Species and environment representation: selecting reserves for the retention of avian diversity in KwaZulu-Natal, South Africa

Multivariate analyses and complementarity-based reserve selection algorithm s were used as tools for delimiting representative conservation areas in Kw aZulu-Natal, South Africa, based on environmental variables and records for 566 breeding and nonbreeding birds. The relationships between species asse mblages and 20 environmental variables were investigated; including climate , topography, vegetation, landscape attributes, and land-use. Detrended cor respondence analysis, canonical correspondence analysis, and hierarchical c lassification suggested that the major gradients of bird species variation are primarily related to climatic variables such as growing season temperat ure and seasonality of precipitation, reflecting the strong;east-west gradi ent in oceanicity in KwaZulu-Natal, and water balance, reflecting a strong north-south gradient in aridity. Hierarchical classification was applied to the ordination data and revealed five representative types of bird communi ties. Four of the five bird communities and 37 species are under-represente d in the existing reserve system, and the environmental processes required for long-term maintenance of birds have not been adequately sampled, showin g that additional reserves are needed to ensure a more functional represent ative system. As an alternative, the inclusion of species spatial turnover structure together with associated environmental gradients were employed in both rarity and richness-based complementary reserve selection algorithms to provide a better representation of birds and environmental processes. Im provements on the current generation of reserve selection techniques derive d by including ordination analysis and spatial structure are discussed. (C) 2001 Elsevier Science Ltd. All rights reserved.