Light partitioning is often invoked as 2 a mechanism for positive plant diversity effects on ecosystem functioning. Yet evidence for an improved distribution of foliage in space or time in 4 diverse plant communities remains scarce and restricted mostly to temperate grasslands. Here we identify the mechanisms through which tree species diversity affects community-level light capture 6 in a biodiversity experiment with tropical trees that displays overyielding, i.e. enhanced biomass production in mixtures. Using a combination of empirical data, mechanistic models and 8 statistical tools, we develop innovative methods to test for the isolated and combined effects of architectural and temporal niche differences among species as well as plastic changes in crown 10 shape within species. We show that all three mechanisms enhanced light capture in mixtures and that temporal niche differences were the most important driver of this result in our seasonal 12 tropical system. Our study mechanistically demonstrates that niche differences and phenotypic plasticity can generate significant biodiversity effects on ecosystem functioning in tropical forests. 14 Keywords: Biodiversity; Ecosystem functioning; Phenology; Niche differences; Tree; Plantation; Light competition; Crown; Sardinilla project; Phenotypic plasticity; Intraspecific diversity; 16 Overyielding

The influence of biodiversity on ecosystem functioning is now well established. However, our ability to predict the ecological consequences of biodiversity changes remains limited by our poor understanding of the mechanisms underlying biodiversity effects. We disentangled the contributions of light competition and residual neighborhood interactions in a 10‐year‐old biodiversity experiment with tropical trees that display overyielding, i.e., higher community‐level yields in mixtures compared with monocultures. We developed models of individual tree growth that partition the effects of neighboring trees into shading and residual effects assumed to reflect primarily belowground interactions. These models reject the hypothesis that reduced light competition in mixtures is the only mechanism driving overyielding. After factoring out the effects of shading, litter production by neighbors was a far better predictor of tree growth than traditional crowding indices; it contributed to overyielding by producing pairwise interactions that ranged from competitive to facilitative, but which, on average, concentrated competition within species. Consistent with litter‐mediated biodiversity effects, the magnitude of overyielding increased over time. Our results provide evidence for diversity effects extending beyond that of light and reveal the neglected role of litter‐mediated interactions among trees.

No abstract