Estimating Squirrel Abundance from Livetrapping Data

Abstract

Estimates of squirrel (Sciurus carolinensis and S. niger) abundance were derived from several methods of population estimation applied to data obtained by livetrapping squirrels on the Waterloo Wildlife Experiment Station in southeastern Ohio, 1962 and 1963. The frequency of capture of marked squirrels suggests that the probability of capture is not the same for all squirrels; as a result a trapped sample typically contains a disproportionately high number of recaptures. Thus, the multiple census methods of Schnabel and of Schumacher produced estimates lower than the number of animals considered to comprise the population. Frequency of capture approximated the geometric distribution. The simplified equation for maximum likelihood estimation (MLE) for the geometric distribution, presented in 1967 by Edwards and Eberhardt, appeared useful for estimating squirrel abundance from livetrapping data, although estimates tended to be somewhat high. The intercept of a line fitted to a logarithmic plot of data on the frequency of capture, using linear regression techniques, gave what appeared to be adequate approximations of the numbers of squirrels in the zero (uncaptured) class. Although estimates derived from MLE for the geometric distribution and from linear regression are based on assumptions not strictly fulfilled by the data, these methods should prove useful until better techniques are developed. MLE for the Poisson distribution appeared to underestimate the zero class. Similarities in results of evaluations of techniques of population estimation for squirrels and rabbits suggest that further research on population estimation may provide findings applicable to a variety of species.