Fig 1.
Maps of (A) Thailand with Nan province highlighted in orange and (B) Nan province elevation and villages (n = 905).
The red dots represent the villages positions. The base layer of the map is available at https://data.humdata.org/dataset/cod-ab-tha.
Table 1.
Description of variables and their use in the models. The area is defined as the 2500 m buffer zone.
Fig 2.
Description of scrub typhus human cases: trend of scrub typhus cases per year.
Fig 3.
Maps of Nan Province: (A) spatial distribution of cumulative scrub typhus cases between 2003 and 2019 and (B) interpolation of mean scrub typhus cases at village level between 2003 and 2019 by kriging using a semi-variogram based on the village position. The base layer map for administrative boundaries is available at https://data.humdata.org/dataset/cod-ab-tha.
Fig 4.
Change in land cover surface area between 2003 and 2019 per land cover class.
The y-axes represent the number of cells, which is a square area of 300 meters by 300 meters.
Table 2.
Results of general additive modelling (GAM) (1) explaining the number of cases of scrub typhus per village in Nan Province using a negative binomial link (theta = 1.075), with approximate significance of smooth terms.
Fig 5.
General Additive Modeling (GAM) (1) results of the selected land covers explaining the number of scrub typhus cases in Nan Province between 2003–2019 by villages and year, using a binomial negative link function (theta = 1.075).
The smoothed variables selected in the best GAM were: (A) year, (B) number of males per village, (C) shrubland cover, (D) mosaic land cover, (E) broadleaf forest cover, (F) needleleaf forest cover and (G) geographical distribution of villages (longitude and latitude).
Fig 6.
Land cover change matrix of Nan Province between 2003 and 2019.
“Wat” = Water area, “Urb” = Urban area, “CropI” = Irrigated cropland, “CropR” = Rainfed cropland, “Grass” = Grassland, “Shrub” = Shrubland, “Mos” = Mosaic, “ForB” = Broadleaf forest, “ForN” = Needleleaf Forest. The coloured bar represents the proportion of land cover of 2019 compared to the 2003 period. For example, Water surface area had not changed and still represents 100% of its surface area of 2003. A 100% of the urbanised areas in 2003 stayed urbanised in 2019, however urban areas also expanded on the cropland irrigated field.
Table 3.
Results of general additive modelling (GAM) (2) of land cover transition explaining the number of cumulative scrub typhus cases per village in Nan Province using a negative binomial link (theta = 2.74), with approximate significance of smooth terms.
Table 4.
Results of general additive modelling (GAM) of land cover transition explaining the number of cumulative cases of scrub typhus per village in Nan Province using a negative binomial link (theta = 2.712), with polynomial adapted terms.
Fig 7.
Results of General Additive Modeling (GAM) (2) of selected land cover transition explaining the total number of scrub typhus cases in Nan Province between 2003–2018 at the village level, using a binomial negative link function (theta = 2.74).
The smoothed variables selected in the best GAM were: (A) transition from shrubland to grassland, (B) transition from shrubland to broadleaf forest, (C) stable broadleaf forest, (D) the number of males per village and (E) geographical distribution of villages (longitude and latitude).