Abstract: Alnus cremastogyne is a predominant tree species in subtropical mountain swamp ecosystems, recognized for its considerable ecological and economic importance. This fast-growing species exhibits a pronounced capacity for carbon sequestration, playing a vital role in regional carbon cycling. Accurate assessment of its biomass is essential for understanding and projecting the carbon sink potential of mountain swamp ecosystems under ongoing regional climate change. In this study, field surveys and laboratory analyses were conducted to measure the biomass of various components of standard A. cremastogyne specimen trees. Using multivariate nonlinear least squares regression, we developed and compared two categories of predictive models: relative growth models and compatible biomass models, designed to estimate biomass at both the component and whole-tree levels. Each model’s predictive accuracy was rigorously evaluated using statistical metrics such as root mean square error (RMSE), mean absolute error (MAE), and the coefficient of determination (R²). Results indicate that relative growth models achieved higher accuracy for estimating specific biomass components, including leaves, roots, branches, stems, crowns, wood, and bark, when modeled independently. Among these, leaf biomass was best represented by a univariate power function of the form W=aD^b . For roots, branches, stems, and crowns, a bivariate power function, W=aD^bH^c , provided the best fit. Wood and bark biomass were most accurately estimated by a bivariate power model of the form W=a(D^2H)^b . In contrast, the compatible biomass model demonstrated superior performance in predicting total individual tree biomass, with the univariate form W=aD^b yielding the highest simulation accuracy. This model ensures the additivity of component biomasses, enhancing reliability in whole-tree assessments. By systematically constructing and comparing these model types, this study identifies the optimized approaches for biomass estimation in A. cremastogyne. The findings provide a robust theoretical foundation for improving the accuracy of forest carbon stock quantification and support the scientific management of subtropical mountain swamp ecosystems under climate change scenarios. This research provides a comprehensive evaluation of biomass modeling approaches for A. cremastogyne in subtropical mountain swamp ecosystems. The findings identify optimal modeling strategies for different biomass components and establish a robust theoretical foundation for improving the accuracy of forest carbon stock quantification. The study offers practical insights for ecosystem management and enhances our capacity to assess carbon sequestration potential in these vulnerable ecosystems, supporting more reliable carbon accounting and informed climate change mitigation strategies in subtropical regions. The demonstrated methodology and model selection framework have broader applications for biomass estimation in similar wetland forest ecosystems worldwide.