Odile T. Bruggisser

Studies on biodiversity–ecosystem functioning (BEF) in highly controlled experiments often yield results incompatible with observations from natural systems: experimental results often reveal positive relationships between diversity and productivity, while for natural systems, zero or even negative relationships have been reported. The discrepancy may arise due to a limited or closed local species pool in experiments, while natural systems in meta-community contexts experience dynamic processes, i.e., colonization and extinctions. In our study, we analysed plant community properties and above-ground biomass within a semi-natural (i.e., not weeded) experiment in an agricultural landscape. Eleven replicates with four different diversity levels were created from a species pool of 20 wildflower species. We found an overall significant negative relationship between total diversity and productivity. This relationship likely resulted from invasion resistance: in plots sown with low species numbers, we observed colonization by low-performing species; colonization increased species richness but did not contribute substantially to productivity. Interestingly, when analysing the biomass of the sown and the colonizer species separately, we observed in both cases positive BEF relationships, while this relationship was negative for the whole system. A structural equation modelling approach revealed that higher biomass of the sown species was linked to higher species richness, while the positive BEF relationship of the colonizers was indirect and constrained by the sown species biomass. Our results suggest that, in semi-natural conditions common in extensive agroecosystems, the negative BEF relationship results from the interplay between local dominant species and colonization from the regional species pool by subordinate species.