Predicting mixed-species litter decomposition with community-aggregated means via the idiosyncratic hypothesis

Background/Question/Methods

Litter decomposition is a key process for understanding biogeochemical cycles. Despite several studies on decomposition rates, it remains difficult to provide large-scale predictions, especially due to the complexity of interactions between species. This study tests the accuracy of the biomass-ratio hypothesis for predicting the decomposition of multispecies litter mixtures, using community aggregated decomposition rates. Specifically, we ask (a) if the biomass-ratio hypothesis introduces a systematic over- or under-estimation of litter decomposition rate in mixture; (b) if so, what is the magnitude of this bias; and (c) if the degree of variability of this bias decreases as the number of species in the mixture increases. Based on leaf functional traits, we chose two sets (trees and herbs) of contrasted species (mixed in combinations of up to 6 species for trees, 4 species for herbs). Litterbags of these mixtures were placed in microcosms under controlled conditions. We then compared observed to expected (from single species rates) mass losses predicted from the biomass-ratio hypothesis, using a linear mixed-effect model. 

Results/Conclusions

For tree mixtures, we observed both positive and negative deviations from expectations across the mixtures that largely cancelled, giving only a small mean bias (decomposition rate 1.3% less than expected). This was not true in the herb mixtures, with strong antagonistic effects (decomposition rate 21.4% less than predicted). The variance in these deviations decreased with increasing species richness in the mixture for both herbs and trees. This "idiosyncratic annulment" allows more accurate predictions in mixtures with greater species richness. Thus, the choice of the spatial scale of a study will be determinant in the prediction of ecosystems processes, such as decomposition rates of litter.