Ecology, more than antibiotics consumption, is the major predictor for the global distribution of aminoglycoside-modifying enzymes

Thank you for submitting your article "Ecology, more than antibiotics consumption, is the major predictor for the global distribution of aminoglycoside-modifying enzymes" for consideration by eLife. Your article has been reviewed by 2 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Meredith Schuman as the Senior Editor. The reviewers have opted to remain anonymous.

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission. In your revision, please directly address the essential revisions. You may draw on the individual reviews to help you address these, but a point-by-point response to each reviewer comment is not needed.

Essential Revisions:

1) The authors should justify their focus on aminoglycosides as opposed to other classes of antibiotics.

2) The reviewers expressed concerns regarding the fact that aminoglycosides were rarely used in humans over the period considered in the study. The authors should address how this is possibly impacting their conclusions regarding antibiotic consumption in humans as a factor driving the spread of aminoglycoside resistance.

3) The authors should provide additional statistical analysis on the possible impact(s) of sampling biases on their conclusions. For example, the authors could use subsampling techniques to evaluate to what extent their results and conclusions are due to biases in their dataset.

1. We have added a paragraph in the introduction to justify our focus on aminoglycosides as opposed to other classes of antibiotics in the introduction (see lines 88-102). On the one hand, aminoglycosides still are widely used as a topic treatment of otitis in humans, as a second-intent drug for patients infected by multi-resistant bacteria, and they still represent a large share of antimicrobials used for farm animals. Thus, aminoglycoside resistance is still a threat to human health, especially when facing a rise of resistance to first-intent treatments, and a threat to food production. Moreover, since most aminoglycosides originate from soil-dwelling bacteria, AMEs should spread between at least three ecological contexts (hospitals, farms, and soil). They potentially represent a good model for resistance to antibiotics that also originated from the environment (e.g. β-lactams).

2. Following suggestions made by Reviewer #2, we have performed complementary analyses of the dataset to properly test the impact of aminoglycoside consumption on the distribution of aminoglycoside resistance genes. First, data on farm animal populations in Europe were collected, in order to normalize aminoglycoside consumption by capita rather than by land area (see lines 993-1011). Second, we had tried to palliate the lack of antibiotic pollution data by assuming that antibiotic concentration remain unchanged as they diffuse from one biome to the other, but Reviewer #2 showed us that this assumption had very little support in the literature. Thus, we created a subset of the European dataset, only including the biomes in which aminoglycoside consumption data should be reliable (clinical, human, and farms samples, see lines 307-334 in the Results section). We argue that restricting the analysis to these biomes should have allowed us to draw reliable estimates of the effect of aminoglycoside consumption on the spread of aminoglycoside resistance. We thus performed an additional model selection on this subset. This new analysis still supports our initial conclusion that variations in aminoglycoside consumption are very weakly associated with variations in AME-gene-carrying bacteria prevalences.

3. The binomial regressions we had performed on the European dataset (as well as on the worldwide dataset) already included a correction for low sampling: this type of models allows the precision of estimates to increase with sample size (the proportion confidence interval being correlated with sqrt(1/n)). However, as argued by Reviewer #1, there might have been sampling biases depending on ecology (e.g. Cyanobacteria being more sampled in water, Enterobacteriaceae in clinical environments, etc.), which could have affected our conclusions. We have thus performed complementary analyses of the European dataset to test this possibility. We splitted the European dataset into two subset: one containing only the most sampled species (relative to each biome), and the other containing the least sampled species (see lines 274-305 in the Results section). By performing two distinct model selections on these two datasets, we found that sample biases did have the effect of overestimating the contribution of ecology to explaining the spread of AMEs. However, our main conclusion still holds true: ecology still is the variable explaining most variation in the prevalence of AME-gene-carrying bacteria, especially compared to the effect of aminoglycoside consumption.

Additionally, as suggested by Reviewer #1, we added Figure 4—figure supplement 2 and Source Data File 3 so readers may have a better idea of sampling biases in our dataset.

4. For better clarity, we simplified the list of acronyms. Moreover, we renamed the list of clusters of AMEs, so that they are grouped by their broad biochemical function (N-acetyltransferase, nucleotidyltransferase, or phosphotransferase).

5. We rephrased several sentences in the discussion that could have suggested that our results would be applicable to antibiotic resistance in general and/or that antibiotic stewardship efforts could be unnecessary.