Species-specific allometric models were developed to estimate aboveground biomass in scrub forests of Avicennia germinans on the Ajuruteua Peninsula, northeastern Amazon coast. Structural characterization of thousands of individuals revealed distinct groups that differ in height, stem diameter, density, and dominance, reflecting strong environmental filtering associated with high salinity and limited water availability. Destructive sampling combined with the evaluation of multiple structural predictors enabled the formulation of robust allometric equations, with diameter and canopy-related variables emerging as the most informative for biomass estimation. Biomass distribution differed markedly among structural groups, emphasizing the role of internal forest heterogeneity in shaping carbon storage patterns. These models reduce uncertainties in biomass assessments for Amazonian dwarf mangroves and provide a stronger methodological foundation for quantifying carbon stocks in environmentally stressed coastal ecosystems, improving the accuracy of blue carbon evaluations and supporting monitoring efforts in regions susceptible to climatic and hydrological change.