forest patches

We quantify the notion of predictable species loss from progressively smaller islands, and apply the quantification to the indigenous forest-dwelling birds of a series of New Zealand islands and to the passerines of the Cyclades Archipelago in the Aegean Sea. The analysis focuses on the reasons why the species-area relationship deviates from a perfect rank-correlation. For both avifaunas, most species are found remarkably predictably: they approximate a pattern in which each species occupies all those and only those islands larger than some species-specific minimum area.

This paper considers how habitat geometry affects New Zealand bird distributions on land-bridge islands, oceanic islands, and forest patches. The data base consists of distributions of 60 native land and freshwater bird species on 31 islands. A theoretical section examines how species incidences should vary with factors such as population density, island area, and dispersal ability, in two cases: immigration possible or impossible. New Zealand bird species are divided into water-crossers and non-crossers on the basis of six types of evidence.

It will be necessary to establish reserves for the conservation of New Zealand's forest avifauna largely in the absence of detailed autecological studies. Hence the empirical findings of island biogeography may provide the best available guide to the reserve size necessary for the preservation of both species communities and individual species.