Tree height prediction models for two forest reserves in Nigeria using mixed-effects approach

Abstract

Height-diameter models for predicting tree height are essential for routine forest inventory. These models can be developed using fixed-or mixed-effects approach. Few studies have applied the mixed-effect approach to developed height prediction model for the natural forest in Nigeria. Therefore, in this study, the mixed-effect modelling approach was used to develop height prediction models for Ikrigon and Cross River South (CRS) Forest Reserves, Nigeria. Data consist of 776 and 438 height-diameter pairs from Ikrigon and CRS Forest Reserves, respectively. Five 2-parameters and five 3-parameters height-diameter models were evaluated including Nalund, Wykoff, Curtis, Meyer, Michaelis-Menten, Chapman-Richards, Ratkowsky, Korf, Logistic and Gompertz. Model fitting was done in two stages: Fixed-effect approach was used in the first stage wherein candidate models were selected and refitted in the second stage using mixed-effect approach. Adjusted coefficient of determination, root mean square error, mean absolute bias, Akaike information and Bayesian information criterion were used to assess the models. The results showed that Gompertz and Meyer models were more consistent. Gompertz and Meyer had adjusted coefficient of determination, root mean square error, mean absolute bias, Akaike information criterion and Bayesian information criterion of 0.642, 4.457, 3.501 and 4591.487, 4638.028; and 0.638, 4.482, 3.541, 4592.008, 4619.933, respectively for Ikrigon and 0.724, 4.076, 3.215, 2536.148 and 2576.970; and 0.711, 4.176, 3.273, 2536.352 and 2560.845, respectively for CRS. The mixed-effect approach improved tree height predicting of the forest stands. These models are recommended for estimating tree height in the forest reserves.