Within-Night Variation in Predictor Importance Highlights Dynamic Nature of Bird Migration.

Ecological forecasting is increasingly important for conservation. Predicting nocturnal bird migration events is a promising vehicle for forecasts but isn't often explored at fine temporal scales. We use weather surveillance radar to examine dynamic drivers of migration in 2-h periods throughout a night. We assess the relative importance of terrestrial, atmospheric and sampling predictors (which relate to radar position and scan timing) across spring and fall. Atmospheric conditions were consistently strong predictors. In contrast, terrestrial predictors contributed relatively little to explaining variation in activity. Sampling variables, such as time after sunset, varied in importance, with the highest influence shortly after sunset. We highlight the temporal variability in predictors of migration, emphasising it as a dynamic process, involving continuous decisions and adjustments rather than following fixed routes. We underscore the value of radar for capturing transitions between habitats while revealing key limitations and opportunities for understanding fine-scale migratory behaviour.