In the past two decades, an increasing number of studies have been performed to analyse the relationship between biodiversity and ecosystem functioning. One conclusion that can be drawn from the research of the last years is that a low diversity in an assemblage is associated with a lowered mean (and an increase in the variance) in many of the ecosystem variables investigated. This is supported not only by individual experiments but also by meta-analyses in the past few years (e.g. Balvanera et al. 2006, Ecology Letters 9: 1146-1156 and Cardinale et al. 2006, Nature, 443: 989-992). One major shortcoming of the current state of knowledge is, however, that for many of the observed biodiversity effects we do not know the underlying mechanisms. Species complementarity, for example, is often deduced only by statistical analyses of species performance in monocultures and mixtures. While the concept of complementarity relies on differences among species in their ecological niches, in particular with respect to resource uptake and resource conversion into biomass, the traits underlying such niche differentiation have rarely been identified. We also do not know the relative importance of niche complementarity for very different traits, e.g. soil resources vs. pathogen or herbivore resistance. Similarly, the selection effect, where individual species dominate and perform better than expected in high-diversity mixtures, relies on differences between species in relevant traits that need to be identified. In general, a high trait diversity is thought to be conducive to complementarity, but at the same time increases the likelihood of including extreme species capable of driving a selection effect. It is currently not known how these two mechanisms depend on the type and magnitude of functional diversity. Associational mechanisms such as facilitation or mutualism also require trait differences that are relevant for species interactions. Without the identification of mechanisms it will be difficult for biodiversity science to be both predictive and quantitative. In our session, theoretical and empirical approaches to the analysis of mechanisms underlying the biodiversity-ecosystem functioning relationship will be presented and discussed. We aim at a mixture of empirical and theoretical talks. After an overview over proposed mechanisms, talks will address, among other topics, resource use complementarity in plant-plant interactions, complementarity mediated by multitrophic interactions or by above-belowground interactions. Theoretical approaches include recent approaches to the statistical analysis of trait space.