Fig 1.
The flow diagram of literature review.
SAPs sandfly-associated pathogens; VL visceral leishmaniasis; CISDCP Chinese Information System for Disease Control and Prevention.
Table 1.
The average AUC of the BRT models and predicted numbers, land areas and population sizes of affected counties for the 12 most prevalent sandfly species in the mainland of China.
Fig 2.
Clustering of sandfly species, and the detection of SAPs in animals and sandflies.
Panel A, the dendrogram displays clusters 1–3 of sandfly species. The features used for clustering are the three properties associated with each predictor in the BRT model. Two of the three properties were shown to indicate possible levels of ecological suitability: i) the relative contributions of the predictors were shown in colors, with dark color corresponding to high contribution; ii) the standardized median value of the predictor was indicated by the numbers in the heatmap (numbers 1–4 indicate the position of this median in reference to the quartiles of this predictor among all counties). For details on the feature selection and generation for clustering, see Text B in S1 Appendix. Panels B–D, the spatial distribution of the three clusters of sandflies. Red solid lines indicate the boundaries of the seven biogeographic regions. Panel E, sandfly species and animals that carry SAPs. The names of the pathogens of VL are shown in blue, and the names of the vectors of VL are shown in red. SAPs sandfly-associated pathogens; BRT boosted regression tree; VL visceral leishmaniasis. Base layers of the maps were downloaded from Resource and Environment Science and Data Center (https://www.resdc.cn/DOI/DOI.aspx?DOIID=120).
Fig 3.
The spatial distribution of SAPs, and the recorded and model-predicted annual incidence of VL at the county level.
Panel A, locations of SAPs detected from sandflies. Panel B, locations of SAPs detected from animals. Panel C, the average annual incidence of VL from 2014–2018 recorded at the county level. Panel D, the model-predicted annual incidence of VL in 2014–2018. The provinces outlined in red, green, and purple correspond to AVL&DT-ZVL endemicity, MT-ZVL endemicity, and endemicity for both types, respectively. The white area represents provinces that are non-endemic and devoid of any risk areas. SAPs sandfly-associated pathogens; VL visceral leishmaniasis; AVL&DT-ZVL anthroponotic VL and desert-type zoonotic VL; MT-ZVL mountain-type zoonotic VL. Base layers of the maps were downloaded from Resource and Environment Science and Data Center (https://www.resdc.cn/DOI/DOI.aspx?DOIID=120).
Table 2.
Projections of the counties, areas and population size potentially affected by VL cases under SSP585 in the mainland of China.
Table 3.
The relative contributions of major factors to the spatial distributions of two types of VL, estimated by two-stage XGBoost models.
Fig 4.
Spatial distribution and changes in model-predicted incidence of VL under SSP585.
Panels A–C, predicted annual incidence of VL in 2021–2040, 2041–2060, and 2061–2080, respectively. Panels D–F, changes in predicted VL risk level from 2014–2018 to 2021–2040, from 2021–2040 to 2041–2060, and from 2041–2060 to 2061–2080, respectively. The white block labeled “no risk” indicates that the VL risk level has not changed compared to the previous period, and both periods are predicted to be “no risk”. VL risk levels were divided into five levels based on quartiles of county-level annual incidences from 2014 to 2018: level 0 (no risk), level 1 (>0 but <25th percentile), level 2 (≥25th percentile but <50th percentile), level 3 (≥50th percentile but <75th percentile), and level 4 (≥75th percentile). The change in VL risk level was calculated as the predicted VL risk level of a specific county in a later period minus the predicted VL risk level in the previous period. VL visceral leishmaniasis. Base layers of the maps were downloaded from Resource and Environment Science and Data Center (https://www.resdc.cn/DOI/DOI.aspx?DOIID=120).