Drivers of Stopover Distribution Across Two Autumn Migration Routes of a Long-Distance Afro-Palearctic Migrant, the...

Across the globe, many long-distance migratory birds have undergone population declines. Understanding the environmental factors which underly migratory stopover locations can inform conservation strategies for such species. Here, we identify environmental predictors of autumn stopover sites used by UK-breeding male cuckoos (Cuculus canorus) migrating to central African wintering grounds via two alternative flyways. Between 2011 and 2021, 88 male cuckoos from UK breeding grounds were fitted with < 5 g satellite tags connected to the Argos satellite-based tracking system. Filtered stopovers were used to build presence-pseudoabsence distribution models for each flyway, using Boosted Regression Tree and Random Forest algorithms with thirteen climatic, habitat, topographic and anthropogenic variables. Performance was evaluated using 10-fold cross-validation, and weighted ensembles were generated. To assess drought sensitivity, we produced predictions for high- and low-drought years based on Standardised Precipitation-Evapotranspiration Indices (SPEI). Models performed well (AUC = 0.79-0.88; TSS = 0.50-0.66) and revealed route-specific environmental predictors. Along the western route, stopover probability was highest in cooler, semi-humid, high-elevation areas with low contemporaneous drought; Iberian mountain ranges emerged as key sites. Along the eastern route, stopover probability was greatest in heterogeneous habitat, with lowlands and uplands favoured over intermediate elevations; the Po watershed appears particularly important. Hunting risk (included to assess anthropogenic threat) was also high at key areas. The spatial extent of stopover regions-and the probability of them being utilised-declined during years of high short-term drought along the western route, and high long-term drought along the eastern route. Whilst Iberian montane stopover regions are sensitive to increasing drought and aridity under climate change, their topographic heterogeneity may provide climatic refuge while limiting intensive land uses. Preference for heterogeneous habitat within the Po watershed highlights the need to retain semi-natural habitat within agricultural landscapes. We recommend route-specific conservation measures and higher resolution models using newly available GPS tags to better understand fine-scale site selection within the regions highlighted here.