Summary 1We established monocultures and 16 unique mixed communities of 12 native grasses, legumes and non-N 2 -fixing forbs. We identified species having a greater or lesser yield in the mixed communities than expected from monoculture data as winners and losers, respectively. To test our hypothesis that performance of the subordinate species (losers) is mainly controlled by light availability, whereas the dominant species (winners) are sensitive to the availability of below-ground resources, we traced the effects of number of species, light transmission in the community and foliar N concentration on plant photosynthesis through leaf C isotope composition ( δ 13 C). 2 Phalaris arundinacea and Phleum pratense , the two tallest grass species, yielded more in mixtures than expected, as, initially, did Dactylis glomerata . Festuca ovina , the smallest grass, and Ranunculus acris , a forb, had smaller yield in mixtures than expected. For most species, observed mixture yields did not deviate significantly from those expected. 3 Decreases in transmitted light decreased δ 13 C in D. glomerata , Lotus corniculatus and Rumex acetosa . The δ 13 C of Trifolium pratense and L. corniculatus was affected by increasing number of species in the plant community even after accounting for the transmitted light. In P. arundinacea , δ 13 C increased with increasing foliar %N, as expected for the tallest, dominant species. 4 Species showing a positive, significant relationship between δ 13 C and transmitted light were relatively low growing and unable to establish dominance in multispecies communities due to shading by larger dominants. 5 We conclude that above-ground competition is crucial in determining C isotope composition among the subordinate species, whereas the dominant species are more strongly affected by below-ground resources. Different factors thus dictate the physiological performance of species according to the size-distribution hierarchy in the community.