Spatial characterization of colonies of the flying fox bat, a carrier of Nipah virus in Thailand

Abstract

Background: A major reservoir of Nipah virus is believed to be the flying fox genus Pteropus, a fruit bat distributed across many of the world's tropical and sub-tropical areas. The emergence of the virus and its zoonotic transmission to livestock and humans have been linked to losses in the bat's habitat. Nipah has been identified in a number of indigenous flying fox populations in Thailand. While no evidence of infection in domestic pigs or people has been found to date, pig farming is an active agricultural sector in Thailand and therefore could be a potential pathway for zoonotic disease transmission from the bat reservoirs. The disease, then, represents a potential zoonotic risk. To characterize the spatial habitat of flying fox populations along Thailand's Central Plain, and to map potential contact zones between flying fox habitats, pig farms and human settlements, we conducted field observation, remote sensing, and ecological niche modeling to characterize flying fox colonies and their ecological neighborhoods. A Potential Surface Analysis was applied to map contact zones among local epizootic actors.

Results: Flying fox colonies are found mainly on Thailand's Central Plain, particularly in locations surrounded by bodies of water, vegetation, and safe havens such as Buddhist temples. High-risk areas for Nipah zoonosis in pigs include the agricultural ring around the Bangkok metropolitan region where the density of pig farms is high.

Conclusions: Passive and active surveillance programs should be prioritized around Bangkok, particularly on farms with low biosecurity, close to water, and/or on which orchards are concomitantly grown. Integration of human and animal health surveillance should be pursued in these same areas. Such proactive planning would help conserve flying fox colonies and should help prevent zoonotic transmission of Nipah and other pathogens.

Figure 1

Study area of flying fox colonies. Study area covering 93,826.2 km² of 23 provinces across western, central, and eastern Thailand (grey); 22 flying foxes’ colonies (red circles); comparing the size and locations of the study area and Thailand map (right).

Figure 2

Flying fox colonies compared to their environments. Comparison among the locations of the flying foxes’ colonies (circle) and variables in the study area: elevation (A); land cover (B); bodies of water (C); and human density (D).

Figure 3

Predicted suitability maps for flying fox colonies on the central plain of Thailand. The maps explained by Bioclim (BC), Domain (DM), Generalized Linear Model (GLM), Generalized Additive Model (GAM), Maximum Entropy Model (MAX), Boosted Regression Tree (BRT), and Random Forest (RF). The large map shows the Ensemble model (EM) output obtained by combining the 7 SDMs weighted by their respective predictive performance.

Figure 4

The predictive performance of 7 species distribution models. Box plots showing the predictive performance of 7 SDMs evaluated using the area under the curve (AUC) of ROC plots for the model sets (left) and test sets (right).

Figure 5

Fitted functions and relative contributions of variables predicted by the BRT. Partial dependence plots show the effect of a predictive variable on the response after accounting for the average effects of all other variables in the model: distance to water (A); distance to temple (B); human density (C); amount of vegetation area within 10 km radius (D); and elevation (E). The relative contributions of each variable from the BRT is shown in (F).

Figure 6

Factors used in mapping NiV risk. Maps of 4 factors used for analyzing the risk map of NiV in the central plain of Thailand: flying fox distribution map (A); distance to the flying foxes colonies (B); pig farm density at the sub-district level (C); house density at the sub-district level (D).

Figure 7

Risk area of NiV in the central plain of Thailand. Risk area of NiV produced by Potential Surface Analysis (PSA) based on i) flying fox distribution map, ii) distance to flying fox colonies, iii) house density and iv) pig farm density. The risk area of NiV for humans obtained from the first 3 factors (A), from all 4 factors (B), and the risk area of NiV for pigs produced by combining factors i, ii and iv (C). The yellow circles show different risk areas between B and C. Risk was low if the summation score was less than $\overline{x}-\sigma$, moderate if the summation score was range between $\overline{x}-\sigma$ and $\overline{x}-\sigma$, and high if the summation score was more than $\overline{x}-\sigma$.