Glyphosate losses through various stages of coffee production and consequences for human exposure.

Abstract

Green coffee beans, rejected for commercial use because of glyphosate contamination, were examined to monitor their glyphosate levels from harvest, through roasting, until various coffee extractions. The green beans, Arabica and Robusta, exhibited glyphosate levels above the EU-MRL (0.14-0.21 mg/kg), representing a worst-case scenario. The beans were roasted to different degrees and subsequently used for different coffee preparations. As a result of roasting (>200 °C), glyphosate contents were reduced, frequently by more than 73%. Remarkably, up to 9% of initial glyphosate was removed together with the silverskin, already at lower temperatures. Filtered and instant coffee beverages prepared from respective coffee samples resulted in virtually quantitative glyphosate transfer. Glyphosate transfer was significantly less for espresso, and ristretto, apparently due to the reduced amounts of water used for extraction. Aminomethylphosphonic acid (AMPA) was not detectable on any process level, confirming that AMPA is not a thermal glyphosate degradation product. In conclusion, compelling evidence is provided that glyphosate contamination becomes considerably reduced during roasting, whereas beverage preparation contributes at best to a minor further reduction. In consequence, even unusually high initial glyphosate loads in green beans are strongly reduced by the roasting process, resulting in a final cup content of <0.4 µg.