Comparison of repeated sample measurements for Ochratoxin A and Total Aflatoxin in pistachio lots

Abstract Molds producing aflatoxin and Ochratoxin A (OTA) are naturally present in the production, handling and/or processing environments for pistachios. Despite extensive investments by US pistachio processors to eliminate these toxins, under select conditions, these molds can produce mycotoxins that contaminate pistachios at trace levels. While the distribution of aflatoxin in pistachio lots (when present) from repeated samples is well understood given long standing regulatory testing requirements for this mycotoxin, such data for OTA had not been studied, as this mycotoxin has only been more recently regulated globally. As such, the distribution of OTA in pistachio lots after repeated sampling, versus equivalent data collected for aflatoxin, was compared. OTA sample test results for 750 United States pistachio lots being considered for European Union (EU) export were collected across three commercial labs in California. Every lot considered had at least two samples submitted for OTA testing. Sample size was a function of lot size and if the nuts were in-shell or shelled, per export guidelines. The variation in repeated sample measures for OTA, as measured by standard deviation (SD) was directly proportional to corresponding means and well described ( ) by a straight line. Comparison with an equivalent plot derived from repeated samples of 4,367 pistachio lots for aflatoxin, found a similar linear response ( ). The two linear fits of SD vs mean for OTA and Total Aflatoxin were not statistically different. This equivalency means that operating characteristic (OC) curves derived from published pistachio aflatoxin sampling data should also effectively describe OTA sampling data in pistachio. Published OC curves for pistachio/aflatoxin were accessed via the online Food and Agricultural Organization (FAO) mycotoxin sampling tool and compared to experimental data. The total variance for aflatoxin at 5 μg/kg predicted by the FAO tool agreed well with experimental data.