Simple and Multiple Linear Regression Models for Predicting Maturity of ‘Mendez#1’ and ‘Hass’ Avocado Fruit Harvested from inside and outside Tree Canopy Positions

Abstract

ABSTRACT This study evaluated the feasibility of using bio-markers to develop statistical models for predicting time to maturity of avocado fruit. The experiment used ‘Mendez#1ʹ and ‘Hass’ as model cultivars. Fruit was sampled bi-weekly from two positions within the tree canopy, namely, inside and outside canopy. Nine fruits per canopy position were sampled from fifteen trees of each cultivar. Mesocarp dry matter content (DM), oil content (OC), D-mannoheptulose and perseitol were assessed from 182 to 294 days after full bloom. Simple and multiple linear regression models were successfully developed to predict the maturity parameters of ‘Mendez#1ʹ and ‘Hass’ avocado fruit. Oil content was the most reliable index for predicting the DM due to the strongest positive and significant correlation (r = 0.76; p < .001). Furthermore, OC remained the only best fit in the DM model after backward elimination that was performed. The DM and oil content (OC) were the most successful variables with a strongest positive and significant (r > 0.7; p < .001) correlation with the number of days to reach maturity, suggesting that these parameters can be used to predict the days to maturity. Therefore, a farmer who keep a record of time after full bloom can use the number of days to predict DM and OC, before starting actual maturity tests. The predictive models from this study can be used to predict OC, DM and time (days) to maturity. D-mannoheptulose and perseitol had non-significant predictive values (p > .05), thus did not fit in both models of predicting avocado maturity.