Abstract Aim In this paper, we adopted a large-scale approach to evaluate the effect of regional richness of forest birds on the number of bird species retained by forest fragments in several localities across Europe. Location We studied bird assemblages in fourteen
forest archipelagos embedded in agricultural matrices from southern Norway to central Spain. Tree composition varied from oak and beech forests of the northern localities to oak and pine xerophitic woodlands of the southern ones. The number of fragments in each forest archipelago ranged from
eighteen to 211. Methods We used the Gleason equation (s = a + z log A; where s and A are, respectively, the species richness and size of forest fragments and z the rate
of species loss) to estimate the species richness for 1- and 15-ha fragments in each archipelago. The regional richness of forest birds was estimated by modelling the geographical distribution of species richness in the European atlas of breeding birds. Results The latitudinal
distribution of regional richness displayed a convex form, with the highest values being in central Europe. Along this gradient, the number of species retained by fragments and the rate of species loss was positively related to regional richness. In addition, the percentage of the regional
pool of species sampled by fragments decreased in the southern localities. Main conclusions Relationships between regional richness of forest birds and richness in fragments seem to explain why fragments in central Europe shelter more species than their southern counterparts.
The decreased ability of southern forest fragments to sample the regional richness of forest birds, could be explained as an effect of the low abundance of many species in the Mediterranean, which could depress their ability to prevent extinction in fragments by a rescue effect. Alternatively,
high beta diversity in the Mediterranean could produce undersampling by fragments of the regional pool of species. These regional differences in the response of bird assemblages to forest fragmentation are used to discuss the usefulness of large-scale, biogeographical approaches in the design
of conservation guidelines.