Invasive ship rats (Rattus rattus) are a major threat to the native species and ecosystems of islands. We used 10 self-resetting traps (A24 rat and stoat traps, Goodnature Ltd., Wellington, NZ), along with existing single kill DOC200 traps at two devices per hectare on a 9.3 hectare island in New Zealand to reduce rat numbers and ideally achieve eradication. Each self-resetting trap was monitored with motion-activated cameras to analyse rat behaviour and A24 kill numbers were documented using Goodnature digital strike counters. The traps were checked on 10 occasions from August 2016 to October 2017. The videos documented initial high rat activity on the island, which reduced over time following initial trapping success. An immediately obvious neophobic response towards the A24 traps was not observed. Rats interacted with the A24 traps within hours after initial deployment and 60% of the traps were triggered in the first night. After three nights, all traps were triggered at least once. While rats interacted with the traps at all times of the year the number of observed trap-triggers was relatively low. High number of interactions resulted in high kill numbers in late spring when population size was increasing and seasonal food abundance had not yet reached its peak. A second peak was observed in late autumn when rat abundance was presumably high. Recruitment of naïve individuals was a probable cause for high kill numbers during the breeding season. In winter, when rat abundance was presumably low, a few individuals were the likely cause for a high number of interactions while kill numbers were low. A knock-down (i.e. suppression from high to low abundance) of rats using both trap types was achieved in the first 100 days. However, kill numbers of A24s declined over time. After the initial suppression, the number of rats killed was insufficient to offset intrinsic population growth and reinvasion from the adjacent coast, thereby preventing eradication.

Lethal trapping of island invasive rodents is a critical practice used by management organizations to protect native biota. Carcass detection from lethal trapping is dependent on the trapping method, carcass palatability, and scavengers present. Goodnature A24 self-resetting rat traps are an effective tool in remote areas and complex terrain because traps can be visited at 4–6 month intervals and produce multiple carcasses during that interval. The goal of this study was to determine whether we are a) underestimating target mortality with carcass counts and b) failing to detect non-target mortality between A24 trap checks at two field sites on the island of Kauai. Both sites have established Goodnature A24 trap grids with 300 A24s deployed by the Kauai Forest Bird Recovery Project (KFBRP), and one site is fenced to exclude invasive ungulates. KFBRP conducts routine trap checks every four months, finding 0–3 rat or mouse carcasses at each trap. We assumed that traps kill more animals than indicated by carcass counts because 75% of traps have counters to record when traps fire, and counter tallies usually exceed carcass counts. In 2018 and 2019, several bird carcasses were found under traps; therefore, we needed to investigate the likelihood that non-target mortality went undetected. In May 2019, we placed 60 carcasses (30 non-native birds and 30 rats) at a fenced site, and 60 carcasses (30 bird, 30 rat) at an unfenced site, in both gulches and uplands. Carcasses were periodically surveyed for 120 days after deployment. The unfenced site had greater removal rates, notably in gulches; 33 of 60 carcasses remained detectable, compared to 52 of 60 carcasses at the fenced site. Bird and rat carcasses did not differ in persistence, and taxon did not affect scavenger preference. These results suggest that significant non-target mortality has not gone undetected in our A24 trap grid because we are likely to detect most target and non-target carcasses after four months in fenced areas, and especially upland unfenced areas. However, we are less likely to detect carcasses in the unfenced gulches where ungulate scavengers are prevalent, and increased monitoring may be needed in such gulches.