ABSTRACT

Crown diameter measurements provide invaluable information for species management, but are left unassessed during inventory because of associated difficulties in measurement, and problems of underestimation from limited visibility. This study investigates the relationship between crown diameter (cd) and diameter at breast height (dbh) of Parkia biglobosa stands in Oyo State, Nigeria, to aid its management. Crown diameter, dbh and stem height (ht) were measured on each Parkia stand, and the collected dataset was randomly split into training and validation datasets, in ratio seven to three. Pairwise correlation analysis was conducted using Pearson’s correlation while test of normality and homogeneity of variance were conducted using Shapiro-Wilk’s (SW) and Goldfeld-Quandt’s (GQ) test at 0.05 alpha level. The relationship between cd and dbh was stronger (r = 0.79) than with cd and ht (r = 0.38). A linear model between cd and dbh had lower Akaike’s Information Criterion (AIC) value (AIC = 3285.55) than when ht was included in the model (AIC = 3287.77). Consequently, ht was excluded from the model. Residual for the cd and dbh model was normal (SW = 0.72, p = 0.652) but was not homoscedastic (GQ = 1.25, p-value = 0.016) when Ordinary Least Squares (OLS) was used. Consequently, the Weighted Least Squares (WLS) and Maximum Likelihood Estimation (MLE) methods of estimation were included to estimate the model’s parameters. Standard Error of the Estimate (SEE) and Root Mean Square Error (RMSE) values were smallest for the WLS estimates (i.e. SEE = 15.871, RMSE = 13.414) compared to the estimates of OLS (i.e. SEE = 18.912, RMSE = 13.448) and MLE (i.e. SEE = 440.130, RMSE = 371.98), respectively which was affirmed by independent samples t-test (p-value = 0.5581). The WLS method of estimation would, therefore, provide a more reliable fit when homoscedasticity is violated.