Seasonal changes in the numerical responses of predators to cyclic vole populations

Theoretical models predict that a delayed density‐dependent mortality factor with a time lag of ca 9 months is able to drive 3–5‐yr population cycles of northern voles. We studied numerical responses of predators in western Finland during 1986–92, in an area with 3‐yr population cycles of voles. Abundances of small mammals were monitored in several farmland areas (each 3 km 2 ) by snap‐trapping in April, June, August, and October (only in 1986–90), and the abundances of avian, mammalian, and reptilian predators by visual censuses during trapping occasions. The 3‐yr cycle studied was a cycle of Microtus voles (field vole M. agrestis and sibling vole M. rossiaemeridionalis ) and their small‐sized predators (small mustelids and vole‐eating birds of prey). The numerical responses of both migratory avian predators and small mustelids to changes in vole densities were more alike than different. In late summer (August), the time lag in the numerical response of all main predators was short (0–4 months), whereas longer time lags prevailed from spring to early summer. The length of the time lag in spring appeared to be related to the length of the winter, which indicates that strong seasonality may create longer time lags to the numerical response of predators at northern latitudes than at more southern latitudes. Our results suggest that, from spring to early summer, predation by migratory avian predators may act in concordance with mustelid predation to produce the long time lag necessary to drive the 3‐yr cycle of voles, whereas almost direct density‐dependent predation by all major predators in late summer may dampen spatial variation in prey densities.