Nearest-neighbour analysis and the prevelance of woody plant competition in South African savannas

Abstract

Inter- and intra-specific competition has long been regarded as asignificant determinant of the structure and function of woody plantcommunitiesin African savannas. The most widely used approach to detect the presence orabsence of competition has been the use of nearest-neighbour methods. Althoughseveral studies have been published reporting the presence of competition inAcacia dominated communities, less than 20 plots have beensampled across all these studies. Results from broad-leaved communities arevariable, and also based on a small sample number. Consequently, this studysought to assess the prevalence of competition from a large number of savannasites (45), and to identify abiotic and biotic factors characterising siteswithcompetition relative to those without. Using the nearest-neighbour method onlyfour sites (9.3%) indicated the presence of inter-specific competition; twowereAcacia dominated communities (18.2% ofAcacia sites) and two were broad-leaved communities(7.7%).These four sites had a significantly higher mean annual rainfall and fewerwoodyspecies than sites without inter-specific competition. There were alsosignificant relationships between the inter-specific competition index and thedominance index and the number of species at a site. In terms of intra-specificcompetition only ten of the 31 sites tested revealed a significant correlationbetween nearest-neighbour distance and summed canopy volume of the twoneighbours. Seven were for Acacia species (55.6% of theAcacia species comparisons) and three were forbroad-leavedspecies (21.4% of the broad-leaved species comparisons). Sites lacking evidenceof intra-specific competition were at a significantly lower slope position andhad a smaller proportion of small stems. The lower prevalence of competition inSouth African savannas detected in this study using the same method as previousstudies is discussed, and the appropriateness of nearest-neighbour analysis isexamined.