Responses to fire differ between S outh A frican and N orth A merican grassland communities

Question Does fire frequency affect mesic grassland plant community structure and composition similarly in N orth A merica and S outh A frica? Location Konza P rairie B iological S tation ( KNZ ), K ansas, USA , and Ukulinga Research Farm ( URF ), KwaZulu‐Natal, South Africa. Methods Plant community structure and composition were compared among annually burned, unburned and intermediate treatments within two long‐term fire frequency manipulation experiments in native grasslands in North America and South Africa using comparable methodology over a 5‐yr period. Because fire may reduce soil nitrogen (N) availability and thus affect plant community structure, N additions were superimposed on the fire treatments as a means of assessing direct vs indirect mechanisms driving responses to fire. Results The total number of species was higher at URF (183) than at KNZ (57). Overall divergence in plant community response to fire frequency occurred despite similar responses to nutrient additions. At KNZ , more frequent fire resulted in dominance by a few, tall, deep‐rooted rhizomatous grasses (e.g. A ndropogon gerardii ). On unburned sites, shorter, more shade‐tolerant species such as P oa pratensis increased in abundance, although A. gerardii remained dominant. Species richness increased with decreasing fire frequency at KNZ . At URF , frequent fire resulted in short, diverse grassland weakly dominated by a range of grass species, including Themeda triandra , Tristachya leucothrix and Hyparrhenia hirta . Decreasing fire frequency reduced species richness and resulted in dominance by a few, relatively tall caespitose grasses such as Aristida junciformis . There was a complete turnover of dominant species between annually burned and unburned treatments at URF , while at KNZ A. gerardii and Sorghastrum nutans occurred across the range of treatments. N addition reduced species richness in both sites. Conclusions Different responses to fire frequency between KNZ and URF are likely linked to the dominant species and their characteristic traits, including height and method of clonal reproduction, with the rhizomatous growth form of A. gerardii dominating the North American grassland. South Africa does not have an equivalent grass species; instead, a range of tufted, non‐rhizomatous species dominate across the fire frequency treatments at URF . Reductions in soil N due to frequent fire did not appear to be a common mechanism driving responses in community composition in these two grasslands.