The significant role of tropical agroforestry (AF) systems in the global carbon budget has increased the need for accurate biomass estimates. The absence of biomass allometric equations limits our understanding of tree biomass and carbon stocks. Most existing allometric equations are developed for natural forest, which restricts their application to the fruit tree species commonly found in homegardens AF systems. This limitation hampers precise biomass estimation, which is essential for effective climate change mitigation. This study aimed to develop species-specific and mixed-species aboveground biomass (AGB) models for the dominant fruit trees in the Agroforestry Systems of the Upper Gibe Region of Ethiopia. We destructively harvested 96 sample trees representing four dominant species, with diameters at breast height (DBH) ranging from 2.5 to 62.8 cm. The models were formulated using DBH, tree height (Ht), and wood basic density (WBD) as predictor variables. We evaluated model performance based on parameter significance, the Akaike Information Criterion (AIC), pseudo-R2, Root Mean Square Error (RMSE), and Mean Absolute Error (MAE). Results indicated that DBH alone was the most effective predictor for Persea americana (AGB = 0.365 × (DBH)2.029). In contrast, other species exhibited better prediction performance when additional variables were considered. For Mangifera indica (AGB = 0.315 × (DBH)2.0922 × (WBD)0.192) and the mixed-species dataset (AGB = 0.839 × (DBH)1.835 × (WBD)0.331), models that combined DBH and wood density (WBD) yielded the highest predictive accuracy. Conversely, Prunus persica (AGB = 0.792 × (DBH)1.907 × (Ht)0.022 × (WBD)0.305) and Psidium guajava (AGB = 0.439 × (DBH)1.604 × (Ht)0.646 × (WBD)0.455) were best predicted using models that incorporated DBH, tree height (Ht), and wood density. Overall, species-specific and mixed-species models outperformed regional and pantropical equations, underscoring the limited effectiveness of generalized models for agroforestry trees. Additionally, multivariable models yielded greater predictive accuracy than single-variable approaches. These results enhance the estimation of biomass and carbon stocks in agroforestry systems. In conclusion, this study presents robust, locally calibrated allometric equations that enhance biomass and carbon assessments. We recommend applying these models for regional carbon accounting, land management, and climate initiatives like REDD + , rather than relying on pantropical equations.

Biomass allometric models for dominant fruit trees in the agroforestry systems of the upper gibe region of Ethiopia

期刊论文
发表日期：

2026年1月27日

查看全文

Gadisa Demie,
Tsehay Tufa

摘要

The significant role of tropical agroforestry (AF) systems in the global carbon budget has increased the need for accurate biomass estimates. The absence of biomass allometric equations limits our understanding of tree biomass and carbon stocks. Most existing allometric equations are developed for natural forest, which restricts their application to the fruit tree species commonly found in homegardens AF systems. This limitation hampers precise biomass estimation, which is essential for effective climate change mitigation. This study aimed to develop species-specific and mixed-species aboveground biomass (AGB) models for the dominant fruit trees in the Agroforestry Systems of the Upper Gibe Region of Ethiopia. We destructively harvested 96 sample trees representing four dominant species, with diameters at breast height (DBH) ranging from 2.5 to 62.8 cm. The models were formulated using DBH, tree height (Ht), and wood basic density (WBD) as predictor variables. We evaluated model performance based on parameter significance, the Akaike Information Criterion (AIC), pseudo-R², Root Mean Square Error (RMSE), and Mean Absolute Error (MAE). Results indicated that DBH alone was the most effective predictor for Persea americana (AGB = 0.365 × (DBH)^2.029). In contrast, other species exhibited better prediction performance when additional variables were considered. For Mangifera indica (AGB = 0.315 × (DBH)^2.0922 × (WBD)^0.192) and the mixed-species dataset (AGB = 0.839 × (DBH)^1.835 × (WBD)^0.331), models that combined DBH and wood density (WBD) yielded the highest predictive accuracy. Conversely, Prunus persica (AGB = 0.792 × (DBH)^1.907 × (Ht)^0.022 × (WBD)^0.305) and Psidium guajava (AGB = 0.439 × (DBH)^1.604 × (Ht)^0.646 × (WBD)^0.455) were best predicted using models that incorporated DBH, tree height (Ht), and wood density. Overall, species-specific and mixed-species models outperformed regional and pantropical equations, underscoring the limited effectiveness of generalized models for agroforestry trees. Additionally, multivariable models yielded greater predictive accuracy than single-variable approaches. These results enhance the estimation of biomass and carbon stocks in agroforestry systems. In conclusion, this study presents robust, locally calibrated allometric equations that enhance biomass and carbon assessments. We recommend applying these models for regional carbon accounting, land management, and climate initiatives like REDD + , rather than relying on pantropical equations.