Competitive interactions between savanna trees.

Abstract

Savannas are socio-economically important ecosystems, which support high floral and faunal diversity. This biome covers large areas of Africa, Australia, South America, India and parts of North America, and is comprised of a mixture of grasses and woody plant biomass. Most empirical studies of savanna ecology have focused primarily on the interactions between trees and grasses, especially at the seedling stage where recruitment is regarded as a key driver of savanna dynamics. However, studies of interactions between woody savanna plants, such as competition and facilitation, are less common in the literature. Considering the increasingly negative effects of woody plant encroachment and global climate change, interactions between woody plants need to be closely monitored and evaluated. In this thesis, I investigated the effects of neighbourhood competition on four dominant tree species from humid savannas (receiving more than 1000 mm of mean annual rainfall, hereafter referred to as “humid species”) and four dominant tree species from mesic savannas (receiving around 650 mm of mean annual rainfall, hereafter referred to as “mesic species”). I employed a greenhouse-based study to examine the effects of neighbour density on the growth, survival and biomass of savanna tree seedling species. I quantified two aspects of competitive ability (competitive effect and response), and compiled competitive hierarchies for both groups. In addition, I correlated competitive ability with several plant traits. Using field surveys of natural stands of Acacia karroo from humid savanna sites across KwaZulu-Natal, South Africa, I examined the spatial patterns and competitive interactions between trees. A greenhouse experiment revealed that mesic species suffered high levels of mortality when exposed to increasing neighbour density, while humid species were relatively unaffected in terms of survival. However, mesic species were able to maintain constant relative growth rates (RGR) despite increasing neighbourhood competition while the RGR of humid species decreased as neighbour density increased. The total biomass of both humid and mesic species also declined as the neighbourhood competition increased. In terms of competitive effect and response, we found that these two aspects of competitive ability were not concordant (i.e. good effect competitors were not necessarily good response competitors). Lastly, we found that plant traits such as specific leaf area and above-ground features (e.g. shoot biomass and leaf number) were significantly related to the competitive response or effect of savanna tree seedlings. Spatial distribution patterns of a dominant humid savanna species, Acacia karroo, revealed that juvenile plants are aggregated, as expected due to facilitation, seed dispersal and vegetative reproduction. However, the regular spacing of larger individuals due to competition and density-dependant mortality were not detected. We found, using nearest neighbour analysis, that trees with closer neighbours had smaller canopy diameters. This suggests that while competitive interactions are present, they may be weak and insufficient to cause mortality, rather resulting in decreased plant performance. Overall, I found that, at the seedling stage, neighbourhood competition was particularly important for both humid and mesic savanna trees. Competitive interactions between mesic seedlings resulted in significantly higher mortality rates, greatly reducing the recruitment of these species. Humid species, although able to successfully recruit, experienced reduced growth rates under dense neighbourhood competition. In the field, patterns of competitive interactions were difficult to detect using spatial statistics alone. However, we did find evidence of weak competitive interactions among humid savanna trees. In summary, competitive interactions were important for all savanna species at the crucial seedling stage. However, field comparisons showed that competitive interactions were relatively weak in A. karroo and resulted in reduced performance rather than differential mortality.