Spatial point pattern of the baobab (Adansonia digitata) in semi-arid savanna, Southeast Zimbabwe

Abstract

Species distribution modelling based on bioclimatic variables to understand habitat suitability and changes in range dominates studies on the spatial distribution of baobabs, which limits our understanding of how the small-scale plant–plant interactions influence the spatial point pattern of this keystone species in semi-arid regions of sub-Saharan Africa. Here, we investigated the spatial distribution of the baobab in two protected areas, Chipinge Safari Area (CSA) and Save Valley Conservancy (SVC) which lie in the semi-arid Southeastern lowveld of Zimbabwe. CSA was previously a sparsely human populated area before its proclamation as a protected area, has a small number of herbivores and baboons, and has no elephants, while SVC has a high density of elephants (1.5 km−2) and large number of herbivores. At each site, 1 km2 plots, SVC (n = 4) and CSA (n = 3), were established for spatial analysis. In each plot, we recorded the spatial location, elevation, height, diameter at breast height, and canopy cover of each baobab. Spatial point-pattern was assessed using both the inhomogeneous transformed L-function for Ripley's K-function and the pair-correlation function. Baobabs in SVC were larger (height, canopy cover, and basal area) relative to those in CSA. At both sites, baobabs were clustered at small spatial scale (0–50 m), indicating that competition was insufficient to cause density-dependent thinning that can lead to even distribution as we expected. The distribution of small baobabs was independent of large baobabs, which reflects the importance of other environmental factors such as site conditions and seed dispersal rather than nurse-protégé on seedling establishment. The random distribution pattern of adult baobabs could be the legacy effects of these random factors and processes at establishment. Overall, our study shows that the distribution of baobabs is not dependent on local plant–plant interactions, but most likely from larger scale random pattern forming factors and processes such as seed dispersal and environmental heterogeneity.