Fig 1.
The results of the March 2020 literature review.
A: Study and sample selection process for the 2020 infection occurrence database. Records were produced via a literature review which was performed on March 2nd 2020, filtering for publications released after October 2015. B: Newly extracted point and polygon occurrence records across Southeast Asia by spatial type. Admin 1 regions are the first subdivision below national, e.g. state or province. Admin 2 regions are the second subdivision below national, e.g. district or regency. C: The number of occurrence samples in each occurrence database by the year the sample was collected. Administrative boundary base maps sourced from the Malaria Atlas Project (CC BY 3.0, [32]) and international boundaries from the US Department of State Large Scale International Boundaries dataset (public domain, [33]).
Table 1.
The set of raster covariate datasets used in model fitting and prediction.
Differences in raster datasets between this work and those used in the 2015 P. knowlesi risk model appear in bold. STRM: Shuttle Radar Topography Mission, MODIS: Moderate Resolution Imaging Spectroradiometer, IGBP: International Geosphere-Biosphere Programme.
Fig 2.
Data included for modelling across the training and evaluation regions and corresponding multivariate environmental similarity surface (MESS).
A: The data-set of occurrence points and polygons used for fitting the boosted regression tree model across the model training region of Malaysia, Brunei, and Singapore. Presence polygons are displayed as the number of polygons covering each given pixel, with this density being proportional to the probability distribution of points sampled from the polygons for each bootstrap. B: The presence and absence records used in the model evaluation process, across the evaluation region of Southeast Asia excluding Malaysia, Brunei and Singapore. C: Multivariate environmental similarity surface (MESS) for the model, where areas shaded in light grey indicate that at least one covariate value at that point is outside the range of values within the training data (extrapolation). Administrative boundary base maps sourced from the Malaria Atlas Project (CC BY 3.0, [32]) and international boundaries from the US Department of State Large Scale International Boundaries dataset (public domain, [33]).
Fig 3.
Predicted P. knowlesi transmission suitability across Southeast Asia.
A: Modelled transmission suitability mean over Southeast Asia across the 500 bootstraps. Results are displayed only where an area is within the range of both a vector and reservoir species necessary for transmission (see Methods), regions outside of this range (displayed as grey) are considered to be very low risk for P. knowlesi transmission. Transmission suitability is a relative measure of the risk of P. knowlesi transmission from known reservoir species (via vector species) to humans. B: Standard deviation of the predicted transmission suitability across the 500 bootstraps. Administrative boundary base maps sourced from the Malaria Atlas Project (CC BY 3.0, [32]) and international boundaries from the US Department of State Large Scale International Boundaries dataset (public domain, [33]).