Evaluating the factors determining long-term post-landslide forest recovery using a spatially explicit forest landscape model

Landslides are major natural disturbances of mountain forests worldwide. Although topographical, biological, edaphic, and landscape factors affect post-landslide forest recovery, their relative importance remains unknown. Additionally, there is no spatially explicit model capable of projecting post-landslide forest recovery by incorporating the environmental heterogeneity among landslide slopes. These challenges make it difficult to develop effective forest restoration plans immediately after a landslide. We evaluated the important characteristics of landslide slopes for aboveground biomass and tree taxonomic composition recovery after shallow landslides. We modified the forest landscape model LANDIS-II to represent forest recovery depending on the environmental heterogeneity of landslide slopes and assessed the simulation results using random forest analysis. The amount of residual vegetation, slope angle, and seed dispersal of pioneer species will mainly determine aboveground biomass and tree taxonomic composition in 2100. Cohorts established after landslides and legacy cohorts respond differently to slope angle and seed dispersal. These contrasting responses and the contribution of legacy cohorts to recovery illustrate the differences in aboveground biomass. Establishment-related variables were more important than growth-related variables. Combined with most landslide slopes having little residual vegetation, the initial establishments from seed dispersal were important even for long-term forest recovery. For natural forest restoration, our results imply that (1) retaining residual vegetation enables rapid forest recovery; (2) forests can recover even on gentle landslide slopes with abundant seed dispersal of pioneer species without residual vegetation; and (3) steep landslides with little residual vegetation and seed dispersal of pioneer species require >100 years for recovery.