Species specific height diameter relationships of dominant tropical rainforest trees in Kakamega Forest Kenya

Abstract

Accurate height–diameter (H–DBH) relationships are fundamental for forest inventory, structural assessment, and biomass estimation in tropical forests, where direct measurement of tree height is often time-consuming, costly, and prone to error. Species-specific allometric models can improve the characterization of forest structure and support sustainable forest monitoring. This study aimed to develop and evaluate species-specific H–DBH models for three dominant tree species in Kakamega Forest, Kenya ( Pterocarya stenoptera , Spathodea campanulata , and Antiaris toxicaria ), and to examine interspecific differences in growth architecture relevant to forest monitoring and sustainable ecosystem management. The study was conducted in a 1-ha permanent plot in Kakamega Forest, western Kenya. Diameter at breast height (DBH ≥ 10 cm) and total tree height were measured using a laser hypsometer. Four candidate models, including two nonlinear models (Weibull and Chapman–Richards) and two linearizable models (Power and Logarithmic), were fitted and compared using adjusted R², root mean square error (RMSE), and Akaike Information Criterion (AIC). Model performance was evaluated using Leave-One-Out Cross-Validation (LOOCV), while bootstrap resampling was used to assess parameter uncertainty. Residual heteroskedasticity was evaluated using the Breusch–Pagan test and corrected where necessary. Model performance differed markedly among species. The Power model provided the strongest predictive performance for Spathodea campanulata (Adj. R² = 0.698) and Antiaris toxicaria (Adj. R² = 0.366). In contrast, all candidate models produced negative adjusted R² values for Pterocarya stenoptera , indicating that none adequately predicted tree height beyond the species mean within the sampled dataset. Cross-validation confirmed substantial interspecific differences in predictive accuracy, while bootstrap analyses indicated greater parameter uncertainty for P. stenoptera , consistent with its small sample size and greater variability in the H–DBH relationship. The results demonstrate contrasting species-specific H–DBH patterns, with S. campanulata showing a stronger height–diameter relationship, A. toxicaria exhibiting comparatively lower explanatory power but lower absolute prediction error, and P. stenoptera showing a weak and highly variable relationship. The species-specific H–DBH models developed in this study provide locally calibrated tools for improving characterization of forest structure in Kakamega Forest. Enhanced precision in forest structural assessment may support future biomass estimation, forest monitoring, and carbon-accounting applications. By strengthening local forest assessment frameworks, the study contributes to evidence-based approaches aligned with SDG 13 (Climate Action) through improved forest monitoring and SDG 15 (Life on Land) through enhanced understanding of forest structure and sustainable ecosystem management. Species-specific H–DBH relationships varied considerably among dominant tropical tree species in Kakamega Forest, demonstrating that model performance is species dependent and that generalized allometric equations may not adequately capture local growth patterns. The developed models provide locally derived tools for forest structural assessment and establish a foundation for future biomass estimation and forest monitoring applications following independent validation.