There are two main hypotheses describing tree grass coexistence in savanna. The first of these was proposed by Walter (1970) suggesting that trees obtain their resources from deep soil layers while grasses are more shallow rooted. The second hypothesis was proposed by Higgins et al (2000) suggesting that trees are prevented from reaching mature size classes through insufficient rainfall or alternatively through fire and herbivory. Our research in the Kruger National park has allowed us to examine these two hypotheses more closely giving us a much broader mechanistic understanding of savanna systems. We now believe that the availability of water and nutrients not only determine grass biomass but also grass species composition. Whether these are in South America, Australia or in southern Africa savanna soils are limiting in nitrogen and phosphorous and this means that both grasses and trees are rooted in the same soil horizons where nutrients are highest.
Grasses rather than trees are superior competitors for these resources and grasses are not affected by fire frequency but fire frequency does control tree biomass. How then do trees survive among competitively superior grasses? Trees avoid competition with grasses by leafing out early and taking up nutrients before grasses are able. Trees are able to establish when grass biomass is low through drought or intense grazing. Funding for the manipulative experiments necessary to develop this understanding was not possible without the Mellon Foundation and here I reflect on how this understanding was achieved over the last 15 years.