Questions: One tenet of biodiversity and ecosystem functioning research is the dependence of ecosystem functions on species-specific differences in traits. Community assembly theory predicts niche differentiation among the constituent species, for which there is as yet no empirical evidence for subtropical forest ecosystems.We thus ask: (1) Which mechanisms promote sapling coexistence in the four abundant early-successional subtropical species Schima superba, Elaeocarpus decipiens, Quercus serrata and Castanea henryi? (2) Do species richness, species composition, species identity and stand density have an impact on growth patterns and crown architecture of tree saplings?

In a field experiment, we manipulated the local neighbourhood of saplings with regard to species richness (1, 2 and 4 species), species composition (monocultures, six two-species combinations and one four-species combination) and stand density (low, intermediate and high).

Pilot Experiment 29°06'N 117°55' E To analyse biomass allocation patterns with regard to stratification (i.e. height layers) and to different constitutents (stem, branches and leaves), the four central individuals per plot were harvested in September 2010 in 50 cm strata starting from ground. Saplings were divided into stem, branches and leaves for each stratum. Biomass was dried at 70° C for 48 h and weighed to 0.01 g precision. Biomass data were logarithmically transformed prior to analyses.To analyse the vertical above-ground biomass distribution, we calculated the cumulative biomass fraction C, i.e. the proportion of cumulative above-ground biomass, summed up from the ground to the height strata hs (50, 100, 150, 200, 250 cm).

trees have been planted in 2009 and harvested in autumn 2010

4 Species: Schima superba, Elaeocarpus decipiens, Quercus serrata, Castanea henryi

The medium density plots of block 4 were harvested in July 2010 by SP1.

The overall aim of this study was to disentangle neighbourhood effects on growth, biomass allocation, crown architecture and branch demography of saplings. Firstly, the complete dataset was used to test for diversity effects by fitting mixed effects models (Model 1) including species richness and density as factorial variables and the initial diameter as fixed effect. The initial diameter was used to account for differences in size at the beginning of the experiment. Secondly, all two species combinations were analysed for species composition. Mixed effects models (Model 2a) were fitted using species composition, density and initial diameter as fixed effect. The analyses with Model 2b were performed for the high density treatment data divided by species to exclude density effects and to test for composition effects on the individual-level of each species. Model 2b contained species composition and initial diameter as fixed effects. Thirdly, mixed effects models (Model 3) for all monocultures were calculated to test for species identity. They were fitted by the predictor variables species identity, density and initial diameter as fixed effects. Random effects for all models were plot nested in block. Model simplification was performed by stepwise backward selection of fixed factors, removing the least significant variables until only significant predictory variables remained (p < 0.05). The significant categorical variables were further examined by a Tukey post-hoc test.