Strategic rat control for restoring populations of native species in forest fragments

Forest fragments have biodiversity value that may be enhanced through management such as control of non-native predators. However, such efforts may be ineffective, and research is needed to ensure that predator control is done strategically. We used Bayesian hierarchical modeling to estimate fragment-specific effects of experimental rat control on a native species targeted for recovery in a New Zealand pastoral landscape. The experiment was a modified BACI (before-after-control-impact) design conducted over 6 years in 19 forest fragments with low-density subpopulations of North Island Robins (Petroica longipes). The aim was to identify individual fragments that not only showed clear benefits of rat control, but also would have a high probability of subpopulation growth even if they were the only fragment managed. We collected data on fecundity, adult and juvenile survival, and juvenile emigration, and modeled the data in an integrated framework to estimate the expected annual growth rate (λ) of each subpopulation with and without rat control. Without emigration, subpopulation growth was estimated as marginal (λ = 0.95-1.05) or negative (λ = 0.74-0.90) without rat control, but it was estimated as positive in all fragments (λ = 1.4-2.1) if rats were controlled. This reflected a 150% average increase in fecundity and 45% average increase in adult female survival. The probability of a juvenile remaining in its natal fragment was 0.37 on average, but varied with fragment connectivity. With juvenile emigration added, 6 fragments were estimated to have a high (>0.8) probability of being self-sustaining (λ > 1) with rat control. The key factors affecting subpopulation growth rates under rat control were low connectivity and stock fencing because these factors were associated with lower juvenile emigration and higher fecundity, respectively. However, there was also substantial random variation in adult survival among fragments, illustrating the importance of hierarchical modeling for fragmentation studies.