Predicting Vegetation Patterning across Climate, Soil, and Topographic Gradients

Abstract:

Vegetation communities in water-limited systems sometimes form periodic patterns, e.g. banded, spotted and labyrinthine distributions of woody and herbaceous plants. Pattern formation is commonly linked to competition and facilitation among plants, and variation in runoff and infiltration capacity in the landscape. Based on previous studies, we expect that climate, soil type, and slope to a large degree influence the type of vegetation pattern found at a specific site. We have analyzed to what extent vegetation patterns on the African continent can be predicted based on available climatic, topographic, and soil data. Our focus is not restricted to periodic patterns in drylands, but encompasses a range of tropical ecosystems from arid to humid. Vegetation patterns observed in remote sensing data can be informative regarding the underlying ecological processes that shape the landscape, not only in strikingly periodic vegetation but also in savannas with randomly located or dispersed vegetation. We use high-resolution multispectral and panchromatic remote sensing data classified into woody, herbaceous, and bare ground components. From these images we extract spatial statistical metrics that define type and degree of vegetation patterning. We then relate variables from climate, soil and topographic datasets to the observed patterns in order to determine how well we can predict vegetation patterning and which climatic and edaphic variables are most informative. We discuss the results and the possible sources of uncertainty in the relationships.