Abstract

Aims. To determine the aggregation scales/sizes and elucidate the assembly mechanisms of local seedling and sapling communities in broad-leaved Korean pine forests.

Methods . Based on regeneration survey data from four broad-leaved Korean pine forest plots, Ripley's g(r) was fitted using a Thomas process with a two-nested-scale function. Hierarchical clustering was used to delineate local communities, followed by calculating the abundance-based and presence-based standardized effect size of mean pairwise leaf trait distances (SES.MPD) and the standardized effect size of species evenness (SES.E). The overall community characteristics were determined by evaluating the degree of distribution tendency (D) values of the three indicators above. Redundancy analysis (RDA) was performed to assess the relationship between local canopy environmental variables and community-weighted mean (CWM) traits.

Results. The small aggregation scale for local seedling and sapling communities was approximately 1 m, while the large scale was approximately 15 m. At the large scale, the seedling communities had 42.623 small patches (1.732 seedlings/patch) on average, and the sapling communities had an average of 20.329 patches (1.149 saplings/patch). Across the 1–15 m scales, both abundance-weighted SES.MPD and SES.E for the seedling and sapling communities were significantly greater than zero. The presence-based SES.MPD was significantly greater than zero across all scales for saplings but was only significant at smaller scales for seedlings. The values of D for the three indicators were slightly greater than one across all spatial scales. The seedling communities exhibited higher D values for abundance-weighted SES.MPD and SES.E than the sapling communities. Additionally, the seedling communities exhibited higher D values for presence-based SES.MPD at smaller aggregation scales, although this trend was reversed at larger scales. The RDA showed that total R2 (0.021–0.245) increased with the spatial scale and was consistently higher in the sapling communities than in the seedling communities. Compared with the seedling communities, the sapling communities demonstrated weaker or even opposite correlations in CWM.LA, SLA, and LW with vertical distribution tendency degrees and local coniferous LAI.

Conclusions. The assembly of local regeneration communities was dominated by stochastic processes. Among deterministic processes, limited dispersal leads to spatial and trait aggregation of regeneration communities across all spatial scales; canopy-driven environmental filtering accumulates with regeneration development, and limiting similarity has a more important role in local sapling communities than in seedling communities at smaller spatial scales. Differences in ontogenetic strategy lead to variation in trait – canopy relationships between local seedling and sapling communities.