Understanding the impact of biodiversity loss on ecosystem functioning and services is of fundamental and practical importance in ecology. Recent studies have generated several interesting findings on the mechanistic linkage between plant functional diversity (PFD) and ecosystem functioning based on experiments using synthetically assembled grassland communities. It remains uncertain, however, whether the results from artificial grasslands can be applied to natural ecosystems. To address this critical issue, here we report on the evidence from part of the ongoing Inner Mongolia Grassland Removal Experiment (IMGRE), which was designed to test how changes in PFD and composition of the community would affect underpinning ecosystem processes along a manipulated PFD gradient generated by removing combinations of plant functional groups (PFGs).

Results/Conclusions

Our results showed that, with the increasing number of PFGs removed: (1) Aboveground net primary productivity (ANPP) decreased while the interannual variability of ANPP increased linearly; (2) Effects of functional diversity and complementarity declined, but no significant relationship between selection effect and PFD was found during three consecutive years. For a given level of PFG loss, the magnitude and direction of both the complementarity and selection effects depended substantially on the PFG composition of the community. The loss of perennial bunchgrasses alone and in combination with other PFGs explained large amounts of variation in ANPP. Our experimental evidence also demonstrated that the strength of compensatory effects associated with nonrandom local extinctions was determined mainly by the differences in recruitment rates and dispersal ability of the remaining species or PFGs. By the third year, perennial rhizome grasses compensated for the loss of perennial bunchgrasses by increasing its population density and biomass production; whereas early successional annual species colonized the open space at low functional diversity plots.  Both perennial bunchgrasses and perennial forbs, in contrast, did not show a consistent compensatory effect for the loss of other PFGs, indicating a high recruitment limitation. Our findings have important implications for predicting ecological impacts of overgrazing and for restoring functional diversity, productivity, and resilience of heavily degraded grassland in Inner Mongolia.