3.4 Disentangling the community assembly processes of abundant and rare taxa
The values of βNTI varied partly from -2 to +2 for abundant taxa, while rare taxa exhibited βNTI values higher than 2 (Fig. 4a). Null model analysis revealed that stochastic processes, especially dispersal limitation (DL, 84%) were the most important processes responsible for the community assembly of abundant taxa, followed by heterogeneous selection (HeS, 15%) (Fig. 4b). By contrast, the deterministic processes were largely responsible for the community assembly of rare taxa (79%), of which the greatest contributor was HeS (77%, Fig. 4b). Taken together, these results suggest that rare taxa are more environmentally constrained and less limited by dispersal than their abundant counterparts.
Mantel tests were performed to examine the relationships between βNTI and the variation of environmental variables in order to infer the relative influences of deterministic and stochastic assembly processes in the abundant and rare subcommunities. Results of Mantel tests demonstrated that forest area and TN were the most important variables associated with βNTI of the abundant (r = 0.116, P = 0.005) and rare (r = 0.171, P = 0.002) subcommunities, respectively (Table 1). Furthermore, the changes in forest area and TN were significantly correlated with pairwise comparisons of βNTI of the abundant (R2 = 0.029, P < 0.001, Fig. S5a) and rare (R2 = 0.041, P < 0.001, Fig. S5b) subcommunities, respectively. Despite the βNTI values of the abundant subcommunity being significantly related to differences in forest area, the distribution of βNTI values were mainly in the region of stochastic assembly (−2 < βNTI < +2). Yet the distribution of βNTI values of the rare subcommunity were mainly in the region of HeS (βNTI > +2) with the changes in sediment TN content.