secr (version 2.5.0)

rbind.capthist: Combine capthist Objects

Description

Form a single capthist object from two or more compatible capthist objects.

Usage

MS.capthist(...)
  rbind.capthist(..., renumber = TRUE, pool = NULL, verify = TRUE)

Arguments

...

one or more capthist objects or lists of capthist objects

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 verify

Value

• 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 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.

Details

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. rbind.capthist 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). For rbind.capthist, the ...argument may be

1. A series of single-session capthist objects, which are pooled to form one new single-session object, or
2. 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. Ifpool = NULL(the default) then all the sessions are pooled to form one single-session capthist object.

Sessions to be pooled must have the same number of capture occasions and use the same detectors (traps). At present there is no function to pool data from different detector arrays. 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 and rbind.capthist look like S3 methods, they aren't. The full function names must be used.

See Also

capthist, subset.capthist

Examples

## 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.capthist(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

## compare two different model fits 
tempfit3 <- secr.fit(temp3, CL = T, 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 = T, buffer = 150, model = list
    (g0 ~ range.size, sigma ~ range.size), trace = FALSE)
predict(tempfit4, newdata = data.frame(range.size = c("large",
    "small")))