We evaluated the association between predator species richness and filovirus spillover in Africa. The results showed that higher species richness in the order Strigiformes and family Colubridae was associated with lower odds of Ebolavirus spillover compared with that in regions with lower predator species richness. Regardless of the approach taken to calculate species richness, this association was robust.
The negative association between predator species richness and the risk of Ebolavirus spillover suggests top–down regulation of Ebolavirus reservoir hosts (i.e., bats) by predators. The crucial roles played by predators in terms of the functional diversity of ecological communities and the control of populations of disease reservoir hosts have been reported previously [9, 10, 12]. The greater the predator species richness (i.e., the numbers of predator species within an area), the greater the cascade effect on prey species. The growing body of research on bat predation is slowly improving our understanding of bat predators and the effects of predation on bat populations [12–14]. Natural bat predators may include birds, snakes, and mammals. Although few vertebrate predators are known to specialize on bats, and bat predation appears to be mostly opportunistic in nature, generalist and opportunistic predators may substantially impact bat ecology [9, 12, 14] via both direct predation and non-lethal cascade effects, also termed trait-mediated indirect interactions. Thus, predators control the abundance, density, and behavior patterns of prey species, eventually reducing the rate of contact between reservoir hosts and humans and thus mitigating the risk of zoonotic spillover [9]. Such suppression is relatively strong in regions wherein ecological diversity is well-maintained.
The predator species richness of the order Strigiformes was significantly and negatively associated with the risk of Ebolavirus spillover. This is consistent with previous studies suggesting that owls are primary predators of bats [13, 14]. Snakes are also supposed to prey on bats, via two strategies: positioning themselves near bat passage routes (i.e., near the entrances to bat roosts) and entering the refuges [12]. Most such behaviors have been reported in tropical regions [34], perhaps because tropical bats roost by hiding among leaves or in open canopies that are accessible to most vertebrate predators. Bat predation is poorly understood; bats fly at night and hide by day. However, it appears that predation of bats by snakes in our study area is more significant than previously thought. More ecological research is required.
Predator species richness was not significantly associated with Ebolavirus cases in models that considered only the species reported to prey on bats. This may be attributable to a lack of information on all bat predators. Although the number of known predators is increasing, such research is limited by the ecological characteristics of bats, which render observations of predation difficult [12]. Also, Marburgvirus occurrences were not consistently associated with predator species richness. The composition of bat species in the Marburgvirus regions may explain these results. Given the high bat diversity in the study region, R. aegyptiacus, the primary reservoir host of Marburgvirus, would not be the dominant bat species there. Therefore, the extent of predator richness may not have had any discernible effect on bat activities [35]. Further studies of bat ecology, diversity, and abundance, especially of R. aegyptiacus, are needed.
Despite the strengths of this ecological study, several limitations should be noted. First, we estimated the richness of predator species using stacked (aggregated) species distribution models. These models may systematically overestimate site-level species richness [36]. Therefore, we adjusted for bias using categorical values of predator species richness. Second, we did not include the temporal variations in species numbers from 2000 to 2021. However, such temporal changes can be ignored because most species considered are classified as IUCN “Least concern” (i.e., low risk of extinction). Third, when measuring species richness, we simply calculated the numbers of species; we excluded the relative abundances of the predator species or the density of predators. Future research should employ other indicators of diversity such as the Simpson diversity index [37]. Fourth, we considered only three bat species (E. franqueti, H. monstrosus, and M. torquate) that tested positive by PCR as primary reservoir hosts of Ebolavirus. Other probable reservoirs (bat species positive using serological methods) could be included in future studies. Fifth, a surveillance bias might exaggerate the results [38]. Due to the non-uniform reporting system for each region, the differences between regions may be smaller than those presented as results in this paper. Finally, our study units were 1˚ × 1˚ grids; the use of a different scale (such as 0.5˚ × 0.5˚) could have affected the results. This is the well-known modifiable area unit problem.