Rubber (Hevea brasiliensis) plantations constitute the largest artificial ecological forest systems in tropical regions of China, while long-term monoculture has significantly reduced biodiversity, particularly among insect communities. Rubber-based agroforestry systems are widely recognized as a promising approach to improving ecosystem functionality. However, the mechanisms by which different intercropping patterns affect insect community dynamics remain poorly understood. This study systematically evaluated the effects of eight rubber-based agroforestry systems on insect community diversity, functional group composition, and associated environmental drivers. Using rubber monoculture as a control, seven rubber-based agroforestry systems were investigated from April 2024 to March 2025. A total of 94,483 insect individuals belonging to 16 orders, 222 families, and 1560 species were recorded. The results indicate that the rubber–fig (Ficus hirta) and rubber–banana (Musa nana) agroforestry systems supported higher insect richness, diversity, and community stability than other systems, while the more complex rubber–coconut (Cocos nucifera)–fig (Ficus hirta) system exhibited a relatively lower value. Functionally, herbivores dominated the rubber monoculture system. The moderately grazed rubber–forage grass (Brachiaria eruciformis)–black goat agroforestry system promoted predators and detritivores, whereas the rubber–konjak (Amorphophallus bulbifer) agroforestry system attracted more omnivores. The permutational multivariate analysis of variance revealed that insect species composition was primarily negatively driven by canopy cover (R² = 14.65%) and management intensity (R² = 11.54%). The ecological benefits of rubber-based agroforestry systems depend not only on crop species diversity but also on vegetation structural complexity and management practices. It is recommended to promote the rubber–banana and rubber–fig agroforestry systems as optimized models and to enhance insect-mediated ecosystem services by maintaining understory vegetation structure, regulating canopy cover, and implementing low-intervention management practices.

Large-scale tropical forest conversion to plantations or cropland severely impacts soil health. However, most studies have relied on pairwise designs, single-type chronosequences, or multi-type comparisons that are often confounded by environmental heterogeneity, thereby overlooking the complexity of successive land-use transitions on a common land-use background. Here, we investigated the responses of soil physicochemical properties and microbial communities to two conversion pathways (NSF → ORF → YRF; NSF → ORF → URF → AL) within a single farm, where shared soil parent material and climate allowed for a direct comparison of land-use types with different conversion pathways. Soil properties and microbial diversity were analyzed using physicochemical methods and high-throughput sequencing, with samples collected from five land-use types (six biological replicates per type). Results showed NSF had the highest soil properties. Along NSF → ORF → YRF, TOC, DOC, TN, IN and C/N declined gradually. In contrast, URF exhibited severe degradation, with AL showing partial recovery. Bacterial diversity remained relatively stable, whereas fungal communities were highly sensitive, with diversity decreasing progressively and dominant phyla shifting with intensified disturbance. DOC, EC and SM were most strongly associated with microbial community ordination, with pH showing no significant association. URF had the highest predicted abundance of C/N/S cycling genes, while NSF maintained a balanced fungal trophic structure. This study provides a comparative assessment of soil properties and microbial communities along two tropical land-use conversion pathways, contributing empirical evidence for sustainable land-use management in tropical regions.