gdistsamp: Fit the generalized distance sampling model of Chandler et al. (2011).

Description

Extends the distance sampling model of Royle et al. (2004) to estimate the probability of being available for detection. Also allows abundance to be modeled using the negative binomial and zero-inflated Poisson distributions.

Usage

gdistsamp(lambdaformula, phiformula, pformula, data, keyfun =
c("halfnorm", "exp", "hazard", "uniform"), output = c("abund",
"density"), unitsOut = c("ha", "kmsq"), mixture = c("P", "NB", "ZIP"), K,
starts, method = "BFGS", se = TRUE, engine=c("C","R"), rel.tol=1e-4, threads=1, ...)

Value

An object of class unmarkedFitGDS.

Arguments

lambdaformula: A right-hand side formula describing the abundance covariates.
phiformula: A right-hand side formula describing the availability covariates.
pformula: A right-hand side formula describing the detection function covariates.
data: An object of class unmarkedFrameGDS
keyfun: One of the following detection functions: "halfnorm", "hazard", "exp", or "uniform." See details.
output: Model either "density" or "abund"
unitsOut: Units of density. Either "ha" or "kmsq" for hectares and square kilometers, respectively.
mixture: Either "P", "NB", or "ZIP" for the Poisson, negative binomial, or zero-inflated Poisson models of abundance.
K: An integer value specifying the upper bound used in the integration.
starts: A numeric vector of starting values for the model parameters.
method: Optimization method used by optim.
se: logical specifying whether or not to compute standard errors.
engine: Either "C" to use fast C++ code or "R" to use native R code during the optimization.
rel.tol: relative accuracy for the integration of the detection function. See integrate. You might try adjusting this if you get an error message related to the integral. Alternatively, try providing different starting values.
threads: Set the number of threads to use for optimization in C++, if OpenMP is available on your system. Increasing the number of threads may speed up optimization in some cases by running the likelihood calculation in parallel. If threads=1 (the default), OpenMP is disabled.
...: Additional arguments to optim, such as lower and upper bounds

Author

Richard Chandler rbchan@uga.edu

Details

Extends the model of Royle et al. (2004) by estimating the probability of being available for detection $\phi$. To estimate this additional parameter, replicate distance sampling data must be collected at each transect. Thus the data are collected at i = 1, 2, ..., R transects on t = 1, 2, ..., T occassions. As with the model of Royle et al. (2004), the detections must be binned into distance classes. These data must be formatted in a matrix with R rows, and JT columns where J is the number of distance classses. See unmarkedFrameGDS for more information about data formatting.

The definition of availability depends on the context. The model is $$M_i \sim \text{Pois}(\lambda)$$ $$N_{i,t} \sim \text{Bin}(M_i, \phi)$$ $$y_{i,1,t}, \dots, y_{i,J,t} \sim \text{Multinomial}(N_{i,t}, \pi_{i,1,t}, \dots, \pi_{i,J,t})$$

If there is no movement, then $M_i$ is local abundance, and $N_{i,t}$ is the number of individuals that are available to be detected. In this case, $\phi=g_0$. Animals might be missed on the transect line because they are difficult to see or detected. This relaxes the assumption of conventional distance sampling that $g_0=1$.

However, when there is movement in the form of temporary emigration, local abundance is $N_{i,t}$; it's the fraction of $M_i$ that are on the plot at time t. In this case, $\phi$ is the temporary emigration parameter, and we need to assume that $g_0=1$ in order to interpret $N_{i,t}$ as local abundance. See Chandler et al. (2011) for an analysis of the model under this form of temporary emigration.

If there is movement and $g_0<1$ then it isn't possible to estimate local abundance at time t. In this case, $M_i$ would be the total number of individuals that ever use plot i (the super-population), and $N_{i,t}$ would be the number available to be detected at time t. Since a fraction of the unavailable individuals could be off the plot, and another fraction could be on the plot, it isn't possible to infer local abundance and density during occasion t.

References

Royle, J. A., D. K. Dawson, and S. Bates. 2004. Modeling abundance effects in distance sampling. Ecology 85:1591-1597.

Chandler, R. B, J. A. Royle, and D. I. King. 2011. Inference about density and temporary emigration in unmarked populations. Ecology 92:1429--1435.

Examples

Run this code



# Simulate some line-transect data

set.seed(36837)

R <- 50 # number of transects
T <- 5  # number of replicates
strip.width <- 50
transect.length <- 100
breaks <- seq(0, 50, by=10)

lambda <- 5 # Abundance
phi <- 0.6  # Availability
sigma <- 30 # Half-normal shape parameter

J <- length(breaks)-1
y <- array(0, c(R, J, T))
for(i in 1:R) {
    M <- rpois(1, lambda) # Individuals within the 1-ha strip
    for(t in 1:T) {
        # Distances from point
        d <- runif(M, 0, strip.width)
        # Detection process
        if(length(d)) {
            cp <- phi*exp(-d^2 / (2 * sigma^2)) # half-normal w/ g(0)<1
            d <- d[rbinom(length(d), 1, cp) == 1]
            y[i,,t] <- table(cut(d, breaks, include.lowest=TRUE))
            }
        }
    }
y <- matrix(y, nrow=R) # convert array to matrix

# Organize data
umf <- unmarkedFrameGDS(y = y, survey="line", unitsIn="m",
    dist.breaks=breaks, tlength=rep(transect.length, R), numPrimary=T)
summary(umf)


# Fit the model
m1 <- gdistsamp(~1, ~1, ~1, umf, output="density", K=50)

summary(m1)


backTransform(m1, type="lambda")
backTransform(m1, type="phi")
backTransform(m1, type="det")

if (FALSE) {
# Empirical Bayes estimates of abundance at each site
re <- ranef(m1)
plot(re, layout=c(10,5), xlim=c(-1, 20))
}

Run the code above in your browser using DataLab