Review Reports - Resident Perceptions of Mosquito Problems Are More Influenced by Landscape Factors than Mosquito Abundance

Round 1

Reviewer 1 Report

The manuscript by Brown and colleagues offers important insights regarding how environment can influence resident perceptions of pest issues. The work is interesting and important but a few critical details, especially regarding the mosquito trapping protocol, require attention. First, the authors need to explain what mosquito traps were used and with what lures. Were the ‘complaint traps’ the same type? What was the relative proportion of complaint traps used (and how did that vary across the different neighborhoods)? If complaint traps were more common in some types of neighborhoods - how was potential bias addressed?

Second, the seasonality of the trapping intensity/frequency must be detailed. Mosquito populations are seasonal and the growth curves might differ (temporally) or between species across an urban gradient such as described. The current analysis assumes that the same part of the seasonal cycle is captured at all sites (and that differences in mosquito species across space are negligible). It is not possible from the current methods text to deduce if these assumptions are plausible. Further, the 1 km radius is quite large for some species and a spatial map showing how abundance varies across the study landscape would be useful.

Many of the key results are open to interpretation and the manuscript would be strengthened if additional information was available to help move from plausible correlation to some greater evidence for causative relationships – which are perhaps over stated in the discussion. For example, youth and lower education both came out as important variables in the model but it isn’t clear if these younger and/or lower ed respondents were dispersed across neighborhoods or clustered in a location. Discussion includes reasons why youth might be more prone to mosquito exposure but this assumes some random sampling of youth across the spatial template (across which mosquito habitat is also likely variable/clustered). I think the addition of spatial maps showing variation in the dependent variable, as well as counts and some other demographics might be needed. If lower education for example is persistently a positive predictor of perception across neighborhoods and not just because a low income/socio-economic status neighborhood is driving that response then the implications for importance are different.

Other points for consideration:

Figure 1. The inset (PWR) neighborhood might be easier to understand if it was visibly connected or extended from the larger map (reduce readers’ need to search).

L151 typo ‘til’

L164 grammar – seems to be disagreement in use of plurals

L175 grammar – seems to be disagreement in use of plurals

L 179 typo ‘a’

L227 40% returned surveys is a decent return rate. How variable was this across different neighborhoods? Are the survey results likely to be dominated by responses from specific types of neighborhoods/demographics? It looks like # was ~20-60 surveys but is this just due to population size or to return rate?

L333 Typo ‘who’

L351 Typo ‘within across’ ?

L392 Missing ‘as’

Author Response

Thank you very much for your thoughtful and helpful comments. Please see below for how we addressed each comment. Additionally, we have attached a document with all reviewer comments and changes so you may have a fuller contextualization of the changes to the manuscript.

Reviewer 4 Comments and Responses:

Author’s Response: Thank you for this comment. We have added additional information about the trapping methods in the paper to clarify these questions. We used the per trap per trap day metric to standardized across the study since traps were set out for different lengths. We have re-written the section to make it clearer that the mosquito abundance information was calculated as the average number of mosquitoes per trap per trap day.

Author’s Response: We would like to thank the reviewer for this thoughtful comment. The routine traps are rotated on a schedule, so each region is sampled for approximately the same amount of time across the year. Since we are averaging across all trap days, the effects of seasonality will be normalized across all traps. We also added language to the methods to clarify trapping methods. As to the second comment, we confirmed with experts about the range of mosquito dispersal and found several papers suggesting 1 km is a reasonable dispersal distance for misquotes. We have clarified this in the text. As to a species distribution map, this is outside of the scope of this research as we are specifically interested in linking social survey data to data collected from Maricopa County vector control. Finally, we also clarified that the 1 km radius matches what many survey respondents consider to be “around their home” and this helps us better link responses to landscape variables.

Author’s Response: We thank the reviewer for these comments. We have softened the language throughout the text to emphasize that our results are all correlative. Since we do not have a mechanistic link, this work is meant to highlight correlations and provide possible explanations that are grounded in previous research and theory. As to the correlation between variables, one of the checks we ran prior to running our boosted regression tree was to look for correlation between variables which we did not find. We also added information to the text to make this more apparent. Additionally, we selected BRT because this method builds multiple regression trees from subsamples of the entire dataset and is robust to spatial autocorrelation. Finally, information on the spatial distribution of these variables across the study neighborhoods is available in the 2017 PASS report and PASS data is also publicly available. We site both the data and the report in this study for those interested in a closer examination of the data.

Figure 1. The inset (PWR) neighborhood might be easier to understand if it was visibly connected or extended from the larger map (reduce readers’ need to search).

Author’s Response: We have remade figure 1 to show multiple study neighborhoods as well as to make the blow out maps clearer and easy to find.

L151 typo ‘til’