Accurate estimation of tree biomass and carbon stocks is essential for evaluating the ecological and economic contributions of agroforestry systems. This study developed allometric equations for estimating component-wise and total biomass and carbon in shea tree based on destructive sampling of 30 trees in Sudanian parklands of Burkina Faso. The objectives were to (i) assess biomass allocation among tree components, (ii) develop allometric models for aboveground and belowground biomass, and (iii) estimate carbon stocks based on tree-level biomass and wood carbon content. Total aboveground biomass ranged from 36.4 to 489.2 kg tree(-)1, and belowground biomass from 9.7 to 142.3 kg tree(-)1. The root-to-shoot ratio averaged 0.29 +/- 0.07 and decreased significantly with increasing crown area, indicating a size-dependent shift in carbon allocation. Generalized Additive Models for Location, Scale, and Shape (GAMLSS) yielded robust models (pseudo-R2 = 0.95 for AGB, 0.89 for BGB) with diameter, height, and crown area as key predictors. Mean carbon concentration was 48.9%, and tree-level carbon stock ranged from 23.2 to 309.5 kg. These equations provide species-specific tools for carbon quantification in dryland agroforestry systems and can support national reporting and land restoration strategies in West Africa.