Citizen science data reveal altitudinal movement and seasonal ecosystem use by hummingbirds in the Andes Mountains

Ensuring connectivity is crucial to protect landscapes but it requires knowledge about how animals use ecosystems throughout the year. However, animal movements remain largely unknown in biodiversity hotspots, even for species that fulfill key ecological roles, as is the case of hummingbirds in the Andes. In the complex topography of mountain slopes, movement of these avian pollinators may occur either between habitat patches with asynchronous plant blooms or across ecosystems that are located within the same elevation bands or along altitudinal gradients. Here, we used two decades (2000-2020) of records from citizen science data and boosted regression trees to predict monthly distributions for 55 hummingbird species in the Andes. We identified shifts in altitudinal distribution between contiguous months and calculated changes in the proportion of predicted distributions occupied by ecosystem types. Our findings reveal substantial altitudinal movement and differences in the proportion of ecosystem types utilized throughout the year that had not been previously reported for several species. Yet the magnitude of altitudinal and ecosystem shifts varies between hummingbird clades, and in some cases changes in the proportion of ecosystem types within estimated distributions occurs with little variation in altitude. All ecosystems across the Andes show temporal changes in hummingbird occurrence, but these are higher in natural landscapes compared to croplands or urban areas. Finally, we used phylogenetic logistic regression to test whether altitudinal and ecosystem shifts affect population trends. We found that higher ecosystem seasonality is more strongly associated with decreasing populations in comparison to altitudinal shifts. Altogether, our study reveals complex patterns of movement in hummingbirds and highlights the importance of ecological connectivity across different ecosystem types. More generally, it demonstrates the opportunity of using citizen science data to increase understanding about species' seasonal occurrences, so that landscapes can be better managed to protect animal movement.Keywords: boosted regression trees, eBird, ecological connectivity, species distribution models