Abstract: Using 16 representative shrub species from the Zhangguangcai Mountain region of Heilongjiang Province, we applied Pearson correlation analysis to identify factors significantly associated with biomass, and subsequently developed predictive models using both dummy-variable and mixed-effects approaches. Our results show that the mixed-effects model consistently outperformed the dummy-variable model in both goodness of fit and predictive accuracy, enabling more precise estimation of shrub biomass. Elevation and crown area emerged as the key determinants of biomass variation. Although dummy-variable models are often regarded as advantageous when categorical predictors are limited, in this study the mixed-effects framework demonstrated superior performance.