TURNOVER OF BREEDING BIRDS IN SMALL FOREST FRAGMENTS - THE SAMPLING COLONIZATION HYPOTHESIS CORROBORATED

A sampling view of the colonization of small habitat fragments by birds is based on the assumption that individual birds acquire territories in the fragments randomly, and their occurrence frequencies follow from abundances in the regional species pool. We tested the following predictions of this view using bird census data from a fragment ''archipelago'' in the southern Finnish taiga: (1) the location of breeding pairs varies randomly among the fragments from year to year, and (2) the pattern of species accrual with increasing sample size is similar to random sampling. The data agreed with the predictions. For most species the distribution of pairs in the fragments agreed with the Poisson prediction. The pattern of species turnover in single fragments agreed with the sampling view so that an average fragment had about 10 pairs of 7 species in a single breeding season, and about 40 pairs of 13 species over the 4-yr period. The sampling model is a ''null'' assumption that should be considered when assessing the occurrence of birds in small habitat fragments. This requirement defines methodological imperatives that have often been violated in ornithological studies; we discuss such methodological problems. The sampling model implies that local turnover in small fragments may be a mechanical reflection of change in territory location from year to year and disconnected from population dynamics. In such a case, apparent ''extinctions'' and ''recolonizations'' in single fragments are ecologically trivial. Assessing the interface between sampling colonization and genuine metapopulation dynamics is an urgent challenge for conservation ecology, particularly in temperate and boreal areas.