Agroforestry ecosystems are an efficient strategy for enhancing soil nutrient conditions and sustainable agricultural development. Soil extracellular enzymes (EEAs) are important drivers of biogeochemical processes. However, changes in EEAs and chemometrics in rubber-based agroforestry systems and their mechanisms of action are still not fully understood. Distribution of EEAs, enzymatic stoichiometry, and microbial nutrient limitation characteristics of rubber plantations under seven planting patterns (RM, rubber monoculture system; AOM, Hevea brasiliensis-Alpinia oxyphylla Miq; PAR, Hevea brasiliensis-Pandanus amaryllifolius Roxb; AKH, Hevea brasiliensis-Alpinia katsumadai Hayata; CAA, Hevea brasiliensis-Coffea Arabica; CCA, Hevea brasiliensis-Cinnamomum cassia (L.) D. Don, and TCA, Hevea brasiliensis-Theobroma Cacao) were analyzed to investigate the metabolic limitations of microorganisms and to identify the primary determinants that restrict nutrient limitation. Compared with rubber monoculture systems, agroforestry ecosystems show increased carbon (C)-acquiring enzyme (EEAC), nitrogen (N)-acquiring enzyme (EEAN), and phosphorus (P)-acquiring enzyme (EEAP) activities. The ecoenzymatic stoichiometry model demonstrated that all seven plantation patterns experienced C and N limitation. Compared to the rubber monoculture system, all agroforestry systems exacerbated the microbial limitations of C and N by reducing the vector angle and increasing vector length. P limitation was not detected in any plantation pattern. In agroforestry systems, progression from herbs to shrubs to trees through intercropping results in a reduction in soil microbial nutrient constraints. This is primarily because of the accumulation of litter and root biomass in tree-based systems, which enhances the soil nutrient content (e.g., soil organic carbon, total nitrogen, total phosphorus, and ammonium nitrogen) and accessibility. Conversely, as soil depth increased, microbial nutrient limitations tended to become more pronounced. Partial least squares path modelling (PLS-PM) indicated that nutrient ratios and soil total nutrient content were the most important factors influencing microbial C limitation (−0.46 and 0.40) and N limitation (−0.30 and −0.42). This study presented novel evidence regarding the constraints and drivers of soil microbial metabolism in rubber agroforestry systems. Considering the constraints of soil nutrients and microbial metabolism, intercropping of rubber trees with arboreal species is recommended over that of herbaceous species to better suit the soil environment of rubber plantation areas on Hainan Island.