Wind and rain govern the intensity of nocturnal bird migration in central Europe - A log-linear regression analysis

For the first time an almost complete sample of the seasonal course of nocturnal autumn migration was recorded by a conically scanning pencil-beam radar and was analysed with respect to weather, using a log-linear regression model. A variable aiming to estimate the number of birds ready to depart for migration due to preceding precipitation was included in the model. The density of nocturnal migration was strongly influenced by wind and by rain. It appears that birds distinguished between favourable and unfavourable wind conditions rather than to grade wind on a continuous scale from least to most favourable. The final model, including variables for actual wind, rain, previous precipitation, and the general seasonal trend explained 70% of the total deviance. Thus, the low correspondence of empirical data with predictions from models on stopover duration and optimal fuel loads with respect to fat deposition rates may be caused by the high impact of local weather conditions on the take off decision of the majority of nocturnal migrants passing through central Europe.