Allometric modeling for fine woody debris biomass estimation in the Gorgan's Shastkolateh Beech Forests

Background and objective: Fine woody debris (FWD) is a significant reservoir of carbon and bioenergy in the Hyrcanian forests. Therefore, monitoring changes in FWD biomass on the forest floor is essential for sustainable management, particularly concerning carbon dioxide emissions and carbon trading in both international and domestic markets. The primary objective of this research is to develop optimal allometric equations for accurately monitoring FWD in one of the beech (Fagus orientalis Lipsky) communities within the Hyrcanian forests of Iran.
Methodology: This study was conducted in the Shastkolateh research forest in Gorgan County, Iran. A random stratified method was employed for direct sampling and weighing of FWD. The FWD was categorized into three diameter classes: 1-2.5 cm, 2.5-4.5 cm, and 4.5-7.5 cm. For each diameter class, length and degree of decay were measured randomly in pure beech and mixed beech stands. Total weight was measured destructively using digital scales. To estimate biomass values of FWD, a power function allometric model based on curve estimation analysis was utilized. The middle diameter, length, dry wood density, and degree of decay of the weighed FWD were introduced as explanatory variables in the model input matrix. These explanatory variables were combined in various ways as independent variables in a log-transformed model. Model validation was conducted using adjusted regression coefficients (Adj.R²), variance inflation factors (VIF), and residual mean square (RMS). A t-test evaluated the significance of parameters in each model, while the fitting of a Loess curve between standardized residuals and estimations was used to assess the validity of the models.
Results: Model development indicated that incorporating middle diameter (D) and length (L) of FWDs into the exponential function of the log-transformed model increased variance explanation from 38% to 73% and reduced RMS from 0.6 to 0.26. Adding wood density (ρ) did not significantly enhance model accuracy or validity. Including degree of decay as a dummy variable slightly improved model accuracy, resulting in the top-ranked model with the highest variance explanation (Adj.R² = 0.78) and lowest RMS (0.22). Furthermore, introducing the biomass surrogate (D² × L × ρ) along with degree of decay considerably increased accuracy, yielding a second-ranked model (Adj.R² = 0.75; RMS = 0.24). Although some parameters were not statistically significant (P > 0.05), no collinearity was detected among the models presented (VIF < 5). The goodness-of-fit data indicated that the Loess curve remained linear around the zero line, with changes in residuals aligning with estimation changes around this line. Findings suggested that biomass estimation for Diospyros lotus L. and unidentified species may be unreliable due to standardized residuals falling outside the zero line range, indicating interdependence in error estimation for these species. Conversely, biomass estimates for F. orientalis and Carpinus betulus L. appeared reliable, with variations not being interdependent.
Conclusion In summary, biomass distribution and changes in FWD values exhibit significant variations based on middle diameter and length. Due to limited ranges of decay and dry wood density, these additional variables did not substantially enhance accuracy in log-transformed models. Approximately 22% to 25% of variance remains unexplained when estimating FWD biomass using selected optimal models, particularly regarding errors (standardized residuals). This discrepancy is most pronounced for D. lotus and unidentified species. Overall, the models developed in this research for pure beech and mixed beech-hornbeam forests demonstrate broad applicability within the study area.
 
Keywords: Bio-energy, carbon trade, carbon sink, fine woody debris, optimum allometric model.