Estimating impacts of poison operations on non-target species using mark-recapture analysis and simulation modelling: an example with saddlebacks

Poison operations to control or eradicate exotic mammals are a key component of conservation management in the South Pacific. They also result in by-kill of native species. It is therefore important to develop reliable methods for estimating by-kill and assessing its impact. The North Island saddleback (Philesturnus carunculatus rufusater), a rare New Zealand forest bird, was reintroduced to Mokoia Island in 1992, and 4 years later there was an aerial drop of cereal pellets containing Brodifacoum aimed at eradicating mice. We used mark-recapture analysis on resighting data collected from 1992 to 1997 to estimate the by-kill of saddlebacks attributable to this poison drop. We nominated a set of candidate models to explain the data, and compared these using Akaike's Information Criterion. The analysis showed that saddleback survival was substantially lower than expected in the 6-week interval after the poison drop, taking age, density dependence, season and random variation into account. Taking expected survival rates into account, the probability of an adult being killed was estimated to be 0.45 (95% CI=0.34-0.56), and the probability of a juvenile being killed was estimated to be 0.35 (95% CI = -0.05-0.75). We then used a simulation model developed for the population to assess the longer-term impact of this mortality. While the mortality set back the expansion of the population by 1-2 years, we predicted that the population would have recovered fairly quickly and ultimately reach the same carrying capacity. Markrecapture analysis permits precise estimates of background survival rates unconfounded by seasonal or random variation. (C) 2002 Elsevier Science Ltd. All rights reserved.