Tree and grass interactions governing the production and distribution of savannas: linking meta-scale patterns to underlying mechanisms

Abstract

Savannas, characterized by the co-dominance of herbaceous and woody vegetation, support an estimated 20% of the global human population and account for ~30% of terrestrial net primary productivity. Interactions among savanna trees and grasses determine important ecosystem functions such as hydrological and biogeochemical cycles and production and transpiration rates, and impact the availability of resources (fuel-wood, grass for livestock) fundamental for human wellbeing. Additionally, interactions among trees may be an important driver of savanna vegetation structure, though few existing studies empirically estimate the intensity and importance of savanna tree-tree interactions. A clear understanding of the mechanisms that govern the coexistence of trees and grasses and their interactions in savanna landscapes is crucial to our ability to predict their responses to changing climatic and anthropogenic disturbance regimes. I present research aimed at advancing our understanding of emergent trends in savanna plant interactions and the underlying mechanisms responsible for observed patterns. First, I present the results of a meta-analysis of empirical studies that quantify the net effect of savanna trees on grass production under tree canopies relative to production away from trees. We found that the effect of trees on subcanopy herbaceous production varies predictably with climate, such that trees in arid savannas generally promote grass growth and trees in mesic regions suppress growth. This finding is consistent with a general theoretical model predicting the relative importance of facilitative processes for species coexistence. Termed the stress gradient hypothesis (SGH), the theory predicts an increasing importance of facilitation with increasing environmental stress, such as high water-stress typical of arid savannas. I then present results from two empirical studies designed to experimentally test the predictions of the SGH and infer mechanistic drivers by relating abiotic covariates to plant growth in the presence and absence of neighbors. In the shortgrass steppe (SGS) in northeastern Colorado, we found a net-neutral effect of shrubs and grasses on the other life form, contrary to expected facilitation. We suggest shrub morphology and interactive effects of topography and soil texture are primarily responsible for observed patterns of growth. At five savanna field sites situated along a rainfall gradient (i.e. water-stress gradient) in Mali, West Africa, we found the net effect of trees on grass growth to be consistent with SGH predictions. Light availability and distance to tree boles best explained shifts in herbaceous production along the rainfall gradient. Lastly, I present results from a longitudinal study in an East African savanna estimating tree growth as a function of the size and distance of neighboring woody competitors. In so doing, we quantified the magnitude of inter-tree competition and inferred its impact on stand spatial structure through spatial point pattern analysis. Overall, this research increases our understanding of biotic interactions among savanna plants. The effects of savanna trees on subcanopy grass production generally conform to the predictions of the SGH, and appear to be mediated by microclimate modification by tree canopies related to light availability and water balance. The effects of grasses on trees along environmental gradients are less clear, though we found net neutral effects on woody growth over one growing season in tropical and temperate shrub-grass systems, suggesting that active competitive and facilitative mechanisms largely offset, or that the effects of grasses on plant-available resources for woody species are negligible. Finally, we found that shrubs aggregate at local scales, despite significant neighbor competition. We suggest competition among woody plants influences production and relative species abundance, but dispersal and establishment bottlenecks are likely more important for landscape-scale spatial structure. These results have important implications for our theoretical understanding of coexistence between woody and herbaceous vegetation. Furthermore, we provide empirical data that can be used to refine and parameterize vegetation models predicting savanna ecosystem processes and the global distribution of mixed tree-grass systems.