Table 1.
Evidence Categories and Possible Scores.
Fig 1.
Covariates used to model the environmental suitability of MAYV.
A. and B. Land surface temperature (LST) night and LST day, respectively; C. Rainfall; D. Enhanced vegetation index (EVI); E. Evergreen forest; F. Urban/built-up land cover; G. Tasseled cap brightness (TCB); H. Elevation; I. Slope. Maps were created in R using shape files from the Natural Earth public domain repository (http://www.naturalearthdata.com/).
Fig 2.
Evidence consensus is presented at the country level for all countries in the study (Fig 2A) and at the first-level administrative division (Fig 2B) for Brazil. Scores are based on health organization status, date of most recent MAYV occurrence; validity of MAYV diagnostic test, recency of MAYV outbreaks or clinical cases, and recency of MAYV occurrence in animals or arthropods. Blue represents very low evidence consensus while red represents very high evidence consensus. Maps were created in R using shape files from the Natural Earth public domain repository (http://www.naturalearthdata.com/).
Fig 3.
Geographic distribution and temporal trend of MAYV occurrence.
The map shows the distribution of the 195 occurrence locations (before the spatial thinning procedure) that were used to construct the boosted regression tree (BRT) model. The color corresponds to the host type of each point (human, animal, or arthropod). The inset chart displays total occurrences that were reported in each year since the initial human case was detected in 1954. The map was created in R using shape files from the Natural Earth public domain repository (http://www.naturalearthdata.com/).
Fig 4.
Map of environmental suitability and prediction uncertainty for MAYV occurrence.
Suitability ranged from blue (0—no suitability) to red (1—very high suitability). We obtained the lower and upper bound of MAYV presence limits by fitting an ensemble of 100 BRT submodels. The base map was sourced from Global Administrative Areas (GADM) version 4.0: https://gadm.org/download_country.html.
Fig 5.
Partial dependence plots of the included variables.
The solid black line represents the average response over 100 sub-models and the gray region represents one standard deviation. Tick marks represent values of each variable at occurrence locations. The y-axis represents the untransformed logit response and x-axis represents the full range of values for each covariate.
Table 2.
Total population living in areas potentially suitable for MAYV transmission (millions).