Using satellite data to assess spatial drivers of bird diversity

Birds are useful indicators of overall biodiversity, which continues to decline globally, despite targets reduce its loss. The aim this paper is understand the importance different spatial drivers for modelling bird distributions. Specifically, it assesses satellite-derived measures habitat productivity, heterogeneity and landscape structure diversity across Great Britain. Random forest (RF) regression used assess extent a combination covariates explain woodland farmland richness. Feature contribution analysis then applied relationships between response variable in final RF models. We show that much variation distributions explained (R2 0.64–0.77) using monthly habitat-specific productivity values (FRAGSTATS) metrics. highlights important species richness diversity, including high grassland during spring birds patch edge length birds. feature provides insight into form relationship when particular driver affects positively or negatively. For example, May 80th percentile Normalized Difference Vegetation Index (NDVI) broadleaved has strong positive effect on NDVI >0.7 negative below. If such as these stable over time, they offer analytical tool understanding comparing influence drivers.