Origin context of trait data matters for predictions of community performance in a grassland biodiversity experiment

Abstract

Plant functional traits may explain the positive relationship between species richness and ecosystem functioning, but species‐level trait variation in response to growth conditions is often ignored in trait‐based predictions of community performance. In a large grassland biodiversity experiment (Jena Experiment), we measured traits on plants grown as solitary individuals, in monocultures or in mixtures. We calculated two measures of community‐level trait composition, i.e., community‐weighted mean traits (CWM) and trait diversity (Rao's quadratic entropy; FD) based on different contexts in which traits were measured (trait origins). CWM and FD values of the different measurement origins were then compared regarding their power to predict community biomass production and biodiversity effects quantified with the additive partitioning method. Irrespective of trait origin, models combining CWM and FD values as predictors best explained community biomass and biodiversity effects. CWM values based on monoculture, mixture‐mean or community‐specific trait data were similarly powerful predictors, but predictions became worse when trait values originated from solitary‐grown individuals. FD values based on monoculture traits were the best predictors of community biomass and net biodiversity effects, while FD values based on community‐specific traits were the best predictors for complementarity and selection effects. Traits chosen as best CWM predictors were not strongly affected by trait origin but traits chosen as best FD predictors varied strongly dependent on trait origin and altered the predictability of community performance. We conclude that by adjusting their functional traits to species richness and even specific community compositions, plants can change community‐level trait compositions, thereby also changing community biomass production and biodiversity effects. Incorporation of these plastic trait adjustments of plants in trait‐based ecology can improve its predictive power in explaining biodiversity–ecosystem functioning relationships.