of the superimposed grid is fixed automatically. If type = "centre"
detectors are included if they lie within the (inflated) polygon. Otherwise, the decision on whether to include a candidate new detector is based on the corner vertices of the cell around the detector (side = spacing
); type = "any"
and type = "all"
have the obvious meanings.
tol
may be negative, in which case the array(s) will be shrunk relative
to the polygon(s).
For irregular polygons the edge cells in the output may be only partially contained within the polygon they represent. Set cell.overlap = TRUE
to retain the proportion of overlap as the ‘usage’ of the new traps object. This can take a few seconds to compute. If ‘usage’ is already defined then the new ‘usage’ is the old multiplied by the proportion of overlap.
Combining type = "any"
and cell.overlap = TRUE
with tol
> 0 can have the odd effect of including some marginal detectors that are assigned zero usage.
With type = "any"
, the sum of the overlap proportions times cell area is equal to the area of the polygons.
A capthist or traps object of the requested detector type, but otherwise carrying forward all attributes of the input. The embedded traps object has a factor covariate ‘polyID’ recording the polygon to which each point detector relates.
Consider the likely number of detectors in the output before you start.
## Not run: ## generate some polygon data pol <- make.poly() CH <- sim.capthist(pol, popn = list(D = 30), detectfn = 'HHN', detectpar = list(lambda0 = 0.3)) plot(CH, border = 10, gridl = FALSE, varycol = FALSE) ## discretize and plot CH1 <- discretize(CH, spacing = 10, output = 'count') plot(CH1, add = TRUE, cappar = list(col = 'orange'), varycol = FALSE, rad = 0) plot(traps(CH1), add = TRUE) # overlay cell boundaries plot(as.mask(traps(CH1)), dots = FALSE, col = NA, meshcol = 'green', add = TRUE) ## show how detections are snapped to new detectors newxy <- traps(CH1)[nearesttrap(xy(CH),traps(CH1)),] segments(xy(CH)[,1], xy(CH)[,2], newxy[,1], newxy[,2]) plot(traps(CH), add = TRUE) # original polygon ## Incomplete overlap pol <- rotate(make.poly(), 45) CH2 <- sim.capthist(pol, popn = list(D = 30), detectfn = 'HHN', detectpar = list(lambda0 = 0.3)) plot(CH2, border = 10, gridl = FALSE, varycol = FALSE) CH3 <- discretize(CH2, spacing = 10, output = 'count', type = 'any', cell.overlap = TRUE, tol=0.05) plot(CH3, add = TRUE, cappar = list(col = 'orange'), varycol = FALSE, rad = 0) plot(traps(CH3), add = TRUE) # overlay cell boundaries and usage msk <- as.mask(traps(CH3)) covariates(msk) <- data.frame(usage = usage(traps(CH3))[,1]) plot(msk, dots = FALSE, cov='usage', meshcol = 'green', add = TRUE) ## End(Not run)
## Not run: ## generate some polygon data pol <- make.poly() CH <- sim.capthist(pol, popn = list(D = 30), detectfn = 'HHN', detectpar = list(lambda0 = 0.3)) plot(CH, border = 10, gridl = FALSE, varycol = FALSE) ## discretize and plot CH1 <- discretize(CH, spacing = 10, output = 'count') plot(CH1, add = TRUE, cappar = list(col = 'orange'), varycol = FALSE, rad = 0) plot(traps(CH1), add = TRUE) # overlay cell boundaries plot(as.mask(traps(CH1)), dots = FALSE, col = NA, meshcol = 'green', add = TRUE) ## show how detections are snapped to new detectors newxy <- traps(CH1)[nearesttrap(xy(CH),traps(CH1)),] segments(xy(CH)[,1], xy(CH)[,2], newxy[,1], newxy[,2]) plot(traps(CH), add = TRUE) # original polygon ## Incomplete overlap pol <- rotate(make.poly(), 45) CH2 <- sim.capthist(pol, popn = list(D = 30), detectfn = 'HHN', detectpar = list(lambda0 = 0.3)) plot(CH2, border = 10, gridl = FALSE, varycol = FALSE) CH3 <- discretize(CH2, spacing = 10, output = 'count', type = 'any', cell.overlap = TRUE, tol=0.05) plot(CH3, add = TRUE, cappar = list(col = 'orange'), varycol = FALSE, rad = 0) plot(traps(CH3), add = TRUE) # overlay cell boundaries and usage msk <- as.mask(traps(CH3)) covariates(msk) <- data.frame(usage = usage(traps(CH3))[,1]) plot(msk, dots = FALSE, cov='usage', meshcol = 'green', add = TRUE) ## End(Not run)
Compute Euclidean distance from each of a set of points to the nearest detector in an array, or return the sequence number of the detector nearest each point.
distancetotrap(X, traps) nearesttrap(X, traps)
distancetotrap(X, traps) nearesttrap(X, traps)
X |
coordinates |
traps |
traps object or 2-column matrix of coordinates |
distancetotrap
returns the distance from each point in X to the
nearest detector in traps
. It may be used to restrict the points
on a habitat mask.
For traps objects with polygon detector type (polygon, polygonX), and for SpatialPolygons, the function sf::st_distance is used internally(from secr 4.5.2).
distancetotrap
returns a vector of distances (assumed to be in metres).
nearesttrap
returns the index of the nearest trap.
## restrict a habitat mask to points within 70 m of traps ## this is nearly equivalent to using make.mask with the ## `trapbuffer' option temptrap <- make.grid() tempmask <- make.mask(temptrap) d <- distancetotrap(tempmask, temptrap) tempmask <- subset(tempmask, d < 70)
## restrict a habitat mask to points within 70 m of traps ## this is nearly equivalent to using make.mask with the ## `trapbuffer' option temptrap <- make.grid() tempmask <- make.mask(temptrap) d <- distancetotrap(tempmask, temptrap) tempmask <- subset(tempmask, d < 70)
S3 class for rasterized fitted density surfaces. A Dsurface is a type of ‘mask’ with covariate(s) for the predicted density at each point.
## S3 method for class 'Dsurface' print(x, scale = 1, ...) ## S3 method for class 'Dsurface' summary(object, scale = 1, ...)
## S3 method for class 'Dsurface' print(x, scale = 1, ...) ## S3 method for class 'Dsurface' summary(object, scale = 1, ...)
x , object
|
Dsurface object to display |
scale |
numeric multiplier for density |
... |
other arguments passed to print method for data frames or summary method for masks |
A Dsurface will usually have been constructed with predictDsurface
.
The ‘scale’ argument may be used to change the units of density from the default (animals / hectare) to animals / km^2 (scale = 100) or animals / 100km^2 (scale = 10000).
A virtual S4 class ‘Dsurface’ is defined to allow the definition of a method for the generic function raster
from the raster package.
predictDsurface
, plot.Dsurface
Plot joint confidence ellipse for two parameters of secr model, or for a distribution of points.
ellipse.secr(object, par = c("g0", "sigma"), alpha = 0.05, npts = 100, plot = TRUE, linkscale = TRUE, add = FALSE, col = palette(), ...) ellipse.bvn(xy, alpha = 0.05, npts = 100, centroid = TRUE, add = FALSE, ...)
ellipse.secr(object, par = c("g0", "sigma"), alpha = 0.05, npts = 100, plot = TRUE, linkscale = TRUE, add = FALSE, col = palette(), ...) ellipse.bvn(xy, alpha = 0.05, npts = 100, centroid = TRUE, add = FALSE, ...)
object |
|
par |
character vector of length two, the names of two ‘beta’ parameters |
alpha |
alpha level for confidence intervals |
npts |
number of points on perimeter of ellipse |
plot |
logical for whether ellipse should be plotted |
linkscale |
logical; if FALSE then coordinates will be backtransformed from the link scale |
add |
logical to add ellipse to an existing plot |
col |
vector of one or more plotting colours |
... |
arguments to pass to plot functions (or polygon() in the case of ellipse.bvn) |
xy |
2-column matrix of coordinates |
centroid |
logical; if TRUE the plotted ellipse is a confidence
region for the centroid of points in |
ellipse.secr
calculates coordinates of a confidence ellipse from
the asymptotic variance-covariance matrix of the beta parameters
(coefficients), and optionally plots it.
If linkscale
== FALSE, the inverse of the appropriate link
transformation is applied to the coordinates of the ellipse, causing it
to deform.
If object
is a list of secr models then one ellipse is
constructed for each model. Colours are recycled as needed.
ellipse.bvn
plots a bivariate normal confidence ellipse for the
centroid of a 2-dimensional distribution of points (default centroid =
TRUE), or a Jennrich and Turner (1969) elliptical home-range model.
A list containing the x and y coordinates is returned invisibly from either function.
Jennrich, R. I. and Turner, F. B. (1969) Measurement of non-circular home range. Journal of Theoretical Biology, 22, 227–237.
ellipse.secr(secrdemo.0)
ellipse.secr(secrdemo.0)
Compute Horvitz-Thompson-like estimate of population density from a
previously fitted spatial detection model, and estimate its sampling
variance using the empirical spatial variance of the number observed
in replicate sampling units. Wrapper functions are provided for
several different scenarios, but all ultimately call
derivednj
. The function derived
also computes
Horvitz-Thompson-like estimates, but it assumes a Poisson or binomial
distribution of total number when computing the sampling variance.
derivednj ( nj, esa, se.esa = NULL, method = c("SRS", "R2", "R3", "local", "poisson", "binomial"), xy = NULL, alpha = 0.05, loginterval = TRUE, area = NULL, independent.esa = FALSE ) derivedMash ( object, sessnum = NULL, method = c("SRS", "local"), alpha = 0.05, loginterval = TRUE) derivedCluster ( object, method = c("SRS", "R2", "R3", "local", "poisson", "binomial"), alpha = 0.05, loginterval = TRUE) derivedSession ( object, method = c("SRS", "R2", "R3", "local", "poisson", "binomial"), xy = NULL, alpha = 0.05, loginterval = TRUE, area = NULL, independent.esa = FALSE ) derivedExternal ( object, sessnum = NULL, nj, cluster, buffer = 100, mask = NULL, noccasions = NULL, method = c("SRS", "local"), xy = NULL, alpha = 0.05, loginterval = TRUE)
derivednj ( nj, esa, se.esa = NULL, method = c("SRS", "R2", "R3", "local", "poisson", "binomial"), xy = NULL, alpha = 0.05, loginterval = TRUE, area = NULL, independent.esa = FALSE ) derivedMash ( object, sessnum = NULL, method = c("SRS", "local"), alpha = 0.05, loginterval = TRUE) derivedCluster ( object, method = c("SRS", "R2", "R3", "local", "poisson", "binomial"), alpha = 0.05, loginterval = TRUE) derivedSession ( object, method = c("SRS", "R2", "R3", "local", "poisson", "binomial"), xy = NULL, alpha = 0.05, loginterval = TRUE, area = NULL, independent.esa = FALSE ) derivedExternal ( object, sessnum = NULL, nj, cluster, buffer = 100, mask = NULL, noccasions = NULL, method = c("SRS", "local"), xy = NULL, alpha = 0.05, loginterval = TRUE)
object |
fitted secr model |
nj |
vector of number observed in each sampling unit (cluster) |
esa |
estimate of effective sampling area ( |
se.esa |
estimated standard error of effective sampling
area ( |
method |
character string ‘SRS’ or ‘local’ |
xy |
dataframe of x- and y- coordinates ( |
alpha |
alpha level for confidence intervals |
loginterval |
logical for whether to base interval on log(N) |
area |
area of region for method = "binomial" (hectares) |
independent.esa |
logical; controls variance contribution from esa (see Details) |
sessnum |
index of session in object$capthist for which output required |
cluster |
‘traps’ object for a single cluster |
buffer |
width of buffer in metres (ignored if |
mask |
mask object for a single cluster of detectors |
noccasions |
number of occasions (for |
derivednj
accepts a vector of counts (nj
), along with
and
. The
argument
esa
may be a scalar or (if se.esa is NULL)
a 2-column matrix with and
for each replicate
(row).
In the special case that
nj
is of length 1, or method
takes the values ‘poisson’ or
‘binomial’, the variance is computed using a theoretical variance
rather than an empirical estimate. The value of method
corresponds to ‘distribution’ in derived
, and defaults to
‘poisson’. For method = 'binomial'
you must specify area
(see Examples).
If independent.esa
is TRUE then independence is assumed among
cluster-specific estimates of esa, and esa variances are summed. The default
is a weighted sum leading to higher overall variance.
derivedCluster
accepts a model fitted to data from clustered
detectors; each cluster is interpreted as a replicate
sample. It is assumed that the sets of individuals sampled by
different clusters do not intersect, and that all clusters have the
same geometry (spacing, detector number etc.).
derivedMash
accepts a model fitted to clustered data that have
been ‘mashed’ for fast processing (see mash
); each
cluster is a replicate sample: the function uses the vector of cluster
frequencies () stored as an attribute of the mashed
capthist
by mash
.
derivedExternal
combines detection parameter estimates from a
fitted model with a vector of frequencies nj
from replicate
sampling units configured as in cluster
. Detectors in
cluster
are assumed to match those in the fitted model with
respect to type and efficiency, but sampling duration
(noccasions
), spacing etc. may differ. The mask
should
match cluster
; if mask
is missing, one will be
constructed using the buffer
argument and defaults from
make.mask
.
derivedSession
accepts a single fitted model that must span
multiple sessions; each session is interpreted as a replicate sample.
Spatial variance is calculated by one of these methods
Method | Description |
"SRS" |
simple random sampling with identical clusters |
"R2" |
variable cluster size cf Thompson (2002:70) estimator for line transects |
"R3" |
variable cluster size cf Buckland et al. (2001) |
"local" |
neighbourhood variance estimator (Stevens and Olsen 2003) SUSPENDED in 4.4.7 |
"poisson" |
theoretical (model-based) variance |
"binomial" |
theoretical (model-based) variance in given area
|
The weighted options R2 and R3 substitute for line length
in the corresponding formulae of Fewster et al. (2009, Eq 3,5). Density is estimated by
where
. The variance of
is estimated as the sum of the cluster-specific variances, assuming independence among clusters. Fewster et al. (2009) found that an alternative estimator for line transects derived by Thompson (2002) performed better when there were strong density gradients correlated with line length (R2 in Fewster et al. 2009, Eq 3).
The neighborhood variance estimator is implemented in package spsurvey and was originally proposed for generalized random tessellation stratified (GRTS) samples. For ‘local’ variance
estimates, the centre of each replicate must be provided in xy
,
except where centres may be inferred from the data. It is unclear whether ‘local’ can or should be used when clusters vary in size.
derivedSystematic
, now defunct, was an experimental function in earlier versions of secr.
Dataframe with one row for each derived parameter (‘esa’, ‘D’) and columns as below
estimate | estimate of derived parameter |
SE.estimate | standard error of the estimate |
lcl | lower 100(1--alpha)% confidence limit |
ucl | upper 100(1--alpha)% confidence limit |
CVn | relative SE of number observed (across sampling units) |
CVa | relative SE of effective sampling area |
CVD | relative SE of density estimate |
The variance of a Horvitz-Thompson-like estimate of density may be
estimated as the sum of two components, one due to uncertainty in the
estimate of effective sampling area () and the
other due to spatial variance in the total number of animals
observed on
replicate sampling units (
). We use a delta-method approximation
that assumes independence of the components:
where . The
estimate of
is model-based while
that of
is design-based. This formulation follows
that of Buckland et al. (2001, p. 78) for conventional distance
sampling. Given sufficient independent replicates, it is a robust way
to allow for unmodelled spatial overdispersion.
There is a complication in SECR owing to the fact that
is a derived quantity (actually an integral)
rather than a model parameter. Its sampling variance
is estimated indirectly in
secr by combining the asymptotic estimate of the covariance
matrix of the fitted detection parameters
with a
numerical estimate of the gradient of
with
respect to
. This calculation is performed in
derived
.
Buckland, S. T., Anderson, D. R., Burnham, K. P., Laake, J. L., Borchers, D. L. and Thomas, L. (2001) Introduction to Distance Sampling: Estimating Abundance of Biological Populations. Oxford University Press, Oxford.
Fewster, R. M. (2011) Variance estimation for systematic designs in spatial surveys. Biometrics 67, 1518–1531.
Fewster, R. M., Buckland, S. T., Burnham, K. P., Borchers, D. L., Jupp, P. E., Laake, J. L. and Thomas, L. (2009) Estimating the encounter rate variance in distance sampling. Biometrics 65, 225–236.
Stevens, D. L. Jr and Olsen, A. R. (2003) Variance estimation for spatially balanced samples of environmental resources. Environmetrics 14, 593–610.
Thompson, S. K. (2002) Sampling. 2nd edition. Wiley, New York.
## The `ovensong' data are pooled from 75 replicate positions of a ## 4-microphone array. The array positions are coded as the first 4 ## digits of each sound identifier. The sound data are initially in the ## object `signalCH'. We first impose a 52.5 dB signal threshold as in ## Dawson & Efford (2009, J. Appl. Ecol. 46:1201--1209). The vector nj ## includes 33 positions at which no ovenbird was heard. The first and ## second columns of `temp' hold the estimated effective sampling area ## and its standard error. ## Not run: signalCH.525 <- subset(signalCH, cutval = 52.5) nonzero.counts <- table(substring(rownames(signalCH.525),1,4)) nj <- c(nonzero.counts, rep(0, 75 - length(nonzero.counts))) temp <- derived(ovensong.model.1, se.esa = TRUE) derivednj(nj, temp["esa",1:2]) ## The result is very close to that reported by Dawson & Efford ## from a 2-D Poisson model fitted by maximizing the full likelihood. ## If nj vector has length 1, a theoretical variance is used... msk <- ovensong.model.1$mask A <- nrow(msk) * attr(msk, "area") derivednj (sum(nj), temp["esa",1:2], method = "poisson") derivednj (sum(nj), temp["esa",1:2], method = "binomial", area = A) ## Set up an array of small (4 x 4) grids, ## simulate a Poisson-distributed population, ## sample from it, plot, and fit a model. ## mash() condenses clusters to a single cluster testregion <- data.frame(x = c(0,2000,2000,0), y = c(0,0,2000,2000)) t4 <- make.grid(nx = 4, ny = 4, spacing = 40) t4.16 <- make.systematic (n = 16, cluster = t4, region = testregion) popn1 <- sim.popn (D = 5, core = testregion, buffer = 0) capt1 <- sim.capthist(t4.16, popn = popn1) fit1 <- secr.fit(mash(capt1), CL = TRUE, trace = FALSE) ## Visualize sampling tempmask <- make.mask(t4.16, spacing = 10, type = "clusterbuffer") plot(tempmask) plot(t4.16, add = TRUE) plot(capt1, add = TRUE) ## Compare model-based and empirical variances. ## Here the answers are similar because the data ## were simulated from a Poisson distribution, ## as assumed by \code{derived} derived(fit1) derivedMash(fit1) ## Now simulate a patchy distribution; note the ## larger (and more credible) SE from derivedMash(). popn2 <- sim.popn (D = 5, core = testregion, buffer = 0, model2D = "hills", details = list(hills = c(-2,3))) capt2 <- sim.capthist(t4.16, popn = popn2) fit2 <- secr.fit(mash(capt2), CL = TRUE, trace = FALSE) derived(fit2) derivedMash(fit2) ## The detection model we have fitted may be extrapolated to ## a more fine-grained systematic sample of points, with ## detectors operated on a single occasion at each... ## Total effort 400 x 1 = 400 detector-occasions, compared ## to 256 x 5 = 1280 detector-occasions for initial survey. t1 <- make.grid(nx = 1, ny = 1) t1.100 <- make.systematic (cluster = t1, spacing = 100, region = testregion) capt2a <- sim.capthist(t1.100, popn = popn2, noccasions = 1) ## one way to get number of animals per point nj <- attr(mash(capt2a), "n.mash") derivedExternal (fit2, nj = nj, cluster = t1, buffer = 100, noccasions = 1) ## Review plots base.plot <- function() { MASS::eqscplot( testregion, axes = FALSE, xlab = "", ylab = "", type = "n") polygon(testregion) } par(mfrow = c(1,3), xpd = TRUE, xaxs = "i", yaxs = "i") base.plot() plot(popn2, add = TRUE, col = "blue") mtext(side=3, line=0.5, "Population", cex=0.8, col="black") base.plot() plot (capt2a, add = TRUE,title = "Extensive survey") base.plot() plot(capt2, add = TRUE, title = "Intensive survey") par(mfrow = c(1,1), xpd = FALSE, xaxs = "r", yaxs = "r") ## defaults ## Weighted variance derivedSession(ovenbird.model.1, method = "R2") ## End(Not run)
## The `ovensong' data are pooled from 75 replicate positions of a ## 4-microphone array. The array positions are coded as the first 4 ## digits of each sound identifier. The sound data are initially in the ## object `signalCH'. We first impose a 52.5 dB signal threshold as in ## Dawson & Efford (2009, J. Appl. Ecol. 46:1201--1209). The vector nj ## includes 33 positions at which no ovenbird was heard. The first and ## second columns of `temp' hold the estimated effective sampling area ## and its standard error. ## Not run: signalCH.525 <- subset(signalCH, cutval = 52.5) nonzero.counts <- table(substring(rownames(signalCH.525),1,4)) nj <- c(nonzero.counts, rep(0, 75 - length(nonzero.counts))) temp <- derived(ovensong.model.1, se.esa = TRUE) derivednj(nj, temp["esa",1:2]) ## The result is very close to that reported by Dawson & Efford ## from a 2-D Poisson model fitted by maximizing the full likelihood. ## If nj vector has length 1, a theoretical variance is used... msk <- ovensong.model.1$mask A <- nrow(msk) * attr(msk, "area") derivednj (sum(nj), temp["esa",1:2], method = "poisson") derivednj (sum(nj), temp["esa",1:2], method = "binomial", area = A) ## Set up an array of small (4 x 4) grids, ## simulate a Poisson-distributed population, ## sample from it, plot, and fit a model. ## mash() condenses clusters to a single cluster testregion <- data.frame(x = c(0,2000,2000,0), y = c(0,0,2000,2000)) t4 <- make.grid(nx = 4, ny = 4, spacing = 40) t4.16 <- make.systematic (n = 16, cluster = t4, region = testregion) popn1 <- sim.popn (D = 5, core = testregion, buffer = 0) capt1 <- sim.capthist(t4.16, popn = popn1) fit1 <- secr.fit(mash(capt1), CL = TRUE, trace = FALSE) ## Visualize sampling tempmask <- make.mask(t4.16, spacing = 10, type = "clusterbuffer") plot(tempmask) plot(t4.16, add = TRUE) plot(capt1, add = TRUE) ## Compare model-based and empirical variances. ## Here the answers are similar because the data ## were simulated from a Poisson distribution, ## as assumed by \code{derived} derived(fit1) derivedMash(fit1) ## Now simulate a patchy distribution; note the ## larger (and more credible) SE from derivedMash(). popn2 <- sim.popn (D = 5, core = testregion, buffer = 0, model2D = "hills", details = list(hills = c(-2,3))) capt2 <- sim.capthist(t4.16, popn = popn2) fit2 <- secr.fit(mash(capt2), CL = TRUE, trace = FALSE) derived(fit2) derivedMash(fit2) ## The detection model we have fitted may be extrapolated to ## a more fine-grained systematic sample of points, with ## detectors operated on a single occasion at each... ## Total effort 400 x 1 = 400 detector-occasions, compared ## to 256 x 5 = 1280 detector-occasions for initial survey. t1 <- make.grid(nx = 1, ny = 1) t1.100 <- make.systematic (cluster = t1, spacing = 100, region = testregion) capt2a <- sim.capthist(t1.100, popn = popn2, noccasions = 1) ## one way to get number of animals per point nj <- attr(mash(capt2a), "n.mash") derivedExternal (fit2, nj = nj, cluster = t1, buffer = 100, noccasions = 1) ## Review plots base.plot <- function() { MASS::eqscplot( testregion, axes = FALSE, xlab = "", ylab = "", type = "n") polygon(testregion) } par(mfrow = c(1,3), xpd = TRUE, xaxs = "i", yaxs = "i") base.plot() plot(popn2, add = TRUE, col = "blue") mtext(side=3, line=0.5, "Population", cex=0.8, col="black") base.plot() plot (capt2a, add = TRUE,title = "Extensive survey") base.plot() plot(capt2, add = TRUE, title = "Intensive survey") par(mfrow = c(1,1), xpd = FALSE, xaxs = "r", yaxs = "r") ## defaults ## Weighted variance derivedSession(ovenbird.model.1, method = "R2") ## End(Not run)
Plot effective sampling area (Borchers and Efford 2008) as a function of increasing buffer width.
esaPlot
was previously called esa.plot
.
esaPlot (object, max.buffer = NULL, spacing = NULL, max.mask = NULL, detectfn, detectpar, noccasions, binomN = NULL, thin = 0.1, poly = NULL, poly.habitat = TRUE, session = 1, plt = TRUE, type = c('density', 'esa', 'meanpdot', 'CVpdot'), n = 1, add = FALSE, overlay = TRUE, conditional = FALSE, ...)
esaPlot (object, max.buffer = NULL, spacing = NULL, max.mask = NULL, detectfn, detectpar, noccasions, binomN = NULL, thin = 0.1, poly = NULL, poly.habitat = TRUE, session = 1, plt = TRUE, type = c('density', 'esa', 'meanpdot', 'CVpdot'), n = 1, add = FALSE, overlay = TRUE, conditional = FALSE, ...)
object |
|
max.buffer |
maximum width of buffer in metres |
spacing |
distance between mask points |
max.mask |
|
detectfn |
integer code or character string for shape of detection function 0 = halfnormal etc. – see detectfn |
detectpar |
list of values for named parameters of detection function |
noccasions |
number of sampling occasions |
binomN |
integer code for discrete distribution (see
|
thin |
proportion of mask points to retain in plot and output |
poly |
matrix of two columns interpreted as the x and y coordinates of a bounding polygon (optional) |
poly.habitat |
logical as in |
session |
vector of session indices (used if |
plt |
logical to plot results |
type |
character, what to plot |
n |
integer number of distinct individuals detected |
add |
logical to add line to an existing plot |
overlay |
logical; if TRUE then automatically |
conditional |
logical; if TRUE the reported mean and CV are conditional on detection
(see |
... |
graphical arguments passed to plot() and lines() |
Effective sampling area (esa) is defined as the integral of net
capture probability () over a
region.
esaPlot
shows the effect of increasing region size on
the value of esa for fixed values of the detection parameters. The
max.buffer
or max.mask
arguments establish the maximum
extent of the region; points (cells) within this mask are sorted by
their distance from the nearest detector. esa(buffer) is
defined as the cumulative sum of
for
, where
is the area associated with each cell.
The default (type = 'density'
) is to plot the reciprocal of esa
multiplied by n
; this is on a more familiar scale (the density
scale) and hence is easier to interpret.
Because esaPlot
uses the criterion 'distance to nearest
detector', max.mask
should be constructed to include all
habitable cells within the desired maximum buffer and no others. This
is achieved with type = "trapbuffer"
in make.mask
. It is
a good idea to set the spacing
argument of make.mask
rather than relying on the default based on nx
. Spacing may be
small (e.g. sigma/10) and the buffer of max.mask
may be quite
large (e.g. 10 sigma), as computation is fast.
Thinning serves to reduce redundancy in the plotted points, and (if
the result is saved and printed) to generate more legible numerical
output. Use thin=1
to include all points.
esaPlot
calls the internal function esaPlotsecr
when
object
is a fitted model. In this case detectfn
,
detectpar
and noccasions
are inferred from
object
.
A dataframe with columns
buffer |
buffer width |
esa |
computed effective sampling area |
density |
n/esa |
pdot |
|
pdotmin |
cumulative minimum ( |
meanpdot |
expected pdot across mask (see |
CVpdot |
CV of pdot across mask (see |
If plt = TRUE
the dataframe is returned invisibly.
The response of effective sampling area to buffer width is just one
possible mask diagnostic; it's fast, graphic, and often
sufficient. mask.check
performs more intensive checks,
usually for a smaller number of buffer widths.
The old argument 'as.density' was superceded by 'type' in 3.1.7.
Borchers, D. L. and Efford, M. G. (2008) Spatially explicit maximum likelihood methods for capture–recapture studies. Biometrics 64, 377–385.
mask
, pdot
, CVpdot
,
make.mask
, mask.check
,
Detection functions
## Not run: ## with previously fitted model esaPlot(secrdemo.0) ## from scratch trps <- make.grid() msk <- make.mask(trps, buffer = 200, spacing = 5, type = "trapbuffer") detectpar <- list(g0 = 0.2, sigma = 25) esaPlot(trps,,, msk, 0, detectpar, nocc = 10, col = "blue") esaPlot(trps,,, msk, 0, detectpar, nocc = 5, col = "green", add = TRUE) esaPlot(trps,,, msk, 0, detectpar, nocc = 5, thin = 0.002, plt = FALSE) ## End(Not run)
## Not run: ## with previously fitted model esaPlot(secrdemo.0) ## from scratch trps <- make.grid() msk <- make.mask(trps, buffer = 200, spacing = 5, type = "trapbuffer") detectpar <- list(g0 = 0.2, sigma = 25) esaPlot(trps,,, msk, 0, detectpar, nocc = 10, col = "blue") esaPlot(trps,,, msk, 0, detectpar, nocc = 5, col = "green", add = TRUE) esaPlot(trps,,, msk, 0, detectpar, nocc = 5, thin = 0.002, plt = FALSE) ## End(Not run)
Computes the expected number of individuals detected across a detector layout or at each cluster of detectors.
expected.n(object, session = NULL, group = NULL, bycluster = FALSE, splitmask = FALSE, ncores = NULL)
expected.n(object, session = NULL, group = NULL, bycluster = FALSE, splitmask = FALSE, ncores = NULL)
object |
|
session |
character session vector |
group |
group – for future use |
bycluster |
logical to output the expected number for clusters of detectors rather than whole array |
splitmask |
logical for computation method (see Details) |
ncores |
integer number of threads to be used for parallel processing |
The expected number of individuals detected is where the integration is a
summation over
object$mask
. is the probability an
individual at
will be detected at least once either on the
whole detector layout (
bycluster = FALSE
) or on the detectors
in a single cluster (see pdot for more on ).
is the expected density at
, given the model.
is
constant (i.e. density surface flat) if
object$CL == TRUE
or
object$model$D == ~1
, and for some other possible models.
If the bycluster
option is selected and detectors are not, in
fact, assigned to clusters then each detector will be treated as a
cluster, with a warning.
Setting ncores = NULL
uses the existing value from the environment variable
RCPP_PARALLEL_NUM_THREADS (see setNumThreads
).
By default, a full habitat mask is used for each cluster. This is the more robust option. Alternatively, the mask may be split into subregions defined by the cells closest to each cluster.
The calculation takes account of any fitted continuous model for spatial variation in density (note Warning).
The expected count (bycluster = FALSE) or a vector of expected counts, one per cluster. For multi-session data, a list of such vectors.
This function changed slightly between 2.1.0 and 2.1.1, and now performs as indicated here when bycluster = TRUE and clusters are not specified.
Clusters of detectors are assumed to be independent (always true with detector types ‘proximity’, ‘count’ etc.). The computed E(n) does not apply when there is competition among clusters of detectors.
The prediction of density at present considers only the base level of density covariates, such as cell-specific habitat variables.
## Not run: expected.n(secrdemo.0) expected.n(secrdemo.0, bycluster = TRUE) expected.n(ovenbird.model.D) ## Clustered design mini <- make.grid(nx = 3, ny = 3, spacing = 50, detector = "proximity") tempgrids <- trap.builder (cluster = mini , method = "all", frame = expand.grid(x = seq(1000, 9000, 2000), y = seq(1000, 9000, 2000)), plt = TRUE) capt <- sim.capthist(tempgrids, popn = list(D = 2)) tempmask <- make.mask(tempgrids, buffer = 100, type = "clusterbuffer") fit <- secr.fit(capt, mask = tempmask, trace = FALSE) En <- expected.n(fit, bycluster = TRUE) ## GoF or overdispersion statistic p <- length(fit$fit$par) y <- cluster.counts(capt) ## scaled by n-p sum((y - En)^2 / En) / (length(En)-p) sum((y - En)^2 / En) / sum(y/En) ## End(Not run)
## Not run: expected.n(secrdemo.0) expected.n(secrdemo.0, bycluster = TRUE) expected.n(ovenbird.model.D) ## Clustered design mini <- make.grid(nx = 3, ny = 3, spacing = 50, detector = "proximity") tempgrids <- trap.builder (cluster = mini , method = "all", frame = expand.grid(x = seq(1000, 9000, 2000), y = seq(1000, 9000, 2000)), plt = TRUE) capt <- sim.capthist(tempgrids, popn = list(D = 2)) tempmask <- make.mask(tempgrids, buffer = 100, type = "clusterbuffer") fit <- secr.fit(capt, mask = tempmask, trace = FALSE) En <- expected.n(fit, bycluster = TRUE) ## GoF or overdispersion statistic p <- length(fit$fit$par) y <- cluster.counts(capt) ## scaled by n-p sum((y - En)^2 / En) / (length(En)-p) sum((y - En)^2 / En) / sum(y/En) ## End(Not run)
Extract movements from a previously simulated multi-session population.
extractMoves(pop, plotn = NULL, add = FALSE, collapse = TRUE, maxradius = Inf, ...)
extractMoves(pop, plotn = NULL, add = FALSE, collapse = TRUE, maxradius = Inf, ...)
pop |
popn object from |
plotn |
integer maximum number of instances to plot at each session |
add |
logical for whether to add to existing plot |
collapse |
logical; if TRUE plots for sessions 2, 3,... are added to the first |
maxradius |
numeric radius for selecting subset of initial locations |
... |
arguments passed to |
This function is mostly used to check the movement simulations.
Moves are constrained by the edge (argument ‘edgemethod’ of sim.popn
). ‘maxradius’ may be set to restrict the extraction to the subset of animals initially near the centroid of the arena in each session.
Plotting uses the graphics function arrows
that has some quirks, such as difficult-to-suppress warnings for zero-length moves. Set code = 0
to suppress arrowheads; length = 0.1
to shorten to 0.1 inches, etc.
List of data frames, one for each session but the last (columns ‘x1’,‘y1’,‘x2’,‘y2’,‘d’).
set.seed(12345) pop3 <- sim.popn(D = 2, core = make.grid(), buffer = 200, nsessions = 3, details = list(lambda = 1.0, movemodel = 'BVE', move.a = 50, edgemethod = 'stop')) m <- extractMoves(pop3, plotn = 10, length = 0.1) mean(unlist(sapply(m, '[', 'd'))) # less than nominal 2 x move.a # For distances closer to nominal for BVE (2 x move.a = 100), # increase size of arena (e.g., buffer = 500) and consider only # central animals (e.g., maxradius = 300).
set.seed(12345) pop3 <- sim.popn(D = 2, core = make.grid(), buffer = 200, nsessions = 3, details = list(lambda = 1.0, movemodel = 'BVE', move.a = 50, edgemethod = 'stop')) m <- extractMoves(pop3, plotn = 10, length = 0.1) mean(unlist(sapply(m, '[', 'd'))) # less than nominal 2 x move.a # For distances closer to nominal for BVE (2 x move.a = 100), # increase size of arena (e.g., buffer = 500) and consider only # central animals (e.g., maxradius = 300).
A place for hints and miscellaneous advice.
Follow the usual procedure for installing from CRAN archive (see menu item Packages | Install package(s)... in Windows). You also need to get the package abind from CRAN.
Like other contributed packages, secr needs to be loaded before
each use e.g.,library(secr)
.
You can learn about changes in the current version with
news(package = "secr")
.
There are three general ways of displaying documentation from within R. Firstly, you can bring up help pages for particular functions from the command prompt. For example:
?secr
or
?secr.fit
Secondly, help.search() lets you ask for a list of the help pages on a vague topic (or just use ?? at the prompt). For example:
?? "linear models"
Thirdly, you can display various secr documents listed in
secr-package
.
Tip: to search all secr help pages open the pdf version of the manual in Acrobat Reader (secr-manual.pdf; see also ?secr) and use <ctrl> F.
There is a support forum at www.phidot.org/forum under
‘DENSITY|secr’ and another at secrgroup. See below for
more R tips. Some specific problems with secr.fit
are covered in
Troubleshooting.
If you get really stuck or find something you think is a bug then please report the problem to one of the online lists.
You may be asked to send an actual dataset - ideally, the simplest one
that exhibits the problem. Use save
to wrap
several R objects together in one .RData file, e.g.,
save("captdata", "secrdemo.0", "secrdemo.b", file =
"mydata.RData")
. Also, paste into the text of your message the output
from packageDescription( "secr" )
.
Strictly speaking, this should not happen if you have specified the same model and likelihood, although you may see a little variation due to the different maximization algorithms. Likelihoods (and estimates) may differ if you use different integration meshes (habitat masks), which can easily happen because the programs differ in how they set up the mesh. If you want to make a precise comparison, save the Density mesh to a file and read it into secr, or vice versa.
Extreme data, especially rare long-distance movements, may be handled
differently by the two programs. The ‘minprob’ component of the
‘details’ argument of secr.fit
sets a lower threshold of
probability for capture histories (smaller values are all set to
minprob), whereas Density has no explicit limit.
There are many ways - see Speed tips and secr-troubleshooting.pdf.
Keep the number of mask points to a minimum and avoid detection covariates with many levels.
Some computations can be run in parallel on multiple processors (most desktops these days have multiple cores). Likelihood calculations in secr.fit
assign capture histories to multiple parallel threads whenever possible.
The number of threads (cores) is controlled by an environment variable set by setNumThreads
or the 'ncores' argument of some functions.
Yes. See ?timevaryingcov. However, a more direct way to control for varying effort is provided - see the 'usage' atribute, which now allows a continuous measure of effort (secr-varyingeffort.pdf).
A tip: covariate models for detection fit more quickly when the covariate takes only a few different values. Use binCovariate
to bin values.
The function findFn
in package sos lets you search CRAN for
R functions by matching text in their documentation.
There is now a vast amount of R advice available on the web. For the terminally frustrated, ‘R inferno’ by Patrick Burns is recommended (https://www.burns-stat.com/pages/Tutor/R_inferno.pdf). "If you are using R and you think you're in hell, this is a map for you".
Method functions for S3 classes cannot be listed in the usual way by typing the function name at the R prompt because they are ‘hidden’ in a namespace. Get around this with getAnywhere(). For example:
getAnywhere(print.secr)
R objects have ‘attributes’ that usually are kept out of sight.
Important attributes are ‘class’ (all objects), ‘dim’ (matrices and
arrays) and ‘names’ (lists). secr hides quite a lot of useful data
as named ‘attributes’. Usually you will use summary and extraction
methods (traps
, covariates
, usage
etc.) to view and change
the attributes of the various classes of object in secr. If you're
curious, you can reveal the lot with ‘attributes’. For example, with
the demonstration capture history data ‘captdata’:
traps(captdata) ## extraction method for `traps'
attributes(captdata) ## all attributes
Also, the function str
provides a compact summary of any object:
str(captdata)
Claeskens, G. and Hjort N. L. (2008) Model Selection and Model Averaging. Cambridge: Cambridge University Press.
General function for estimating a variance inflation factor () from observed counts.
Fletcher.chat (observed, expected, np, verbose = TRUE, type = c('Fletcher', 'Wedderburn', 'both'), multinomial = FALSE)
Fletcher.chat (observed, expected, np, verbose = TRUE, type = c('Fletcher', 'Wedderburn', 'both'), multinomial = FALSE)
observed |
integer vector of observed counts, or a list of such vectors |
expected |
numeric vector of expected counts |
np |
integer number of parameters estimated |
verbose |
logical; if TRUE returns extended output |
type |
character |
multinomial |
logical; if TRUE, one df is subtracted for the constraint |
Fletcher.chat
applies the overdispersion formula of Fletcher (2012) or computes the conventional (Wedderburn 1974) variance inflation factor . It is used by
chat.nk
and adjustVarD
. The inputs ‘observed’ and ‘expected’ are vectors of counts (e.g., number of distinct individuals per detector); ‘observed’ may also be a list of such vectors, possibly simulated.
Output depends on ‘verbose’, ‘observed’ and ‘type’:
– if ‘observed’ is a list of nk vectors (usually generated by simulation) then the output is a vector of (Fletcher or Wedderburn) values, one element for each component of ‘observed’, unless type = "both" when the output is a list of two such vectors. Argument ‘verbose’ is ignored.
– if ‘observed’ is a simple vector then ‘verbose’ output is a list comprising input values, various summary statistics, and the computed Fletcher overdispersion (‘chat’). The statistic ‘cX2’ is the conventional variance inflation factor of Wedderburn (1974) – . For
verbose = FALSE
, a single estimate of is returned when
type = "Fletcher"
or type = "Wedderburn"
, otherwise a vector of the two estimates.
Fletcher, D. (2012) Estimating overdispersion when fitting a generalized linear model to sparse data. Biometrika 99, 230–237.
Wedderburn, R. W. M. (1974) Quasi-likelihood functions, generalized linear models, and the Gauss-Newton method. Biometrika 61, 439–47.
Display contours of the probability density function for the estimated
location of one or more activity centres (AC), compute values for
particular points X, or compute mode of pdf. The pdf is given by
, where
is the probability density
of range centres across the mask (Borchers and Efford 2008).
These functions were previously named fxi.secr
, fxi.contour
and fxi.mode
.
## S3 method for class 'secr' fxi(object, i = NULL, sessnum = 1, X = NULL, ncores = NULL, ...) fxiContour (object, i = 1, sessnum = 1, border = 100, nx = 64, levels = NULL, p = seq(0.1,0.9,0.1), plt = TRUE, add = FALSE, fitmode = FALSE, plotmode = FALSE, fill = NULL, output = c('list','sf','SPDF'), ncores = NULL, ...) fxiMode(object, i = 1, sessnum = 1, start = NULL, ncores = NULL, ...)
## S3 method for class 'secr' fxi(object, i = NULL, sessnum = 1, X = NULL, ncores = NULL, ...) fxiContour (object, i = 1, sessnum = 1, border = 100, nx = 64, levels = NULL, p = seq(0.1,0.9,0.1), plt = TRUE, add = FALSE, fitmode = FALSE, plotmode = FALSE, fill = NULL, output = c('list','sf','SPDF'), ncores = NULL, ...) fxiMode(object, i = 1, sessnum = 1, start = NULL, ncores = NULL, ...)
object |
a fitted secr model |
i |
integer or character vector of individuals (defaults to all in |
sessnum |
session number if |
border |
width of blank margin around the outermost detectors |
nx |
dimension of interpolation grid in x-direction |
levels |
numeric vector of confidence levels for Pr(X|wi) |
p |
numeric vector of contour levels as probabilities |
plt |
logical to plot contours |
add |
logical to add contour(s) to an existing plot |
fitmode |
logical to refine estimate of mode of each pdf |
plotmode |
logical to plot mode of each pdf |
X |
2-column matrix of x- and y- coordinates (defaults to mask) |
fill |
vector of colours to fill contours (optional) |
output |
character; format of output (list, sf or SpatialPolygonsDataFrame) |
ncores |
integer number of threadss to be used for parallel processing |
start |
vector of x-y coordinates for maximization |
... |
additional arguments passed to |
fxiContour
computes contours of AC probability density for one
or more detection histories. Increase nx
for smoother
contours. If levels
is not set, contour levels are set
to approximate the confidence levels in p
.
fxi
computes the AC probability density for one or more
detection histories; X
may contain coordinates for one or
several points; a dataframe or vector (x then y) will be coerced to a
matrix.
fxiMode
attempts to find the x- and y-coordinates
corresponding to the maximum of the AC pdf for a single detection history
(i.e. i
is of length 1). fxiMode
calls
nlm
.
fxiContour
with fitmode = TRUE
calls fxiMode
for each individual. Otherwise, the reported mode is an approximation
(mean of coordinates of highest contour).
If i
is character it will be matched to row names of
object$capthist (restricted to the relevant session in the case of a
multi-session fit); otherwise it will be interpreted as a row number.
Values of the pdf are normalised by dividing by the
integral of
over the habitat mask in
object
. (In secr >= 4.0 may differ from previous versions).
Setting ncores = NULL
uses the existing value from the environment variable
RCPP_PARALLEL_NUM_THREADS (see setNumThreads
).
If start
is not provided to fit.mode
then (from 2.9.4) the weighted mean of
all detector sites is used (see Warning below).
The ... argument gives additional control over a contour plot; for
example, set drawlabels = FALSE
to suppress contour labels.
fxi
–
Vector of probability densities
fxiContour
(output = 'list') –
Coordinates of the plotted contours are returned as a list with one component per polygon. The list is returned invisibly if plt = TRUE.
An additional component ‘mode’ reports the x-y coordinates of the highest point of each pdf (see Details).
fxiContour
(output = 'SPDF') –
Contours are returned as a SpatialPolygonsDataFrame (see package sp) with one component per animal. The attributes dataframe has two columns, the x- and y-coordinates of the mode. The SpatialPolygonsDataFrame is returned invisibly if plt = TRUE.
fxiContour
(output = 'sf') – simple features 'sf' object, as for SPDF.
fxiMode
–
List with components ‘x’ and ‘y’
fxiMode
may fail to find the true mode unless a good starting
point is provided. Note that the distribution may have multiple modes and
only one is reported. The default value of start
before secr 2.9.4
was the first detected location of the animal.
From secr 2.8.3, these functions work with both homogeneous
and inhomogeneous Poisson density models, and fxi
accepts
vector-valued i
.
See fxTotal
for a surface summed across individuals.
Borchers, D. L. and Efford, M. G. (2008) Spatially explicit maximum likelihood methods for capture–recapture studies. Biometrics 64, 377–385.
## Not run: fxi(secrdemo.0, i = 1, X = c(365,605)) ## contour first 5 detection histories plot(secrdemo.0$capthist) fxiContour (secrdemo.0, i = 1:5, add = TRUE, plotmode = TRUE, drawlabels = FALSE) ## extract modes only ## these are more reliable than those from fit.mode called directly as ## they use a contour-based approximation for the starting point fxiout <- fxiContour (secrdemo.0, i = 1:5, plt = FALSE, fitmode = TRUE) t(sapply(fxiout, "[[", "mode")) ## using fill colours ## lty = 0 suppresses contour lines ## nx = 256 ensures smooth outline plot(traps(captdata)) fxiContour(secrdemo.0, i = 1:5, add = TRUE, p = c(0.5,0.95), drawlabels = FALSE, nx = 256, fill = topo.colors(4), lty = 0) ## output as simple features sf <- fxiContour(secrdemo.0, i = 1:3, plt = FALSE, p = c(0.5,0.95), nx = 256, output = 'sf', fitmode = TRUE) ## save as ESRI shapefile testsf.shp etc. library(sf) st_write(sf, 'testsf.shp') ## plot contours and modes plot(st_as_sfc(sf)) # outline only points(sf$modex, sf$modey) ## output as SpatialPolygonsDataFrame spdf <- fxiContour(secrdemo.0, i = 1:3, plt = FALSE, p = c(0.5,0.95), nx = 256, output = 'SPDF', fitmode = TRUE) sp::plot(spdf) points(data.frame(spdf)) ## End(Not run)
## Not run: fxi(secrdemo.0, i = 1, X = c(365,605)) ## contour first 5 detection histories plot(secrdemo.0$capthist) fxiContour (secrdemo.0, i = 1:5, add = TRUE, plotmode = TRUE, drawlabels = FALSE) ## extract modes only ## these are more reliable than those from fit.mode called directly as ## they use a contour-based approximation for the starting point fxiout <- fxiContour (secrdemo.0, i = 1:5, plt = FALSE, fitmode = TRUE) t(sapply(fxiout, "[[", "mode")) ## using fill colours ## lty = 0 suppresses contour lines ## nx = 256 ensures smooth outline plot(traps(captdata)) fxiContour(secrdemo.0, i = 1:5, add = TRUE, p = c(0.5,0.95), drawlabels = FALSE, nx = 256, fill = topo.colors(4), lty = 0) ## output as simple features sf <- fxiContour(secrdemo.0, i = 1:3, plt = FALSE, p = c(0.5,0.95), nx = 256, output = 'sf', fitmode = TRUE) ## save as ESRI shapefile testsf.shp etc. library(sf) st_write(sf, 'testsf.shp') ## plot contours and modes plot(st_as_sfc(sf)) # outline only points(sf$modex, sf$modey) ## output as SpatialPolygonsDataFrame spdf <- fxiContour(secrdemo.0, i = 1:3, plt = FALSE, p = c(0.5,0.95), nx = 256, output = 'SPDF', fitmode = TRUE) sp::plot(spdf) points(data.frame(spdf)) ## End(Not run)
The summed probability densities of both observed and unobserved individuals are computed for a fitted model and dataset.
Function fx.total
was replaced by method fxTotal
in secr 5.0.0.
## S3 method for class 'secr' fxTotal(object, sessnum = 1, mask = NULL, ncores = NULL, ...)
## S3 method for class 'secr' fxTotal(object, sessnum = 1, mask = NULL, ncores = NULL, ...)
object |
a fitted secr model |
sessnum |
session number if |
mask |
x- and y- coordinates of points at which density will be computed |
ncores |
integer number of threads to be used for parallel processing |
... |
other arguments passed to |
This function calls fxi
for each detected animal and
overlays the results to obtain a summed probability density surface D.fx
for the locations of the home-range centres of detected individuals.
A separate calculation using pdot
provides the expected
spatial distribution of undetected animals, as another density
surface: crudely, D.nc(X) = D(X) * ( 1 – pdot(X)).
The pointwise sum of the two surfaces is sometimes used to represent the spatial distrbution of the population, but see Notes.
Setting ncores = NULL
uses the existing value from the environment variable
RCPP_PARALLEL_NUM_THREADS (see setNumThreads
).
An object of class ‘Dsurface’ (a variety of mask) with a ‘covariates’ attribute that is a dataframe with columns –
D.fx |
sum of |
D.nc |
expected density of undetected (‘not caught’) individuals |
D.sum |
sum of D.fx and D.nc |
All densities are in animals per hectare (the ‘scale’ argument of
plot.Dsurface
allows the units to be varied later).
The surface D.sum represents what is known from the data about a specific realisation of the spatial point process for home range centres: varying the intensity of sampling will change its shape. It is not an unbiased estimate of a biologically meaningful density surface. The surface will always tend to lack relief towards the edge of a habitat mask where the main or only contribution is from D.nc.
Borchers, D. L. and Efford, M. G. (2008) Spatially explicit maximum likelihood methods for capture–recapture studies. Biometrics 64, 377–385.
## Not run: tmp <- fxTotal(secrdemo.0) ## to plot we must name one of the covariates: ## the Dsurface default 'D.0' causes an error plot(tmp, covariate = 'D.sum', col = terrain.colors(16), plottype = 'shaded') plot(tmp, covariate = 'D.sum', col = 'white', add = TRUE, plottype = 'contour') if (interactive()) { spotHeight(tmp, prefix = 'D.sum') } fxsurface <- fxTotal(ovenbird.model.D, sessnum = 3) plot(fxsurface, covariate = 'D.sum') ## End(Not run)
## Not run: tmp <- fxTotal(secrdemo.0) ## to plot we must name one of the covariates: ## the Dsurface default 'D.0' causes an error plot(tmp, covariate = 'D.sum', col = terrain.colors(16), plottype = 'shaded') plot(tmp, covariate = 'D.sum', col = 'white', add = TRUE, plottype = 'contour') if (interactive()) { spotHeight(tmp, prefix = 'D.sum') } fxsurface <- fxTotal(ovenbird.model.D, sessnum = 3) plot(fxsurface, covariate = 'D.sum') ## End(Not run)
Forms grid cells centred on input points.
gridCells(x, cellsize = spacing(x), crs = NA)
gridCells(x, cellsize = spacing(x), crs = NA)
x |
matrix or dataframe with x- and y-coordinates |
cellsize |
length of gridcell side |
crs |
crs description suitable for |
The argument x will often be a traps or mask object with spacing attribute. Otherwise cellsize
must be provided.
See make.grid
for jittered locations within grid cells.
A simple features (sf) object of class ‘sfc_MULTIPOLYGON’.
crs
may be the integer EPSG code (e.g. 3578 Yukon Albers).
plot(gridCells(traps(captdata))) plot(traps(captdata), add = TRUE)
plot(gridCells(traps(captdata))) plot(traps(captdata), add = TRUE)
The argument hcov
in secr.fit
is used to fit a hybrid
mixture model. ‘Hybrid’ refers to a flexible combination of latent
classes (as in a finite mixture) and known classes (cf groups or
sessions). A hybrid mixture model includes a parameter ‘pmix’ for the
mixing proportion and optionally allows detection parameters to be
modelled as class-specific ( ~ h2). This is particularly useful for
modelling sex ratio and sex differences in detection, and matches the
Bayesian sex-specific model of Gardner et al. (2010).
For observed animals all of unknown class the model is identical to a finite mixture (i.e. latent-class) model. For observed animals all of known class, the classes are no longer ‘latent’ and the model is equivalent to a grouped model with an additional binomial factor for class membership.
hcov
identifies a single individual covariate (the class
covariate) that should be a factor with two levels, or contain
character values that will be coerced to a factor (e.g., ‘f’,
‘m’). Missing values (NA) are used for individuals of unknown
class. If hcov
has more than two levels, all but the first two
levels are converted to NA (but see exception for h3 models below).
It is assumed that the probability of recording a missing value for the class covariate is independent of the true class membership (e.g., sex equally likely to be recorded for males and females).
A hybrid mixture model is fitted whenever hcov
is not
NULL. Mixture models include a parameter ‘pmix’, the mixing
proportion. If the covariate identified by hcov
is missing (''
or NA) for all individuals and a mixture term (h2 or h3)
appears in the detection model (e.g., g0 ~ h2) then a conventional
finite mixture model is fitted (cf Pledger 2000, Borchers & Efford
2008).
As with finite mixture models, any detection parameter (g0, sigma etc.) may be modelled as depending on mixture class by model specifications such as (g0 ~ h2, sigma ~ h2). See Examples.
In general hcov
has been designed for two classes and two
classes are assumed if neither ‘h2’ nor ‘h3’ appears in the model
formulae. However, there is a small exception: hcov
may have
three non-missing levels if ‘h3’ appears in a model formula. Note
that h2 cannot be combined with h3; h3 is for advanced use only and
has not been fully tested.
The number of fitted parameters is the same as the corresponding finite mixture model if mixture terms (‘h2’, ‘h3’) appear in the model formulae. Otherwise (no mixture terms) estimating pmix requires a single extra parameter. The estimate of pmix then depends solely on the observed class proportions in the covariate, and the beta variance-covariance matrix will show zero covariance of pmix with other detection parameters.
Variation in the parameter pmix may be modelled across sessions i.e., models such as pmix ~ session or pmix ~ Session are valid, as are formulae involving session covariates defined in the sessioncov argument of secr.fit.
If no mixture term appears in the formula for pmix then one is added automatically (usually ‘h2’). This serves mostly to keep track of values in the output.
Attempting to model pmix as a function of individual covariates or other within-session terms (t, b etc.) will cause an error.
When you display a fitted secr model the parameter estimates are in a
final section headed 'Fitted (real) parameters evaluated at base
levels of covariates'. The same output may be obtained by calling the
predict
method directly. Calling predict
has the advantage
that you can obtain estimates for levels of the covariates other than
the base levels, by specifying newdata
. An example below shows
how to specify h2 in newdata
. [Note: predict
is generic, and you must consult ?predict.secr to see the help for the
specific implementation of this method for fitted secr objects].
The output from predict.secr
for a mixture model is a list with
one component for each (possibly latent) class. Each row corresponds
to a fitted real parameter: ordinarily these include the detection
parameters (e.g., g0, sigma) and the mixing proportion (pmix).
In the case of a model fitted by maximizing the full likelihood
(CL = FALSE
), density D will also appear in the output. Note
that only one parameter for density is estimated, the total density
across classes. This total density figure appears twice in the
output, once for each class.
The standard error (SE.estimate) is shown for each parameter. These
are asymptotic estimates back-transformed from the link scale. The
confidence limits are also back-transformed from the link scale (95%
CI by default; vary alpha
in predict.secr
if you want
e.g. 90% CI).
The mixing proportion pmix depends on the composition of the sample
with respect to hcov
and the detection model. For a null
detection model the mixing proportion is exactly the proportion in the
sample, with appropriate binomial confidence limits. Otherwise, the
mixing proportion adjusts for class differences in the probability and
scale of detection (see Examples).
The preceding refers to the default behaviour when pmix ~ h2. It is possible also to fix the mixing proportion at any arbitrary value (e.g., fixed = list(pmix = 0.5) for 1:1 sex ratio).
On output the classes are tagged with the factor levels of hcov
,
regardless of how few or how many individuals were actually of known
class. If only a small fraction were of known class, and there is
cryptic variation unrelated to hcov
, then the association
between the fitted classes and the nominal classes (i.e. levels of
hcov
) may be weak, and should not be trusted.
Hybrid mixture models are incompatible with groups as presently implemented.
The hcov likelihood conditions on the number of known-class
individuals. A model fitted with hcov = NULL
or with a
different hcov covariate has in effect a different data set, and
likelihoods, deviances or AICs cannot be compared. AIC can be used to
compare models provided they all have the same hcov covariate in the
call to secr.fit
, whether or not h2 appears in the model
definition.
The likelihood of the hybrid mixture model is detailed in an appendix of the vignette secr-finitemixtures.pdf.
Borchers, D.L. and Efford, M.G. (2008) Spatially explicit maximum likelihood methods for capture–recapture studies. Biometrics 64, 377–385.
Gardner, B., Royle, J.A., Wegan, M.T., Rainbolt, R. and Curtis, P. (2010) Estimating black bear density using DNA data from hair snares. Journal of Wildlife Management 74, 318–325.
Pledger, S. (2000) Unified maximum likelihood estimates for closed capture–recapture models using mixtures. Biometrics 56, 434–442.
## Not run: ## house mouse dataset, morning trap clearances ## 81 female, 78 male, 1 unknown morning <- subset(housemouse, occ = c(1,3,5,7,9)) summary(covariates(morning)) ## speedy model fitting with coarse mask mmask <- make.mask(traps(morning), buffer = 20, nx = 32) ## assuming equal detection of males and females ## fitted sex ratio p(female) = 0.509434 = 81 / (81 + 78) fit.0 <- secr.fit(morning, hcov = "sex", mask = mmask, trace = FALSE) predict(fit.0) ## allowing sex-specific detection parameters ## this leads to new estimate of sex ratio fit.h2 <- secr.fit(morning, hcov = "sex", mask = mmask, trace = FALSE, model = list(g0 ~ h2, sigma ~ h2)) predict(fit.h2) ## specifying newdata for h2 - equivalent to predict(fit.h2) predict(fit.h2, newdata = data.frame(h2 = factor(c('f','m')))) ## conditional likelihood fit of preceding model ## estimate of sex ratio does not change fit.CL.h2 <- secr.fit(morning, hcov = "sex", mask = mmask, trace = FALSE, CL = TRUE, model = list(g0 ~ h2, sigma ~ h2)) predict(fit.CL.h2) ## did sexes differ in detection parameters? fit.CL.0 <- secr.fit(morning, hcov = "sex", mask = mmask, trace = FALSE, CL = TRUE, model = list(g0 ~ 1, sigma ~ 1)) LR.test(fit.CL.h2, fit.CL.0) ## did sex ratio deviate from 1:1? fit.CL.h2.50 <- secr.fit(morning, hcov = "sex", mask = mmask, trace = FALSE, CL = TRUE, model = list(g0 ~ h2, sigma ~ h2), fixed = list(pmix = 0.5)) LR.test(fit.CL.h2, fit.CL.h2.50) ## did sexes show extra-compensatory variation in lambda0? ## (Efford and Mowat 2014) fit.CL.a0 <- secr.fit(morning, hcov = "sex", mask = mmask, trace = FALSE, CL = TRUE, model = list(a0 ~ 1, sigma ~ h2)) LR.test(fit.CL.h2, fit.CL.a0) ## trend in ovenbird sex ratio, assuming sex-specific detection omask <- make.mask(traps(ovenCH), buffer = 300, nx = 32) fit.sextrend <- secr.fit(ovenCH, model = list(g0~h2, sigma~h2, pmix~Session), hcov = "Sex", CL = TRUE, mask = omask, trace = FALSE) predict(fit.sextrend)[1:5] ## End(Not run)
## Not run: ## house mouse dataset, morning trap clearances ## 81 female, 78 male, 1 unknown morning <- subset(housemouse, occ = c(1,3,5,7,9)) summary(covariates(morning)) ## speedy model fitting with coarse mask mmask <- make.mask(traps(morning), buffer = 20, nx = 32) ## assuming equal detection of males and females ## fitted sex ratio p(female) = 0.509434 = 81 / (81 + 78) fit.0 <- secr.fit(morning, hcov = "sex", mask = mmask, trace = FALSE) predict(fit.0) ## allowing sex-specific detection parameters ## this leads to new estimate of sex ratio fit.h2 <- secr.fit(morning, hcov = "sex", mask = mmask, trace = FALSE, model = list(g0 ~ h2, sigma ~ h2)) predict(fit.h2) ## specifying newdata for h2 - equivalent to predict(fit.h2) predict(fit.h2, newdata = data.frame(h2 = factor(c('f','m')))) ## conditional likelihood fit of preceding model ## estimate of sex ratio does not change fit.CL.h2 <- secr.fit(morning, hcov = "sex", mask = mmask, trace = FALSE, CL = TRUE, model = list(g0 ~ h2, sigma ~ h2)) predict(fit.CL.h2) ## did sexes differ in detection parameters? fit.CL.0 <- secr.fit(morning, hcov = "sex", mask = mmask, trace = FALSE, CL = TRUE, model = list(g0 ~ 1, sigma ~ 1)) LR.test(fit.CL.h2, fit.CL.0) ## did sex ratio deviate from 1:1? fit.CL.h2.50 <- secr.fit(morning, hcov = "sex", mask = mmask, trace = FALSE, CL = TRUE, model = list(g0 ~ h2, sigma ~ h2), fixed = list(pmix = 0.5)) LR.test(fit.CL.h2, fit.CL.h2.50) ## did sexes show extra-compensatory variation in lambda0? ## (Efford and Mowat 2014) fit.CL.a0 <- secr.fit(morning, hcov = "sex", mask = mmask, trace = FALSE, CL = TRUE, model = list(a0 ~ 1, sigma ~ h2)) LR.test(fit.CL.h2, fit.CL.a0) ## trend in ovenbird sex ratio, assuming sex-specific detection omask <- make.mask(traps(ovenCH), buffer = 300, nx = 32) fit.sextrend <- secr.fit(ovenCH, model = list(g0~h2, sigma~h2, pmix~Session), hcov = "Sex", CL = TRUE, mask = omask, trace = FALSE) predict(fit.sextrend)[1:5] ## End(Not run)
Returns the first or last parts of secr objects
## S3 method for class 'mask' head(x, n=6L, ...) ## S3 method for class 'Dsurface' head(x, n=6L, ...) ## S3 method for class 'traps' head(x, n=6L, ...) ## S3 method for class 'capthist' head(x, n=6L, ...) ## S3 method for class 'mask' tail(x, n=6L, ...) ## S3 method for class 'Dsurface' tail(x, n=6L, ...) ## S3 method for class 'traps' tail(x, n=6L, ...) ## S3 method for class 'capthist' tail(x, n=6L, ...)
## S3 method for class 'mask' head(x, n=6L, ...) ## S3 method for class 'Dsurface' head(x, n=6L, ...) ## S3 method for class 'traps' head(x, n=6L, ...) ## S3 method for class 'capthist' head(x, n=6L, ...) ## S3 method for class 'mask' tail(x, n=6L, ...) ## S3 method for class 'Dsurface' tail(x, n=6L, ...) ## S3 method for class 'traps' tail(x, n=6L, ...) ## S3 method for class 'capthist' tail(x, n=6L, ...)
x |
‘mask’, ‘traps’ or ‘capthist’ object |
n |
a single integer. If positive, size for the resulting object: number of elements for a vector (including lists), rows for a matrix or data frame or lines for a function. If negative, all but the n last/first number of elements of x. |
... |
other arguments passed to |
These custom S3 methods retain the class of the target object, unlike the default methods applied to ‘mask’, ‘Dsurface’, ‘traps’ or ‘capthist’ objects.
An object of the same class as x, but (usually) fewer rows.
head(possummask)
head(possummask)
Some ad hoc measures of home range size may be calculated in secr from capture–recapture data:
dbar
is the mean distance between consecutive capture locations,
pooled over individuals (e.g. Efford 2004). moves
returns the
raw distances.
MMDM
(for ‘Mean Maximum Distance Moved’) is the average maximum
distance between detections of each individual i.e. the observed range
length averaged over individuals (Otis et al. 1978).
ARL
(or ‘Asymptotic Range Length’) is obtained by fitting an
exponential curve to the scatter of observed individual range length vs
the number of detections of each individual (Jett and Nichols 1987: 889).
RPSV
(for ‘Root Pooled Spatial Variance’) is a measure of the 2-D
dispersion of the locations at which individual animals are detected,
pooled over individuals (cf Calhoun and Casby 1958, Slade and Swihart 1983).
moves
reports the distance between successive detections of each animal.
centroids
reports the averaged coordinates of each animal's detections
ORL
reports the observed range length of each animal, the maximum
distance between any two detections.
trapsPerAnimal
tabulates the number of animals recorded at 1, 2, ..., K detectors
dbar(capthist, userdist = NULL, mask = NULL) MMDM(capthist, min.recapt = 1, full = FALSE, userdist = NULL, mask = NULL) ARL(capthist, min.recapt = 1, plt = FALSE, full = FALSE, userdist = NULL, mask = NULL) moves(capthist, userdist = NULL, mask = NULL, names = FALSE) RPSV(capthist, CC = FALSE) ORL(capthist, userdist = NULL, mask = NULL) centroids(capthist) trapsPerAnimal(capthist)
dbar(capthist, userdist = NULL, mask = NULL) MMDM(capthist, min.recapt = 1, full = FALSE, userdist = NULL, mask = NULL) ARL(capthist, min.recapt = 1, plt = FALSE, full = FALSE, userdist = NULL, mask = NULL) moves(capthist, userdist = NULL, mask = NULL, names = FALSE) RPSV(capthist, CC = FALSE) ORL(capthist, userdist = NULL, mask = NULL) centroids(capthist) trapsPerAnimal(capthist)
capthist |
object of class |
userdist |
function or matrix with user-defined distances |
mask |
habitat mask passed to userdist function, if required |
names |
logical; should results be ordered alphanumerically by row names? |
min.recapt |
integer minimum number of recaptures for a detection history to be used |
plt |
logical; if TRUE observed range length is plotted against number of recaptures |
full |
logical; set to TRUE for detailed output |
CC |
logical for whether to use Calhoun and Casby formula |
dbar
is defined as –
When CC = FALSE
, RPSV
is defined as –
.
Otherwise (CC = TRUE
), RPSV
uses the formula of Calhoun
and Casby (1958) with a different denominator –
.
The Calhoun and Casby formula (offered from 2.9.1) correctly estimates
when trapping is on an infinite, fine grid, and is preferred
for this reason. The original RPSV
(
CC = FALSE
) is retained as the default for compatibility with
previous versions of secr.
RPSV
has a specific role as a proxy for
detection scale in inverse-prediction estimation of density (Efford
2004, 2023).
RPSV
is used in autoini
to obtain plausible starting
values for maximum likelihood estimation.
MMDM
and ARL
discard data from detection histories
containing fewer than min.recapt
+1 detections.
The userdist
option is included for exotic non-Euclidean cases
(see e.g. secr.fit
details). RPSV is not defined for
non-Euclidean distances.
If capthist
comprises standalone telemetry data (all detector 'telemetry')
then calculations are performed on the telemetry coordinates. If capthist
combines telemetry data and conventional detections (‘multi’, ‘proximity’ etc.)
then only the conventional data are summarised.
Movements are reliably reported by moves
only if there is a maximum of one detection per animal per occasion. The sequence of detections within any occasion is not known; where these occur the sequence used by moves
is arbitrary (sequence follows detector index).
For dbar
, MMDM
, ARL
and RPSV
–
Scalar distance in metres, or a list of such values if capthist
is a multi-session list.
The full
argument may be used with MMDM
and ARL
to
return more extensive output, particularly the observed range length for
each detection history.
For moves
–
List with one component for each animal, a vector of distances, or numeric(0) if the animal is detected only once. A list of such lists if capthist
is a multi-session list.
For centroids
–
For a single-session capthist, a matrix of two columns, the x- and y-coordinates of the centroid of the detections of each animal. The number of detections is returned as the attribute ‘Ndetections’, a 1-column matrix.
For a multi-session capthist, a 3-D array as before, but with a third dimension for the session. Centroid coordinates are missing (NA) if the animal was not detected in a session. The attribute ‘Ndetections’ with the number of detections per animal and session is a matrix.
For trapsPerAnimal
–
A vector with the number of animals detected at k detectors.
All measures are affected by the arrangement of detectors. dbar
is also affected quite strongly by serial correlation in the sampled
locations. Using dbar
with ‘proximity’ detectors raises a problem
of interpretation, as the original sequence of multiple detections
within an occasion is unknown. RPSV is a value analogous to the standard
deviation of locations about the home range centre.
The value returned by dbar
for ‘proximity’ or ‘count’ detectors
is of little use because multiple detections of an individual within an
occasion are in arbitrary order.
Inclusion of these measures in the secr package does not mean they are
recommended for general use! It is usually better to use a spatial
parameter from a fitted model (e.g., of the
half-normal detection function). Even then, be careful that
is not ‘contaminated’ with behavioural effects (e.g.
attraction of animal to detector) or ‘detection at a distance’.
The argument 'names' was added in 3.0.1. The default names = FALSE
causes a change in behaviour from that version onwards.
Calhoun, J. B. and Casby, J. U. (1958) Calculation of home range and density of small mammals. Public Health Monograph. No. 55. U.S. Government Printing Office.
Efford, M. G. (2004) Density estimation in live-trapping studies. Oikos 106, 598–610.
Efford, M. G. (2023) ipsecr: An R package for awkward spatial capture–recapture data. Methods in Ecology and Evolution In press.
Jett, D. A. and Nichols, J. D. (1987) A field comparison of nested grid and trapping web density estimators. Journal of Mammalogy 68, 888–892.
Otis, D. L., Burnham, K. P., White, G. C. and Anderson, D. R. (1978) Statistical inference from capture data on closed animal populations. Wildlife Monographs 62, 1–135.
Slade, N. A. and Swihart, R. K. (1983) Home range indices for the hispid cotton rat (Sigmodon hispidus) in Northeastern Kansas. Journal of Mammalogy 64, 580–590.
dbar(captdata) RPSV(captdata) RPSV(captdata, CC = TRUE) centr <- centroids(captdata) plot(traps(captdata), border = 20 ) text(centr[,1], centr[,2], attr(centr, 'Ndetections')) text(centr[,1]+2, centr[,2]+3, rownames(captdata), cex = 0.6, adj = 0)
dbar(captdata) RPSV(captdata) RPSV(captdata, CC = TRUE) centr <- centroids(captdata) plot(traps(captdata), border = 20 ) text(centr[,1], centr[,2], attr(centr, 'Ndetections')) text(centr[,1]+2, centr[,2]+3, rownames(captdata), cex = 0.6, adj = 0)
Data from multiple searches for flat-tailed horned lizards (Phrynosoma mcalli) on a plot in Arizona, USA.
hornedlizardCH
hornedlizardCH
The flat-tailed horned lizard (Phrynosoma mcalli) is a desert lizard found in parts of southwestern Arizona, southeastern California and northern Mexico. There is considerable concern about its conservation status. The species is cryptically coloured and has the habit of burying under the sand when approached, making it difficult or impossible to obtain a complete count (Grant and Doherty 2007).
K. V. Young conducted a capture–recapture survey of flat-tailed horned lizards 25 km south of Yuma, Arizona, in the Sonoran Desert. The habitat was loose sand dominated by creosote bush and occasional bur-sage and Galletta grass. A 9-ha plot was surveyed 14 times over 17 days (14 June to 1 July 2005). On each occasion the entire 300 m x 300 m plot was searched for lizards. Locations within the plot were recorded by handheld GPS. Lizards were captured by hand and marked individually on their underside with a permanent marker. Marks are lost when the lizard sheds, but this happens infrequently and probably caused few or no identification errors during the 2.5-week study.
A total of 68 individuals were captured 134 times. Exactly half of the individuals were recaptured at least once.
Royle and Young (2008) analysed the present dataset to demonstrate a method for density estimation using data augmentation and MCMC simulation. They noted that the plot size was much larger than has been suggested as being practical in operational monitoring efforts for this species, that the plot was chosen specifically because a high density of individuals was present, and that high densities typically correspond to less movement in this species. The state space in their analysis was a square comprising the searched area and a 100-m buffer (J. A. Royle pers. comm.).
The detector type for these data is ‘polygonX’ and there is a single
detector (the square plot). The data comprise a capture history matrix
(the body of hornedlizardCH
) and the x-y coordinates of each
positive detection (stored as an attribute that may be displayed with
the ‘xy’ function); the ‘traps’ attribute of hornedlizardCH
contains the vertices of the plot. See
secr-datainput.pdf for guidance on
data input.
Non-zero entries in a polygonX capture-history matrix indicate the number of the polygon containing the detection. In this case there was just one polygon, so entries are 0 or 1. No animal can appear more than once per occasion with the polygonX detector type, so there is no need to specify ‘binomN = 1’ in secr.fit.
Object | Description |
hornedlizardCH | single-session capthist object |
Royle and Young (2008) and J. A. Royle (pers. comm.), with additional information from K. V. Young (pers. comm.).
Efford, M. G. (2011) Estimation of population density by spatially explicit capture–recapture analysis of data from area searches. Ecology 92, 2202–2207.
Grant, T. J. and Doherty, P. F. (2007) Monitoring of the flat-tailed horned lizard with methods incorporating detection probability. Journal of Wildlife Management 71, 1050–1056
Marques, T. A., Thomas, L. and Royle, J. A. (2011) A hierarchical model for spatial capture–recapture data: Comment. Ecology 92, 526–528.
Royle, J. A. and Young, K. V. (2008) A hierarchical model for spatial capture–recapture data. Ecology 89, 2281–2289.
capthist
, detector
, reduce.capthist
plot(hornedlizardCH, tracks = TRUE, varycol = FALSE, lab1 = TRUE, laboff = 6, border = 10, title = "Flat-tailed Horned Lizards (Royle & Young 2008)") table(table(animalID(hornedlizardCH))) traps(hornedlizardCH) ## show first few x-y coordinates head(xy(hornedlizardCH)) ## Not run: ## Compare default (Poisson) and binomial models for number ## caught FTHL.fit <- secr.fit(hornedlizardCH) FTHLbn.fit <- secr.fit(hornedlizardCH, details = list(distribution = "binomial")) collate(FTHL.fit, FTHLbn.fit)[,,,"D"] ## Collapse occasions (does not run faster) hornedlizardCH.14 <- reduce(hornedlizardCH, newoccasions = list(1:14), outputdetector = "polygon") FTHL14.fit <- secr.fit(hornedlizardCH.14, binomN = 14) ## End(Not run)
plot(hornedlizardCH, tracks = TRUE, varycol = FALSE, lab1 = TRUE, laboff = 6, border = 10, title = "Flat-tailed Horned Lizards (Royle & Young 2008)") table(table(animalID(hornedlizardCH))) traps(hornedlizardCH) ## show first few x-y coordinates head(xy(hornedlizardCH)) ## Not run: ## Compare default (Poisson) and binomial models for number ## caught FTHL.fit <- secr.fit(hornedlizardCH) FTHLbn.fit <- secr.fit(hornedlizardCH, details = list(distribution = "binomial")) collate(FTHL.fit, FTHLbn.fit)[,,,"D"] ## Collapse occasions (does not run faster) hornedlizardCH.14 <- reduce(hornedlizardCH, newoccasions = list(1:14), outputdetector = "polygon") FTHL14.fit <- secr.fit(hornedlizardCH.14, binomN = 14) ## End(Not run)
Data of H. N. Coulombe from live trapping of feral house mice (Mus musculus) in a salt marsh, California, USA.
housemouse
housemouse
H. N. Coulombe conducted a live-trapping study on an outbreak of feral house mice in a salt marsh in mid-December 1962 at Ballana Creek, Los Angeles County, California. A square 10 x 10 grid was used with 100 Sherman traps spaced 3 m apart. Trapping was done twice daily, morning and evening, for 5 days.
The dataset was described by Otis et al. (1978) and distributed with their CAPTURE software (now available from https://eesc.usgs.gov/mbr/software/capture.shtml). Otis et al. (1978 p. 62, 68) cite Coulombe's unpublished 1965 master's thesis from the University of California, Los Angeles, California.
The data are provided as a single-session capthist
object. There
are two individual covariates: sex (factor levels ‘f’, ‘m’) and age
class (factor levels ‘j’, ‘sa’, ‘a’). The sex of two animals is not
available (NA); it is necessary to drop these records for analyses
using ‘sex’ unless missing values are specifically allowed, as in hcov
.
The datasets were originally in the CAPTURE ‘xy complete’ format which for each detection gives the ‘column’ and ‘row’ numbers of the trap (e.g. ‘ 9 5’ for a capture in the trap at position (x=9, y=5) on the grid). Trap identifiers have been recoded as strings with no spaces by inserting zeros (e.g. ‘0905’ in this example).
Sherman traps are designed to capture one animal at a time, but the data include 30 double captures and one occasion when there were 4 individuals in a trap at one time. The true detector type therefore falls between ‘single’ and ‘multi’. Detector type is set to ‘multi’ in the distributed data objects.
Otis et al. (1978) report various analyses including a closure test on the full data, and model selection and density estimation on data from the mornings only.
File ‘examples’ distributed with program CAPTURE.
Otis, D. L., Burnham, K. P., White, G. C. and Anderson, D. R. (1978) Statistical inference from capture data on closed animal populations. Wildlife Monographs 62, 1–135.
plot(housemouse, title = paste("Coulombe (1965), Mus musculus,", "California salt marsh"), border = 5, rad = 0.5, gridlines = FALSE) morning <- subset(housemouse, occ = c(1,3,5,7,9)) summary(morning) ## drop 2 unknown-sex mice known.sex <- subset(housemouse, !is.na(covariates(housemouse)$sex)) ## reveal multiple captures table(trap(housemouse), occasion(housemouse)) ## Not run: ## assess need to distinguish morning and afternoon samples housemouse.0 <- secr.fit (housemouse, buffer = 20) housemouse.ampm <- secr.fit (housemouse, model = g0~tcov, buffer = 20, timecov = c(0,1,0,1,0,1,0,1,0,1)) AIC(housemouse.0, housemouse.ampm) ## End(Not run)
plot(housemouse, title = paste("Coulombe (1965), Mus musculus,", "California salt marsh"), border = 5, rad = 0.5, gridlines = FALSE) morning <- subset(housemouse, occ = c(1,3,5,7,9)) summary(morning) ## drop 2 unknown-sex mice known.sex <- subset(housemouse, !is.na(covariates(housemouse)$sex)) ## reveal multiple captures table(trap(housemouse), occasion(housemouse)) ## Not run: ## assess need to distinguish morning and afternoon samples housemouse.0 <- secr.fit (housemouse, buffer = 20) housemouse.ampm <- secr.fit (housemouse, model = g0~tcov, buffer = 20, timecov = c(0,1,0,1,0,1,0,1,0,1)) AIC(housemouse.0, housemouse.ampm) ## End(Not run)
Functions called internally by secr and exported but not usually called by users.
boundarytoSF (poly) Dfn2(designD, beta = NULL, ...)
boundarytoSF (poly) Dfn2(designD, beta = NULL, ...)
poly |
data to define one or more polygons |
designD |
dataframe of density design data (output from |
beta |
numeric vector of beta values (see Details for NULL) |
... |
other arguments (not used) |
The function boundarytoSF
converts various
possible polygon input formats to a standard form (sfc).
Possible inputs are:
Input | From | Note |
2-column matrix or dataframe | base R | |
SpatialPolygons | sp | |
SpatialPolygonsDataFrame | sp | |
SpatVector | terra | |
sf | sf | geometry type POLYGON or MULTIPOLYGON |
sfc | sf | geometry type POLYGON or MULTIPOLYGON |
Matrix input defines a single polygon.
Dfn2
is supplied automatically as 'details' argument Dfn in
secr.fit
when the switch Dlambda is set to TRUE for the
multi-session trend reparameterization of density. Dfn2
uses beta = NULL to return the required number of density coefficients
(beta parameters) in the model.
boundarytoSF
– Spatial object of sf class sfc, containing
a geometry set of type POLYGON or MULTIPOLYGON. NULL input results in NULL output.
Dfn2
– Vector of density values on the link scale, suitable for the internal array (mask x groups x sessions).
Hijmans, R. J. (2022) terra: Spatial Data Analysis. R package version 1.5-14. https://rspatial.org/terra/
Pebesma, E. (2018) Simple features for R: standardized support for spatial vector data. The R Journal 10(1), 439–446. https://doi.org/10.32614/RJ-2018-009
Pebesma, E.J. and Bivand, R. S. (2005) Classes and methods for spatial data in R. R News 5(2), 9–13. https://cran.r-project.org/doc/Rnews/Rnews_2005-2.pdf.
pointsInPolygon
,
secr-spatialdata.pdf,
predictDlambda
,
secr-trend.pdf,
## Not run: poly <- cbind(x = c(0,6,6,0,0), y = c(0,0,6,6,0)) secr:::boundarytoSF(poly) ## End(Not run)
## Not run: poly <- cbind(x = c(0,6,6,0,0), y = c(0,0,6,6,0)) secr:::boundarytoSF(poly) ## End(Not run)
Functions for data manipulation
intervals(object, ...) intervals(object) <- value sessionlabels(object, ...) sessionlabels(object) <- value
intervals(object, ...) intervals(object) <- value sessionlabels(object, ...) sessionlabels(object) <- value
object |
capthist object |
value |
vector of intervals or primary session labels |
... |
other arguments (not used) |
intervals
extracts the ‘interval’ attribute if it exists.
The attribute ‘intervals’ is set automatically by the secr function
join
.
sessionlabels
provides session names for the primary sessions encoded
in a “single-session” capthist object (e.g., the result of join
)
that has an intervals attribute. The names are used by some summary functions
in the package openCR (M. Efford unpubl.) (m.array
, JS.counts
).
The function session
has a different purpose: labelling
sessions in a multi-session capthist object. However, session
names of multi-session input are used automatically by join
to construct the
sessionlabels
attribute of the resulting single-session object.
For intervals
, a numeric vector of time intervals, one less than the number of occasions (secondary sessions).
For sessionlabels
, a character vector of primary session names.
There is a naming conflict with the intervals function in nlme.
singlesessionCH <- join(ovenCH) intervals(singlesessionCH) sessionlabels(singlesessionCH)
singlesessionCH <- join(ovenCH) intervals(singlesessionCH) sessionlabels(singlesessionCH)
Make a single-session capthist object from a list of single-session objects, or a multi-session capthist object.
join(object, remove.dupl.sites = TRUE, tol = 0.001, sites.by.name = FALSE, drop.sites = FALSE, intervals = NULL, sessionlabels = NULL, timevaryingcov = NULL) unjoin(object, intervals, ...)
join(object, remove.dupl.sites = TRUE, tol = 0.001, sites.by.name = FALSE, drop.sites = FALSE, intervals = NULL, sessionlabels = NULL, timevaryingcov = NULL) unjoin(object, intervals, ...)
object |
list of single-session objects, or a multi-session
capthist object [ |
remove.dupl.sites |
logical; if TRUE then a single record is retained for each trap site used in multiple input sessions |
tol |
absolute distance in metres within which sites are considered identical |
sites.by.name |
logical; if TRUE and |
drop.sites |
logical; if TRUE then site information is discarded |
intervals |
vector of times between sessions (join) or occasions (unjoin; zero indicates same session) |
sessionlabels |
vector of session names |
timevaryingcov |
character vector of covariate names |
... |
other arguments passed to |
join
The input sessions are assumed to be of the same detector type and to have the same attributes (e.g., covariates should be present for all or none).
The number of occasions (columns) in the output is equal to the sum of the number of occasions in each input.
Duplicates may be defined either as sites within a given distance (tol
) or sites with the same name (sites.by.name = TRUE
). Using site names is faster.
For non-spatial analyses it is efficient to drop the third dimension and discard the traps attribute (drop.sites = TRUE
).
A new dataframe of individual covariates is formed using the covariates for the first occurrence of each animal.
If timevaryingcov
is given then for each name a new covariate is generated for each session and populated with values observed in that session, or NA if the animal was not detected. A ‘timevaryingcov’ (list) attribute is created that associates each set of new session-specific columns with the corresponding old name, so that it may be used in formulae (see timevaryingcov
).
Attributes xy and signal are handled appropriately, as is trap usage.
unjoin
The input grouping of occasions (columns) into sessions is specified via
intervals
. This is a vector of length one less than the number of
occasions (columns) in object
. Elements greater than zero
indicate a new session.
The intervals
argument may be omitted if object
has a
valid ‘intervals’ attribute, as in the output from join
.
For join
, a single-session capthist object. The vector attribute ‘intervals’ records the
distinction between occasions that are adjacent in the input (interval =
0) and those that are in consecutive sessions (e.g., interval = 1); ‘intervals’
has length one less than the number of occasions.
For unjoin
, a multi-session capthist object. Sessions are named
with integers.
Do not confuse unjoin
with split.capthist
which
splits by row (animal) rather than by column (occasion).
Occasions survive intact; to pool occasions use
reduce.capthist
.
join
was modified in version 2.9.5 to check whether the
components of ‘object’ all used the same detectors (‘traps’) (putting
aside differences in usage). If the traps are identical and
remove.dupl.sites = TRUE then the resulting ‘capthist’ uses the common list of
detectors, with a usage attribute formed by concatenating the usage
columns of the input. This is faster than the previous filtering
algorithm using ‘tol’; the older algorithm is still used if the traps differ.
Problems may be encountered with large datasets. These may be alleviated by setting sites.by.name = TRUE (if matching sites have matching names, avoiding the need for coordinate matching) or drop.sites = TRUE (if only non-spatial data are required for openCR).
joined.ovenCH <- join (ovenCH) summary(joined.ovenCH) attr(joined.ovenCH, "intervals") summary(unjoin(joined.ovenCH)) ## Not run: ## suppose the 5-year ovenbird covariates include a column for weight ## (here generated as random numbers) for (i in 1:5) covariates(ovenCH[[i]])$wt <- runif(nrow(ovenCH[[i]])) ## construct single-session version of data for openCR ## identify 'wt' as varying across years ovenCHj <- join(ovenCH, timevaryingcov = 'wt') head(covariates(ovenCHj)) timevaryingcov(ovenCHj) ## Use example: openCR.fit(ovenCHj, model = p~wt) ## End(Not run)
joined.ovenCH <- join (ovenCH) summary(joined.ovenCH) attr(joined.ovenCH, "intervals") summary(unjoin(joined.ovenCH)) ## Not run: ## suppose the 5-year ovenbird covariates include a column for weight ## (here generated as random numbers) for (i in 1:5) covariates(ovenCH[[i]])$wt <- runif(nrow(ovenCH[[i]])) ## construct single-session version of data for openCR ## identify 'wt' as varying across years ovenCHj <- join(ovenCH, timevaryingcov = 'wt') head(covariates(ovenCHj)) timevaryingcov(ovenCHj) ## Use example: openCR.fit(ovenCHj, model = p~wt) ## End(Not run)
Computes the overlap index of Efford et al. (2016) from various inputs, including fitted models.
kfn(object)
kfn(object)
object |
fitted secr model, numeric vector, matrix, dataframe |
kfn
simply computes , where
is the sigma parameter of a fitted halfnormal detection function and
is the corresponding density estimate. The factor of 1/100 adjusts for the units used in secr (sigma in metres; D in animals per hectare).
Input may be in any of these forms
vector with D and sigma in the first and second positions.
matrix with each row as in (1)
dataframe such as produced by predict.secr
with rows ‘D’ and ‘sigma’, and column ‘estimate’.
fitted secr model
a list of any of the above
Numeric vector with elements ‘D’, ‘sigma’ and ‘k’, or a matrix with these columns.
The index should not be taken too literally as a measure of overlap: it represents the overlap expected if activity centres are randomly distributed and if home ranges have bivariate normal utilisation. Thus it does not measure overlap due to social behaviour etc. except as that affects home range size.
The index may be estimated directly using the sigmak parameterization
(i.e., when sigmak appears in the model for secr.fit
).
This provides SE and confidence limits for sigmak (= ). However,
the directly estimated value of sigmak lacks the unit correction and is
therefore 100
the value from
kfn
.
Efford, M. G., Dawson, D. K., Jhala, Y. V. and Qureshi, Q. (2016) Density-dependent home-range size revealed by spatially explicit capture–recapture. Ecography 39, 676–688.
predict.secr
, secr.fit
, details
kfn(secrdemo.0) ## compare ## fitk <- secr.fit(captdata, model = sigmak~1, buffer = 100, trace = FALSE) ## predict(fitk)
kfn(secrdemo.0) ## compare ## fitk <- secr.fit(captdata, model = sigmak~1, buffer = 100, trace = FALSE) ## predict(fitk)
This function is a wrapper for secr.fit
that allows multiple models to be fitted.
list.secr.fit (..., constant = list(), prefix = "fit", names = NULL)
list.secr.fit (..., constant = list(), prefix = "fit", names = NULL)
... |
varying arguments of |
constant |
list of named arguments held constant |
prefix |
character prefix for automatic names |
names |
character names of output |
The ... argument may be one or several vectors of the same length that refer to a different named argument of secr.fit
. secr.fit
is called with the constant arguments plus the first value in each vector, then the second value, etc. The logic follows mapply
.
Each of the ... arguments may also be a named argument with a single value, although the compound values should be wrapped in list(), passed by name (in quotes), or placed in the 'constant' list to avoid misinterpretation. For example, the capthist argument of secr.fit
should be be wrapped in list() or " " if it is placed outside 'constant'.
'prefix' is used only if 'names' is not supplied.
An secrlist
of the successful model fits (see secr.fit
).
In the special case that ‘constant’ specifies ‘details’ with component LLonly = TRUE,
the list of values from secr.fit
without checking or modification.
This function replaces the previous function par.secr.fit
: since the introduction of multi-threading in secr 4.0 it is no longer efficient to use parallel worker processes.
secr.fit
,
AIC.secr
,
predict.secr
## Not run: # fit two detection models fits <- list.secr.fit (model = c(g0~1, g0~b), constant = list(captdata, trace = FALSE)) AIC(fits) # alternatively, fits <- list.secr.fit ('captdata', model = c(g0~1, g0~b), trace = FALSE) AIC(fits) # replacing par.derived and par.region.N: lapply(fits, derived) lapply(fits, region.N) ## End(Not run)
## Not run: # fit two detection models fits <- list.secr.fit (model = c(g0~1, g0~b), constant = list(captdata, trace = FALSE)) AIC(fits) # alternatively, fits <- list.secr.fit ('captdata', model = c(g0~1, g0~b), trace = FALSE) AIC(fits) # replacing par.derived and par.region.N: lapply(fits, derived) lapply(fits, region.N) ## End(Not run)
LLsurface is a generic function to calculate log likelihood over a grid of values of two coefficients (beta parameters) from a fitted model and optionally make an approximate contour plot of the log likelihood surface.
A method is provided for secr objects.
LLsurface(object, ...) ## S3 method for class 'secr' LLsurface(object, betapar = c("g0", "sigma"), xval = NULL, yval = NULL, centre = NULL, realscale = TRUE, plot = TRUE, plotfitted = TRUE, ncores = NULL, ...)
LLsurface(object, ...) ## S3 method for class 'secr' LLsurface(object, betapar = c("g0", "sigma"), xval = NULL, yval = NULL, centre = NULL, realscale = TRUE, plot = TRUE, plotfitted = TRUE, ncores = NULL, ...)
object |
fitted model, |
betapar |
character vector giving the names of two beta parameters |
xval |
vector of numeric values for x-dimension of grid |
yval |
vector of numeric values for y-dimension of grid |
centre |
vector of central values for all beta parameters |
realscale |
logical. If TRUE input and output of x and y is on the untransformed (inverse-link) scale. |
plot |
logical. If TRUE a contour plot is produced |
plotfitted |
logical. If TRUE the MLE from |
ncores |
integer number of threads for parallel processing |
... |
other arguments passed to |
centre
is set by default to the fitted values of the beta
parameters in object
. This has the effect of holding parameters
other than those in betapar
at their fitted values.
If xval
or yval
is not provided then 11 values are set at
equal spacing between 0.8 and 1.2 times the values in centre
(on
the ‘real’ scale if realscale
= TRUE and on the ‘beta’ scale
otherwise).
Contour plots may be customized by passing graphical parameters through the ... argument.
Setting ncores = NULL
uses the existing value from the environment variable
RCPP_PARALLEL_NUM_THREADS (see setNumThreads
).
A matrix of the log likelihood evaluated at each
grid point (rows x, columns y), invisibly if plot = TRUE
.
Failed evaluations return NA.
LLsurface
works for named ‘beta’ parameters rather than
‘real’ parameters. The default realscale = TRUE
only works for
beta parameters that share the name of the real parameter to which
they relate i.e. the beta parameter for the base level of the real
parameter. This is because link functions are defined for real
parameters not beta parameters.
The contours are approximate because they rely on interpolation. See Examples for a more reliable way to compare the likelihood at the MLE with nearby points on the surface.
## Not run: LLsurface(secrdemo.CL, xval = seq(0.16,0.40,0.02), yval = 25:35, nlevels = 20) ## now verify MLE ## click on MLE and apparent `peak' if (interactive()) { xy <- locator(2) LLsurface(secrdemo.CL, xval = xy$x, yval = xy$y, plot = FALSE) } ## End(Not run)
## Not run: LLsurface(secrdemo.CL, xval = seq(0.16,0.40,0.02), yval = 25:35, nlevels = 20) ## now verify MLE ## click on MLE and apparent `peak' if (interactive()) { xy <- locator(2) LLsurface(secrdemo.CL, xval = xy$x, yval = xy$y, plot = FALSE) } ## End(Not run)
Transform real values to the logit scale, and the inverse.
logit(x) invlogit(y)
logit(x) invlogit(y)
x |
vector of numeric values in (0,1) (possibly a probability) |
y |
vector of numeric values |
The logit transformation is defined as for
.
Numeric value on requested scale.
logit
is equivalent to qlogis
, and invlogit
is equivalent to plogis
(both R functions in the stats package).
logit
and invlogit
are used in secr because they are slightly more robust to bad input, and their names are more memorable!
logit(0.5) invlogit(logit(0.2))
logit(0.5) invlogit(logit(0.2))
Compute the constant multinomial component of the SECR log likelihood
logmultinom(capthist, grp = NULL)
logmultinom(capthist, grp = NULL)
capthist |
|
grp |
factor defining group membership, or a list (see Details) |
For a particular dataset and grouping, the multinomial coefficient is
a constant; it does not depend on the parameters and may be ignored
when maximizing the likelihood to obtain parameter
estimates. Nevertheless, the log likelihood reported by
secr.fit
includes this component unless the
detector type is ‘signal’, ‘polygon’, ‘polygonX’, ‘transect’ or
‘transectX’ (from 2.0.0).
If grp
is NULL then all animals are assumed to belong to one
group. Otherwise, the length of grp
should equal the number of rows of
capthist
.
grp
may also be any vector that can be coerced
to a factor. If capthist
is a multi-session capthist object
then grp
should be a list with one factor per session.
If capture histories are not assigned to groups the value is the logarithm of
where is the total number of
capture histories and
...
are the frequencies with
which each of the
unique capture histories were observed.
If capture histories are assigned to groups the value is the
logarithm of
where is the number of capture histories of
group
and
...
are the frequencies with
which each of the
unique capture histories were observed for
group
.
For multi-session data, the value is the sum of the single-session values. Both session structure and group structure therefore affect the value computed. Users will seldom need this function.
The numeric value of the log likelihood component.
Borchers, D. L. and Efford, M. G. (2008) Spatially explicit maximum likelihood methods for capture–recapture studies. Biometrics 64, 377–385.
Efford, M. G., Borchers D. L. and Byrom, A. E. (2009) Density estimation by spatially explicit capture–recapture: likelihood-based methods. In: D. L. Thompson, E. G. Cooch and M. J. Conroy (eds) Modeling Demographic Processes in Marked Populations. Springer. Pp. 255–269.
## no groups logmultinom(stoatCH)
## no groups logmultinom(stoatCH)
Compute likelihood ratio test to compare two fitted models, one nested within the other.
LR.test(model1, model2)
LR.test(model1, model2)
model1 |
fitted model |
model2 |
fitted model |
The fitted models must be of a class for which there is a logLik
method (e.g., ‘secr’ or ‘lm’). Check with methods("logLik")
.
The models must be nested (no check is performed - this is up to the user), but either model1 or model2 may be the more general model.
The models must also be compatible by the criteria of AICcompatible
.
The test statistic is twice the difference of the maximized likelihoods. It is compared to a chi-square distribution with df equal to the number of extra parameters in the more complex model.
Object of class ‘htest’, a list with components
statistic |
value the test statistic |
parameter |
degrees of freedom of the approximate chi-squared distribution of the test statistic |
p.value |
probability of test statistic assuming chi-square distribution |
method |
character string indicating the type of test performed |
data.name |
character string with names of models compared |
AICcompatible
,
AIC.secr
,
score.test
## two pre-fitted models AIC (secrdemo.0, secrdemo.b) LR.test (secrdemo.0, secrdemo.b)
## two pre-fitted models AIC (secrdemo.0, secrdemo.b) LR.test (secrdemo.0, secrdemo.b)
Form a capthist
object from a data frame of capture records and a traps
object.
make.capthist(captures, traps, fmt = c("trapID", "XY"), noccasions = NULL, covnames = NULL, bysession = TRUE, sortrows = TRUE, cutval = NULL, tol = 0.01, snapXY = FALSE, noncapt = "NONE", signalcovariates)
make.capthist(captures, traps, fmt = c("trapID", "XY"), noccasions = NULL, covnames = NULL, bysession = TRUE, sortrows = TRUE, cutval = NULL, tol = 0.01, snapXY = FALSE, noncapt = "NONE", signalcovariates)
captures |
dataframe of capture records in one of two possible formats (see Details) |
traps |
object of class |
fmt |
character string for capture format. |
noccasions |
number of occasions on which detectors were operated |
covnames |
character vector of names for individual covariate fields |
bysession |
logical, if true then ID are made unique by session |
sortrows |
logical, if true then rows are sorted in ascending order of animalID |
cutval |
numeric, threshold of signal strength for ‘signal’ detector type |
tol |
numeric, snap tolerance in metres |
snapXY |
logical; if TRUE then fmt = 'XY' uses nearest trap within tol for non-polygon detectors |
noncapt |
character value; animal ID used for ‘no captures’ |
signalcovariates |
character vector of field names from ‘captures’ |
make.capthist
is the most flexible way to prepare data for
secr.fit
. See read.capthist
for a more streamlined
way to read data from text files for common detector types. Each row of
the input data frame captures
represents a detection on one
occasion. The capture data frame may be formed from a text file with
read.table
.
Input formats are based on the Density software (Efford 2012; see also
secr-datainput.pdf). If fmt =
"XY"
the required fields are (session, ID, occasion, x, y) in that
order. If fmt = "trapID"
the required fields are (session, ID,
occasion, trap), where trap
is the numeric index of the relevant
detector in traps
. session
and ID
may be
character-, vector- or factor-valued; other required fields are
numeric. Fields are matched by position (column number), not by
name. Columns after the required fields are interpreted as individual
covariates that may be continuous (e.g., size) or categorical (e.g.,
age, sex).
If captures
has data from multiple sessions then traps
may
be either a list of traps
objects, one per session, or a single
traps
object that is assumed to apply throughout. Similarly,
noccasions
may be a vector specifying the number of occasions in
each session.
Covariates are assumed constant for each individual; the first
non-missing value is used. The length of covnames
should equal the
number of covariate fields in captures
.
bysession
takes effect when the same individual is detected in
two or more sessions: TRUE results in one capture history per session,
FALSE has the effect of generating a single capture history (this is not
appropriate for the models currently provided in secr).
Deaths are coded as negative values in the occasion field of
captures
. Occasions should be numbered 1, 2, ..., noccasions. By
default, the number of occasions is the maximum value of ‘occasion’ in
captures
.
Signal strengths may be provided in the fifth (fmt = trapID) or sixth (fmt = XY) columns. Detections with signal strength missing (NA) or below ‘cutval’ are discarded.
A session may result in no detections. In this case a null line is
included in captures
using the animal ID field given by
noncapt
, the maximum occasion number, and any trapID (e.g. "sess1
NONE 5 1" for a 5-occasion session) (or equivalently "sess1 NONE 5 10
10" for fmt = XY).
Nonspatial data (Session, AnimalID, Occasion and possibly individual covariates) may be entered by omitting the ‘traps’ argument or setting it to NULL.
An object of class capthist
(a matrix or array of
detection data with attributes for detector positions etc.). For
‘single’ and ‘multi’ detectors this is a matrix with one row per animal
and one column per occasion (dim(capthist)=c(nc,noccasions)); each
element is either zero (no detection) or a detector number (the row
number in traps
not the row name). For ‘proximity’
detectors capthist
is an array of values {-1, 0, 1} and
dim(capthist)=c(nc,noccasions,ntraps). The number of animals nc
is determined from the input, as is noccasions
if it is not specified.
traps
, covariates
and other data are retained as
attributes of capthist
.
Deaths during the experiment are represented as negative values in capthist
.
For ‘signal’ and ‘signalnoise’ detectors, the columns of captures
identified in signalcovariates
are saved along with signal
strength measurements in the attribute ‘signalframe’.
If the input has data from multiple sessions then the output is an
object of class c("capthist", "list") comprising a list of single-session
capthist
objects.
make.capthist
requires that the data for captures
and
traps
already exist as R objects. To read data from external
(text) files, first use read.table
and read.traps
, or try
read.capthist
for a one-step solution.
Prior to secr 4.4.0, occasional valid records for "multi" and "single" detectors were rejected as duplicates.
From secr 4.5.0, ‘snapXY’ works for transects as well as point detectors.
Efford, M. G. (2012) DENSITY 5.0: software for spatially explicit capture–recapture. Department of Mathematics and Statistics, University of Otago, Dunedin, New Zealand. https://www.otago.ac.nz/density/.
capthist
, traps
,
read.capthist
, secr.fit
,
sim.capthist
## peek at demonstration data head(captXY) head(trapXY) demotraps <- read.traps(data = trapXY) demoCHxy <- make.capthist (captXY, demotraps, fmt = "XY") demoCHxy ## print method for capthist plot(demoCHxy) ## plot method for capthist summary(demoCHxy) ## summary method for capthist ## To enter `count' data without manually repeating rows ## need a frequency vector f, length(f) == nrow(captXY) n <- nrow(captXY) f <- sample (1:5, size = n, prob = rep(0.2,5), replace = TRUE) ## repeat rows as required... captXY <- captXY[rep(1:n, f),] counttraps <- read.traps(data = trapXY, detector = "count") countCH <- make.capthist (captXY, counttraps, fmt = "XY")
## peek at demonstration data head(captXY) head(trapXY) demotraps <- read.traps(data = trapXY) demoCHxy <- make.capthist (captXY, demotraps, fmt = "XY") demoCHxy ## print method for capthist plot(demoCHxy) ## plot method for capthist summary(demoCHxy) ## summary method for capthist ## To enter `count' data without manually repeating rows ## need a frequency vector f, length(f) == nrow(captXY) n <- nrow(captXY) f <- sample (1:5, size = n, prob = rep(0.2,5), replace = TRUE) ## repeat rows as required... captXY <- captXY[rep(1:n, f),] counttraps <- read.traps(data = trapXY, detector = "count") countCH <- make.capthist (captXY, counttraps, fmt = "XY")
A lacework design comprises a square grid with detectors placed at regular distances along the grid lines (Efford unpubl.). This requires fewer detectors than uniform coverage at close spacing and is simpler than clustered designs, while providing good spatial coverage and protection from alignment bias (Efford 2019).
make.lacework(region, spacing = c(100, 20), times = NULL, origin = NULL, rotate = 0, radius = NULL, detector = "multi", keep.design = TRUE)
make.lacework(region, spacing = c(100, 20), times = NULL, origin = NULL, rotate = 0, radius = NULL, detector = "multi", keep.design = TRUE)
region |
dataframe or SpatialPolygonsDataFrame with coordinates of perimeter |
spacing |
numeric 2-vector with major (grid) and minor spacings, or minor spacing only |
times |
numeric ratio major:minor spacing if spacing length 1 |
origin |
numeric vector giving x- and y-cooordinates of fixed grid origin (origin is otherwise random) |
rotate |
numeric; number of degrees by which to rotate design clockwise about centroid of region bounding box |
radius |
numeric; detectors are dropped if they are further than this from a crossing |
detector |
character detector type – see |
keep.design |
logical; if TRUE then input argument values are retained |
It is tidy for the major spacing (spacing[1]
) to be a multiple of the minor spacing (spacing[2]
); precisely one detector is then placed at each grid intersection. This outcome may also be achieved by providing only the minor spacing in the spacing
argument and specifying an integer value for times
.
In general it is better not to specify origin
. Specifying both origin
and rotate
may result in incomplete coverage, as the desired grid is relative to the bounding box of the rotated region.
Set radius
< spacing[1]/2 to break lacework into multiple cross-shaped arrays centred on the intersections (crossing points) and truncated at radius
metres (assuming you follow advice and express all linear measurements in metres).
The number of detectors should not exceed 5000.
An secr traps object. The attribute ‘crossings’ is a 2-column matrix with the coordinates of the intersection points. If keep.design
is TRUE then the input argument values are retained in attribute ‘design’ (a list with first component function = 'make.lacework'
).
Efford, M. G. (2019) Non-circular home ranges and the estimation of population density. Ecology 100, e02580. https://esajournals.onlinelibrary.wiley.com/doi/10.1002/ecy.2580
trps <- make.lacework(possumarea, c(1000,100), rotate = 45, detector = 'proximity') plot(trps, gridspace = 1000) lines(possumarea) points(attr(trps, 'crossings'), pch = 16)
trps <- make.lacework(possumarea, c(1000,100), rotate = 45, detector = 'proximity') plot(trps, gridspace = 1000) lines(possumarea) points(attr(trps, 'crossings'), pch = 16)
Construct a habitat mask object for spatially explicit capture-recapture. A mask object is a set of points with optional attributes.
make.mask(traps, buffer = 100, spacing = NULL, nx = 64, ny = 64, type = c("traprect", "trapbuffer", "pdot", "polygon", "clusterrect", "clusterbuffer", "rectangular", "polybuffer"), poly = NULL, poly.habitat = TRUE, cell.overlap = c("centre","any","all"), keep.poly = TRUE, check.poly = TRUE, pdotmin = 0.001, random.origin = FALSE, ...)
make.mask(traps, buffer = 100, spacing = NULL, nx = 64, ny = 64, type = c("traprect", "trapbuffer", "pdot", "polygon", "clusterrect", "clusterbuffer", "rectangular", "polybuffer"), poly = NULL, poly.habitat = TRUE, cell.overlap = c("centre","any","all"), keep.poly = TRUE, check.poly = TRUE, pdotmin = 0.001, random.origin = FALSE, ...)
traps |
object of class |
buffer |
width of buffer in metres |
spacing |
spacing between grid points (metres) |
nx |
number of grid points in ‘x’ direction |
ny |
number of grid points in ‘y’ direction (type = ‘rectangular’) |
type |
character string for method |
poly |
bounding polygon to which mask should be clipped (see Details) |
poly.habitat |
logical for whether poly represents habitat or its inverse (non-habitat) |
cell.overlap |
character string for cell vertices used to determine overlap with ‘poly’ |
keep.poly |
logical; if TRUE any bounding polygon is saved as the attribute ‘polygon’ |
check.poly |
logical; if TRUE a warning is given for traps that lie outside a bounding polygon |
pdotmin |
minimum detection probability for inclusion in mask when |
random.origin |
logical; if TRUE the mask coordinates are jittered |
... |
additional arguments passed to |
The ‘traprect’ method constructs a grid of points in the rectangle
formed by adding a buffer strip to the minimum and maximum x-y
coordinates of the detectors in traps
. Both ‘trapbuffer’ and
‘pdot’ start with a ‘traprect’ mask and drop some points.
The ‘trapbuffer’ method restricts the grid to points within distance
buffer
of any detector.
The ‘pdot’ method restricts the grid to points for which the net
detection probability (see
pdot
) is at least pdotmin
. Additional parameters
are used by pdot
(detectpar, noccasions). Set these with the
... argument; otherwise make.mask
will silently use the
arbitrary defaults. pdot
is currently limited to a halfnormal
detection function.
The ‘clusterrect’ method constructs a grid of rectangular submasks
centred on ‘clusters’ of detectors generated with
trap.builder
(possibly indirectly by
make.systematic
). The ‘clusterbuffer’ method resembles
‘trapbuffer’, but is usually faster when traps are arranged in clusters
because it starts with a ‘clusterrect’ mask.
The ‘rectangular’ method constructs a simple rectangular mask with the given nx, ny and spacing.
The ‘polybuffer’ method constructs a mask by buffering around the polygon specified in the ‘poly’ argument. If that inherits from ‘SpatialPolygons’ or ‘sfc’ then the buffering is performed with sf::st_buffer. Otherwise, buffering is approximate, based on the distance to points on an initial discretized mask enclosed by ‘poly’ (points at half the current ‘spacing’).
If poly
is specified, points outside poly
are dropped (unless type = "polybuffer"). The default is to require only the centre to lie within poly
; use cell.overlap = "all"
to require all cell corners to lie within poly
, or cell.overlap = "any"
to accept cells with any corner in poly
. The ‘polygon’ method places points on a rectangular grid clipped to the
polygon (buffer
is not used). Thus ‘traprect’ is equivalent to
‘polygon’ when poly
is supplied. poly
may be either
a matrix or dataframe of two columns interpreted as x and y coordinates, or
an object from package ‘sf’ with polygon geometries, or
a SpatialPolygons or SpatialPolygonsDataFrame object as defined in the package ‘sp’, possibly imported by reading a shapefile.
If spacing
is not specified then it is determined by dividing the
range of the x coordinates (including any buffer) by nx
.
random.origin
shifts the origin of the mask by a uniform random displacement within a spacing x spacing grid cell, while ensuring that the mask also satisfies the buffer requirement. random.origin
is available only for ‘traprect’, ‘trapbuffer’, ‘polygon’, and ‘rectangular’ types, and spacing
must be specified.
An object of class mask
. When keep.poly = TRUE
,
poly
and poly.habitat
are saved as attributes of the
mask.
A warning is displayed if type = "pdot"
and the buffer is too small to
include all points with > pdotmin.
A habitat mask is needed to fit an SECR model and for some related
computations. The default mask settings in secr.fit
may be good
enough, but it is preferable to use make.mask
to construct a mask
in advance and to pass that mask as an argument to secr.fit
.
The function bufferContour
displays the extent of one or more
‘trapbuffer’ zones - i.e. the effect of buffering the detector array
with varying strip widths.
mask
, read.mask
, subset.mask
,
pdot
, bufferContour
, deleteMaskPoints
,
as.mask
temptrap <- make.grid(nx = 10, ny = 10, spacing = 30) ## default method: traprect tempmask <- make.mask(temptrap, spacing = 5) plot(tempmask) summary (tempmask) ## make irregular detector array by subsampling ## form mask by `trapbuffer' method temptrap <- subset (temptrap, sample(nrow(temptrap), size = 30)) tempmask <- make.mask (temptrap, spacing = 5, type = "trapbuffer") plot (tempmask) plot (temptrap, add = TRUE) ## Not run: ## form mask by "pdot" method temptrap <- make.grid(nx = 6, ny = 6) tempmask <- make.mask (temptrap, buffer = 150, type = "pdot", pdotmin = 0.0001, detectpar = list(g0 = 0.1, sigma = 30), noccasions = 4) plot (tempmask) plot (temptrap, add = TRUE) ## Using an ESRI polygon shapefile for clipping (shapefile ## polygons may include multiple islands and holes). library(sf) shpfilename <- system.file("extdata/possumarea.shp", package = "secr") possumarea <- st_read(shpfilename) possummask2 <- make.mask(traps(possumCH), spacing = 20, buffer = 250, type = "trapbuffer", poly = possumarea) par(mar = c(1,6,6,6), xpd = TRUE) plot (possummask2, ppoly = TRUE) plot(traps(possumCH), add = TRUE) par(mar = c(5,4,4,2) + 0.1, xpd = FALSE) ## if the polygon delineates non-habitat ... seaPossumMask <- make.mask(traps(possumCH), buffer = 1000, type = "traprect", poly = possumarea, poly.habitat = FALSE) plot(seaPossumMask) plot(traps(possumCH), add = TRUE) ## this mask is not useful! ## End(Not run)
temptrap <- make.grid(nx = 10, ny = 10, spacing = 30) ## default method: traprect tempmask <- make.mask(temptrap, spacing = 5) plot(tempmask) summary (tempmask) ## make irregular detector array by subsampling ## form mask by `trapbuffer' method temptrap <- subset (temptrap, sample(nrow(temptrap), size = 30)) tempmask <- make.mask (temptrap, spacing = 5, type = "trapbuffer") plot (tempmask) plot (temptrap, add = TRUE) ## Not run: ## form mask by "pdot" method temptrap <- make.grid(nx = 6, ny = 6) tempmask <- make.mask (temptrap, buffer = 150, type = "pdot", pdotmin = 0.0001, detectpar = list(g0 = 0.1, sigma = 30), noccasions = 4) plot (tempmask) plot (temptrap, add = TRUE) ## Using an ESRI polygon shapefile for clipping (shapefile ## polygons may include multiple islands and holes). library(sf) shpfilename <- system.file("extdata/possumarea.shp", package = "secr") possumarea <- st_read(shpfilename) possummask2 <- make.mask(traps(possumCH), spacing = 20, buffer = 250, type = "trapbuffer", poly = possumarea) par(mar = c(1,6,6,6), xpd = TRUE) plot (possummask2, ppoly = TRUE) plot(traps(possumCH), add = TRUE) par(mar = c(5,4,4,2) + 0.1, xpd = FALSE) ## if the polygon delineates non-habitat ... seaPossumMask <- make.mask(traps(possumCH), buffer = 1000, type = "traprect", poly = possumarea, poly.habitat = FALSE) plot(seaPossumMask) plot(traps(possumCH), add = TRUE) ## this mask is not useful! ## End(Not run)
A spatial coverage design places one cluster of detectors in each compact subregion of a region of interest. Equal subregions are determined by k-means clustering of pixels (Walvoort et al. 2010).
make.spcosa(n, cluster, region, rotation = 0, randomize = FALSE, maxtries = 100, keep.mask = FALSE, ...)
make.spcosa(n, cluster, region, rotation = 0, randomize = FALSE, maxtries = 100, keep.mask = FALSE, ...)
n |
integer number of subregions |
cluster |
traps object defining a cluster of detectors |
region |
boundary of region of interest |
rotation |
numeric angular rotation of each cluster (negative for random) |
randomize |
logical; if TRUE then cluster is located at random within subregion |
maxtries |
integer maximum attempts to find random location |
keep.mask |
logical; if TRUE then a habitat mask with covariate ‘stratum’ |
... |
other arguments passed to |
The region may be specified in any form acceptable as the ‘poly’ argument of make.mask
(see also boundarytoSF).
The ... argument determines the coarseness of the discretization used to define the subregions, via the ‘nx’ or ‘spacing’ arguments of make.mask
.
By default (randomize = FALSE) clusters are centred at subregion centroids. Otherwise (randomize = TRUE) clusters are centred in a randomly selected cell of each subregion, subject to the constraint that all detectors fall within the subregion. An error results if no cluster meeting the constraint is found in ‘maxtries’ attempts.
Slightly different partitions of ‘poly’ are generated depending on the value of the random seed, so for consistency this should be first set with set.seed
.
The argument ‘rotation’ is applied separately to each cluster, as in trap.builder
and unlike the argument ‘rotate’ of make.systematic
.
A traps object with n x nrow(cluster) detectors.
Walvoort, D., Brus, D., and de Gruijter, J. (2010) An R package for spatial coverage sampling and random sampling from compact geographical strata by k-means. Computers & Geosciences 36:1261–1267.
make.mask
,
trap.builder
,
boundarytoSF
,
traps
# preliminaries polygonfile <- system.file("extdata/possumarea.txt", package = "secr") poly <- read.table(polygonfile, header = TRUE) subgrid <- make.grid(3,3, spacing = 80) set.seed(123) # nx and keep.mask refer to the discretized region of interest tr <- make.spcosa(n = 5, subgrid, poly, nx = 32, randomize = TRUE, keep.mask = TRUE) plot(attr(tr,'mask'), dots = FALSE, cov = 'stratum', legend = FALSE) plot(tr, add = TRUE) lines(poly)
# preliminaries polygonfile <- system.file("extdata/possumarea.txt", package = "secr") poly <- read.table(polygonfile, header = TRUE) subgrid <- make.grid(3,3, spacing = 80) set.seed(123) # nx and keep.mask refer to the discretized region of interest tr <- make.spcosa(n = 5, subgrid, poly, nx = 32, randomize = TRUE, keep.mask = TRUE) plot(attr(tr,'mask'), dots = FALSE, cov = 'stratum', legend = FALSE) plot(tr, add = TRUE) lines(poly)
A rectangular grid of clusters within a polygonal region.
make.systematic(n, cluster, region, spacing = NULL, origin = NULL, originoffset = c(0,0), chequerboard = c('all','black','white'), order = c('x', 'y', 'xb', 'yb'), rotate = 0, centrexy = NULL, keep.design = TRUE, ...)
make.systematic(n, cluster, region, spacing = NULL, origin = NULL, originoffset = c(0,0), chequerboard = c('all','black','white'), order = c('x', 'y', 'xb', 'yb'), rotate = 0, centrexy = NULL, keep.design = TRUE, ...)
n |
integer approximate number of clusters (see Details) |
cluster |
traps object defining a single cluster |
region |
dataframe or SpatialPolygonsDataFrame with coordinates of perimeter |
spacing |
scalar distance between cluster centres |
origin |
vector giving x- and y-cooordinates of fixed grid origin (origin is otherwise random) |
originoffset |
numeric; 2-vector (x,y offsets); see Details |
chequerboard |
logical; if not ‘all’ then alternate clusters are omitted |
order |
character; sort order for clusters (see Details) |
rotate |
numeric; number of degrees by which to rotate entire design clockwise about centroid of region bounding box |
centrexy |
numeric; 2-vector for centre of rotation, if any |
keep.design |
logical; if TRUE then input argument values are retained |
... |
arguments passed to |
region
may be any shape.
region
may be one of the spatial classes described
in boundarytoSF
. Otherwise,
region
should be a dataframe with columns ‘x’ and ‘y’.
spacing
may be a vector with separate values for spacing in x-
and y- directions. If spacing
is provided then n
is ignored.
If n
is a scalar, the spacing of clusters is determined from
the area of the bounding box of region
divided by the requested
number of clusters (this does not necessarily result in exactly n
clusters). If n
is a vector of two integers these are taken to be
the number of columns and the number of rows.
After preparing a frame of cluster centres, make.systematic
calls trap.builder
with method = ‘all’; ... allows the
arguments ‘rotation’, ‘edgemethod’, ‘plt’, and ‘detector’ to be
passed. Setting the trap.builder
arguments frame
,
method
, and samplefactor
has no effect.
Note the distinction between argument rotate
and the trap.builder
argument rotation
that is applied separately to each cluster.
If origin
is not specified then a random uniform origin is chosen within a box (width = spacing) placed with its bottom left corner at the bottom left of the bounding box of region
, shifted by originoffset
. Before version 3.1.8 the behaviour of make.systematic
was equivalent to originoffset = c(wx,wy)
where wx,wy
are the cluster half widths.
chequerboard = "black"
retains black ‘squares’ and chequerboard = "white"
retains white ‘squares’, where the lower left cluster in the candidate rectangle of cluster origins is black, as on a chess board. The effect is the same as increasing spacing by sqrt(2) and rotating through 45 degrees.
order
determines the ordering of clusters in the resulting traps object. The options are a subset of those for ID
argument of make.grid
:
Option | Sort order |
x | column-dominant |
y | row-dominant |
xb | column-dominant boustrophedonical (alternate columns reversed) |
yb | row-dominant boustrophedonical (alternate rows reversed) |
rotate
rotates the array about the given centre (default is centroid of the bounding box of region
).
A single-session ‘traps’ object.
From 3.2.0 these additional attributes are set –
origin | coordinates of grid origin |
centres | coordinates of true cluster centres (cf cluster.centres ) |
originbox | vertices of rectangular spatial sampling frame for random origin |
From 4.2.0 if keep.design
is TRUE then the input argument values are retained in attribute ‘design’ (a list with first component function = 'make.systematic'
).
Do not confuse with the simpler function make.grid
,
which places single detectors in a rectangular array.
trap.builder
,
make.lacework
,
cluster.centres
mini <- make.grid(nx = 2, ny = 2, spacing = 100) region <- cbind(x=c(0,2000,2000,0), y=c(0,0,2000,2000)) temp <- make.systematic(25, mini, region, plt = TRUE) temp <- make.systematic(c(6, 6), mini, region, plt = TRUE, rotation = -1) ## Example using shapefile "possumarea.shp" in ## "extdata" folder. By default, each cluster is ## a single multi-catch detector ## Not run: library(sf) shpfilename <- system.file("extdata/possumarea.shp", package = "secr") possumarea <- st_read(shpfilename) possumgrid <- make.systematic(spacing = 100, region = possumarea, plt = TRUE) ## or with 2 x 2 clusters possumgrid2 <- make.systematic(spacing = 300, cluster = make.grid(nx = 2, ny = 2, spacing = 100), region = possumarea, plt = TRUE, edgemethod = "allinside") ## label clusters text(cluster.centres(possumgrid2), levels(clusterID (possumgrid2)), cex=0.7) ## If you have GPSBabel installed and on the Path ## then coordinates can be projected and uploaded ## to a GPS with `writeGPS', which also requires the ## package `proj4'. Defaults are for a Garmin GPS ## connected by USB. if (interactive()) { writeGPS(possumgrid, proj = "+proj=nzmg") } ## End(Not run)
mini <- make.grid(nx = 2, ny = 2, spacing = 100) region <- cbind(x=c(0,2000,2000,0), y=c(0,0,2000,2000)) temp <- make.systematic(25, mini, region, plt = TRUE) temp <- make.systematic(c(6, 6), mini, region, plt = TRUE, rotation = -1) ## Example using shapefile "possumarea.shp" in ## "extdata" folder. By default, each cluster is ## a single multi-catch detector ## Not run: library(sf) shpfilename <- system.file("extdata/possumarea.shp", package = "secr") possumarea <- st_read(shpfilename) possumgrid <- make.systematic(spacing = 100, region = possumarea, plt = TRUE) ## or with 2 x 2 clusters possumgrid2 <- make.systematic(spacing = 300, cluster = make.grid(nx = 2, ny = 2, spacing = 100), region = possumarea, plt = TRUE, edgemethod = "allinside") ## label clusters text(cluster.centres(possumgrid2), levels(clusterID (possumgrid2)), cex=0.7) ## If you have GPSBabel installed and on the Path ## then coordinates can be projected and uploaded ## to a GPS with `writeGPS', which also requires the ## package `proj4'. Defaults are for a Garmin GPS ## connected by USB. if (interactive()) { writeGPS(possumgrid, proj = "+proj=nzmg") } ## End(Not run)
Construct a rectangular array of detectors (trapping grid) or a circle of detectors or a polygonal search area.
make.grid(nx = 6, ny = 6, spacex = 20, spacey = spacex, spacing = NULL, detector = "multi", originxy = c(0,0), hollow = FALSE, ID = "alphay", leadingzero = TRUE, markocc = NULL) make.circle (n = 20, radius = 100, spacing = NULL, detector = "multi", originxy = c(0,0), IDclockwise = TRUE, leadingzero = TRUE, markocc = NULL) make.poly (polylist = NULL, x = c(-50,-50,50,50), y = c(-50,50,50,-50), exclusive = FALSE, verify = TRUE) make.transect (transectlist = NULL, x = c(-50,-50,50,50), y = c(-50,50,50,-50), exclusive = FALSE) make.telemetry (xy = c(0,0))
make.grid(nx = 6, ny = 6, spacex = 20, spacey = spacex, spacing = NULL, detector = "multi", originxy = c(0,0), hollow = FALSE, ID = "alphay", leadingzero = TRUE, markocc = NULL) make.circle (n = 20, radius = 100, spacing = NULL, detector = "multi", originxy = c(0,0), IDclockwise = TRUE, leadingzero = TRUE, markocc = NULL) make.poly (polylist = NULL, x = c(-50,-50,50,50), y = c(-50,50,50,-50), exclusive = FALSE, verify = TRUE) make.transect (transectlist = NULL, x = c(-50,-50,50,50), y = c(-50,50,50,-50), exclusive = FALSE) make.telemetry (xy = c(0,0))
nx |
number of columns of detectors |
ny |
number of rows of detectors |
spacex |
distance between detectors in ‘x’ direction (nominally in metres) |
spacey |
distance between detectors in ‘y’ direction (nominally in metres) |
spacing |
distance between detectors (x and y directions) |
detector |
character value for detector type - "single", "multi" etc. |
originxy |
vector origin for x-y coordinates |
hollow |
logical for hollow grid |
ID |
character string to control row names |
leadingzero |
logical; if TRUE numeric rownames are padded with leading zeros |
markocc |
integer vector of marking or sighting codes; see |
n |
number of detectors |
radius |
radius of circle (nominally in metres) |
IDclockwise |
logical for numbering of detectors |
polylist |
list of dataframes with coordinates for polygons |
transectlist |
list of dataframes with coordinates for transects |
x |
x coordinates of vertices |
y |
y coordinates of vertices |
exclusive |
logical; if TRUE animal can be detected only once per occasion |
verify |
logical if TRUE then the resulting traps object is
checked with |
xy |
vector with coordinates of arbitrary point (e.g., centroid of fixes) |
make.grid
generates coordinates for nx.ny
traps at
separations spacex
and spacey
. If spacing
is
specified it replaces both spacex
and spacey
. The
bottom-left (southwest) corner is at originxy
. For a hollow grid,
only detectors on the perimeter are retained. By default, identifiers
are constructed from a letter code for grid rows and an integer value
for grid columns ("A1", "A2",...). ‘Hollow’ grids are always numbered
clockwise in sequence from the bottom-left corner. Other values of
ID
have the following effects:
ID | Effect |
numx | column-dominant numeric sequence |
numy | row-dominant numeric sequence |
numxb | column-dominant boustrophedonical numeric sequence (try it!) |
numyb | row-dominant boustrophedonical numeric sequence |
alphax | column-dominant alphanumeric |
alphay | row-dominant alphanumeric |
xy | combine column (x) and row(y) numbers |
‘xy’ adds leading zeros as needed to give a string of constant length with no blanks.
make.circle
generates coordinates for n traps in a circle centred
on originxy
. If spacing
is specified then it overrides the
radius
setting; the radius is adjusted to provide the requested
straightline distance between adjacent detectors. Traps are numbered
from the trap due east of the origin, either clockwise or anticlockwise
as set by IDclockwise
.
Polygon vertices may be specified with x
and y
in the case
of a single polygon, or as polylist
for one or more polygons. Each
component of polylist
is a dataframe with columns ‘x’ and ‘y’.
polylist
takes precedence. make.poly
automatically closes
the polygon by repeating the first vertex if the first and last vertices
differ.
Transects are defined by a sequence of vertices as for polygons, except that they are not closed.
make.telemetry
builds a simple traps object for the 'telemetry' detector type. The attribute 'telemetrytype' is set to "independent".
Specialised functions for arrays using a triangular grid are described
separately (make.tri
, clip.hex
).
An object of class traps
comprising a data frame of x- and
y-coordinates, the detector type ("single", "multi", or "proximity" etc.),
and possibly other attributes.
Several methods are provided for manipulating detector arrays - see traps
.
Efford, M. G. (2012) DENSITY 5.0: software for spatially explicit capture–recapture. Department of Mathematics and Statistics, University of Otago, Dunedin, New Zealand. https://www.otago.ac.nz/density/.
Efford, M. G., Borchers D. L. and Byrom, A. E. (2009) Density estimation by spatially explicit capture-recapture: likelihood-based methods. In: D. L. Thomson, E. G. Cooch and M. J. Conroy (eds) Modeling Demographic Processes in Marked Populations. Springer, New York. Pp. 255–269.
read.traps
,detector
,
trap.builder
,make.systematic
,
print.traps
, plot.traps
,
traps
, make.tri
,
addTelemetry
demo.traps <- make.grid() plot(demo.traps) ## compare numbering schemes par (mfrow = c(2,4), mar = c(1,1,1,1), xpd = TRUE) for (id in c("numx", "numy", "alphax", "alphay", "numxb", "numyb")) { temptrap <- make.grid(nx = 7, ny = 5, ID = id) plot (temptrap, border = 10, label = TRUE, offset = 7, gridl = FALSE) } temptrap <- make.grid(nx = 7, ny = 5, hollow = TRUE) plot (temptrap, border = 10, label = TRUE, gridl = FALSE) plot(make.circle(n = 20, spacing = 30), label = TRUE, offset = 9) summary(make.circle(n = 20, spacing = 30)) ## jitter locations randomly within grid square ## and plot over `mask' # see also ?gridCells tr0 <- tr <- make.grid(nx = 7, ny = 7, spacing = 30) tr[] <- jitter(unlist(tr), amount = spacing(tr)/2) plot(as.mask(tr0), dots = FALSE, mesh = 'white') plot(tr, add = TRUE)
demo.traps <- make.grid() plot(demo.traps) ## compare numbering schemes par (mfrow = c(2,4), mar = c(1,1,1,1), xpd = TRUE) for (id in c("numx", "numy", "alphax", "alphay", "numxb", "numyb")) { temptrap <- make.grid(nx = 7, ny = 5, ID = id) plot (temptrap, border = 10, label = TRUE, offset = 7, gridl = FALSE) } temptrap <- make.grid(nx = 7, ny = 5, hollow = TRUE) plot (temptrap, border = 10, label = TRUE, gridl = FALSE) plot(make.circle(n = 20, spacing = 30), label = TRUE, offset = 9) summary(make.circle(n = 20, spacing = 30)) ## jitter locations randomly within grid square ## and plot over `mask' # see also ?gridCells tr0 <- tr <- make.grid(nx = 7, ny = 7, spacing = 30) tr[] <- jitter(unlist(tr), amount = spacing(tr)/2) plot(as.mask(tr0), dots = FALSE, mesh = 'white') plot(tr, add = TRUE)
Construct an array of detectors on a triangular grid and optionally select a hexagonal subset of detectors.
make.tri (nx = 10, ny = 12, spacing = 20, detector = "multi", originxy = c(0,0)) clip.hex (traps, side = 20, centre = c(50, 60*cos(pi/6)), fuzz = 1e-3, ID = "num", ...)
make.tri (nx = 10, ny = 12, spacing = 20, detector = "multi", originxy = c(0,0)) clip.hex (traps, side = 20, centre = c(50, 60*cos(pi/6)), fuzz = 1e-3, ID = "num", ...)
nx |
number of columns of detectors |
ny |
number of rows of detectors |
spacing |
distance between detectors (x and y directions) |
detector |
character value for detector type - "single", "multi" etc. |
originxy |
vector origin for x-y coordinates |
traps |
traps object |
side |
length of hexagon side |
centre |
x-y coordinates of hexagon centre |
fuzz |
floating point fuzz value |
ID |
character string to control row names |
... |
other parameters passed to subset.traps (not used) |
make.tri
generates coordinates for nx.ny
traps at
separations spacing
. The bottom-left (southwest) corner is at
originxy
. Identifiers are numeric. See make.grid
for further explanation.
clip.hex
clips a grid of detectors, retaining only
those within a bounding hexagon. Detectors are re-labelled according to
ID
as follows:
ID | Effect |
NULL | no change |
num | numeric sequence |
alpha | letter for`shell'; number within shell |
An object of class traps
comprising a data frame of x- and
y-coordinates, the detector type ("single", "multi", or "proximity" etc.),
and possibly other attributes.
Several methods are provided for manipulating detector arrays - see traps
.
tri.grid <- make.tri(spacing = 10) plot(tri.grid, border = 5) hex <- clip.hex(tri.grid, side = 30, ID = "alpha") plot (hex, add = TRUE, detpar = list(pch = 16, cex = 1.4), label = TRUE, offset = 2.5 )
tri.grid <- make.tri(spacing = 10) plot(tri.grid, border = 5) hex <- clip.hex(tri.grid, side = 30, ID = "alpha") plot (hex, add = TRUE, detpar = list(pch = 16, cex = 1.4), label = TRUE, offset = 2.5 )
makeStart()
wraps the code previously internal to secr.fit()
.
It will not usually be called directly.
makeStart(start = NULL, parindx, capthist, mask, detectfn, link, details = NULL, fixed = NULL, CL = FALSE, anypoly = FALSE, anytrans = FALSE, alltelem = FALSE, sighting = FALSE)
makeStart(start = NULL, parindx, capthist, mask, detectfn, link, details = NULL, fixed = NULL, CL = FALSE, anypoly = FALSE, anytrans = FALSE, alltelem = FALSE, sighting = FALSE)
start |
optional starting values as described for |
parindx |
list with one component per real parameter giving corresponding indices of coefficients (beta parameters) |
capthist |
|
mask |
|
detectfn |
integer code or character string for shape of detection function 0 = halfnormal, 1 = hazard rate etc. – see detectfn |
link |
list of link function names ("log", "logit", "identity") for each real parameter |
details |
list of additional arguments (see |
fixed |
list with optional components corresponding to real parameters giving the scalar value to which the parameter is to be fixed |
CL |
logical, if true then model does not include density |
anypoly |
logical, TRUE if any polygon detectors |
anytrans |
logical, TRUE if any transect detectors |
alltelem |
logical, TRUE if any telemetry detectors |
sighting |
logical, TRUE if any sighting data |
If ‘start’ is a previously fitted model then only ‘start’ and ‘parindx’ are required.
Numeric vector with one value for each coefficient (beta parameter) in model.
makeStart(secrdemo.0, list(D = 1, g0 = 2:3, sigma = 4))
makeStart(secrdemo.0, list(D = 1, g0 = 2:3, sigma = 4))
Encapsulate a habitat mask for spatially explicit capture–recapture. See also secr-habitatmasks.pdf.
A habitat mask serves four main purposes in spatially explicit
capture–recapture. Firstly, it defines an outer limit to the area of
integration; habitat beyond the mask may be occupied, but animals there
should have negligible chance of being detected (see pdot
and below). Secondly, it distinguishes sites in the vicinity of the
detector array that are ‘habitat’ (i.e. have the potential to be
occupied) from ‘non-habitat’. Thirdly, it discretizes continuous habitat
as a list of points. Each point is notionally associated with a cell
(pixel) of uniform density. Discretization allows the SECR likelihood to
be evaluated by summing over grid cells. Fourthly, the x-y coordinates
of the mask and any habitat covariates may be used to build spatial
models of density. For example, a continuous or categorical habitat
covariate ‘cover’ measured at each point on the mask might be used in a
formula for density such as D cover.
In relation to the first purpose, the definition of ‘negligible’ is
fluid. Any probability less than 0.001 seems OK in the sense of not
causing noticeable bias in density estimates, but this depends on the
shape of the detection function (fat-tailed functions such as 'hazard
rate' are problematic). New tools for evaluating masks appeared in
secr 1.5 (mask.check
, esaPlot
), and
suggest.buffer
automates selection of a buffer width.
Mask points are stored in a data frame with columns ‘x’ and ‘y’. The number of rows equals the number of points.
Possible mask attributes
Attribute | Description |
type | `traprect', `trapbuffer', `pdot', `polygon', `clusterrect', `clusterbuffer' (see make.mask) or `user' |
polygon | vertices of polygon defining habitat boundary, for type = `polygon' |
pdotmin | threshold of p.(X) for type = `pdot' |
covariates | dataframe of site-specific covariates |
meanSD | data frame with centroid (mean and SD) of x and y coordinates |
area | area (ha) of the grid cell associated with each point |
spacing | nominal spacing (metres) between adjacent points |
boundingbox | data frame of 4 rows, the vertices of the bounding box of all grid cells in the mask |
Attributes other than covariates
are generated automatically by
make.mask
. Type ‘user’ refers to masks input from a text file
with read.mask
.
A virtual S4 class ‘mask’ is defined to allow the definition of a method
for the generic function raster
from the raster package.
A habitat mask is needed by secr.fit
, but one will be
generated automatically if none is provided. You should be aware of this
and check that the default settings (e.g. buffer
) are
appropriate.
make.mask
, read.mask
,
mask.check
, esaPlot
,
suggest.buffer
, secr.fit
mask.check
evaluates the effect of varying buffer width and
mask spacing on either the likelihood or density estimates from
secr.fit().
mask.check(object, buffers = NULL, spacings = NULL, poly = NULL, LLonly = TRUE, realpar = NULL, session = 1, file = NULL, drop = "", tracelevel = 0, ...)
mask.check(object, buffers = NULL, spacings = NULL, poly = NULL, LLonly = TRUE, realpar = NULL, session = 1, file = NULL, drop = "", tracelevel = 0, ...)
object |
object of class ‘capthist’ or ‘secr’ |
buffers |
vector of buffer widths |
spacings |
vector of mask spacings |
poly |
matrix of two columns, the x- and y-coordinates of a bounding polygon (optional) |
LLonly |
logical; if TRUE then only the log likelihood is computed |
realpar |
list of parameter values |
session |
vector of session indices (used if |
file |
name of output file (optional) |
drop |
character vector: names of fitted secr object to omit |
tracelevel |
integer for level of detail in reporting (0,1,2) |
... |
other arguments passed to secr.fit |
Masks of varying buffer width and spacing are constructed with the
‘trapbuffer’ method in make.mask
, using the detector locations
(‘traps’) from either a capthist object or a previous execution of
secr.fit
. Default values are provided for buffers
and
spacings
if object
is of class ‘secr’ (respectively c(1,
1.5, 2) and c(1, 0.75, 0.5) times the values in the existing
mask). The default for buffers
will not work if a detector is
on the mask boundary, as the inferred buffer is then 0.
Variation in the mask may be assessed for its effect on –
the log-likelihood evaluated for given values of the
parameters (LLonly = TRUE
)
estimates from maximizing the likelihood with each mask
(LLonly = FALSE
)
realpar
should be a list with one named component for each real
parameter (see Examples). It is relevant only if LLonly =
TRUE
. realpar
may be omitted if object
is of class
‘secr’; parameter values are then extracted from object
.
session
should be an integer or character vector suitable for
indexing sessions in object
, or in object$capthist
if
object
is a fitted model. Each session is considered
separately; a model formula that refers to session or uses session
covariates will cause an error.
If file
is specified then detailed
results (including each model fit when LLonly = FALSE
) are
saved to an external .RData file. Loading this file creates or
overwrites object(s) in the workspace: mask.check.output
if
LLonly = TRUE
, otherwise mask.check.output
and
mask.check.fit
. For multiple sessions these are replaced by
lists with one component per session (mask.check.outputs
and
mask.check.fits
). The drop
argument is passed to
trim
and applied to each fitted model; use it to save
space, at the risk of limiting further computation on the fitted
models.
tracelevel>0
causes more verbose reporting of progress during
execution.
The ... argument may be used to override existing settings in
object
- for example, a conditional likelihood fit (CL =
T
) may be selected even if the original model was fitted by
maximizing the full likelihood.
Array of log-likelihoods (LLonly = TRUE
) or estimates
(LLonly = FALSE
) for each combination of buffers
and
spacings
. The array has 3 dimensions if LLonly = FALSE
and both buffers
and spacings
have multiple levels;
otherwise it collapses to a matrix. Rows generally represent
buffers
, but rows represent spacings
if a single buffer
is specified.
mask.check()
may fail if object
is a fitted ‘secr’ model
and a data object named in the original call of secr.fit()
(i.e. object$call
) is no longer in the working environment
(secr.fit
arguments capthist, mask, verify & trace are
exempt). Fix by any of (1) applying mask.check
directly to the
‘capthist’ object, specifying other arguments (buffers
,
spacings
, realpar
) as needed, (2) re-fitting the model
and running mask.check
in the same environment, (3) specifying
the offending argument(s) in ..., or (4) re-creating the required
data objects(s) in the working environment, possibly from saved inputs
in object
(e.g., mytimecov <- myfit$timecov
).
When LLonly = TRUE
the functionality of mask.check
resembles that of the ‘Tools | ML SECR log likelihood’ menu option in
Density 5. The help page in Density 5 for ML SECR 2-D integration (see
index) may be helpful.
Warning messages from secr.fit
are suppressed. ‘capthist’
data provided via the object
argument are checked with
verify.capthist
if tracelevel > 0
.
The likelihood-only method is fast, but not definitive. It is
reasonable to aim for stability in the third decimal place of the log
likelihood. Slight additional variation in the likelihood may cause
little change in the estimates; the only way to be sure is to check
these by setting LLonly = FALSE
.
The performance of a mask depends on the detection function; be sure
to set the detectfn
argument appropriately. The hazard rate
function has a fat tail that can be problematic.
When provided with an ‘secr’ object
, mask.check
constructs a default vector of buffer widths as multiples of the
buffer used in object
even though that value is not saved
explicitly. For this trick, detector locations in
traps(object$capthist)
are compared to the bounding box of
object$mask
; the base level of buffer width is the maximum
possible within the bounding box.
Borchers, D. L. and Efford, M. G. (2008) Spatially explicit maximum likelihood methods for capture–recapture studies. Biometrics 64, 377–385.
Efford, M. G. (2012) DENSITY 5.0: software for spatially explicit capture–recapture. Department of Mathematics and Statistics, University of Otago, Dunedin, New Zealand. https://www.otago.ac.nz/density/.
## Not run: ## from a capthist object, specifying almost everything mask.check (possumCH, spacings = c(20, 30), buffers =c(200, 300), realpar = list(g0 = 0.2, sigma = 50), CL = TRUE) ## from a fitted model, using defaults mask.check (stoat.model.HN) ## LL did not change with varying buffer (rows) or spacing (cols): ## 78.125 58.59375 39.0625 ## 1000 -144.0015 -144.0015 -144.0015 ## 1500 -144.0017 -144.0017 -144.0017 ## 2000 -144.0017 -144.0017 -144.0017 ## fit new models for each combination of buffer & spacing, ## and save fitted models to a file mask.check (stoat.model.HN, buffers = 1500, spacings = c(40,60,80), LLonly = FALSE, file = "test", CL = TRUE) ## look in more detail at the preceding fits ## restores objects `mask.check.output' and `mask.check.fit' load("test.RData") lapply(mask.check.fit, predict) lapply(mask.check.fit, derived) ## multi-session data mask.check(ovenbird.model.1, session = c("2005","2009")) ## clipping mask txtfilename <- system.file("extdata/possumarea.txt", package = "secr") possumarea <- read.table(txtfilename, header = TRUE) mask.check (possum.model.0, spacings = c(20, 30), buffers = c(200, 300), poly = possumarea, LLonly = FALSE, file = "temp", CL = TRUE) ## review fitted models load ("temp.RData") par(mfrow = c(2,2), mar = c(1,4,4,4)) for (i in 1:4) { plot(traps(mask.check.fit[[i]]$capthist), border = 300, gridlines = FALSE) plot(mask.check.fit[[i]]$mask, add = TRUE) lines(possumarea) text ( 2698618, 6078427, names(mask.check.fit)[i]) box() } par(mfrow = c(1,1), mar = c(5,4,4,2) + 0.1) ## defaults ## End(Not run)
## Not run: ## from a capthist object, specifying almost everything mask.check (possumCH, spacings = c(20, 30), buffers =c(200, 300), realpar = list(g0 = 0.2, sigma = 50), CL = TRUE) ## from a fitted model, using defaults mask.check (stoat.model.HN) ## LL did not change with varying buffer (rows) or spacing (cols): ## 78.125 58.59375 39.0625 ## 1000 -144.0015 -144.0015 -144.0015 ## 1500 -144.0017 -144.0017 -144.0017 ## 2000 -144.0017 -144.0017 -144.0017 ## fit new models for each combination of buffer & spacing, ## and save fitted models to a file mask.check (stoat.model.HN, buffers = 1500, spacings = c(40,60,80), LLonly = FALSE, file = "test", CL = TRUE) ## look in more detail at the preceding fits ## restores objects `mask.check.output' and `mask.check.fit' load("test.RData") lapply(mask.check.fit, predict) lapply(mask.check.fit, derived) ## multi-session data mask.check(ovenbird.model.1, session = c("2005","2009")) ## clipping mask txtfilename <- system.file("extdata/possumarea.txt", package = "secr") possumarea <- read.table(txtfilename, header = TRUE) mask.check (possum.model.0, spacings = c(20, 30), buffers = c(200, 300), poly = possumarea, LLonly = FALSE, file = "temp", CL = TRUE) ## review fitted models load ("temp.RData") par(mfrow = c(2,2), mar = c(1,4,4,4)) for (i in 1:4) { plot(traps(mask.check.fit[[i]]$capthist), border = 300, gridlines = FALSE) plot(mask.check.fit[[i]]$mask, add = TRUE) lines(possumarea) text ( 2698618, 6078427, names(mask.check.fit)[i]) box() } par(mfrow = c(1,1), mar = c(5,4,4,2) + 0.1) ## defaults ## End(Not run)
MCgof
implements and extends the Monte Carlo resampling method of Choo et al. (2024) to emulate Bayesian posterior predictive checks (Gelman et al. 1996, Royle et al. 2014). Initial results suggest the approach is more informative than the deviance-based test proposed by Borchers and Efford (2008) and implemented in secr.test
.
However, the tests have limited power.
MCgof
is under development. The structure of the output may change and
bugs may be found. See Warning below for exclusions.
## S3 method for class 'secr' MCgof(object, nsim = 100, statfn = NULL, testfn = NULL, seed = NULL, ncores = 1, clustertype = c("PSOCK", "FORK"), usefxi = TRUE, useMVN = TRUE, Ndist = NULL, quiet = FALSE, debug = FALSE, ...) ## S3 method for class 'secrlist' MCgof(object, nsim = 100, statfn = NULL, testfn = NULL, seed = NULL, ncores = 1, clustertype = c("PSOCK", "FORK"), usefxi = TRUE, useMVN = TRUE, Ndist = NULL, quiet = FALSE, debug = FALSE, ...)
## S3 method for class 'secr' MCgof(object, nsim = 100, statfn = NULL, testfn = NULL, seed = NULL, ncores = 1, clustertype = c("PSOCK", "FORK"), usefxi = TRUE, useMVN = TRUE, Ndist = NULL, quiet = FALSE, debug = FALSE, ...) ## S3 method for class 'secrlist' MCgof(object, nsim = 100, statfn = NULL, testfn = NULL, seed = NULL, ncores = 1, clustertype = c("PSOCK", "FORK"), usefxi = TRUE, useMVN = TRUE, Ndist = NULL, quiet = FALSE, debug = FALSE, ...)
object |
secr fitted model or |
nsim |
integer number of replicates |
statfn |
function to extract summary statistics from capture histories |
testfn |
function to compare observed and expected counts |
seed |
integer seed |
ncores |
integer for number of parallel cores |
clustertype |
character cluster type for parallel::makeCluster |
usefxi |
logical; if FALSE then AC are simulated de novo from the density process rather than using information on the detected individuals |
useMVN |
logical; if FALSE parameter values are fixed at the MLE rather than drawn from multivariate normal distribution |
Ndist |
character; distribution of number of unobserved AC (optional) |
quiet |
logical; if FALSE then a progress bar (ncores=1) and final timing are shown |
debug |
integer; if >0 then the browser is started at one of 4 points in code |
... |
other arguments passed to testfn |
At each replicate parameter values are sampled from the multivariate-normal sampling distribution of the fitted model. The putative location of each detected individual is drawn from the spatial distribution implied by its observations and the resampled parameters (see fxi
); locations of undetected individuals are simulated from the complement of pdot(x) times D(x).
New detections are simulated under the model for individuals at the simulated locations, along with the expected numbers. Detections form a capthist object, a 3-D array with dimensions for individual , occasion
and detector
*. Thus for each replicate and detected individual there are the original observations
, simulated observations
, and expected counts
. Two discrepancy statistics are calculated for each replicate – observed vs expected counts, and simulated vs expected counts – and a record is kept of which of these discrepancy statistics is the larger (indicating poorer fit).
* Notation differs slightly from Choo et al. (2024), using for occasion and
for detector to be consistent with usage in secr and elsewhere (e.g., Borchers and Fewster 2016).
The default discrepancy (testfn
) is the Freeman-Tukey statistic as in Choo et al. (2024) and Royle et al. (2014) (see also Brooks, Catchpole and Morgan 2000). The statistic has this general form for counts
with expected value
:
The key output of MCgof
is the proportion of replicates in which the simulated discrepancy exceeds the observed discrepancy. For perfect fit this will be about 0.5, and for poor fit it will approach zero.
By default, tests are performed separately for three types of count: the numbers of detections of each individual (yi), at each detector (yk), and for each individual at each detector (yik) extracted by the default statfn
from the margins of the observed and simulated capture histories.
|
individual x detector | |
|
individual | |
|
detector | |
Parallel processing is offered using multiple cores (CPUs) through the package parallel when ncores > 1. This differs from the usual multithreading paradigm in secr and does not rely on the environment variable set by setNumThreads
except that, if ncores = NULL, ncores will be set to the value from setNumThreads
. The cluster type "FORK" is available only on Unix-like systems; it can require large amounts of memory, but is generally fast. A small value of ncores>1 may be optimal, especially With cluster type "PSOCK".
‘usefxi’ and ‘useMVN’ may be used to drop key elements of the Choo et al. (2024) approach - they are provided for demonstration only.
‘Ndist’ refers to the distribution of the number of unobserved AC, conditional on the expected number where
is the resampled density,
the mask area, and
the number of detected individuals. By default ‘Ndist’ depends on the distribution component of the ‘details’ argument of the fitted model (“poisson" for Poisson
, “fixed"" for binomial
).
‘debug’ may be used to view intermediate data at certain points in MCgof() numbered 1 to 5. Examine the code of secr:::MCgof.secr or secr:::simfxiAC for these points. Debugging requires ‘ncores = 1’.
The RNGkind
of the random number generator is set internally for consistency across platforms.
The ... argument may be used to pass 'np' and 'verbose' to ‘Fletcher.chat' used as ’testfn'.
Invisibly returns an object of class ‘MCgof’ with components -
nsim |
as input |
statfn |
as input or default |
testfn |
as input or default |
all |
list of outputs: for each statistic, a 3 x nsim matrix. Rows correspond to Tobs, Tsim, and a binary indicator for Tsim > Tobs |
proctime |
execution time in seconds |
For secrlist input the value returned is a list of ‘MCgof’ objects.
Not all models are covered and some are untested. These models are specifically excluded -
multi-session models
models with groups
conditional likelihood
polygon, transect, telemetry or signal detectors
non-binary behavioural responses
This implementation extends the work of Choo et al. (2024) in these respects -
detector types ‘multi’ and ‘count’ are allowed
the model may include variation among detectors
the model may include behavioural responses
2-class finite mixture and hybrid mixture models are both allowed.
Murray Efford and Yan Ru Choo
Borchers, D. L. and Efford, M. G. (2008) Spatially explicit maximum likelihood methods for capture–recapture studies. Biometrics 64, 377–385.
Borchers, D. L. and Fewster, R. M. (2016) Spatial capture–recapture models. Statistical Science 31, 219–232.
Brooks, S. P., Catchpole, E. A. and Morgan, B. J. T. (2000) Bayesian animal survival estimation. Statistical Science 15, 357–376.
Choo, Y. R., Sutherland, C. and Johnston, A. (2024) A Monte Carlo resampling framework for implementing goodness-of-fit tests in spatial capture-recapture model Methods in Ecology and Evolution DOI: 10.1111/2041-210X.14386.
Gelman, A., Meng, X.-L., and Stern, H. (1996) Posterior predictive assessment of model fitness via realized discrepancies. Statistica Sinica 6, 733–807.
Royle, J. A., Chandler, R. B., Sollmann, R. and Gardner, B. (2014) Spatial capture–recapture. Academic Press.
Parallel,
secr.test
,
plot.MCgof
,
hist.MCgof
,
summary.MCgof
tmp <- MCgof(secrdemo.0) summary(tmp) par(mfrow = c(1,3), pty = 's') plot(tmp)
tmp <- MCgof(secrdemo.0) summary(tmp) par(mfrow = c(1,3), pty = 's') plot(tmp)
AIC- or AICc-weighted average of estimated ‘real’ or ‘beta’ parameters from multiple fitted secr models, and the tabulation of estimates.
## S3 method for class 'secr' modelAverage(object, ..., realnames = NULL, betanames = NULL, newdata = NULL, alpha = 0.05, dmax = 10, covar = FALSE, average = c("link", "real"), criterion = c("AIC","AICc"), CImethod = c("Wald", "MATA"), chat = NULL) ## S3 method for class 'secrlist' modelAverage(object, ..., realnames = NULL, betanames = NULL, newdata = NULL, alpha = 0.05, dmax = 10, covar = FALSE, average = c("link", "real"), criterion = c("AIC","AICc"), CImethod = c("Wald", "MATA"), chat = NULL)
## S3 method for class 'secr' modelAverage(object, ..., realnames = NULL, betanames = NULL, newdata = NULL, alpha = 0.05, dmax = 10, covar = FALSE, average = c("link", "real"), criterion = c("AIC","AICc"), CImethod = c("Wald", "MATA"), chat = NULL) ## S3 method for class 'secrlist' modelAverage(object, ..., realnames = NULL, betanames = NULL, newdata = NULL, alpha = 0.05, dmax = 10, covar = FALSE, average = c("link", "real"), criterion = c("AIC","AICc"), CImethod = c("Wald", "MATA"), chat = NULL)
object |
secr or secrlist object |
... |
other secr objects |
realnames |
character vector of real parameter names |
betanames |
character vector of beta parameter names |
newdata |
optional dataframe of values at which to evaluate models |
alpha |
alpha level for confidence intervals |
dmax |
numeric, the maximum AIC or AICc difference for inclusion in confidence set |
covar |
logical, if TRUE then return variance-covariance matrix |
average |
character string for scale on which to average real parameters |
criterion |
character, information criterion to use for model weights |
CImethod |
character, type of confidence interval (see Details) |
chat |
numeric optional variance inflation factor for quasi-AIC weights |
Models to be compared must have been fitted to the same data and use the
same likelihood method (full vs conditional). If realnames
=
NULL and betanames
= NULL then all real parameters will be
averaged; in this case all models must use the same real parameters. To
average beta parameters, specify betanames
(this is ignored if a
value is provided for realnames
). See predict.secr
for an explanation of the optional argument newdata
;
newdata
is ignored when averaging beta parameters.
Model-averaged estimates for parameter are given by
where the subscript refers to a specific
model and the
are AIC or AICc weights (see
AIC.secr
for details). Averaging of real parameters may be
done on the link scale before back-transformation
(average="link"
) or after back-transformation
(average="real"
).
Models for which dAIC > dmax
(or dAICc > dmax
) are given a
weight of zero and effectively are excluded from averaging.
Also,
where and the variances are asymptotic estimates
from fitting each model
. This follows Burnham and Anderson
(2004) rather than Buckland et al. (1997).
Two methods are offered for confidence intervals. The default ‘Wald’
uses the above estimate of variance. The alternative ‘MATA’
(model-averaged tail area) avoids estimating a weighted variance and
is thought to provide better coverage at little cost in increased
interval length (Turek and Fletcher 2012). Turek and Fletcher (2012)
also found averaging with AIC weights (here criterion = 'AIC'
)
preferable to using AICc weights, even for small
samples. CImethod
does not affect the reported standard errors.
If 'chat' is provided then quasi-AIC or quasi-AICc weights are used, depending on the value of 'criterion'.
For modelAverage
, an array of model-averaged estimates, their
standard errors, and a % confidence
interval. The interval for real parameters is backtransformed from the
link scale. If there is only one row in
newdata
or beta
parameters are averaged or averaging is requested for only one parameter
then the array is collapsed to a matrix. If covar = TRUE
then a
list is returned with separate components for the estimates and the
variance-covariance matrices.
modelAverage
replaces the deprecated function model.average
whose name conflicted with a method in RMark.
Buckland S. T., Burnham K. P. and Augustin, N. H. (1997) Model selection: an integral part of inference. Biometrics 53, 603–618.
Burnham, K. P. and Anderson, D. R. (2002) Model Selection and Multimodel Inference: A Practical Information-Theoretic Approach. Second edition. New York: Springer-Verlag.
Burnham, K. P. and Anderson, D. R. (2004) Multimodel inference - understanding AIC and BIC in model selection. Sociological Methods & Research 33, 261–304.
Turek, D. and Fletcher, D. (2012) Model-averaged Wald confidence intervals. Computational statistics and data analysis 56, 2809–2815.
modelAverage
,
AIC.secr
,
secr.fit
,
collate
## Compare two models fitted previously ## secrdemo.0 is a null model ## secrdemo.b has a learned trap response modelAverage(secrdemo.0, secrdemo.b) modelAverage(secrdemo.0, secrdemo.b, betanames = c("D","g0","sigma")) ## In this case we find the difference was actually trivial... ## (subscripting of output is equivalent to setting fields = 1)
## Compare two models fitted previously ## secrdemo.0 is a null model ## secrdemo.b has a learned trap response modelAverage(secrdemo.0, secrdemo.b) modelAverage(secrdemo.0, secrdemo.b, betanames = c("D","g0","sigma")) ## In this case we find the difference was actually trivial... ## (subscripting of output is equivalent to setting fields = 1)
Logical function to distinguish objects that span multiple sessions
## Default S3 method: ms(object, ...) ## S3 method for class 'mask' ms(object, ...) ## S3 method for class 'secr' ms(object, ...)
## Default S3 method: ms(object, ...) ## S3 method for class 'mask' ms(object, ...) ## S3 method for class 'secr' ms(object, ...)
object |
any object |
... |
other arguments (not used) |
The test applied varies with the type of object. The default method uses
inherits(object, "list")
.
logical, TRUE if object
contains data for multiple sessions
ms(ovenCH) ms(ovenbird.model.1) ms(ovenCH[[1]])
ms(ovenCH) ms(ovenbird.model.1) ms(ovenCH[[1]])
Internal function used to generate a dataframe containing design data for the base levels of all predictors in an secr object.
## Default S3 method: makeNewData(object, all.levels = FALSE, ...) ## S3 method for class 'secr' makeNewData(object, all.levels = FALSE, bytrap = FALSE, ...)
## Default S3 method: makeNewData(object, all.levels = FALSE, ...) ## S3 method for class 'secr' makeNewData(object, all.levels = FALSE, bytrap = FALSE, ...)
object |
fitted secr model object |
all.levels |
logical; if TRUE then all levels of factors are included |
bytrap |
logical; if TRUE then all detectors are included |
... |
other arguments (not used) |
makeNewData
is used by predict
in lieu of
user-specified ‘newdata’. There is seldom any need to call the function
makeNewData
directly.
With ‘bytrap’ there is a row for each detector, with columns for the corresponding levels of any detector covariates that appear in the model.
A dataframe with one row for each session and group, and columns for the
predictors used by object$model
.
## from previously fitted model makeNewData(secrdemo.b)
## from previously fitted model makeNewData(secrdemo.b)
Non-target detections and interference events may be recorded in a binary detector x occasion matrix attached as an attribute to a single-session capthist object, or to each component of a multi-session capthist object.
Models fitted by secr make no use of these data. They may be used in ipsecr.
From secr 4.5.6, a summary of nontarget data is reported by the
summary
method for capthist objects, and the verify
method reports clashes between detections and nontarget data.
Extraction and replacement funcions are provided from secr 4.5.7 on.
nontarget(object, ...) nontarget(object) <- value
nontarget(object, ...) nontarget(object) <- value
object |
capthist object |
... |
other argments (not used) |
value |
numeric binary matrix (rows = detectors, columns = occasions) |
The order of rows should match the order of detectors in traps(object)
.
Matrix entries should be zero for trap x occasion combinations that were not used
(see usage) or for which there is a corresponding detection.
value
is coerced to a matrix before assignment.
For nontarget()
, a matrix or list of matrices.
plot.capthist
, summary.capthist
set.seed(123) ch <- captdata # traps that caught something caught <- t(apply(ch, 2:3, sum)) # construct artificial nontarget data # (positive for half the traps that caught nothing) nontarget(ch) <- (1-caught) * (runif(500)>0.5) head(caught) head(nontarget(ch)) # the summary method recognises the 'nontarget' attribute summary(ch)$nontarget
set.seed(123) ch <- captdata # traps that caught something caught <- t(apply(ch, 2:3, sum)) # construct artificial nontarget data # (positive for half the traps that caught nothing) nontarget(ch) <- (1-caught) * (runif(500)>0.5) head(caught) head(nontarget(ch)) # the summary method recognises the 'nontarget' attribute summary(ch)$nontarget
Displays a graphic key to the occasions corresponding to petals in a petal plot.
occasionKey(capthist, noccasions, rad = 3, x, y, px = 0.9, py = 0.9, title = 'Occasion', ...)
occasionKey(capthist, noccasions, rad = 3, x, y, px = 0.9, py = 0.9, title = 'Occasion', ...)
capthist |
single-session capthist object |
noccasions |
number of petals (if |
rad |
distance of petal centre from key centre |
x |
numeric x coordinate for centre of key |
y |
numeric y coordinate for centre of key |
px |
x position as fraction of user coordinates |
py |
y position as fraction of user coordinates |
title |
character |
... |
other arguments passed to |
Either capthist
or noccasions
is required. It is assumed that a plot exists.
Graphic arguments in ... are applied to both the title and the occasion numbers.
The key will be added to an existing plot. No value is returned.
plot(captdata, border = 50) occasionKey(captdata, rad = 8, cex = 0.8)
plot(captdata, border = 50) occasionKey(captdata, rad = 8, cex = 0.8)
Data from a multi-year mist-netting study of ovenbirds (Seiurus aurocapilla) at a site in Maryland, USA.
ovenCH ovenCHp ovenbird.model.1 ovenbird.model.D ovenmask
ovenCH ovenCHp ovenbird.model.1 ovenbird.model.D ovenmask
From 2005 to 2009 D. K. Dawson and M. G. Efford conducted a capture–recapture survey of breeding birds in deciduous forest at the Patuxent Research Refuge near Laurel, Maryland, USA. The forest was described by Stamm, Davis & Robbins (1960), and has changed little since. Analyses of data from previous mist-netting at the site by Chan Robbins were described in Efford, Dawson & Robbins (2004) and Borchers & Efford (2008).
Forty-four mist nets (12 m long, 30-mm mesh) spaced 30 m apart on the perimeter of a 600-m x 100-m rectangle were operated for approximately 9 hours on each of 9 or 10 non-consecutive days during late May and June in each year. Netting was passive (i.e. song playback was not used to lure birds into the nets). Birds received individually numbered bands, and both newly banded and previously banded birds were released at the net where captured. Sex was determined in the hand from the presence of a brood patch (females) or cloacal protuberance (males). A small amount of extra netting was done by other researchers after the main session in some years.
This dataset comprises all records of adult (after-hatch-year) ovenbirds caught during the main session in each of the five years 2005–2009. One ovenbird was killed by a predator in the net in 2009, as indicated by a negative net number in the dataset. Sex was determined in the hand from the presence of a brood patch (females) or cloacal protuberance (males). Birds are listed by their band number (4-digit prefix, ‘.’, and 5-digit number).
The data are provided as a multi-session capthist
object
‘ovenCHp’. Sex is coded as a categorical individual covariate ("M"
or "F").
Recaptures at the same site within a day are not included in this dataset,
so ovenCHp
has detector type ‘proximity’. Previous versions of secr
provided only a trimmed version of these data, retaining only one capture
per bird per day (ovenCH
with detector type ‘multi’). That may be
obtained from ovenCHp
as shown in the examples.
Although several individuals were captured in more than one year, no use is made of this information in the analyses presently offered in secr.
An analysis of the data for males in the first four years showed that they tended to avoid nets after their first capture within a season (Dawson & Efford 2009). While the species was present consistently, the number of detections in any one year was too small to give reliable estimates of density; pooling of detection parameters across years helped to improve precision.
Included with the data are a mask and two models fitted to ovenCH
as in
Examples.
Object | Description |
ovenCH | multi-session capthist object (as multi-catch) |
ovenCHp | multi-session capthist object (as binary proximity) |
ovenbird.model.1 | fitted secr model -- null |
ovenbird.model.D | fitted secr model -- trend in density across years |
ovenmask | mask object |
D. K. Dawson ([email protected]) and M. G. Efford unpublished data.
Borchers, D. L. and Efford, M. G. (2008) Spatially explicit maximum likelihood methods for capture-recapture studies. Biometrics 64, 377–385.
Dawson, D. K. and Efford, M. G. (2009) Bird population density estimated from acoustic signals. Journal of Applied Ecology 46, 1201–1209.
Efford, M. G., Dawson, D. K. and Robbins C. S. (2004) DENSITY: software for analysing capture-recapture data from passive detector arrays. Animal Biodiversity and Conservation 27, 217–228.
Stamm, D. D., Davis, D. E. and Robbins, C. S. (1960) A method of studying wild bird populations by mist-netting and banding. Bird-Banding 31, 115–130.
## commands used to create ovenCH from the input files ## "netsites0509.txt" and "ovencapt.txt" ## for information only - these files not distributed # netsites0509 <- read.traps(file = "netsites0509.txt", # skip = 1, detector = "proximity") # temp <- read.table("ovencapt.txt", colClasses=c("character", # "character", "numeric", "numeric", "character")) # ovenCHp <- make.capthist(temp, netsites0509, covnames = "Sex") # ovenCHp <- reduce(ovenCHp, dropunused = FALSE) # drop repeat detections par(mfrow = c(1,5), mar = c(1,1,4,1)) plot(ovenCHp, tracks = TRUE, varycol = TRUE) par(mfrow = c(1,1), mar = c(5,4,4,2) + 0.1) ## defaults counts(ovenCHp, "n") ## Not run: ## trimmed version of data - for consistency with earlier versions ovenCH <- reduce(ovenCHp, outputdetector = "multi", dropunused = FALSE) ## array constant over years, so build mask only once ovenmask <- make.mask(traps(ovenCH)[["2005"]], type = "pdot", buffer = 400, spacing = 15, detectpar = list(g0 = 0.03, sigma = 90), nocc = 10) ## fit constant-density model ovenbird.model.1 <- secr.fit(ovenCH, mask = ovenmask) ## fit temporal trend in density (Session capitalized) ovenbird.model.D <- secr.fit(ovenCH, mask = ovenmask, model = list(D ~ Session)) ## compare pre-fitted models AIC(ovenbird.model.1, ovenbird.model.D) ## End(Not run)
## commands used to create ovenCH from the input files ## "netsites0509.txt" and "ovencapt.txt" ## for information only - these files not distributed # netsites0509 <- read.traps(file = "netsites0509.txt", # skip = 1, detector = "proximity") # temp <- read.table("ovencapt.txt", colClasses=c("character", # "character", "numeric", "numeric", "character")) # ovenCHp <- make.capthist(temp, netsites0509, covnames = "Sex") # ovenCHp <- reduce(ovenCHp, dropunused = FALSE) # drop repeat detections par(mfrow = c(1,5), mar = c(1,1,4,1)) plot(ovenCHp, tracks = TRUE, varycol = TRUE) par(mfrow = c(1,1), mar = c(5,4,4,2) + 0.1) ## defaults counts(ovenCHp, "n") ## Not run: ## trimmed version of data - for consistency with earlier versions ovenCH <- reduce(ovenCHp, outputdetector = "multi", dropunused = FALSE) ## array constant over years, so build mask only once ovenmask <- make.mask(traps(ovenCH)[["2005"]], type = "pdot", buffer = 400, spacing = 15, detectpar = list(g0 = 0.03, sigma = 90), nocc = 10) ## fit constant-density model ovenbird.model.1 <- secr.fit(ovenCH, mask = ovenmask) ## fit temporal trend in density (Session capitalized) ovenbird.model.D <- secr.fit(ovenCH, mask = ovenmask, model = list(D ~ Session)) ## compare pre-fitted models AIC(ovenbird.model.1, ovenbird.model.D) ## End(Not run)
Data from an acoustic survey of ovenbirds (Seiurus aurocapilla) at a site in Maryland, USA.
signalCH ovensong.model.1 ovensong.model.2
signalCH ovensong.model.1 ovensong.model.2
In June 2007 D. K. Dawson and M. G. Efford used a moving 4-microphone array
to survey breeding birds in deciduous forest at the Patuxent Research
Refuge near Laurel, Maryland, USA. The data for ovenbirds were used to
demonstrate a new method for analysing acoustic data (Dawson and Efford
2009). See ovenbird
for mist-netting data from the same
site over 2005–2009, and for other background.
Over five days, four microphones were placed in a square (21-m side) centred at each of 75 points in a rectangular grid (spacing 50 m); on each day points 100 m apart were sampled sequentially. Recordings of 5 minutes duration were made in .wav format on a 4-channel digital sound recorder.
The data are estimates of average power on each channel (microphone) for the first song of each ovenbird distinguishable in a particular 5-minute recording. Power was estimated with the sound analysis software Raven Pro 1.4 (Charif et al. 2008), using a window of 0.7 s duration and frequencies between 4200 and 5200 Hz, placed manually at the approximate centre of each ovenbird song. Sometimes this frequency range was obscured by insect noise so an alternative 1000-Hz range was measured and the values were adjusted by regression.
The data are provided as a single-session, single-occasion
capthist
object signalCH
. The ‘signal’ attribute contains
the power measurement in decibels for each detected sound on each
channel where the power threshold is exceeded. As the threshold signal
(attribute cutval
= 35) is less than any signal value in this
dataset, all detection histories are complete (1,1,1,1) across
microphones. For analysis Dawson and Efford applied a higher threshold
that treated weaker signals as ‘not detected’ (see Examples).
The row names of signalCH
(e.g. "3755AX") are formed from a
4-digit number indicating the sampling location (one of 75 points on a
50-m grid) and a letter A–D to distinguish individual ovenbirds within
a 5-minute recording; ‘X’ indicates power values adjusted by regression.
The default model for sound attenuation is a log-linear decline with
distance from the source (linear decline on dB scale). Including a
spherical spreading term in the sound attenuation model causes the
likelihood surface to become multimodal in this case. Newton-Raphson,
the default maximization method in secr.fit
, is particularly
inclined to settle on a local maximum; in the example below we use a set
of starting values that have been found by trial and error to yield the
global maximum.
Two fitted models are included (see Examples for details).
Object | Description |
signalCH | capthist object |
ovensong.model.1 | fitted secr model -- spherical spreading |
ovensong.model.2 | fitted secr model -- no spherical spreading |
D. K. Dawson ([email protected]) and M. G. Efford unpublished data.
Charif, R. A., Waack, A. M. and Strickman, L. M. (2008) Raven Pro 1.3 User's Manual. Cornell Laboratory of Ornithology, Ithaca, New York.
Dawson, D. K. and Efford, M. G. (2009) Bird population density estimated from acoustic signals. Journal of Applied Ecology 46, 1201–1209.
Efford, M. G., Dawson, D. K. and Borchers, D. L. (2009) Population density estimated from locations of individuals on a passive detector array. Ecology 90, 2676–2682.
capthist
, ovenbird
, Detection functions
summary(signalCH) traps(signalCH) signal(signalCH) ## apply signal threshold signalCH.525 <- subset(signalCH, cutval = 52.5) ## Not run: ## models with and without spherical spreading omask <- make.mask(traps(signalCH), buffer = 200) ostart <- c(log(20), 80, log(0.1), log(2)) ovensong.model.1 <- secr.fit( signalCH.525, mask = omask, start = ostart, detectfn = 11 ) ovensong.model.2 <- secr.fit( signalCH.525, mask = omask, start = ostart, detectfn = 10 ) ## End(Not run) ## compare fit of models AIC(ovensong.model.1, ovensong.model.2) ## density estimates, dividing by 75 to allow for replication collate(ovensong.model.1, ovensong.model.2)[1,,,"D"]/75 ## plot attenuation curves cf Dawson & Efford (2009) Fig 5 pars1 <- predict(ovensong.model.1)[c("beta0", "beta1"), "estimate"] pars2 <- predict(ovensong.model.2)[c("beta0", "beta1"), "estimate"] attenuationplot(pars1, xval=0:150, spherical = TRUE, ylim = c(40,110)) attenuationplot(pars2, xval=0:150, spherical = FALSE, add = TRUE, col = "red") ## spherical spreading only pars1[2] <- 0 attenuationplot(pars1, xval=0:150, spherical = TRUE, add = TRUE, lty=2)
summary(signalCH) traps(signalCH) signal(signalCH) ## apply signal threshold signalCH.525 <- subset(signalCH, cutval = 52.5) ## Not run: ## models with and without spherical spreading omask <- make.mask(traps(signalCH), buffer = 200) ostart <- c(log(20), 80, log(0.1), log(2)) ovensong.model.1 <- secr.fit( signalCH.525, mask = omask, start = ostart, detectfn = 11 ) ovensong.model.2 <- secr.fit( signalCH.525, mask = omask, start = ostart, detectfn = 10 ) ## End(Not run) ## compare fit of models AIC(ovensong.model.1, ovensong.model.2) ## density estimates, dividing by 75 to allow for replication collate(ovensong.model.1, ovensong.model.2)[1,,,"D"]/75 ## plot attenuation curves cf Dawson & Efford (2009) Fig 5 pars1 <- predict(ovensong.model.1)[c("beta0", "beta1"), "estimate"] pars2 <- predict(ovensong.model.2)[c("beta0", "beta1"), "estimate"] attenuationplot(pars1, xval=0:150, spherical = TRUE, ylim = c(40,110)) attenuationplot(pars2, xval=0:150, spherical = FALSE, add = TRUE, col = "red") ## spherical spreading only pars1[2] <- 0 attenuationplot(pars1, xval=0:150, spherical = TRUE, add = TRUE, lty=2)
A dataset from long-term capture-recapture trapping of brushtail possums Trichosurus vulpecula in New Zealand.
OVpossumCH
OVpossumCH
A multi-session capthist object of 6 sessions. Sessions are labeled 49–54, corresponding to February 1996, June 1996, September 1996, February 1997, June 1997 and September 1997.
Brushtail possums are 2-4 kg largely arboreal marsupials that have become pests of forests and farmland in New Zealand since their introduction from Australia in the nineteenth century. Their population dynamics in mixed native forest have been studied by capture-recapture in the Orongorongo Valley near Wellington since 1966 (e.g. Crawley 1973, Efford 1998, Efford and Cowan 2004).
From 1996 to 2006, a grid of 167 traps set on the ground at 30-m spacing was operated in an area of about 14 ha for 5 consecutive days three times each year (Efford and Cowan 2004). Each trap could catch only one animal, with rare exceptions when a young ‘backrider’ entered the trap with its mother. All animals were tagged and tattooed for individual identification and released at the site of capture.
A broad shingle riverbed forms a natural boundary on two sides of the study grid. Much of the grid lies on a gently sloping old alluvial fan and recent terraces, but to the southeast the valley side rises steeply and, except where cut by streams, supports beech forest (Nothofagus truncata and Nothofagus solandri solandri) rather than the mixed broadleaf forest of the valley floor.
This dataset relates to six five-day trapping sessions in 1996 and 1997, a time of high and declining density. Possums are long-lived (up to about 15 years) and as adults restrict their movements to a home range of 1-10 ha. Breeding is seasonal, resulting in an influx of newly independent juveniles in the first trapping of each calendar year.
The dataset includes individual covariates not provided by Efford (2012): ‘sex’ (F or M) and ‘Ageclass’ (1 for first year, 2 for older).
A coarse habitat map is provided for the immediate vicinity of the trapping grid as the shapefile ‘OVforest.shp’ in the package ‘extdata’ folder. This distinguishes two forest classes (‘beech’ and ‘non-beech’), and leaves out the shingle riverbed. The distinction between ‘beech’ and ‘non-beech’ is mapped only to a distance of about 120 m from the outermost traps. A text file 'leftbank.txt' in the same folder contains the x- and y- coordinates of the adjoining bank of the Orongorongo River. All coordinates relate to the old New Zealand Map Grid (NZMG), since replaced by the New Zealand Transverse Mercator grid (NZTM2000).
The example code shows how to import the shapefile as a sp
SpatialPolygonsDataFrame object and use it to construct a mask for
secr.fit
.
Efford (2012) and unpublished data.
Crawley, M. C. (1973) A live-trapping study of Australian brush-tailed possums, Trichosurus vulpecula (Kerr), in the Orongorongo Valley, Wellington, New Zealand. Australian Journal of Zoology 21, 75–90.
Efford, M. G. (1998) Demographic consequences of sex-biased dispersal in a population of brushtail possums. Journal of Animal Ecology 67, 503–517.
Efford, M. G. (2012) DENSITY 5.0: software for spatially explicit capture-recapture. Department of Mathematics and Statistics, University of Otago, Dunedin, New Zealand. https://www.otago.ac.nz/density/
Efford, M. G. and Cowan, P. E. (2004) Long-term population trend of Trichosurus vulpecula in the Orongorongo Valley, New Zealand. In: The Biology of Australian Possums and Gliders. Edited by R. L. Goldingay and S. M. Jackson. Surrey Beatty & Sons, Chipping Norton. Pp. 471–483.
Ward, G. D. (1978) Habitat use and home range of radio-tagged opossums Trichosurus vulpecula (Kerr) in New Zealand lowland forest. In: The ecology of arboreal folivores. Edited by G. G. Montgomery. Smithsonian Institute Press. Washington, D.C. Pp. 267–287.
## Not run: library(sf) summary(OVpossumCH, terse = TRUE) ovtrap <- traps(OVpossumCH[[1]]) ## retrieve and plot the forest map OVforest <- st_read(system.file("extdata/OVforest.shp", package = "secr")) ## omit forest across the river by selecting only 1,2 OVforest <- OVforest[1:2,] forestcol <- terrain.colors(6)[c(4,2)] plot(st_as_sfc(OVforest), col = forestcol) plot(ovtrap, add = TRUE) ## construct a mask ovmask <- make.mask(ovtrap, buffer = 120, type = 'trapbuffer', poly = OVforest, spacing = 7.5, keep.poly = FALSE) ovmask <- addCovariates(ovmask, OVforest) ## display mask par(mar = c(0,0,0,8)) plot(ovmask, covariate = 'forest', dots = FALSE, col = forestcol) plot(ovtrap, add = TRUE) ## add the left bank of the Orongorongo River lines(read.table(system.file("extdata/leftbank.txt", package = "secr"))) ## End(Not run)
## Not run: library(sf) summary(OVpossumCH, terse = TRUE) ovtrap <- traps(OVpossumCH[[1]]) ## retrieve and plot the forest map OVforest <- st_read(system.file("extdata/OVforest.shp", package = "secr")) ## omit forest across the river by selecting only 1,2 OVforest <- OVforest[1:2,] forestcol <- terrain.colors(6)[c(4,2)] plot(st_as_sfc(OVforest), col = forestcol) plot(ovtrap, add = TRUE) ## construct a mask ovmask <- make.mask(ovtrap, buffer = 120, type = 'trapbuffer', poly = OVforest, spacing = 7.5, keep.poly = FALSE) ovmask <- addCovariates(ovmask, OVforest) ## display mask par(mar = c(0,0,0,8)) plot(ovmask, covariate = 'forest', dots = FALSE, col = forestcol) plot(ovtrap, add = TRUE) ## add the left bank of the Orongorongo River lines(read.table(system.file("extdata/leftbank.txt", package = "secr"))) ## End(Not run)
From version 4.0 secr uses multi-threading in C++ (package RcppParallel, Allaire et al. 2021) to speed likelihood evaluation and hence model fitting in secr.fit
. Detection histories are distributed over threads. Setting ncores = NULL
in functions with multi-threading uses the existing value from the environment variable RCPP_PARALLEL_NUM_THREADS (see setNumThreads
).
These functions use multi-threading and call setNumThreads
internally:
autoini |
confint.secr |
derived.secr |
esa.secr |
fxi and related functions |
pdot |
region.N |
score.test |
secr.fit |
These functions use multi-threading without calling setNumThreads
:
LLsurface.secr |
mask.check |
expected.n |
secr.test |
sim.secr |
Other functions may use multithreading indirectly through a call to one of these functions. Examples are suggest.buffer
(autoini
), esaPlot
(pdot
), and bias.D
(pdot
).
NOTE: The mechanism for setting the number of threads changed between versions 4.1.0 and 4.2.0. The default number of cores is now capped at 2 to meet CRAN requirements. Setting ncores = NULL
previously specified one less than the maximum number of cores.
Earlier versions of secr made more limited use of multiple cores (CPUs)
through the package parallel. The functions par.secr.fit
, par.derived
, and par.region.N
are now deprecated because they were too slow. list.secr.fit
replaces par.secr.fit
‘Unit’ refers to the unit of work sent to each worker process. As a guide, a ‘large’ benefit means >60% reduction in process time with 4 CPUs.
parallel offers several different mechanisms, bringing together the functionality of multicore and snow. The mechanism used by secr is the simplest available, and is expected to work across all operating systems. Technically, it relies on Rscript and communication between the master and worker processes is via sockets. As stated in the parallel documentation "Users of Windows and Mac OS X may expect pop-up dialog boxes from the firewall asking if an R process should accept incoming connections". You may possibly get warnings from R about closing unused connections. These can safely be ignored.
Use parallel::detectCores()
to get
an idea of how many cores are available on your machine; this may (in
Windows) include virtual cores over and above the number of physical
cores. See RShowDoc("parallel", package = "parallel") in core R for
explanation.
In secr.fit
the output component ‘proctime’ misrepresents the
elapsed processing time when multiple cores are used.
It appears that multicore operations in secr using parallel may fail if the packages snow and snowfall are also loaded. The error message is obscure:
“Error in UseMethod("sendData") : no applicable method for 'sendData' applied to an object of class "SOCK0node"”
Allaire, J. J., Francois, R., Ushey, K., Vandenbrouck, G., Geelnard, M. and Intel (2021) RcppParallel: Parallel Programming Tools for 'Rcpp'. R package version 5.1.2. https://CRAN.R-project.org/package=RcppParallel.
## Not run: sessionInfo() # R version 4.3.0 (2023-04-21 ucrt) # Platform: x86_64-w64-mingw32/x64 (64-bit) # Running under: Windows 11 x64 (build 22621) # on Dell-XPS 8950 Intel i7-12700K # ... # see stackoverflow suggestion for microbenchmark list argument # https://stackoverflow.com/questions/32950881/how-to-use-list-argument-in-microbenchmark library(microbenchmark) options(digits = 4) ## benefit from multi-threading in secr.fit jobs <- lapply(seq(2,8,2), function(nc) bquote(suppressWarnings(secr.fit(captdata, trace = FALSE, ncores = .(nc))))) microbenchmark(list = jobs, times = 10, unit = "seconds") # [edited output] # Unit: seconds # expr min lq mean median uq max neval # ncores = 2 1.75880 2.27978 2.6680 2.7450 3.0960 3.4334 10 # ncores = 4 1.13549 1.16280 1.6120 1.4431 2.0041 2.4018 10 # ncores = 6 0.88003 0.98215 1.2333 1.1387 1.5175 1.6966 10 # ncores = 8 0.78338 0.90033 1.5001 1.0406 1.2319 4.0669 10 ## sometimes (surprising) lack of benefit with ncores>2 msk <- make.mask(traps(ovenCH[[1]]), buffer = 300, nx = 25) jobs <- lapply(c(1,2,4,8), function(nc) bquote(secr.fit(ovenCH, trace = FALSE, ncores = .(nc), mask = msk))) microbenchmark(list = jobs, times = 10, unit = "seconds") # [edited output] # Unit: seconds # expr min lq mean median uq max neval # ncores = 1 12.5010 13.4951 15.674 15.304 16.373 21.723 10 # ncores = 2 10.0363 11.8634 14.407 13.726 16.782 22.966 10 # ncores = 4 8.6335 10.3422 13.085 12.449 15.729 17.914 10 # ncores = 8 8.5286 9.9008 10.751 10.736 10.796 14.885 10 ## and for simulation... jobs <- lapply(seq(2,8,2), function(nc) bquote(sim.secr(secrdemo.0, nsim = 20, tracelevel = 0, ncores = .(nc)))) microbenchmark(list = jobs, times = 10, unit = "seconds") # [edited output] # Unit: seconds # expr min lq mean median uq max neval # ncores = 2 48.610 49.932 59.032 52.485 54.730 119.905 10 # ncores = 4 29.480 29.996 30.524 30.471 31.418 31.612 10 # ncores = 6 22.583 23.594 24.148 24.354 24.644 25.388 10 # ncores = 8 19.924 20.651 25.581 21.002 21.696 51.920 10 ## and log-likelihood surface jobs <- lapply(seq(2,8,2), function(nc) bquote(suppressMessages(LLsurface(secrdemo.0, ncores = .(nc))))) microbenchmark(list = jobs, times = 10, unit = "seconds") # [edited output] # Unit: seconds # expr min lq mean median uq max neval # ncores = 2 20.941 21.098 21.290 21.349 21.471 21.619 10 # ncores = 4 14.982 15.125 15.303 15.263 15.449 15.689 10 # ncores = 6 13.994 14.299 14.529 14.342 14.458 16.515 10 # ncores = 8 13.597 13.805 13.955 13.921 14.128 14.353 10 ## End(Not run)
## Not run: sessionInfo() # R version 4.3.0 (2023-04-21 ucrt) # Platform: x86_64-w64-mingw32/x64 (64-bit) # Running under: Windows 11 x64 (build 22621) # on Dell-XPS 8950 Intel i7-12700K # ... # see stackoverflow suggestion for microbenchmark list argument # https://stackoverflow.com/questions/32950881/how-to-use-list-argument-in-microbenchmark library(microbenchmark) options(digits = 4) ## benefit from multi-threading in secr.fit jobs <- lapply(seq(2,8,2), function(nc) bquote(suppressWarnings(secr.fit(captdata, trace = FALSE, ncores = .(nc))))) microbenchmark(list = jobs, times = 10, unit = "seconds") # [edited output] # Unit: seconds # expr min lq mean median uq max neval # ncores = 2 1.75880 2.27978 2.6680 2.7450 3.0960 3.4334 10 # ncores = 4 1.13549 1.16280 1.6120 1.4431 2.0041 2.4018 10 # ncores = 6 0.88003 0.98215 1.2333 1.1387 1.5175 1.6966 10 # ncores = 8 0.78338 0.90033 1.5001 1.0406 1.2319 4.0669 10 ## sometimes (surprising) lack of benefit with ncores>2 msk <- make.mask(traps(ovenCH[[1]]), buffer = 300, nx = 25) jobs <- lapply(c(1,2,4,8), function(nc) bquote(secr.fit(ovenCH, trace = FALSE, ncores = .(nc), mask = msk))) microbenchmark(list = jobs, times = 10, unit = "seconds") # [edited output] # Unit: seconds # expr min lq mean median uq max neval # ncores = 1 12.5010 13.4951 15.674 15.304 16.373 21.723 10 # ncores = 2 10.0363 11.8634 14.407 13.726 16.782 22.966 10 # ncores = 4 8.6335 10.3422 13.085 12.449 15.729 17.914 10 # ncores = 8 8.5286 9.9008 10.751 10.736 10.796 14.885 10 ## and for simulation... jobs <- lapply(seq(2,8,2), function(nc) bquote(sim.secr(secrdemo.0, nsim = 20, tracelevel = 0, ncores = .(nc)))) microbenchmark(list = jobs, times = 10, unit = "seconds") # [edited output] # Unit: seconds # expr min lq mean median uq max neval # ncores = 2 48.610 49.932 59.032 52.485 54.730 119.905 10 # ncores = 4 29.480 29.996 30.524 30.471 31.418 31.612 10 # ncores = 6 22.583 23.594 24.148 24.354 24.644 25.388 10 # ncores = 8 19.924 20.651 25.581 21.002 21.696 51.920 10 ## and log-likelihood surface jobs <- lapply(seq(2,8,2), function(nc) bquote(suppressMessages(LLsurface(secrdemo.0, ncores = .(nc))))) microbenchmark(list = jobs, times = 10, unit = "seconds") # [edited output] # Unit: seconds # expr min lq mean median uq max neval # ncores = 2 20.941 21.098 21.290 21.349 21.471 21.619 10 # ncores = 4 14.982 15.125 15.303 15.263 15.449 15.689 10 # ncores = 6 13.994 14.299 14.529 14.342 14.458 16.515 10 # ncores = 8 13.597 13.805 13.955 13.921 14.128 14.353 10 ## End(Not run)
Compute spatially explicit net probability of detection for individual(s) at given coordinates (pdot).
pdot(X, traps, detectfn = 0, detectpar = list(g0 = 0.2, sigma = 25, z = 1), noccasions = NULL, binomN = NULL, userdist = NULL, ncores = NULL) CVpdot(..., conditional = FALSE)
pdot(X, traps, detectfn = 0, detectpar = list(g0 = 0.2, sigma = 25, z = 1), noccasions = NULL, binomN = NULL, userdist = NULL, ncores = NULL) CVpdot(..., conditional = FALSE)
X |
vector or 2-column matrix of coordinates |
traps |
|
detectfn |
integer code for detection function q.v. |
detectpar |
a named list giving a value for each parameter of detection function |
noccasions |
number of sampling intervals (occasions) |
binomN |
integer code for discrete distribution (see
|
userdist |
user-defined distance function or matrix (see userdist) |
ncores |
integer number of threads |
... |
arguments passed to |
conditional |
logical; if TRUE then computed mean and CV are conditional on detection |
If traps
has a usage attribute then noccasions
is
set accordingly; otherwise it must be provided.
The probability computed is where
the product is over the detectors in
traps
, excluding any not
used on a particular occasion. The per-occasion detection function
is halfnormal (0) by default, and is assumed not to vary
over the
occasions.
From 4.6.11, the detection parameters g0, lambda0 and sigma for point detectors may be detector- and occasion-specific. This is achieved by providing a vector of values that is replicated internally to fill a matrix with dimensions ntraps x noccasions (i.e. in trap order for occasion 1, then occasion 2 etc.)
For detection functions (10) and (11) the signal threshold ‘cutval’ should be
included in detectpar
, e.g., detectpar = list(beta0 = 103, beta1
= -0.11, sdS = 2, cutval = 52.5)
.
The calculation is not valid for single-catch traps because
is reduced by competition between animals.
userdist
cannot be set if ‘traps’ is any of polygon, polygonX,
transect or transectX. if userdist
is a function requiring
covariates or values of parameters ‘D’ or ‘noneuc’ then X
must
have a covariates attribute with the required columns.
Setting ncores = NULL
uses the existing value from the environment variable
RCPP_PARALLEL_NUM_THREADS (see setNumThreads
).
CVpdot
returns the expected mean and CV of pdot across the points listed in X
, assuming uniform population density. X
is usually a habitat mask. See Notes for details.
For pdot
, a vector of probabilities, one for each row in X.
For CVpdot
, a named vector with elements ‘meanpdot’ and ‘CVpdot’.
CVpdot
computes the mean and variance
of the location-specific overall detection probability
as follows.
For uniform density and conditional = FALSE
, is merely a scaling factor independent of
.
If conditional = TRUE
then .
The coefficient of variation is CV = .
secr
,
make.mask
,
Detection functions
,
pdotContour
,
CV
## Not run: temptrap <- make.grid() ## per-session detection probability for an individual centred ## at a corner trap. By default, noccasions = 5. pdot (c(0,0), temptrap, detectpar = list(g0 = 0.2, sigma = 25), noccasions = 5) msk <- make.mask(temptrap, buffer = 100) CVpdot(msk, temptrap, detectpar = list(g0 = 0.2, sigma = 25), noccasions = 5) ## End(Not run)
## Not run: temptrap <- make.grid() ## per-session detection probability for an individual centred ## at a corner trap. By default, noccasions = 5. pdot (c(0,0), temptrap, detectpar = list(g0 = 0.2, sigma = 25), noccasions = 5) msk <- make.mask(temptrap, buffer = 100) CVpdot(msk, temptrap, detectpar = list(g0 = 0.2, sigma = 25), noccasions = 5) ## End(Not run)
For a telemetry dataset, either as a standalone capthist object with
detector type ‘telemetryonly’ or the xylist attribute of a combined capthist
object resulting from addTelemetry
, determine the proportion of
fixes of each individual that lie within a set of polygons. Typically
used to obtain the proportion of fixes on a trapping grid, hence
‘proportion on grid’.
PG(CH, poly = NULL, includeNULL = FALSE, plt = FALSE, ...)
PG(CH, poly = NULL, includeNULL = FALSE, plt = FALSE, ...)
CH |
capthist object including telemetry locations |
poly |
polygon object (see |
includeNULL |
logical; if TRUE then missing values are returned for animals without telemetry data |
plt |
logical; if TRUE then poly and telemetry locations are plotted |
... |
other arguments passed to |
By default poly
is obtained by applying
bufferContour
with arguments ... to the traps
attribute of CH
. Note that either a positive buffer
argument or convex = TRUE
is needed for the polygon to have
area > 0.
If plt = TRUE
, bufferContour
is used to plot
poly
and the points are overplotted (open circles outside,
filled circles inside). To control the framing of the plot, create an
initial plot (e.g., with plot.traps, setting the border
argument) and use add = TRUE
(see Examples).
Numeric vector of proportions. If includeNULL = TRUE
length
equal to number of animals (rows) in CH
; otherwise length is
the number of animals for which there is telemetry data (because
xylist may cover only a subset of animals in CH
).
Grant, T. J. and Doherty, P. F. (2007) Monitoring of the flat-tailed horned lizard with methods incorporating detection probability. Journal of Wildlife Management 71, 1050–1056
addTelemetry
, bufferContour
,
pointsInPolygon
## Not run: olddir <- setwd('d:/density communication/combining telemetry and secr/possums') CvilleCH <- read.capthist('CVILLE summer captures 4occ.txt', 'CVILLE detectors summer 4occ.txt', detector = 'single') CvilleGPS <- read.telemetry('CVILLE GPS Combined 4occ.txt') CvilleGPSnew <- read.telemetry('CVILLE summer GPS New occasions.txt') setwd(olddir) CvilleBoth <- addTelemetry(CvilleCH, CvilleGPSnew) plot(CvilleBoth, border = 400) PG(CvilleBoth, buffer = 100, convex = TRUE, plt = TRUE, add = TRUE, col = 'red') ################################################################### ## this code computes an area-adjusted density estimate ## cf Grant and Doherty 2007 PGD <- function (CH, estimator = 'h2', ...) { pg <- PG(CH, ...) PGbar <- mean(pg) N <- closedN(CH, estimator) A <- polyarea(bufferContour(traps(CH), ...)[[1]]) Dhat <- N$Nhat / A * PGbar varDhat <- (N$Nhat^2 * var(pg) + PGbar^2 * N$seNhat^2) / A^2 c(Dhat = Dhat, seDhat = sqrt(varDhat)) } plot(traps(CvilleBoth), border = 400) PGD(CvilleBoth, buffer = 0, convex = TRUE, plt = TRUE, add = TRUE) PGD(CvilleBoth, est='null', buffer = 0, convex = TRUE, plt = FALSE) ################################################################### ## this code generates a PG summary for telemetry records randomly ## translated and rotated, keeping the centres within a habitat mask randomPG <- function(CH, poly = NULL, mask, reorient = TRUE, nrepl = 1, seed = 12345, ...) { moveone <- function(xy, newcentre) { xy <- sweep(xy,2,apply(xy,2,mean)) if (reorient) ## random rotation about centre xy <- rotate(xy, runif(1)*360) sweep(xy,2,unlist(newcentre), "+") } onerepl <- function(r) { ## r is dummy for replicate centres <- sim.popn(D = D, core = mask, model2D = "IHP", Ndist = "fixed") xyl <- mapply(moveone, xyl, split(centres, rownames(centres))) attr(CH, 'xylist') <- xyl ## substitute random placement PG(CH = CH , poly = poly, plt = FALSE, ...) } set.seed(seed) if (!requireNamespace('sf')) stop ("requires package sf") if (is.null(poly)) { poly <- bufferContour (traps(CH), ...) poly <- lapply(poly, as.matrix) poly <- sf::st_sfc(sf::st_polygon(poly)) } xyl <- telemetryxy(CH) D <- length(xyl) / maskarea(mask) sapply(1:nrepl, onerepl) } mask <- make.mask (traps(CvilleBoth), buffer = 400, type = "trapbuffer") pg <- randomPG (CvilleBoth, mask = mask, buffer = 100, convex = TRUE, nrepl = 20) apply(pg, 1, mean) ################################################################### ## End(Not run)
## Not run: olddir <- setwd('d:/density communication/combining telemetry and secr/possums') CvilleCH <- read.capthist('CVILLE summer captures 4occ.txt', 'CVILLE detectors summer 4occ.txt', detector = 'single') CvilleGPS <- read.telemetry('CVILLE GPS Combined 4occ.txt') CvilleGPSnew <- read.telemetry('CVILLE summer GPS New occasions.txt') setwd(olddir) CvilleBoth <- addTelemetry(CvilleCH, CvilleGPSnew) plot(CvilleBoth, border = 400) PG(CvilleBoth, buffer = 100, convex = TRUE, plt = TRUE, add = TRUE, col = 'red') ################################################################### ## this code computes an area-adjusted density estimate ## cf Grant and Doherty 2007 PGD <- function (CH, estimator = 'h2', ...) { pg <- PG(CH, ...) PGbar <- mean(pg) N <- closedN(CH, estimator) A <- polyarea(bufferContour(traps(CH), ...)[[1]]) Dhat <- N$Nhat / A * PGbar varDhat <- (N$Nhat^2 * var(pg) + PGbar^2 * N$seNhat^2) / A^2 c(Dhat = Dhat, seDhat = sqrt(varDhat)) } plot(traps(CvilleBoth), border = 400) PGD(CvilleBoth, buffer = 0, convex = TRUE, plt = TRUE, add = TRUE) PGD(CvilleBoth, est='null', buffer = 0, convex = TRUE, plt = FALSE) ################################################################### ## this code generates a PG summary for telemetry records randomly ## translated and rotated, keeping the centres within a habitat mask randomPG <- function(CH, poly = NULL, mask, reorient = TRUE, nrepl = 1, seed = 12345, ...) { moveone <- function(xy, newcentre) { xy <- sweep(xy,2,apply(xy,2,mean)) if (reorient) ## random rotation about centre xy <- rotate(xy, runif(1)*360) sweep(xy,2,unlist(newcentre), "+") } onerepl <- function(r) { ## r is dummy for replicate centres <- sim.popn(D = D, core = mask, model2D = "IHP", Ndist = "fixed") xyl <- mapply(moveone, xyl, split(centres, rownames(centres))) attr(CH, 'xylist') <- xyl ## substitute random placement PG(CH = CH , poly = poly, plt = FALSE, ...) } set.seed(seed) if (!requireNamespace('sf')) stop ("requires package sf") if (is.null(poly)) { poly <- bufferContour (traps(CH), ...) poly <- lapply(poly, as.matrix) poly <- sf::st_sfc(sf::st_polygon(poly)) } xyl <- telemetryxy(CH) D <- length(xyl) / maskarea(mask) sapply(1:nrepl, onerepl) } mask <- make.mask (traps(CvilleBoth), buffer = 400, type = "trapbuffer") pg <- randomPG (CvilleBoth, mask = mask, buffer = 100, convex = TRUE, nrepl = 20) apply(pg, 1, mean) ################################################################### ## End(Not run)
Display a plot of detection (capture) histories or telemetry data over a map of the detectors.
## S3 method for class 'capthist' plot(x, rad = 5, hidetraps = FALSE, tracks = FALSE, title = TRUE, subtitle = TRUE, add = FALSE, varycol = TRUE, icolours = NULL, randcol = FALSE, lab1cap = FALSE, laboffset = 4, ncap = FALSE, splitocc = NULL, col2 = "green", type = c("petal", "n.per.detector", "n.per.cluster", "sightings", "centres", "telemetry", "nontarget"), cappar = list(cex = 1.3, pch = 16, col = "blue"), trkpar = list(col = "blue", lwd = 1), labpar = list(cex = 0.7, col = "black"), ...) plotMCP(x, add = FALSE, col = "black", fill = NA, lab1cap = FALSE, laboffset = 4, ncap = FALSE, ...)
## S3 method for class 'capthist' plot(x, rad = 5, hidetraps = FALSE, tracks = FALSE, title = TRUE, subtitle = TRUE, add = FALSE, varycol = TRUE, icolours = NULL, randcol = FALSE, lab1cap = FALSE, laboffset = 4, ncap = FALSE, splitocc = NULL, col2 = "green", type = c("petal", "n.per.detector", "n.per.cluster", "sightings", "centres", "telemetry", "nontarget"), cappar = list(cex = 1.3, pch = 16, col = "blue"), trkpar = list(col = "blue", lwd = 1), labpar = list(cex = 0.7, col = "black"), ...) plotMCP(x, add = FALSE, col = "black", fill = NA, lab1cap = FALSE, laboffset = 4, ncap = FALSE, ...)
x |
an object of class |
rad |
radial displacement of dot indicating each capture event from the detector location (used to separate overlapping points) |
hidetraps |
logical indicating whether trap locations should be displayed |
tracks |
logical indicating whether consecutive locations of individual animals should be joined by a line |
title |
logical or character string for title |
subtitle |
logical or character string for subtitle |
add |
logical for whether to add to existing plot |
varycol |
logical for whether to distinguish individuals by colour |
icolours |
vector of individual colours (when |
randcol |
logical to use random colours ( |
lab1cap |
logical for whether to label the first capture of each animal |
laboffset |
distance by which to offset labels from points |
ncap |
logical to display the number of detections per trap per occasion |
splitocc |
optional occasion from which second colour is to be used |
col2 |
second colour (used with |
type |
character string ("petal", "n.per.detector" or "n.per.cluster") |
cappar |
list of named graphical parameters for detections (passed to |
trkpar |
list of named graphical parameters for tracks (passed to |
labpar |
list of named graphical parameters for labels (passed to |
... |
arguments passed to |
col |
vector of line colour numbers or names (plotMCP only) |
fill |
vector of fill colour numbers or names (plotMCP only) |
By default, a ‘petal’ plot is generated in the style of Density (Efford 2012)
using eqscplot
from the MASS library.
If type =
"n.per.detector"
or type = "n.per.cluster"
the result is a
colour-coded plot of the number of individuals at each unit, pooled over
occasions.
If type = "sightings"
the sightings of unmarked animals are
displayed on a petal-like plot (requires mark-resight data) (see also sightingPlot
).
If type = "centres"
then a single point is plotted for each animal, jittered on each axis by a random amount (limits +/- rad
/2).
If type = "telemetry"
and the ‘telemetryxy’ attribute is not NULL then the telemetry locations are plotted.
If type = "nontarget"
and the ‘nontarget’ attribute is not NULL then the nontarget captures or interference events are plotted.
If title
= FALSE no title is displayed; if title
= TRUE,
the session identifer is used for the title.
If subtitle
= FALSE no subtitle is displayed; if subtitle
= TRUE, the subtitle gives the numbers of occasions, detections and
individuals.
If x
is a multi-session capthist object then a separate plot is
produced for each session. Use par(mfrow = c(nr, nc))
to allow a
grid of plots to be displayed simultaneously (nr rows x nc columns).
These arguments are used only for petal plots: rad
,
tracks
, varycol
, randcol
, lab1cap
,
laboffset
, ncap
, splitocc
, col2
,
trkpar
, and labpar
. Call occasionKey
to add a key to the petals. From 5.0.1 a warning is issued if rad exceeds
3% of the detector span in either x- or y-dimensions.
If icolours = NULL
and varycol = TRUE
then a vector of
colours is generated automatically as topo.colors((nrow(x)+1) * 1.5).
If there are too few values in icolours
for the number of
individuals then colours will be re-used.
plotMCP
plots minimum convex polygons of individual location
data over a base plot of detector locations. Usually the data are
telemetry locations in the xylist attribute of the capthist
object; if this is not present and x
is a polygon search
capthist then the individual xy data are plotted.
For type = "petal"
, the number of detections in x
.
For type = "sightings"
, the number of sightings of unmarked animals in x
.
For type = "n.per.detector"
or type = "n.per.cluster"
, a
dataframe with data for a legend (see Examples).
plotMCP
invisibly returns a list in which each component is a
2-column (x,y) dataframe of boundary coordinates for one individual.
Efford, M. G. (2012) DENSITY 5.0: software for spatially explicit capture–recapture. Department of Mathematics and Statistics, University of Otago, Dunedin, New Zealand. https://www.otago.ac.nz/density/.
capthist
,
occasionKey
,
plot.traps
,
sightingPlot
demotrap <- make.grid() tempcapt <- sim.capthist(demotrap, popn = list(D = 5, buffer = 50), detectpar = list(g0 = 0.15, sigma = 30)) plot(tempcapt, border = 10, rad = 3, tracks = TRUE, lab1cap = TRUE, laboffset = 2.5) ## type = n.per.cluster ## generate some captures testregion <- data.frame(x = c(0,2000,2000,0), y = c(0,0,2000,2000)) popn <- sim.popn (D = 10, core = testregion, buffer = 0, model2D = "hills", details = list(hills = c(-2,3))) t1 <- make.grid(nx = 1, ny = 1) t1.100 <- make.systematic (cluster = t1, spacing = 100, region = testregion) capt <- sim.capthist(t1.100, popn = popn, noccasions = 1) ## now plot captures ... temp <- plot(capt, title = "Individuals per cluster", type = "n.per.cluster", hidetraps = FALSE, gridlines = FALSE, cappar = list(cex = 1.5)) if (interactive()) { ## add legend; click on map to place top left corner legend (locator(1), pch = 21, pt.bg = temp$colour, pt.cex = 1.3, legend = temp$legend, cex = 0.8) } ## Not run: ## try varying individual colours - requires RColorBrewer library(RColorBrewer) plot(infraCH[[2]], icolours = brewer.pal(12, "Set3"), tracks = TRUE, bg = "black", cappar = list(cex = 2), border = 10, rad = 2, gridlines = FALSE) ## generate telemetry data te <- make.telemetry() tr <- make.grid(detector = "proximity") totalpop <- sim.popn(tr, D = 20, buffer = 100) tepop <- subset(totalpop, runif(nrow(totalpop)) < 0.05) teCH <- sim.capthist(te, popn = tepop, renumber=FALSE, detectfn = "HHN", detectpar = list(lambda0 = 3, sigma = 25)) plot(teCH, type = 'telemetry', tracks = TRUE) ## simple "centres" example ## polygon data require 'hazard' detection function 14:19 CH <- sim.capthist(make.poly(), nocc = 20, detectfn = 'HHN', detectpar = list(lambda0 = 1, sigma = 10)) plot(CH, cappar = list(col = 'orange'), varycol = FALSE, border = 10) plot(CH, type = 'centres', add = TRUE, rad = 0) ## End(Not run)
demotrap <- make.grid() tempcapt <- sim.capthist(demotrap, popn = list(D = 5, buffer = 50), detectpar = list(g0 = 0.15, sigma = 30)) plot(tempcapt, border = 10, rad = 3, tracks = TRUE, lab1cap = TRUE, laboffset = 2.5) ## type = n.per.cluster ## generate some captures testregion <- data.frame(x = c(0,2000,2000,0), y = c(0,0,2000,2000)) popn <- sim.popn (D = 10, core = testregion, buffer = 0, model2D = "hills", details = list(hills = c(-2,3))) t1 <- make.grid(nx = 1, ny = 1) t1.100 <- make.systematic (cluster = t1, spacing = 100, region = testregion) capt <- sim.capthist(t1.100, popn = popn, noccasions = 1) ## now plot captures ... temp <- plot(capt, title = "Individuals per cluster", type = "n.per.cluster", hidetraps = FALSE, gridlines = FALSE, cappar = list(cex = 1.5)) if (interactive()) { ## add legend; click on map to place top left corner legend (locator(1), pch = 21, pt.bg = temp$colour, pt.cex = 1.3, legend = temp$legend, cex = 0.8) } ## Not run: ## try varying individual colours - requires RColorBrewer library(RColorBrewer) plot(infraCH[[2]], icolours = brewer.pal(12, "Set3"), tracks = TRUE, bg = "black", cappar = list(cex = 2), border = 10, rad = 2, gridlines = FALSE) ## generate telemetry data te <- make.telemetry() tr <- make.grid(detector = "proximity") totalpop <- sim.popn(tr, D = 20, buffer = 100) tepop <- subset(totalpop, runif(nrow(totalpop)) < 0.05) teCH <- sim.capthist(te, popn = tepop, renumber=FALSE, detectfn = "HHN", detectpar = list(lambda0 = 3, sigma = 25)) plot(teCH, type = 'telemetry', tracks = TRUE) ## simple "centres" example ## polygon data require 'hazard' detection function 14:19 CH <- sim.capthist(make.poly(), nocc = 20, detectfn = 'HHN', detectpar = list(lambda0 = 1, sigma = 10)) plot(CH, cappar = list(col = 'orange'), varycol = FALSE, border = 10) plot(CH, type = 'centres', add = TRUE, rad = 0) ## End(Not run)
Plot a habitat mask either as points or as an image
plot. Colours maybe used to show the value of one mask covariate.
## S3 method for class 'mask' plot(x, border = 20, add = FALSE, covariate = NULL, axes = FALSE, dots = TRUE, col = "grey", breaks = 10, meshcol = NA, ppoly = TRUE, polycol = "red", legend = TRUE, ...) ## S3 method for class 'Dsurface' plot(x, covariate, group = NULL, plottype = "shaded", scale = 1, ...) ## S3 method for class 'Rsurface' plot(x, covariate = "Resource", plottype = "shaded", scale = 1, ...) spotHeight (object, prefix = NULL, dec = 2, point = FALSE, text = TRUE, sep = ", ", session = 1, scale = 1, ...)
## S3 method for class 'mask' plot(x, border = 20, add = FALSE, covariate = NULL, axes = FALSE, dots = TRUE, col = "grey", breaks = 10, meshcol = NA, ppoly = TRUE, polycol = "red", legend = TRUE, ...) ## S3 method for class 'Dsurface' plot(x, covariate, group = NULL, plottype = "shaded", scale = 1, ...) ## S3 method for class 'Rsurface' plot(x, covariate = "Resource", plottype = "shaded", scale = 1, ...) spotHeight (object, prefix = NULL, dec = 2, point = FALSE, text = TRUE, sep = ", ", session = 1, scale = 1, ...)
x , object
|
mask or Dsurface object |
border |
width of blank display border (metres) |
add |
logical for adding mask points to an existing plot |
covariate |
name (as character string in quotes) or column number of a covariate to use for colouring |
axes |
logical for plotting axes |
dots |
logical for plotting mask points as dots, rather than as square pixels |
col |
colour(s) to use for plotting |
breaks |
an integer or a numeric vector – see |
meshcol |
colour for pixel borders (NA for none) |
ppoly |
logical for whether the bounding polygon should be plotted (if ‘poly’ specified) |
polycol |
colour for outline of polygon ( |
legend |
logical; if TRUE a legend is plotted |
... |
other arguments passed to |
group |
group for which plot required, if more than 1 |
plottype |
character string c("dots", "shaded", "contour", "persp") |
scale |
numeric multiplier for density or other numeric covariate
(see |
prefix |
character vector for name(s) of covariate(s) to retrieve |
dec |
number of decimal places for rounding density |
point |
logical for whether to plot point |
text |
logical for whether to place density label on plot |
sep |
character separator for elements if length(prefix)>1 |
session |
session number or identifier |
The argument dots
of plot.mask
selects between two
distinct types of plot (dots and shaded (coloured) pixels).
plot.Dsurface
and plot.Rsurface
offer contour and
perspective plots in addition to the options in plot.mask
. It may
take some experimentation to get what you want - see
contour
and persp
.
For plot.Dsurface the default value of ‘covariate’ is ‘D’ unless the Dsurface has a ‘parameter’ attribute of ‘noneuc’,
If using a covariate or Dsurface or Rsurface to colour dots or pixels, the
col
argument should be a colour vector of length equal to the
number of levels (the default palette from 2.9.0 is terrain.colors
, and this
palette will also be used whenever there are too few levels in the
palette provided; see Notes for more on palettes). Border lines around
pixels are drawn in ‘meshcol’. Set this to NA to eliminate pixel
borders.
If a covariate
is specified in a call to plot.Dsurface
then
that covariate will be plotted instead of density. This is a handy way
to contour a covariate (contouring is not available in plot.mask
).
If ‘breaks’ is an integer then the range of the covariate is divided into this number of equal intervals. Alternatively, ‘breaks’ may be a vector of break points (length one more than the number of intervals). This gives more control and often ‘prettier’
spotHeight
may be used to interrogate a plot produced with
plot.Dsurface
or plot.Rsurface
, or by plot.mask
if
the mask has covariates. prefix
defaults to ‘density.’ for
Dsurface objects and to '' (all covariates) for mask objects. The
predicted density or covariate at the nearest point is returned when the
user clicks on the plot. Multiple values may be displayed (e.g.,
prefix = c("lcl","ucl")
if Dsurface includes confidence
limits). Click outside the mask or hit the Esc key to
end. spotHeight
deals with one session at a time.
Legend plotting is enabled only when a covariate is specified. It uses
legend
when dots = TRUE
and
strip.legend
otherwise.
If covariate
is specified and plottype = "shaded"
then
plot.mask
invisibly returns a character vector of the intervals
defined by ‘breaks’ (useful for plotting a legend).
If plottype = "persp"
then plot.mask
invisibly returns a the
perspective matrix that may be used to add to the plot with
trans3d
.
spotHeight
invisibly returns a dataframe of the extracted
values and their coordinates.
plot.mask()
acquired the argument ‘legend’ in version 2.9.0,
and other changes (e.g., breaks = 10
) may alter the output.
Contouring requires a rectangular grid; if a Dsurface is not
rectangular then plot.Dsurface with plottype = "contour"
triggers a call to
rectangularMask
.
The colour palettes topo.colors
, heat.colors
and
terrain.colors
may be viewed with the demo.pal
function in
the Examples code of their help page palettes.
The package RColorBrewer is a good source of palettes. Try
display.brewer.all()
and e.g., col = brewer.pal(7, "YlGn")
.
colours
,
mask
,
Dsurface
,
rectangularMask
,
contour
persp
strip.legend
# simple temptrap <- make.grid() tempmask <- make.mask(temptrap) plot (tempmask) ## Not run: ## restrict to points over an arbitrary detection threshold, ## add covariate, plot image and overlay traps tempmask <- subset(tempmask, pdot(tempmask, temptrap, noccasions = 5)>0.001) covariates (tempmask) <- data.frame(circle = exp(-(tempmask$x^2 + tempmask$y^2)/10000) ) plot (tempmask, covariate = "circle", dots = FALSE, axes = TRUE, add = TRUE, breaks = 8, col = terrain.colors(8), mesh = NA) plot (temptrap, add = TRUE) ## add a legend par(cex = 0.9) covrange <- range(covariates(tempmask)$circle) step <- diff(covrange)/8 colourlev <- terrain.colors(9) zlev <- format(round(seq(covrange[1],covrange[2],step),2)) legend (x = "topright", fill = colourlev, legend = zlev, y.intersp = 0.8, title = "Covariate") title("Colour mask points with p.(X) > 0.001") mtext(side=3,line=-1, "g0 = 0.2, sigma = 20, nocc = 5") ## Waitarere possum density surface extrapolated across region regionmask <- make.mask(traps(possumCH), buffer = 1000, spacing = 10, poly = possumremovalarea) dts <- distancetotrap(regionmask, possumarea) covariates(regionmask) <- data.frame(d.to.shore = dts) shorePossums <- predictDsurface(possum.model.Ds, regionmask) ## plot as coloured pixels with white lines colourlev <- terrain.colors(7) plot(shorePossums, breaks = seq(0,3.5,0.5), plottype = "shaded", poly = FALSE, col = colourlev, mesh = NA) plot(traps(possumCH), add = TRUE, detpar = list(col = "black")) polygon(possumremovalarea) ## check some point densities spotHeight(shorePossums, dec = 1, col = "black") ## add a legend zlev <- format(seq(0,3,0.5), digits = 1) legend (x = "topright", fill = colourlev, legend = paste(zlev,"--"), y.intersp = 1, title = "Density / ha") ## End(Not run)
# simple temptrap <- make.grid() tempmask <- make.mask(temptrap) plot (tempmask) ## Not run: ## restrict to points over an arbitrary detection threshold, ## add covariate, plot image and overlay traps tempmask <- subset(tempmask, pdot(tempmask, temptrap, noccasions = 5)>0.001) covariates (tempmask) <- data.frame(circle = exp(-(tempmask$x^2 + tempmask$y^2)/10000) ) plot (tempmask, covariate = "circle", dots = FALSE, axes = TRUE, add = TRUE, breaks = 8, col = terrain.colors(8), mesh = NA) plot (temptrap, add = TRUE) ## add a legend par(cex = 0.9) covrange <- range(covariates(tempmask)$circle) step <- diff(covrange)/8 colourlev <- terrain.colors(9) zlev <- format(round(seq(covrange[1],covrange[2],step),2)) legend (x = "topright", fill = colourlev, legend = zlev, y.intersp = 0.8, title = "Covariate") title("Colour mask points with p.(X) > 0.001") mtext(side=3,line=-1, "g0 = 0.2, sigma = 20, nocc = 5") ## Waitarere possum density surface extrapolated across region regionmask <- make.mask(traps(possumCH), buffer = 1000, spacing = 10, poly = possumremovalarea) dts <- distancetotrap(regionmask, possumarea) covariates(regionmask) <- data.frame(d.to.shore = dts) shorePossums <- predictDsurface(possum.model.Ds, regionmask) ## plot as coloured pixels with white lines colourlev <- terrain.colors(7) plot(shorePossums, breaks = seq(0,3.5,0.5), plottype = "shaded", poly = FALSE, col = colourlev, mesh = NA) plot(traps(possumCH), add = TRUE, detpar = list(col = "black")) polygon(possumremovalarea) ## check some point densities spotHeight(shorePossums, dec = 1, col = "black") ## add a legend zlev <- format(seq(0,3,0.5), digits = 1) legend (x = "topright", fill = colourlev, legend = paste(zlev,"--"), y.intersp = 1, title = "Density / ha") ## End(Not run)
Plot, summary and print methods for MCgof objects.
## S3 method for class 'MCgof' plot(x, counts = 'all', overlay = NULL, maxT = NULL, main = NULL, cex = 0.9, ...) ## S3 method for class 'MCgof' hist(x, counts = 'all', maxT = NULL, main = NULL, cex = 0.9, ...) ## S3 method for class 'MCgof' summary(object, ...) ## S3 method for class 'MCgof' print(x, ...)
## S3 method for class 'MCgof' plot(x, counts = 'all', overlay = NULL, maxT = NULL, main = NULL, cex = 0.9, ...) ## S3 method for class 'MCgof' hist(x, counts = 'all', maxT = NULL, main = NULL, cex = 0.9, ...) ## S3 method for class 'MCgof' summary(object, ...) ## S3 method for class 'MCgof' print(x, ...)
x |
MCgof object |
counts |
character vector of marginal counts for which statistics are to be plotted |
overlay |
MCgof object |
maxT |
numeric maximum plotted value of statistic |
main |
character vector of labels (see Details) |
cex |
numeric size of labels and points |
... |
other arguments passed by the plot method to |
object |
MCgof object |
We start with a 3-D capthist array with dimensions corresponding to individuals (i), occasions (j) and detectors (k). The possible marginal counts for the default ‘statfn’ in MCgof
are designated –
Count | Margin | Cell value |
yik | individual x detector | |
yi | individual | |
yk | detector | |
The plot method displays a scatterplot of discrepancies for observed and simulated data (one point per replicate) (Gelman et al. 1996).
If ‘overlay’ is provided then the results are overlaid on the initial plot. Points should be distinguished by specifying a different colour (col) or symbol (pch) with the ... argument.
‘main’ is a vector of labels used as headers; the names should include all components of ‘statfn’. Setting main = "" suppresses headers.
The hist method displays a histogram of the ratio Tobs/Tsim.
The summary method returns a matrix of values in which the columns correspond to the different statistics (default yik, yi, yk) and the rows are
median discrepancy Tobs
median discrepancy Tsim
proportion Tobs>Tsim
number of valid results
Gelman, A., Meng, X.-L., and Stern, H. (1996) Posterior predictive assessment of model fitness via realized discrepancies. Statistica Sinica 6, 733–807.
Display animal locations from a popn
object.
## S3 method for class 'popn' plot(x, add = FALSE, frame = TRUE, circles = NULL, collapse = FALSE, seqcol = NULL, ...)
## S3 method for class 'popn' plot(x, add = FALSE, frame = TRUE, circles = NULL, collapse = FALSE, seqcol = NULL, ...)
x |
object of class |
add |
logical to add points to an existing plot |
frame |
logical to add frame or polygon within which points were simulated |
circles |
vector giving the radii if circles are to be plotted |
collapse |
logical; if TRUE then multiple sessions are overlaid |
seqcol |
color used for first detection when collapse = TRUE (optional) |
... |
arguments passed to |
If circles
is provided then a circle of the given radius is
plotted for each animal using the symbols
function. The
arguments fg
and bg
may be used to control the colour of
the perimeter and the fill of each circle (see Examples).
For a multi-session popn with turnover
, collapse = TRUE
allows successive locations to be joined with (type = 'o'
or type = 'l'
).
seqcol
may be a single color, a vector of colours (one per session),
or a vector of two colours, one for the first and one for all later sessions
in which each animal was detected.
If ... includes 'col' then 'collapse' must be specified to avoid confusion, even for single-session data (see Examples).
temppopn <- sim.popn(D = 5, expand.grid( x = c(0,100), y = c(0,100))) # specify collapse to avoid partial match of col plot(temppopn, pch = 16, collapse = FALSE, col = "blue") plot(temppopn, circles = 20, bg = "tan", fg = "white") plot(temppopn, pch = 16, cex = 0.5, add = TRUE)
temppopn <- sim.popn(D = 5, expand.grid( x = c(0,100), y = c(0,100))) # specify collapse to avoid partial match of col plot(temppopn, pch = 16, collapse = FALSE, col = "blue") plot(temppopn, circles = 20, bg = "tan", fg = "white") plot(temppopn, pch = 16, cex = 0.5, add = TRUE)
Plot detection functions using estimates of parameters in an secr object, or as provided by the user.
## S3 method for class 'secr' plot(x, newdata = NULL, add = FALSE, sigmatick = FALSE, rgr = FALSE, limits = FALSE, alpha = 0.05, xval = 0:200, ylim = NULL, xlab = NULL, ylab = NULL, ...) ## S3 method for class 'secrlist' plot(x, newdata = NULL, add = FALSE, sigmatick = FALSE, rgr = FALSE, limits = FALSE, alpha = 0.05, xval = 0:200, ylim = NULL, xlab = NULL, ylab = NULL, ..., overlay = TRUE) detectfnplot (detectfn, pars, details = NULL, add = FALSE, sigmatick = FALSE, rgr = FALSE, hazard = FALSE, xval = 0:200, ylim = NULL, xlab = NULL, ylab = NULL, ...) attenuationplot (pars, add = FALSE, spherical = TRUE, xval = 0:200, ylim = NULL, xlab = NULL, ylab = NULL, ...)
## S3 method for class 'secr' plot(x, newdata = NULL, add = FALSE, sigmatick = FALSE, rgr = FALSE, limits = FALSE, alpha = 0.05, xval = 0:200, ylim = NULL, xlab = NULL, ylab = NULL, ...) ## S3 method for class 'secrlist' plot(x, newdata = NULL, add = FALSE, sigmatick = FALSE, rgr = FALSE, limits = FALSE, alpha = 0.05, xval = 0:200, ylim = NULL, xlab = NULL, ylab = NULL, ..., overlay = TRUE) detectfnplot (detectfn, pars, details = NULL, add = FALSE, sigmatick = FALSE, rgr = FALSE, hazard = FALSE, xval = 0:200, ylim = NULL, xlab = NULL, ylab = NULL, ...) attenuationplot (pars, add = FALSE, spherical = TRUE, xval = 0:200, ylim = NULL, xlab = NULL, ylab = NULL, ...)
x |
an |
newdata |
dataframe of data to form estimates |
add |
logical to add curve(s) to an existing plot |
sigmatick |
logical; if TRUE the scale parameter sigma is shown by a vertical line |
rgr |
logical; if TRUE a scaled curve r.g(r) is plotted instead of g(r) |
hazard |
logical; if TRUE the hazard of detection is plotted instead of probability |
limits |
logical; if TRUE pointwise confidence limits are drawn |
alpha |
alpha level for confidence intervals |
xval |
vector of distances at for which detection to be plotted |
ylim |
vector length 2 giving limits of y axis |
xlab |
label for x axis |
ylab |
label for y axis |
... |
arguments to pass to |
overlay |
logical; if TRUE then automatically |
detectfn |
integer code or character string for shape of detection function 0 = halfnormal etc. – see detectfn |
pars |
list, vector or matrix of parameter values |
details |
list of ancillary parameters |
spherical |
logical for whether to include spherical spreading term |
newdata
is usually NULL, in which case one curve is plotted for
each session and group. Otherwise, predict.secr
is used to form
estimates and plot a curve for each row in newdata
.
If axis labels are not provided they default to ‘Distance (m)’ and ‘Detection probability’ or ‘Detection lambda’.
detectfnplot
is an alternative in which the user nominates the
type of function and provides parameter values. pars
maybe a list
as from detectpar
; it is first coerced to a numeric vector
with unlist
. Parameter values must be in the expected order
(e.g. g0, sigma, z). If pars
is a matrix then a separate
curve is plotted with the parameter values in each row.
For detectfnplot
the signal threshold parameters ‘cutval’ and
‘spherical’ should be provided in details
(see examples).
Approximate confidence limits for g(r) are calculated using a numerical
first-order delta-method approximation to the standard error at each
xval
. The distribution of g(r) is assumed to be normal on the logit scale for non-hazard functions (detectfn 0:13). For hazard detection functions (detectfn 14:18) the hazard is assumed (from version 3.1.1) to be distributed normally on the log scale. Limits are back-transformed to the probability scale g(r).
attenuationplot
plots the expected decline in signal strength
with distance, given parameters and
for a log-linear model of sound attenuation.
plot.secr
invisibly returns a dataframe of the plotted values (or
a list of dataframes in the case that newdata
has more than one
row).
Detection functions
, plot
, secr
plot (secrdemo.b, xval = 0:100, ylim = c(0, 0.4)) ## Add recapture probability plot (secrdemo.b, newdata = data.frame(b = 1), add = TRUE, col = "red") ## signal strength detection: 70dB at source, attenuation ## 0.3dB/m, sdS 5dB; detection threshold 40 dB. detectfnplot (detectfn = 10, c(70, -0.3, 5), details = list(cutval = 40)) ## add a function with louder source and spherical spreading... detectfnplot (detectfn = 11, c(110, -0.3, 5), details = list(cutval = 40), add = TRUE, col = "red") ## matching sound attenuation curves; `spherical-only' dashed line attenuationplot (c(70, -0.3), spherical = FALSE, ylim=c(-10,110)) attenuationplot (c(110, 0), spherical = TRUE, add=TRUE, lty=2) attenuationplot (c(110, -0.3), spherical = TRUE, add = TRUE, col = "red")
plot (secrdemo.b, xval = 0:100, ylim = c(0, 0.4)) ## Add recapture probability plot (secrdemo.b, newdata = data.frame(b = 1), add = TRUE, col = "red") ## signal strength detection: 70dB at source, attenuation ## 0.3dB/m, sdS 5dB; detection threshold 40 dB. detectfnplot (detectfn = 10, c(70, -0.3, 5), details = list(cutval = 40)) ## add a function with louder source and spherical spreading... detectfnplot (detectfn = 11, c(110, -0.3, 5), details = list(cutval = 40), add = TRUE, col = "red") ## matching sound attenuation curves; `spherical-only' dashed line attenuationplot (c(70, -0.3), spherical = FALSE, ylim=c(-10,110)) attenuationplot (c(110, 0), spherical = TRUE, add=TRUE, lty=2) attenuationplot (c(110, -0.3), spherical = TRUE, add = TRUE, col = "red")
Map the locations of detectors (traps).
## S3 method for class 'traps' plot(x, border = 100, label = FALSE, offset = c(6,6), add = FALSE, hidetr = FALSE, detpar = list(), txtpar = list(), bg = "white", gridlines = !add, gridspace = 100, gridcol = "grey", markused = FALSE, markvarying = FALSE, markvertices = FALSE, labelclusters = FALSE, frame = NULL, ...)
## S3 method for class 'traps' plot(x, border = 100, label = FALSE, offset = c(6,6), add = FALSE, hidetr = FALSE, detpar = list(), txtpar = list(), bg = "white", gridlines = !add, gridspace = 100, gridcol = "grey", markused = FALSE, markvarying = FALSE, markvertices = FALSE, labelclusters = FALSE, frame = NULL, ...)
x |
a |
border |
width of blank margin around the outermost detectors |
label |
logical indicating whether a text label should appear by each detector |
offset |
vector displacement of label from point on x and y axes |
add |
logical to add detectors to an existing plot |
hidetr |
logical to suppress plotting of detectors |
detpar |
list of named graphical parameters for detectors (passed to |
txtpar |
list of named graphical parameters for labels (passed to |
bg |
background colour |
gridlines |
logical for plotting grid lines |
gridspace |
spacing of gridlines |
gridcol |
colour of gridlines |
markused |
logical to distinguish detectors used on at least one occasion |
markvarying |
logical to distinguish detectors whose usage varies among occasions |
markvertices |
logical or 0,1,2 for plotting transect or polygon points |
labelclusters |
logical to label clusters |
frame |
data defining a boundary polygon (see |
... |
arguments to pass to |
offset
may also be a scalar value for equal displacement on the x
and y axes. The hidetr
option is most likely to be used when
plot.traps is called by plot.capthist. See par
and
colours
for more information on setting graphical
parameters. The initial values of graphical parameters are restored on
exit.
Axes are not labeled. Use axis
and mtext
if
necessary.
markvertices
determines whether the vertices of each transect or
polygon will be emphasised by overplotting a point symbol
(detpar$pch). Value may be logical (TRUE, FALSE) or integer (0 = no
points, 1 = terminal vertices only, 2 = all vertices).
From 4.4.0, polygon detectors are shaded with detpar$col and outlined (border) with detpar$fg. Use detpar$col = NA for no shading (transparent polygons).
labelclusters
requires x
to have attributes ‘clusterID’ and
‘clustertrap’.
A boundary polygon is plotted in black if frame
is not NULL.
None
temptrap <- make.grid() plot (temptrap, detpar = list(pch = 16, col = "blue"), label = TRUE, offset = 7)
temptrap <- make.grid() plot (temptrap, detpar = list(pch = 16, col = "blue"), label = TRUE, offset = 7)
Plots the outer edge of a mask.
plotMaskEdge(mask, plt = TRUE, add = FALSE, ...)
plotMaskEdge(mask, plt = TRUE, add = FALSE, ...)
mask |
secr habitat mask object |
plt |
logical; if TRUE the edge is plotted |
add |
logical; if TRUE the line is added to an existing plot |
... |
other line plotting arguments passed to |
May be slow.
A numeric matrix of 4 columns is returned invisibly. The columns may be used as arguments x0, y0, x1, y1 in a call to segments().
A bug in secr <3.2.2 caused some internal lines to appear when the mask spacing was not an integer.
## Not run: plot(possummask) plotMaskEdge (possummask, add = TRUE) ## End(Not run)
## Not run: plot(possummask) plotMaskEdge (possummask, add = TRUE) ## End(Not run)
Compute the profile likelihood of a finite mixture model for a user-specified range of values for the mixing parameter. This provides a check on multimodality.
pmixProfileLL(CH, model = list(g0 ~ h2, sigma ~ h2), CL = TRUE, pmvals = seq(0.01, 0.99, 0.01), pmi = 5, ...)
pmixProfileLL(CH, model = list(g0 ~ h2, sigma ~ h2), CL = TRUE, pmvals = seq(0.01, 0.99, 0.01), pmi = 5, ...)
CH |
capthist object |
model |
model as in |
CL |
logical as in in |
pmvals |
numeric vector of values for mixing parameter ‘pmix’ |
pmi |
integer index of ‘pmix’ in vector of coefficients (beta parameters) for the specified model |
... |
other arguments passed to |
Choosing the wrong value for pmi results in the error message "invalid fixed beta - require NP-vector". The easiest way to find the value of pmi
is to inspect the
output from a previously fitted mixture model - either count the coefficients
or check fit$parindx$pmix (for a model named ‘fit’). It is assumed that ‘pmix’ is the last real
parameter in the model, and that pmix is constant.
Numeric vector of profile likelihoods.
This is slow to execute and the results are hard to interpret. Use only if you are confident.
## Not run: pmvals <- seq(0.02,0.99,0.02) mask <- make.mask(traps(ovenCH[[1]]), nx = 32, buffer = 100) ## only g0 ~ h2, so reduce pmi from 5 to 4 outPL <- pmixProfileLL(ovenCH[[1]], model = list(g0~h2), mask = mask, pmvals, CL = TRUE, trace = FALSE, pmi = 4) plot(pmvals, outPL, xlim = c(0,1), xlab = 'Fixed pmix', ylab = 'Profile log-likelihood') ## End(Not run)
## Not run: pmvals <- seq(0.02,0.99,0.02) mask <- make.mask(traps(ovenCH[[1]]), nx = 32, buffer = 100) ## only g0 ~ h2, so reduce pmi from 5 to 4 outPL <- pmixProfileLL(ovenCH[[1]], model = list(g0~h2), mask = mask, pmvals, CL = TRUE, trace = FALSE, pmi = 4) plot(pmvals, outPL, xlim = c(0,1), xlab = 'Fixed pmix', ylab = 'Profile log-likelihood') ## End(Not run)
Determines which of a set of points lie inside a closed polygon or at least one of a set of polygons
pointsInPolygon(xy, poly, logical = TRUE)
pointsInPolygon(xy, poly, logical = TRUE)
xy |
2-column matrix or dataframe of x-y coordinates for points to assess |
poly |
2-column matrix or dataframe containing perimeter points of polygon, or a SpatialPolygonsDataFrame object from package sp, or a ‘mask’ object (see Warning) |
logical |
logical to control the output when ‘poly’ is a mask (see Details) |
If poly
is a SpatialPolygonsDataFrame object then the method
over
is used from sp. This allows multiple polygons and
polygons with holes.
If poly
is an secr ‘mask’ object then xy
is discretized
and matched to the cells in poly
. If logical = FALSE
then the returned value is a vector of integer indices to the row in
‘poly’ corresponding to each row of ‘xy’; otherwise the result is a
vector of logical values.
Otherwise, the algorithm is adapted from some code posted on the S-news list by Peter Perkins (23/7/1996). The polygon should be closed (last point same as first).
Vector of logical or integer values, one for each row in xy
If poly
is a mask object then its cells must be
aligned to the x- and y- axes
## 100 random points in unit square xy <- matrix(runif(200), ncol = 2) ## triangle centred on (0.5, 0.5) poly <- data.frame(x = c(0.2,0.5,0.8,0.2), y = c(0.2,0.8,0.2,0.2)) plot(xy, pch = 1 + pointsInPolygon(xy, poly)) lines(poly)
## 100 random points in unit square xy <- matrix(runif(200), ncol = 2) ## triangle centred on (0.5, 0.5) poly <- data.frame(x = c(0.2,0.5,0.8,0.2), y = c(0.2,0.8,0.2,0.2)) plot(xy, pch = 1 + pointsInPolygon(xy, poly)) lines(poly)
Area of a single closed polygon (simple x-y coordinate input) or of multiple polygons, possibly with holes.
polyarea(xy, ha = TRUE)
polyarea(xy, ha = TRUE)
xy |
dataframe or list with components ‘x’ and ‘y’, or a SpatialPolygons or SpatialPolygonsDataFrame object from package sp, or an sf object with polygon data |
ha |
logical if TRUE output is converted from square metres to hectares |
For sf, sfc, SpatialPolygons or SpatialPolygonsDataFrame objects, the package sf is used.
A scalar.
polyarea(make.grid(hollow = TRUE))
polyarea(make.grid(hollow = TRUE))
Encapsulate the locations of a set of individual animals.
An object of class popn
records the locations of a set of
individuals, together with ancillary data such as their sex. Often used
for a realisation of a spatial point process (e.g. homogeneous Poisson)
with known density (intensity). Locations are stored in a data frame
with columns ‘x’ and ‘y’.
A popn
object has attributes
covariates | data frame with numeric, factor or character variables to be used as individual covariates |
model2D | 2-D distribution ("poisson", "cluster", "IHP", "linear" etc.) |
Ndist | distribution of number of individuals ("poisson", "fixed") |
boundingbox | data frame of 4 rows, the vertices of the rectangular area |
The number of rows in covariates
must match the length of
x
and y
. See sim.popn
for more information
on Ndist
and model2D
.
The popn
class is used only occasionally: it is not central to spatially explicit capture recapture.
sim.popn
, plot.popn
, transformations
Data from a trapping study of brushtail possums at Waitarere, North Island, New Zealand.
possumCH possumarea possumremovalarea possummask possum.model.0 possum.model.Ds
possumCH possumarea possumremovalarea possummask possum.model.0 possum.model.Ds
Brushtail possums (Trichosurus vulpecula) are an unwanted invasive species in New Zealand. Although most abundant in forests, where they occasionally exceed densities of 15 / ha, possums live wherever there are palatable food plants and shelter.
Efford et al. (2005) reported a live-trapping study of possums in Pinus radiata plantation on coastal sand dunes. The 300-ha site at Waitarere in the North Island of New Zealand was a peninsula, bounded on one side by the sea and on two other sides by the Manawatu river. Cage traps were set in groups of 36 at 20-m spacing around the perimeter of five squares, each 180 m on a side. The squares (‘hollow grids’) were centred at random points within the 300-ha area. Animals were tagged and released daily for 5 days in April 2002. Subsequently, leg-hold trapping was conducted on a trapping web centred on each square (data not reported here), and strenuous efforts were made to remove all possums by cyanide poisoning and further leghold trapping across the entire area. This yielded a density estimate of 2.26 possums / ha.
Traps could catch at most one animal per day. The live-trapped animals comprised 46 adult females, 33 adult males, 10 immature females and 11 immature males; sex and/or age were not recorded for 4 individuals (M. Coleman unpubl. data). These counts do not sum to the number of capture histories - see Note. One female possum was twice captured at two sites on one day, having entered a second trap after being released; one record in each pair was selected arbitrarily and discarded.
The data are provided as a single-session capthist
object
‘possumCH’. ‘possummask’ is a matching mask object - see Examples.
Two fitted models are provided for illustration.
The dataframe possumarea
contains boundary coordinates of a
habitat polygon that is used to clip possummask
at the shore
(from secr 1.5). possumarea
comprises a single polygon
representing the extent of terrestrial vegetation to the west, north and
east, and an arbitrary straight southern boundary. The boundary is also
included as a shapefile and as a text file (‘possumarea.shp’ etc. and
‘possumarea.txt’ in the package ‘extdata’ folder). See Examples in
make.mask
.
The dataframe possumremovalarea
contains boundary coordinates of
another polygon, the nominal removal area of Efford et al. (2005 Fig. 1)
(from secr 2.3).
Object | Description |
possumCH | capthist object |
possummask | mask object |
possumarea | habitat perimeter |
possumremovalarea | nominal boundary of removal region |
possum.model.0 | fitted secr model -- null |
possum.model.Ds | fitted secr model -- distance to shore |
A significant problem with the data used by Efford et al. (2005) was noticed recently. Five capture histories in possumCH are for animals that had lost a previous tag. A further three histories may also have been animals that were tagged previously or mis-recorded. Analyses that treat each previously tagged animal as a new individual are in error (this includes the published analyses, the pre-fitted models described here, and those in the vignette secr-densitysurfaces.pdf). All eight questionable histories are now indicated in possumCH with the logical covariate ‘prev.tagged’.
Methods have not yet been developed to adjust for tag loss in SECR models.
Landcare Research, New Zealand.
Borchers, D.L. and Efford, M.G. (2008) Spatially explicit maximum likelihood methods for capture-recapture studies. Biometrics 64, 377–385.
Efford, M. G., Dawson, D. K. and Robbins C. S. (2004) DENSITY: software for analysing capture-recapture data from passive detector arrays. Animal Biodiversity and Conservation 27, 217–228.
Efford, M. G., Warburton, B., Coleman, M. C. and Barker, R. J. (2005) A field test of two methods for density estimation. Wildlife Society Bulletin 33, 731–738.
plot(possummask) plot(possumCH, tracks = TRUE, add = TRUE) plot(traps(possumCH), add = TRUE) lines(possumarea) summary(possumCH) ## compare & average pre-fitted models AIC(possum.model.0, possum.model.Ds) modelAverage(possum.model.0, possum.model.Ds) ## Not run: ## Roughly estimate tag-loss error by dropping dubious histories ## i.e. restrict to "not previously tagged" NPT <- !covariates(possumCH)$prev.tagged possum.model.0.NPT <- secr.fit(subset(possumCH,NPT), mask = possummask, trace = FALSE) predict(possum.model.0)[1,2]/ predict(possum.model.0.NPT)[1,2] ## ...about 9% ## End(Not run)
plot(possummask) plot(possumCH, tracks = TRUE, add = TRUE) plot(traps(possumCH), add = TRUE) lines(possumarea) summary(possumCH) ## compare & average pre-fitted models AIC(possum.model.0, possum.model.Ds) modelAverage(possum.model.0, possum.model.Ds) ## Not run: ## Roughly estimate tag-loss error by dropping dubious histories ## i.e. restrict to "not previously tagged" NPT <- !covariates(possumCH)$prev.tagged possum.model.0.NPT <- secr.fit(subset(possumCH,NPT), mask = possummask, trace = FALSE) predict(possum.model.0)[1,2]/ predict(possum.model.0.NPT)[1,2] ## ...about 9% ## End(Not run)
Evaluate a spatially explicit capture–recapture model. That is, compute the ‘real’ parameters corresponding to the ‘beta’ parameters of a fitted model for arbitrary levels of any variables in the linear predictor.
## S3 method for class 'secr' predict(object, newdata = NULL, realnames = NULL, type = c("response", "link"), se.fit = TRUE, alpha = 0.05, savenew = FALSE, ...) ## S3 method for class 'secrlist' predict(object, newdata = NULL, realnames = NULL, type = c("response", "link"), se.fit = TRUE, alpha = 0.05, savenew = FALSE, ...) ## S3 method for class 'secr' detectpar(object, ..., byclass = FALSE, bytrap = FALSE)
## S3 method for class 'secr' predict(object, newdata = NULL, realnames = NULL, type = c("response", "link"), se.fit = TRUE, alpha = 0.05, savenew = FALSE, ...) ## S3 method for class 'secrlist' predict(object, newdata = NULL, realnames = NULL, type = c("response", "link"), se.fit = TRUE, alpha = 0.05, savenew = FALSE, ...) ## S3 method for class 'secr' detectpar(object, ..., byclass = FALSE, bytrap = FALSE)
object |
|
newdata |
optional dataframe of values at which to evaluate model |
realnames |
character vector of real parameter names |
type |
character; type of prediction required. The default ("response") provides estimates of the ‘real’ parameters. |
se.fit |
logical for whether output should include SE and confidence intervals |
alpha |
alpha level for confidence intervals |
savenew |
logical for whether newdata should be saved |
... |
other arguments passed to |
byclass |
logical; if TRUE values are returned for each latent class in a mixture model, or class in a hybrid mixture (hcov) model |
bytrap |
logical; if TRUE values are returned for each detector |
The variables in the various linear predictors are described in
secr-models.pdf and listed for the particular model in the
vars
component of object
.
Optional newdata
should be a dataframe with a column for each of
the variables in the model (see ‘vars’ component of object
). If
newdata
is missing then a dataframe is constructed automatically.
Default newdata
are for a naive animal on the first occasion;
numeric covariates are set to zero and factor covariates to their base
(first) level. From secr 3.1.4 the argument ‘all.levels’ may be passed
to makeNewData
; if TRUE then the default newdata includes
all factor levels.
realnames
may be used to select a subset of parameters.
Standard errors for parameters on the response (real) scale are by the delta method (Lebreton et al. 1992), and confidence intervals are backtransformed from the link scale.
The value of newdata
is optionally saved as an attribute.
detectpar
is used to extract the detection parameter estimates
from a simple model to pass to functions such as
esaPlot
. detectpar
calls predict.secr
. Parameters
will be evaluated by default at base levels of the covariates, although
this may be overcome by passing a one-line newdata
to
predict
via the ... argument. Groups and mixtures are a
headache for detectpar
: it merely returns the estimated detection
parameters of the first group or mixture.
If the ‘a0’ parameterization has been used in secr.fit
(i.e.,
object$details$param == 3
) then detectpar
automatically
backtransforms (a0, sigma) to (g0, sigma) or (lambda0, sigma) depending
on the value of object$detectfn
.
When se.fit
= FALSE, a dataframe identical to newdata
except for the addition of one column for each ‘real’ parameter. Otherwise, a list with one component for each row in newdata
. Each component is a dataframe with one row for each ‘real’ parameter (density, g0, sigma, b) and columns as below
link | link function |
estimate | estimate of real parameter |
SE.estimate | standard error of the estimate |
lcl | lower 100(1--alpha)% confidence limit |
ucl | upper 100(1--alpha)% confidence limit |
When newdata
has only one row, the structure of the list is
‘dissolved’ and the return value is one data frame.
For detectpar
, a list with the estimated values of detection
parameters (e.g., g0 and sigma if detectfn = "halfnormal"). In the case
of multi-session data the result is a list of lists (one list per
session).
predictDsurface
should be used for predicting density at many
points from a model with spatial variation. This deals automatically
with scaling of x- and y-coordinates, and is much is faster than
predict.secr. The resulting Dsurface object has its own plot method.
The argument ‘scaled’ was removed from both predict methods in version 2.10 as the scaleg0 and scalesigma features had been superceded by other parameterisations.
Overdispersion results in confidence intervals that are too narrow. See adjustVarD
for a partial solution.
Lebreton, J.-D., Burnham, K. P., Clobert, J. and Anderson, D. R. (1992) Modeling survival and testing biological hypotheses using marked animals: a unified approach with case studies. Ecological Monographs 62, 67–118.
secr.fit
,
predictDsurface
,
adjustVarD
,
makeNewData
## load previously fitted secr model with trap response ## and extract estimates of `real' parameters for both ## naive (b = 0) and previously captured (b = 1) animals predict (secrdemo.b, newdata = data.frame(b = 0:1)) ## OR from secr 3.1.4 predict (secrdemo.b, all.levels = TRUE) temp <- predict (secrdemo.b, all.levels = TRUE, save = TRUE) attr(temp, "newdata") detectpar(secrdemo.0)
## load previously fitted secr model with trap response ## and extract estimates of `real' parameters for both ## naive (b = 0) and previously captured (b = 1) animals predict (secrdemo.b, newdata = data.frame(b = 0:1)) ## OR from secr 3.1.4 predict (secrdemo.b, all.levels = TRUE) temp <- predict (secrdemo.b, all.levels = TRUE, save = TRUE) attr(temp, "newdata") detectpar(secrdemo.0)
Predict density at each point on a raster mask from a fitted secr model.
predictDsurface(object, mask = NULL, se.D = FALSE, cl.D = FALSE, alpha = 0.05, parameter = c('D', 'noneuc'))
predictDsurface(object, mask = NULL, se.D = FALSE, cl.D = FALSE, alpha = 0.05, parameter = c('D', 'noneuc'))
object |
fitted secr object |
mask |
secr mask object |
se.D |
logical for whether to compute prediction SE |
cl.D |
logical for whether to compute confidence limits |
alpha |
alpha level for 100(1 – alpha)% confidence intervals |
parameter |
character for real parameter to predict |
Predictions use the linear model for density on the link scale in the
fitted secr model ‘object’, or the fitted user-defined function, if
that was specified in secr.fit
.
If ‘mask’ is NULL then predictions are for the mask component of ‘object’.
SE and confidence limits are computed only if specifically requested. They are not available for user-defined density functions.
Density is adjusted automatically for the number of clusters in
‘mashed’ models (see mash
).
Object of class ‘Dsurface’ inheriting from ‘mask’. Predicted densities are added to the covariate dataframe (attribute ‘covariates’) as column(s) with prefix ‘D.’ If the model uses multiple groups, multiple columns will be distinguished by the group name (e.g., "D.F" and "D.M"). If groups are not defined the column is named "D.0".
For multi-session models the value is a multi-session mask.
The pointwise prediction SE is saved as a covariate column prefixed ‘SE.’ (or multiple columns if multiple groups). Confidence limits are likewise saved with prefixes ‘lcl.’ and ‘ucl.’.
plot.Dsurface
, secr.fit
, predict.secr
## use canned possum model shorePossums <- predictDsurface(possum.model.Ds) par(mar = c(1,1,1,6)) plot(shorePossums, plottype = "shaded", polycol = "blue", border = 100) plot(traps(possumCH), detpar = list(col = "black"), add = TRUE) par(mar = c(5,4,4,2) + 0.1) ## reset to default ## extract and summarise summary(covariates(shorePossums)) ## Not run: ## extrapolate to a new mask; add covariate needed by model; plot regionmask <- make.mask(traps(possumCH), buffer = 1000, spacing = 10, poly = possumremovalarea) dts <- distancetotrap(regionmask, possumarea) covariates(regionmask) <- data.frame(d.to.shore = dts) regionPossums <- predictDsurface(possum.model.Ds, regionmask, se.D = TRUE, cl.D = TRUE) par(mfrow = c(1,2), mar = c(1,1,1,6)) plot(regionPossums, plottype = "shaded", mesh = NA, breaks = 20) plot(regionPossums, plottype = "contour", add = TRUE) plot(regionPossums, covariate = "SE", plottype = "shaded", mesh = NA, breaks = 20) plot(regionPossums, covariate = "SE", plottype = "contour", add = TRUE) ## confidence surfaces plot(regionPossums, covariate = "lcl", breaks = seq(0,3,0.2), plottype = "shaded") plot(regionPossums, covariate = "lcl", plottype = "contour", add = TRUE, levels = seq(0,2.7,0.2)) title("lower 95% surface") plot(regionPossums, covariate = "ucl", breaks=seq(0,3,0.2), plottype = "shaded") plot(regionPossums, covariate = "ucl", plottype = "contour", add = TRUE, levels = seq(0,2.7,0.2)) title("upper 95% surface") ## annotate with CI par(mfrow = c(1,1)) plot(regionPossums, plottype = "shaded", mesh = NA, breaks = 20) plot(traps(possumCH), add = TRUE, detpar = list(col = "black")) if (interactive()) { spotHeight(regionPossums, dec = 1, pre = c("lcl","ucl"), cex = 0.8) } ## perspective plot pm <- plot(regionPossums, plottype = "persp", box = FALSE, zlim = c(0,3), phi=30, d = 5, col = "green", shade = 0.75, border = NA) lines(trans3d (possumremovalarea$x, possumremovalarea$y, rep(1,nrow(possumremovalarea)), pmat = pm)) par(mfrow = c(1,1), mar = c(5, 4, 4, 2) + 0.1) ## reset to default ## compare estimates of region N ## grid cell area is 0.01 ha sum(covariates(regionPossums)[,"D.0"]) * 0.01 region.N(possum.model.Ds, regionmask) ## End(Not run)
## use canned possum model shorePossums <- predictDsurface(possum.model.Ds) par(mar = c(1,1,1,6)) plot(shorePossums, plottype = "shaded", polycol = "blue", border = 100) plot(traps(possumCH), detpar = list(col = "black"), add = TRUE) par(mar = c(5,4,4,2) + 0.1) ## reset to default ## extract and summarise summary(covariates(shorePossums)) ## Not run: ## extrapolate to a new mask; add covariate needed by model; plot regionmask <- make.mask(traps(possumCH), buffer = 1000, spacing = 10, poly = possumremovalarea) dts <- distancetotrap(regionmask, possumarea) covariates(regionmask) <- data.frame(d.to.shore = dts) regionPossums <- predictDsurface(possum.model.Ds, regionmask, se.D = TRUE, cl.D = TRUE) par(mfrow = c(1,2), mar = c(1,1,1,6)) plot(regionPossums, plottype = "shaded", mesh = NA, breaks = 20) plot(regionPossums, plottype = "contour", add = TRUE) plot(regionPossums, covariate = "SE", plottype = "shaded", mesh = NA, breaks = 20) plot(regionPossums, covariate = "SE", plottype = "contour", add = TRUE) ## confidence surfaces plot(regionPossums, covariate = "lcl", breaks = seq(0,3,0.2), plottype = "shaded") plot(regionPossums, covariate = "lcl", plottype = "contour", add = TRUE, levels = seq(0,2.7,0.2)) title("lower 95% surface") plot(regionPossums, covariate = "ucl", breaks=seq(0,3,0.2), plottype = "shaded") plot(regionPossums, covariate = "ucl", plottype = "contour", add = TRUE, levels = seq(0,2.7,0.2)) title("upper 95% surface") ## annotate with CI par(mfrow = c(1,1)) plot(regionPossums, plottype = "shaded", mesh = NA, breaks = 20) plot(traps(possumCH), add = TRUE, detpar = list(col = "black")) if (interactive()) { spotHeight(regionPossums, dec = 1, pre = c("lcl","ucl"), cex = 0.8) } ## perspective plot pm <- plot(regionPossums, plottype = "persp", box = FALSE, zlim = c(0,3), phi=30, d = 5, col = "green", shade = 0.75, border = NA) lines(trans3d (possumremovalarea$x, possumremovalarea$y, rep(1,nrow(possumremovalarea)), pmat = pm)) par(mfrow = c(1,1), mar = c(5, 4, 4, 2) + 0.1) ## reset to default ## compare estimates of region N ## grid cell area is 0.01 ha sum(covariates(regionPossums)[,"D.0"]) * 0.01 region.N(possum.model.Ds, regionmask) ## End(Not run)
Print method for capthist
objects.
## S3 method for class 'capthist' print(x, ..., condense = FALSE, sortrows = FALSE)
## S3 method for class 'capthist' print(x, ..., condense = FALSE, sortrows = FALSE)
x |
|
... |
arguments to pass to |
condense |
logical, if true then use condensed format for 3-D data |
sortrows |
logical, if true then sort output by animal |
The condense
option may be used to format data from proximity
detectors in a slightly more readable form. Each row then presents the
detections of an individual in a particular trap, dropping rows (traps)
at which the particular animal was not detected.
Invisibly returns a dataframe (condense = TRUE) or array in the format printed.
## simulated detections of simulated default population of 5/ha print(sim.capthist(make.grid(nx=5,ny=3)))
## simulated detections of simulated default population of 5/ha print(sim.capthist(make.grid(nx=5,ny=3)))
Print results from fitting a spatially explicit capture–recapture model or generate a list of summary values.
## S3 method for class 'secr' print(x, newdata = NULL, alpha = 0.05, deriv = FALSE, call = TRUE, ...) ## S3 method for class 'secr' summary(object, newdata = NULL, alpha = 0.05, deriv = FALSE, ...)
## S3 method for class 'secr' print(x, newdata = NULL, alpha = 0.05, deriv = FALSE, call = TRUE, ...) ## S3 method for class 'secr' summary(object, newdata = NULL, alpha = 0.05, deriv = FALSE, ...)
x |
|
object |
|
newdata |
optional dataframe of values at which to evaluate model |
alpha |
alpha level |
deriv |
logical for calculation of derived D and esa |
call |
logical; if TRUE the call is printed |
... |
other arguments optionally passed to derived.secr |
Results from print.secr
are potentially complex and depend upon the analysis (see
below). Optional newdata
should be a dataframe with a column for
each of the variables in the model. If newdata
is missing then a
dataframe is constructed automatically. Default newdata
are for
a naive animal on the first occasion; numeric covariates are set to zero
and factor covariates to their base (first) level. Confidence intervals
are 100 (1 – alpha) % intervals.
call | the function call (optional) |
version,time | secr version, date and time fitting started, and elapsed time |
Detector type | `single', `multi', `proximity' etc. |
Detector number | number of detectors |
Average spacing | |
x-range | |
y-range | |
New detector type | as fitted when details$newdetector specified |
N animals | number of distinct animals detected |
N detections | number of detections |
N occasions | number of sampling occasions |
Mask area | |
Model | model formula for each `real' parameter |
Fixed (real) | fixed real parameters |
Detection fn | detection function type (halfnormal or hazard-rate) |
N parameters | number of parameters estimated |
Log likelihood | log likelihood |
AIC | Akaike's information criterion |
AICc | AIC with small sample adjustment (Burnham and Anderson 2002) |
Beta parameters | coef of the fitted model, SE and confidence intervals |
vcov | variance-covariance matrix of beta parameters |
Real parameters | fitted (real) parameters evaluated at base levels of covariates |
Derived parameters | derived estimates of density and mean effective sampling area (optional) |
Derived parameters (see derived
) are computed only if
deriv = TRUE
.
The summary
method constructs a list of outputs similar to those printed by the print
method, but somewhat more concise and re-usable:
versiontime | secr version, and date and time fitting started |
traps | detector summary |
capthist | capthist summary |
mask | mask summary |
modeldetails | miscellaneous model characteristics (CL etc.) |
AICtable | single-line output of AIC.secr |
coef | table of fitted coefficients with CI |
predicted | predicted values (`real' parameter estimates) |
derived | output of derived.secr (optional) |
Burnham, K. P. and Anderson, D. R. (2002) Model selection and multimodel inference: a practical information-theoretic approach. Second edition. New York: Springer-Verlag.
## load & print previously fitted null (constant parameter) model print(secrdemo.0) summary(secrdemo.0) ## combine AIC tables from list of summaries do.call(AIC, lapply(list(secrdemo.b, secrdemo.0), summary)) ## Not run: print(secrdemo.CL, deriv = TRUE) ## End(Not run)
## load & print previously fitted null (constant parameter) model print(secrdemo.0) summary(secrdemo.0) ## combine AIC tables from list of summaries do.call(AIC, lapply(list(secrdemo.b, secrdemo.0), summary)) ## Not run: print(secrdemo.CL, deriv = TRUE) ## End(Not run)
Print method for traps
objects.
## S3 method for class 'traps' print(x, ...)
## S3 method for class 'traps' print(x, ...)
x |
|
... |
arguments to pass to |
print(make.grid(nx = 5, ny = 3))
print(make.grid(nx = 5, ny = 3))
The Modified Random Cluster algorithm of Saura and Martinez-Millan (2000) is used to generate a mask object representing patches of contiguous ‘habitat’ cells (pixels) within a ‘non-habitat’ matrix (‘non-habitat’ cells are optionally dropped). Spatial autocorrelation (fragmentation) of habitat patches is controlled via the parameter ‘p’. ‘A’ is the expected proportion of ‘habitat’ cells.
randomDensity
is a wrapper for randomHabitat
that may be used as input to sim.popn
.
randomHabitat(mask, p = 0.5, A = 0.5, directions = 4, minpatch = 1, drop = TRUE, covname = "habitat", plt = FALSE, seed = NULL) randomDensity(mask, parm)
randomHabitat(mask, p = 0.5, A = 0.5, directions = 4, minpatch = 1, drop = TRUE, covname = "habitat", plt = FALSE, seed = NULL) randomDensity(mask, parm)
mask |
secr mask object to use as template |
p |
parameter to control fragmentation |
A |
parameter for expected proportion of habitat |
directions |
integer code for adjacency (rook's move 4 or queen's move 8) |
minpatch |
integer minimum size of patch |
drop |
logical for whether to drop non-habitat cells |
covname |
character name of covariate when |
plt |
logical for whether intermediate stages should be plotted |
seed |
either NULL or an integer that will be used in a call to |
parm |
list of arguments for |
Habitat is simulated within the region defined by the cells of mask. The region may be non-rectangular.
The algorithm comprises stages A-D:
A. Randomly select proportion p
of cells from the input mask
B. Cluster selected cells with any immediate neighbours as defined by
directions
C. Assign clusters to ‘non-habitat’ (probability 1–A) and ‘habitat’ (probability A)
D. Cells not in any cluster from (B) receive the habitat class of the majority of the <=8 adjacent cells assigned in (C), if there are any; otherwise they are assigned at random (with probabilities 1–A, A).
Fragmentation declines, and cluster size increases, as p increases up to the ‘percolation threshold’ which is about 0.59 in the default case (Saura and Martinez-Millan 2000 p.664).
If minpatch > 1
then habitat patches of less than minpatch
cells are converted to non-habitat, and vice versa. This is likely to
cause the proportion of habitat to deviate from A
.
If drop = FALSE
a binary-valued (0/1) covariate with the
requested name is included in the output mask, which has the same extent
as the input. Otherwise, non-habitat cells are dropped and no covariate
is added.
The argument ‘parm’ for randomDensity
is a list with average density D and an optional subset of named values to override the defaults (p = 0.5, A = 0.5, directions = 4, minpatch = 1, plt = FALSE, seed = NULL). ‘rescale’ is a further optional component of ‘parm’; if ‘rescale = TRUE’ then the pixel-specific densities are adjusted upwards by the factor 1/A to maintain the same expected number of activity centres as if the nominal density applied throughout. Arguments ‘mask’ and ‘drop’ of randomHabitat
are substituted automatically.
For randomHabitat –
An object of class ‘mask’. By default (drop = TRUE
) this
has fewer rows (points) than the input mask.
The attribute “type” is a character string formed from paste('MRC p=',p, ' A=',A, sep='')
.
The RNG seed is stored as attribute ‘seed’ (see secrRNG).
For randomDensity –
A vector of cell-specific densities.
Single-linkage clustering and adjacency operations use functions
‘clump’ and ‘adjacency’ of the package raster; ‘clump’ also
requires package igraph0 (raster still uses this
deprecated version). Optional plotting
of intermediate stages (plt = TRUE
) uses the plot method for
rasterLayers in raster.
A non-rectangular input mask is padded out to a rectangular rasterLayer for operations in raster; cells added as padding are ultimately dropped.
The procedure of Saura and Martinez-Millan (2000) has been followed as far as possible, but this implementation may not match theirs in every detail.
This implementation allows only two habitat classes. The parameter A is the expected value of the habitat proportion; the realised habitat proportion may differ quite strongly from A, especially for large p (e.g., p > 0.5).
Anisotropy is not implemented; it would require skewed adjacency filters (i.e. other than rook- or queen-move filters) that are not available in raster.
Gaussian random fields provide an alternative method for simulating
random habitats (e.g., rLGCP option in sim.popn
).
Hijmans, R. J. and van Etten, J. (2011) raster: Geographic analysis and modeling with raster data. R package version 1.9-33. https://CRAN.R-project.org/package=raster.
Saura, S. and Martinez-Millan, J. (2000) Landscape patterns simulation with a modified random clusters method. Landscape Ecology, 15, 661–678.
## Not run: tempmask <- make.mask(nx = 100, ny = 100, spacing = 20) mrcmask <- randomHabitat(tempmask, p = 0.4, A = 0.4) plot(mrcmask, dots = FALSE, col = "green") pop <- sim.popn(10, mrcmask, model2D = "IHP") plot(pop, add = TRUE) # OR plot(sim.popn(D = randomDensity, core = tempmask, model2D = "IHP", details = list(D = 10, p = 0.4, A = 0.4, plt = TRUE)), add = TRUE, frame = FALSE) ## plot intermediate steps A, C, D opar <- par(mfrow = c(1,3)) mrcmask <- randomHabitat(tempmask, p = 0.4, A = 0.4, plt = TRUE) par(opar) ## keep non-habitat cells mrcmask <- randomHabitat(tempmask, p = 0.4, A = 0.4, drop = FALSE) plot(mrcmask, covariate = "habitat", dots = FALSE, col = c("grey","green"), breaks = 2) ## effect of purging small patches opar <- par(mfrow=c(1,2)) mrcmask <- randomHabitat(tempmask, p = 0.4, A = 0.4, minpatch = 1) plot(mrcmask, dots = FALSE, col ="green") mrcmask <- randomHabitat(tempmask, p = 0.4, A = 0.4, minpatch = 5) plot(mrcmask, dots = FALSE, col ="green") par(opar) ## End(Not run)
## Not run: tempmask <- make.mask(nx = 100, ny = 100, spacing = 20) mrcmask <- randomHabitat(tempmask, p = 0.4, A = 0.4) plot(mrcmask, dots = FALSE, col = "green") pop <- sim.popn(10, mrcmask, model2D = "IHP") plot(pop, add = TRUE) # OR plot(sim.popn(D = randomDensity, core = tempmask, model2D = "IHP", details = list(D = 10, p = 0.4, A = 0.4, plt = TRUE)), add = TRUE, frame = FALSE) ## plot intermediate steps A, C, D opar <- par(mfrow = c(1,3)) mrcmask <- randomHabitat(tempmask, p = 0.4, A = 0.4, plt = TRUE) par(opar) ## keep non-habitat cells mrcmask <- randomHabitat(tempmask, p = 0.4, A = 0.4, drop = FALSE) plot(mrcmask, covariate = "habitat", dots = FALSE, col = c("grey","green"), breaks = 2) ## effect of purging small patches opar <- par(mfrow=c(1,2)) mrcmask <- randomHabitat(tempmask, p = 0.4, A = 0.4, minpatch = 1) plot(mrcmask, dots = FALSE, col ="green") mrcmask <- randomHabitat(tempmask, p = 0.4, A = 0.4, minpatch = 5) plot(mrcmask, dots = FALSE, col ="green") par(opar) ## End(Not run)
Methods to convert secr object to a RasterLayer object.
## S4 method for signature 'mask' raster(x, covariate, values = 1, crs = NA) ## S4 method for signature 'Dsurface' raster(x, covariate, values = 1, crs = NA) ## S4 method for signature 'mask' rast(x, covariate, values = 1, crs = "") ## S4 method for signature 'Dsurface' rast(x, covariate, values = 1, crs = "")
## S4 method for signature 'mask' raster(x, covariate, values = 1, crs = NA) ## S4 method for signature 'Dsurface' raster(x, covariate, values = 1, crs = NA) ## S4 method for signature 'mask' rast(x, covariate, values = 1, crs = "") ## S4 method for signature 'Dsurface' rast(x, covariate, values = 1, crs = "")
x |
mask or Dsurface object |
covariate |
character name of covariate to provide values for RasterLayer |
values |
numeric values for RasterLayer |
crs |
character or object of class CRS. Optional PROJ.4 type description of a Coordinate Reference System (map projection). |
There are two ways to specify the values to be used. If covariate
is provided then
the values of the corresponding covariate of the mask or Dsurface are used. Otherwise,
values
is duplicated to the required number of rows.
The resulting RasterLayer may optionally include a PROJ.4 map projection
defined via crs
. The specification may be very simple (as in the
example below) or complex, including an explicit datum and other
arguments. Projections are used by sf, terra, raster,
sp and other packages. See raster
for
further explanation and links.
The S3 classes ‘mask’ and ‘Dsurface’ are defined in secr as virtual S4 classes. This enables these extensions to the list of S4 methods defined in raster and terra.
Although these methods work ‘standalone’, it is currently necessary to load the terra or raster package to do much with the result (e.g., plot it).
RasterLayer (raster)
SpatRaster (rast)
Prior to secr 2.9.5 these methods could fail unpredictably because an intermediate array was badly dimensioned due to truncation of a floating point value.
## Not run: shorePossums <- predictDsurface(possum.model.Ds) tmp <- raster(shorePossums, covariate = "D.0") library(raster) plot(tmp, useRaster = FALSE) ## alternative with same result tmp <- raster(shorePossums, values = covariates(shorePossums)$D.0) ## set the projection ## here the crs PROJ.4 spec refers simply to the old NZ metric grid tmp <- raster(shorePossums, "D.0", crs = "+proj=nzmg") ## check the projection proj4string(tmp) # mask to rast dtsrast <- rast(possummask, covariate = 'd.to.shore') plot(dtsrast) # Dsurface to rast class(shorePossums) <- c('mask', 'data.frame') # or change "mask" to "Dsurface" above Drast <- rast(shorePossums, covariate = 'D.0') plot(Drast) ## End(Not run)
## Not run: shorePossums <- predictDsurface(possum.model.Ds) tmp <- raster(shorePossums, covariate = "D.0") library(raster) plot(tmp, useRaster = FALSE) ## alternative with same result tmp <- raster(shorePossums, values = covariates(shorePossums)$D.0) ## set the projection ## here the crs PROJ.4 spec refers simply to the old NZ metric grid tmp <- raster(shorePossums, "D.0", crs = "+proj=nzmg") ## check the projection proj4string(tmp) # mask to rast dtsrast <- rast(possummask, covariate = 'd.to.shore') plot(dtsrast) # Dsurface to rast class(shorePossums) <- c('mask', 'data.frame') # or change "mask" to "Dsurface" above Drast <- rast(shorePossums, covariate = 'D.0') plot(Drast) ## End(Not run)
Form a single capthist
object from two or more compatible capthist
objects.
MS.capthist(...) append.capthist(..., synchronous = TRUE) ## S3 method for class 'capthist' rbind(..., renumber = TRUE, pool = NULL, verify = TRUE)
MS.capthist(...) append.capthist(..., synchronous = TRUE) ## S3 method for class 'capthist' rbind(..., renumber = TRUE, pool = NULL, verify = TRUE)
... |
one or more |
synchronous |
logical; if TRUE occasions are assumed to coincide |
renumber |
logical, if TRUE assigns new composite individual ID |
pool |
list of vectors of session indices or names |
verify |
logical, if TRUE the output is checked with |
MS.capthist
concatenates the sessions in the input objects as
one multi-session capthist object. Each session may use a different
detector array (traps) and a different number of sampling
occasions. Session names are derived implicitly from the inputs, or
may be given explicitly (see Examples); if any name is duplicated, all
will be replaced with sequential integers. The ... argument may include
lists of single-session capthist objects.
append.capthist
constructs a single-session capthist assuming the inputs
are independent capthist objects. Individuals and detectors are renumbered. Occasions are assumed to be synchronous by default. This is an alternative to rbind when the detector array varies.
The rbind
method for capthist objects is used to pool capture data
from more than one
session into a single session. The number of rows in the output
session is the sum of the number of rows in the input sessions
(i.e. each animal appears in only one session). Sessions to be pooled with
rbind
must have the same number of capture occasions and use the
same detectors (traps). At present there is no function to pool capthist
data from different detector arrays. For this it is recommended that you
merge the input files and rebuild the capthist object from scratch.
For rbind.capthist
, the ... argument may be
A series of single-session capthist objects, which are pooled to form one new single-session object, or
One multi-session capthist object, when the components of
‘pool’ are used to define combinations of old sessions; e.g. pool =
list(A=1:3, B=4:5)
produces an object with two sessions (named ‘A’
and ‘B’) from 5 old ones. If pool = NULL
(the default) then all the
sessions are pooled to form one single-session capthist object.
The names of arguments other than ... should be given in full. If
renumber = TRUE
(the default), the session name will be prepended
to the animal ID before pooling: animals 1, 2 and 3 in Session A will
become A.1, A.2 and A.3, while those in Session B become B.1, B.2 and
B.3. This ensures that each animal has a unique ID. If renumber =
FALSE
, the animal IDs will not change.
Other attributes (xy, signal) are handled appropriately. If the signal threshold (attribute ‘cutval’) differs among sessions, the maximum is used and detections of lower signal strength are discarded.
The use of rbind.capthist
to concatenate sessions is now deprecated:
use MS.capthist
.
Although MS.capthist
looks like an S3 method, it isn't. The full function name must be used. rbind.capthist
became an S3 method in secr 3.1, so it is called as rbind
alone.
For MS.capthist
, a multi-session object of class ‘capthist’ with
number of sessions equal to the number of sessions in the objects in
....
For append.capthist
, a single-session of class ‘capthist’. When ‘synchronous’ the dimensions are () where (
) are the numbers of individuals, occasions and detectors in the i-th input. Otherwise the dimensions are (
). An index to the original component is saved in the traps covariate ‘sub’.
For rbind.capthist
, either an object of class ‘capthist’ with one
session formed by pooling the sessions in the input objects, or a
capthist object with more than one session, each formed by pooling
groups of sessions defined by the ‘pool’ argument. Covariate columns that appear in
all input sessions are retained in the output.
## extend a multi-session object ## we fake the 2010 data by copying from 2005 ## note how we name the appended session fakeCH <- ovenCH[["2005"]] MS.capthist(ovenCH, "2010" = fakeCH) ## simulate sessions for 2-part mixture temptrap <- make.grid(nx = 8, ny = 8) temp1 <- sim.capthist(temptrap, detectpar = list(g0 = 0.1, sigma = 40)) temp2 <- sim.capthist(temptrap, detectpar = list(g0 = 0.2, sigma = 20)) ## concatenate sessions temp3 <- MS.capthist(large.range = temp1, small.range = temp2) summary(temp3) ## session-specific movement statistic RPSV(temp3) ## pool sessions temp4 <- rbind(temp1, temp2) summary(temp4) RPSV(temp4) ## compare mixture to sum of components ## note `detectors visited' is not additive for 'multi' detector ## nor is `detectors used' (summary(temp1)$counts + summary(temp2)$counts) - summary(temp4)$counts ## Not run: ## compare two different model fits tempfit3 <- secr.fit(temp3, CL = TRUE, buffer = 150, model = list (g0 ~ session, sigma ~ session), trace = FALSE) predict(tempfit3) ## if we can tell which animals had large ranges... covariates(temp4) <- data.frame(range.size = rep(c("large", "small"), c(nrow(temp1), nrow(temp2)))) tempfit4 <- secr.fit(temp4, CL = TRUE, buffer = 150, model = list (g0 ~ range.size, sigma ~ range.size), trace = FALSE) predict(tempfit4, newdata = data.frame(range.size = c("large", "small"))) ## polygon data pol1 <- make.poly() pol2 <- make.poly(x = c(50,50,150,150)) ch1 <- sim.capthist(pol1, popn = list(D = 30), detectfn = 'HHN', detectpar = list(lambda0 = 0.3)) ch2 <- sim.capthist(pol2, popn = list(D = 30), detectfn = 'HHN', detectpar = list(lambda0 = 0.3)) plot(ch1); plot(pol2, add = TRUE); plot(ch2, add = TRUE) ## End(Not run)
## extend a multi-session object ## we fake the 2010 data by copying from 2005 ## note how we name the appended session fakeCH <- ovenCH[["2005"]] MS.capthist(ovenCH, "2010" = fakeCH) ## simulate sessions for 2-part mixture temptrap <- make.grid(nx = 8, ny = 8) temp1 <- sim.capthist(temptrap, detectpar = list(g0 = 0.1, sigma = 40)) temp2 <- sim.capthist(temptrap, detectpar = list(g0 = 0.2, sigma = 20)) ## concatenate sessions temp3 <- MS.capthist(large.range = temp1, small.range = temp2) summary(temp3) ## session-specific movement statistic RPSV(temp3) ## pool sessions temp4 <- rbind(temp1, temp2) summary(temp4) RPSV(temp4) ## compare mixture to sum of components ## note `detectors visited' is not additive for 'multi' detector ## nor is `detectors used' (summary(temp1)$counts + summary(temp2)$counts) - summary(temp4)$counts ## Not run: ## compare two different model fits tempfit3 <- secr.fit(temp3, CL = TRUE, buffer = 150, model = list (g0 ~ session, sigma ~ session), trace = FALSE) predict(tempfit3) ## if we can tell which animals had large ranges... covariates(temp4) <- data.frame(range.size = rep(c("large", "small"), c(nrow(temp1), nrow(temp2)))) tempfit4 <- secr.fit(temp4, CL = TRUE, buffer = 150, model = list (g0 ~ range.size, sigma ~ range.size), trace = FALSE) predict(tempfit4, newdata = data.frame(range.size = c("large", "small"))) ## polygon data pol1 <- make.poly() pol2 <- make.poly(x = c(50,50,150,150)) ch1 <- sim.capthist(pol1, popn = list(D = 30), detectfn = 'HHN', detectpar = list(lambda0 = 0.3)) ch2 <- sim.capthist(pol2, popn = list(D = 30), detectfn = 'HHN', detectpar = list(lambda0 = 0.3)) plot(ch1); plot(pol2, add = TRUE); plot(ch2, add = TRUE) ## End(Not run)
Form a single popn
object from two or more existing popn
objects, or a list.
## S3 method for class 'popn' rbind(..., renumber = TRUE)
## S3 method for class 'popn' rbind(..., renumber = TRUE)
... |
one or more |
renumber |
logical for whether row names in the new object should be set to the row indices |
An attempt to combine objects will fail if they conflict in their covariates
attributes.
From secr 3.1 this is an S3 method and list input is not allowed.
An object of class popn
with number of rows equal to the sum of the rows in the input objects.
## generate and combine two subpopulations trapobj <- make.grid() p1 <- sim.popn(D = 3, core = trapobj) p2 <- sim.popn(D = 2, core = trapobj) covariates(p1) <- data.frame(size = rep("small", nrow(p1))) covariates(p2) <- data.frame(size = rep("large", nrow(p2))) pop <- rbind(p1,p2) ## or pop <- do.call(rbind, list(p1,p2))
## generate and combine two subpopulations trapobj <- make.grid() p1 <- sim.popn(D = 3, core = trapobj) p2 <- sim.popn(D = 2, core = trapobj) covariates(p1) <- data.frame(size = rep("small", nrow(p1))) covariates(p2) <- data.frame(size = rep("large", nrow(p2))) pop <- rbind(p1,p2) ## or pop <- do.call(rbind, list(p1,p2))
Form a single traps
object from two or more existing traps
objects.
## S3 method for class 'traps' rbind(..., renumber = TRUE, addusage, checkdetector = TRUE, suffix = TRUE)
## S3 method for class 'traps' rbind(..., renumber = TRUE, addusage, checkdetector = TRUE, suffix = TRUE)
... |
one or more |
renumber |
logical for whether row names in the new object should be set to the row indices |
addusage |
integer vector; if specified and the inputs lack usage attributes then a binary usage attribute will be generated with the given number of occasions for each input |
checkdetector |
logical; if TRUE then variation in the detector attribute triggers a warning |
suffix |
logical; if TRUE then suffix to the row names indicates source |
An attempt to combine objects will fail if they conflict in their
covariates
attributes. Differences in the usage
attribute are handled as follows. If usage
is missing for all
inputs and addusage = TRUE
is specified then usage codes are
generated automatically (positive for the specified number of
occasions). If usage
is specified for one input but not
other(s), the missing values are constructed assuming all detectors
were operated for the maximum number of occasions in any input. If
inputs differ in the number of ‘usage’ columns (occasions), the
smaller matrices are padded with ‘zero’ columns to the maximum number
of columns in any input.
... may be a single multi-session traps object (from 2.10.0).
By default (and always prior to 3.1.1) row names include a suffix (e.g., ".1", or ".2") to indicate the original object (first, second etc.). A suffix is added automatically to all names if any name is duplicated, and a warning is generated.
An object of class traps
with number of rows equal to the sum of the rows in the input objects.
## nested hollow grids hollow1 <- make.grid(nx = 8, ny = 8, hollow = TRUE) hollow2 <- shift(make.grid(nx = 6, ny = 6, hollow = TRUE), c(20, 20)) nested <- rbind (hollow1, hollow2) plot(nested, gridlines = FALSE, label = TRUE)
## nested hollow grids hollow1 <- make.grid(nx = 8, ny = 8, hollow = TRUE) hollow2 <- shift(make.grid(nx = 6, ny = 6, hollow = TRUE), c(20, 20)) nested <- rbind (hollow1, hollow2) plot(nested, gridlines = FALSE, label = TRUE)
Data in the DENSITY formats for capture data and trap layouts may be
imported as a capthist
object for analysis in secr. Data in
a capthist
object may also be exported in these formats for use
in DENSITY (Efford 2012). read.capthist
inputs data from text
files and constructs a capthist
object in one step using the
functions read.traps
and make.capthist
. Data may also be
read from Excel spreadsheets if the package readxl is installed (see
secr-datainput.pdf).
read.capthist(captfile, trapfile, detector = "multi", fmt = c("trapID","XY"), noccasions = NULL, covnames = NULL, trapcovnames = NULL, cutval = NULL, verify = TRUE, noncapt = "NONE", tol = 0.01, snapXY = FALSE, markocc = NULL, ...) write.capthist(object, filestem = deparse(substitute(object)), sess = "1", ndec = 2, covariates = FALSE, tonumeric = TRUE, ...)
read.capthist(captfile, trapfile, detector = "multi", fmt = c("trapID","XY"), noccasions = NULL, covnames = NULL, trapcovnames = NULL, cutval = NULL, verify = TRUE, noncapt = "NONE", tol = 0.01, snapXY = FALSE, markocc = NULL, ...) write.capthist(object, filestem = deparse(substitute(object)), sess = "1", ndec = 2, covariates = FALSE, tonumeric = TRUE, ...)
captfile |
name of capture data file |
trapfile |
name of trap layout file or (for a multi-session captfile) a vector of file names, one for each session |
detector |
character value for detector type (‘single’, ‘multi’, ‘proximity’, etc.) |
fmt |
character value for capture format (‘trapID’ or ‘XY’) |
noccasions |
number of occasions on which detectors were operated |
covnames |
character vector of names for individual covariate fields in ‘captfile’ |
trapcovnames |
character vector of names for detector covariate fields in ‘trapfile’ |
cutval |
numeric, threshold of signal strength for ‘signal’ detector type |
verify |
logical if TRUE then the resulting capthist object is
checked with |
noncapt |
character value; animal ID used for ‘no captures’ |
tol |
numeric, snap tolerance in metres |
snapXY |
logical; if TRUE then fmt = 'XY' uses nearest trap within tol |
markocc |
integer vector distinguishing marking occasions (1) from sighting occasions (0) |
... |
other arguments passed to |
object |
|
filestem |
character value used to form names of output files |
sess |
character session identifier |
ndec |
number of digits after decimal point for x,y coordinates |
covariates |
logical or a character vector of covariates to export |
tonumeric |
logical for whether factor and character covariates should be converted to numeric values on output |
read.capthist
captfile
should record one detection on each line. A detection
comprises a session identifier, animal identifier, occasion number (1,
2,...,S where S is the number of occasions), and a
detector identifier (fmt = "trapID"
) or X- and Y-coordinates
(fmt = "XY"
). Each line of trapfile
has a detector
identifier and its X- and Y-coordinates. In either file type the
identifiers (labels) may be numeric or alphanumeric values. Values
should be separated by blanks or tabs unless (i) the file name ends in
‘.csv’ or (ii) sep = ","
is passed in ..., in which case commas
are assumed. Blank lines and any text after ‘#’ are ignored. For further
details see secr-datainput.pdf,
make.capthist
and ‘Data formats’ in the help for DENSITY.
The noccasions
argument is needed only if there were no
detections on the final occasion; it may be a positive integer (constant
across all sessions) or a vector of positive integers, one for each
session. covnames
is needed only when captfile
includes
individual covariates. Likewise for trapcovnames
and
detector covariates. Values of noccasions
and covnames
are passed directly to make.capthist
, and trapcovnames
is
passed to read.traps
.
A session identifier is required even for single-session
capture data. In the case of data from multiple sessions,
trapfile
may be a vector of file names, one for each session.
Additional data may be coded as for DENSITY. Specifically,
captfile
may include extra columns of individual covariates, and
trapfile
may code varying usage of each detector over occasions
and detector covariates.
markocc
is needed only if sightings of unmarked animals are potentially
recorded on some occasions. If the data span multiple sessions with differing
combinations of marking and sighting occasions then markocc
may be a
list with one vector per session.
The function read.telemetry
is a simplified version of
read.capthist
for telemetry data.
write.capthist
For a single-session analysis, DENSITY requires one text file of
capture data and one text file with detector coordinates (the 'trap
layout' file). write.capthist
constructs names for these files
by appending ‘capt.txt’ and ‘trap.txt’ to filestem
which
defaults to the name of the capthist object. If filestem
is
empty then output goes to the console.
If object
contains multiple sessions with differing
traps
then a separate trap layout file is exported for each
session and each file name includes the session name. All capture data
are exported to one file regardless of the number of sessions. The
DENSITY format used is ‘TrapID’ except when x-y coordinates are
specific to a detection (i.e., polygon and transect detectors).
covariates
controls the export of both detector and individual
covariates. If it is TRUE or FALSE then it is taken to apply to
both. A vector of covariate names is used as a lookup for both
detector and capthist covariate fields: covariates are exported if
their name matches; this may be used to export any combination of
(uniquely named) detector and capthist covariates.
Existing text files will be replaced without warning. In the case of a
multi-session capthist file, session names are taken from
object
rather than sess
. Session names are truncated to
17 characters with blanks and commas removed.
To export data in comma-delimited (‘.csv’) format, pass sep =
","
in .... The resulting files have extension ‘.csv’ rather than
‘.txt’ and may be opened with spreadsheet software.
write.capthist
does not work for mark–resight data.
The original DENSITY formats accommodate ‘single’, ‘multi’ and ‘proximity’
data. Data for the newer detector types (‘count’, ‘signal’, ‘polygon’,
‘polygonX’, ‘transect’, ‘transectX’ and ‘telemetryonly’) may be input using the
DENSITY formats with minor variations. They may also be output with
write.capthist
, but a warning is given that DENSITY does not
understand these data types. See detector
and
secr-datainput.pdf for more.
The ... argument is useful for some special cases. For example, if
your input uses ‘;’ instead of ‘#’ for comments (‘;’ is also valid in
DENSITY) then set comment.char = ";"
in read.capthist
.
In a similar fashion, write comma- or tab-separated values by
setting sep = ","
or sep = "\t"
respectively.
The arguments of count.fields
are a subset of those of
read.table
so ... is limited to any of {sep, quote,
skip, blank.lines.skip, comment.char}.
If you fail to set fmt
correctly in read.capthist
then the
error message from verify
may be uninformative.
Efford, M. G. (2012) DENSITY 5.0: software for spatially explicit capture–recapture. Department of Mathematics and Statistics, University of Otago, Dunedin, New Zealand https://www.otago.ac.nz/density/.
read.telemetry
, read.traps
, make.capthist
,
write.captures
, write.traps
, read.table
## export ovenbird capture histories ## the files "ovenCHcapt.txt" and "ovenCHtrap.txt" are ## placed in the current folder (check with getwd() or dir()) ## Not run: write.capthist(ovenCH) ## End(Not run)
## export ovenbird capture histories ## the files "ovenCHcapt.txt" and "ovenCHtrap.txt" are ## placed in the current folder (check with getwd() or dir()) ## Not run: write.capthist(ovenCH) ## End(Not run)
Read coordinates of points on a habitat mask from a text file.
read.mask(file = NULL, data = NULL, spacing = NULL, columns = NULL, ...)
read.mask(file = NULL, data = NULL, spacing = NULL, columns = NULL, ...)
file |
character string with name of text file |
data |
dataframe |
spacing |
spacing of grid points in metres |
columns |
character vector naming the columns to save as covariates |
... |
other arguments to pass to |
For file input, the x and y coordinates are usually the first two values
on each line, separated by white space. If the file starts with a line
of column headers and ‘header = TRUE’ is passed to read.table
in
the ... argument then ‘x’ and ‘y’ need not be the first two fields.
data
is an alternative input route if the x and y coordinates
already exist in R as columns in a dataframe. Only one of data
or file
should be specified.
The grid cell size spacing
should be provided if known. If it is
not provided then an attempt is made to infer it from the minimum
spacing of points. This can be slow and may demand more memory than is
available. In rare cases (highly fragmented masks) it may also yield the
wrong answer.
From 2.3.0, additional columns in the input are saved as covariates. The
default (columns = NULL
) is to save all columns.
object of class mask
with type ‘user’
read.mask
creates a single-session mask. If used in
secr.fit
with a multi-session capthist object a
single-session mask will be replicated to the number of sessions. This
is appropriate if all sessions relate to the same geographical
region. If the ‘sessions’ relate to different regions you will need to
construct a multi-session mask as a list of single-session
masks (e.g. mask <- list(mask1, mask2, mask3)
).
## Replace file name with a valid local name and remove `#' # read.mask (file = "c:\\myfolder\\mask.txt", # spacing = 3, header = TRUE) ## "mask.txt" should have lines like this # x y # 265 265 # 268 265 # ...
## Replace file name with a valid local name and remove `#' # read.mask (file = "c:\\myfolder\\mask.txt", # spacing = 3, header = TRUE) ## "mask.txt" should have lines like this # x y # 265 265 # 268 265 # ...
A shortcut function for constructing a telemetry capthist object from a file of
telemetry fixes. Telemetry data are generally similar in format to polygon data
(see also addTelemetry
).
read.telemetry(file = NULL, data = NULL, covnames = NULL, verify = TRUE, ...)
read.telemetry(file = NULL, data = NULL, covnames = NULL, verify = TRUE, ...)
file |
character name of text file |
data |
data.frame containing coordinate data (alternative to |
covnames |
character vector of names for individual covariates |
verify |
logical for whether to check input |
... |
other arguments passed to countfields, read.table etc. |
Input data may be in a text file (argument file
) or a dataframe
(argument data
). Data should be in the XY format for function 'read.capthist'
i.e. the first 5 columns should be Session, ID, Occasion, X, Y. Further columns are
treated as individual covariates.
No ‘traps’ input is required. A traps object is generated automatically.
An secr capthist object including attribute ‘telemetryxy’ with the x-y coordinates, and a ‘traps’ object with detector type = ‘telemetry’
## Not run: olddir <- setwd('D:/bears/alberta') ## peek at raw data head(readLines('gps2008.txt')) gps2008CH <- read.telemetry("gps2008.txt") setwd(olddir) plot( gps2008CH, gridsp = 10000) head(gps2008CH) secr.fit(gps2008CH, start = log(4000), detectfn = 'HHN', details = list(telemetryscale = 1e12)) ## End(Not run)
## Not run: olddir <- setwd('D:/bears/alberta') ## peek at raw data head(readLines('gps2008.txt')) gps2008CH <- read.telemetry("gps2008.txt") setwd(olddir) plot( gps2008CH, gridsp = 10000) head(gps2008CH) secr.fit(gps2008CH, start = log(4000), detectfn = 'HHN', details = list(telemetryscale = 1e12)) ## End(Not run)
Construct an object of class traps
with detector locations from a text file or data frame. Usage per occasion and covariates may be included. Data may also be read from an Excel spreadsheet (see secr-datainput.pdf).
read.traps(file = NULL, data = NULL, detector = "multi", covnames = NULL, binary.usage = TRUE, markocc = NULL, trapID = NULL, ...)
read.traps(file = NULL, data = NULL, detector = "multi", covnames = NULL, binary.usage = TRUE, markocc = NULL, trapID = NULL, ...)
file |
character string with name of text file |
data |
data frame of detector coordinates |
detector |
character string for detector type |
covnames |
character vector of names for detector covariate fields |
binary.usage |
logical; if FALSE will read usage fields as continuous effort |
markocc |
integer vector distinguishing marking occasions (1) from sighting occasions (0) |
trapID |
character column containing detector names (see Details) |
... |
other arguments to pass to |
Reads a text file in which the first column is a character string (see Note) identifying a detector and the next two columns are its x- and y-coordinates, separated by white space. The coordinates optionally may be followed by a string of codes ‘0’ or ‘1’ indicating whether the detector was operated on each occasion. Trap-specific covariates may be added at the end of the line preceded by ‘/’. This format is compatible with the Density software (Efford 2012), except that all detectors are assumed to be of the same type (usage codes greater than 1 are treated as 1), and more than one covariate may be specified.
If file
is missing then x-y coordinates will be taken instead
from data
, which should include columns ‘x’ and ‘y’. Row names of
data
are read as detector identifiers unless trapID
is specified.
This option does not allow for covariates
or usage
,
but they maybe added later.
detector
specifies the behaviour of the detector following Efford
et al. (2009). ‘single’ refers to a trap that is able to catch at most
one animal at a time; ‘multi’ refers to a trap that may catch more than
one animal at a time. For both ‘single’ and ‘multi’ detectors a trapped
animals can appear at only one detector per occasion. Detectors of type
‘proximity’, such as camera traps and hair snags for DNA sampling, allow
animals to be recorded at several detectors on one occasion. See
detector
for further detector types.
For polygon and transect detector types, each line corresponds to a vertex and starts with a code to identify the polygon or transect (hence the same code appears on 2 or more lines). For input from a dataframe the code column should be named ‘polyID’. Also, usage and covariates are for the polygon or transect as a whole and not for each vertex. Usage and covariates are appended to the end of the line, just as for point detectors (traps etc.). The usage and covariates for each polygon or transect are taken from its first vertex. Although the end-of-line strings of other vertices are not used, they cannot be blank and should use the same spacing as the first vertex.
An object of class traps
comprising a data frame of x- and
y-coordinates, the detector type (‘single’, ‘multi’, ‘proximity’,
‘count’, ‘polygon’ etc.), and possibly other attributes.
Detector names, which become row names in the traps object, should not contain underscores.
Prior to 4.3.1 the function did not read usage or covariates from xls or data input.
Efford, M. G. (2012) DENSITY 5.0: software for spatially explicit capture–recapture. Department of Mathematics and Statistics, University of Otago, Dunedin, New Zealand. https://www.otago.ac.nz/density/.
Efford, M. G., Borchers D. L. and Byrom, A. E. (2009) Density estimation by spatially explicit capture-recapture: likelihood-based methods. In: D. L. Thomson, E. G. Cooch and M. J. Conroy (eds) Modeling Demographic Processes in Marked Populations. Springer, New York. Pp. 255–269.
## Not run: ## "trap.txt" should have lines like this # 1 365 365 # 2 365 395 # 3 365 425 # etc. ## in following, replace file name with a valid local name filename <- paste0(system.file("extdata", package = "secr"), '/trap.txt') tr1 <- read.traps (filename, detector = "proximity") summary(tr1) ## Or if we have a dataframe of coordinates... mytrapdf <- data.frame(x = c(365,365,365), y = c(365,395,425), row.names = c('A','B','C')) mytrapdf # x y # A 365 365 # B 365 395 # C 365 425 ## ...then we can convert it to a `traps' object with tr2 <- read.traps(data = mytrapdf) summary(tr2) ## End(Not run)
## Not run: ## "trap.txt" should have lines like this # 1 365 365 # 2 365 395 # 3 365 425 # etc. ## in following, replace file name with a valid local name filename <- paste0(system.file("extdata", package = "secr"), '/trap.txt') tr1 <- read.traps (filename, detector = "proximity") summary(tr1) ## Or if we have a dataframe of coordinates... mytrapdf <- data.frame(x = c(365,365,365), y = c(365,395,425), row.names = c('A','B','C')) mytrapdf # x y # A 365 365 # B 365 395 # C 365 425 ## ...then we can convert it to a `traps' object with tr2 <- read.traps(data = mytrapdf) summary(tr2) ## End(Not run)
Convert a mask or Dsurface with an irregular outline into a mask or Dsurface with a rectangular outline and the same bounding box. This enables contour plotting.
rectangularMask(mask)
rectangularMask(mask)
mask |
object of class mask or Dsurface |
The covariates of new points are set to missing. The operation may be
reversed (nearly) with subset(rectmask, attr(rectmask, "OK"))
.
The results are unpredictable if the mask has been rotated.
A rectangular mask or Dsurface with the same ‘area’, ‘boundingbox’,
‘meanSD’, ‘polygon’ and ‘polygon.habitat’ attributes as mask
. A
logical vector attribute ‘OK’ is added identifying the points
inherited from mask
.
rMask <- rectangularMask(possummask) plot(rMask) plot(possummask, add = TRUE, col = "blue")
rMask <- rectangularMask(possummask) plot(rMask) plot(possummask, add = TRUE, col = "blue")
Combine columns in a matrix-like object to create a new data set using the first non-zero value.
reduce (object, ...) ## Default S3 method: reduce(object, columns, ...)
reduce (object, ...) ## Default S3 method: reduce(object, columns, ...)
object |
object that may be coerced to a matrix |
columns |
list in which each component is a vector of subscripts for columns to be pooled |
... |
other arguments (not used currently) |
The first element of columns
defines the columns of object
for the first new column, the second for the second new column etc.
This is a generic method. More useful methods exist for capthist
and
traps
objects.
A matrix with number of columns equal to length(columns)
.
capthist
, reduce.capthist
, reduce.traps
## matrix with random zeros temp <- matrix(runif(20), nc = 4) temp[sample(20,10)] <- 0 temp reduce(temp, list(1:2, 3:4))
## matrix with random zeros temp <- matrix(runif(20), nc = 4) temp[sample(20,10)] <- 0 temp reduce(temp, list(1:2, 3:4))
Use these methods to combine data from multiple occasions or multiple detectors in a
capthist
or traps
object, creating a new data set and possibly converting
between detector types.
## S3 method for class 'traps' reduce(object, newtraps = NULL, newoccasions = NULL, span = NULL, rename = FALSE, newxy = c('mean', 'first'), ...) ## S3 method for class 'capthist' reduce(object, newtraps = NULL, span = NULL, rename = FALSE, newoccasions = NULL, by = 1, outputdetector = NULL, select = c("last","first","random"), dropunused = TRUE, verify = TRUE, sessions = NULL, ...)
## S3 method for class 'traps' reduce(object, newtraps = NULL, newoccasions = NULL, span = NULL, rename = FALSE, newxy = c('mean', 'first'), ...) ## S3 method for class 'capthist' reduce(object, newtraps = NULL, span = NULL, rename = FALSE, newoccasions = NULL, by = 1, outputdetector = NULL, select = c("last","first","random"), dropunused = TRUE, verify = TRUE, sessions = NULL, ...)
object |
|
newtraps |
list in which each component is a vector of subscripts for detectors to be pooled |
newoccasions |
list in which each component is a vector of subscripts for occasions to be pooled |
span |
numeric maximum span in metres of new detector |
rename |
logical; if TRUE the new detectors will be numbered from 1, otherwise a name will be constructed from the old detector names |
newxy |
character; coordinates when detectors grouped with ‘newtraps’ |
by |
number of old occasions in each new occasion |
outputdetector |
character value giving detector type for output (defaults to input) |
select |
character value for method to resolve conflicts |
dropunused |
logical, if TRUE any never-used detectors are dropped |
verify |
logical, if TRUE the |
sessions |
vector of session indices or names (optional) |
... |
other arguments passed by reduce.capthist to reduce.traps, or by reduce.traps to hclust |
reduce.traps –
Grouping may be specified explicitly via newtraps
, or
implicitly by span
.
If span
is specified a clustering of detector sites will be
performed with hclust
and detectors will be assigned to
groups with cutree
. The default algorithm in hclust
is complete linkage, which tends to yield compact, circular clusters;
each will have diameter less than or equal to span
.
newxy = 'first'
selects the coordinates of the first detector in a
group defined by ‘newtraps’, rather then the average of all detectors in group.
reduce.capthist –
The first component of newoccasions
defines the columns of
object
for new occasion 1, the second for new occasion 2, etc. If
newoccasions
is NULL then all occasions are output. Subscripts in a
component of newoccasions
that do not match an occasion in the input
are ignored. When the output detector is one of the trap types
(‘single’, ‘multi’), reducing capture occasions can result in locational
ambiguity for individuals caught on more than one occasion, and for
single-catch traps there may also be conflicts between individuals at
the same trap. The method for resolving conflicts among ‘multi’
detectors is determined by select
which should be one of ‘first’,
‘last’ or ‘random’. With ‘single’ detectors select
is ignored and
the method is: first, randomly select* one trap per animal per day;
second, randomly select* one animal per trap per day; third, when
collapsing multiple days use the first capture, if any, in each
trap.
Usage data in the traps
attribute are also pooled if present;
usage is summed over contributing occasions and detectors. If there is
no 'usage' attribute in the input, and outputdetector
is one of
'count', 'polygon', 'transect' and 'telemetry', a homogeneous (all-1's)
'usage' attribute is first generated for the input.
* i.e., in the case of a single capture, use that capture; in the case of multiple ‘competing’ captures draw one at random.
If newoccasions
is not provided then old occasions are grouped into
new occasions as indicated by the by
argument. For example, if
there are 15 old occasions and by = 5
then new occasions will be
formed from occasions 1:5, 6:10, and 11:15. A warning is given when the
number of old occasions is not a multiple of by
as then the final
new occasion will comprise fewer old occasions.
dropunused = TRUE
has the possibly unintended effect of dropping
whole occasions on which there were no detections.
A special use of the by
argument is to combine all occasions into
one for each session in a multi-session dataset. This is done by setting
by = "all"
.
reduce.capthist
may be used with non-spatial capthist objects
(NULL 'traps' attribute) by setting verify = FALSE
.
reduce.traps –
An object of class traps with detectors combined according to
newtraps
or span
. The new object has an attribute
‘newtrap’, a vector of length equal to the original number of
detectors. Each element in newtrap is the index of the new detector to
which the old detector was assigned (see Examples).
The object has no clusterID or clustertrap attribute.
reduce.capthist –
An object of class capthist with number of occasions (columns) equal to
length(newoccasions)
; detectors may simulataneously be aggregated
as with reduce.traps
. The detector type is inherited from object
unless a new type is specified with the argument
outputdetector
.
The argument named ‘columns’ was renamed to ‘newoccasions’ in version 2.5.0, and arguments were added to reduce.capthist for the pooling of detectors. Old code should work as before if all arguments are named and ‘columns’ is changed.
The reduce method may be used to re-assign the detector type (and
hence data format) of a capthist object without combining occasions or
detectors. Set the object
and outputdetector
arguments
and leave others at their default values.
Automated clustering can produce unexpected outcomes. In particular,
there is no guarantee that clusters will be equal in size. You should
inspect the results of reduce.traps especially when using span
.
reduce.traps
is not implemented for polygons or transects.
The function discretize
converts polygon data to
point-detector (multi, proximity or count) data.
capthist
, subset.capthist
,
discretize
, hclust
, cutree
tempcapt <- sim.capthist (make.grid(nx = 6, ny = 6), nocc = 6) class(tempcapt) pooled.tempcapt <- reduce(tempcapt, newocc = list(1,2:3,4:6)) summary (pooled.tempcapt) pooled.tempcapt2 <- reduce(tempcapt, by = 2) summary (pooled.tempcapt2) ## collapse multi-session dataset to single-session 'open population' onesess <- join(reduce(ovenCH, by = "all")) summary(onesess) # group detectors within 60 metres plot (traps(captdata)) plot (reduce(captdata, span = 60), add = TRUE) # plot linking old and new old <- traps(captdata) new <- reduce(old, span = 60) newtrap <- attr(new, "newtrap") plot(old, border = 10) plot(new, add = TRUE, detpar = list(pch = 16), label = TRUE) segments (new$x[newtrap], new$y[newtrap], old$x, old$y) ## Not run: # compare binary proximity with collapsed binomial count # expect TRUE for each year for (y in 1:5) { CHA <- abs(ovenCHp[[y]]) ## abs() to ignore one death usage(traps(CHA)) <- matrix(1, 44, ncol(CHA)) CHB <- reduce(CHA, by = 'all', output = 'count') # summary(CHA, terse = TRUE) # summary(CHB, terse = TRUE) fitA <- secr.fit(CHA, buffer = 300, trace = FALSE) fitB <- secr.fit(CHB, buffer = 300, trace = FALSE, binomN = 1, biasLimit = NA) A <- predict(fitA)[,-1] B <- predict(fitB)[,-1] cat(y, ' ', all(abs(A-B)/A < 1e-5), '\n') } ## multi-session fit ## expect TRUE overall CHa <- ovenCHp for (y in 1:5) { usage(traps(CHa[[y]])) <- matrix(1, 44, ncol(CHa[[y]])) CHa[[y]][,,] <- abs(CHa[[y]][,,]) } CHb <- reduce(CHa, by = 'all', output = 'count') summary(CHa, terse = TRUE) summary(CHb, terse = TRUE) fita <- secr.fit(CHa, buffer = 300, trace = FALSE) fitb <- secr.fit(CHb, buffer = 300, trace = FALSE, binomN = 1, biasLimit = NA) A <- predict(fita)[[1]][,-1] B <- predict(fitb)[[1]][,-1] all(abs(A-B)/A < 1e-5) ## End(Not run)
tempcapt <- sim.capthist (make.grid(nx = 6, ny = 6), nocc = 6) class(tempcapt) pooled.tempcapt <- reduce(tempcapt, newocc = list(1,2:3,4:6)) summary (pooled.tempcapt) pooled.tempcapt2 <- reduce(tempcapt, by = 2) summary (pooled.tempcapt2) ## collapse multi-session dataset to single-session 'open population' onesess <- join(reduce(ovenCH, by = "all")) summary(onesess) # group detectors within 60 metres plot (traps(captdata)) plot (reduce(captdata, span = 60), add = TRUE) # plot linking old and new old <- traps(captdata) new <- reduce(old, span = 60) newtrap <- attr(new, "newtrap") plot(old, border = 10) plot(new, add = TRUE, detpar = list(pch = 16), label = TRUE) segments (new$x[newtrap], new$y[newtrap], old$x, old$y) ## Not run: # compare binary proximity with collapsed binomial count # expect TRUE for each year for (y in 1:5) { CHA <- abs(ovenCHp[[y]]) ## abs() to ignore one death usage(traps(CHA)) <- matrix(1, 44, ncol(CHA)) CHB <- reduce(CHA, by = 'all', output = 'count') # summary(CHA, terse = TRUE) # summary(CHB, terse = TRUE) fitA <- secr.fit(CHA, buffer = 300, trace = FALSE) fitB <- secr.fit(CHB, buffer = 300, trace = FALSE, binomN = 1, biasLimit = NA) A <- predict(fitA)[,-1] B <- predict(fitB)[,-1] cat(y, ' ', all(abs(A-B)/A < 1e-5), '\n') } ## multi-session fit ## expect TRUE overall CHa <- ovenCHp for (y in 1:5) { usage(traps(CHa[[y]])) <- matrix(1, 44, ncol(CHa[[y]])) CHa[[y]][,,] <- abs(CHa[[y]][,,]) } CHb <- reduce(CHa, by = 'all', output = 'count') summary(CHa, terse = TRUE) summary(CHb, terse = TRUE) fita <- secr.fit(CHa, buffer = 300, trace = FALSE) fitb <- secr.fit(CHb, buffer = 300, trace = FALSE, binomN = 1, biasLimit = NA) A <- predict(fita)[[1]][,-1] B <- predict(fitb)[[1]][,-1] all(abs(A-B)/A < 1e-5) ## End(Not run)
Estimate the expected and realised populations in a region,
using a fitted spatially explicit capture–recapture model. Density is
assumed to follow an inhomogeneous Poisson process in two
dimensions. Expected is the volume under a fitted density
surface; realised
is the number of individuals within the region
for the current realisation of the process (cf Johnson et al. 2010; see
Note).
region.N(object, ...) ## S3 method for class 'secr' region.N(object, region = NULL, spacing = NULL, session = NULL, group = NULL, se.N = TRUE, alpha = 0.05, loginterval = TRUE, keep.region = FALSE, nlowerbound = TRUE, RN.method = "poisson", pooled.RN = FALSE, ncores = NULL, ...) ## S3 method for class 'secrlist' region.N(object, region = NULL, spacing = NULL, session = NULL, group = NULL, se.N = TRUE, alpha = 0.05, loginterval = TRUE, keep.region = FALSE, nlowerbound = TRUE, RN.method = "poisson", pooled.RN = FALSE, ncores = NULL, ...)
region.N(object, ...) ## S3 method for class 'secr' region.N(object, region = NULL, spacing = NULL, session = NULL, group = NULL, se.N = TRUE, alpha = 0.05, loginterval = TRUE, keep.region = FALSE, nlowerbound = TRUE, RN.method = "poisson", pooled.RN = FALSE, ncores = NULL, ...) ## S3 method for class 'secrlist' region.N(object, region = NULL, spacing = NULL, session = NULL, group = NULL, se.N = TRUE, alpha = 0.05, loginterval = TRUE, keep.region = FALSE, nlowerbound = TRUE, RN.method = "poisson", pooled.RN = FALSE, ncores = NULL, ...)
object |
|
region |
mask object defining the possibly non-contiguous region for which population size is required, or vector polygon(s) (see Details) |
spacing |
spacing between grid points (metres) if region mask is constructed on the fly |
session |
character session |
group |
group – for future use |
se.N |
logical for whether to estimate SE( |
alpha |
alpha level for confidence intervals |
loginterval |
logical for whether to base interval on log(N) |
keep.region |
logical for whether to save the raster region |
nlowerbound |
logical for whether to use n as lower bound when computing log interval for realised N |
RN.method |
character string for method used to calculate realised N (RN) and its sampling variance. ‘poisson’ or ‘MSPE’. |
pooled.RN |
logical; if TRUE the estimate of realised N for a multi-session model is computed as if for combined sampling with all detectors (see Details) |
ncores |
integer number of threads to be used for parallel processing |
... |
other arguments (not used) |
If the density surface of the fitted model is flat
(i.e. object$model$D == ~1
or object$CL == TRUE
) then
is simply the density multiplied by the area of
region
,
and the standard error is also a simple product. In the conditional
likelihood case, the density and standard error are obtained by first
calling derived
.
If, on the other hand, the density has been modelled then the density
surface is predicted at each point in region
and is
obtained by discrete summation. Pixel size may have a minor effect on
the result - check by varying
spacing
. Sampling variance is
determined by the delta method, using a numerical approximation to the
gradient of with respect to each beta parameter.
The region may be defined as a mask object (if omitted, the mask
component of object
will be used). Alternatively, region
may be a SpatialPolygonsDataFrame object (see package sp), and a
raster mask will be constructed on the fly using the specified
spacing. See make.mask
for an example importing a
shapefile to a SpatialPolygonsDataFrame.
Note: The option of specifying a polygon rather than a mask for
region
does not work if the density model in object
uses
spatial covariates: these must be passed in a mask.
Group-specific N has yet to be implemented.
Population size is adjusted automatically for the number of clusters
in ‘mashed’ models (see mash
). However, the population
size reported is that associated with a single cluster unless
regionmask
is specified.
pooled.RN = TRUE
handles the special case of a multi-session
model in which the region of interest spans several patches (i.e.,
sampling in each session is localised within region
. This is not
yet fully implemented.
Setting ncores = NULL
uses the existing value from the environment variable
RCPP_PARALLEL_NUM_THREADS (see setNumThreads
).
Use par.region.N
to apply region.N
in parallel to
several models.
If se.N = FALSE
, the numeric value of expected population size,
otherwise, a dataframe with rows ‘E.N’ and ‘R.N’, and columns as
below.
estimate | estimate of N (expected or realised, depending on row) |
SE.estimate | standard error of estimated N |
lcl | lower 100(1--alpha)% confidence limit |
ucl | upper 100(1--alpha)% confidence limit |
n | total number of individuals detected |
For multiple sessions, the value is a list with one component per session, each component as above.
If keep.region = TRUE
then the mask object for the region is
saved as the attribute ‘region’ (see Examples).
The area in hectares of the region is saved as attribute ‘regionarea’.
The estimates of expected and realised are generally very
similar, or identical, but realised
usually has lower
estimated variance, especially if the
detected animals
comprise a large fraction.
Realised is given by
(the second term
represents undetected animals). This definition strictly holds only
when region B is at least as large as the region of integration used
to fit the model; only with this condition can we be sure all
detected animals have centres within B. The sampling variance of
, technically a mean square prediction error (Johnson et al.
2010), is approximated by summing the expected Poisson variance of the
true number of undetected animals and a delta-method estimate of its
sampling variance, obtained as for
.
By default, a shortcut is used to compute the sampling variance of
realised . With this option (RN.method = ‘poisson’) the
sampling variance is the sampling variance of
minus the
estimate of
(representing Poisson process variance). This
has been found to give reliable confidence intervals in simulations
(Efford and Fewster 2013).
If RN.method is neither ‘MSPE’ nor ‘poisson’ (ignoring case) then
the estimate of expected is also used for realised
,
and the ‘poisson’ shortcut variance is used.
Johnson et al. (2010) use the notation for expected
and
for realised
in region
.
In our case, the relative SE (CV) of is the same as that
for the estimated density
if
has been estimated using
the Poisson distribution option in
secr.fit
or
derived()
. If has been estimated with the binomial
distribution option, its relative SE for simple models will be the
same as that of
, assuming that
is the full extent
of the original mask.
Borchers, D. L. and Efford, M. G. (2008) Spatially explicit maximum likelihood methods for capture–recapture studies. Biometrics 64, 377–385.
Efford, M. G. and Fewster, R. M. (2013) Estimating population size by spatially explicit capture–recapture. Oikos 122, 918–928.
Johnson, D. S., Laake, J. L. and Ver Hoef, J. M. (2010) A model-based approach for making ecological inference from distance sampling data. Biometrics 66, 310–318.
secr.fit
, derived
, make.mask
,
expected.n
, closedN
## Not run: ## routine examples using arbitrary mask from model fit region.N(secrdemo.0) region.N(secrdemo.CL) region.N(ovenbird.model.D) ## region defined as vector polygon ## retain and plot region mask temp <- region.N(possum.model.0, possumarea, spacing = 40, keep.region = TRUE) temp plot (attr(temp, "region"), type = "l") ## End(Not run)
## Not run: ## routine examples using arbitrary mask from model fit region.N(secrdemo.0) region.N(secrdemo.CL) region.N(ovenbird.model.D) ## region defined as vector polygon ## retain and plot region mask temp <- region.N(possum.model.0, possumarea, spacing = 40, keep.region = TRUE) temp plot (attr(temp, "region"), type = "l") ## End(Not run)
A single-session capthist object is formed by RMarkInput into a dataframe that may be passed directly to RMark.
RMarkInput(object, grouped = FALSE, covariates = TRUE) unRMarkInput(df, covariates = TRUE)
RMarkInput(object, grouped = FALSE, covariates = TRUE) unRMarkInput(df, covariates = TRUE)
object |
secr capthist object |
grouped |
logical for whether to replace each group of identical capture histories with a single line |
covariates |
logical or character vector; see Details |
df |
dataframe with fields ‘ch’ and ‘freq’ |
To convert a multi-session object first collapse the sessions with
join
.
If covariates
is TRUE then all columns of individual covariates
in the input are appended as columns in the output. If
covariates
is a character-valued vector then only the specified
covariates will be appended.
If both grouped
and covariates
are specified in
RMarkInput
, grouped
will be ignored, with a warning.
For RMarkInput –
Dataframe with fields ch and freq. ‘ch’ is a character string of 0's
and 1's. If grouped = FALSE
the rownames are retained and the
value of ‘freq’ is 1 or -1. Negative values of ‘freq’ indicate
removal.
The dataframe also includes individual covariates specified with
covariates
.
The attribute ‘intervals’ is copied from ‘object’, if present; otherwise it is set to a vector of zeros (indicating a closed-population sample).
For unRMarkInput –
A single-session capthist object with no traps attribute and hence no detector type (i.e. non-spatial capture histories). Covariates are copied as requested.
From secr 4.6.6, missing values (.) in input capture histories are converted to NA in the output, with a warning. The resulting capthist is unusable until the NAs are removed.
In versions before 2.4.0, a spurious occasion was added by
RMarkInput
when grouped = FALSE. Thanks to Jeff Stetz for
spotting this.
The default value for grouped
changed to FALSE in secr 2.4.0.
Laake, J. and Rexstad E. (2008) Appendix C. RMark - an alternative approach to building linear models in MARK. In: Cooch, E. and White, G. (eds) Program MARK: A Gentle Introduction. 6th edition. Most recent edition available at www.phidot.org/software/mark/docs/book/.
## ovenCH is a 5-year mist-netting dataset ovenRD <- RMarkInput (join(ovenCH)) head(ovenRD) unRMarkInput(ovenRD) RMarkInput(deermouse.ESG, covariates = FALSE, grouped = TRUE) RMarkInput(deermouse.ESG, covariates = TRUE) ## Not run: ## fit robust-design model in RMark (MARK must be installed) library(RMark) MarkPath <- 'c:/MARK' ## adjust for your installation ovenRD.data <- process.data(ovenRD, model = "Robust", time.interval = attr(ovenRD, "intervals")) ovenRD.model <- mark(data = ovenRD.data, model = "Robust", model.parameters = list(p = list(formula = ~1, share = TRUE), GammaDoublePrime = list(formula = ~1), GammaPrime = list(formula = ~1), f0 = list(formula = ~1))) cleanup(ask = FALSE) ## End(Not run)
## ovenCH is a 5-year mist-netting dataset ovenRD <- RMarkInput (join(ovenCH)) head(ovenRD) unRMarkInput(ovenRD) RMarkInput(deermouse.ESG, covariates = FALSE, grouped = TRUE) RMarkInput(deermouse.ESG, covariates = TRUE) ## Not run: ## fit robust-design model in RMark (MARK must be installed) library(RMark) MarkPath <- 'c:/MARK' ## adjust for your installation ovenRD.data <- process.data(ovenRD, model = "Robust", time.interval = attr(ovenRD, "intervals")) ovenRD.model <- mark(data = ovenRD.data, model = "Robust", model.parameters = list(p = list(formula = ~1, share = TRUE), GammaDoublePrime = list(formula = ~1), GammaPrime = list(formula = ~1), f0 = list(formula = ~1))) cleanup(ask = FALSE) ## End(Not run)
Precision of parameter estimates from an SECR model, expressed as relative standard error.
RSE(fit, parm = NULL, newdata = NULL)
RSE(fit, parm = NULL, newdata = NULL)
fit |
secr or openCR fitted model |
parm |
character; names of one or more real parameters (default all) |
newdata |
dataframe of covariates for |
The relative standard error (RSE) of parameter is
.
For a parameter estimated using a log link with single coefficient , the RSE is also
.
This formula is used wherever applicable.
Named vector of RSE, or matrix if newdata has more than one row.
The less explicit abbreviation CV has been used for the same quantity (sometimes expressed as a percentage). CV is used also for the relative standard deviation of a distribution.
Efford, M. G. and Boulanger, J. 2019. Fast evaluation of study designs for spatially explicit capture–recapture. Methods in Ecology and Evolution 10, 1529–1535.
RSE(secrdemo.0)
RSE(secrdemo.0)
Creates a smoothed resource surface from a covariate of a mask. Smoothing entails summing the value in each pixel weighted by a detection kernel centred on the focal pixel. The detection kernel represents home-range utilization with spatial scale sigma. The resulting surface is equivalent to the denominator used by Royle et al. (2013) to normalize site-specific detection.
Rsurface(mask, sigma, usecov = NULL, alpha2 = 1, detectfn = 'HHN', z = 1, inverse = FALSE, scale = TRUE)
Rsurface(mask, sigma, usecov = NULL, alpha2 = 1, detectfn = 'HHN', z = 1, inverse = FALSE, scale = TRUE)
mask |
secr habitat mask object (single-session) |
sigma |
numeric spatial scale of home range model |
alpha2 |
numeric coefficient of spatial covariate |
usecov |
character name of resource covariate |
detectfn |
integer or character code for detection function |
z |
numeric shape parameter of home range model |
inverse |
logical; if TRUE the reciprocal of smoothed resource is returned |
scale |
logical; not used |
detectfn
may be uniform (‘UN’) or one of the
cumulative-hazard functions (‘HHN’, ‘HHR’, ‘HEX’,
‘HAN’, ‘HCG’) (or integer codes 4, 14:18; see
detectfn).
The default ‘HHN’ corresponds to a halfnormal function on the hazard scale, or a bivariate circular normal home range.
If usecov
is not named then it takes the value 1.0 for all points
on the mask and zero otherwise.
The Rsurface can be used implicitly to normalize detection probability when
fitting a model with detector-specific covariate equal to
usecov
(see details, but the process is intricate and not
fully documented).
An object with class c(‘Rsurface’, ‘mask’, ‘data.frame’) and covariate ‘Resource’
(other covariates are retained from the input mask). The attribute
‘scale’ is 1.0 if scale = FALSE
; otherwise it is the average of the
resource over the masked area.
Consider a focal pixel s and another point in the habitat mask
x, with distance . Weights are given by a kernel
. Typically the kernel
will be halfnormal
(detectfn = ‘HHN’) or exponential
(detectfn =
‘HEX’) (see detectfn for other possibilities).
If represents the covariate value at point
x, the summed resource availability at s is given by
This corresponds to the denominator of eqn 4 in Royle et al. (2013).
By default, the numerical values reported by Rsurface
are not raw
values. If
scale = TRUE
, values are standardized by
dividing by the mean: where
is the number of
pixels. Values of
are centred on 1.0.
If inverse = TRUE
, the numeric values are
or
as determined by
scale
.
Royle, J. A., Chandler, R. B., Sun, C. C. and Fuller, A. K. (2013) Integrating resource selection information with spatial capture–recapture. Methods in Ecology and Evolution 4, 520–530.
mask
, plot.Rsurface
,
spotHeight
, details
## create binary covariate (0 outside habitat) msk <- make.mask(traps(possumCH), buffer = 800) covariates(msk) <- data.frame(z = as.numeric(pointsInPolygon (msk,possumarea))) ## derive and plot "resource availability" Rs <- Rsurface(msk, sigma = 100, usecov = 'z') plot(Rs, plottype = 'contour', col = topo.colors(10)) lines(possumarea) if (interactive()) { spotHeight(Rs, dec = 2) }
## create binary covariate (0 outside habitat) msk <- make.mask(traps(possumCH), buffer = 800) covariates(msk) <- data.frame(z = as.numeric(pointsInPolygon (msk,possumarea))) ## derive and plot "resource availability" Rs <- Rsurface(msk, sigma = 100, usecov = 'z') plot(Rs, plottype = 'contour', col = topo.colors(10)) lines(possumarea) if (interactive()) { spotHeight(Rs, dec = 2) }
Compute score tests comparing a fitted model and a more general alternative model.
score.test(secr, ..., betaindex = NULL, trace = FALSE, ncores = NULL, .relStep = 0.001, minAbsPar = 0.1) score.table(object, ..., sort = TRUE, dmax = 10)
score.test(secr, ..., betaindex = NULL, trace = FALSE, ncores = NULL, .relStep = 0.001, minAbsPar = 0.1) score.table(object, ..., sort = TRUE, dmax = 10)
secr |
fitted secr model |
... |
one or more alternative models OR a fitted secr model |
trace |
logical. If TRUE then output one-line summary at each evaluation of the likelihood |
ncores |
integer number of threads for parallel processing |
.relStep |
see |
minAbsPar |
see |
betaindex |
vector of indices mapping fitted values to parameters in the alternative model |
object |
score.test object or list of such objects |
sort |
logical for whether output rows should be in descending order of AIC |
dmax |
threshold of dAIC for inclusion in model set |
Score tests allow fast model selection (e.g. Catchpole & Morgan 1996).
Only the simpler model need be fitted. This implementation uses the
observed information matrix, which may sometimes mislead (Morgan et al.
2007). The gradient and second derivative of the likelihood function are
evaluated numerically at the point in the parameter space of the second
model corresponding to the fit of the first model. This operation uses
the function fdHess
of the nlme package; the likelihood
must be evaluated several times, but many fewer times than would be
needed to fit the model. The score statistic is an approximation to the
likelihood ratio; this allows the difference in AIC to be estimated.
Covariates are inferred from components of the reference model
secr
. If the new models require additional covariates these may
usually be added to the respective component of secr
.
Mapping of parameters between the fitted and alternative models
sometimes requires user intervention via the betaindex
argument.
For example betaindex
= c(1,2,4) is the correct mapping when
comparing the null model (D1, g0
1,
sigma
1) to one with a behavioural effect on g0
(D
1, g0
b, sigma
1).
The arguments .relStep
and minAbsPar
control the numerical
gradient calculation and are passed directly to
fdHess
. More investigation is needed to determine
optimal settings.
score.table
summarises one or more score tests in the form of a
model comparison table. The ... argument here allows the inclusion of
additional score test objects (note the meaning differs from
score.test
). Approximate AIC values are used to compute relative
AIC model weights for all models within dmax AIC units of the best
model.
If ncores = NULL
then the existing value from the environment variable
RCPP_PARALLEL_NUM_THREADS is used (see setNumThreads
).
An object of class ‘score.test’ that inherits from ‘htest’, a list with components
statistic |
the value the chi-squared test statistic (score statistic) |
parameter |
degrees of freedom of the approximate chi-squared distribution of the test statistic (difference in number of parameters H0, H1) |
p.value |
probability of test statistic assuming chi-square distribution |
method |
a character string indicating the type of test performed |
data.name |
character string with null hypothesis, alternative hypothesis and arguments to function call from fit of H0 |
H0 |
simpler model |
np0 |
number of parameters in simpler model |
H1 |
alternative model |
H1.beta |
coefficients of alternative model |
AIC |
Akaike's information criterion, approximated from score statistic |
AICc |
AIC with small-sample adjustment of Hurvich & Tsai 1989 |
If ... defines several alternative models then a list of score.test objects is returned.
The output from score.table
is a dataframe with one row per model, including the reference model.
This implementation is experimental. The AIC values, and values derived from them, are approximations that may differ considerably from AIC values obtained by fitting and comparing the respective models. Use of the observed information matrix may not be optimal.
Weights were based on AICc rather than AIC prior to version 5.0.0.
Catchpole, E. A. and Morgan, B. J. T. (1996) Model selection of ring-recovery models using score tests. Biometrics 52, 664–672.
Hurvich, C. M. and Tsai, C. L. (1989) Regression and time series model selection in small samples. Biometrika 76, 297–307.
McCrea, R. S. and Morgan, B. J. T. (2011) Multistate mark-recapture model selection using score tests. Biometrics 67, 234–241.
Morgan, B. J. T., Palmer, K. J. and Ridout, M. S. (2007) Negative score test statistic. American statistician 61, 285–288.
## Not run: AIC (secrdemo.0, secrdemo.b) st <- score.test (secrdemo.0, g0 ~ b) st score.table(st) ## adding a time covariate to separate occasions (1,2) from (3,4,5) secrdemo.0$timecov <- data.frame(t2 = factor(c(1,1,2,2,2))) st2 <- score.test (secrdemo.0, g0 ~ t2) score.table(st,st2) ## End(Not run)
## Not run: AIC (secrdemo.0, secrdemo.b) st <- score.test (secrdemo.0, g0 ~ b) st score.table(st) ## adding a time covariate to separate occasions (1,2) from (3,4,5) secrdemo.0$timecov <- data.frame(t2 = factor(c(1,1,2,2,2))) st2 <- score.test (secrdemo.0, g0 ~ t2) score.table(st,st2) ## End(Not run)
These functions are no longer available in secr.
# Defunct in 5.0.0 derivedSystematic() # Defunct in 4.6.2 (2023-09-30) model.average() # Defunct in 4.5.10 (2023-03-10) ip.secr() pfn() # Defunct in 4.4.2 (2021-05-04) make.newdata() # Defunct in 4.4.0 (2021-05-01) secr.make.newdata() # Defunct in 4.0.0 (2019-10-27) read.SPACECAP() write.SPACECAP()
# Defunct in 5.0.0 derivedSystematic() # Defunct in 4.6.2 (2023-09-30) model.average() # Defunct in 4.5.10 (2023-03-10) ip.secr() pfn() # Defunct in 4.4.2 (2021-05-04) make.newdata() # Defunct in 4.4.0 (2021-05-01) secr.make.newdata() # Defunct in 4.0.0 (2019-10-27) read.SPACECAP() write.SPACECAP()
Some of these have stubs which report that they are defunct, but most have been removed completely (apart from being documented here).
model.average
is replaced by a method for ‘secr’ and ‘secrlist’ objects of the generic modelAverage
. The internal code is essentially the same for model.average
and modelAverage.secrlist
. The generic avoids a name conflict with RMark and is also used in openCR.
ip.secr
and pfn
have been superceded by ipsecr.fit
and
proxy.ms
in package ipsecr.
Internal functions secr.make.newdata
and make.newdata
were
replaced with makeNewData
generic from 4.4.2.
SPACECAP was removed from the CRAN archive on 2019-08-31.
These functions will be removed from future versions of secr or have been renamed.
par.secr.fit (arglist, ncores = 1, seed = NULL, trace = TRUE, logfile = "logfile.txt", prefix = "fit.", LB = FALSE, save.intermediate = FALSE)
par.derived (secrlist, ncores = 1, ...)
par.region.N (secrlist, ncores = 1, ...)
fx.total (object, sessnum = 1, mask = NULL, ncores = NULL, ...)
fxi.secr (object, i = NULL, sessnum = 1, X = NULL, ncores = NULL, ...)
buffer.contour (traps, buffer, nx = 64, convex = FALSE, ntheta = 100, plt = TRUE, add = FALSE, poly = NULL, poly.habitat = TRUE, fill = NULL, ...)
pdot.contour (traps, border = NULL, nx = 64, detectfn = 0, detectpar = list(g0 = 0.2, sigma = 25, z = 1), noccasions = NULL, binomN = NULL, levels = seq(0.1, 0.9, 0.1), poly = NULL, poly.habitat = TRUE, plt = TRUE, add = FALSE, fill = NULL, ...)
esa.plot (object, max.buffer = NULL, spacing = NULL, max.mask = NULL, detectfn, detectpar, noccasions, binomN = NULL, thin = 0.1, poly = NULL, poly.habitat = TRUE, session = 1, plt = TRUE, type = c('density', 'esa', 'meanpdot', 'CVpdot'), n = 1, add = FALSE, overlay = TRUE, conditional = FALSE, ...)
fxi.contour (object, i = 1, sessnum = 1, border = 100, nx = 64, levels = NULL, p = seq(0.1,0.9,0.1), plt = TRUE, add = FALSE, fitmode = FALSE, plotmode = FALSE, fill = NULL, output = c('list','sf','SPDF'), ncores = NULL, ...)
fxi.mode (object, i = 1, sessnum = 1, start = NULL, ncores = NULL, ...)
Since the introduction of multi-threading in secr 4.0 it is no longer efficient to use parallel worker processes.
list.secr.fit
replaces par.secr.fit
.
Functions par.derived and par.region.N can be replaced by a simple call of lapply (see Examples in list.secr.fit
).
Some functions have been renamed to avoid the ambiguous ".":
Old | New |
fxi.secr | fxi (S3 generic with method for secr objects) |
fxi.contour | fxiContour |
fxi.mode | fxiMode |
fx.total | fxTotal |
esa.plot | esaPlot |
pdot.contour | pdotContour |
buffer.contour | bufferContour |
secr-defunct
, list.secr.fit
, secr-version5
This document explains changes in secr 5.0. Version 5.0 is compatible in most respects with earlier versions, but a few names and one default have been changed without warning. See the NEWS file for a complete list of the changes over time.
Several new generic functions are defined, with methods specifically for ‘secr’
fitted models (esa
, fxi
, fxTotal
).
Some functions with "." in their name have been renamed to avoid confusion with methods for generics.
Where possible, the old names have been deprecated (tagged with a warning), and will continue to work for a while.
Old | New |
buffer.contour |
bufferContour |
esa.plot |
esaPlot |
fxi.contour |
fxiContour |
fxi.mode |
fxiMode |
fxi.secr |
fxi (generic) |
fx.total |
fxTotal (generic) |
pdot.contour |
pdotContour |
fxi.secr
has been replaced by the generic fxi
. Thus instead of fxi.secr(secrdemo.0, i = 1, X = c(365,605))
use fxi(secrdemo.0, i = 1, X = c(365,605))
.
AIC and related functions now default to criterion = "AIC" instead of criterion = "AICc".
Some of us have been uneasy for a long time about blanket use of the AICc small-sample adjustment to AIC (Hurvich and Tsai 1989). Royle et al. (2014) expressed doubts because the sample size itself is poorly defined. AICc is widely used, but AIC may be better for model averaging even when samples are small (Turek and Fletcher 2012; Fletcher 2019, p. 60).
blackbearCH
secr 5.0 includes a new black bear DNA hair snag dataset from the Great Smoky Mountains, Tennessee (thanks to J. Laufenberg, F. van Manen and J. Clark).
MCgof
The method of Choo et al. (2024) for emulating the Bayesian p-value
goodness-of-fit test (Gelman 1996, Royle et al. 2014) has been implemented
as the generic MCgof
with a method for ‘secr’ fitted models.
I thank Yan Ru Choo for his assistance.
This is a new approach and should be used with caution. Bugs may yet be found, and the power of the tests is limited.
These extensions allow MCgof
to cover a wider range of models:
detectpar
optionally returns values for each detector
pdot
accepts detector- and occasion-specific detection parameters
The code for area-search and transect-search models (detector types ‘polygonX’,
‘polygon’, ‘transectX’, ‘transect’) has been streamlined with a view to
removing it to another package. Simulation for these models (functions
sim.capthist
, sim.detect
) will remain in secr, but uses native R
functions rather than RcppNumerical of Qiu et al. (2023).
The undocumented detection function ‘HPX’ has been removed.
Choo, Y. R., Sutherland, C. and Johnston, A. (2024) A Monte Carlo resampling framework for implementing goodness-of-fit tests in spatial capture-recapture models Methods in Ecology and Evolution DOI: 10.1111/2041-210X.14386.
Efford, M. G. (2024) secr: Spatially explicit capture-recapture models. R package version 5.0.0. https://CRAN.R-project.org/package=secr
Fletcher, D. (2019) Model averaging. SpringerBriefs in Statistics. Berlin: Springer-Verlag.
Hurvich, C. M. and Tsai, C. L. (1989) Regression and time series model selection in small samples. Biometrika 76, 297–307.
Gelman, A., Meng, X.-L., and Stern, H. (1996) Posterior predictive assessment of model fitness via realized discrepancies. Statistica Sinica 6, 733–807.
Qiu, Y., Balan, S., Beall, M., Sauder, M., Okazaki, N. and Hahn, T. (2023) RcppNumerical: 'Rcpp' Integration for Numerical Computing Libraries. R package version 0.6-0. https://CRAN.R-project.org/package=RcppNumerical
Royle, J. A., Chandler, R. B., Sollmann, R. and Gardner, B. (2014) Spatial capture–recapture. Academic Press.
Turek, D. and Fletcher, D. (2012) Model-averaged Wald confidence intervals. Computational statistics and data analysis 56, 2809–2815.
Internal functions used by secr.fit
.
secr.design.MS (capthist, models, timecov = NULL, sessioncov = NULL, groups = NULL, hcov = NULL, dframe = NULL, naive = FALSE, CL = FALSE, keep.dframe = FALSE, full.dframe = FALSE, ignoreusage = FALSE, contrasts = NULL, ...) make.lookup (tempmat) insertdim (x, dimx, dims)
secr.design.MS (capthist, models, timecov = NULL, sessioncov = NULL, groups = NULL, hcov = NULL, dframe = NULL, naive = FALSE, CL = FALSE, keep.dframe = FALSE, full.dframe = FALSE, ignoreusage = FALSE, contrasts = NULL, ...) make.lookup (tempmat) insertdim (x, dimx, dims)
capthist |
|
models |
list of formulae for parameters of detection |
timecov |
optional dataframe of values of time (occasion-specific) covariate(s). |
sessioncov |
optional dataframe of values of session-specific covariate(s). |
groups |
optional vector of one or more variables with which to
form groups. Each element should be the name of a factor variable in
the |
hcov |
character name of an individual (capthist) covariate for known class membership in h2 models |
dframe |
optional data frame of design data for detection parameters |
naive |
logical if TRUE then modelled detection probability is for a naive animal (not caught previously); if FALSE then detection probability is contingent on individual's history of detection |
CL |
logical; TRUE for model to be fitted by maximizing the conditional likelihood |
keep.dframe |
logical; if TRUE the dataframe of design data is included in the output |
full.dframe |
logical; if FALSE then padding rows are purged from
output dframe (ignored if |
ignoreusage |
logical; if TRUE any usage attribute of traps(capthist) is ignored |
contrasts |
contrast specification as for |
... |
other arguments passed to the R function
|
tempmat |
matrix for which row lookup required |
x |
vector of character, numeric or factor values |
dimx |
vector of notional dimensions for x to fill in target array |
dims |
vector of notional dimensions of target array |
These are internal secr functions that you are unlikely ever to use.
Each real parameter is notionally different for each unique combination
of session, individual, occasion, detector and latent class, i.e., for
sessions,
individuals,
occasions,
detectors and
latent classes
there are potentially
different values. Actual models always predict a much reduced set
of distinct values, and the number of rows in the design matrix is
reduced correspondingly; a parameter index array allows these to
retrieved for any combination of session, individual, occasion and
detector.
The keep.dframe
option is provided for the rare occasions that a
user may want to check the data frame that is an intermediate step in
computing each design matrix with model.matrix
(i.e. the
data argument of model.matrix
).
... may be used to pass contrasts.arg
to model.matrix
.
For secr.design.MS
, a list with the components
designMatrices |
list of reduced design matrices, one for each real detection parameter |
parameterTable |
index to row of the reduced design matrix for
each real detection parameter; dim(parameterTable) = c(uniquepar, np),
where uniquepar is the number of unique combinations of paramater
values (uniquepar < |
PIA |
Parameter Index Array - index to row of parameterTable for a given session, animal, occasion and detector; dim(PIA) = c(R,n,S,K,M) |
R |
number of sessions |
If models
is empty then all components are NULL except for PIA
which is an array of 1's (M set to 1).
Optionally (keep.dframe = TRUE
) -
dframe |
dataframe of design data, one column per covariate, one row for each c(R,n,S,K,M). For multi-session models n, S, and K refer to the maximum across sessions |
validdim |
list giving the valid dimensions (n, S, K, M) before padding |
For make.lookup
, a list with components
lookup |
matrix of unique rows |
index |
indices in lookup of the original rows |
For insertdim
, a vector with length prod(dims) containing the
values replicated according to dimx.
secr.design.MS (captdata, models = list(g0 = ~b))$designMatrices secr.design.MS (captdata, models = list(g0 = ~b))$parameterTable ## peek at design data constructed for learned response model head(captdata) temp <- secr.design.MS (captdata, models = list(g0 = ~b), keep.dframe = TRUE) a1 <- temp$dframe$animal == 1 & temp$dframe$detector %in% 8:10 temp$dframe[a1,] ## ... and trap specific learned response model temp <- secr.design.MS (captdata, models = list(g0 = ~bk), keep.dframe = TRUE) a1 <- temp$dframe$animal == 1 & temp$dframe$detector %in% 8:10 temp$dframe[a1,] ## place values 1:6 in different dimensions insertdim(1:6, 1:2, c(2,3,6)) insertdim(1:6, 3, c(2,3,6))
secr.design.MS (captdata, models = list(g0 = ~b))$designMatrices secr.design.MS (captdata, models = list(g0 = ~b))$parameterTable ## peek at design data constructed for learned response model head(captdata) temp <- secr.design.MS (captdata, models = list(g0 = ~b), keep.dframe = TRUE) a1 <- temp$dframe$animal == 1 & temp$dframe$detector %in% 8:10 temp$dframe[a1,] ## ... and trap specific learned response model temp <- secr.design.MS (captdata, models = list(g0 = ~bk), keep.dframe = TRUE) a1 <- temp$dframe$animal == 1 & temp$dframe$detector %in% 8:10 temp$dframe[a1,] ## place values 1:6 in different dimensions insertdim(1:6, 1:2, c(2,3,6)) insertdim(1:6, 3, c(2,3,6))
Estimate animal population density with data from an array of passive
detectors (traps) by fitting a spatial detection model by maximizing the
likelihood. Data must have been assembled as an object of class
capthist
. Integration is by summation over the grid of points in
mask
.
secr.fit (capthist, model = list(D~1, g0~1, sigma~1), mask = NULL, buffer = NULL, CL = FALSE, detectfn = NULL, binomN = NULL, start = NULL, link = list(), fixed = list(), timecov = NULL, sessioncov = NULL, hcov = NULL, groups = NULL, dframe = NULL, details = list(), method = "Newton-Raphson", verify = TRUE, biasLimit = 0.01, trace = NULL, ncores = NULL, ...)
secr.fit (capthist, model = list(D~1, g0~1, sigma~1), mask = NULL, buffer = NULL, CL = FALSE, detectfn = NULL, binomN = NULL, start = NULL, link = list(), fixed = list(), timecov = NULL, sessioncov = NULL, hcov = NULL, groups = NULL, dframe = NULL, details = list(), method = "Newton-Raphson", verify = TRUE, biasLimit = 0.01, trace = NULL, ncores = NULL, ...)
capthist |
|
mask |
|
buffer |
scalar mask buffer radius if |
CL |
logical, if true then the model is fitted by maximizing the conditional likelihood |
detectfn |
integer code or character string for shape of detection function 0 = halfnormal, 1 = hazard rate etc. – see detectfn |
binomN |
integer code for distribution of counts (see Details) |
start |
vector of initial values for beta parameters, or |
link |
list with optional components corresponding to ‘real’ parameters (e.g., ‘D’, ‘g0’, ‘sigma’), each a character string in {"log", "logit", "identity", "sin"} for the link function of one real parameter |
fixed |
list with optional components corresponding to real parameters giving the scalar value to which the parameter is to be fixed |
model |
list with optional components each symbolically defining a linear predictor for one real parameter using |
timecov |
optional dataframe of values of time (occasion-specific) covariate(s). |
sessioncov |
optional dataframe of values of session-specific covariate(s). |
hcov |
character name of individual covariate for known membership of mixture classes. |
groups |
optional vector of one or more variables with which to form groups. Each element should be the name of a factor variable in the |
dframe |
optional data frame of design data for detection parameters |
details |
list of additional settings, mostly model-specific (see Details) |
method |
character string giving method for maximizing log likelihood |
verify |
logical, if TRUE the input data are checked with |
biasLimit |
numeric threshold for predicted relative bias due to buffer being too small |
trace |
logical, if TRUE then output each evaluation of the likelihood, and other messages |
ncores |
integer number of threads to use for parallel processing |
... |
other arguments passed to the maximization function |
secr.fit
fits a SECR model by maximizing the likelihood. The
likelihood depends on the detector type ("multi", "proximity", "count",
"polygon" etc.) of the traps
attribute of capthist
(Borchers and Efford 2008, Efford, Borchers and Byrom 2009, Efford,
Dawson and Borchers 2009, Efford 2011). The ‘multi’ form of the
likelihood is also used, with a warning, when detector type = "single"
(see Efford et al. 2009 for justification).
The default model
is null (model = list(D~1, g0~1,
sigma~1)
for detectfn = 'HN'
and CL = FALSE
), meaning
constant density and detection probability). The set of variables
available for use in linear predictors includes some that are
constructed automatically (t, T, b, B, bk, Bk, k, K), group (g), and
others that appear in the covariates
of the input data. See also
usage
for varying effort, timevaryingcov
to
construct other time-varying detector covariates, and secr-models.pdf
and secr-overview.pdf for more on
defining models.
buffer
and mask
are alternative ways to define the region
of integration (see mask). If mask
is not specified then a
mask of type "trapbuffer" will be constructed automatically using the
specified buffer width in metres.
hcov
is used to define a hybrid mixture model, used especially to
model sex differences (see hcov
). (Allows some animals to
be of unknown class).
The length of timecov
should equal the number of sampling
occasions (ncol(capthist)
). Arguments timecov
,
sessioncov
and groups
are used only when needed for terms
in one of the model specifications. Default link
is list(D="log",
g0="logit", sigma="log")
.
If start
is missing then autoini
is used for D, g0
and sigma, and other beta parameters are set initially to arbitrary
values, mostly zero. start
may be a previously fitted model. In
this case, a vector of starting beta values is constructed from the old
(usually nested) model and additional betas are set to zero. Mapping of
parameters follows the default in score.test
, but user
intervention is not allowed. From 2.10.0 the new and old models need not
share all the same ‘real’ parameters, but any new real parameters, such
as ‘pmix’ for finite mixture models, receive a starting value of 0 on
the link scale (remembering e.g., invlogit(0) = 0.5 for parameter ‘pmix’).
binomN
(previously a component of details
) determines the
distribution that is fitted for the number of detections of an individual
at a particular detector, on a particular occasion, when the detectors
are of type ‘count’, ‘polygon’ or ‘transect’:
binomN > 1 — binomial with size binomN
binomN = 1 — binomial with size determined by usage
binomN = 0 — Poisson
The default with these detectors is to fit a Poisson distribution.
details
is used for various specialized settings listed below. These are
described separately - see details
.
autoini | session to use for starting values (default 1) |
centred | centre x-y coordinates |
chat | overdispersion of sighting counts Tu, Tm |
chatonly | compute overdispersion for Tu and Tm, then exit |
contrasts | coding of factor predictors |
convexpolygon | allows non-convex polygons (slower) |
Dfn | reparameterization of density model (seldom used directly) |
Dlambda | switch density reparameterization to trend model |
distribution | binomial vs Poisson N |
externalpdot | name of mask covariate to substitute for (relative density) |
fastproximity | special handling of binary proximity detectors |
fixedbeta | specify fixed beta parameter(s) |
grain | grain argument of RcppParallel::parallelFor |
hessian | variance method |
ignoreusage | override usage in traps object of capthist |
intwidth2 | controls optimise when only one parameter |
knownmarks | known or unknown number of marked animals in sighting-only model |
LLonly | compute one likelihood for values in start |
maxdistance | distance threshold for selective mask |
miscparm | starting values for extra parameters fitted via userdist function |
newdetector | detector type to override detector(traps(capthist)) |
nsim | number of simulations to compute overdispersion |
param | optional parameterisation code |
relativeD | optional relative density conditional on |
savecall | optionally suppress saving of call |
telemetrytype | treat telemetry data as independent, dependent or concurrent |
normalize | rescale detection to individual range use |
usecov | spatial covariate of use for normalization |
userdist | user-provided distance function or matrix |
Setting ncores = NULL
uses the existing value from the environment variable
RCPP_PARALLEL_NUM_THREADS (see setNumThreads
).
A mark-resight model is fitted if the markocc
attribute of the capthist
‘traps’ object includes sighting occasions. See the vignette
secr-markresight.pdf
for a full account.
If method = "Newton-Raphson"
then nlm
is
used to maximize the log likelihood (minimize the negative log
likelihood); otherwise optim
is used with the
chosen method ("BFGS", "Nelder-Mead", etc.). If maximization fails a
warning is given appropriate to the method.
From secr 2.5.1, method = "none"
may be used to skip likelihood
maximization and compute only the hessian for the current dataset at the
values in start, and the corresponding variance-covariance matrix of
beta parameters. The computation uses fdHess from nlme.
If verify
= TRUE then verify
is called to check
capthist and mask; analysis is aborted if "errors" are found. Some
conditions that trigger an "error" are benign (e.g., no detections in
some sessions of a multi-session study of a sparse population); use
verify = FALSE
to avoid the check. See also Note.
If buffer
is used rather than mask
, and biasLimit
is valid, then the estimated density is checked for bias due to the
choice of buffer. A warning is generated when buffer
appears
to be too small (predicted RB(D-hat) > biasLimit
, default 1%
relative bias). The prediction uses bias.D
. No check
is performed when mask
is specified, when biasLimit
is 0,
negative or NA, or when the detector type is "polygon", "transect",
"polygonX" or "transectX".
Function list.secr.fit
is a way to fit several models at once.
When details$LLonly = TRUE
a single log-likelihood is returned, with attributes
npar |
number of parameters to be estimated, |
preptime |
elapsed setup time in seconds, |
LLtime |
elapsed time for single likelihood evaluation, exclusive of setup. |
Otherwise, secr.fit
returns an object of class secr representing the fitted SECR model. This has
components
call |
function call) |
capthist |
saved input |
mask |
saved input |
detectfn |
saved input |
CL |
saved input |
timecov |
saved input |
sessioncov |
saved input |
hcov |
saved input |
groups |
saved input |
dframe |
saved input |
designD |
design matrix for density model; may be NULL |
designNE |
design matrix for noneuc model; may be NULL |
design |
reduced design matrices for detection parameters, parameter table and parameter
index array for actual animals (see |
design0 |
reduced design matrices for detection parameters, parameter table and parameter
index array for ‘naive’ animal (see |
start |
vector of starting values for beta parameters |
link |
list with one component for each real parameter (typically ‘D’, ‘g0’, ‘sigma’),giving the name of the link function used for each real parameter. |
fixed |
saved input |
parindx |
list with one component for each real parameter giving the indices of the ‘beta’ parameters associated with each real parameter |
model |
saved input |
details |
saved input |
vars |
vector of unique variable names in |
betanames |
names of beta parameters |
realnames |
names of fitted (real) parameters |
fit |
list describing the fit (output from |
beta.vcv |
variance-covariance matrix of beta parameters |
smoothsetup |
list of objects specifying smooths in mgcv |
learnedresponse |
logical; TRUE if any learned response in detection model |
version |
secr version number |
starttime |
character string of date and time at start of fit |
proctime |
processor time for model fit, in seconds |
The environment variable RCPP_PARALLEL_NUM_THREADS is updated if an integer value is provided for ncores
.
** Mark-resight data formats and models are experimental in secr 2.10.0 and subject to change **
One system of units is used throughout secr. Distances are in metres and areas are in hectares (ha). The unit of density is animals per hectare. 1 ha = 10000 m^2 = 0.01 km^2. To convert density to animals / km^2, multiply by 100.
When you display an ‘secr’ object by typing its name at the command
prompt, you implicitly call its ‘print’ method print.secr
, which
in turn calls predict.secr
to tabulate estimates of the ‘real’
parameters. Confidence limits (lcl, ucl) are for a 100(1-alpha)%
interval, where alpha defaults to 0.05 (95% interval); alpha may be
varied in print.secr
or predict.secr
.
AIC
, logLik
and vcov
methods are also
provided. Take care with using AIC: not all models are comparable (see
Notes section of AIC.secr
) and large differences in AIC
may relate to trivial differences in estimated density.
derived
is used to compute the derived parameters ‘esa’
(effective sampling area) and ‘D’ (density) for models fitted by
maximizing the conditional likelihood (CL = TRUE).
Components ‘version’ and ‘starttime’ were introduced in version 1.2.7, and recording of the completion time in ‘fitted’ was discontinued.
The Newton-Raphson algorithm is fast, but it sometimes fails to compute the information matrix correctly, causing some or all standard errors to be set to NA. This usually indicates a major problem in fitting the model, and parameter estimates should not be trusted. See Troubleshooting.
The component D in output was replaced with N from version 2.3. Use
region.N
to obtain SE or confidence intervals for N-hat,
or to infer N for a different region.
Prior to version 2.3.2 the buffer bias check could be switched off by
setting verify = FALSE
. This is now done by setting
biasLimit = 0
or biasLimit = NA
.
Borchers, D. L. and Efford, M. G. (2008) Spatially explicit maximum likelihood methods for capture–recapture studies. Biometrics 64, 377–385.
Efford, M. G. (2004) Density estimation in live-trapping studies. Oikos 106, 598–610.
Efford, M. G. (2011) Estimation of population density by spatially explicit capture–recapture with area searches. Ecology 92, 2202–2207.
Efford, M. G., Borchers D. L. and Byrom, A. E. (2009) Density estimation by spatially explicit capture–recapture: likelihood-based methods. In: D. L. Thompson, E. G. Cooch and M. J. Conroy (eds) Modeling Demographic Processes in Marked Populations. Springer. Pp. 255–269.
Efford, M. G., Dawson, D. K. and Borchers, D. L. (2009) Population density estimated from locations of individuals on a passive detector array. Ecology 90, 2676–2682.
Detection functions,
AIC.secr
,
capthist
,
details
,
derived
,
hcov
,
mask
,
list.secr.fit
,
predict.secr
,
print.secr
,
region.N
,
Speed tips
Troubleshooting
userdist
usage
,
vcov.secr
,
verify
,
## Not run: ## construct test data (array of 48 `multi-catch' traps) detectors <- make.grid (nx = 6, ny = 8, detector = "multi") detections <- sim.capthist (detectors, popn = list(D = 10, buffer = 100), detectpar = list(g0 = 0.2, sigma = 25)) ## fit & print null (constant parameter) model secr0 <- secr.fit (detections) secr0 ## uses print method for secr ## compare fit of null model with learned-response model for g0 secrb <- secr.fit (detections, model = g0~b) AIC (secr0, secrb) ## typical result ## model detectfn npar logLik AIC AICc dAIC AICwt ## secr0 D~1 g0~1 sigma~1 halfnormal 3 -347.1210 700.242 700.928 0.000 0.7733 ## secrb D~1 g0~b sigma~1 halfnormal 4 -347.1026 702.205 703.382 2.454 0.2267 ## End(Not run)
## Not run: ## construct test data (array of 48 `multi-catch' traps) detectors <- make.grid (nx = 6, ny = 8, detector = "multi") detections <- sim.capthist (detectors, popn = list(D = 10, buffer = 100), detectpar = list(g0 = 0.2, sigma = 25)) ## fit & print null (constant parameter) model secr0 <- secr.fit (detections) secr0 ## uses print method for secr ## compare fit of null model with learned-response model for g0 secrb <- secr.fit (detections, model = g0~b) AIC (secr0, secrb) ## typical result ## model detectfn npar logLik AIC AICc dAIC AICwt ## secr0 D~1 g0~1 sigma~1 halfnormal 3 -347.1210 700.242 700.928 0.000 0.7733 ## secrb D~1 g0~b sigma~1 halfnormal 4 -347.1026 702.205 703.382 2.454 0.2267 ## End(Not run)
Simple Monte-Carlo goodness-of-fit tests for full-likelihood SECR models. The approach is to calculate a statistic from either the raw data or a fitted model, and to relate this to the distribution of the statistic under the original fitted model. The distribution is estimated by simulating data from the model, and possibly re-fitting the model to each simulated dataset.
The suitability of different test statistics has yet to be assessed.
An alternative and possibly more informative set of tests is provided
in function MCgof
that implements the approach of
Choo et al. (2024) for a range of models.
secr.test(object, nsim = 99, statfn, fit = FALSE, seed = NULL, ncores = NULL, tracelevel = 1, ...)
secr.test(object, nsim = 99, statfn, fit = FALSE, seed = NULL, ncores = NULL, tracelevel = 1, ...)
object |
a fitted secr model |
nsim |
integer number of replicates |
statfn |
function to compute a numeric vector of one or more statistics from a single-session ‘capthist’ object or from a fitted model (see Details) |
fit |
logical; if TRUE the model is re-fitted to each simulated dataset |
seed |
either NULL or an integer that will be used in a call to |
ncores |
integer number of threads for parallel processing |
tracelevel |
see |
... |
other arguments passed to statfn, if needed |
The test statistic(s) may be computed either on a dataset or on a
fitted model, as determined by the argument fit
. The single
argument expected by statfn
should be either a ‘capthist’ object
(fit = FALSE
) or an ‘secr’ object (fit = TRUE
).
The default statistic when fit = FALSE
is the proportion of
individuals observed on only one occasion, which is equivalent to
statfn = function(CH) c(f1 = sum(apply(abs(CH) > 0,1,sum) == 1) /
nrow(CH))
. Repeat detections on one occasion at the same or different
detectors are not counted. The default statistic is therefore not
appropriate for some data, specifically from ‘count’ or ‘polygon’
detectors with few occasions or only one.
The default statistic when fit = TRUE
is the deviance divided by
the residual degrees of freedom (i.e., statfn = function(object)
c(devdf = deviance(object) / df.residual(object))
).
The reported probability (p) is the rank of the observed value in the
vector combining the observed value and simulated values, divided by
(nsim + 1). Ranks are computed with rank
using the default
ties.method = "average"
.
Simulations take account of the usage attribute of detectors in the original capthist object, given that usage was defined and ignoreusage was not set.
Setting ncores = NULL
uses the existing value from the environment variable
RCPP_PARALLEL_NUM_THREADS (see setNumThreads
).
statfn
may return a vector of statistics for each observed or
simulated dataset or model: then the value of each statistic will be
calculated for every simulated dataset, and summarised. If fit =
TRUE
the vector of statistics may include both those computed on the
raw data (object$capthist) and the fitted model.
An object of class c('secrtest','list')
with components:
object |
as input |
nsim |
as input |
statfn |
as input |
fit |
as input |
seed |
as input |
output |
list comprising the simulated values, the observed value(s), and estimated probabilities |
For multi-session input when fit = FALSE
, ‘output’ is a list
in which each session provides one component.
Print and plot methods are provided for ‘secrtest’ objects.
simulate.secr
is always used to simulate the raw data, but
simulate.secr
does not work for all types of fitted
model. Models fitted by maximizing the likelihood conditional on
(
CL = TRUE
in secr.fit
) potentially include
individual covariates whose distribution in the population is
unknown. This precludes simulation, and conditional-likelihood models
in general are therefore not covered by secr.test
.
Other exclusions include exotic non-binary behavioural responses (“bn", “bkn", “bkc", “Bkc" - but these are generally undocumented in any case).
If fit = TRUE
then sim.secr
is used.
At each simulation a new population is generated across the extent of the original mask. If the extent is unduly large then time will be wasted simulating the possibility of detection for many essentially undetectable animals. This is an argument for keeping the mask tight - large enough only to avoid mask-induced bias.
Choo, Y. R., Sutherland, C. and Johnston, A. (2024) A Monte Carlo resampling framework for implementing goodness-of-fit tests in spatial capture-recapture model Methods in Ecology and Evolution DOI: 10.1111/2041-210X.14386.
MCgof
,
print.secrtest
,
plot.secrtest
,
simulate.secr
,
sim.secr
,
deviance.secr
## Not run: secr.test(secrdemo.0, nsim = 99) secr.test(ovenbird.model.1, nsim = 20) ## example combining raw data summary and model fit ## assumes single-session bothfn <- function(object) { CH <- object$capthist f1 <- sum(apply(abs(CH) > 0, 1, sum) == 1) / nrow(CH) devdf <- deviance(object) / df.residual(object) c(f1 = f1, devdf = devdf) } test <- secr.test (secrdemo.0, nsim = 19, statfn = bothfn, fit = TRUE) test plot(test, main = '') ## End(Not run)
## Not run: secr.test(secrdemo.0, nsim = 99) secr.test(ovenbird.model.1, nsim = 20) ## example combining raw data summary and model fit ## assumes single-session bothfn <- function(object) { CH <- object$capthist f1 <- sum(apply(abs(CH) > 0, 1, sum) == 1) / nrow(CH) devdf <- deviance(object) / df.residual(object) c(f1 = f1, devdf = devdf) } test <- secr.test (secrdemo.0, nsim = 19, statfn = bothfn, fit = TRUE) test plot(test, main = '') ## End(Not run)
Demonstration data from program Density are provided as text
files in the ‘extdata’ folder, as raw dataframes (trapXY
,
captXY
), and as a combined capthist
object
(captdata
) ready for input to secr.fit
.
The fitted models are objects of class secr
formed by
secrdemo.0 <- secr.fit (captdata)
secrdemo.b <- secr.fit (captdata, model = list(g0 = ~b))
secrdemo.CL <- secr.fit (captdata, CL = TRUE)
data(secrdemo)
data(secrdemo)
The raw data are 235 fictional captures of 76 animals over 5 occasions in 100 single-catch traps 30 metres apart on a square grid with origin at (365,365).
Dataframe trapXY
contains the data from the Density input file
‘trap.txt’, and captXY
contains the data from ‘capt.txt’ (Efford
2012).
The fitted models use a halfnormal detection function and the likelihood for multi-catch traps (expect estimates of g0 to be biased because of trap saturation Efford et al. 2009). The first is a null model (i.e. parameters constant) and the second fits a learned trap response.
Object | Description |
captXY | data.frame of capture data |
trapXY | data.frame of trap locations |
captdata | capthist object |
secrdemo.0 | fitted secr model -- null |
secrdemo.b | fitted secr model -- g0 trap response |
secrdemo.CL | fitted secr model -- null, conditional likelihood |
Efford, M. G. (2012) DENSITY 5.0: software for spatially explicit capture–recapture. Department of Mathematics and Statistics, University of Otago, Dunedin, New Zealand. https://www.otago.ac.nz/density/.
Efford, M. G., Borchers D. L. and Byrom, A. E. (2009) Density estimation by spatially explicit capture-recapture: likelihood-based methods. In: D. L. Thomson, E. G. Cooch and M. J. Conroy (eds) Modeling Demographic Processes in Marked Populations. Springer, New York. Pp. 255–269.
## Not run: ## navigate to folder with raw data files olddir <- setwd (system.file("extdata", package="secr")) ## construct capthist object from raw data captdata <- read.capthist ("capt.txt", "trap.txt", fmt = "XY", detector = "single") ## generate demonstration fits secrdemo.0 <- secr.fit (captdata) secrdemo.CL <- secr.fit (captdata, CL = TRUE) secrdemo.b <- secr.fit (captdata, model = list(g0 ~ b)) ## restore previous setting setwd(olddir) ## End(Not run) ## display the null model fit, using the print method for secr secrdemo.0 ## compare fit of models AIC(secrdemo.0, secrdemo.b) ## display estimates for the two models (single session) collate(secrdemo.0, secrdemo.b)[1,,,]
## Not run: ## navigate to folder with raw data files olddir <- setwd (system.file("extdata", package="secr")) ## construct capthist object from raw data captdata <- read.capthist ("capt.txt", "trap.txt", fmt = "XY", detector = "single") ## generate demonstration fits secrdemo.0 <- secr.fit (captdata) secrdemo.CL <- secr.fit (captdata, CL = TRUE) secrdemo.b <- secr.fit (captdata, model = list(g0 ~ b)) ## restore previous setting setwd(olddir) ## End(Not run) ## display the null model fit, using the print method for secr secrdemo.0 ## compare fit of models AIC(secrdemo.0, secrdemo.b) ## display estimates for the two models (single session) collate(secrdemo.0, secrdemo.b)[1,,,]
The use of random number seeds in secr is explained.
R provides several kinds of random number generator (RNG) in the base package (see RNG). These are used both explicitly, in functions such as runif
and rnorm
, and implicitly (sample
).
A seed suitable for any kind of RNG is held in a vector of 626 integers named .Random.seed
. The vector is not to be modified directly by users. Instead, to start a reproducible stream of random numbers, the user calls set.seed
with a single non-null integer argument. This has the effect of initialising .Random.seed
. The value of .Random.seed
may nevertheless be stored and restored to reset the RNG state.
set.seed
with a NULL argument initialises .Random.seed
to an indeterminate (time- and process-dependent) value. The same happens if a random number function is called before .Random.seed
has been set.
The ‘official’ approach to setting and storing the RNG seed is shown in code and documentation for the generic function simulate
in the stats package.
The generic has argument ‘seed’ with default NULL.
If ‘seed’ is non-null then set.seed
is called.
The returned value has an attribute “seed” whose value is either (i) if specified, the integer value of the ‘seed’ argument (with its own attribute “kind” from RNGkind), or (ii) the original vector .Random.seed
.
On exit the RNG state in .Random.seed
is reset to the value that applied when the function was called.
For NULL seed input, the saved RNGstate may be used to reset .Random.seed (see Examples).
Many functions in secr call on random numbers, sometimes in unexpected places. For example autoini
selects a random sample to thin points and speed computation. In most functions there is no provision for direct control of the random number state: users won't usually care, and if they do then set.seed
may be called for the particular R session.
However, control of the RNG seed is required for reproducible data generation in simulation functions. These functions typically have a ‘seed’ argument that is used internally in a call to set.seed
. Handling of seeds in the simulation functions of secr largely follows stats::simulate
as described in the preceding section.
The relevant functions are –
Function | Default | Saved attribute | Note |
randomHabitat |
NULL | seed or RNGstate | |
secr.test |
NULL | seed or RNGstate | calls and retains seed from simulate.secr |
sim.capthist |
NULL | seed or RNGstate | |
sim.resight |
NULL | seed or RNGstate | Seed may be passed in ... argument |
sim.popn |
NULL | seed or RNGstate | |
sim.secr |
NULL | seed or RNGstate | |
simulate.secr |
NULL | seed or RNGstate | S3 method called by sim.secr
|
Setting seed = NULL
in any of these functions has the effect of continuing the existing random number stream; it is not the same as calling set.seed(NULL)
.
Two models are used for parallel processing in secr, corresponding to multi-threading with package RcppParallel (e.g. secr.fit
) and parallel cores in package parallel (e.g. chat.nk
).
In the parallel model the L'Ecuyer pseudorandom generator is used to provide a separate random number stream for each core (see clusterSetRNGStream
).
When using Rcpp the state of the random number generator is set in C++ with the call
RNGScope scope;
that automatically resets the state of the generator on exit (Eddelbuettel 2013 p. 115).
Random number streams in separate RcppParallel threads are (probably) not independent. Thus there are potential issues with RNG calls in multi-threaded code. However, in secr 4.0 all RNG calls in C++ code are outside multi-threaded contexts, with the exception of simulations allowing for overdispersion in mark–resight estimates (Rcpp exported function sightingchatcpp). The implications for mark-resight estimates have not been explored, and it is unclear whether more elaborate solutions are needed.
Eddelbuettel, D. 2013. Seamless R and C++ integration with Rcpp. Springer.
Parallel
,
set.seed
,
simulate
,
sim.capthist
,
sim.popn
,
sim.resight
,
secr.test
,
simulate.secr
## Not run: lmfit <- lm(speed ~ dist, data = cars) ## 1. NULL seed r1 <- simulate(lmfit, seed = NULL) r2 <- simulate(lmfit, seed = NULL) ## restore RNGstate, assuming RNGkind unchanged .Random.seed <- attr(r1, "seed") r3 <- simulate(lmfit, seed = NULL) r1[1:6,1] r2[1:6,1] r3[1:6,1] ## 2. explicit seed r4 <- simulate(lmfit, seed = 123) r5 <- simulate(lmfit, seed = attr(r4, "seed")) r4[1:6,1] r5[1:6,1] ## End(Not run)
## Not run: lmfit <- lm(speed ~ dist, data = cars) ## 1. NULL seed r1 <- simulate(lmfit, seed = NULL) r2 <- simulate(lmfit, seed = NULL) ## restore RNGstate, assuming RNGkind unchanged .Random.seed <- attr(r1, "seed") r3 <- simulate(lmfit, seed = NULL) r1[1:6,1] r2[1:6,1] r3[1:6,1] ## 2. explicit seed r4 <- simulate(lmfit, seed = 123) r5 <- simulate(lmfit, seed = attr(r4, "seed")) r4[1:6,1] r5[1:6,1] ## End(Not run)
S3 class for results from secr.test
.
## S3 method for class 'secrtest' print(x, terse = TRUE, ...) ## S3 method for class 'secrtest' plot(x, stat, ...)
## S3 method for class 'secrtest' print(x, terse = TRUE, ...) ## S3 method for class 'secrtest' plot(x, stat, ...)
x |
secrtest object from |
terse |
logical; if TRUE only p values are displayed |
stat |
character; names of statistics to plot (default: all) |
... |
other arguments passed to hist by |
An ‘secrtest’ object is output from secr.test
.
plot.secrtest
plots a histogram of the simulated values.
If plot.secrtest
is applied to an object with more than one
statistic then multiple plots are produced, so a multi-figure layout
should be prepared (par(mfrow = c(1,2))
for 2 plots side by
side). Include the hist
argument main = ''
to suppress the
ugly plot labels, and ensure each statistic is named by statfn
so
that the x-axis is labelled correctly (See the Examples in help for
secr.test
).
## Not run: tmp <- secr.test(ovenbird.model.1) if (inherits(tmp, 'secrtest')) { tmp ## terse print print(tmp, terse = FALSE) par(mfrow = c(1,5)) plot(tmp, main = '', xlim=c(0,1), breaks=seq(0,1,0.05)) par(mfrow = c(1,1)) ## reset to default } ## End(Not run)
## Not run: tmp <- secr.test(ovenbird.model.1) if (inherits(tmp, 'secrtest')) { tmp ## terse print print(tmp, terse = FALSE) par(mfrow = c(1,5)) plot(tmp, main = '', xlim=c(0,1), breaks=seq(0,1,0.05)) par(mfrow = c(1,1)) ## reset to default } ## End(Not run)
Extract or replace the session names of a capthist
object.
session(object, ...) session(object) <- value
session(object, ...) session(object) <- value
object |
object with ‘session’ attribute e.g. |
value |
character vector or vector that may be coerced to character, one value per session |
... |
other arguments (not used) |
Replacement values will be coerced to character.
a character vector with one value for each session in capthist
Like Density, secr uses the term ‘session’ for a closed-population sample. A session usually includes data from several closely-spaced capture occasions (often consecutive days). Each 'primary session' in the ‘robust’ design of Pollock (1982) would be treated as a session in secr. secr also uses ‘session’ for independent subsets of the capture data distinguished by characteristics other than sampling time (as above). For example, two grids trapped simultaneously could be analysed as distinct sessions if (i) they were far enough apart that there was negligible prospect of the same animal being caught on both grids, and (ii) there was interest in comparing estimates from the two grids, or fitting a common detection model.
The log likelihood for a session model is the sum of the separate session log likelihoods. Although this assumes independence of sampling, parameters may be shared across sessions, or session-specific parameter values may be functions of session-level covariates. For many purposes, ‘sessions’ are equivalent to ‘groups’. For multi-session models the detector array and mask are specified separately for each session. Group models are therefore generally simpler to implement. On the other hand, sessions offer more flexibility in defining and evaluating between-session models, including trend models.
Pollock, K. H. (1982) A capture-recapture design robust to unequal probability of capture. Journal of Wildlife Management 46, 752–757.
session(captdata)
session(captdata)
Set or report the number of cores to be used for multi-threaded operations. A wrapper for the RcppParallel function setThreadOptions (Allaire et al. 2019).
setNumThreads(ncores, ...)
setNumThreads(ncores, ...)
ncores |
integer number of threads to use |
... |
other arguments passed to |
If ncores
is NULL then the current value of the environment variable RCPP_PARALLEL_NUM_THREADS is used. RCPP_PARALLEL_NUM_THREADS defaults to 2 at the start of a session (assuming at least 2 logical cores available).
Calling setNumThreads()
with no arguments is a handy way to check how many threads are in use.
The value of RCPP_PARALLEL_NUM_THREADS is also reset when a multi-threaded function such as secr.fit
is called with a non-NULL value of the ncores argument. This value applies in later calls of secr.fit
with ncores = NULL
until changed.
The new value of the environment variable RCPP_PARALLEL_NUM_THREADS.
The mechanism for setting the number of threads changed between versions 4.1.0 and 4.2.0. The default number of cores is now capped at 2 to meet CRAN requirements. Setting ncores = NULL
previously specified one less than the number of available cores.
Allaire, J. J., Francois, R., Ushey, K., Vandenbrouck, G., Geelnard, M. and Intel (2019) RcppParallel: Parallel Programming Tools for 'Rcpp'. R package version 4.4.4. https://CRAN.R-project.org/package=RcppParallel.
Parallel,
setThreadOptions
Sys.getenv
# determine current number of threads setNumThreads() ## Not run: # set new number of threads setNumThreads(7) # a call to secr.fit that specifies 'ncores' also sets the # number of threads, as we see here fit <- secr.fit(captdata, trace = FALSE, ncores = 8) setNumThreads() ## End(Not run)
# determine current number of threads setNumThreads() ## Not run: # set new number of threads setNumThreads(7) # a call to secr.fit that specifies 'ncores' also sets the # number of threads, as we see here fit <- secr.fit(captdata, trace = FALSE, ncores = 8) setNumThreads() ## End(Not run)
Extract or replace the markocc
attribute of a traps
object that distinguishes marking occasions from sighting
occasions. Also, extract or replace the attributes Tu, Tm and Tn of a capthist
object, used for storing counts of sightings. All attributes are
optional, but Tu, Tm and Tn require markocc to be specified.
markocc(object, ...) markocc(object) <- value sighting(object) Tu(object, ...) Tu(object) <- value Tm(object, ...) Tm(object) <- value Tn(object, ...) Tn(object) <- value
markocc(object, ...) markocc(object) <- value sighting(object) Tu(object, ...) Tu(object) <- value Tm(object, ...) Tm(object) <- value Tn(object, ...) Tn(object) <- value
object |
|
value |
numeric matrix of detectors x occasions, or a vector (see Details) |
... |
other arguments (not used) |
For replacement of markocc, ‘value’ should be a vector of integers indicating the occasions on which animals are sighted only (0) or marked or recaptured (1).
For replacement of Tu
, Tm
or Tn
, ‘value’ may be a scalar
(total count) or a detectors x occasions matrix.
markocc(object) returns the markocc vector of the traps
object. markocc(object)
may be NULL.
Tu
, Tm
and Tn
return the respective attributes of a capthist object, or
NULL if they are unspecified.
sighting(object) returns TRUE if the markocc attribute indicates at least one sighting-only occasion.
traps
,
addSightings
,
sightingPlot
,
secr-markresight.pdf
Extract or replace signal attributes of a ‘capthist’ object.
signalframe(object) signalframe(object) <- value ## S3 method for class 'capthist' signal(object, ...) ## S3 method for class 'capthist' noise(object, ...) signal(object) <- value noise(object) <- value
signalframe(object) signalframe(object) <- value ## S3 method for class 'capthist' signal(object, ...) ## S3 method for class 'capthist' noise(object, ...) signal(object) <- value noise(object) <- value
object |
a ‘capthist’ object |
value |
replacement value (see Details) |
... |
other arguments (not used) |
Signal attributes of a ‘capthist’ object are stored in a dataframe called the signalframe. This has one row per detection. The signalframe includes the primary field ‘signal’ and an unlimited number of other fields. To extract the signal field alone use the signal method.
These functions extract data on detections, ignoring occasions when an animal was not detected. Detections are ordered by occasion, animalID and trap.
Replacement values must precisely match object
in number of
detections and in their order.
For signalframe
, a dataframe containing signal data and
covariates, one row per detection. The data frame has one row per
detection. See signalmatrix
for a matrix with one row per
cue and columns for different microphones.
For signal
and noise
, a numeric vector with one element per detection.
If object
has multiple sessions, the result is a list with one
component per session.
## ovensong dataset has very simple signalframe head(signalframe(signalCH))
## ovensong dataset has very simple signalframe head(signalframe(signalCH))
Produce sound x microphone matrix, possibly with sound covariates as extra columns.
signalmatrix(object, noise = FALSE, recodezero = FALSE, prefix = "Ch", signalcovariates = NULL, names = NULL)
signalmatrix(object, noise = FALSE, recodezero = FALSE, prefix = "Ch", signalcovariates = NULL, names = NULL)
object |
object inheriting from secr class ‘capthist’ |
noise |
logical; if TRUE, noise is extracted instead of signal |
recodezero |
logical; if TRUE zero signals are set to NA |
prefix |
character value used to form channel names |
signalcovariates |
character vector of covariate names from signalframe to add as columns |
names |
character vector of column names |
This function extracts signal or noise data from a capthist object, where it is stored in the ‘signalframe’ attribute, as a more natural sound x microphone table. There is no equivalent replacement function.
The signalcovariates
argument may be used to specify additional
columns of the signal frame to collapse and add as columns to the right
of the actual signal data. Ordinarily there will be multiple rows in
signalframe for each row in the output; the covariate value is taken
from the first matching row.
If names
is not provided, column names are constructed from the
detector names. If the length of names
is less than the number of
columns, simple numerical names are constructed.
A dataframe with dim = c(n,K+j) where n is the number of separate sounds, K is the number of microphones, and j is the number of covariates (by default j = 0).
## use 'secr' ovenbird data signalmatrix(signalCH)
## use 'secr' ovenbird data signalmatrix(signalCH)
Create a set of capture or marking-and-resighting histories by simulated sampling of a 2-D population using an array of detectors.
sim.capthist(traps, popn = list(D = 5, buffer = 100, Ndist = "poisson"), detectfn = 0, detectpar = list(), noccasions = 5, nsessions = 1, binomN = NULL, exactN = NULL, p.available = 1, renumber = TRUE, seed = NULL, maxperpoly = 100, chulltol = 0.001, userdist = NULL, savepopn = FALSE) sim.resight(traps, popn = list(D = 5, buffer = 100, Ndist = "poisson"), ..., pID = 1, unmarked = TRUE, nonID = TRUE, unresolved = FALSE, unsighted = TRUE, pmark = 0.5, Nmark = NULL, markingmask = NULL)
sim.capthist(traps, popn = list(D = 5, buffer = 100, Ndist = "poisson"), detectfn = 0, detectpar = list(), noccasions = 5, nsessions = 1, binomN = NULL, exactN = NULL, p.available = 1, renumber = TRUE, seed = NULL, maxperpoly = 100, chulltol = 0.001, userdist = NULL, savepopn = FALSE) sim.resight(traps, popn = list(D = 5, buffer = 100, Ndist = "poisson"), ..., pID = 1, unmarked = TRUE, nonID = TRUE, unresolved = FALSE, unsighted = TRUE, pmark = 0.5, Nmark = NULL, markingmask = NULL)
traps |
|
popn |
locations of individuals in the population to be sampled, either as
a |
detectfn |
integer code or character string for shape of detection function 0 = halfnormal etc. – see detectfn |
detectpar |
list of values for named parameters of detection function |
noccasions |
number of occasions to simulate |
nsessions |
number of sessions to simulate |
binomN |
integer code for distribution of counts (see Details) |
exactN |
integer number of telemetry fixes per occasion |
p.available |
vector of one or two probabilities (see Details) |
renumber |
logical for whether output rows should labeled sequentially (TRUE) or retain the numbering of the population from which they were drawn (FALSE) |
seed |
either NULL or an integer that will be used in a call to |
maxperpoly |
integer maximum number of detections of an individual in one polygon or transect on any occasion |
chulltol |
numeric buffer (m) for polygon around telemetry locations |
userdist |
user-defined distance function or matrix (see details) |
savepopn |
logical; if TRUE then the popn (input or simulated) is saved as an attribute |
... |
arguments to pass to |
pID |
probability of individual identification for marked animals |
unmarked |
logical, if TRUE unmarked individuals are not recorded during ‘sighting’ |
nonID |
logical, if TRUE then unidentified marked individuals are not recorded during ‘sighting’ |
unresolved |
logical, if TRUE then individuals of unresolved mark status are not recorded during ‘sighting’ |
unsighted |
logical, if TRUE and sighting only then capthist includes all-zero histories |
pmark |
numeric probability that an individual is ‘pre-marked’ (see Details) |
Nmark |
number of individuals to be ‘pre-marked’ (see Details) |
markingmask |
|
If popn
is not of class ‘popn’ then a homogeneous Poisson
population with the desired density (animals/ha) is first simulated over
the rectangular area of the bounding box of traps
plus a buffer
of the requested width (metres). The detection algorithm depends on the
detector type of traps
. For ‘proximity’ detectors, the actual
detection probability of animal i at detector j is the
naive probability given by the detection function. For ‘single’ and
‘multi’ detectors the naive probability is modified by competition
between detectors and, in the case of ‘single’ detectors, between animals. See
Efford (2004) and other papers below for details.
Detection parameters in detectpar
are specific to the detection
function, which is indicated by detectfn
.
Parameters may vary with time - for this provide a vector of length
noccasions
. The g0 parameter may vary both by time and detector
- for this provide a matrix with noccasions
rows and as many
columns as there are detectors. The default detection parameters are
list(g0 = 0.2, sigma = 25, z = 1)
.
The default is to simulate a single session. This may be overridden by
providing a list of populations to sample (argument popn
) or by
specifying nsessions
> 1 (if both then the number of sessions must
match). Using nsessions
> 1 results in replicate samples of
populations with the same density etc. as specified directly in the
popn
argument.
binomN
determines the statistical distribution of the number of
detections of an individual at a particular ‘count’ detector or polygon
on a particular occasion. A Poisson distribution is indicated by
binomN = 0
; see secr.fit
for more. The distribution
is always Bernoulli (binary) for ‘proximity’ and ‘signal’ detectors.
If exactN
is not specified or zero then the number of telemetry
fixes is a random variable determined by the other detection settings.
p.available
specifies temporary non-availability for detection in
multi-session simulations. If a single probability is specified then
temporary non-availability is random (independent from session to
session). If two probabilities are given then non-availability is
Markovian (dependent on previous state) and the two values are for
animals available and not available at the preceding session. In the
Markovian case, availability in the first session is assigned at random
according to the equilibrium probability p2 / (1 - p1 + p2). Incomplete
availability is not implemented for sampling lists of populations.
detectpar
may include a component ‘truncate’ for the distance
beyond which detection probability is set to zero. By default this value
is NULL (no specific limit).
detectpar
may also include a component ‘recapfactor’ for a
general learned trap response. For ‘single’ and ‘multi’ detector types
the probability of detection changes by this factor for all occasions
after the occasion of first capture. Attempted use with other detector
types causes an error. If recapfactor x g(d) > 1.0, g(d) is truncated at
1.0. Other types of response (site-specific bk, Markovian B) are not
allowed.
If popn
is specified by an object of class ‘popn’ then any
individual covariates will be passed on; the covariates
attribute
of the output is otherwise set to NULL.
The random number seed is managed as in simulate
.
chulltol
is used only when simulating telemetry locations. By
default, a new 'traps' polygon is generated as the convex hull of the
simulated locations, with a slight (1 mm) added buffer to ensure
boundary points are within the polygon. Buffering is suppressed if
chulltol
is NA or negative.
userdist
cannot be set if ‘traps’ is any of polygon, polygonX,
transect or transectX.
sim.resight
generates mark-resight data. The ‘markocc’ attribute
of ‘traps’ indicates the occasions which are for sighting-only (0) or
marking and recapture (1). The number
of occasions is determined by markocc
. sim.capthist
is first
called with the arguments ‘traps’ and .... The detector type of ‘traps’
should be ‘proximity’ or ‘count’ for sighting occasions (markocc = 0). The detector type need not be the same for marking and sighting occasions ('multi' is allowed on marking occasions). If ... includes a non-null ‘seed’ the
random seed is reset in sim.resight
and not passed to
sim.capthist
.
A special case arises when all occasions are sighting-only. Then it is assumed that individuals in the population are marked prior to the start of sampling with a known spatial distribution (i.e. marking does not follow a spatial detection model). By default, animals throughout the buffered area are pre-marked with probability pmark
. If Nmark
is specified then a sample of size Nmark
will be selected for marking, overriding pmark
.
The marked population may be restricted to a subset of the space spanned by popn
by specifying markingmask
, which may have a further covariate ‘marking’ to vary the intensity of marking.
For sim.capthist
, an object of class capthist
, a 3-dimensional array
with additional attributes. Rows represent
individuals and columns represent occasions; the third dimension, codes the number of
detections at each detector (zero or one for trap detectors (‘single’, ‘multi’) and binary proximity detectors.
The initial state of the R random number generator is stored in the ‘seed’ attribute.
For sim.resight
, an object of class capthist
for which the traps object has a markocc attribute (marking occasions), and there are further attributes Tu (sightings of unmarked animals) and Tm (sightings of marked but not
identified animals).
External code is called to speed the simulations. The present version
assumes a null model, i.e., naive detection probability is constant
except for effects of distance and possibly time (using vector-valued
detection parameters from 1.2.10). You can, however, use
rbind.capthist
to combine detections of population
subclasses (e.g. males and females) simulated with different parameter
values. This is not valid for detector type "single" because it fails to
allow for competition for traps between subclasses. Future versions may
allow more complex models.
truncate
has no effect (i) when using a uniform detection
function with radius (sigma
) <= truncate
and (ii) with
signal strength detection (detectfn 10, 11). Note that truncated
detection functions are provided for de novo simulation, but are not
available when fitting models with in secr.fit
or simulating from
a fitted model with sim.secr
.
maxperpoly
limits the size of the array allocated for
detections in C code; an error results if the is number is exceeded.
Prior to 2.10.0 sim.resight
interpreted length-2 vectors of detection parameters as referring to marking and sighting occasions; this feature has been discontinued.
Borchers, D. L. and Efford, M. G. (2008) Spatially explicit maximum likelihood methods for capture–recapture studies. Biometrics 64, 377–385.
Efford, M. G. (2004) Density estimation in live-trapping studies. Oikos 106, 598–610.
Efford, M. G., Borchers D. L. and Byrom, A. E. (2009) Density estimation by spatially explicit capture-recapture: likelihood-based methods. In: D. L. Thomson, E. G. Cooch and M. J. Conroy (eds) Modeling Demographic Processes in Marked Populations. Springer, New York. Pp. 255–269.
Efford, M. G., Dawson, D. K. and Borchers, D. L. (2009) Population density estimated from locations of individuals on a passive detector array. Ecology 90, 2676–2682.
sim.popn
, capthist
, traps
,
popn
, Detection functions
, simulate
,
Tu
, Tm
## simple example ## detector = "multi" (default) temptrap <- make.grid(nx = 6, ny = 6, spacing = 20) sim.capthist (temptrap, detectpar = list(g0 = 0.2, sigma = 20)) ## with detector = "proximity", there may be more than one ## detection per individual per occasion temptrap <- make.grid(nx = 6, ny = 6, spacing = 20, detector = "proximity") summary(sim.capthist (temptrap, detectpar = list(g0 = 0.2, sigma = 20))) ## marking on occasions 1, 3 only temptrap <- make.grid(nx = 6, ny = 6, spacing = 20, detector = 'proximity') markocc(temptrap) <- c(1,0,1,0,0) CH <- sim.resight (temptrap, detectpar = list(g0 = 0.2, sigma = 20)) summary(CH) ## multiple sessions grid4 <- make.grid(nx = 2, ny = 2) temp <- sim.capthist (grid4, popn = list(D = 1), nsessions = 20) summary(temp, terse = TRUE) ## unmarked or presence types # grid <- make.grid(nx = 10, ny = 10, detector = "unmarked") # CH <- sim.capthist (grid, noccasions = 5) # CH ## "presence" and "unmarked" data are stored as "count" data ## behaviour is controlled by detector type, e.g. # detector(traps(CH)) <- "presence" # CH
## simple example ## detector = "multi" (default) temptrap <- make.grid(nx = 6, ny = 6, spacing = 20) sim.capthist (temptrap, detectpar = list(g0 = 0.2, sigma = 20)) ## with detector = "proximity", there may be more than one ## detection per individual per occasion temptrap <- make.grid(nx = 6, ny = 6, spacing = 20, detector = "proximity") summary(sim.capthist (temptrap, detectpar = list(g0 = 0.2, sigma = 20))) ## marking on occasions 1, 3 only temptrap <- make.grid(nx = 6, ny = 6, spacing = 20, detector = 'proximity') markocc(temptrap) <- c(1,0,1,0,0) CH <- sim.resight (temptrap, detectpar = list(g0 = 0.2, sigma = 20)) summary(CH) ## multiple sessions grid4 <- make.grid(nx = 2, ny = 2) temp <- sim.capthist (grid4, popn = list(D = 1), nsessions = 20) summary(temp, terse = TRUE) ## unmarked or presence types # grid <- make.grid(nx = 10, ny = 10, detector = "unmarked") # CH <- sim.capthist (grid, noccasions = 5) # CH ## "presence" and "unmarked" data are stored as "count" data ## behaviour is controlled by detector type, e.g. # detector(traps(CH)) <- "presence" # CH
Simulate a point process representing the locations of individual animals.
sim.popn (D, core, buffer = 100, model2D = c("poisson", "cluster", "IHP", "coastal", "hills", "linear", "even", "rLGCP", "rThomas"), buffertype = c("rect", "concave", "convex"), poly = NULL, covariates = list(sex = c(M = 0.5, F = 0.5)), number.from = 1, Ndist = c("poisson", "fixed", "specified"), nsessions = 1, details = NULL, seed = NULL, keep.mask = model2D %in% c("IHP", "linear"), Nbuffer = NULL, age = FALSE, ...) tile(popn, method = "reflect")
sim.popn (D, core, buffer = 100, model2D = c("poisson", "cluster", "IHP", "coastal", "hills", "linear", "even", "rLGCP", "rThomas"), buffertype = c("rect", "concave", "convex"), poly = NULL, covariates = list(sex = c(M = 0.5, F = 0.5)), number.from = 1, Ndist = c("poisson", "fixed", "specified"), nsessions = 1, details = NULL, seed = NULL, keep.mask = model2D %in% c("IHP", "linear"), Nbuffer = NULL, age = FALSE, ...) tile(popn, method = "reflect")
D |
density animals / hectare (10 000 m^2) (see Details for IHP case) |
core |
data frame of points defining the core area |
buffer |
buffer radius about core area |
model2D |
character string for 2-D distribution |
buffertype |
character string for buffer type |
poly |
bounding polygon (see Details) |
covariates |
list of named covariates or function to generate covariates |
number.from |
integer ID for animal |
Ndist |
character string for distribution of number of individuals |
nsessions |
number of sessions to simulate |
details |
optional list with additional parameters |
seed |
either NULL or an integer that will be used in a call to |
keep.mask |
logical; if TRUE and model2D %in% c('IHP','linear')
then |
Nbuffer |
numeric number of individuals to simulate (possibly non-integer) |
age |
logical; if TRUE then age covariate added for multisession popn with turnover |
... |
arguments passed to subset if poly is not NULL |
popn |
popn object |
method |
character string "reflect" or "copy" |
core
must contain columns ‘x’ and ‘y’; a traps
object is
suitable. For buffertype = "rect"
, animals are simulated in the
rectangular area obtained by extending the bounding box of core
by buffer
metres to top and bottom, left and right. This box has
area . If
model2D = 'poisson'
the buffer type may also be ‘convex’ (points within a buffered convex polygon) or ‘concave’ (corresponding to a mask of type ‘trapbuffer’); these buffer types use bufferContour
.
Covariates may be specified in either of two ways. In the first, each element of covariates
defines a categorical (factor) covariate with the given probabilities of membership in each class. In the second, the 'covariates' argument is a function (or a character value naming a function) that takes a dataframe of x and y coordinates as its sole argument; the function should return a dataframe with the same number of rows that will be used as the covariates attribute (secr >= 4.6.7).
A notional random covariate ‘sex’ is generated by default.
Ndist should usually be ‘poisson’ or ‘fixed’. The number of individuals has
expected value
. If
is non-integer then Ndist = "fixed"
results in
, with probabilities set to yield
individuals on average. The option ‘specified’ is undocumented;
it is used in some open-population simulations.
If model2D = "cluster"
then the simulated population approximates a Neyman-Scott
clustered Poisson distribution. Ancillary parameters are passed as
components of details
: details$mu is the expected number of
individuals per cluster and details$hsigma is the spatial scale
() of a 2-D kernel for location within each cluster.
The algorithm is
Determine the number of clusters (parents) as a random Poisson variate
with
Locate each parent by drawing uniform random x- and y-coordinates
Determine number of offspring for each parent by drawing from a Poisson distribution with mean mu
Locate offspring by adding random normal error to each parent coordinate
Apply toroidal wrapping to ensure all offspring locations are inside the buffered area
A special cluster option is selected if details$clone = "constant": then each parent is cloned exactly details$mu times.
Toroidal wrapping is a compromise. The result is more faithful to the Neyman-Scott distribution if the buffer is large enough that only a small proportion of the points are wrapped.
If model2D = "IHP"
then an inhomogeneous Poisson distribution is
simulated. core
should be a habitat mask and D
should be one of –
a vector of length equal to the number of cells (rows)
in core
,
the name of a covariate in core
that contains
cell-specific densities (animals / hectare),
a function to generate the intensity of the distribution at each mask point, or
a constant.
If a function, D
should take two arguments, a habitat mask and a list of parameter values ('core' and 'details' are passed internally as these arguments).
The number
of individuals in each cell is either (i) Poisson-distributed with mean
where
is the cell area (an attribute of the mask)
(
Ndist = "poisson"
) or (ii) multinomial with size and
relative cell probabilities given by D (
Ndist =
"fixed"
). buffertype
and buffer
are ignored, as the
extent of the population is governed entirely by the mask in
core
.
If model2D = "linear"
then a linear population is simulated as
for model2D = "IHP"
, except that core
should be a
linearmask object from package secrlinear, and density (D) is
expressed in animals per km. The documentation of secrlinear
should be consulted for further detail (e.g. the wrapper function
sim.linearpopn
).
If model2D = "coastal"
then a form of inhomogeneous Poisson
distribution is simulated in which the x- and y-coordinates are drawn from
independent Beta distributions. Default parameters generate the
‘coastal’ distribution used by Fewster and Buckland (2004) for
simulations of line-transect distance sampling (x ~ Beta(1, 1.5), y ~
Beta(5, 1), which places 50% of the population in the ‘northern’ 13%
of the rectangle). The four Beta parameters may be supplied in the
vector component Beta of the ‘details’ list (see Examples). The Beta
parameters (1,1) give a uniform distribution. Coordinates are scaled to
fit the limits of a sampled rectangle, so this method assumes buffertype
= "rect".
If model2D = "hills"
then a form of inhomogeneous Poisson
distribution is simulated in which intensity is a sine curve in the x-
and y- directions (density varies symmetrically between 0 and 2 x D
along each axis). The number of hills in each direction (default 1) is
determined by the ‘hills’ component of the ‘details’ list (e.g. details
= list(hills=c(2,3)) for 6 hills). If either number is negative then
alternate rows will be offset by half a hill. Displacements of the
entire pattern to the right and top are indicated by further elements of
the ‘hills’ component (e.g. details = list(hills=c(1,1,0.5,0.5)) for 1
hill shifted half a unit to the top right; coordinates are wrapped, so
the effect is to split the hill into the four corners). Negative
displacements are replaced by runif(1). Density is zero at the edge when
the displacement vector is (0,0) and rows are not offset.
If model2D = "even"
then the buffered area is divided into square cells with side sqrt(10000/D) and one animal is located at a random uniform location within each cell. If the height or width is not an exact multiple of the cell side then one whole extra row or column of cells is added; animals located at random in these cells are discarded if they fall outside the original area.
From secr 4.6.2, sim.popn
provides an interface to two simulation functions from spatstat (Baddeley et al. 2015): rLGCP
and rThomas
.
If model2D = "rLGCP"
then a log-gaussian Cox process is simulated within the buffered area. Function rLGCP
in spatstat calls functions from RandomFields (Schlather et al. 2015; see Notes). Certain options are fixed: the correlation function is RMexp from RandomFields, and there is no provision for covariate effects. Clipping to a polygon (poly) and fixed-N (Ndist = "fixed") are not supported. The algorithm first constructs the log spatial intensity as a realisation of a Gaussian random field; one realisation of an IHP with that intensity is then simulated.
The parameters for model2D = "rLGCP"
are the scalar density (D) and the variance and spatial scale of the random field (passed as details arguments ‘var’ and ‘scale’). The variance is on the log scale; the mean on the log scale is computed internally as mu = log(D) - var/2. var = 0 results in a random uniform (Poisson) distribution. When details$saveLambda = TRUE, the discretized intensity function is saved as the attribute "Lambda", a habitat mask with covariate "Lambda" that may be used to construct further IHP realisations (see Examples).
If model2D = "rThomas"
then a Thomas process is simulated. This is a special case of the Neyman-Scott process in which each parent gives rise to a Poisson number of offspring (see Notes). The expected number of offspring per parent and the spatial scatter about each parent are specified by the details arguments ‘mu’ and ‘scale’. Argument ‘kappa’ of rThomas
(density of parent process) is computed as D/mu/1e4. Other arguments remain at their defaults, including ‘expand’ (4 * scale). A dataframe of parent locations is saved in attribute ‘parents’. The intensity surface for each realisation is saved in attribute 'Lambda' when details$saveLambda = TRUE.
If poly
is specified, points outside poly
are
dropped. poly
may be one of the types descrbed in
boundarytoSF
.
The subset
method is called internally when poly
is used;
the ... argument may be used to pass values for keep.poly
and
poly.habitat
.
Multi-session populations may be generated with nsessions > 1
.
Multi-session populations may be independent or generated by per capita
turnover from a starting population. In the ‘independent’ case
(details$lambda
not specified) D or Nbuffer may be a vector of length equal to
nsessions
. Turnover is controlled by survival, growth rate and movement
parameters provided as components of details
and described in turnover.
The optional covariate 'age' is the number of sessions from the session of recruitment.
The random number seed is managed as in simulate.lm
.
Function tile
replicates a popn pattern by either reflecting or
copying and translating it to fill a 3 x 3 grid.
An object of class c("popn", "data.frame")
a data frame with columns ‘x’ and ‘y’. Rows correspond to individuals. Individual covariates (optional) are stored
as a data frame attribute. The initial state of the R random number generator is
stored in the ‘seed’ attribute.
If model2D = "linear"
the output is of class c("linearpopn",
"popn", "data.frame")
.
If model2D = "IHP"
or model2D = "linear"
the value of
core
is stored in the ‘mask’ attribute.
Package RandomFields is not currently on CRAN. It may be installed with this code:
install.packages("RandomFields", repos = c("https://spatstat.r-universe.dev",
"https://cloud.r-project.org"))
model2D = "rThomas"
and model2D = "cluster"
(the builtin Neyman-Scott implementation) are equivalent. There may be some subtle differences. The spatstat implementation is usually to be preferred.
Baddeley, A., Rubak, E., and Turner, R. 2015. Spatial Point Patterns: Methodology and Applications with R. Chapman and Hall/CRC Press, London. ISBN 9781482210200, https://www.routledge.com/Spatial-Point-Patterns-Methodology-and-Applications-with-R/Baddeley-Rubak-Turner/p/book/9781482210200/.
Fewster, R. M. and Buckland, S. T. 2004. Assessment of distance sampling estimators. In: S. T. Buckland, D. R. Anderson, K. P. Burnham, J. L. Laake, D. L. Borchers and L. Thomas (eds) Advanced distance sampling. Oxford University Press, Oxford, U. K. Pp. 281–306.
Schlather, M., Malinowski, A., Menck, P. J., Oesting, M. and Strokorb, K. 2015. Analysis, simulation and prediction of multivariate random fields with package RandomFields. Journal of Statistical Software, 63, 1–25. URL https://www.jstatsoft.org/v63/i08/.
popn
, plot.popn
,
randomHabitat
, turnover,
simulate
temppop <- sim.popn (D = 10, expand.grid(x = c(0,100), y = c(0,100)), buffer = 50) ## plot, distinguishing "M" and "F" plot(temppop, pch = 1, cex= 1.5, col = c("green","red")[covariates(temppop)$sex]) ## add a continuous covariate ## assumes covariates(temppop) is non-null covariates(temppop)$size <- rnorm (nrow(temppop), mean = 15, sd = 3) summary(covariates(temppop)) ## Neyman-Scott cluster distribution (see also rThomas) par(xpd = TRUE, mfrow=c(2,3)) for (h in c(5,15)) for (m in c(1,4,16)) { temppop <- sim.popn (D = 10, expand.grid(x = c(0,100), y = c(0,100)), model2D = "cluster", buffer = 100, details = list(mu = m, hsigma = h)) plot(temppop) text (50,230,paste(" mu =",m, "hsigma =",h)) } par(xpd = FALSE, mfrow=c(1,1)) ## defaults ## Inhomogeneous Poisson distribution xy <- secrdemo.0$mask$x + secrdemo.0$mask$y - 900 tempD <- xy^2 / 1000 plot(sim.popn(tempD, secrdemo.0$mask, model2D = "IHP")) ## Coastal distribution in 1000-m square, homogeneous in ## x-direction arena <- data.frame(x = c(0, 1000, 1000, 0), y = c(0, 0, 1000, 1000)) plot(sim.popn(D = 5, core = arena, buffer = 0, model2D = "coastal", details = list(Beta = c(1, 1, 5, 1)))) ## Hills plot(sim.popn(D = 100, core = arena, model2D = "hills", buffer = 0, details = list(hills = c(-2,3,0,0))), cex = 0.4) ## tile demonstration pop <- sim.popn(D = 100, core = make.grid(), model2D = "coastal") par(mfrow = c(1,2), mar = c(2,2,2,2)) plot(tile(pop, "copy")) polygon(cbind(-100,200,200,-100), c(-100,-100,200,200), col = "red", density = 0) title("copy") plot(tile(pop, "reflect")) polygon(cbind(-100,200,200,-100), c(-100,-100,200,200), col = "red", density = 0) title("reflect") ## Not run: ## simulate from inhomogeneous fitted density model regionmask <- make.mask(traps(possumCH), type = "polygon", spacing = 20, poly = possumremovalarea) dts <- distancetotrap(regionmask, possumarea) covariates(regionmask) <- data.frame(d.to.shore = dts) dsurf <- predictDsurface(possum.model.Ds, regionmask) possD <- covariates(dsurf)$D.0 posspop <- sim.popn(D = possD, core = dsurf, model = "IHP") plot(regionmask, dots = FALSE, ppoly = FALSE) plot(posspop, add = TRUE, frame = FALSE) plot(traps(possumCH), add = TRUE) ## randomHabitat demonstration ## - assumes igraph has been installed # The wrapper function randomDensity may be passed to generate # a new habitat map each time sim.popn is called. The `details' argument # of sim.popn is passed to randomDensity as the `parm' argument. tempmask <- make.mask(nx = 100, ny = 100, spacing = 20) pop <- sim.popn(D = randomDensity, core = tempmask, model2D = "IHP", details = list(D = 10, p = 0.4, A = 0.5)) plot(attr(pop, 'mask'), cov = 'D', dots = FALSE) plot(pop, add = TRUE) ## rLGCP demonstration ## - assumes spatstat and RandomFields have been installed if (requireNamespace("spatstat") && requireNamespace("RandomFields")) { msk <- make.mask(traps(captdata)) # details argument 'spacing' ensures core matches Lambda below pop <- sim.popn(D = 20, core = msk, buffer = 0, model2D = "rLGCP", details = list(var=1, scale = 30, saveLambda = TRUE), seed = 1234) plot(pop) plot(traps(captdata), add = TRUE) # another IHP realisation from same LGCP intensity surface lgcp <- attr(pop, 'Lambda') pop2 <- sim.popn(D = 'Lambda', core = lgcp, model2D = "IHP") plot (lgcp, covariate = "Lambda", dots = FALSE) plot (pop2, add = TRUE, frame = FALSE) # check input and output masks match summary(lgcp) summary(msk) } ## End(Not run)
temppop <- sim.popn (D = 10, expand.grid(x = c(0,100), y = c(0,100)), buffer = 50) ## plot, distinguishing "M" and "F" plot(temppop, pch = 1, cex= 1.5, col = c("green","red")[covariates(temppop)$sex]) ## add a continuous covariate ## assumes covariates(temppop) is non-null covariates(temppop)$size <- rnorm (nrow(temppop), mean = 15, sd = 3) summary(covariates(temppop)) ## Neyman-Scott cluster distribution (see also rThomas) par(xpd = TRUE, mfrow=c(2,3)) for (h in c(5,15)) for (m in c(1,4,16)) { temppop <- sim.popn (D = 10, expand.grid(x = c(0,100), y = c(0,100)), model2D = "cluster", buffer = 100, details = list(mu = m, hsigma = h)) plot(temppop) text (50,230,paste(" mu =",m, "hsigma =",h)) } par(xpd = FALSE, mfrow=c(1,1)) ## defaults ## Inhomogeneous Poisson distribution xy <- secrdemo.0$mask$x + secrdemo.0$mask$y - 900 tempD <- xy^2 / 1000 plot(sim.popn(tempD, secrdemo.0$mask, model2D = "IHP")) ## Coastal distribution in 1000-m square, homogeneous in ## x-direction arena <- data.frame(x = c(0, 1000, 1000, 0), y = c(0, 0, 1000, 1000)) plot(sim.popn(D = 5, core = arena, buffer = 0, model2D = "coastal", details = list(Beta = c(1, 1, 5, 1)))) ## Hills plot(sim.popn(D = 100, core = arena, model2D = "hills", buffer = 0, details = list(hills = c(-2,3,0,0))), cex = 0.4) ## tile demonstration pop <- sim.popn(D = 100, core = make.grid(), model2D = "coastal") par(mfrow = c(1,2), mar = c(2,2,2,2)) plot(tile(pop, "copy")) polygon(cbind(-100,200,200,-100), c(-100,-100,200,200), col = "red", density = 0) title("copy") plot(tile(pop, "reflect")) polygon(cbind(-100,200,200,-100), c(-100,-100,200,200), col = "red", density = 0) title("reflect") ## Not run: ## simulate from inhomogeneous fitted density model regionmask <- make.mask(traps(possumCH), type = "polygon", spacing = 20, poly = possumremovalarea) dts <- distancetotrap(regionmask, possumarea) covariates(regionmask) <- data.frame(d.to.shore = dts) dsurf <- predictDsurface(possum.model.Ds, regionmask) possD <- covariates(dsurf)$D.0 posspop <- sim.popn(D = possD, core = dsurf, model = "IHP") plot(regionmask, dots = FALSE, ppoly = FALSE) plot(posspop, add = TRUE, frame = FALSE) plot(traps(possumCH), add = TRUE) ## randomHabitat demonstration ## - assumes igraph has been installed # The wrapper function randomDensity may be passed to generate # a new habitat map each time sim.popn is called. The `details' argument # of sim.popn is passed to randomDensity as the `parm' argument. tempmask <- make.mask(nx = 100, ny = 100, spacing = 20) pop <- sim.popn(D = randomDensity, core = tempmask, model2D = "IHP", details = list(D = 10, p = 0.4, A = 0.5)) plot(attr(pop, 'mask'), cov = 'D', dots = FALSE) plot(pop, add = TRUE) ## rLGCP demonstration ## - assumes spatstat and RandomFields have been installed if (requireNamespace("spatstat") && requireNamespace("RandomFields")) { msk <- make.mask(traps(captdata)) # details argument 'spacing' ensures core matches Lambda below pop <- sim.popn(D = 20, core = msk, buffer = 0, model2D = "rLGCP", details = list(var=1, scale = 30, saveLambda = TRUE), seed = 1234) plot(pop) plot(traps(captdata), add = TRUE) # another IHP realisation from same LGCP intensity surface lgcp <- attr(pop, 'Lambda') pop2 <- sim.popn(D = 'Lambda', core = lgcp, model2D = "IHP") plot (lgcp, covariate = "Lambda", dots = FALSE) plot (pop2, add = TRUE, frame = FALSE) # check input and output masks match summary(lgcp) summary(msk) } ## End(Not run)
Simulate a spatially distributed population, sample from that population with an array of detectors, and optionally fit an SECR model to the simulated data.
## S3 method for class 'secr' simulate(object, nsim = 1, seed = NULL, maxperpoly = 100, chat = 1, poponly = FALSE, ...) sim.secr(object, nsim = 1, extractfn = function(x) c(deviance = deviance(x), df = df.residual(x)), seed = NULL, maxperpoly = 100, data = NULL, tracelevel = 1, hessian = c("none", "auto", "fdHess"), start = object$fit$par, ncores = NULL, ...) sim.detect(object, popnlist, maxperpoly = 100, renumber = TRUE, expected = FALSE, dropzeroCH = TRUE)
## S3 method for class 'secr' simulate(object, nsim = 1, seed = NULL, maxperpoly = 100, chat = 1, poponly = FALSE, ...) sim.secr(object, nsim = 1, extractfn = function(x) c(deviance = deviance(x), df = df.residual(x)), seed = NULL, maxperpoly = 100, data = NULL, tracelevel = 1, hessian = c("none", "auto", "fdHess"), start = object$fit$par, ncores = NULL, ...) sim.detect(object, popnlist, maxperpoly = 100, renumber = TRUE, expected = FALSE, dropzeroCH = TRUE)
object |
a fitted secr model |
nsim |
integer number of replicates |
seed |
either NULL or an integer that will be used in a call to |
maxperpoly |
integer maximum number of detections of an individual in one polygon or transect on any occasion |
chat |
real value for overdispersion parameter |
poponly |
logical; if TRUE then only populations are simulated |
ncores |
integer number of threads used by |
extractfn |
function to extract output values from fitted model |
data |
optional list of simulated data saved from previous call to |
tracelevel |
integer for level of detail in reporting (0,1,2) |
hessian |
character or logical controlling the computation of the Hessian matrix |
start |
vector of starting ‘beta’ values for |
... |
other arguments (not used by simulate, passed to ‘extractfn’ by sim.secr) |
popnlist |
list of popn objects |
renumber |
logical; if TRUE then output animals are renumbered |
expected |
logical; if TRUE then the array of expected counts is saved as an attribute |
dropzeroCH |
logical; if TRUE then all-zero capture histories are dropped |
For each replicate, simulate.secr
calls sim.popn
to
generate session- and group-specific realizations of the (possibly
inhomogeneous) 2-D Poisson distribution fitted in object
, across
the habitat mask(s) in object
. Group subpopulations are combined
using rbind.popn
within each session; information to
reconstruct groups is retained in the individual-level factor
covariate(s) of the resulting popn
object (corresponding to
object$groups
). Unless ‘poponly = TRUE’ each population is then sampled
using the fitted
detection model and detector (trap) array(s) in object
.
The random number seed is managed as in simulate.lm
.
Certain model types are not supported by simulate.secr
. These
include models fitted using conditional likelihood (object$CL =
TRUE
), telemetry models and exotic behavioural response models.
Detector type is determined by detector(traps(object$capthist))
.
sim.secr
is a wrapper function. If data = NULL
(the
default) then it calls simulate.secr
to generate nsim
new datasets. If
data
is provided then nsim
is taken to be
length(data)
. secr.fit
is called to fit the original model
to each new dataset. Results are summarized according to the
user-provided function extractfn
. The default extractfn
returns the deviance and its degrees of freedom; a NULL value for
extractfn
returns the fitted secr objects after
trim
ming to reduce bulk. Simulation uses the detector type
of the data, even when another likelihood is fitted (this is the case
with single-catch data, for which a multi-catch likelihood is fitted).
Warning messages from secr.fit
are suppressed.
extractfn
should be a function that takes an secr
object
as its only argument.
tracelevel=0
suppresses most messages; tracelevel=1
gives a
terse message at the start of each fit; tracelevel=2
also sets
‘details$trace = TRUE’ for secr.fit
, causing each likelihood
evaluation to be reported.
hessian
controls computation of the Hessian matrix from which
variances and covariances are obtained. hessian
replaces the
value in object\$details
. Options are "none" (no variances),
"auto" (the default) or "fdhess" (see secr.fit
). It is OK
(and faster) to use hessian="none"
unless extractfn
needs
variances or covariances. Logical TRUE and FALSE are interpreted by
secr.fit
as "auto" and "none".
If ncores = NULL
then the existing value from the environment variable
RCPP_PARALLEL_NUM_THREADS is used (see setNumThreads
).
sim.capthist
is a more direct way to simulate data from a null
model (i.e. one with constant parameters for density and detection), or
from a time-varying model.
sim.detect
is a function used internally that will not usually be
called directly.
For simulate.secr
,
if poponly = TRUE, a list of populations (‘popn’ objects)
if poponly = FALSE, a list of data sets (‘capthist’ objects). This list has class c("secrdata", "list")
The initial state of the random number generator (roughly, the value of .Random.seed) is stored as the attribute ‘seed’.
The value from sim.secr
depends on extractfn
: if that
returns a numeric vector of length n.extract
then the value is a
matrix with dim = c(nsim, n.extract)
(i.e., the matrix has one
row per replicate and one column for each extracted value). Otherwise,
the value returned by sim.secr
is a list with one component per
replicate (strictly, an object of class = c("secrlist", "list")
). Each
simulated fit may be retrieved in toto by specifying
extractfn = identity
, or slimmed down by specifying
extractfn = NULL
or extractfn = trim
, which are
equivalent.
For either form of output from sim.secr
the initial state of the
random number generator is stored as the attribute ‘seed’.
For sim.detect
a list of ‘capthist’ objects.
sim.secr
does not work for mark–resight models.
The value returned by simulate.secr
is a list of ‘capthist’
objects; if there is more than one session, each ‘capthist’ is itself a
sort of list .
The classes ‘secrdata’ and ‘secrlist’ are used only to override the ugly
and usually unwanted printing of the seed attribute. However, a few
other methods are available for ‘secrlist’ objects
(e.g. plot.secrlist
).
The default value for start
in sim.secr
is the previously
fitted parameter vector. Alternatives are NULL or object$start
.
sim.capthist
, secr.fit
,
simulate
, secr.test
## Not run: ## previously fitted model simulate(secrdemo.0, nsim = 2) ## The following has been superceded by secr.test() ## this would take a long time... sims <- sim.secr(secrdemo.0, nsim = 99) deviance(secrdemo.0) devs <- c(deviance(secrdemo.0),sims$deviance) quantile(devs, probs=c(0.95)) rank(devs)[1] / length(devs) ## to assess bias and CI coverage extrfn <- function (object) unlist(predict(object)["D",-1]) sims <- sim.secr(secrdemo.0, nsim = 50, hessian = "auto", extractfn = extrfn) sims ## with a larger sample, could get parametric bootstrap CI quantile(sims[,1], c(0.025, 0.975)) ## End(Not run)
## Not run: ## previously fitted model simulate(secrdemo.0, nsim = 2) ## The following has been superceded by secr.test() ## this would take a long time... sims <- sim.secr(secrdemo.0, nsim = 99) deviance(secrdemo.0) devs <- c(deviance(secrdemo.0),sims$deviance) quantile(devs, probs=c(0.95)) rank(devs)[1] / length(devs) ## to assess bias and CI coverage extrfn <- function (object) unlist(predict(object)["D",-1]) sims <- sim.secr(secrdemo.0, nsim = 50, hessian = "auto", extractfn = extrfn) sims ## with a larger sample, could get parametric bootstrap CI quantile(sims[,1], c(0.025, 0.975)) ## End(Not run)
Data from a study of skinks (Oligosoma infrapunctatum and O. lineoocellatum) in New Zealand.
infraCH lineoCH LStraps
infraCH lineoCH LStraps
Lizards were studied over several years on a steep bracken-covered hillside on Lake Station in the Upper Buller Valley, South Island, New Zealand. Pitfall traps (sunken cans baited with a morsel of fruit in sugar syrup) were set in two large grids, each 11 x 21 traps nominally 5 meters apart, surveyed by tape and compass (locations determined later with precision surveying equipment - see Examples). Three diurnal lizard species were trapped: Oligosoma infrapunctatum, O. lineoocellatum and O. polychroma (Scincidae). The smallest species O. polychroma was seldom caught and these data are not included. The two other species are almost equal in average size (about 160 mm total length); they are long-lived and probably mature in their second or third year. The study aimed to examine their habitat use and competitive interactions.
Traps were set for 12 3-day sessions over 1995–1996, but some sessions yielded very few captures because skinks were inactive, and some sessions were incomplete for logistical reasons. The data are from sessions 6 and 7 in late spring (17–20 October 1995 and 14–17 November 1995). Traps were cleared daily; the few skinks present when traps were closed on the morning of the fourth day are treated as Day 3 captures. Individuals were marked uniquely by clipping one toe on each foot. Natural toe loss caused some problems with long-term identification; captures were dropped from the dataset when identity was uncertain. Released animals were occasionally recaptured in a different trap on the same day; these records were also discarded.
The data are provided as two two-session capthist
objects
‘infraCH’ and ‘lineoCH’. Also included is ‘LStraps’, the traps
object with the coordinates and covariates of the trap sites (these data
are also embedded in each of the capthist
objects). Pitfall traps
are multi-catch traps so detector(LStraps)
= ‘multi’.
Habitat data for each trap site are included as a dataframe of trap
covariates in LStraps
. Ground cover and vegetation were recorded
for a 1-m radius plot at each trap site. The dataframe also gives the
total number of captures of each species by site on 31 days between
April 1995 and March 1996, and the maximum potential annual solar
radiation calculated from slope and aspect (Frank and Lee 1966). Each
site was assigned to a habitat class by fuzzy clustering (Kaufman
and Rousseauw 1990; package cluster) of a distance matrix using
the ground cover, vegetation and solar radiation variables. Sites in
class 1 were open with bare ground or low-canopy vegetation including
the heath-like Leucopogon fraseri and grasses; sites in class 2
had more-closed vegetation, lacking Leucopogon fraseri and with a
higher canopy that often included Coriaria arborea. Site
variables are listed with definitions in the attribute
habitat.variables
of LStraps
(see Examples).
Object | Description |
infraCH | multi-session capthist object O. infrapunctatum |
lineoCH | multi-session capthist object O. lineoocellatum |
LStraps | traps object -- Lake Station grids |
M. G. Efford, B. W. Thomas and N. J. Spencer unpublished data.
Efford, M. G., Spencer, N. J., Thomas, B. W., Mason, R. F. and Williams, P. In prep. Distribution of sympatric skink species in relation to habitat.
Frank, E. C. and Lee , R. (1966) Potential solar beam irradiation on slopes. United States Forest Service Research Paper RM-118.
Kaufman, L. and Rousseauw, P. J. (1990) Finding groups in data: an introduction to cluster analysis. John Wiley & Sons, New York.
Spencer, N. J., Thomas, B. W., Mason, R. F. and Dugdale, J. S. (1998) Diet and life history variation in the sympatric lizards Oligosoma nigriplantare polychroma and Oligosoma lineoocellatum. New Zealand Journal of Zoology 25: 457–463.
summary (infraCH) summary (lineoCH) ## check mean distance to nearest trap etc. summary(LStraps) ## LStraps has several site covariates; terse descriptions are in ## an extra attribute that may be displayed thus attr(LStraps, "habitat.variables") ## For density modelling we need covariate values at each point in the ## habitat mask. This requires both on-grid interpolation and ## extrapolation beyond the grids. One (crude) possibility is to ## extrapolate a mask covariate from a covariate of the nearest trap: LSmask <- make.mask(LStraps, buffer = 30, type = "trapbuffer") temp <- nearesttrap(LSmask, LStraps) habclass <- covariates(LStraps)$class[temp] habclass <- factor (habclass, levels = c(1,2)) covariates(LSmask) <- data.frame(habclass) ## plot mask with colour-coded covariate par(fg = "white") ## white pixel borders plot (LSmask, covariate = "habclass", dots = FALSE, axes = FALSE, col = c("yellow", "green"), border = 0) plot(LStraps, add = TRUE, detpar = list(pch = 16)) par(fg = "black") ## default
summary (infraCH) summary (lineoCH) ## check mean distance to nearest trap etc. summary(LStraps) ## LStraps has several site covariates; terse descriptions are in ## an extra attribute that may be displayed thus attr(LStraps, "habitat.variables") ## For density modelling we need covariate values at each point in the ## habitat mask. This requires both on-grid interpolation and ## extrapolation beyond the grids. One (crude) possibility is to ## extrapolate a mask covariate from a covariate of the nearest trap: LSmask <- make.mask(LStraps, buffer = 30, type = "trapbuffer") temp <- nearesttrap(LSmask, LStraps) habclass <- covariates(LStraps)$class[temp] habclass <- factor (habclass, levels = c(1,2)) covariates(LSmask) <- data.frame(habclass) ## plot mask with colour-coded covariate par(fg = "white") ## white pixel borders plot (LSmask, covariate = "habclass", dots = FALSE, axes = FALSE, col = c("yellow", "green"), border = 0) plot(LStraps, add = TRUE, detpar = list(pch = 16)) par(fg = "black") ## default
From version 2.9.0, the model formulae provided to secr.fit
may
include smooth terms as specified for the mgcv function ‘gam’,
with some restrictions. Smooth terms may be used for both density and
detection parameters.
The specification of smooth terms is explained in
formula.gam
. Only a subset of options are relevant
to ‘secr’. Penalized splines are not available. The smooth function
may be ‘s’ or ‘te’.
The ‘wiggliness’ of the curve is controlled by the argument k, which in this implementation is set by the user. The argument ‘fx’ should be set to TRUE.
See also the example in secr-densitysurfaces.pdf.
Regression splines are a very flexible way to represent non-linear responses in generalized additive models (e.g., mgcv, Wood 2006). Borchers and Kidney (2014) have shown how they may be used to model 2-dimensional trend in density in secrgam, an R package that extends secr. Their approach is to use mgcv to construct regression spline basis functions from mask x- and y-coordinates, and possibly additional mask covariates, and to pass these as covariates to secr. The idea of using mgcv to construct the basis functions is applied within secr from version 2.9.
Smooth semi-parametric responses are also useful for modelling variation in detection parameters such as g0 and sigma over time, or in response to individual or detector-level covariates, when (1) a linear or other parametric response is arbitrary or implausible, and (2) sampling spans a range of times or levels of the covariate(s).
For a concrete example, consider a population sampled monthly for a year (i.e., 12 ‘sessions’). If home range size varies seasonally then the parameter sigma may vary in a more-or-less sinusoidal fashion. A linear trend is obviously inadequate, and a quadratic is not much better. However, a sine curve is hard to fit (we would need to estimate its phase, amplitude, mean and spatial scale) and assumes the increase and decrease phases are equally steep. An extreme solution is to treat month as a factor and estimate a separate parameter for each level (month). A smooth (semi-parametric) curve may capture the main features of seasonal variation with fewer parameters.
This implementation of smooth models results in large fitted objects, on account of the need to store setup information from mgcv. It is also vulnerable to future changes in mgcv.
Expect that the implementation will change in later versions of
secr, and that smooth models fitted in the this version will not
necessarily be compatible with predict
and
predictDsurface
in later versions.
Setting the intercept of a smooth to zero is not a canned option in mgcv, and is not offered in secr. It may be achieved by placing a knot at zero and hacking the matrix of basis functions to drop the corresponding column, plus some more jiggling.
Borchers, D. L. and Kidney, D. (2014) Flexible density surface estimation for spatially explicit capture–recapture surveys. Technical Report, University of St Andrews.
Wood, S. N. (2006) Generalized Additive Models: An Introduction with R. Chapman and Hall/CRC.
## Not run: ## smooth density surface possum.model.sxy <- secr.fit(possumCH, mask = possummask, model = D ~ s(x,y, k = 6, fx = TRUE), trace = FALSE) fittedsurface <- predictDsurface(possum.model.sxy) par(mar = c(1,1,1,6)) plot(fittedsurface) plot(fittedsurface, plottype = 'contour', add = TRUE) par(mar = c(5,4,4,2) + 0.1) ## reset to default ## Now try smooth on g0 ## For the smooth we use 'Session' which is coded numerically (0:4) ## rather than the factor 'session' ('2005', '2006', '2007', '2008', ## '2009') ovenbird.model.g0 <- secr.fit(ovenCH, mask = ovenmask, model = g0 ~ session, trace = FALSE) ovenbird.model.sg0 <- secr.fit(ovenCH, mask = ovenmask, model = g0 ~ s(Session, k = 3, fx = TRUE), trace = FALSE) AIC(ovenbird.model.g0, ovenbird.model.sg0) ## Or over occasions within a session... fit.sT3 <- secr.fit(captdata, model = g0 ~ s(T, k = 3, fx = TRUE), trace = FALSE) pred <- predict(fit.sT3, newdata = data.frame(T = 0:4)) plot(sapply(pred, '[', 'g0', 'estimate')) ## End(Not run)
## Not run: ## smooth density surface possum.model.sxy <- secr.fit(possumCH, mask = possummask, model = D ~ s(x,y, k = 6, fx = TRUE), trace = FALSE) fittedsurface <- predictDsurface(possum.model.sxy) par(mar = c(1,1,1,6)) plot(fittedsurface) plot(fittedsurface, plottype = 'contour', add = TRUE) par(mar = c(5,4,4,2) + 0.1) ## reset to default ## Now try smooth on g0 ## For the smooth we use 'Session' which is coded numerically (0:4) ## rather than the factor 'session' ('2005', '2006', '2007', '2008', ## '2009') ovenbird.model.g0 <- secr.fit(ovenCH, mask = ovenmask, model = g0 ~ session, trace = FALSE) ovenbird.model.sg0 <- secr.fit(ovenCH, mask = ovenmask, model = g0 ~ s(Session, k = 3, fx = TRUE), trace = FALSE) AIC(ovenbird.model.g0, ovenbird.model.sg0) ## Or over occasions within a session... fit.sT3 <- secr.fit(captdata, model = g0 ~ s(T, k = 3, fx = TRUE), trace = FALSE) pred <- predict(fit.sT3, newdata = data.frame(T = 0:4)) plot(sapply(pred, '[', 'g0', 'estimate')) ## End(Not run)
This function splits the transects in a ‘transect’ or ‘transectX’ traps object into multiple shorter sections. The function may also be applied directly to a capthist object based on transect data. This makes it easy to convert detection data collected along linear transects to point detection data (see Example).
snip(object, from = 0, by = 1000, length.out = NULL, keep.incomplete = TRUE, tol = 0.01)
snip(object, from = 0, by = 1000, length.out = NULL, keep.incomplete = TRUE, tol = 0.01)
object |
secr ‘traps’ or ‘capthist’ object based on transects |
from |
numeric starting posiiton (m) |
by |
numeric length of new transects (m) |
length.out |
numeric number of new transects, as alternative to ‘by’ |
keep.incomplete |
logical; if TRUE then initial or terminal sections of each original transect that are less than ‘by’ will be retained in the output |
tol |
numeric tolerance for xyontransect (capthist only) |
If a positive length.out
is specified, by
will be computed as
(transectlength(object) - from) / length.out
.
A ‘traps’ or ‘capthist’ object, according to the input.
If keep.incomplete == FALSE
animals and detections from the
snip
does not work for mark–resight data.
x <- seq(0, 4*pi, length = 41) temptrans <- make.transect(x = x*100, y = sin(x)*300) plot (snip(temptrans, by = 200), markvertices = 1) ## Not run: ## simulate some captures tempcapt <- sim.capthist(temptrans, popn = list(D = 2, buffer = 300), detectfn = 'HHN', binomN = 0, detectpar = list(lambda0 = 0.5, sigma = 50)) ## snip capture histories tempCH <- snip(tempcapt, by = 20) ## collapse from 'transect' to 'count', discarding location within transects tempCH <- reduce(tempCH, outputdetector = "count") ## fit secr model and examine H-T estimates of density ## fails with detectfn = 'HN' fit <- secr.fit(tempCH, buffer = 300, CL = TRUE, detectfn = 'HHN', trace = FALSE) derived(fit) ## also, may split an existing transect into equal lengths ## same result: plot(snip(temptrans, by = transectlength(temptrans)/10), markvertices = 1) plot(snip(temptrans, length.out = 10), markvertices = 1) ## End(Not run)
x <- seq(0, 4*pi, length = 41) temptrans <- make.transect(x = x*100, y = sin(x)*300) plot (snip(temptrans, by = 200), markvertices = 1) ## Not run: ## simulate some captures tempcapt <- sim.capthist(temptrans, popn = list(D = 2, buffer = 300), detectfn = 'HHN', binomN = 0, detectpar = list(lambda0 = 0.5, sigma = 50)) ## snip capture histories tempCH <- snip(tempcapt, by = 20) ## collapse from 'transect' to 'count', discarding location within transects tempCH <- reduce(tempCH, outputdetector = "count") ## fit secr model and examine H-T estimates of density ## fails with detectfn = 'HN' fit <- secr.fit(tempCH, buffer = 300, CL = TRUE, detectfn = 'HHN', trace = FALSE) derived(fit) ## also, may split an existing transect into equal lengths ## same result: plot(snip(temptrans, by = transectlength(temptrans)/10), markvertices = 1) plot(snip(temptrans, length.out = 10), markvertices = 1) ## End(Not run)
Rows are sorted by fields in covariates or by a provided sort key of length equal to the number of rows.
## S3 method for class 'capthist' sort(x, decreasing = FALSE, by = "", byrowname = TRUE,...) ## S3 method for class 'mask' sort(x, decreasing = FALSE, by = "", byrowname = TRUE,...)
## S3 method for class 'capthist' sort(x, decreasing = FALSE, by = "", byrowname = TRUE,...) ## S3 method for class 'mask' sort(x, decreasing = FALSE, by = "", byrowname = TRUE,...)
x |
|
decreasing |
logical. Should the sort be increasing or decreasing? |
by |
character vector (names of covariates) or data frame whose columns will be used as sort keys |
byrowname |
logical. Should row name be used as a final sort key? |
... |
other arguments (not used) |
For multi-session capthist
objects only the named covariate form
is suitable as the number of rows varies between sessions.
If requested, rows are sorted by rowname within by
. The effect of
the defaultsis to sort by rowname.
The attribute markingpoints of a mask object is removed if present, as it is no longer meaningful.
capthist
or mask
object with sorted rows; any relevant attributes are
also sorted (covariates, signal, xy)
sort(ovenCH, by = "Sex") covariates(ovenCH)[["2005"]] covariates(sort(ovenCH, by = "Sex"))[["2005"]]
sort(ovenCH, by = "Sex") covariates(ovenCH)[["2005"]] covariates(sort(ovenCH, by = "Sex"))[["2005"]]
Extract or replace the spacing attribute of a detector array or mask.
spacing(object, ...) spacing(object) <- value ## S3 method for class 'traps' spacing(object, ..., recalculate = FALSE) ## S3 method for class 'mask' spacing(object, ..., recalculate = FALSE)
spacing(object, ...) spacing(object) <- value ## S3 method for class 'traps' spacing(object, ..., recalculate = FALSE) ## S3 method for class 'mask' spacing(object, ..., recalculate = FALSE)
object |
object with ‘spacing’ attribute e.g. |
value |
numeric value for spacing |
... |
other arguments (not used) |
recalculate |
logical; if TRUE compute average spacing afresh |
The ‘spacing’ attribute of a detector array is the average distance from one detector to the nearest other detector.
The attribute was not always set by make.grid()
and
read.traps()
in versions of secr before 1.5.0. If the
attribute is found to be NULL then spacing
will compute it on the
fly.
scalar numeric value of mean spacing, or a vector if object
has multiple sessions
temptrap <- make.grid(nx = 6, ny = 8) spacing(temptrap)
temptrap <- make.grid(nx = 6, ny = 8) spacing(temptrap)
A list of ways to make secr.fit
run faster.
Check the extent and spacing of the habitat mask that you are using.
Execution time is roughly proportional to the number of mask points
(nrow(mymask)
). Default settings can lead to very large masks
for detector arrays that are elongated ‘north-south’ because the number
of points in the east-west direction is fixed. Compare results with a
much sparser mask (e.g., nx = 32 instead of nx = 64).
If you don't need to model variation in density over space or time then consider maximizing the conditional likelihood in secr.fit (CL = TRUE). This reduces the complexity of the optimization problem, especially where there are several sessions and you want session-specific density estimates (by default, derived() returns a separate estimate for each session even if the detection parameters are constant across sessions).
Do you really need to fit all those complex models? Chasing down small decrements in AIC is so last-century. Remember that detection parameters are mostly nuisance parameters, and models with big differences in AIC may barely differ in their density estimates. This is a good topic for further research - we seem to need a ‘focussed information criterion’ (Claeskens and Hjort 2008) to discern the differences that matter. Be aware of the effects that can really make a difference: learned responses (b, bk etc.) and massive unmodelled heterogeneity.
Use score.test() to compare nested models. At each stage this requires only the more simple model to have been fitted in full; further processing is required to obtain a numerical estimate of the gradient of the likelihood surface for the more complex model, but this is much faster than maximizing the likelihood. The tradeoff is that the score test is only approximate, and you may want to later verify the results using a full AIC comparison.
Suppose you are fitting models to multiple separate datasets that fit the general description of ‘sessions’. If you are fitting separate detection parameters to each session (i.e., you do not need to pool detection information), and you are not modelling trend in density across sessions, then it is much quicker to fit each session separately than to try to do it all at once. See Examples.
If your detectors are arranged in similar clusters (e.g., small square
grids) then try the function mash
.
Full data from ‘proximity’ detectors has dimensions n x S x K (n is
number of individuals, S is number of occasions, K is number of
traps). If the data are sparse (i.e. multiple detections of an
individual on one occasion are rare) then it is efficient to treat
proximity data as multi-catch data (dimension n x S, maximum of one
detection per occasion). Use reduce(proxCH, outputdetector =
"multi")
.
Most computers these days have multiple processors and these will be used by secr if the user sets ncores
greater than one in secr.fit
, sim.secr
and some other functions. If ncores = NULL
then the existing value from the environment variable RCPP_PARALLEL_NUM_THREADS is used (see setNumThreads
).
Categorical (factor) covariates with many levels and continuous covariates that take many values are not handled efficiently in secr.fit, and can dramatically slow down analyses and increase memory requirements.
Setting typsize manually in the call of 'secr.fit' can speed up fitting when magnitudes on the link scale are very different (for example, when an identity link is used for density and density is very small or very large).
Model fitting is not needed to assess power. The precision of estimates
from secr.fit can be predicted without laboriously fitting models to
simulated datasets. Just use method = "none"
to obtain the asymptotic
variance at the known parameter values for which data have been
simulated (e.g. with sim.capthist()).
Suppress computation of standard errors by derived(). For a model fitted by conditional likelihood (CL = TRUE) the subsequent computation of derived density estimates can take appreciable time. If variances are not needed (e.g., when the aim is to predict the bias of the estimator across a large number of simulations) it is efficient to set se.D = FALSE in derived().
It is tempting to save a list with the entire ‘secr’ object from each simulated fit, and to later extract summary statistics as needed. Be aware that with large simulations the overheads associated with storage of the list can become very large. The solution is to anticipate the summary statistics you will want and save only these.
Claeskens, G. and Hjort N. L. (2008) Model Selection and Model Averaging. Cambridge: Cambridge University Press.
## Not run: ## compare timing of combined model with separate single-session models ## for 5-session ovenbird mistnetting data: 2977/78 = 38-fold difference setNumThreads(7) system.time(fit1 <- secr.fit(ovenCH, buffer = 300, trace = FALSE, model = list(D ~ session, g0 ~ session, sigma ~ session))) ## user system elapsed ## 1837.71 31.81 730.56 system.time(fit2 <- lapply (ovenCH, secr.fit, buffer = 300, trace = FALSE)) ## user system elapsed ## 43.74 0.46 11.13 ## ratio of density estimates collate(fit1)[,1,1,"D"] / sapply(fit2, function(x) predict(x)["D","estimate"]) ## session=2005 session=2006 session=2007 session=2008 session=2009 ## 1.0000198 1.0000603 0.9999761 0.9999737 0.9999539 ## End(Not run)
## Not run: ## compare timing of combined model with separate single-session models ## for 5-session ovenbird mistnetting data: 2977/78 = 38-fold difference setNumThreads(7) system.time(fit1 <- secr.fit(ovenCH, buffer = 300, trace = FALSE, model = list(D ~ session, g0 ~ session, sigma ~ session))) ## user system elapsed ## 1837.71 31.81 730.56 system.time(fit2 <- lapply (ovenCH, secr.fit, buffer = 300, trace = FALSE)) ## user system elapsed ## 43.74 0.46 11.13 ## ratio of density estimates collate(fit1)[,1,1,"D"] / sapply(fit2, function(x) predict(x)["D","estimate"]) ## session=2005 session=2006 session=2007 session=2008 session=2009 ## 1.0000198 1.0000603 0.9999761 0.9999737 0.9999539 ## End(Not run)
Data of A. E. Byrom from a study of stoats (Mustela erminea) in New Zealand. Individuals were identified from DNA in hair samples.
stoatCH stoat.model.HN stoat.model.EX
stoatCH stoat.model.HN stoat.model.EX
The data are from a pilot study of stoats in red beech (Nothofagus fusca) forest in the Matakitaki Valley, South Island, New Zealand. Sticky hair-sampling tubes (n = 94) were placed on a 3-km x 3-km grid with 500-m spacing between lines and 250-m spacing along lines. Tubes were baited with rabbit meat and checked daily for 7 days, starting on 15 December 2001. Stoat hair samples were identified to individual using DNA microsatellites amplified by PCR from follicular tissue (Gleeson et al. 2010). Six loci were amplified and the mean number of alleles was 7.3 per locus. Not all loci could be amplified in 27% of samples. A total of 40 hair samples were collected (Gleeson et al. 2010), but only 30 appear in this dataset; the rest presumably did not yield sufficient DNA for genotyping.
The data are provided as a single-session capthist
object
‘stoatCH’. Hair tubes are ‘proximity’ detectors which allow
an individual to be detected at multiple detectors on one occasion
(day), but there are no multiple detections in this dataset and for
historical reasons the data are provided as detector type ‘multi’. Two
pre-fitted models are included: stoat.model.HN
and stoat.model.EX
.
Object | Description |
stoatCH | capthist object |
stoat.model.HN | fitted secr model -- null, halfnormal detection function |
stoat.model.EX | fitted secr model -- null, exponential detection function |
The log-likelihood values reported for these data by secr.fit
differ by a constant from those published by Efford et al. (2009)
because the earlier version of DENSITY used in that analysis did not
include the multinomial coefficient, which in this case is log(20!) or
about +42.336. The previous analysis also used a coarser habitat mask
than the default in secr (32 x 32 rather than 64 x 64) and this
slightly alters the log-likelihood and AIC
values.
Fitting the hazard-rate detection function previously required the shape parameter z (or b) to be fixed, but the model can be fitted in secr without fixing z. However, the hazard rate function can cause problems owing to its long tail, and it is not recommended. The check on the buffer width, usually applied automatically on completion of secr.fit, causes an error and must be suppressed with biasLimit = NA (see Examples).
Gleeson et al. (2010) address the question of whether there is enough variability at the sampled microsatellite loci to distinguish individuals. The reference to 98 sampling sites in that paper is a minor error (A. E. Byrom pers. comm.).
Efford, M. G., Borchers D. L. and Byrom, A. E. (2009) Density estimation by spatially explicit capture-recapture: likelihood-based methods. In: D. L. Thomson, E. G. Cooch and M. J. Conroy (eds) Modeling Demographic Processes in Marked Populations. Springer, New York. Pp. 255–269.
Gleeson, D. M., Byrom, A. E. and Howitt, R. L. J. (2010) Non-invasive methods for genotyping of stoats (Mustela erminea) in New Zealand: potential for field applications. New Zealand Journal of Ecology 34, 356–359. Available on-line at https://newzealandecology.org/nzje/2936/.
capthist
, Detection functions
,
secr.fit
summary(stoatCH) ## Not run: stoat.model.HN <- secr.fit(stoatCH, buffer = 1000, detectfn = 0) # this generates an error unless we use biasLimit = NA # to suppress the default bias check stoat.model.EX <- secr.fit(stoatCH, buffer = 1000, detectfn = 2) confint(stoat.model.HN, "D") ## Profile likelihood interval(s)... ## lcl ucl ## D 0.01275125 0.04055662 ## End(Not run) ## plot fitted detection functions xv <- seq(0,800,10) plot(stoat.model.EX, xval = xv, ylim = c(0,0.12), limits = FALSE, lty = 2) plot(stoat.model.HN, xval = xv, limits = FALSE, lty = 1, add = TRUE) ## review density estimates collate(stoat.model.HN, stoat.model.EX, realnames = "D", perm = c(2,3,4,1)) modelAverage(stoat.model.HN, stoat.model.EX, realnames = "D")
summary(stoatCH) ## Not run: stoat.model.HN <- secr.fit(stoatCH, buffer = 1000, detectfn = 0) # this generates an error unless we use biasLimit = NA # to suppress the default bias check stoat.model.EX <- secr.fit(stoatCH, buffer = 1000, detectfn = 2) confint(stoat.model.HN, "D") ## Profile likelihood interval(s)... ## lcl ucl ## D 0.01275125 0.04055662 ## End(Not run) ## plot fitted detection functions xv <- seq(0,800,10) plot(stoat.model.EX, xval = xv, ylim = c(0,0.12), limits = FALSE, lty = 2) plot(stoat.model.HN, xval = xv, limits = FALSE, lty = 1, add = TRUE) ## review density estimates collate(stoat.model.HN, stoat.model.EX, realnames = "D", perm = c(2,3,4,1)) modelAverage(stoat.model.HN, stoat.model.EX, realnames = "D")
This function is used with shaded plots to display a legend.
strip.legend(xy, legend, col, legendtype = c("breaks", "intervals", "other"), tileborder = NA, height = 0.5, width = 0.06, inset = 0.06, text.offset = 0.02, text.cex = 0.9, xpd = TRUE, scale = 1, title = "", box = NA, box.col = par()$bg)
strip.legend(xy, legend, col, legendtype = c("breaks", "intervals", "other"), tileborder = NA, height = 0.5, width = 0.06, inset = 0.06, text.offset = 0.02, text.cex = 0.9, xpd = TRUE, scale = 1, title = "", box = NA, box.col = par()$bg)
xy |
location of legend (see Details) |
legend |
character vector (see Details) |
col |
vector of colour values |
legendtype |
character |
tileborder |
colour of lines around each tile in the colour strip. Use NA for none. |
height |
height of colour strip as a fraction of the plot dimensions |
width |
width of colour strip as a fraction of the plot dimensions |
inset |
spacing between legend and outside plot boundary, as a fraction of the plot dimensions |
text.offset |
spacing between colour strip and text, as a fraction of the plot dimensions |
text.cex |
size of text font |
xpd |
logical, if TRUE the legend will use the margins of the plot |
scale |
numeric; each value x will be displayed as scale * x |
title |
text displayed above legend |
box |
colour of frame, if framed, otherwise NA |
box.col |
colour of background, if framed, otherwise ignored |
The location of the legend is determined by xy
which may be one
of the character values "topright", "topleft", "bottomright",
"bottomleft", "right", "left", or the x-y coordinates (in
user units) of the top-left corner of the colour strip. Coordinates may
be given as a vector or a list, and the output from
locator
(1) is suitable.
For more on colours, see notes in plot.mask
and
colors
and terrain.colors
If legendtype = 'breaks'
then labels are placed at the class
boundaries; otherwise, the labels are centred vertically. If
legendtype = 'breaks'
or legendtype = 'intervals'
then
numeric values are extracted from the input, otherwise the text strings
in legend
are used as provided.
The legend itself may be provided as a vector of values or as the class
labels output from plot.mask
. Class labels are generated
by cut
in the form ‘(0,20]’, ‘(20,40]’, etc. These are
parsed to construct either breaks (0,20,40,...) or intervals (‘0-20’,
‘20-40’,...) as requested in the legendtype
argument.
box
may also be TRUE/FALSE; if TRUE the foreground colour is used par()$fg
.
Invisibly returns a vector of user coordinates for the left, right, bottom and top of the colour strip.
From secr 2.9.0, the default behaviour of plot.mask
is to
call strip.legend
to display a legend in the top right of the
plot, labeled at breaks.
temptrap <- make.grid() tempmask <- make.mask(temptrap) covariates (tempmask) <- data.frame(circle = exp(-(tempmask$x^2 + tempmask$y^2)/10000) ) tmpleg <- plot (tempmask, covariate = "circle", dots = FALSE, breaks = 10, legend = FALSE) strip.legend (xy = 'topright', col = terrain.colors(10), legend = tmpleg, title = "Test plot") if (interactive()) { ## a custom axis using the returned values par(mar = c(2,2,2,6)) plot (tempmask, covariate = "circle", dots = FALSE, breaks = 10, legend = FALSE) b <- strip.legend (locator(1), col = terrain.colors(10), legendtype = "other", legend = " ", title = "Test plot", height = 0.3, box = NA) axis(side = 4, pos = b[2]+5, at = seq(b[4], b[3], length = 3), lab = seq(0,1,0.5), las = 1, tck = -0.02) par(mar = c(5,4,4,2) + 0.1) ## reset to default }
temptrap <- make.grid() tempmask <- make.mask(temptrap) covariates (tempmask) <- data.frame(circle = exp(-(tempmask$x^2 + tempmask$y^2)/10000) ) tmpleg <- plot (tempmask, covariate = "circle", dots = FALSE, breaks = 10, legend = FALSE) strip.legend (xy = 'topright', col = terrain.colors(10), legend = tmpleg, title = "Test plot") if (interactive()) { ## a custom axis using the returned values par(mar = c(2,2,2,6)) plot (tempmask, covariate = "circle", dots = FALSE, breaks = 10, legend = FALSE) b <- strip.legend (locator(1), col = terrain.colors(10), legendtype = "other", legend = " ", title = "Test plot", height = 0.3, box = NA) axis(side = 4, pos = b[2]+5, at = seq(b[4], b[3], length = 3), lab = seq(0,1,0.5), las = 1, tck = -0.02) par(mar = c(5,4,4,2) + 0.1) ## reset to default }
Create a new capthist
object or list of objects by selecting rows (individuals), columns (occasions) and traps from an existing capthist
object.
## S3 method for class 'capthist' subset(x, subset = NULL, occasions = NULL, traps = NULL, sessions = NULL, cutval = NULL, dropnullCH = TRUE, dropnullocc = FALSE, dropunused = TRUE, droplowsignals = TRUE, dropNAsignals = FALSE, cutabssignal = TRUE, renumber = FALSE, ...) ## S3 method for class 'capthist' split(x, f, drop = FALSE, prefix = "S", bytrap = FALSE, byoccasion = FALSE, bysession = FALSE, ...)
## S3 method for class 'capthist' subset(x, subset = NULL, occasions = NULL, traps = NULL, sessions = NULL, cutval = NULL, dropnullCH = TRUE, dropnullocc = FALSE, dropunused = TRUE, droplowsignals = TRUE, dropNAsignals = FALSE, cutabssignal = TRUE, renumber = FALSE, ...) ## S3 method for class 'capthist' split(x, f, drop = FALSE, prefix = "S", bytrap = FALSE, byoccasion = FALSE, bysession = FALSE, ...)
x |
object of class |
subset |
vector of subscripts to select rows (individuals) (see Details for variations) |
occasions |
vector of subscripts to select columns (occasions) |
traps |
vector of subscripts to select detectors (traps) |
sessions |
vector of subscripts to select sessions |
cutval |
new threshold for signal strength |
dropnullCH |
logical for whether null (all-zero) capture histories should be dropped |
dropnullocc |
logical for whether occasions with no detections should be dropped |
dropunused |
logical for whether never-used detectors should be dropped |
droplowsignals |
logical for whether cutval should be applied at each microphone rather than to sound as a whole |
dropNAsignals |
logical for whether detections with missing signal should be dropped |
cutabssignal |
logical for whether to apply cutval to absolute signal strength or the difference between signal and noise |
renumber |
logical for whether row.names should be replaced with sequence number in new |
f |
factor or object that may be coerced to a factor |
drop |
logical indicating if levels that do not occur should be dropped (if f is a factor) |
prefix |
a character prefix to be used for component names when values of f are numeric |
bytrap |
logical; if TRUE then each level of f identifies traps to include |
byoccasion |
logical; if TRUE then each level of f identifies occasions to include |
bysession |
logical; if TRUE then each level of f identifies sessions of a multisession capthist to include |
... |
other arguments passed to subset.capthist (split.capthist) or to optional subset function (subset.capthist) |
Subscript vectors may be either logical- (length equal to the relevant
dimension of x
), character- or integer-valued. Subsetting is
applied to attributes (e.g. covariates
, traps
) as
appropriate. The default action is to include all animals, occasions,
and detectors if the relevant argument is omitted.
When traps
is provided, detections at other detectors are set to
zero, as if the detector had not been used, and the corresponding rows
are removed from traps
. If the detector type is ‘proximity’ then
selecting traps also reduces the third dimension of the capthist array.
split
generates a list in which each component is a
capthist
object. Each component corresponds to a level of
f
. Multi-session capthists are accepted in secr >= 4.4.0;
f
should then be a list of factors with one component per session
and the same levels in all.
To combine (pool) occasions use reduce.capthist
. There is
no equivalent of unlist
for lists of capthist
objects.
The effect of droplowsignals = FALSE
is to retain below-threshold
measurements of signal strength on all channels (microphones) as long as
the signal is above cutval
on at least one. In this case all
retained sounds are treated as detected on all microphones. This fails
when signals are already missing on some channels.
Subsetting is awkward with multi-session input when the criterion is an individual covariate. See the Examples for one way this can be tackled.
capthist
object with the requested subset of observations, or a
list of such objects (i.e., a multi-session capthist
object).
List input results in list output, except when a single session is
selected.
split.capthist
does not work for mark–resight data.
capthist
,
rbind.capthist
,
reduce.capthist
tempcapt <- sim.capthist (make.grid(nx = 6, ny = 6), noccasions = 6) summary(subset(tempcapt, occasions = c(1,3,5))) ## Consider `proximity' detections at a random subset of detectors ## This would not make sense for `multi' detectors, as the ## excluded detectors influence detection probabilities in ## sim.capthist. tempcapt2 <- sim.capthist (make.grid(nx = 6, ny = 6, detector = "proximity"), noccasions = 6) tempcapt3 <- subset(tempcapt2, traps = sample(1:36, 18, replace = FALSE)) summary(tempcapt3) plot(tempcapt3) tempcapt4 <- split (tempcapt2, f = sample (c("A","B"), nrow(tempcapt2), replace = TRUE)) summary(tempcapt4) ## Split out captures on alternate rows of a grid tempcapt5 <- split(captdata, f = rep(1:2, 50), bytrap = TRUE) summary(tempcapt5) ## Divide one session into two by occasion tempcapt6 <- split(captdata, f = factor(c(1,1,2,2,2)), byoccasion = TRUE) summary(tempcapt6) ## Applying a covariate criterion across all sessions of a ## multi-session capthist object e.g. selecting male ovenbirds from the ## 2005--2009 ovenCH dataset. We include a restriction on occasions ## to demonstrate the use of 'MoreArgs'. Note that mapply() creates a ## list, and the class of the output must be restored manually. ovenCH.males <- mapply(subset, ovenCH, subset = lapply(ovenCH, function(x) covariates(x)$Sex == "M"), MoreArgs = list(occasions = 1:5)) class(ovenCH.males) <- class(ovenCH) summary(ovenCH.males, terse = TRUE) ## A simpler approach using a function to define subset subsetfn <- function(x, sex) covariates(x)$Sex == sex ovenCH.males <- subset(ovenCH, subset = subsetfn, sex = "M") summary(ovenCH.males, terse = TRUE)
tempcapt <- sim.capthist (make.grid(nx = 6, ny = 6), noccasions = 6) summary(subset(tempcapt, occasions = c(1,3,5))) ## Consider `proximity' detections at a random subset of detectors ## This would not make sense for `multi' detectors, as the ## excluded detectors influence detection probabilities in ## sim.capthist. tempcapt2 <- sim.capthist (make.grid(nx = 6, ny = 6, detector = "proximity"), noccasions = 6) tempcapt3 <- subset(tempcapt2, traps = sample(1:36, 18, replace = FALSE)) summary(tempcapt3) plot(tempcapt3) tempcapt4 <- split (tempcapt2, f = sample (c("A","B"), nrow(tempcapt2), replace = TRUE)) summary(tempcapt4) ## Split out captures on alternate rows of a grid tempcapt5 <- split(captdata, f = rep(1:2, 50), bytrap = TRUE) summary(tempcapt5) ## Divide one session into two by occasion tempcapt6 <- split(captdata, f = factor(c(1,1,2,2,2)), byoccasion = TRUE) summary(tempcapt6) ## Applying a covariate criterion across all sessions of a ## multi-session capthist object e.g. selecting male ovenbirds from the ## 2005--2009 ovenCH dataset. We include a restriction on occasions ## to demonstrate the use of 'MoreArgs'. Note that mapply() creates a ## list, and the class of the output must be restored manually. ovenCH.males <- mapply(subset, ovenCH, subset = lapply(ovenCH, function(x) covariates(x)$Sex == "M"), MoreArgs = list(occasions = 1:5)) class(ovenCH.males) <- class(ovenCH) summary(ovenCH.males, terse = TRUE) ## A simpler approach using a function to define subset subsetfn <- function(x, sex) covariates(x)$Sex == sex ovenCH.males <- subset(ovenCH, subset = subsetfn, sex = "M") summary(ovenCH.males, terse = TRUE)
Retain selected rows of a mask
object.
## S3 method for class 'mask' subset(x, subset, ...) ## S3 method for class 'mask' split(x, f, drop = FALSE, clusters = NULL, na.rm = TRUE, ...) ## S3 method for class 'mask' rbind(...)
## S3 method for class 'mask' subset(x, subset, ...) ## S3 method for class 'mask' split(x, f, drop = FALSE, clusters = NULL, na.rm = TRUE, ...) ## S3 method for class 'mask' rbind(...)
x |
|
subset |
numeric or logical vector to select rows of mask |
f |
factor or object that may be coerced to a factor |
drop |
logical indicating if levels that do not occur should be dropped (if f is a factor) |
clusters |
list of traps objects, each defining a cluster (alternative to f) |
na.rm |
logical; if TRUE then cells with NA value of f are dropped |
... |
two or more |
The subscripts in subset
may be of type integer, character or
logical as described in Extract
.
The split
method may use either a factor f with one value for each row or a list of clusters, each a traps object. The output mask corresponding to each cluster is the subset of the original mask points that lie within buffer of a trap within the cluster; buffer is computed as the maximum distance between a mask point in x
and any detector in clusters
. Sub-masks specified with clusters
may overlap.
Covariates are ignored by rbind.mask
.
na.rm
was introduced in 5.1.1. Previously NA cells were included.
na.rm
is not used when clusters
are defined.
For subset
, an object of class ‘mask’ with only the requested
subset of rows and ‘type’ attribute set to ‘subset’.
For split
, a list of mask objects.
For rbind
, an object of class ‘mask’ with all unique rows from
the masks in ..., and ‘type’ attribute set to ‘rbind’.
The spacing attribute is carried over from the input (it is not updated automatically). In the case of very sparse masks (i.e. those with isolated points) this may lead to an unexpected value for this attribute. (Automatic updating requires excessive computation time and/or memory for very large masks).
tempmask <- make.mask(make.grid()) OK <- (tempmask$x + tempmask$y) > 100 tempmask <- subset(tempmask, subset = OK) plot(tempmask)
tempmask <- make.mask(make.grid()) OK <- (tempmask$x + tempmask$y) > 100 tempmask <- subset(tempmask, subset = OK) plot(tempmask)
Retain selected rows of a popn object.
## S3 method for class 'popn' subset(x, subset = NULL, sessions = NULL, poly = NULL, poly.habitat = TRUE, keep.poly = TRUE, renumber = FALSE, ...)
## S3 method for class 'popn' subset(x, subset = NULL, sessions = NULL, poly = NULL, poly.habitat = TRUE, keep.poly = TRUE, renumber = FALSE, ...)
x |
|
subset |
vector to subscript the rows of |
sessions |
vector to subscript sessions if |
poly |
bounding polygon (see Details) |
poly.habitat |
logical for whether poly represents habitat or its inverse (non-habitat) |
keep.poly |
logical; if TRUE any bounding polygon is saved as the attribute ‘polygon’ |
renumber |
logical for whether to renumber rows in output |
... |
arguments passed to other functions |
The subscripts in subset
may be of type integer, character or
logical as described in Extract
. By default, all rows are
retained.
In the case of a multi-session popn object (a list of populations),
subset
may be a list with one component for the subscripts in
each new session.
If poly
is specified, points outside poly
are
dropped. poly
may be one of the types descrbed in
boundarytoSF
.
An object of class popn
with only the requested subset of rows.
Subsetting is applied to the covariates
attribute if this is
present. Attributes ‘Ndist’ and ‘model2D’ are set to NULL.
temppop <- sim.popn (D = 10, expand.grid(x = c(0,100), y = c(0,100)), buffer = 50) ## 50% binomial sample of simulated population temppops <- subset(temppop, runif(nrow(temppop)) < 0.5) plot(temppop) plot(temppops, add = TRUE, pch = 16)
temppop <- sim.popn (D = 10, expand.grid(x = c(0,100), y = c(0,100)), buffer = 50) ## 50% binomial sample of simulated population temppops <- subset(temppop, runif(nrow(temppop)) < 0.5) plot(temppop) plot(temppops, add = TRUE, pch = 16)
Retain selected rows of a traps object.
## S3 method for class 'traps' subset(x, subset = NULL, occasions = NULL, ...) ## S3 method for class 'traps' split(x, f, drop = FALSE, prefix = "S", byoccasion = FALSE, ...)
## S3 method for class 'traps' subset(x, subset = NULL, occasions = NULL, ...) ## S3 method for class 'traps' split(x, f, drop = FALSE, prefix = "S", byoccasion = FALSE, ...)
x |
|
subset |
vector to subscript the rows of |
occasions |
vector to subscript columns in |
... |
arguments passed to other functions or to optional subset function (subset.traps) |
f |
factor or object that may be coerced to a factor |
drop |
logical indicating if levels that do not occur should be dropped (if f is a factor) |
prefix |
a character prefix to be used for component names when values of f are numeric |
byoccasion |
logical ; if TRUE then f is used to split occasions |
The subscripts in subset
may be of type integer, character or
logical as described in Extract
. By default, all rows are retained.
In the case of ‘polygon’ and ‘transect’ detectors, subsetting is done at
the level of whole polygons or transects. subset
should therefore
have the same length as levels(polyID(x))
or
levels(transectID(x))
.
split
generates a list in which each component is a traps
object. Each component corresponds to a level of f
. The argument
x
of split
cannot be a list (i.e. x
must be a
single-session traps object).
If the levels of f
are numeric, from version 2.10.3 a leading zero is inserted in the names of the output list to maintain the sort order.
An object of class traps
with only the requested subset of rows.
Subsetting is applied to usage
and covariates
attributes
if these are present.
Splitting with byoccasion = TRUE
produces a list of traps
objects, each with usage codes for a subset of occasions. Traps not used
on any occasion within a session are automatically dropped from that
session.
split.traps
does not work for mark–resight data.
## odd-numbered traps only, using modulo operator temptrap <- make.grid(nx = 7, ny = 7) t2 <- subset(temptrap, as.logical(1:nrow(temptrap) %% 2)) plot(t2) ## this works also for even number of rows, but must change 'outer' call temptrap <- make.grid(nx = 8, ny = 8) t3 <- subset(temptrap, !as.logical(outer(1:8,1:8,'+')%%2)) plot(t3)
## odd-numbered traps only, using modulo operator temptrap <- make.grid(nx = 7, ny = 7) t2 <- subset(temptrap, as.logical(1:nrow(temptrap) %% 2)) plot(t2) ## this works also for even number of rows, but must change 'outer' call temptrap <- make.grid(nx = 8, ny = 8) t3 <- subset(temptrap, !as.logical(outer(1:8,1:8,'+')%%2)) plot(t3)
Determines a suitable buffer width for an integration mask. The
‘buffer’ in question defines a concave polygon around a detector array
constructed using make.mask
with type = "trapbuffer"
. The
method relies on an approximation to the bias of maximum likelihood
density estimates (M. Efford unpubl).
suggest.buffer(object, detectfn = NULL, detectpar = NULL, noccasions = NULL, ignoreusage = FALSE, ncores = NULL, RBtarget = 0.001, interval = NULL, binomN = NULL, ...) bias.D (buffer, traps, detectfn, detectpar, noccasions, binomN = NULL, control = NULL)
suggest.buffer(object, detectfn = NULL, detectpar = NULL, noccasions = NULL, ignoreusage = FALSE, ncores = NULL, RBtarget = 0.001, interval = NULL, binomN = NULL, ...) bias.D (buffer, traps, detectfn, detectpar, noccasions, binomN = NULL, control = NULL)
object |
single-session ‘secr’, ‘traps’ or ‘capthist’ object |
detectfn |
integer code or character string for shape of detection function 0 = halfnormal etc. – see detectfn |
detectpar |
list of values for named parameters of detection function – see detectpar |
noccasions |
number of sampling occasions |
ignoreusage |
logical for whether to discard usage information from
|
ncores |
integer number of threads to use for parallel processing |
RBtarget |
numeric target for relative bias of density estimate |
interval |
a vector containing the end-points of the interval to be searched |
binomN |
integer code for distribution of counts (see
|
... |
other argument(s) passed to |
buffer |
vector of buffer widths |
traps |
‘traps’ object |
control |
list of mostly obscure numerical settings (see Details) |
The basic input style of suggest.buffer
uses a ‘traps’ object and
a detection model specified by ‘detectpar’, ‘detectfn’ and ‘noccasions’,
plus a target relative bias (RB). A numerical search is conducted for
the buffer width that is predicted to deliver the requested RB. If
interval
is omitted it defaults to (1, 100S) where S is the
spatial scale of the detection function (usually
detectpar$sigma
). An error is reported if the required buffer
width is not within interval
. This often happens with
heavy-tailed detection functions (e.g., hazard-rate): choose another
function, a larger RBtarget
or a wider interval
.
Setting ncores = NULL
uses the existing value from the environment variable
RCPP_PARALLEL_NUM_THREADS (see setNumThreads
).
Convenient alternative input styles are –
secr
object containing a fitted model. Values of ‘traps’,
‘detectpar’, ‘detectfn’ and ‘noccasions’ are extracted from
object
and any values supplied for these arguments are ignored.
capthist
object containing raw data. If detectpar
is
not supplied then autoini
is used to get ‘quick and dirty’
values of g0
and sigma
for a halfnormal detection
function. noccasions
is ignored. autoini
tends to
underestimate sigma
, and the resulting buffer also tends to be
too small.
bias.D
is called internally by suggest.buffer
.
suggest.buffer
returns a scalar value for the suggested buffer
width in metres, or a vector of such values in the case of a
multi-session object
.
bias.D
returns a dataframe with columns buffer
and RB.D
(approximate bias of density estimate using finite buffer width,
relative to estimate with infinite buffer).
The algorithm in bias.D
uses one-dimensional numerical
integration of a polar approximation to site-specific detection
probability. This uses a further 3-part linear approximation for the
length of contours of distance-to-nearest-detector () as a
function of
.
The approximation seems to work well for a compact detector array, but
it should not be taken as an estimate of the bias for any other purpose:
do not report RB.D
as "the relative bias of the density
estimate". RB.D
addresses only the effect of using a finite
buffer. The effect of buffer width on final estimates should be checked
with mask.check
.
The default buffer type in make.mask
, and hence in
secr.fit
, is ‘traprect’, not ‘trapbuffer’, but a buffer width
that is adequate for ‘trapbuffer’ is always adequate for ‘traprect’.
control
contains various settings of little interest to the
user.
The potential components of control
are –
method = 1
code for method of modelling p.(X) as a function of buffer (q(r))
bfactor = 20
q(r) vs p.(X) calibration mask buffer width in multiples of trap spacing
masksample = 1000
maximum number of points sampled from calibration mask
spline.df = 10
effective degrees of freedom for
smooth.spline
ncores = NULL
integer number of cores
mask
, make.mask
, mask.check
, esaPlot
## Not run: temptraps <- make.grid() detpar <- list(g0 = 0.2, sigma = 25) suggest.buffer(temptraps, "halfnormal", detpar, 5) suggest.buffer(secrdemo.0) suggest.buffer(ovenCH[[1]]) RB <- bias.D(50:150, temptraps, "halfnormal", detpar, 5) plot(RB) detpar <- list(g0 = 0.2, sigma = 25, z=5) RB <- bias.D(50:150, temptraps, "hazard rate", detpar, 5) lines(RB) ## compare to esa plot esaPlot (temptraps, max.buffer = 150, spacing = 4, detectfn = 0, detectpar = detpar, noccasions = 5, type = "density") ## compare detection histories and fitted model as input suggest.buffer(captdata) suggest.buffer(secrdemo.0) ## End(Not run)
## Not run: temptraps <- make.grid() detpar <- list(g0 = 0.2, sigma = 25) suggest.buffer(temptraps, "halfnormal", detpar, 5) suggest.buffer(secrdemo.0) suggest.buffer(ovenCH[[1]]) RB <- bias.D(50:150, temptraps, "halfnormal", detpar, 5) plot(RB) detpar <- list(g0 = 0.2, sigma = 25, z=5) RB <- bias.D(50:150, temptraps, "hazard rate", detpar, 5) lines(RB) ## compare to esa plot esaPlot (temptraps, max.buffer = 150, spacing = 4, detectfn = 0, detectpar = detpar, noccasions = 5, type = "density") ## compare detection histories and fitted model as input suggest.buffer(captdata) suggest.buffer(secrdemo.0) ## End(Not run)
Concise description of capthist
object.
## S3 method for class 'capthist' summary(object, terse = FALSE, moves = FALSE, tpa = FALSE, ...) ## S3 method for class 'summary.capthist' print(x, ...) counts(CHlist, counts = "M(t+1)")
## S3 method for class 'capthist' summary(object, terse = FALSE, moves = FALSE, tpa = FALSE, ...) ## S3 method for class 'summary.capthist' print(x, ...) counts(CHlist, counts = "M(t+1)")
object |
|
terse |
logical; if TRUE return only summary counts |
moves |
logical; if TRUE then summary includes detected movements |
tpa |
logical; if TRUE then summary includes number of detectors per animal |
x |
|
... |
arguments passed to other functions |
CHlist |
capthist object, especially a multi-session object |
counts |
character vector of count names |
These counts are reported by summary.capthist
n | number of individuals detected on each occasion |
u | number of individuals detected for the first time on each occasion |
f | number of individuals detected exactly f times |
M(t+1) | cumulative number of individuals detected |
losses | number of individuals reported as not released on each occasion |
detections | number of detections, including within-occasion `recaptures' |
traps visited | number of detectors at which at least one detection was recorded |
traps set | number of detectors, excluding any `not set' in usage attribute of traps attribute |
The last two rows are dropped if the data are nonspatial (object has no traps attribute).
Movements are as reported by moves
. When terse = TRUE
the number of non-zero moves is reported. The temporal sequence of detections at ‘proximity’ and ‘count’ detectors is not recorded in the capthist object, so the movement statistics are not to be taken too seriously. The problem is minimised when detections are sparse (seldom more than one per animal per occasion), and does not occur with 'single' or 'multi' detectors.
The ‘tpa’ option provides the frequency distribution of detectors per animal. When terse = TRUE
the number of animals at >= 2 detectors is reported ('Animal2').
counts
may be used to return the specified counts in a compact
session x occasion table. If more than one count is named then a list is returned
with one component for each type of count.
From summary.capthist
, an object of class
summary.capthist
, a list with at least these components
detector |
|
ndetector |
number of detectors |
xrange |
range of x coordinates of detectors |
yrange |
range of y coordinates of detectors |
spacing |
mean distance from each trap to nearest other trap |
counts |
matrix of summary counts (rows) by occasion (columns). See Details. |
dbar |
mean recapture distance |
RPSV |
root pooled spatial variance |
or, when terse = TRUE
, a vector (single session) or dataframe (multiple sessions) of counts (Occasions, Detections, Animals, Detectors, and optionally Moves and Animals2).
A summary of individual covariates is provided if these are present (from secr 4.0.1).
A summary of interference/non-target captures is provided if there is a nontarget attribute (from secr 4.5.5).
temptrap <- make.grid(nx = 5, ny = 3) summary(sim.capthist(temptrap)) summary(sim.capthist(temptrap))$counts["n",] summary(captdata, moves = TRUE)
temptrap <- make.grid(nx = 5, ny = 3) summary(sim.capthist(temptrap)) summary(sim.capthist(temptrap))$counts["n",] summary(captdata, moves = TRUE)
Concise summary of a mask
object.
## S3 method for class 'mask' summary(object, ...) ## S3 method for class 'summary.mask' print(x, ...)
## S3 method for class 'mask' summary(object, ...) ## S3 method for class 'summary.mask' print(x, ...)
object |
|
x |
|
... |
other arguments (not used) |
The bounding box is the smallest rectangular area with edges parallel to the x- and y-axes that contains all points and their associated grid cells. A print method is provided for objects of class summary.mask
.
Object of class ‘summary.mask’, a list with components
detector |
character string for detector type ("single","multi","proximity") |
type |
mask type ("traprect", "trapbuffer", "pdot", "polygon", "user", "subset") |
nmaskpoints |
number of points in mask |
xrange |
range of x coordinates |
yrange |
range of y coordinates |
meanSD |
dataframe with mean and SD of x, y, and each covariate |
spacing |
nominal spacing of points |
cellarea |
area (ha) of grid cell associated with each point |
bounding box |
dataframe with x-y coordinates for vertices of bounding box |
covar |
summary of each covariate |
tempmask <- make.mask(make.grid()) ## left to right gradient covariates (tempmask) <- data.frame(x = tempmask$x) summary(tempmask)
tempmask <- make.mask(make.grid()) ## left to right gradient covariates (tempmask) <- data.frame(x = tempmask$x) summary(tempmask)
Concise summary of a popn
object.
## S3 method for class 'popn' summary(object, collapse = FALSE, ...) ## S3 method for class 'summary.popn' print(x, ...)
## S3 method for class 'popn' summary(object, collapse = FALSE, ...) ## S3 method for class 'summary.popn' print(x, ...)
object |
|
collapse |
logical; if TRUE multi-session popn objects are treated as a single open population |
x |
|
... |
other arguments (not used) |
By default each component of a multisession object is summarised separately. If collapse = TRUE
then turnover and movements are collated across sessions, matching individuals by rownames.
For summary.popn
, an object of class ‘summary.popn’ with various components. For a multisession object and collapse = TRUE the descriptors include the numbers of new individuals (recruits) and lost individuals (deaths), and matrices showing the status of each animal in each session (‘status’ codes 0 not recruited yet; 1 alive; -1 dead) and movement from previous session if alive then (‘movements’).
grid <- make.grid(8,8) turnover <- list(phi = 0.8, lambda = 1) pop <- sim.popn(Nbuffer = 200, core = grid, buffer = 200, Ndist = 'fixed', nsessions = 5, details = turnover) summary(pop, collapse = TRUE)
grid <- make.grid(8,8) turnover <- list(phi = 0.8, lambda = 1) pop <- sim.popn(Nbuffer = 200, core = grid, buffer = 200, Ndist = 'fixed', nsessions = 5, details = turnover) summary(pop, collapse = TRUE)
Concise description of traps
object.
## S3 method for class 'traps' summary(object, getspacing = TRUE, covariates = FALSE, ...) ## S3 method for class 'summary.traps' print(x, terse = FALSE, ...)
## S3 method for class 'traps' summary(object, getspacing = TRUE, covariates = FALSE, ...) ## S3 method for class 'summary.traps' print(x, terse = FALSE, ...)
object |
|
getspacing |
logical to calculate spacing of detectors from scratch |
covariates |
logical; if true each covariate is summarised |
x |
|
terse |
if TRUE suppress printing of usage and covariate summary |
... |
arguments passed to other functions |
When object
includes both categorical (factor) covariates and
usage
, usage is tabulated for each level of the covariates.
Computation of spacing
(mean distance to nearest trap) is slow
and may hit a memory limit when there are many traps. In this case, turn
off the computation with getspacing
= FALSE.
An object of class summary.traps
, a list with elements
detector |
|
ndetector |
number of detectors |
xrange |
range of x coordinates |
yrange |
range of y coordinates |
spacing |
mean distance from each trap to nearest other trap |
usage |
table of usage by occasion |
covar |
summary of covariates |
demo.traps <- make.grid() summary(demo.traps) ## uses print method for summary.traps object
demo.traps <- make.grid() summary(demo.traps) ## uses print method for summary.traps object
Extract or replace time varying covariates
timevaryingcov(object, ...) timevaryingcov(object) <- value
timevaryingcov(object, ...) timevaryingcov(object) <- value
object |
an object of class |
value |
a list of named vectors |
... |
other arguments (not used) |
The timevaryingcov attribute is a list of one or more named vectors. Each vector identifies a subset of columns of covariates(object), one for each occasion. If character values are used they should correspond to covariate names.
In secr models, time-varying covariates are restricted to traps objects. Time-varying (session-specific) individual covariates may be used in openCR. The following remarks apply to time-varying traps covariates.
The name of the vector may be used in a model formula; when the model is fitted, the value of the trap covariate on a particular occasion is retrieved from the column indexed by the vector.
For replacement, if object
already has a usage
attribute, the length of each vector in value
must match exactly
the number of columns in usage(object)
.
When converting a multi-session capthist object into a robust-design “single-session” object with function join
the argument ‘timevaryingcov’ is used to collate
covariate values across sessions in a form suitable for inclusion in
openCR models (see join
).
timevaryingcov(object)
returns the timevaryingcov attribute of
object
(may be NULL).
It is usually better to model varying effort directly, via the usage attribute (see secr-varyingeffort.pdf).
Models for data from detectors of type ‘multi’, ‘polygonX’ or ‘transectX’ take much longer to fit when detector covariates of any sort are used.
Time-varying covariates are not available with the (default) 'fastproximity' option.
See secr-varyingeffort.pdf for input of detector covariates from a file.
# make a trapping grid with simple covariates temptrap <- make.grid(nx = 6, ny = 8, detector = "multi") covariates (temptrap) <- data.frame(matrix( c(rep(1,48*3),rep(2,48*2)), ncol = 5)) head(covariates (temptrap)) # identify columns 1-5 as daily covariates timevaryingcov(temptrap) <- list(blockt = 1:5) timevaryingcov(temptrap) ## Not run: # default density = 5/ha, noccasions = 5 CH <- sim.capthist(temptrap, detectpar = list(g0 = c(0.15, 0.15, 0.15, 0.3, 0.3), sigma = 25)) fit.1 <- secr.fit(CH, trace = FALSE) fit.tvc2 <- secr.fit(CH, model = g0 ~ blockt, trace = FALSE) # because variation aligns with occasions, we get the same with: fit.t2 <- secr.fit(CH, model = g0 ~ tcov, timecov = c(1,1,1,2,2), trace = FALSE) predict(fit.t2, newdata = data.frame(tcov = 1:2)) predict(fit.tvc2, newdata = data.frame(blockt = 1:2)) # now model some more messy variation covariates (traps(CH))[1:10,] <- 3 fit.tvc3 <- secr.fit(CH, model = g0 ~ blockt, trace = FALSE) AIC(fit.tvc2, fit.t2, fit.tvc3) # fit.tvc3 is the 'wrong' model ## End(Not run)
# make a trapping grid with simple covariates temptrap <- make.grid(nx = 6, ny = 8, detector = "multi") covariates (temptrap) <- data.frame(matrix( c(rep(1,48*3),rep(2,48*2)), ncol = 5)) head(covariates (temptrap)) # identify columns 1-5 as daily covariates timevaryingcov(temptrap) <- list(blockt = 1:5) timevaryingcov(temptrap) ## Not run: # default density = 5/ha, noccasions = 5 CH <- sim.capthist(temptrap, detectpar = list(g0 = c(0.15, 0.15, 0.15, 0.3, 0.3), sigma = 25)) fit.1 <- secr.fit(CH, trace = FALSE) fit.tvc2 <- secr.fit(CH, model = g0 ~ blockt, trace = FALSE) # because variation aligns with occasions, we get the same with: fit.t2 <- secr.fit(CH, model = g0 ~ tcov, timecov = c(1,1,1,2,2), trace = FALSE) predict(fit.t2, newdata = data.frame(tcov = 1:2)) predict(fit.tvc2, newdata = data.frame(blockt = 1:2)) # now model some more messy variation covariates (traps(CH))[1:10,] <- 3 fit.tvc3 <- secr.fit(CH, model = g0 ~ blockt, trace = FALSE) AIC(fit.tvc2, fit.t2, fit.tvc3) # fit.tvc3 is the 'wrong' model ## End(Not run)
Flip (reflect), rotate or slide (translate) an array of points. Methods
are provided for ‘traps’ objects that ensure other attributes are
retained. The methods may be used with rbind.traps
to
create complex geometries.
flip (object, lr = FALSE, tb = FALSE, ...) rotate (object, degrees, centrexy = NULL, ...) shift (object, shiftxy, ...) ## S3 method for class 'traps' flip(object, lr = FALSE, tb = FALSE, ...) ## S3 method for class 'traps' rotate(object, degrees, centrexy = NULL, ...) ## S3 method for class 'traps' shift(object, shiftxy, ...) ## S3 method for class 'popn' flip(object, lr = FALSE, tb = FALSE, ...) ## S3 method for class 'popn' rotate(object, degrees, centrexy = NULL, ...) ## S3 method for class 'popn' shift(object, shiftxy, ...) ## S3 method for class 'mask' shift(object, shiftxy, ...)
flip (object, lr = FALSE, tb = FALSE, ...) rotate (object, degrees, centrexy = NULL, ...) shift (object, shiftxy, ...) ## S3 method for class 'traps' flip(object, lr = FALSE, tb = FALSE, ...) ## S3 method for class 'traps' rotate(object, degrees, centrexy = NULL, ...) ## S3 method for class 'traps' shift(object, shiftxy, ...) ## S3 method for class 'popn' flip(object, lr = FALSE, tb = FALSE, ...) ## S3 method for class 'popn' rotate(object, degrees, centrexy = NULL, ...) ## S3 method for class 'popn' shift(object, shiftxy, ...) ## S3 method for class 'mask' shift(object, shiftxy, ...)
object |
a 2-column matrix or object that can be coerced to a matrix |
lr |
either logical for whether array should be flipped left-right, or numeric value for x-coordinate of axis about which it should be flipped left-right |
tb |
either logical for whether array should be flipped top-bottom, or numeric value for y-coordinate of axis about which it should be flipped top-bottom |
degrees |
clockwise angle of rotation in degrees |
centrexy |
vector with xy coordinates of rotation centre |
shiftxy |
vector of x and y displacements |
... |
other arguments (not used) |
flip
reflects points about a vertical or horizontal axis. Logical
values for lr
or tb
indicate that points should be flipped
about the mean on the relevant axis. Numeric values indicate the
particular axis value(s) about which points should be flipped. The
default arguments result in no change.
shift
shifts the location of each point by the desired amount
on each axis.
rotate
rotates the array about a designated point. If
centrexy
is NULL then rotation is about (0,0)
(rotate.default
), the array centre (rotate.traps
), or the
centre of the bounding box (rotate.popn
).
A matrix or object of class ‘traps’ or ‘popn’ with the coordinates of each point transformed as requested.
temp <- matrix(runif (20) * 2 - 1, nc = 2) ## flip temp2 <- flip(temp, lr = 1) plot(temp, xlim=c(-1.5,4), ylim = c(-1.5,1.5), pch = 16) points (temp2, pch = 1) arrows (temp[,1], temp[,2], temp2[,1], temp2[,2], length = 0.1) abline(v = 1, lty = 2) ## rotate temp2 <- rotate(temp, 25) plot(temp, xlim=c(-1.5,1.5), ylim = c(-1.5,1.5), pch = 16) points (0,0, pch=2) points (temp2, pch = 1) arrows (temp[,1], temp[,2], temp2[,1], temp2[,2], length = 0.1) ## shiftxy temp2 <- shift(temp, c(0.1, 0.1)) plot(temp, xlim=c(-1.5,1.5), ylim = c(-1.5,1.5), pch = 16) points (0,0, pch=2) points (temp2, pch = 1) arrows (temp[,1], temp[,2], temp2[,1], temp2[,2], length = 0.1) ## flip.traps par(mfrow = c(1,2), xpd = TRUE) traps1 <- make.grid(nx = 8, ny = 6, ID = "numxb") traps2 <- flip (traps1, lr = TRUE) plot(traps1, border = 5, label = TRUE, offset = 7, gridl = FALSE) plot(traps2, border = 5, label = TRUE, offset = 7, gridl = FALSE) par(mfrow = c(1,1), xpd = FALSE) ## rotate.traps hollow1 <- make.grid(nx = 8, ny = 8, hollow = TRUE) nested <- rbind (hollow1, rotate(hollow1, 45, c(70, 70))) plot(nested, gridlines = FALSE) ## shift.traps hollow1 <- make.grid(nx = 8, ny = 8, hollow = TRUE) hollow2 <- shift(make.grid(nx = 6, ny = 6, hollow = TRUE), c(20, 20)) nested <- rbind (hollow1, hollow2) plot(nested, gridlines = FALSE, label = TRUE)
temp <- matrix(runif (20) * 2 - 1, nc = 2) ## flip temp2 <- flip(temp, lr = 1) plot(temp, xlim=c(-1.5,4), ylim = c(-1.5,1.5), pch = 16) points (temp2, pch = 1) arrows (temp[,1], temp[,2], temp2[,1], temp2[,2], length = 0.1) abline(v = 1, lty = 2) ## rotate temp2 <- rotate(temp, 25) plot(temp, xlim=c(-1.5,1.5), ylim = c(-1.5,1.5), pch = 16) points (0,0, pch=2) points (temp2, pch = 1) arrows (temp[,1], temp[,2], temp2[,1], temp2[,2], length = 0.1) ## shiftxy temp2 <- shift(temp, c(0.1, 0.1)) plot(temp, xlim=c(-1.5,1.5), ylim = c(-1.5,1.5), pch = 16) points (0,0, pch=2) points (temp2, pch = 1) arrows (temp[,1], temp[,2], temp2[,1], temp2[,2], length = 0.1) ## flip.traps par(mfrow = c(1,2), xpd = TRUE) traps1 <- make.grid(nx = 8, ny = 6, ID = "numxb") traps2 <- flip (traps1, lr = TRUE) plot(traps1, border = 5, label = TRUE, offset = 7, gridl = FALSE) plot(traps2, border = 5, label = TRUE, offset = 7, gridl = FALSE) par(mfrow = c(1,1), xpd = FALSE) ## rotate.traps hollow1 <- make.grid(nx = 8, ny = 8, hollow = TRUE) nested <- rbind (hollow1, rotate(hollow1, 45, c(70, 70))) plot(nested, gridlines = FALSE) ## shift.traps hollow1 <- make.grid(nx = 8, ny = 8, hollow = TRUE) hollow2 <- shift(make.grid(nx = 6, ny = 6, hollow = TRUE), c(20, 20)) nested <- rbind (hollow1, hollow2) plot(nested, gridlines = FALSE, label = TRUE)
Construct detector layouts comprising small arrays (clusters) replicated across space, possibly at a probability sample of points.
trap.builder (n = 10, cluster, region = NULL, frame = NULL, method = c("SRS", "GRTS", "all", "rank"), edgemethod = c("clip", "allowoverlap", "allinside", "anyinside", "centreinside"), samplefactor = 2, ranks = NULL, rotation = NULL, detector, exclude = NULL, exclmethod = c("clip", "alloutside", "anyoutside", "centreoutside"), plt = FALSE, add = FALSE, ...) mash (object, origin = c(0,0), clustergroup = NULL, ...) cluster.counts (object) cluster.centres (object)
trap.builder (n = 10, cluster, region = NULL, frame = NULL, method = c("SRS", "GRTS", "all", "rank"), edgemethod = c("clip", "allowoverlap", "allinside", "anyinside", "centreinside"), samplefactor = 2, ranks = NULL, rotation = NULL, detector, exclude = NULL, exclmethod = c("clip", "alloutside", "anyoutside", "centreoutside"), plt = FALSE, add = FALSE, ...) mash (object, origin = c(0,0), clustergroup = NULL, ...) cluster.counts (object) cluster.centres (object)
n |
integer number of clusters (ignored if method = "all") |
cluster |
traps object |
region |
bounding polygon(s) |
frame |
data frame of points used as a finite sampling frame |
method |
character string (see Details) |
edgemethod |
character string (see Details) |
samplefactor |
oversampling to allow for rejection of edge clusters (multiple of n) |
ranks |
vector of relative importance (see Details) |
rotation |
angular rotation of each cluster about centre (degrees) |
detector |
character detector type (see |
exclude |
polygon(s) from which detectors are to be excluded |
exclmethod |
character string (see Details) |
plt |
logical: should array be plotted? |
add |
logical: add to existing plot |
object |
single-session multi-cluster capthist object, or traps
object for |
origin |
new coordinate origin for detector array |
clustergroup |
list of vectors subscripting the clusters to be mashed |
... |
other arguments passed by trap.builder to spsurvey::grts (e.g., mindis) and by mash to make.capthist (e.g., sortrows) |
The detector array in cluster
is replicated n
times and translated to centres sampled from the area sampling frame
in region
or the finite sampling frame in frame
. Each
cluster may be rotated about its centre either by a fixed number of
degrees (rotation
positive), or by a random angle (rotation
negative).
If the cluster
argument is not provided then single detectors of
the given type are placed according to the design.
The sampling frame is finite (the points in frame
) whenever
frame
is not NULL. If region
and frame
are both
specified, sampling uses the finite frame but sites may be clipped
using the polygon.
region
and exclude
may be a two-column matrix or
dataframe of x-y coordinates for the boundary, or one of the other polygon
sources listed in boundarytoSF
(these allow multiple polygons).
method
may be "SRS", "GRTS", "all" or "rank". "SRS" takes a simple
random sample (without replacement in the case of a finite sampling
frame). "GRTS" takes a spatially representative sample using the
‘generalized random tessellation stratified’ (GRTS) method of Stevens
and Olsen (2004). "all" replicates cluster
across all points in
the finite sampling frame. "rank" selects n
sites from
frame
on the basis of their ranking on the vector ‘ranks’,
which should have length equal to the number of rows in
frame
; ties are resolved by drawing a site at random.
Options for edgemethod
are –
edgemethod |
Description |
"clip" |
reject any individual detectors outside region |
"allowoverlap" |
no action |
"allinside" |
reject whole cluster if any component is outside region |
"anyinside" |
reject whole cluster if no component is inside region
|
"centreinside" |
reject whole cluster if centre outside region , and clip to region
|
Similarly, exclmethod
may be "clip" (reject individual detectors),
"alloutside" (reject whole cluster if any component is outside exclude
) etc.
Sufficient additional samples ((samplefactor--1) * n
) must be drawn to
allow for replacement of any rejected clusters; otherwise, an error is reported
(‘not enough clusters within polygon’).
GRTS samples require function grts
in version >= 5.3.0 of package spsurvey
(Dumelle et al. 2022). More sophisticated stratified designs may be specified by using grts
directly.
mash
collapses a multi-cluster capthist object as if all
detections were made on a single cluster. The new detector coordinates
in the ‘traps’ attribute are for a single cluster with (min(x),
min(y)) given by origin
. clustergroup
optionally selects
one or more groups of clusters to mash; if length(clustergroup)
> 1
then a multisession capthist object will be generated, one
‘session’ per clustergroup. By default, all clusters are mashed.
mash
discards detector-level covariates and occasion-specific
‘usage’, with a warning.
cluster.counts
returns the number of distinct
individuals detected per cluster in a single-session multi-cluster
capthist object.
cluster.centres
returns the centroid of the detector locations in each cluster. When clusters have been truncated these differ from the attribute centres
set by make.systematic
.
trap.builder
produces an object of class ‘traps’.
plt = TRUE
causes a plot to be displayed, including the polygon
or finite sampling frame as appropriate.
mash
produces a capthist object with the same number of rows as
the input but different detector numbering and ‘traps’. An attribute
‘n.mash’ is a vector of the numbers recorded at each cluster; its
length is the number of clusters. An attribute ‘centres’ is a
dataframe containing the x-y coordinates of the cluster centres. The
predict
method for secr objects and the function derived
both recognise and adjust for mashing.
cluster.counts
returns a vector with the number of individuals
detected at each cluster.
cluster.centres
returns a dataframe of x- and y-coordinates.
The function make.systematic
should be used to generate
systematic random layouts. It calls trap.builder
.
The sequence number of the cluster to which each detector belongs, and
its within-cluster sequence number, may be retrieved with the
functions clusterID
and clustertrap
.
Dumelle, M., Kincaid, T. M., Olsen, A. R., and Weber, M. H. (2021). spsurvey: Spatial Sampling Design and Analysis. R package version 5.2.0.
Stevens, D. L., Jr., and Olsen, A. R. (2004) Spatially-balanced sampling of natural resources. Journal of the American Statistical Association 99, 262–278.
make.grid
, traps
,
make.systematic
,
clusterID
,
clustertrap
## solitary detectors placed randomly within a rectangle tempgrid <- trap.builder (n = 10, method = "SRS", region = cbind(x = c(0,1000,1000,0), y = c(0,0,1000,1000)), plt = TRUE) ## one detector in each 100-m grid cell - ## a form of stratified simple random sample ## see also Examples in ?make.grid origins <- expand.grid(x = seq(0, 900, 100), y = seq(0, 1100, 100)) XY <- origins + runif(10 * 12 * 2) * 100 temp <- trap.builder (frame = XY, method = "all", detector = "multi") ## same as temp <- read.traps(data = XY) plot(temp, border = 0) ## default grid is 100 m ## Not run: ## simulate some data ## regular lattice of mini-arrays minigrid <- make.grid(nx = 3, ny = 3, spacing = 50, detector = "proximity") tempgrid <- trap.builder (cluster = minigrid , method = "all", frame = expand.grid(x = seq(1000, 5000, 2000), y = seq(1000, 5000, 2000)), plt = TRUE) tempcapt <- sim.capthist(tempgrid, popn = list(D = 10)) cluster.counts(tempcapt) cluster.centres(tempgrid) ## "mash" the CH summary(mash(tempcapt)) ## compare timings (estimates are near identical) tempmask1 <- make.mask(tempgrid, type = "clusterrect", buffer = 200, spacing = 10) fit1 <- secr.fit(tempcapt, mask = tempmask1, trace = FALSE) tempmask2 <- make.mask(minigrid, spacing = 10) fit2 <- secr.fit(mash(tempcapt), mask = tempmask2, trace = FALSE) ## density estimate is adjusted automatically ## for the number of mashed clusters (9) predict(fit1) predict(fit2) fit1$proctime fit2$proctime ## SRS excluding detectors from a polygon region <- cbind(x = c(0,6000,6000,0,0), y = c(0,0,6000,6000,0)) exclude <- cbind(x = c(3000,7000,7000,3000,3000), y = c(2000,2000,4000,4000,2000)) newgrid <- trap.builder (n = 40, cluster = minigrid, method = "SRS", edgemethod = "allinside", region = region, exclude = exclude, exclmethod = "alloutside", plt = TRUE) ## two-phase design: preliminary sample across region, ## followed by selection of sites for intensive grids arena <- data.frame(x = c(0,2000,2000,0), y = c(0,0,2500,2500)) t1 <- make.grid(nx = 1, ny = 1) t4 <- make.grid(nx = 4, ny = 4, spacing = 50) singletraps <- make.systematic (n = c(8,10), cluster = t1, region = arena) CH <- sim.capthist(singletraps, popn = list(D = 2)) plot(CH, type = "n.per.cluster", title = "Number per cluster") temp <- trap.builder(10, frame = traps(CH), cluster = t4, ranks = cluster.counts(CH), method = "rank", edgemethod = "allowoverlap", plt = TRUE, add = TRUE) ## GRTS sample of mini-grids within a rectangle ## GRTS uses package 'spsurvey' >= 5.3.0 minigrid <- make.grid(nx = 3, ny = 3, spacing = 50, detector = "proximity") region <- cbind(x = c(0,6000,6000,0,0), y = c(0,0,6000,6000,0)) if (requireNamespace("spsurvey", versionCheck = list(version = ">=5.3.0"))) { tempgrid <- trap.builder (n = 20, cluster = minigrid, region = region, plt = TRUE, method = "GRTS") # specifying minimum distance between cluster origins tempgrid2 <- trap.builder (n = 20, cluster = minigrid, region = region, plt = TRUE, method = "GRTS", mindis = 500, maxtry = 10) # use spsurvey::warnprnt() to view warnings (e.g., maxtry inadequate) } ## End(Not run)
## solitary detectors placed randomly within a rectangle tempgrid <- trap.builder (n = 10, method = "SRS", region = cbind(x = c(0,1000,1000,0), y = c(0,0,1000,1000)), plt = TRUE) ## one detector in each 100-m grid cell - ## a form of stratified simple random sample ## see also Examples in ?make.grid origins <- expand.grid(x = seq(0, 900, 100), y = seq(0, 1100, 100)) XY <- origins + runif(10 * 12 * 2) * 100 temp <- trap.builder (frame = XY, method = "all", detector = "multi") ## same as temp <- read.traps(data = XY) plot(temp, border = 0) ## default grid is 100 m ## Not run: ## simulate some data ## regular lattice of mini-arrays minigrid <- make.grid(nx = 3, ny = 3, spacing = 50, detector = "proximity") tempgrid <- trap.builder (cluster = minigrid , method = "all", frame = expand.grid(x = seq(1000, 5000, 2000), y = seq(1000, 5000, 2000)), plt = TRUE) tempcapt <- sim.capthist(tempgrid, popn = list(D = 10)) cluster.counts(tempcapt) cluster.centres(tempgrid) ## "mash" the CH summary(mash(tempcapt)) ## compare timings (estimates are near identical) tempmask1 <- make.mask(tempgrid, type = "clusterrect", buffer = 200, spacing = 10) fit1 <- secr.fit(tempcapt, mask = tempmask1, trace = FALSE) tempmask2 <- make.mask(minigrid, spacing = 10) fit2 <- secr.fit(mash(tempcapt), mask = tempmask2, trace = FALSE) ## density estimate is adjusted automatically ## for the number of mashed clusters (9) predict(fit1) predict(fit2) fit1$proctime fit2$proctime ## SRS excluding detectors from a polygon region <- cbind(x = c(0,6000,6000,0,0), y = c(0,0,6000,6000,0)) exclude <- cbind(x = c(3000,7000,7000,3000,3000), y = c(2000,2000,4000,4000,2000)) newgrid <- trap.builder (n = 40, cluster = minigrid, method = "SRS", edgemethod = "allinside", region = region, exclude = exclude, exclmethod = "alloutside", plt = TRUE) ## two-phase design: preliminary sample across region, ## followed by selection of sites for intensive grids arena <- data.frame(x = c(0,2000,2000,0), y = c(0,0,2500,2500)) t1 <- make.grid(nx = 1, ny = 1) t4 <- make.grid(nx = 4, ny = 4, spacing = 50) singletraps <- make.systematic (n = c(8,10), cluster = t1, region = arena) CH <- sim.capthist(singletraps, popn = list(D = 2)) plot(CH, type = "n.per.cluster", title = "Number per cluster") temp <- trap.builder(10, frame = traps(CH), cluster = t4, ranks = cluster.counts(CH), method = "rank", edgemethod = "allowoverlap", plt = TRUE, add = TRUE) ## GRTS sample of mini-grids within a rectangle ## GRTS uses package 'spsurvey' >= 5.3.0 minigrid <- make.grid(nx = 3, ny = 3, spacing = 50, detector = "proximity") region <- cbind(x = c(0,6000,6000,0,0), y = c(0,0,6000,6000,0)) if (requireNamespace("spsurvey", versionCheck = list(version = ">=5.3.0"))) { tempgrid <- trap.builder (n = 20, cluster = minigrid, region = region, plt = TRUE, method = "GRTS") # specifying minimum distance between cluster origins tempgrid2 <- trap.builder (n = 20, cluster = minigrid, region = region, plt = TRUE, method = "GRTS", mindis = 500, maxtry = 10) # use spsurvey::warnprnt() to view warnings (e.g., maxtry inadequate) } ## End(Not run)
An object of class traps
encapsulates a set of detector (trap)
locations and related data. A method of the same name extracts or
replaces the traps
attribute of a capthist
object.
traps(object, ...) traps(object) <- value
traps(object, ...) traps(object) <- value
object |
a |
value |
|
... |
other arguments (not used). |
An object of class traps
holds detector (trap) locations as a
data frame of x-y coordinates. Trap identifiers are used as row names.
The required attribute ‘detector’ records the type of detector
("single", "multi" or "proximity" etc.; see detector
for
more).
Other possible attributes of a traps
object are:
spacing |
mean distance to nearest detector |
spacex |
|
spacey |
|
covariates |
dataframe of trap-specific covariates |
clusterID |
identifier of the cluster to which each detector belongs |
clustertrap |
sequence number of each trap within its cluster |
usage |
a traps x occasions matrix of effort (may be binary 0/1) |
markocc |
integer vector distinguishing marking occasions (1) from sighting occasions (0) |
newtrap |
vector recording aggregation of detectors by
reduce.traps
|
If usage is specified, at least one detector must be ‘used’ (usage non-zero) on each occasion.
Various array geometries may be constructed with functions such as
make.grid
and make.circle
, and these may be
combined or placed randomly with trap.builder
.
Generic methods are provided to select rows
(subset.traps
), combine two or more arrays
(rbind.traps
), aggregate detectors
(reduce.traps
), shift an array
(shift.traps
), or rotate an array
(rotate.traps
).
The attributes usage
and covariates
may be extracted or
replaced using generic methods of the same name.
Efford, M. G. (2012) DENSITY 5.0: software for spatially explicit capture–recapture. Department of Mathematics and Statistics, University of Otago, Dunedin, New Zealand. https://www.otago.ac.nz/density/.
Efford, M. G., Borchers D. L. and Byrom, A. E. (2009) Density estimation by spatially explicit capture-recapture: likelihood-based methods. In: D. L. Thomson, E. G. Cooch and M. J. Conroy (eds) Modeling Demographic Processes in Marked Populations. Springer, New York. Pp. 255–269.
make.grid
, read.traps
,
rbind.traps
, reduce.traps
,
plot.traps
, secr.fit
,
spacing
, detector
,
covariates
, trap.builder
,
as.mask
demotraps <- make.grid(nx = 8, ny = 6, spacing = 30) demotraps ## uses print method for traps summary (demotraps) plot (demotraps, border = 50, label = TRUE, offset = 8, gridlines=FALSE) ## generate an arbitrary covariate `randcov' covariates (demotraps) <- data.frame(randcov = rnorm(48)) ## overplot detectors that have high covariate values temptr <- subset(demotraps, covariates(demotraps)$randcov > 0.5) plot (temptr, add = TRUE, detpar = list (pch = 16, col = "green", cex = 2))
demotraps <- make.grid(nx = 8, ny = 6, spacing = 30) demotraps ## uses print method for traps summary (demotraps) plot (demotraps, border = 50, label = TRUE, offset = 8, gridlines=FALSE) ## generate an arbitrary covariate `randcov' covariates (demotraps) <- data.frame(randcov = rnorm(48)) ## overplot detectors that have high covariate values temptr <- subset(demotraps, covariates(demotraps)$randcov > 0.5) plot (temptr, add = TRUE, detpar = list (pch = 16, col = "green", cex = 2))
Extract or replace attributes of an object of class ‘traps’.
polyID(object) polyID(object) <- value transectID(object) transectID(object) <- value searcharea(object) transectlength(object)
polyID(object) polyID(object) <- value transectID(object) transectID(object) <- value searcharea(object) transectlength(object)
object |
a ‘traps’ object |
value |
replacement value (see Details) |
The ‘polyID’ and ‘transectID’ functions assign and extract the attribute of a ‘traps’ object that relates vertices (rows) to particular polygons or transects. The replacement value should be a factor of length equal to nrow(object).
The ‘searcharea’ of a ‘polygon’ traps object is a vector of the areas of the component polygons in hectares. This value is read-only.
The ‘transectlength’ of a ‘transect’ traps object is a vector of the lengths of the component transects in metres. This value is read-only.
polyID
- a factor with one level per polygon. If the object does not have a polyID attribute then a factor with one level for each detector.
searcharea
- numeric value of polygon areas, in hectares.
transectlength
- numeric value of transect lengths, in metres.
## default is a single polygon temp <- make.grid(detector = "polygon", hollow = TRUE) polyID(temp) plot(temp) ## split in two temp <- make.grid(detector = "polygon", hollow = TRUE) polyID(temp) <- factor(rep(c(1,2),rep(10,2))) plot(temp)
## default is a single polygon temp <- make.grid(detector = "polygon", hollow = TRUE) polyID(temp) plot(temp) ## split in two temp <- make.grid(detector = "polygon", hollow = TRUE) polyID(temp) <- factor(rep(c(1,2),rep(10,2))) plot(temp)
Functions for multi-session density trend analysis.
predictDlambda(object, alpha = 0.05)
predictDlambda(object, alpha = 0.05)
object |
multi-session secr object output from secr.fit |
alpha |
alpha level for confidence intervals |
Usage is described in secr-trend.pdf. Briefly, setting details argument 'Dlambda' in 'secr.fit
causes the density model (D~xxx) to be interpreted as a session-specific trend model with parameters for the initial density (D1) and each subsequent session-on-session change in density .
A table of session-specific estimates (initial D, subsequent ) with SE and confidence intervals.
# a model with constant lambda msk <- make.mask(traps(ovenCH[[1]]), buffer = 300, nx = 25) fit <- secr.fit(ovenCH, model = D~1, mask = msk, trace = FALSE, details = list(Dlambda = TRUE), ncores = 2) predictDlambda(fit)
# a model with constant lambda msk <- make.mask(traps(ovenCH[[1]]), buffer = 300, nx = 25) fit <- secr.fit(ovenCH, model = D~1, mask = msk, trace = FALSE, details = list(Dlambda = TRUE), ncores = 2) predictDlambda(fit)
Drop unwanted components from a list
object, usually to save space.
## Default S3 method: trim(object, drop, keep) ## S3 method for class 'secr' trim(object, drop = c("call", "mask", "designD", "designNE", "design", "design0"), keep = NULL) ## S3 method for class 'secrlist' trim(object, drop = c("call", "mask", "designD", "designNE", "design", "design0"), keep = NULL)
## Default S3 method: trim(object, drop, keep) ## S3 method for class 'secr' trim(object, drop = c("call", "mask", "designD", "designNE", "design", "design0"), keep = NULL) ## S3 method for class 'secrlist' trim(object, drop = c("call", "mask", "designD", "designNE", "design", "design0"), keep = NULL)
object |
a list object |
drop |
vector identifying components to be dropped |
keep |
vector identifying components to be kept |
drop
may be a character vector of names or a numeric vector of
indices. If both drop
and keep
are given then the action
is conservative, dropping only components in drop
and not in
keep
.
Be warned that some further operations on fitted secr objects become impossible once you have discarded the default components.
a list retaining selected components.
names(secrdemo.0) names(trim(secrdemo.0)) object.size(secrdemo.0) object.size(trim(secrdemo.0)) object.size(trim(secrlist(secrdemo.0, secrdemo.b)))
names(secrdemo.0) names(trim(secrdemo.0)) object.size(secrdemo.0) object.size(trim(secrdemo.0)) object.size(trim(secrlist(secrdemo.0, secrdemo.b)))
Although secr.fit
is quite robust, it does not always
work. Inadequate data or an overambitious model occasionally cause
numerical problems in the algorithms used for fitting the model, or
problems of identifiability, as described for capture–recapture models
in general by Gimenez et al. (2004). Here are some tips that may help
you.
This page has largely been superceded by secr-troubleshooting.pdf.
Most likely you have infeasible starting values for the
parameters. try some alternatives, specifying them manually with the
start
argument.
This usually means the model did not fit and the estimates should not be trusted. Extremely large variances or standard errors also indicate problems.
Try another maximization method (method = "Nelder-Mead"
is more robust than the default). The same maximum likelihood should
be found regardless of method, so AIC values are comparable across
methods.
Repeat the maximization with different starting values. You can use
secr.fit(..., start = last.model)
where last.model
is a
previously fitted secr object.
If you think the estimates are right but there was a problem
in computing the variances, try re-running secr.fit() with the
previous model as starting values (see preceding) and set
method = "none"
. This bypasses maximization and computes the
variances afresh using fdHess
from nlme.
Try a finer mask (e.g., vary argument nx
in
make.mask
). Check that the extent of the mask matches your
data.
The maximization algorithms work poorly when the beta coefficients
are of wildly different magnitude. This may happen when using
covariates: ensure beta coefficients are similar (within a factor of
5–10 seems adequate, but this is not based on hard evidence) by scaling
any covariates you provide. This can be achieved by setting the
typsize
argument of nlm
or the parscale
control argument of optim
.
Examine the model. Boundary values (e.g., g0 near 1.0) may give
problems. In the case of more complicated models you may gain insight by
fixing the value of a difficult-to-estimate parameter (argument
fixed
).
See also the section ‘Potential problems’ in secr-densitysurfaces.pdf.
This condition does not invariably indicate a failure of model fitting. Proceed with caution, checking as suggested in the preceding section.
A feature of the maximization algorithm used by default in nlm
is that it takes a large step in the parameter space early on in the
maximization. The step may be so large that it causes floating point
underflow or overflow in one or more real parameters. This can be
controlled by passing the ‘stepmax’ argument of nlm
in the
... argument of secr.fit
(see first example). See also the
previous point about scaling of covariates.
This is a problem particularly when using individual covariates in a model fitted by maximizing the conditional likelihood. The memory required is then roughly proportional to the product of the number of individuals, the number of occasions, the number of detectors and the number of latent classes (for finite-mixture models). When maximizing the full-likelihood, substitute ‘number of groups’ for 'number of individuals'. [The limit is reached in external C used for the likelihood calculation, which uses the R function ‘R_alloc’.]
The mash
function may be used to reduce the number of
detectors when the design uses many identical and independent
clusters. Otherwise, apply your ingenuity to simplify your model,
e.g., by casting ‘groups’ as ‘sessions’. Memory is less often an issue
on 64-bit systems (see link below).
These models have known problems due to multimodality of the likelihood. See secr-finitemixtures.pdf.
Gimenez, O., Viallefont, A., Catchpole, E. A., Choquet, R. and Morgan, B. J. T. (2004) Methods for investigating parameter redundancy. Animal Biodiversity and Conservation 27, 561–572.
sim.popn
can simulate a multi-session population with known between-session survival, recruitment and movement probabilities. The parameter settings to achieve this are passed to sim.popn
in its ‘details’ argument. Components of ‘details’ that are relevant to turnover are described below; see sim.popn
for others.
Multi-session populations are generated in sim.popn
whenever its argument ‘nsessions’ is greater than 1. If details$lambda
remains NULL (the default) then the population for each successive session is generated de novo from the given density model (model2D, D etc.). If a value is specified for details$lambda
then only the first population is generated de novo; remaining populations are generated iteratively with probabilistic mortality, recruitment and movement as described here.
sim.popn
details argumentComponent | Description | Default |
phi |
per capita survival rate |
0.7 |
survmodel |
probability model for number of survivors | ``binomial" |
lambda |
finite rate of increase |
none |
recrmodel |
probability model for number of recruits | ``poisson" |
superN |
optional superpopulation size for `multinomial' recruitment model | see below |
Nrecruits |
number of recruits to add at t+1 for `specified' recruitment model | 0 |
movemodel |
``static", ``uncorrelated", ``normal", ``exponential", ``t2D" or a user function | ``static" |
move.a |
first parameter of movement kernel (replacing sigma.m) | 0 |
move.b |
second parameter of movement kernel | 1 |
edgemethod |
treatment of animals that cross the boundary | ``wrap" |
sigma.m |
deprecated in 3.2.1; use move.a | 0 |
wrap |
deprecated in 3.1.6; use edgemethod | TRUE i.e. edgemethod = ``wrap" |
Survival is usually thought of as a Bernoulli process (outcome 0 or 1 for each individual) so the number of survivors is a binomial variable (survmodel = "binomial"). Another approach is to fix the proportion surviving, but this can be done exactly only when
is an integer. A (slightly ad hoc) solution is to randomly choose between the two nearest integers with probability designed in the long term (over many sessions) to give the required
(survmodel = "discrete").
Per capita recruitment () is the difference between lambda and phi (
), which must be non-negative (phi > lambda causes an error). The number of recruits B is a random variable whose probability distribution is controlled by details$recrmodel:
Value | Probability model |
"constantN" | Exact replacement of animals that die (B = ) |
"binomial" | Binomial number of recruits (B ~ bin( ) |
"poisson" | Poisson number of recruits (B ~ pois( )) |
"discrete" | Minimum-variance number of recruits (see Survival) |
"multinomial" | The POPAN model, conditioning on superpopulation size (e.g., Schwarz and Arnason 1996)) |
"specified" | Add the number of recruits specified in Nrecruits (may be vector) |
In the case of binomial recruitment there is a maximum of one recruit per existing individual, so lambda <= (phi+1). Multinomial recruitment requires a value for the superpopulation size. This may be provided as the details component "superN". If not specified directly, a value is inferred by projecting a trial initial (simulated) population using the specified phi and lambda.
Specifying the integer number of recruits in each year (recrmodel ‘specified’) overrides the value of lambda, but a non-null value should be given for lambda.
Individuals may shift their home range centre between sessions. Movement probability is governed by a circular kernel specified by ‘movemodel’ and the parameter values ‘move.a’ and ‘move.b’ (optional). By default there is no movement between sessions (movemodel = "static"). Other options are
``IND" | ``uncorrelated" | individuals are randomly assigned a new, independent location within the buffered area |
``BVN" | ``normal" | bivariate normal (Gaussian) kernel with parameter move.a (previously called sigma.m) |
``BVE" | ``exponential" | negative exponential (Laplace) kernel with parameter move.a |
``BVT" | ``t2D" | circular 2-dimensional t-distribution with scale parameter move.a and shape parameter move.b = df/2 (2Dt of Clark et al. 1999) |
``RDE" | exponential distribution of radial distance (Ergon & Gardner, 2014) | |
``RDG" | gamma distribution of radial distance (Ergon & Gardner, 2014) | |
``RDL" | log-normal distribution of radial distance (Ergon & Gardner, 2014) | |
(parameterized with move.a = exp(mu), move.b = 1/CV^2 = 1 / (exp(SD^2) - 1) | ||
The package openCR >=1.4.0 provides functions for constructing and plotting these kernels and summarising their properties (make.kernel
; plot
and summary
methods for kernel objects). The secr function extractMoves
is useful for checking simulations of movement.
Models IND, BVN, BVE, and RDE may also be zero-inflated (suffix “zi"). The parameter ‘move.a’ (INDzi) or ‘move.b’ (BVNzi, BVEzi, RDEzi) is the zero-inflation probability. See Examples.
In secr <3.2.1 sigma.m was also used to indicate two special cases; these continue to work but may be discontinued in the future:
sigma.m = 0 corresponds to movemodel = ‘static’
sigma.m < 0 corresponds to movemodel = ‘uncorrelated’
In secr >= 4.4.0, the ‘movemodel’ component may also be a user-provided function with these characteristics: two or three arguments, the first being the number of centres to be moved (e.g., n) and the others parameters of the dispersal distribution (e.g., a,b); the function should return a matrix of n rows and 2 columns, the displacements in the x- and y-directions. The output is a set of random points from the bivariate dispersal kernel. The function will be called with the current number of centres and parameter values move.a and move.b as needed.
If movement takes an animal across the boundary of the arena (buffered area) in sim.popn
the component "edgemethod" comes into play. By default, locations are toroidally wrapped i.e. the animal re-joins the population on the opposing edge. Other options are “clip” (discard), “clipandreplace” (assign new identity at original location), “stop” (stop just inside the boundary), “reflect” (bounce off edges to the limit of the dispersal), “normalize” = “truncate” (truncate kernel and scale probability to 1.0) and “none" (allow centres outside the buffered area). The “normalize” option (new in secr 4.3.3) can take longer as it repeatedly relocates each individual until its destination lies within the bounding box, up to a maximum of 500 attempts.
Clark, J. S, Silman, M., Kern, R., Macklin, E. and HilleRisLambers, J. (1999) Seed dispersal near and far: patterns across temperate and tropical forests. Ecology 80, 1475–1494.
Nathan , R., Klein, E., Robledo-Arnuncio, J. J. and Revilla, E. (2012) Dispersal kernels: a review. In: J Clobert et al. Dispersal Ecology and Evolution. Oxford University Press. Pp 187–210.
par (mfrow = c(2,3), mar = c(1,1,1,1)) ## birth and death only grid <- make.grid(nx = 7, ny = 4, detector = 'proximity', spacing = 10) pop <- sim.popn (Nbuffer = 100, core = grid, nsessions = 6, details = list(lambda = 0.8, phi = 0.6)) sapply(pop, nrow) ## how many individuals? plot(pop) ## movement only pop2 <- sim.popn (Nbuffer = 100, core = grid, nsessions = 6, details = list(lambda = 1, phi = 1, movemodel = 'normal', move.a = 10, edgemethod = "wrap")) pop3 <- sim.popn (Nbuffer = 100, core = grid, nsessions = 6, details = list(lambda = 1, phi = 1, movemodel = 'normal', move.a = 10, edgemethod = "clip")) pop4 <- sim.popn (Nbuffer = 100, core = grid, nsessions = 10, details = list(lambda = 1, phi = 1, movemodel = 'normal', move.a = 10, edgemethod = "stop")) sapply(pop2, nrow) ## how many individuals? plot(pop2) ## show effect of edgemethod -- ## first session blue, last session red cols <- c('blue',rep('white',4),'red') par (mfrow=c(1,2)) plot(pop2, collapse = TRUE, seqcol = cols) plot(pop3, collapse = TRUE, seqcol = cols) ## zero-inflated movement ## move.b is zero-inflation probability pop5 <- sim.popn (Nbuffer = 1000, core = grid, nsessions = 6, details = list(lambda = 1, phi = 1, movemodel = 'RDEzi', move.a = 50, move.b = 0.5, edgemethod = "none")) mean(do.call(rbind,extractMoves(pop5))$d) # approx 50 * 0.5
par (mfrow = c(2,3), mar = c(1,1,1,1)) ## birth and death only grid <- make.grid(nx = 7, ny = 4, detector = 'proximity', spacing = 10) pop <- sim.popn (Nbuffer = 100, core = grid, nsessions = 6, details = list(lambda = 0.8, phi = 0.6)) sapply(pop, nrow) ## how many individuals? plot(pop) ## movement only pop2 <- sim.popn (Nbuffer = 100, core = grid, nsessions = 6, details = list(lambda = 1, phi = 1, movemodel = 'normal', move.a = 10, edgemethod = "wrap")) pop3 <- sim.popn (Nbuffer = 100, core = grid, nsessions = 6, details = list(lambda = 1, phi = 1, movemodel = 'normal', move.a = 10, edgemethod = "clip")) pop4 <- sim.popn (Nbuffer = 100, core = grid, nsessions = 10, details = list(lambda = 1, phi = 1, movemodel = 'normal', move.a = 10, edgemethod = "stop")) sapply(pop2, nrow) ## how many individuals? plot(pop2) ## show effect of edgemethod -- ## first session blue, last session red cols <- c('blue',rep('white',4),'red') par (mfrow=c(1,2)) plot(pop2, collapse = TRUE, seqcol = cols) plot(pop3, collapse = TRUE, seqcol = cols) ## zero-inflated movement ## move.b is zero-inflation probability pop5 <- sim.popn (Nbuffer = 1000, core = grid, nsessions = 6, details = list(lambda = 1, phi = 1, movemodel = 'RDEzi', move.a = 50, move.b = 0.5, edgemethod = "none")) mean(do.call(rbind,extractMoves(pop5))$d) # approx 50 * 0.5
Before version 3.0, the internal data format for data from exclusive detectors (single, multi, proximityX, transectX) was a matrix with one row per detected animal and one column per sampling occasion; each cell was either zero or the number of the detector at which the animal was detected (with switched sign if the animal died). The format for data from proximity and other detectors was a 3-dimensional array (third dimension corresponding to detectors) that allowed more than one detection per animal per occasion.
From secr 3.0 all capthist data use the 3-D format internally. This simplifies a lot of the coding, and enables mixing of detector types within a session. The constraint that only one detection is allowed per animal per occasion at exclusive detectors is imposed by verify().
The data input functions (read.capthist etc.) automatically generate objects in the new format. Objects created and saved under earlier versions should be converted if they relate to the ‘exclusive’ detector types listed above.
updateCH(object)
updateCH(object)
object |
capthist object |
The function reduce.capthist is applied with the nominal detector type as the outputdetector.
Object with same class as the input.
Updating has the side effect of discarding invalid supernumerary detections (e.g. if there were two detections of an animal on one occasion, only one will be included).
# if we had the old ovenCH ! sapply(ovenCH, dim) sapply(updateCH(ovenCH), dim)
# if we had the old ovenCH ! sapply(ovenCH, dim) sapply(updateCH(ovenCH), dim)
Extract or replace usage (effort) information of a traps
object
(optional).
## S3 method for class 'traps' usage(object, noccasions = NULL, ...) usage(object) <- value
## S3 method for class 'traps' usage(object, noccasions = NULL, ...) usage(object) <- value
object |
|
noccasions |
integer number of occasions (optional) |
value |
numeric matrix of detectors x occasions |
... |
other arguments (not used) |
In secr versions before 2.5.0, usage was defined as a binary value
(1 if trap used on occasion
, zero otherwise).
In later versions, usage may take nonnegative real values and will be interpreted as effort. This corresponds to the constant T_s used for the duration of sampling by Borchers and Efford (2008). Effort is modelled as a known linear coefficient of detection probability on the hazard scale (secr-varyingeffort.pdf; Efford et al. 2013).
For replacement of usage, various forms are possible for value
:
- a matrix in which the number of rows of value
exactly
matches the number of traps K in object
- a vector of two values, the usage (typically 1) and the number of occasions S (a K x S matrix will be filled with the first value)
- a vector of R+1 values where R is the number of sessions in a multi-session object and elements 2..R+1 correspond to the numbers of occasions S1, S2,... in each session
- the usage only (typically 1) (only works when replacing an existing usage matrix with known number of occasions).
usage(object) returns the usage matrix of the traps
object. usage(object)
may be NULL. If noccasions
is provided
and there is no pre-existing matrix then a matrix of all ones will be generated.
At present, assignment of usage to the traps objects of a multisession capthist object results in the loss of session names from the latter.
Efford, M. G., Borchers D. L. and Mowat, G. (2013) Varying effort in capture–recapture studies. Methods in Ecology and Evolution 4, 629–636.
traps
,
usagePlot
,
read.capthist
,
addSightings
demo.traps <- make.grid(nx = 6, ny = 8) ## random usage over 5 occasions usage(demo.traps) <- matrix (sample(0:1, 48*5, replace = TRUE, p = c(0.5,0.5)), nc = 5) usage(demo.traps) summary(demo.traps) usage(traps(ovenCH)) <- c(1,9,10,10,10,10) ## restore lost names names(ovenCH) <- 2005:2009
demo.traps <- make.grid(nx = 6, ny = 8) ## random usage over 5 occasions usage(demo.traps) <- matrix (sample(0:1, 48*5, replace = TRUE, p = c(0.5,0.5)), nc = 5) usage(demo.traps) summary(demo.traps) usage(traps(ovenCH)) <- c(1,9,10,10,10,10) ## restore lost names names(ovenCH) <- 2005:2009
usagePlot
displays variation in effort (usage) over detectors as a
bubble plot (circles with radius scaled so that area is proportional to
effort).
sightingPlot
displays spatial variation in the number of sightings at each detector as a
bubble plot (circles with radius scaled so that area is proportional to either the average number per occasion or the total over occasions.
usagePlot(object, add = FALSE, occasions = NULL, col = "black", fill = FALSE, scale = 2, metres = TRUE, rad = 5, ...) sightingPlot(object, type = c("Detections", "Tu", "Tm", "Tn"), add = FALSE, occasions = "ALL", mean = TRUE, col = "black", fill = FALSE, scale = 2, metres = TRUE, dropunused = TRUE, title = type, legend = c(1, 2, 4, 8), px = 0.95, py = 0.95, ...)
usagePlot(object, add = FALSE, occasions = NULL, col = "black", fill = FALSE, scale = 2, metres = TRUE, rad = 5, ...) sightingPlot(object, type = c("Detections", "Tu", "Tm", "Tn"), add = FALSE, occasions = "ALL", mean = TRUE, col = "black", fill = FALSE, scale = 2, metres = TRUE, dropunused = TRUE, title = type, legend = c(1, 2, 4, 8), px = 0.95, py = 0.95, ...)
object |
traps object with usage attribute |
add |
logical; if FALSE plot.traps is called to create a base plot |
occasions |
integer number(s) of the occasion(s) for which effort is plotted, "ALL", or NULL |
col |
character or integer colour value |
fill |
logical; if TRUE the circle is filled with the line colour |
scale |
numeric value used to scale radius |
metres |
logical; if TRUE scale is a value in metres (see Details) |
rad |
numeric; radial displacement of symbol centre for each occasion from true detector location (metres) |
... |
other arguments passed to plot.traps |
type |
character to choose among sighting types and detections of marked animals |
mean |
logical; if TRUE then the plotted value is the average over occasions, otherwise the sum |
dropunused |
logical; if TRUE then detectors are omitted when they were unused on |
title |
character |
legend |
numeric values for which legend circles will be drawn |
px |
legend x position as fraction of user coordinates |
py |
legend y position as fraction of user coordinates |
The behaviour of usagePlot
is described first. By default (occasion = NULL
) circles representing usage on each
occasion are plotted around the detector location at distance
rad
, as in the petal plot of
plot.capthist
. Otherwise, the usage on a single
specified occasion, or summed over occasions
(length(occasion)>1
, or occasion = "ALL"
), is plotted as
a circle centred at the detector location.
Package sp provides an alternative to usagePlot
(see Examples).
sightingPlot
may be used to display either detections of marked animals (whether or not occasions
refers to sighting occasions) or any of the sighting attributes (unmarked sightings ‘Tu’, marked, unidentified sightings ‘Tm’, or other uncertain sightings ‘Tn’).
If py
is of length 2 then the values determine the vertical spread of symbols in the legend.
For both functions –
The metres
argument switches between two methods. If metres
= TRUE
, the symbols
function is used with inches = FALSE
to plot
circles with radius scaled in the units of object
(i.e. metres;
scale
is then the radius in metres of the symbol for a detector
with usage = 1.0). Otherwise, plotting uses points
; this has the
advantage of producing better filled circles, but a suitable value of
scale must be found by trial and error.
No value is returned by usagePlot
.
sightingPlot
invisibly returns a ‘traps’ object with a covariate ‘f’ holding the plotted values.
usage
, symbols
, bubble
,
sightings
simgrid <- make.grid(nx = 10, ny = 10, detector = "proximity") usage(simgrid) <- matrix(rep(1:10, 50), nrow = 100, ncol = 5) usagePlot(simgrid, border = 20, scale = 1.5, fill = FALSE, metres = FALSE) # It is hard to get the legend just right # here is one attempt legend (x = -50, y = 185, legend = c(1,2,5,10), pch = 1, pt.cex = c(1,2,5,10)^0.5 * 1.5, x.intersp = 3, y.intersp = 1.8, adj = 1, bty = "n", title = "Usage") usagePlot(simgrid, occasion = NULL, border = 20, scale = 1.5, fill = FALSE, metres = FALSE) ## Not run: # bubble plot in package 'sp' library(sp) simgrid$usage <- usage(simgrid)[,1] ## occasion 1 class(simgrid) <- "data.frame" coordinates(simgrid) <- c("x","y") bubble(simgrid) ## End(Not run)
simgrid <- make.grid(nx = 10, ny = 10, detector = "proximity") usage(simgrid) <- matrix(rep(1:10, 50), nrow = 100, ncol = 5) usagePlot(simgrid, border = 20, scale = 1.5, fill = FALSE, metres = FALSE) # It is hard to get the legend just right # here is one attempt legend (x = -50, y = 185, legend = c(1,2,5,10), pch = 1, pt.cex = c(1,2,5,10)^0.5 * 1.5, x.intersp = 3, y.intersp = 1.8, adj = 1, bty = "n", title = "Usage") usagePlot(simgrid, occasion = NULL, border = 20, scale = 1.5, fill = FALSE, metres = FALSE) ## Not run: # bubble plot in package 'sp' library(sp) simgrid$usage <- usage(simgrid)[,1] ## occasion 1 class(simgrid) <- "data.frame" coordinates(simgrid) <- c("x","y") bubble(simgrid) ## End(Not run)
Non-Euclidean distances have a variety of uses, some obscure. You
probably do not need them unless you have data from linear habitats,
covered in the forthcoming package secrlinear. On the other hand,
they open up some intriguing possibilities for the advanced user. The
key is to provide an appropriate value for the component ‘userdist’ of
the details
argument of secr.fit
.
details$userdist
is either a function to compute distances
between detectors and mask points, or a pre-computed matrix of such
distances. Pre-computing assumes the matrix is static (i.e. fixed and
not dependent on any estimated coefficients). The functions
edist
and nedist
are useful for computing
static matrices of Euclidean or non-Euclidean distances (the latter is
useful when there are barriers to movement).
If details$userdist
is a function then it should take the form
userdist(xy1, xy2, mask)
xy1 |
2-column matrix of x-y coordinates of |
xy2 |
2-column matrix of x-y coordinates of |
mask |
habitat mask defining a non-Euclidean habitat geometry |
The matrix returned by the function must have exactly rows and
columns. The function name may be almost anything you like.
The non-Euclidean habitat geometry may or may not require access to local density (D), local (mask) covariates, and the estimation of additional coefficients (beta variables). In order that secr.fit can assemble these data, there is a mechanism for the user to indicate which, if any, variables are required: when called with no arguments the function should return a character vector of variable names. These may include covariates of ‘mask’, the dynamically computed density 'D', and a special real parameter ‘noneuc’ for which one or more coefficients will be fitted.
‘noneuc’ is like 'D' in that it may be modelled as a function of any mask covariates, session, Session, x, y, etc. The actual meaning attributed to ‘noneuc’ depends entirely on how it is used inside the function.
The function may require no variables and not require estimation of additional coefficients. This is the case for a simple linear geometry as described in documentation for the package ‘secrlinear’.
Value | Interpretation |
'' | no covariates etc. required |
'D' | density at each mask point |
'noneuc' | a multi-purpose real parameter |
defined for each mask point | |
c('D', 'noneuc') | both of the preceding |
c('noneuc','habclass') | both noneuc and the mask covariate 'habclass' |
The last case does not estimate a coefficient for habclass, it merely makes the raw value available to whatever algorithm you implement.
The ‘xy2’ and ‘mask’ parameters of the userdist function overlap in practice: xy1 and xy2 only define the points between which distances are required, whereas mask is a carrier for any and all additional information needed by the algorithm.
Full documentation of the secr capability for non-Euclidean distances is in the separate document secr-noneuclidean.pdf, which includes example code for the analysis of Sutherland et al. (2015).
User-specified distances are compatible with some but not all features of secr. Functions with a ‘userdist’ argument are certainly compatible, and others may be.
With a static userdist, region.N
will generally not calculate population size for a region other than the original mask. If you want to supply a new mask in the ‘region’ argument, replace x$details$userdist with a distance matrix appropriate to the new mask, where ‘x’ is the name of the fitted model.
User-specified distances cannot be used with polygon or transect detectors.
When using sim.capthist
to simulate detections of a new
population from sim.popn
you must provide userdist
as a function rather than a matrix. This is because new animals are not
restricted to locations on the ‘mask’ grid.
Sutherland, C., Fuller, A. K. and Royle, J. A. (2015) Modelling non-Euclidean movement and landscape connectivity in highly structured ecological networks. Methods in Ecology and Evolution 6, 169–177.
## see secr-noneuclidean.pdf
## see secr-noneuclidean.pdf
Minor functions.
getMeanSD(xy) maskarea(mask, sessnum = 1) masklength(mask, sessnum = 1) masksize(mask, sessnum = 1) edist(xy1, xy2) nedist(xy1, xy2, mask, inf = Inf, ...) rlnormCV(n, mean, cv)
getMeanSD(xy) maskarea(mask, sessnum = 1) masklength(mask, sessnum = 1) masksize(mask, sessnum = 1) edist(xy1, xy2) nedist(xy1, xy2, mask, inf = Inf, ...) rlnormCV(n, mean, cv)
xy |
2-column matrix or dataframe |
xy1 |
2-column matrix or dataframe |
xy2 |
2-column matrix or dataframe |
mask |
mask or linearmask object |
sessnum |
integer; for multi-session masks, the number of the session |
inf |
numeric value to use for +infinity |
... |
other arguments for |
n |
number of observations |
mean |
mean on natural scale |
cv |
coefficient of variation on natural scale |
getmeanSD
is used by make.mask
to standardize
mask coordinates.
For masklength
the input should be a linear mask from secrlinear.
edist
computes the Euclidean distance between each point in xy1
and each point in xy2. (This duplicates the functionality of ‘rdist’
in package fields).
nedist
computes the non-Euclidean distance between each point
in xy1 and each point in xy2, in two dimensions. The calculation uses
gdistance (van Etten 2017; see also Csardi & Nepusz 2006): a
transition layer is formed representing the connections between
adjacent points in mask
. By default, points within a 16-point
neighbourhood are considered ‘adjacent’. Distances are obtained by
Dijkstra's (1959) algorithm as least cost paths through the graph of
all points in the mask.
nedist
has some subtle options. If ‘mask’ is missing then the
transition layer will be formed from ‘xy2’. If ‘mask’ has a covariate
named ‘noneuc’ then this will be used to weight distances. The ...
argument of nedist
allows the user to vary arguments of
transition
(defaults transitionFunction =
mean and directions = 16). Be warned this can lead to unexpected
results! Point pairs that are completely separated receive the
distance +Inf unless a finite value is provided for the argument
‘inf’. See
secr-noneuclidean.pdf
for uses of nedist
.
rlnormCV
is a wrapper for rlnorm
that computes its
meanlog and sdlog arguments from the mean and CV on the natural scale:
, and
.
For getMeanSD
, a dataframe with columns ‘x’ and ‘y’ and two
rows, mean and SD.
For maskarea
, the summed area of mask cells in hectares (ha).
For masklength
, the summed length of mask cells in kilometers (km).
For masksize
, whichever of area or length is appropriate.
For edist
and nedist
, a matrix with dim = c(nrow(xy1), nrow(xy2)).
For rlnormCV
a vector of random deviates.
Dijkstra, E. W. (1959) A note on two problems in connexion with graphs. Numerische Mathematik, 1, 269–271.
Csardi, G. and Nepusz, T. (2006) The igraph software package for complex network research. InterJournal, 1695. https://igraph.org
van Etten, J. (2017) R package gdistance: Distances and routes on geographical grids. Journal of Statistical Software, 76(1), 1–21. doi:10.18637/jss.v076.i13
getMeanSD(possummask)
getMeanSD(possummask)
Variance-covariance matrix of beta or real parameters from fitted secr model.
## S3 method for class 'secr' vcov(object, realnames = NULL, newdata = NULL, byrow = FALSE, ...)
## S3 method for class 'secr' vcov(object, realnames = NULL, newdata = NULL, byrow = FALSE, ...)
object |
secr object output from the function |
realnames |
vector of character strings for names of ‘real’ parameters |
newdata |
dataframe of predictor values |
byrow |
logical for whether to compute covariances among ‘real’ parameters for each row of new data, or among rows for each real parameter |
... |
other arguments (not used) |
By default, returns the matrix of variances and covariances among the estimated model coefficients (beta parameters).
If realnames
and newdata
are specified, the result is
either a matrix of variances and covariances for each ‘real’ parameter
among the points in predictor-space given by the rows of newdata
or among real parameters for each row of newdata
. Failure to
specify newdata
results in a list of variances only.
A matrix containing the variances and covariances among beta parameters
on the respective link scales, or a list of among-parameter variance-covariance
matrices, one for each row of newdata
, or a list of among-row variance-covariance
matrices, one for each ‘real’ parameter.
## previously fitted secr model vcov(secrdemo.0)
## previously fitted secr model vcov(secrdemo.0)
Check that the data and attributes of an object are internally consistent to avoid crashing functions such as secr.fit
## Default S3 method: verify(object, report, ...) ## S3 method for class 'traps' verify(object, report = 2, ...) ## S3 method for class 'capthist' verify(object, report = 2, tol = 0.01, ...) ## S3 method for class 'mask' verify(object, report = 2, ...)
## Default S3 method: verify(object, report, ...) ## S3 method for class 'traps' verify(object, report = 2, ...) ## S3 method for class 'capthist' verify(object, report = 2, tol = 0.01, ...) ## S3 method for class 'mask' verify(object, report = 2, ...)
object |
an object of class ‘traps’, ‘capthist’ or ‘mask’ |
report |
integer code for level of reporting to the console. 0 = no report, 1 = errors only, 2 = full. |
tol |
numeric tolerance for deviations from transect line (m) |
... |
other arguments (not used) |
Checks are performed specific to the class of ‘object’. The default method is called when no specific method is available (i.e. class not ‘traps’, ‘capthist’ or ‘mask’), and does not perform any checks.
verify.capthist
No ‘traps’ component
Invalid ‘traps’ component reported by verify.traps
No live detections
Missing values not allowed in capthist
Live detection(s) after reported dead
Empty detection histories (except concurrent telemetry and all-sighting data)
More than one capture in single-catch trap(s)
More than one detection per detector per occasion at proximity detector(s)
Signal detector signal(s) less than threshold or invalid threshold
Number of rows in ‘traps’ object not compatible with reported detections
Number of rows in dataframe of individual covariates differs from capthist
Number of occasions in usage matrix differs from capthist
Detections at unused detectors
Number of coordinates does not match number of detections (‘polygon’, ‘polygonX’, ‘transect’ or ‘transectX’ detectors)
Coordinates of detection(s) outside polygons (‘polygon’ or ‘polygonX’ detectors)
Coordinates of detection(s) do not lie on any transect (‘transect’ or ‘transectX’ detectors)
Row names (animal identifiers) not unique
Levels of factor covariate(s) differ between sessions
verify.traps
Missing detector coordinates not allowed
Number of rows in dataframe of detector covariates differs from expected
Number of detectors in usage matrix differs from expected
Occasions with no used detectors
Polygons overlap
Polygons concave east-west (‘polygon’ detectors)
PolyID missing or not factor
Polygon detector is concave in east-west direction
Levels of factor trap covariate(s) differ between sessions
verify.mask
Valid x and y coordinates
Number of rows in covariates dataframe differs from expected
Levels of factor mask covariate(s) differ between sessions
Earlier errors may mask later errors: fix & re-run.
A list with the component errors
, a logical value indicating
whether any errors were found. If object
contains multi-session
data then session-specific results are contained in a further list
component bysession
.
Full reporting is the same as ‘errors only’ except that a message is posted when no errors are found.
capthist
, secr.fit
, shareFactorLevels
verify(captdata) ## create null (complete) usage matrix, and mess it up temptraps <- make.grid() usage(temptraps) <- matrix(1, nr = nrow(temptraps), nc = 5) usage(temptraps)[,5] <- 0 verify (temptraps) ## create mask, and mess it up tempmask <- make.mask(temptraps) verify(tempmask) tempmask[1,1] <- NA verify(tempmask)
verify(captdata) ## create null (complete) usage matrix, and mess it up temptraps <- make.grid() usage(temptraps) <- matrix(1, nr = nrow(temptraps), nc = 5) usage(temptraps)[,5] <- 0 verify (temptraps) ## create mask, and mess it up tempmask <- make.mask(temptraps) verify(tempmask) tempmask[1,1] <- NA verify(tempmask)
Export detections or detector layout or mask to a text file in format suitable for input to DENSITY.
write.captures(object, file = "", deblank = TRUE, header = TRUE, append = FALSE, sess = "1", ndec = 2, covariates = FALSE, tonumeric = TRUE, ...) write.traps(object, file = "", deblank = TRUE, header = TRUE, ndec = 2, covariates = FALSE, ...) write.mask(object, file = "", header = TRUE, ndec = 2, covariates = TRUE, ...)
write.captures(object, file = "", deblank = TRUE, header = TRUE, append = FALSE, sess = "1", ndec = 2, covariates = FALSE, tonumeric = TRUE, ...) write.traps(object, file = "", deblank = TRUE, header = TRUE, ndec = 2, covariates = FALSE, ...) write.mask(object, file = "", header = TRUE, ndec = 2, covariates = TRUE, ...)
object |
|
file |
character name of output file |
deblank |
logical; if TRUE remove any blanks from character string used to identify detectors |
header |
logical; if TRUE output descriptive header |
append |
logical; if TRUE output is appended to an existing file |
sess |
character session identifier |
ndec |
number of digits after decimal point for x,y coordinates |
covariates |
logical or a character vector of covariates to export |
tonumeric |
logical for whether factor and character covariates should be converted to numeric values on output |
... |
other arguments passed to |
Existing file will be replaced without warning if append =
FALSE
. In the case of a multi-session capthist file, session names
are taken from object
rather than sess
.
write.capthist
is generally simpler to use if you want to export
both the capture data and trap layout from a capthist
object.
By default individual covariates are not exported. When exported they are repeated for each detection of an individual. Factor covariates are coerced to numeric before export.
For write.mask
, header = TRUE
also causes column names to be exposed.
None
write.captures (captdata)
write.captures (captdata)
Upload a set of point locations as waypoints to a GPS unit connected
by USB or via a serial port. Intended primarily for detector locations
in a traps object. Uses the GPSBabel package which must have been
installed. Coordinates are first inverse-projected to latitude and
longitude using function st_transform
from sf.
writeGPS(xy, o = "garmin", F = "usb:", proj = "+proj=nzmg")
writeGPS(xy, o = "garmin", F = "usb:", proj = "+proj=nzmg")
xy |
2-column matrix or dataframe of x-y coordinates |
o |
character output format (see GPSBabel documentation) |
F |
character for destination (see Details) |
proj |
character string describing projection |
This function is derived in part from readGPS
in maptools.
For users of Garmin GPS units, useful values of o
are "garmin"
for direct upload via USB or serial ports, and "gdb" for a file in
Mapsource database format.
F
may be "usb:" or "com4:" etc. for upload via USB or serial
ports, or the name of a file to create.
The proj
argument may be complex. For further information see the
Examples and the vignette
secr-spatialdata.pdf.
If proj
is an empty string then coordinates are assumed already to
be latitudes (column 1) and longitudes (column 2).
Waypoint names are derived from the rownames of xy
.
No value is returned. The effect is to upload waypoints to an attached GPS or file.
GPSBabel is available free online. Remember to add it to the path. On Windows this means following something like Settings > Control panel > System > Advanced settings > Environment variables > (select Path) Edit and adding ";C:/Program Files (x86)/gpsbabel" to the end (without the quotes).
## Example using shapefile "possumarea.shp" in ## "extdata" folder. As 'cluster' is not specified, ## the grid comprises single multi-catch detectors. ## Not run: ## test for availability of GPSBabel if (nzchar(Sys.which("gpsbabel"))) { library(sf) shpfilename <- system.file("extdata/possumarea.shp", package = "secr") possumarea <- st_read(shpfilename) possumgrid <- make.systematic(spacing = 100, region = possumarea, plt = TRUE) ## May upload directly to GPS... # writeGPS(possumgrid, proj = "+proj=nzmg") ## ...or save as Mapsource file writeGPS(possumgrid, o = "gdb", F = "tempgrid.gdb", proj = "+proj=nzmg") ## If `region' had been specified in another projection we ## would need to specify this as in Proj.4. Here is a ## hypothetical example for New Zealand Transverse Mercator ## with datum NZGD2000 (EPSG:2193) NZTM <- paste("+proj=tmerc +lat_0=0 +lon_0=173 +k=0.9996", "+x_0=1600000 +y_0=10000000 +ellps=GRS80", " +towgs84=0,0,0,0,0,0,0 +units=m +no_defs") # writeGPS(possumgridNZTM, o = "gdb", F = "tempNZTM.txt", # proj = NZTM) ## Or to upload coordinates from UTM Zone 18 in eastern ## Maryland, USA... # writeGPS(MarylandUTMgrid, proj = # "+proj=utm +zone=18 +ellps=WGS84") } ## End(Not run)
## Example using shapefile "possumarea.shp" in ## "extdata" folder. As 'cluster' is not specified, ## the grid comprises single multi-catch detectors. ## Not run: ## test for availability of GPSBabel if (nzchar(Sys.which("gpsbabel"))) { library(sf) shpfilename <- system.file("extdata/possumarea.shp", package = "secr") possumarea <- st_read(shpfilename) possumgrid <- make.systematic(spacing = 100, region = possumarea, plt = TRUE) ## May upload directly to GPS... # writeGPS(possumgrid, proj = "+proj=nzmg") ## ...or save as Mapsource file writeGPS(possumgrid, o = "gdb", F = "tempgrid.gdb", proj = "+proj=nzmg") ## If `region' had been specified in another projection we ## would need to specify this as in Proj.4. Here is a ## hypothetical example for New Zealand Transverse Mercator ## with datum NZGD2000 (EPSG:2193) NZTM <- paste("+proj=tmerc +lat_0=0 +lon_0=173 +k=0.9996", "+x_0=1600000 +y_0=10000000 +ellps=GRS80", " +towgs84=0,0,0,0,0,0,0 +units=m +no_defs") # writeGPS(possumgridNZTM, o = "gdb", F = "tempNZTM.txt", # proj = NZTM) ## Or to upload coordinates from UTM Zone 18 in eastern ## Maryland, USA... # writeGPS(MarylandUTMgrid, proj = # "+proj=utm +zone=18 +ellps=WGS84") } ## End(Not run)