Abstract
BACKGROUND: Many migrant birds have been returning to Europe earlier in spring since the 1980s. This has been attributed mostly to an earlier onset of spring in Europe, but we found the timing of Willow Warblers' passage to be influenced by climate indices for Africa as much as those for Europe. Willow Warblers' spring passage through northern Europe involves populations from different wintering quarters in Africa. We therefore expected that migration timing in the early, middle and late periods of spring would be influenced sequentially by climate indices operating in different parts of the winter range. METHODS: Using data from daily mistnetting in 1 April-15 May over 1982-2017 at Bukowo (Poland, Baltic Sea coast), we derived an Annual Anomaly (AA, in days) of Willow Warbler spring migration. We decomposed this anomaly into three main periods (1-26 April, 27 April-5 May, 6-15 May); one-third of migrants in each period. We modelled three sequential time series of spring passage using calendar year and 15 large-scale climate indices averaged over the months of Willow Warblers' life stages in the year preceding spring migration as explanatory variables in multiple regression models. Nine climate variables were selected in the best models. We used these nine explanatory variables and calculated their partial correlations in models for nine overlapping sub-periods of AA. The pattern of relationships between AA in these nine sub-periods of spring and the nine climate variables indicated how spring passage had responded to the climate. We recommend this method for the study of birds' phenological responses to climate change. RESULTS: The Southern Oscillation Index and Indian Ocean Dipole in Aug-Oct showed large partial correlations early in the passage, then faded in importance. For the Sahel Precipitation Index (PSAH) and Sahel Temperature Anomaly (TSAH) in Aug-Oct partial correlations occurred early then peaked in mid-passage; for PSAH (Nov-March) correlations peaked at the end of passage. NAO and local temperatures (April-May) showed low correlations till late April, which then increased. For the Scandinavian Index (Jun-Jul) partial correlations peaked in mid-passage. Year was not selected in any of the best models, indicating that the climate variables alone accounted for Willow Warblers' multiyear trend towards an earlier spring passage. DISCUSSION: Climate indices for southern and eastern Africa dominated relationships in early spring, but western African indices dominated in mid- and late spring. We thus concluded that Willow Warblers wintering in southern and eastern Africa dominated early arrivals, but those from western Africa dominated later. We suggest that drivers of phenological shifts in avian migration are related to changes in climate at remote wintering grounds and at stopovers, operating with climate change in the north, especially for species with complex and long-distance migration patterns.