Abstract
House mice are among the most widely distributed mammals in the world, and adversely affect a wide range of indigenous biota. Suppressing mouse populations, however, is difficult and expensive. Cost-effective suppression requires knowing how low to reduce mouse numbers to achieve biodiversity outcomes, but these targets are usually unknown or not based on evidence. We derived density-impact functions (DIFs) for mice and small indigenous fauna in a tussock grass/shrubland ecosystem. We related two indices of mouse abundance to five indices of indigenous lizard and invertebrate abundance measured inside and outside mammal-resistant fences. Eight of 22 DIFs were significantly non-linear, with positive responses of skinks ( Oligosoma maccanni , O. polychroma ) and ground wētā ( Hemiandrus spp.) only where mice were not detected or scarce ( Woodworthia spp.) were rarely detected where mice were present. A further 9 DIFs were not differentiated from null models, but patterns were consistent with impacts at 5% mouse abundance. This study suggests that unless mouse control programmes commit to very low abundances, they risk little return for effort. Impact studies of invasive house mice are largely restricted to island ecosystems. Studies need to be extended to other ecosystems and species to confirm the universality or otherwise of these highly non-linear DIFs.