Geographical and interannual variability in biomass partitioning in grassland ecosystems: a synthesis of field data.

Biomass partitioning is an important variable in terrestrial ecosystem carbon modeling. However, geographical and interannual variability in f(BNPP), defined as the fraction of belowground net primary productivity (BNPP) to total NPP, and its relationship with climatic variables, have not been explored. Here we addressed these issues by synthesizing 94 site-year field biomass data at 12 grassland sites around the world from a global NPP database and from the literature. Results showed that f(BNPP) varied from 0.40 to 0.86 across 12 sites. In general, savanna and humid savanna ecosystems had smaller f(BNPP) but larger interannual variability in f(BNPP), and cold desert steppes had larger f(BNPP) but smaller interannual variability. While mean f(BNPP) at a site decreased significantly with increasing mean annual temperature and precipitation across sites, no consistent temporal response of f(BNPP) with annual temperature and precipitation was found within sites. Based on these results, both geographical variability in f(BNPP) and the divergent responses of f(BNPP) with climatic variables at geographical and temporal scales should be considered in global C modeling.