| Summary: | Many theoretical capture-recapture estimation models have been introduced into applied field ecology in the last twenty years, but only a few of these models have been tested in field situations. The objective of this thesis was to evaluate capture-recapture estimators when applied to a cyclic snowshoe hare population in the Kluane Lake area in the Southern Yukon. The estimation models and model selection routine of program CAPTURE, and the Jolly-Seberopen model were evaluated. Two independent approaches were used to evaluate estimators: 1) Island populations of known size were used to determine estimator bias and study factors that affect hare capture probabilities, and 2) An individual-based spatial Monte Carlo simulation model was used to evaluate estimator robustness to sample biases caused by a cyclic snowshoe hare population.
Two islands were used for studies of estimator bias. Results from both islands suggest that the CAPTURE heterogeneity models Mh (jackknife), Mh (Chao), and Mth(time/heterogeneity) were approximately unbiased for the island population of hares. All other CAPTURE models displayed a negative bias. The program CAPTURE model selection routine picked models of different bias for each trapping period. Island studies of variation in hare capture probabilities documented a strong relationship between hare movement patterns and capture probabilities on an individual and population level. The strong contribution of sampling factors such as trap placement, and time of sampling in the variation of hare capture probabilities was also documented.
|
|---|
