Tree height-diameter and yield functions for Gmelina arborea (roxb.) stand in Edondon Gmelina plantation, Cross River state, Nigeria

Abstract

This study involved developments of models for predicting tree heights and stem volumes for Gmelina arborea stand at Edondon plantation in Obubra Local Government Area (LGA) of Cross Rivers State in southern Nigeria. Simple random sampling technique was adopted for plot location. Forty (40) 25 m × 25 m (0.16ha) temporary plots were used. The tree growth variables measured on all trees with diameter at breast height (Dbh) ≥10 cm included Dbh; diameters at the base, middle and top as well the tree total height. Data were analyzed using descriptive statistics and regression analysis. Linear, logarithmic, polynomial, power and exponential height-diameter and stem volume models were fitted to the dataset. The predictor was tree Dbh (cm). The developed models were assessed using coefficient of determination (R²) and root mean square error (RMSE). The significance of each of the models was evaluated using one-way analysis of variance. Model validations were done using t-test and mean bias. The results for the individual tree growth variables revealed that the mean Dbh, height, stem volume and basal area were 36.31±14.36 cm, 20.99±6.46 m, 0.12±0.09 m² and 0.96±0.61 m³, respectively. On stand basis, the mean basal area and stem volume were 1.99 m²/ha and 15.41m³/ha, respectively. All the height-diameter models presented in this study had very high R² low RMSE values. Moreover, the selected models were significant (P <0.05). Among the models, logarithmic model was the best with R²; RMSE values of 0.91; 0.1293, and the least -suitable model was the exponential with R²; RMSE values of 0.75; 2.2683. All the yield models tried in this study consistently gave poor results with very low R² values. The best among the stem volume models was power model with R²; RMSE values of 0.40; 0.3787, and the least-suitable was the polynomial with R²; RMSE values of 0.19; 0.8399. The result of model validation for the heightdiameter functions revealed that there were no significant differences in mean observed and the predicted height values under liner, logarithmic, polynomial and power functions (P>0.05). However, mean observed and predicted tree height values significantly differed under exponential function (P< 0.05). Results of model validations for yield models showed that there were no significant difference between the observed and the predicted stem volumes under the linear and the logarithmic models (P>0.0). Nevertheless, mean observed and the predicted stem volumes significantly differed under the polynomial, power and exponential functions (P<0.05). The result revealed that the mean basal area per hectare in the stand were below the recommended values for a well-stocked forest or planted stand in Nigeria.

Keywords: Tree height, stem volume, stand density, prediction models