Identification of weather parameters related to regional population outbreak risk of common voles (Microtus arvalis) in Eastern Germany

Abstract

Context Common vole (Microtus arvalis) populations can increase to several thousand individuals per hectare during outbreaks. In central Europe such outbreaks usually extend across large areas but there can be significant regional differences in outbreak intensity, general outbreak risk and associated crop damage. Aims We tested whether weather parameters can be used to explain the regional variability in outbreak risk of common voles in an area of Eastern Germany where common vole outbreaks are common. Method Suitable weather parameters were identified by principal component analysis (PCA). Time series of common vole abundance from 50 locations across 36 500 km2 sampled in 1973–97 were related to weather parameters selected by PCA and multiple linear regression. A hierarchical cluster analysis was used on relevant weather parameters to display the temporal and spatial variability in vole abundance. An overlay of risk class transformed abundances allowed for the identification of appropriate threshold values to define vole outbreaks. Key results Weather parameters were closely related to the variation in regional outbreak risk of common voles. Mostly weather parameters in winter and early spring were identified to be highly important. All risk thresholds tested revealed similar patterns for the distribution of risk classes across locations and years. While most years of very low or very high outbreak risk clustered well according to weather parameters, some cases of medium risk classes did not cluster well. Conclusions Weather parameters especially in winter and early spring are related to common vole outbreak risk in the following autumn. This is the case for extremely high and low outbreak risks and is largely independent of the choice of particular threshold values for outbreak risk. Implications Weather parameters could be used to develop automated forecast systems at the spatial resolution of single weather stations. Combined with other parameters that are easily available, such as information on soil characteristics, such forecasts might be as reliable as more complex biological models developed in the past.