Differential spatial responses of rodents to masting on forest sites with differing disturbance history

Mast seeding, the synchronized interannual variation in seed production of trees, is a well‐known bottom‐up driver for population densities of granivorous forest rodents. Such demographic effects also affect habitat preferences of the animals: After large seed production events, reduced habitat selectivity can lead to spillover from forest patches into adjacent alpine meadows or clear‐cuts, as has been reported for human‐impacted forests. In unmanaged, primeval forests, however, gaps created by natural disturbances are typical elements, yet it is unclear whether the same spillover dynamics occur under natural conditions. To determine whether annual variation in seed production drives spillover effects in naturally formed gaps, we used 14 years of small mammal trapping data combined with seed trap data to estimate population densities of Apodemus spp. mice and bank voles (Myodes glareolus) on 5 forest sites with differing disturbance history. The study sites, located in a forest dominated by European beech (Fagus sylvatica), Norway spruce (Picea abies), and silver fir (Abies alba), consisted of two primeval forest sites with small canopy gaps, two sites with larger gaps (after an avalanche event and a windthrow event), and a managed forest stand with closed canopy as a control. Hierarchical Bayesian N‐mixture models revealed a strong influence of seed rain on small rodent abundance, which were site‐specific for M. glareolus but not for Apodemus spp. Following years of moderate or low seed crop, M. glareolus avoided open habitat patches but colonized those habitats in large numbers after full mast events, suggesting that spillover events also occur in unmanaged forests, but not in all small rodents. The species‐ and site‐specific characteristics of local density responding to food availability have potentially long‐lasting effects on forest gap regeneration dynamics and should be addressed in future studies. We present novel hierarchical Bayesian N‐mixture models of unique count data from 14 years of small mammal live trapping investigating functional responses of dominant rodent taxa to different habitat types and mast seeding, including sites of natural disturbance events and sites in the largest remaining primeval forest of the Alps.