Modeling forest bird distributions with bioclimatic, forest composition, and GEDI fusion forest structure variables: the importance of ecoregional perspectives

This is the authors' abstract. We don't add key points for this paper.

Understanding how species–habitat relationships vary across large spatial extents is critical for conservation and management. Woodpeckers are a wide-ranging guild whose roles as keystone species often make them a focus of management activities. Using woodpeckers as a focal group, we sought first to examine distributions of 11 woodpecker species in the Northwest forested mountain ecoregion (NWFM). We modeled eBird encounter rates as a function of modeled forest structural data based upon the global ecosystem dynamics investigation (GEDI) instrument (hereafter referred to as GEDI-fusion data), bioclimatic variables, and other remotely sensed variables using RandomForest. Subsequently, we compared results from the NWFM to a similar effort in the Marine West Coast ecoregion. Our model performance within the NWFM generally ranged from good-excellent (AUCs ranging from 0.65 to 0.89). Bioclimatic variables were the most important variables across all species at our home range scales of analyses, with GEDI-fusion data amongst the top predictors. The importance of GEDI-fusion data in models varied by species, and different (and sometimes contradictory) relationships were found for the same structural variable for the same species when compared between ecoregions. This study emphasizes the importance of identifying how species–habitat relationships may differ across ecoregions. Additionally, this work highlights the growing body of literature that leverages large scale animal data (e.g. eBird data) with relatively fine scale forest structure data derived from the GEDI instrument.