• Version
• Download 10
• File Size 723.03 KB
• Create Date 1 January 2001
• Download

The accuracy of population estimates derived from mark-recapture sampling will be compromised when marked animals are more or less likely to be recaptured than unmarked animals. We used a 'test' population of radiocollared black bears (Ursus americanus) to identify the sources and extent of sampling biases in trapping, camera-trapping, and hunter harvest. We investigated whether and how sex, age, family status, and percent of time on the study area affected the likelihood of bears (in this test population) being sampled by each of these methods and calculated biases in resulting population estimates. Vulnerability to trapping and camera sampling varied by sex and age; trapping was biased toward adult females without cubs and subadult males (3-5 years old) and against juvenile females (1-2 years old) and adult males. Bears present in the study area >50% of the time were trapped and camera-trapped more often than those that spent less time there. All sampling methods showed bias toward particular individuals, irrespective of sex, age, or time spent in the study area. Bears that were initially radiocollared in dens, without being trapped, were less likely to be trapped in future years than those that were initially radiocollared via trapping. Radiocollared bears trapped or photographed 1 summer were more likely than others to be trapped or photographed again the next summer or to be shot by hunters in the fall. This linkage between the marked and recaptured samples caused population estimates to be biased low. When we treated previously radiocollared bears that were trapped or camera-trapped 1 summer as a 'marked' sample, and bears so sampled the following summer or shot by hunters the following fall as a 'recapture' sample, in 12 of 13 cases we underestimated the known size of the population of radiocollared bears by 12-47%. We discuss ways to reduce bias, but warn that bias is likely inescapable.