Context-dependent vegetation dynamics in an African savanna

Abstract

Understanding the spatio-temporal dynamics of ecological systems is fundamental to their successful management and conservation. Much research and debate has focused on identifying underlying drivers of vegetation change in savannas, yet few have considered the influence of spatial context and heterogeneity. Our goal was to develop deeper understanding of woody vegetation spatio-temporal dynamics through spatially explicit utilization of historical aerial photography and airborne LiDAR (light detection and ranging). We first assessed temporal change in woody vegetation cover through object-based image analysis of an aerial photography record that spanned 59 years from 1942 to 2001. Secondly, we tested the spatial relationships between environmental variables and patterns of woody structure and dynamics at broad (100 ha), medium (10 ha) and fine-scales (1 ha) through canonical correspondence analysis (CCA). Finally, we used LiDAR derived vegetation heights to explore current woody vegetation structure in the context of historical patterns of change. Total percentage woody cover was stable over time, but woody dynamics were highly variable at smaller scales and displayed distinct spatial trends across the landscape. Losses of woody cover on the diverse alluvial substrates were countered by increases of cover on the hillslopes. Analysis of current woody structure in the context of historical change revealed that the increases took place in the form of shrub encroachment and not the replacement of tall trees. We infer that mammalian herbivory contributed substantially to the losses on lowland alluvial soils, whilst shrub encroachment on the upland hillslopes likely stemmed from changes in fire regime and climate. Deeper reflection on spatial variability is needed in the debate around drivers of change in savanna systems, as spatial patterns of change revealed that different drivers underlie vegetation dynamics in different landscape contexts. Spatial heterogeneity needs explicit consideration in the exploration of pattern–process relationships in ecological systems.