Nonlinear effects of climate on boreal rodent dynamics: mild winters do not negate high‐amplitude cycles

Small rodents are key species in many ecosystems. In boreal and subarctic environments, their importance is heightened by pronounced multiannual population cycles. Alarmingly, the previously regular rodent cycles appear to be collapsing simultaneously in many areas. Climate change, particularly decreasing snow quality or quantity in winter, is hypothesized as a causal factor, but the evidence is contradictory. Reliable analysis of population dynamics and the influence of climate thereon necessitate spatially and temporally extensive data. We combined data on vole abundances and climate, collected at 33 locations throughout Finland from 1970 to 2011, to test the hypothesis that warming winters are causing a disappearance of multiannual vole cycles. We predicted that vole population dynamics exhibit geographic and temporal variation associated with variation in climate; reduced cyclicity should be observed when and where winter weather has become milder. We found that the temporal patterns in cyclicity varied between climatically different regions: a transient reduction in cycle amplitude in the coldest region, low‐amplitude cycles or irregular dynamics in the climatically intermediate regions, and strengthening cyclicity in the warmest region. Our results did not support the hypothesis that mild winters are uniformly leading to irregular dynamics in boreal vole populations. Long and cold winters were neither a prerequisite for high‐amplitude multiannual cycles, nor were mild winters with reduced snow cover associated with reduced winter growth rates. Population dynamics correlated more strongly with growing season than with winter conditions. Cyclicity was weakened by increasing growing season temperatures in the cold, but strengthened in the warm regions. High‐amplitude multiannual vole cycles emerge in two climatic regimes: a winter‐driven cycle in cold, and a summer‐driven cycle in warm climates. Finally, we show that geographic climatic gradients alone may not reliably predict biological responses to climate change.