Variations in dendrometric and fruiting characters of Vitellaria paradoxa populations and multivariate models for estimation of fruit yield

摘要

The fruit of Vitellaria paradoxa is an ideal raw material in cosmetic, pharmaceutical and confectionery industries. There are no accurate data on annual fruit yield due to the lack of objective assessment tools. The objectives of this study were to develop fruit yield prediction models based on dendrometric and fruiting variables, to examine variations in these variables between upland and lowland populations in Burkina Faso, and associations between these variables. A total of 191 fruiting trees were selected according to crown accessibility, and 17 dendrometric and fruiting variables were recorded. The fruit yield, expressed in number of fruits per tree, fresh and dry weights of fruits, was assessed by collecting fruits dropped overnight until the end of the fruiting period. Fruit yield prediction models were derived for each population using partial least squares regression. The results showed significant differences in dendrometric and fruiting variables between populations (P < 0.01). The lowland population had the highest values for most of the dendrometric variables while fruiting variables were the highest for the upland population. A strong significant correlation (P < 0.01) was found between number of shoots and fruiting variables. Within individual trees, fruit yield was lowest for the bottom part of the crown and the section of the crown with north-east orientation. Fruit yield parameters were successfully predicted based on selected dendrometric and fruiting variables (prediction error = 0.092 and 0.125 for upland and lowland populations, respectively). All fruiting variables, number of shoots and crown attributes had the highest influence on the models.