3.2 Geographic distribution patterns
The taxonomic and phylogenetic similarities of both abundant and rare taxa exhibited significantly negative relationships with corresponding geographic distance (all P < 0.05, Fig. 2a and 2b). However, the slopes of distance-decay of compositional and phylogenetic similarity were steeper for abundant taxa (Bray-Curtis, 0.0084; βMNTD, 0.0013) than those for rare taxa (Bray-Curtis, 0.0038; βMNTD, 0.0003), indicating a stronger influence of dispersal limitation for abundant taxa. Moreover, significantly higher taxonomic and phylogenetic β-diversity values were found in rare taxa than for the corresponding abundant taxa (Wilcoxon rank-sum test, P < 0.001, Fig. 2c and 2d). Additionally, significant correlations between community composition and phylogeny were observed in both the abundant and rare subcommunities (all P < 0.001, Fig. S3). The correlation in the rare subcommunity (R2 = 0.829) was much stronger than that in the abundant subcommunity (R2 = 0.149).
VPA results revealed that the pure effect of spatial variables was consistently higher than that of environmental factors for both the abundant (16.9% versus 3.5%) and rare (7.9% versus 1.5%) subcommunities (Fig. S4). The pure effect of land use types accounted for the smallest portions of the variability in both the abundant and rare subcommunities. However, of all the environmental variables and land use types examined, forest area by itself explained 7.7% of the variation in the abundant subcommunity, more than any of the other individual sediment variables (Table S3).