Contributions of woody and herbaceous vegetation to tropical savanna ecosystem productivity: a quasi-global estimate

To estimate the relative contributions of woody and herbaceous vegetation to savanna productivity, we measured the C-13/C-12 isotopic ratios of leaves from trees, shrubs, grasses and the surface soil carbon pool for 22 savannas in Australia, Brazil and Ghana covering the full savanna spectrum ranging from almost pure grassland to closed woodlands on all three continents. All trees and shrubs sampled were of the C-3 pathway and all grasses of the C-4 pathway with the exception of Echinolaena inflexa (Poir.) Chase, a common C-3 grass of the Brazilian cerrado. By comparing the carbon isotopic compositions of the plant and carbon pools, a simple model relating soil delta C-13 to the relative abundances of trees + shrubs (woody plants) and grasses was developed. The model suggests that the relative proportions of a savanna ecosystem's total foliar projected cover attributable to grasses versus woody plants is a simple and reliable index of the relative contributions of grasses and woody plants to savanna net productivity. Model calibrations against woody tree canopy cover made it possible to estimate the proportion of savanna productivity in the major regions of the world attributable to trees + shrubs and grasses from ground-based observational maps of savanna woodiness. Overall, it was estimated that 59% of the net primary productivity (N-p) of tropical savannas is attributable to C-4 grasses, but that this proportion varies significantly within and between regions. The C-4 grasses make their greatest relative contribution to savanna N-p in the Neotropics, whereas in African regions, a greater proportion of savanna N-p is attributable to woody plants. The relative contribution of C-4 grasses in Australian savannas is intermediate between those in the Neotropics and Africa. These differences can be broadly ascribed to large scale differences in soil fertility and rainfall.