Forest structure analysis and yield models for tree species in Okeluse Forest Reserve, Ose LGA, Ondo state, Nigeria

To attain a sustainable forest management decision, accurate information is needed on structure and growth of major forest components. This information can be derived by using functional forest models. Therefore, the study involved formulating site-specific models for yield prediction in Okeluse Forest Reserve, Ondo State, Nigeria. Systematic sampling technique was adopted with 20 1-km-long transects laid at 500 m intervals. Four plots (50 × 50 m) were laid alternately at 250 m intervals, along each of the transects. A total of 80 sample plots were used for the study. All trees with diameters (Dbh) ≥10 cm were measured for diameter, height, stem quality and crown length. All measured trees were identified to species-level. Data were analysed, using descriptive statistics and regression analysis. The yield-prediction models were evaluated, using root mean square error (RMSE), , Akaike information criterion (AIC) and F-test. Models were validated, using paired-sample t-test and absolute bias. Seventeen tree species, belonging to 12 families, were encountered. Lophira alata occurred most with RD of 15.9%. Four (4) of the species encountered (i.e., Afzelia africana, Entandrophragma cylindricum, Lophira alata and Pterygota macrocarpa) were vulnerable. Seven (7) of the tree species (i.e., Albizia lebbeck, Anogeissus leiocarpus, Antiaris toxicaria, Blighia sapida, Ceiba pentandra, Mansonia altissima and Triplochyton scleroxylon) were least concern. The mean species importance value index was 75.74±25.59. Average basal area and stem volume were 31.47±8.91 m2 and 125.89±35.64 m3/ha, respectively. Most of the trees had low slenderness coefficient, indicating excellent stability and low susceptibility damage. The best yield-prediction model was of the form: V= -55.234 + 0.9863(lnDbh) -2.114BA2 + 0.3492MHT with R2, RMSE, AIC and absolute bias of 88.5%, 1.02, 57.62 and 0.210, respectively. The model also showed no significant difference in the mean observed and predicted yield values. It is therefore recommended for future predictions in the study area.