As highlighted during the first round of review, this is a very nice study presented in a great manuscript. The authors have addressed the drawbacks noticed by the reviewers and corrected typos. The article is suitable for publication, I recommend it!

This is an important paper, I would even say a much awaited paper, that paves the way for environmentally-relevant ecotoxicological studies on the effects of pesticide mixtures on wildlife. As the authors correctly point out, although several ecotoxicological studies have examined pesticide mixtures, the pesticides in the mixtures are usually selected based on general use over large territories or countries, or targeting some crop types, and may not correspond to relevant mixtures to which particular species of wildlife are exposed. The approach that the authors take is highly innovative and useful. It makes use of the postcodes of pesticide buyers for 279 different molecules in France. After making a number of well explained and reasonable assumptions, they group the 5,361 postcodes in 18 groups that are discriminated by a small number of pesticides. Some pesticides, like glyphosate, are found in most or all 18 groups and called core pesticides, but others are associated to just one or a few of these groups and termed discriminant pesticides. The approach allows associating each of the 18 groups to a specific crop, or assemblage of crops, hence, for example, differenciating Brittany from the Southern regions, where vineyards are concentrated. 

The limitations of the study are well explained and dealt with in the cautious interpretation of the data. For example, it may be that a pesticide purchased in a given postcode area at a given date is not used in the year or in the postcode in which it was purchased. The authors eliminated chemicals purchased by the SNCF that centralizes its purchases for use across the French territory. Overall, the data are credible and the conclusions prudent. 

The power of this study is that this approach will allow ecotoxicologists to study mixtures of pesticides to which a given species is likely exposed in a given part of its distribution range. The approach taken by the authors is sufficiently well described to be applied by others in other regions and countries where a database of pesticide sales is kept.

From an ecotoxicological perspective, this study is exciting and opens new avenues to research on the impacts of pesticide use at the landscape level. The approach could even be coupled with georeferenced data on biodiversity to further orient studies of the effects of these mixtures. 

I do not have the competence to make a judgment on the statistical approach, and that part will need to be assessed by someone with the relevant expertise.

Finally, the style is clear and the English is excellent. A pleasure to read. I attach a pdf to indicate a few corrections.