Ecological relevance in honeybee pesticide risk assessment: developing context-dependent scenarios to manage uncertainty

We reported 1 year ago (Henry et al., 2012a) that a sublethal dose of thiamethoxam, a neonicotinoid pesticide used on some common flowering crops, reduces the ability of exposed honeybees to find their way back to the colony. More recently, Guez (2013) have raised concerns about the relevance of our study design. The obvious intention of the author was to discredit our study, both by seeking after inconsistencies in calculations and gathering new arguments against the experimental dose of thiamethoxam. In both case, we would like to carry out major rectifications of his statements. We fully agree that there is still place for improvement in the way one may estimate pesticide exposure levels in honeybees, and Guez (2013) brought interesting thoughts in that direction. However, this should be done with respect to the recent advance on that topic (ANSES, 2012; Cresswell and Thompson, 2012; EFSA, 2012a,b, 2013; Henry et al., 2012b) and with respect to the need for hierarchizing the sources of uncertainty. Herein, we show that the overestimation issue raised by Guez has already been addressed before and with an even greater magnitude of uncertainty. But at the heart of the debate is the way to deal with uncertainty in estimates of honeybee field exposure to pesticides. In that respect, we further underline (1) the need to properly hierarchize uncertainties using conventional ecological scaling approaches and (2) the need to investigate context-dependent exposure scenarios, whereby worse-case situations are explicitly linked with their spatial or temporal occurrence likelihood. Our main concern about Guez’s comment is that the conclusions on the peer review of the risk assessment for bees, published consecutively to our study, were basically ignored (ANSES, 2012; EFSA, 2012a,b, 2013). The updated expertise and field measurements of nectar residues intake returned exposure levels of 0.10–0.33 ng.bee−1.day−1 (ANSES, 2012), 0.184–5.888 ng.bee−1.day−1 (EFSA, 2012a) and 0.553–2.903 ng.bee−1.day−1 (EFSA, 2013) for winter oilseed rape treated with thiamethoxam. Based on these results, one must admit that uncertainties in field exposure levels are so high that they should be the prime focus of risk assessors interested in refining honeybee exposure scenarios. Depending on the considered scenario, our 1.34 ng dose may actually overestimate by a factor of 13.4 or underestimate by a factor of 4.4 the real field exposure level. This actually overwhelms the uncertainty levels raised by Guez due to temporal variations in sugar content (<6-fold overestimations, Figure 1A in Guez, 2013) and due to spatial variations in daylight time available for foraging (Figure 1B in Guez, 2013). Although Guez made valuable efforts to challenge our study design, there is a higher level of uncertainty to deal with first.