COS 10-4 - Agents on the spatial patterns and species richness of natural forest communities in South Korea

Abstract: Understanding the spatial patterns of dominant vegetation types is essential for comprehending ecosystem components. Thus, the imprecise distribution boundaries generated by some expert opinions and limited data make it difficult for proactive conservation strategies to be effective. In this study, we clarified the boundaries of forest communities in South Korea and analyzed the factors affecting them. We asked: 1) Can the dominant tree species serve as effective indicators for delineating major forest communities at the national level? 2) If so, what are the abiotic drivers that affect the distribution and species richness of each community?

We used the fuzzy c-means clustering algorithm using a national forest type map containing 32 main tree species classes to improve the forest community map. Using a classification tree from R software with the package “evtree”, we identified key bioclimatic variables that determine the habitat of plants. We also investigated how the proportions and species richness of forest communities differed depending on the topography, soil properties, and impact of human activities in the areas where each community is located.

The forests were classified into four groups, each with predominant tree species: Pinus densiflora(most common pine), Pinus thunbergii(coastal pine), oaks, and other broadleaved tree species. The primary climate predictors, which are the summer precipitation and winter temperature, derived potential distributions that well distinguish coastal forest from inland. The difference in elevation was statistically significant, and oaks were in the highest zones, followed by each group of broadleaved trees, P. densiflora, and P. thunbergii. High-altitude and north-facing areas had relatively higher species richness than lower and south-facing areas. The limestone area had a high proportion of pine trees, whereas the gneiss area had a high proportion of oaks and other broadleaved trees. In areas with deep ( > 60cm) and moist soils with a high silt content, oaks and other broadleaved trees were found to predominate over pines. In the plains where human activities appear, P. thunbergii, evergreen broadleaved trees, and broadleaved trees were widespread. Meanwhile, in the mountains, more pines were found in areas closer to human development, and more oaks were found further away from those areas.

Our results provide more precise information for interpreting the distribution of plant species at the community level across the landscape. Moreover, this approach can offer an objective system for classifying plant communities and analyzing the impacts of abiotic factors, and can further help predict future range of communities.