Scale-Dependent Influence of Topography-Based Hydrologic Features on Patterns of Woody Plant Encroachment in Savanna Landscapes

Abstract

Rainfall in drylands is erratic. Topographic features of landscapes can dampen or amplify temporal variability by spatially influencing patterns of water loss and accumulation. The extent to which portions of a landscape may differentially capture or retain scarce water and nutrient resources is an important determinant of vegetation patterns, particularly with respect to the distribution of woody plants. We therefore hypothesized that historic changes in woody cover on landscapes experiencing similar climate and disturbance regimes would vary with catena-to-catena (hillslope-to-hillslope) variation in topography-based hydrologic features. We tested this hypothesis by comparing topographic wetness index (TWI) values on replicate landscapes where woody plant abundance has increased over the past 100 yr. These landscapes are characterized by savanna parklands on coarse-textured upland portions of catenas that grade (1–3% slopes) into closed-canopy woodlands on fine-textured (lowland) portions of catenas. TWI values for woody and herbaceous communities were comparable within uplands, suggesting factors unrelated to surface/subsurface hydrology determine patterns of woody cover in these catena locations. TWI values for upland savanna parklands were significantly lower than those of closed-canopy woodlands occupying catena footslopes. Furthermore, uplands adjoining historically static woodland boundaries had lower TWI values than those where woodland boundaries had moved upslope 2.1 m yr � 1 from 1976 to 1995. Results suggest runoff–runon relationships influence patterns of woody plant cover and change at the catena scale and may override constraints imposed by soil texture. As a result, changes in woody cover potentially accompanying changes in disturbance regimes, climate or atmospheric chemistry are likely to be constrained by topoedaphic settings. Models of vegetation dynamics may therefore need to explicitly account for rainfall–topography–soil texture relationships and associated scale-dependent mechanisms to accurately predict rates and patterns of change in woody and herbaceous plant abundance.