Comparison of capture–recapture estimators of snowshoe hare populations

We used two island populations of snowshoe hares (Lepus americanus) in the Kluane Lake area of the Yukon Territory of Canada to evaluate capture – recapture estimators. These islands were intensively sampled, allowing us to enumerate the actual population size. Population size estimates were calculated using the programs CAPTURE and JOLLY, and estimators were compared for bias characteristics. Results from both islands suggest that the CAPTURE heterogeneity models Mh (jackknife), Mh (Chao), and Mth (time – heterogeneity) and the Jolly – Seber model were approximately unbiased. All other CAPTURE models displayed a negative bias. The CAPTURE model selection routine picked estimation models of different biases for each trapping period, an undesirable result. We conclude that it is best to use one robust estimator such as the Mh (jackknife) with snowshoe hare data.