Hierarchical processes define spatial pattern of avian assemblages restricted and endemic to the arid Karoo, South Africa

Abstract Aim To identify and quantify biotic and abiotic factors associated with the regional gradients in the distribution and abundance of bird communities restricted and endemic to the Succulent and Nama Karoo biomes of South Africa. Location The arid Nama and Succulent Karoo biomes in South Africa. Methods The quarter degree grid cell (QDGC) was used to extract environmental data, while the bird data previously atlased, was linked to the same geo‐referenced system, using a geographical information system (GIS). Bird species were grouped into different life‐history assemblages. A quantitative, systematic analysis of the different bird communities spanning the Karoo was undertaken to examine contributions of broad‐ and local‐scale physical environmental and biotic factors to regional variations in the species composition, using multivariate statistical and spatial analytical tools. These included two indirect gradient methods; principal components analysis (PCA) and detrended correspondence analysis (DCA), and two direct gradient methods; canonical correspondence analysis (CCA) and redundancy analysis (RDA). Results Principal components analysis results showed that the selected environmental variables accounted for about 85% of the variation in the region. The first two principal gradients defined regional temperature seasonality and variability especially in winter and summer. The third principal gradient mainly defined summer rainfall areas in association with the coefficient of variation of rain and regional primary production, while the fourth gradient defined winter rainfall areas, growth days and elements of landscape structure. CCA/RDA analysis produced shortened hierarchically ranked explanatory variables for each bird assemblage. Stepwise gradient analysis results showed summer rain, rainfall concentration, topographic heterogeneity and annual evapotranspiration, as the most important climate variables explaining species occurrence. Landscape, in terms of percentage transformation, morphology, coefficient of variation of primary productivity and distance between suitable habitat patches, were also important, but to a lesser degree. Total variation explained (TVE) by the supplied variables was between 23 and 37% of variation. Less than 20% of TVE was the intrinsic spatial component of environmental influence, indicating that any unmeasured factors were independent of spatial structuring. For all the eight bird assemblages, climate contributed most to TVE (24–57%). Landscape characteristics (human‐induced transformation, vegetation in terms of size if grassy clumps and the average distances between them) contributed theleast to TVE for all the different assemblages (0–6%), especially the granivorous assemblage where it was not significant at all (0%). Seasonal extremes and variability were more important in explaining species gradients than were annual climatic conditions, with the exception of annual potential evapotranspiration. Main conclusions This study was able to synthesize species environment relations at the broad scale and demonstrated the association of arid zone endemic species occurrence with climate extremes and seasonality. Given the predicted climate change scenarios for South Africa, this regional gradient study provides a quantitative ecological basis for finer scale modelling and analysis, developing regional strategies for conserving biodiversity as well as predicting and planning for the effects of global climate change. However, most importantly, it clearly showed that bird species restricted and endemic to the arid Karoo biome may be more sensitive to climate rather than vegetation structure as previously thought.