Estimating the abundance of non-solitary animals : the effects of contagion on mark-recapture and removal methods
Date
1996
Authors
Ott, Peter K. (Peter Kurt)
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
The removal method and mark-recapture experiment are two common techniques for estimating animal abundance based on capture data. In this thesis, the closed population estimation procedures for both models (under the assumption of constant catchability for all animals on all sampling occasions) are extended to allow for the possibility that the animals within the population reside in non-solitary groups. This is accomplished two different ways. The first is to construct a log likelihood based on a contagious distribution such as the Polya-Eggenberger or Dirichlet. Each model contains an unknown parameter meant to reflect the degree of contagion caused by animal clustering. The second is to derive quasi-likelihood estimating equations for an overdispersed model. The unknown overdispersion parameter is estimated using two different techniques, and these are used to adjust the confidence intervals of the other parameters. The new models are tested using both real data and the results of simulation (artificial populations of various degrees of clustering). The contagious distribution approach provided conservative estimates of population size, and likelihood profile-based confidence intervals were not always possible due to the shape of the likelihood surface. The quasi-likelihood technique yielded estimates very close to those from the classical model but had slightly wider confidence intervals, which were further broadened when the population was exhibiting significant overdispersion (attributable to clustering). These preliminary results reinforce the general belief that the effect of animal clustering has little effect on the classical estimate of population size but the standard confidence intervals are too narrow and should be wider.