Neighbourhood analyses of tree seed predation by introduced rodents in a New Zealand temperate rainforest

House mice Mus musculus and other introduced rodents represent a novel source of predation on tree seeds in New Zealand forests. In the northern temperate forests where these rodents are native, spatial and temporal variation in tree seed production can result in dramatic fluctuations in the distribution and abundance of seed predators, with subsequent feedbacks on the distribution and abundance of seedlings. We use neighbourhood models to examine variation in rodent predation on seeds of 4 tree species of the temperate rainforests of New Zealand as a function of 1) spatial variation in local canopy composition and 2) spatial and temporal variation in mouse activity. We placed seeds throughout mapped stands of mixed forests in alluvial valley bottoms and on elevated marine terraces in the Waitutu Forest, South Island. The risk of predation on seeds of 2 dominant canopy trees - rimu Dacrydium cupressinum and mountain beech Nothofagus solandri var. cliffortioides- peaked in neighbourhoods dominated by those species and by silver beech N. menziesii, particularly in a year of plentiful seed rain from these species. The risk of predation on rimu and beech seed was also related to measures of local mouse activity. These relationships suggest that the highest local abundance of mice was concentrated in rimu and beech neighbourhoods because of the food provided by seed rain from those trees. Predation on seed of miro Prumnopitys ferruginea, which is eaten by rats but not mice, was low in rimu neighbourhoods and where mouse activity was high. These patterns may reflect spatial segregation in the activity of rats versus mice within stands. Our results suggest that the spatial distribution of canopy trees translates into predictable patterns of variation in mouse activity and seed predation. Heterogeneity in rodent activity and seed predation within stands may have important implications for tree population dynamics. © Ecography.