While coexistence in plant communities is frequently explained by effects of resource niche partitioning, the Janzen-Connell (J-C) hypothesis is an alternative approach that has been assumed as a major ecological mechanism explaining high species richness levels, in particular, in tropical forest ecosystems. Central components of the J-C hypothesis are non-competitive effects of density- and distance-dependence, thereby two drivers that contribute independently to species coexistence, but that are ultimately linked in the field. Here, we make use of the forest Biodiversity-Ecosystem Functioning project in subtropical China (BEF-China) to estimate density and distance dependence effects by means of a reciprocal tree seedling transplant experiment. Using monocultures of ten and seven tree species, respectively planted at two different sites, juveniles of all species were grown in their own (home) and in all other monocultures (away), thereby testing for distance effect, just as in three different levels of planting density, testing for density effects. In addition, we repeated a similar set-up in a nearby common garden experiment, where we added a 'shadow' treatment to simulate different light conditions induced by the canopy layer. The general aim of the common garden experiment is to test for additional density dependent intraspecific competition for light in the absence of host-specific agents necessary for J-C effects. Density dependent patterns that are prevalent in the main experiment, while not being prevalent in the common garden experiment can be considered true J-C effects (see "Common Garden Experiment: Seedling addition experiment - growth and biomass data" for further details)

We implemented a reciprocal transplant experiment with juvenile plants using monocultures of 11 different tree species and following a home-away design, i.e. planting transplants in their own monoculture (home) and in a different one (away). The species represent a subset of the Main Experiment’s species pool and had been selected according to their representation at the two sites and finally included 6 evergreen and 5 deciduous species. From the species pool, 10 species were implemented at site A and 7 species at site B, thus 6 species were used on both sites. For each species, 3 monoculture plots were selected at each site and used for subplot implementation according to the project-specific experimental setup. The general design of BEF-China provides two ‘true’ replicates of each species’ monocultures at the plot level (i.e. where all 20 * 20 planted individuals belong to the same species). As a third replicate, we used 2-species mixtures including the respective target species and therein, all subplots were established in between four individuals of the target species. All site-specific target species were planted in suplots either established in their own monoculture (home: 3 replicates each) or in monocultures of the other species within site (away: 9 species * 3 replicates, i.e. 27 plots site A and 18 for site B, respectively). Three monocultures in total, (1 in site A and 2 in site B) could not be realised. This approach was used to test for distance dependence. In order to test for effects of density-dependence, each subplot, was then subdivided in split-plots where tree seedlings were planted in conspecific communities at different density levels of 4, 9 and 25 individuals each. Accordingly, in target plots at site A, each five subplots (hosting each two species) were implemented, whereas at site B, the number of subplots per plot was lower because of the lower number of species tested. Thus, in each 29 and 19 plots (for site A and B, respectively), 5 and 4 subplots (for site A and B, respectively) were randomly allocated with each subplot 1.5 m x 1.0 m in size. Each splitplots occupied an area of 0.5 x 0.5 m in size. Species × density combinations were randomly allocated within subplot.

Plots of the main experiment on Site A and B