Site-based vs. species-based analyses of long-term farmland bird datasets: Implications for conservation policy evaluations

Conservation of Europe’s biodiversity increasingly depends on funds invested within Natura 2000 farmland. Performance of these investments is estimated by the official Farmland Bird Index indicator, that merges species-specific trends for farmland species estimated with the standard TRIM method. We here reanalyze the long-term datasets used to calculate the Spanish Farmland Bird Index by computing abundance and richness of selected bird groups at the point census scale rather than by merging species’ trends. We test whether community trends at site scales differed according to agricultural habitat types (annual, perennial, and mosaic croplands) and locations inside or outside Natura 2000 sites, using both the TRIM method and generalized mixed models. Site-based analyses showed a general increase in bird abundance and richness outside the Natura 2000 network, and a general decrease in perennial and mosaic croplands inside it. Increasing trends were due to non-farmland birds occupying farmland, as farmland species showed significant decreasing trends overall, especially inside Natura 2000 sites and for steppic birds. Trends for threatened birds in annual cropland located inside Nature 2000 were positive, but trends for threatened farmland birds were negative overall, especially in mosaic croplands. Results were qualitatively consistent among statistical methods, although quantitative estimates varied widely among methods, habitats, Natura 2000 location, and relevant bird groups. Site-based analyses of long-term databases confirmed overall trends detected by species-based official reports, and complement them by suggesting additional reasons for failures at reverting negative trends in farmland biodiversity. Regionally-targeted conservation measures should be developed and/or extended to improve these results, and their results monitored at the farm scale to complement the low spatial resolution of volunteer-based bird monitoring schemes. Combination of broad–scale citizen science programs with cause-effect, finer-scale studies will help disentangle the causes of the observed patterns to develop better and more efficient recommendations for conservation measures in farmland areas.