Comparison of Open and Closed Models for Cotton Rat Population Estimates

Abstract

Population estimates of hispid cotton rats (Sigmodon hispidus) in 2 southern Florida sugarcane fields were obtained from capture-recapture data. A model for population estimation that assumes an open population of individuals with equal probabilities of capture (Jolly-Seber) and a sequence of models (program CAPTURE) that assumes a closed population of individuals with varying capture probabilities yielded similar estimates, despite considerable heterogeneity of capture probabilities over 8-day trapping periods. The trends of the open and closed model estimates and 2 population indices, minimum number known to be alive and captures per 100 trap-nights, were similar: population size in both fields was reduced following harvest, remained low during spring and early summer, and increased in late summer and throughout the fall. Minimum number known to be alive appeared to be a more sensitive index to population fluctuation than captures per 100 trap-nights. J. WILDL. MANAGE. 46(1):156-163 Despite considerable literature on statistical methods of population estimation based on capture-recapture data, many researchers prefer to use indices of abundance or simple estimators such as the Lincoln Index. Indices such as minimum number of animals known to be alive, number of individuals trapped, and captures per 100-trap nights are commonly used. A gap exists between the practical use of various models for population estimation, each with their restrictive assumptions, and an understanding of how capture-recapture data for a given species meet or fail to meet the models' assumptions. Otis et al. (1978), drawing from recent work by Pollock (1974) and Burnham and Overton (1978), developed a number of models that allow unequal capture probabilities in capture-recapture studies, and formulated computer program CAPTURE to test the fit of these models to capture-recapture data, select the most appropriate model for a given data set, and compute the estimate of population size under that model. Such an approach gives the biologist an objective means of learning which assumption(s) the data violate most seriously when using specific models and which model(s), if any, fit the data. Otis et al.'s (1978) models assume a closed population, that is, recruitment, immigration, emigration, and mortality are not allowed, and this may limit their usefulness for many studies. However, when trapping periods are short in relation to the opportunity for the occurrence of such factors, and when the area trapped is sufficiently large, this assumption can be approximately met. Smallmammal studies conducted in a fairly w ll-defined area for a period of a week, for example, may often meet the closure criterion adequately. In contrast to the above closed models, the Jolly-Seber (J-S) model (Jolly 1965; Seber 1965, 1973) is a so-called open population model, allowing the occurrence of immigration, emigration, recruitment, and mortality during the trapping period. The model assumes, however, that capture probabilities vary only by trapping occasion, and thus individual heterogeneity among animals or behavioral response to capture are not permitted. Both the J-S model and the models of Otis et al. (1978) assume geographic 156 J. Wildl. Manage. 46(1):1982 This content downloaded from 207.46.13.103 on Thu, 20 Oct 2016 04:32:10 UTC All use subject to http://about.jstor.org/terms ESTIMATING SIZE OF COTTON RAT POPULATIONS* Lefebvre et al. 157 closure; that is, the concept of a population, occupying a defined geographical area and made up of an absolute number of individuals, is implicit in both types of models. Our study investigates use of program CAPTURE on cotton rat capture-recapture data, and compares estimates obtained with J-S estimates and with the population indices minimum number of animals known to be alive (MNA) and captures per 100 trap-nights. D. Decker, U.S. Fish and Wildlife Service, conducted much of the rat handling and assisted in data preparation. C. Ingram assisted in developing trapping design, N. Shafer assisted in obtaining Jolly-Seber estimates, and W. Braley, T. O'Brien, D. Buecker, P. Gall, W. Maddox, M. Sandsberry, K. Simpson, D. Steffen, and S. Williams assisted in setting traps and data recording. K. Burnham and D. Anderson reviewed an earlier version and made several helpful suggestions.