A meta-analysis using a logit non-linear mixed effects model for ‘Hass’ avocado postharvest performance data

Abstract

‘Hass’ avocado fruit quality data collected from storage trials in New Zealand during the 2002, 2003 and 2004 seasons have been assimilated into a model to describe relationships between the postharvest environment (storage temperature and duration), fruit dry matter at harvest, and the incidence of pathological and physiological disorders in fruit after cold storage. Similar or related disorders were grouped into three categories: disorders in unripe fruit, pathological disorders of ripe fruit (rots) and physiological disorders of ripe fruit. A logistic non-linear mixed effects model and its simplified version with normal errors on the logit linear predictor gave a good general description of relationship between postharvest disorders and dry matter at harvest, storage temperature and storage duration, but not for predictions of disorder for individual orchards. The capacity of the model to predict disorder incidence was limited largely by the variability among orchards and seasons. It is concluded that while the mixed effects model does describe well the relationships of the disorder categories with storage temperature, storage duration and dry matter, any predictions of disorder at the orchard level are likely to have large prediction errors because of the high variability among orchards and seasons. The challenge in the future is to determine, and incorporate into the model, those factors that contribute to the large variation among orchards and seasons.