Model forms used to predict tree biomass.

Dry Afromontane forests in Ethiopia are crucial for carbon sequestration; however, the absence of robust biomass and carbon stock estimation models hinders accurate assessment. This study addresses this limitation by developing and validating site-specific, multispecies biomass estimation models for Wof-Washa plantation and natural forests. Biometric data were collected from 127 harvested trees representing seven dominant species from both plantation and natural forests. Aboveground biomass (AGB) was regressed against diameter at breast height (DBH) as the sole predictor, with stepwise inclusion of total height (H), crown area (CA), and wood density (ρ). Weighted nonlinear least squares regression was performed to fit new models for each forest, their performance was evaluated using the root mean square error (rRMSE), pseudo-R2, and relative mean prediction error (rMPE %). The best-selected model using DBH and H explained 90% and 95% of the variation in the AGB of plantation and natural forests, respectively. This model produced the lowest bias (rMPE = 5.9% for plantation and 2.5% for natural forests) compared to pan-tropical models. Our findings demonstrated that our optimal model provides accurate AGB predictions at plot and landscape levels. This confirms that the models can provide sufficiently reliable estimations of carbon stocks, indicating the potential for national carbon accounting and thereby enhancing decision-making in the study forests. Therefore, the findings of this research contribute directly to enhancing the accuracy of carbon dynamic monitoring and supporting sustainable forest management, a crucial component in global efforts to combat climate change.