Online resources to accompany the McCrea & Morgan text
This short course will outline recent research developments undertaken by ecological statisticians from the University of Kent who collaborate closely with members of DICE. The presentations are intended for quantitative ecologists.
The four main themes which will be covered are stopover capture-recapture modelling, mixture modelling, N-mixture modelling and removal modelling.
These extensions of basic capture-recapture and modelling approaches for count data have wide application in ecology, and the theory will be applied to data on butterflies, reptiles and birds. Practical aspects of how to use this cutting-edge methodology will be provided through computer demonstrations.
Introductory Slides are available here: Introduction
Classic capture-recapture models for estimation of population size assume that a population is closed, so that there are no arrivals or departures of individuals during the study period. For some species, such as migratory birds, this is not a realistic assumption. Stopover models generalise the traditional Jolly-Seber model through the estimation of arrival and retention probabilities of individuals within the study. The retention probabilities can also depend on the duration of stay which has particular value for studies at breeding sites for migratory species.
Stopover modelling slides are available here: Stopover
Matlab code for stopover models is available here: stopover.m
The standard model for removal survey data results in a simple geometric distribution for the resulting counts over time. However, some taxa - such as reptiles and amphibians - yield count data that require far more complex modelling. We examine two extensions of removal models: (i) the incorporation of climatic covariates to account for variability in capture probability and (ii) a flexible framework which allows new cohorts of individuals to arrive at the population during the depletion study.
Removal modelling slides are available here: Removal
Many ecological data require description by mixture models, eg., the counts of migrant birds at Dungeness bird observatory, and the national counts of bivoltine butterflies, such as the Common blue. Here we illustrate recent work in the latter case. A simple model is found to provide a convenient description of butterfly count data over sites and years. The model is compared with a stopover alternative as well as the use of Generalised Additive Models for producing temporal indices for individual species and also an indication of the effect of global warming on phenology. A computer demonstration reveals the speed of analysis.
Mixture modelling slides are available here: Mixture
The N-mixture model is widely used to estimate the abundance of a population in the presence of unknown detection probability, from a set of counts subject to spatial and temporal replication. Research at Kent has shown the equivalence of N-mixture and multivariate Poisson and negative binomial models, which leads to powerful new computational methods. It is shown that when detection probability and the number of sampling occasions are small, infinite estimates of abundance can arise. The methods are illustrated using data on Hermann's tortoise.
N-mixture modelling slides are available here: N-mixture