Published online by Cambridge University Press: 11 May 2010
To produce conservation indices from bird community data, with the aim of identifying and comparing areas of regional conservation concern, we selected bird community samples randomly over a large study area, without including habitat features a priori in the sampling procedure. Subsequently, we subjected the samples to a cluster analysis to identify a posteriori bird communities to be indexed together with their associated habitat and distribution/abundance on a regional scale. We identified 15 bird community types and were able to characterize each of them by a community conservation index (CCI) taking into account three different components of rarity: habitat selectivity, geographical distribution, and abundance. A cluster of four communities was associated with lowland farmland and woodland; another cluster of five was found at higher average altitudes and degree of urbanization; a third one of six showed a montane distribution. This third cluster showed much higher CCI values in comparison with the first two suggesting that the index here proposed is very sensitive to a high community selectivity and a restricted distribution/abundance.