Abstract
Wildlife populations often exhibit unequal catchability between subgroups such as males and females. This heterogeneity of capture probabilities can bias both population size and sex ratio estimates. Several authors have suggested that this problem can be overcome by treating males and females as separate populations and calculating a population estimate for each of them. However, this suggestion has received little testing, and many researchers do not implement it. Therefore, we used two simulations to test the utility of this method. One simulated a closed population, while the other simulated an open population and used the robust design to calculate population sizes. We tested both simulations with multiple levels of heterogeneity, and we used a third simulation to test several methods for detecting heterogeneity of capture probabilities. We found that treating males and females as separate populations produced more accurate population and sex ratio estimates. The benefits of this method were particularly pronounced for sex ratio estimates. When males and females were included as a single population, the sex ratio estimates became inaccurate when even slight heterogeneity was present, but when males and females were treated separately, the estimates were accurate even when large biases were present. Nevertheless, treating males and females separately reduced precision, and this method may not be appropriate when capture and recapture rates are low. None of the methods for detecting heterogeneity were robust, and we do not recommend that researchers rely on them. Rather, we suggest separating populations by sex, age, or other subgroups whenever sample sizes permit.