Assessing habitat/landscape predictors of bird diversity in managed deciduous forests: a seasonal and guild-based approach

In this study we focused on deciduous managed forests in northwest Italy and investigated relationships between bird diversity and forest descriptors both at the habitat and landscape level through a seasonal, guild-based approach. Birds were censused by point counts in randomly selected plots of 50 m radius. Each census plot was described in habitat structure terms by using the Bitterlich variable plot-radius method. Landscape metrics (both at class- and landscape-level, according to Fragstats' terminology) were measured within 500 m radius circle of each plot using an existing GIS database. We first identified specialist and generalist species according to the presence/absence of each species in relation to habitat structure measurements using generalized additive models (GAMs), then we tested for the effects of habitat and landscape variables on guild parameters (namely, bird species richness, Shannon diversity and abundance) using generalized linear models (GLMs). Our analyses showed that several habitat and landscape predictors affected bird diversity. However, in general, guild parameters relative to specialists were especially influenced by oak biomass (a habitat predictor) and oak core area (a class-level predictor) irrespective of the season, while parameters of generalists were influenced by oak core area in winter and perimeter/area ratio of oak patches in spring. Landscape-level predictors significantly affected winter bird abundance only. Results clearly showed that the retention of native oaks both at the habitat and the landscape level is the key-factor for the conservation of diversity of specialist bird species in deciduous, originally oak-dominated woods. Results also suggest that studies focusing on different periods should be preferred to mono-season surveys and that a guild-based approach should be preferred to an holistic one taking into account all species.