Capture probability and heterogeneity of trap response in stoats (Mustela erminea)

Abstract

The technique most widely used to control the stoat, an introduced predator in New Zealand, is to set Fenn (kill) traps, usually in lines or (less often) in grids. There has been no analysis of trap response in stoats, nor of the extent of potential variation in probability of capture. We report the results of an analysis of mark–recapture data recorded from stoats observed during a period of high stoat and mouse density in January 1980 in the Eglinton and Hollyford Valleys (northern Fiordland), using livetraps set in lines at one per 400 m over 14 km in each valley. Over 8 days of trapping (1–11 January), 89 stoats were tagged. The daily probability of first capture for all ages, both sexes, was 0.14 (with 95% confidence intervals 0.07–0.25) and of recapture 0.10 (0.07–0.14).

We also analysed a new set of mark–recapture data collected during a period of very low mouse density in the Grebe Valley (southern Fiordland) in December 2000, using 19 live-traps set in a line at one per kilometre over 20 km. In this study 21 adult stoats and no young of year were tagged. The daily probability of first capture for adult males was 0.12 (0.04–0.31), and of recapturing them, 0.15 (0.10–0.23). A month later, in late January 2001, 68 Fenn traps set at four per kilometre caught 48 previously unmarked stoats, plus 12 of the 21 marked stoats released alive.

Heterogeneity in probability of recapture was investigated by taking a longer subset of the 1980 data (1–17 January) and grouping individuals by sex and age. In the best closed-captures models, ranked using AIC_c, first-capture probability was similar for all stoats (0.17 (0.12–0.24)), and evidence of variation in the probability of recapture between age and sex classes was present but weak. The confidence limits around the recapture probabilities for adult males and females overlapped completely. Recapture probabilities for young-of-the-year males remained about the same (0.14 (0.11–0.19)), while the recapture probability of young-of-the-year females halved after first capture (0.07 (0.04–0.11)).

Pledger's finite-mixture models demonstrating individual heterogeneity in trappability produced lower AIC_c values than the closed-captures models partitioning variation in recapture probability by age and sex alone. The observed heterogeneity in trap response is therefore not due only to variable individual response to traps, but is also to opportunity, as might be expected in data collected from a line of traps where the edge effect on trap-encounter rate is high. However, the extent to which trap-encounter rate helps to explain the observed heterogeneity is unknown. Indeed, there may be other sources of individual heterogeneity that are not related to age/sex or to trap-encounter rate, and this is a potential problem for wildlife managers using conventional trap lines to remove stoats to protect native species.