Review 3: "Yellow Fever in Ghana: Predicting Emergence and Ecology from Historical Outbreaks"

RR:C19 Evidence Scale rating by reviewer:

• Strong. The main study claims are very well-justified by the data and analytic methods used. There is little room for doubt that the study produced has very similar results and conclusions as compared with the hypothetical ideal study. The study’s main claims should be considered conclusive and actionable without reservation.

***************************************

Review: Recurring outbreaks of Yellow Fever (YF) in Ghana are driven by a confluence of human and ecological factors and represent a high priority public health risk to Ghana and the region. Drawing on the historical record, the authors draw inferences about the conditions under which zoonotic outbreaks have occurred and identify regions at risk of future outbreaks. Subnational risk maps can inform public health policy, including efficient resource deployment and improved surveillance of both human populations and animal reservoirs.

This manuscript well-organized and clearly presented. Particular strengths include careful attention to local knowledge, a nuanced understanding of the statistical methods, and a thoughtful ecological analysis and interpretation of both the study system and the principle results. This study fills important gaps in the disease ecology of sub-Saharan Africa, a populous (and growing) area of the world that has received comparatively less public health funding and attention. 

There aren’t any major methodological issues present, but some notable weaknesses and a few areas which could benefit from clarity are as follows:

• L224-5: It's not clear why species *richness* was chosen rather than NHP host density. Indeed, in some zoonotic systems, spillover seems to inversely correspond with species richness (ie rodents). The authors should clearly support their reasoning for their choice.

• The authors first use the term "occurrence" in the methods, alongside "outbreaks" in Table 2 and methods. A clear delineation of how an occurrence *differs* from an outbreak would significant improve manuscript clarity. Related: including points showing occurrences on at least one of the maps (e.g. Fig 1) would help orient the reader. 

• Lines 167-173 reference citation #23, which appears outdated and doesn't support many of the manuscript's assertions. A quick search turned up several good resources, such as Agwu et al. 2016 (Acta Tropica, Volume 161, p18-25).

Some minor issues:

• Ref #24 is referred to repeatedly. A bit more exposition at first introduction re who/what/when would help contextualize results.

• L198-201: Split into 2 sentences? Tough to parse as is.

• L224-5: any literature on differential NHP host competence? Why not use presence of each species as a separate presence/absence? A bit more detail here would help readers with limited background in NHPs. 

• L236: "a jackknife procedure" - a bit more detail here would be appreciated.

• L243: "percent contribution was deemed less important" - by what measure? If it's not important, why include it?

• L246-8: methods here are a bit tough to follow.

• L297: explain "minimum training presence threshold" in text here (in addition to Fig caption). Does this differ from "minimum training presence cumulative threshold" (L250)? Related, a bit more narrative exposition on how Figs 3 and 4 are related would be appreciated (and possibly mentioned in respective caption(s)).

• L295-313: results from multiple models are included in a long list. Consider breaking into shorter paragraphs and adding some transition text to orient the reader. 

• L424-5: "AUC values can be inflated" - please cite.