Landscape pattern and census area as determinants of the diversity of farmland avifauna in Estonia

Landscape structure affects farmland bird species richness and diversity at different scales. We develop a predictive, metric-based model for farmland bird species richness in different bird guilds and study how the research area size influences the relationship between bird diversity and landscape metrics. Thirty research locations (randomly selected, each containing four counting points) were located in three counties of Estonia. In early summer 2002 and 2004, two bird counts per point were carried out in each research location. Three landscape metrics (patch density, edge density and Shannon’s diversity index) were calculated at three spatial scales: 100- and 200-m buffer zones around the count points and in a 1 km² at 1 m × 1 m grain size. Using generalized additive mixed models with repeated measures, we found in most cases that the proportion of variance explained between the bird variables and landscape metrics increased with an increased study area thus demonstrating the effect of scale. Thus, a larger research area is recommended for landscape metrics. At large scales, however, all used landscape metrics were equally good predictors for bird species richness.