A significant fraction of the terrestrial biosphere is comprised of biomes containing tree-grass mixtures. Predictions about how changes in precipitation regimes will alter tree:grass ratios in these biomes requires a mechanistic understanding of how trees and grasses partition and compete for water. Rooting separation as a function of depth constitutes one important axis through which resource partitioning can occur, but there is a surprising lack of knowledge about the extent to which this separation varies across large-scale environmental gradients, and what it implies for water availability and tree-grass competition and coexistence. We conducted a study of tree and grass rooting depth along a mean annual precipitation (MAP) gradient extending from ~ 450 to ~ 750 mm to examine patterns of tree-grass functional rooting differences in Kruger National Park, South Africa. We used stable isotopes from soil and stem water to make inferences about relative differences in rooting depth between these two functional groups. We then used a simple mechanistic model of water uptake and vegetation biomass dynamics to explore the implications of these differences for tree-grass competitive interactions at an intensively-sampled site.
Results/Conclusions
We found clear differences in functional rooting depth between grasses and trees across the MAP gradient, with grasses generally exhibiting shallower functional rooting profiles than trees. We also found that trees tended to become more shallow-rooted as a function of MAP, to the extent that trees and grasses had similar rooting depth at the wettest sites. Our results reconcile previously conflicting evidence for rooting separation in this system, and suggest that by reducing rooting depth and increasing water use from the shallowest soil layers, trees become more competitive against grasses as MAP increases. These results have important implications for understanding tree-grass dynamics under altered precipitation scenarios, suggesting that the relative advantage of trees over grasses as regimes shift towards fewer but larger rainfall events may be strongly dependent on the amount of baseline precipitation.