Fitting nonlinear double sigmoid models to macauba fruit growth data using Bayesian inference

ABSTRACT: Macauba (Acrocomia aculeata) is a palm tree native to the tropical regions of the Americas. It is used as a food and applied in the cosmetic and pharmaceutical industries. Macauba fruit is a source of raw material for oil and bioenergy production. Determining the ideal time to harvest macauba fruit is essential to ensure maximum quality and optimize the post-harvest and storage processes. The present study characterized the accumulation of total dry mass and mesocarp dry mass of macauba fruit over time. We used Bayesian inference to estimate the parameters of the nonlinear double logistic and double Gompertz regression models to find the model that best describes the data. Fruit samples were collected from one week after spathe opening and continuing weekly until natural fruit drop, at 62 weeks after anthesis, and dried in a forced ventilation oven. Statistical analyses were conducted in the R software, and the weighted resampling method was used. The Criterion of density Predictive Ordered (CPO) and deviance information criterion (DIC) were used to select the models. The double Gompertz model best fitted the data. The fruit mesocarp showed no difference in growth rate between the first and second stages of development. In the final stage, it reached values close to 9 g. The total dry mass showed faster growth in the first stage, with values close to 34 g in the final stage.