Bayesiari regression of shelling stock pistachio prevalence in link with aflatoxin contamination and climate

摘要

Aflatoxin contamination in pistachios is a threat for human health due to the fact that these mycotoxins, produced by Aspergillus fungi, have genotoxic carcinogenic effects, with no threshold. Climate and geographic conditions are very important determinants of the prevalence and distribution of Aspergillus species. Aflatoxin contamination level and incidence rate in pistachios is related to climate, as the contaminated pistachios are mainly found among the ones with a ruptured hull. When the hull ruptures in the orchard, the kernel is exposed to invasion by Aspergillus fungi and insects that are contamination vectors. More precisely, the hull rupture incidence is possibly in link with temperature and/or rainfall during a specific period. The aim of this work was to assess this link thanks to Bayesian simulations. The Bayesian method was used for regression analysis, as it improves classical estimators in terms of the precision of estimates. The hull rupture incidence under study results in lower value pistachios after processing, as they consist of dark stained and shell damaged nuts, categorized into shelling stock pistachios. The shelling stock data used were the ones from US Pistachio production yearly statistics by county from 2005 to 2012. The prior distributions of the parameters were calculated for years 2005 to 2010 and were combined to the likelihood equation. The result was a posterior estimate of the parameters, using data from 2011 and 2012. Posterior estimates were obtained for counties having a steppe climate: Kern, Fresno, Kings and Madera. These preliminary results show that the Bayesian method improves the estimation of model parameters in comparison to classical statistical analysis. The method is going to be extended to other counties in order to determine the link between climate and shelling stock pistachios.