Ecohydrological significance of observed power-law distributions of tree canopy cluster sizes for savanna vegetation
Abstract
Savanna ecosystems are characterized by sparse tree cover, the pattern of which can provide insight into the processes by which these systems are shaped and maintained. Since water is often considered to be the limiting factor for savanna vegetation, we used mean wet season rainfall (MWSR) as the basis for the selection of six sites along the Kalahari Transect, a north-south aridity gradient in southern Africa. MWSR ranged from 879 mm to 216 mm from the wettest to most arid site, and the corresponding fractional tree cover ranged from 0.65 to 0.04. IKONOS satellite images, having a resolution of 4m in the multi-spectral bands, were acquired for each of these sites and tree canopies were identified by merging the remotely-sensed normalized difference vegetation index data with information taken from ground-based surveys. Analysis of the tree canopy cluster size distributions showed the wettest site to deviate from a power-law distribution as a result of the fractional cover being above the four-neighbor percolation threshold for a square lattice. Four of the five other sites, however, were characterized by distinct power-law distributions, despite having substantially disparate fractional covers. Simple random neutral models were insufficient in describing the tree clustering patterns of the observed savanna landscape, leading us to seek fundamental interactions that could explain the consistent spatial organization. In particular, our cellular automata approach focuses on water use by the savanna vegetation, and how annual variability in this limiting resource can lead to emergent pattern formation.
- Publication:
-
AGU Spring Meeting Abstracts
- Pub Date:
- May 2004
- Bibcode:
- 2004AGUSM.B41A..17S
- Keywords:
-
- 1640 Remote sensing;
- 1869 Stochastic processes;
- 9305 Africa