An Evaluation of the Removal Method of Estimating Animal Populations

Publisher Website
Google Scholar
Abstract
The removal method assumes that constant trapping effort is applied for k trappings, with removal of captured animals from the population. Also assumed are & stationary population of size N, a common probability, p, of capture during a single trapping for all animals exposed to capture, and that this probability of capture does not change from trapping to trapping. A rapid graphical method of obtaining maximum likelihood estimates of N and p from trapping results was presented, as well as formulas for the standard errors of the estimates. The asymptotic precision of this method of estimation was determined. It was observed that for fixed total trapping effort, only a slight gain in precision results from increasing the number of trappings with decreased effort per trapping. The precision of the method was shown to be poor, e.g. for N of 1000 or less, at least 40% of the total population must be trapped in order to obtain a coefficient of variation of 30% or less. Experimental sampling showed close similarity between the maximum likelihood and regression methods of estimation; also that with p=.4 for three "trappings", fairly close approximation to normality and the asymptotic variance was attained by the distributions of estimates of N=200 and larger. Chi-square tests of the assumption of a constant probability of capture, performed on trapping data collected in the North American Census of Small Mammals, showed significance at the two per cent level.

This publication has 4 references indexed in Scilit: