Human activities on the Tibetan Plateau have severely degraded native ecosystems, necessitating effective restoration strategies such as turf transplantation. However, the early-stage ecological impacts of this method, particularly microbial responses to polygonal soil cracks at turf interfaces, remain poorly understood. We conducted a 3-year field experiment in southeastern Tibet to compare microbial resource limitation and use efficiency between normal turf (NT) and degraded polygonal crack areas (DT). Results showed that turf transplantation reduced vegetation cover (by 19.5–41.3%) and diversity compared to natural grasslands, while polygonal cracks expanded 4.8-fold by the third year. Soil pH declined, but carbon (175% increase in DOC) and nitrogen availability rose in NT, contrasting with nutrient depletion in DT. Enzyme stoichiometry revealed intensified microbial carbon limitation in both NT and DT, shifting from C-N limitation to C-P limitation in NT and from C-P limitation to C-N limitation in DT. Carbon use efficiency (CUE) and nitrogen use efficiency (NUE) decreased by 12–18% in DT as crack area increased, driven by enzyme-mediated nutrient imbalances and microbial metabolic trade-offs. Structural equation modeling indicated polygonal cracks altered enzyme stoichiometry ratios, directly reducing CUE and NUE. These findings highlight that polygonal cracking disrupts microbial resource partitioning, exacerbating nutrient constraints and impairing carbon sequestration during early-stage turf restoration. Mitigating crack formation is critical for enhancing the sustainability of alpine grassland rehabilitation.