Biodiversity loss is often positively associated with habitat perturbation 47,48, and both have been linked to increased infectious disease risk 9, 49–52; however, previous studies have failed to decouple the effects of habitat degradation on biodiversity and disease transmission 2,53. Our study system of tropical cave-dwelling bat communities provided such an opportunity, given that cave complexity, not habitat perturbation, explained bat community composition. This relationship was consistent across seasons. For instance, we found rich communities of 12 bat species in highly degraded habitats associated with complex caves but depauperate communities of two bat species in pristine environments associated with simple caves. Past studies from the Philippines and Brazil have shown similar results 19,41,42, highlighting the influence of cave complexity on local bat community assemblage structure across fragmented landscapes 22 and maintenance of bat-provided ecosystem services 24. Despite finding that bat richness did not decline with perturbation, we did find lower bat densities in degraded habitats. This may indicate that degraded habitats have reduced capacity to sustain large bat populations, potentially a warning sign of future population declines. Although caves are more vulnerable to anthropogenic disturbance than many other ecosystems 43,54,55, they are rarely considered in regional or landscape-level conservation policy 43, a situation our results stress must change.
The potential for biodiversity to affect disease risk relies on two primary criteria 13. First, the order in which species assemble is at least partly deterministic, not fully stochastic, resulting in a nested community structure, where low-diversity communities constitute a subset of higher-diversity communities 29, as seen in our study system and other cave-dwelling bat studies 56–58. Second, there is an association between host competence (i.e., ability to amplify and transmit pathogens to others) and extirpation risk 59. Hence, heterogeneity in host competence affects disease prevalence 60, where changes favoring highly competent species are expected to increase transmission. In contrast, changes favoring non-competent host species would lead to reduced transmission 61. Mechanistically, resilient species may exhibit fast life histories and invest less in immune defense than species more sensitive to disturbance 30,31,62; therefore, host competence may be shaped by trade-offs with life history strategy 63,64. Alternatively, pathogens may be better adapted to the most abundant species 65, making host competence the unique outcome of a particular host-pathogen co-evolutionary history 53. Nonetheless, the latter criteria assume host competence is a species-specific fixed trait. Yet, evidence suggests that host competence may be a plastic trait with intraspecific variation 66,67, where environmental variables and seasonality can have additional influences 68,69. Thus, anthropogenic habitat perturbation may also influence host species' competence 70. Considering this, we evaluated not only the effects of species richness but also habitat quality on disease prevalence in cave-dwelling bat communities. Since bats roosting in caves surrounded by degraded landscapes may be forced to forage in poor quality habitats or commute longer distances to reach suitable habitats, potentially increasing energetic costs with downstream effects on health 22,26,32,33. Tropical cave-dwelling bats provide us with a unique opportunity to assess variation in multiple pathogen prevalence in a multi-host system along a diversity and perturbation gradient.
In most studied systems, greater species richness seems to reduce pathogen prevalence 3; however, this pattern was not general for our selected pathogens. While Bartonella prevalence was negatively associated with species richness (i.e., dilution effect 61), Trypanosoma and microfilaria prevalence were positively associated (i.e., amplification effect 61), and Leptospira prevalence showed no association with species richness. Others have previously argued that biodiversity-disease relationships may be idiosyncratic and context-dependent 4,6, 71–73, and our results provide empirical support to these views. For example, another study of Trypanosoma in bats from Panama found that pathogen prevalence decreased as habitat patch size increased, where bigger patches sustained higher host diversity and average biomass 74. However, it is worth noting that we did not evaluate the effects of host identity, and a dilution effect may occur owing to the presence of a particular host species rather than just community richness 29. Host species may differ in their susceptibility and competence, and their sole presence in a community could alter patterns in disease prevalence 75, which may differ across pathogens.
Although it is well recognized that host communities may be infected by multiple pathogens 76, few studies have evaluated the effects of biodiversity on multi-pathogen prevalence in the same set of host 77,78. This approach allows us to identify potential pathogen traits that lead to similar or distinct outcomes 2. In our case, we found varied responses to species richness, which may be explained by pathogen variation in transmission mode, immune responses elicited, or host range of these target pathogens 72,79. These dissimilarities are further supported by the differential response of these four pathogens to seasonality (i.e., Bartonella prevalence was higher in the wet season, but microfilaria and Leptospira prevalence were higher in the dry season, and Trypanosoma prevalence did not differ between seasons). We encourage future research to build on these insights to improve our understanding of the transmission dynamics of the target pathogens and the life history and competence of the hosts in this system 14.
Beyond the direct effects of bat species richness on pathogen prevalence, another key finding of our study was the direct and indirect effects of habitat quality and cave metrics on pathogen prevalence, further demonstrating that context is relevant. For instance, cave complexity indirectly increased pathogen prevalence by increasing species richness (for Trypanosoma and microfilaria) but also had a direct negative effect on microfilaria prevalence. An opposing pattern was shown for Bartonella. Similarly, habitat quality only indirectly affected Bartonella prevalence through bat density but had a consistently direct effect on Bartonella, Trypanosoma, and microfilaria. Mechanistically, this direct effect may be due to effects on the pathogens or necessary vectors, e.g., survival rates. Further, it reveals that confounding factors in observational studies may be high because the same habitat perturbations and cave metrics that create host diversity gradients also alter transmission dynamics. Thus, using habitat degradation as a proxy for diversity in biodiversity-disease research and as the apparent link between disturbance and disease can be inaccurate 80.
Another critical challenge in empirically studying the relationship between biodiversity and disease risk revolves around separating the effects of host diversity and density 13,81,82. To address this, we estimated bat density in each cave to evaluate its direct effect on pathogen prevalence. Interestingly, bat density only affected Bartonella prevalence, but the relationship was positive in the dry season and negative in the wet. Contrary to predictions from mathematical models of an amplification effect for all density-dependent transmission when communities are additive and a dilution effect for all frequency-dependent transmission 1, our results show an opposing pattern. Moreover, host density is generally expected to promote pathogen transmission 83; however, higher densities of non-competent hosts may reduce infection rates through lower encounter rates 9 or via competition with competent hosts 84. Thus, although bat density remained constant between seasons, shifts in bat physiology (including immunity), demography, and/or relative species abundance may explain these divergent patterns.
Seasonality affects resource availability and, consequently, the timing of reproductive pulses in tropical systems, further impacting bat physiology in ways that may increase pathogen transmission 40,85,86. For example, data collected during a longitudinal study of Rousettus aegyptiacus in Uganda revealed distinct pulses of Marburg virus infection in newly susceptible six-month-old bats, coinciding with the biannual birthing season and the peak of the rainy season, subsequently increasing the risk of human infection 86. In our study, intriguing patterns arose in the wet season, with habitat quality positively affecting pathogen prevalence. Perhaps, as in the Marburg virus example, birth pulses are related to higher resource availability in the wet seasons and in high-quality habitats 40. Our results highlight that multi-year, longitudinal studies across seasons are needed to understand the influence of seasonality on pathogen dynamics in multi-pathogen, multi-host systems.
While we acknowledge the limitations of observational field studies like ours caused by confounding factors (i.e., the environmental gradient simultaneously affecting host species diversity and other aspects of transmission), our results highlight the critical importance of caves for bat conservation and the need for longitudinal studies of cave-dwelling bat disease dynamics 20. We also demonstrate that pathogen dynamics within communities vary across perturbation gradients, seasons, and pathogens, highlighting that predicted pathogen risk based on presence and relative abundance of host species have the potential to be misleading. By developing a distinctive framework using a multi-host, multi-parasite system, our work provides insight into the complex patterns of pathogen dynamics and bat community ecology in changing environments with application to biodiversity conservation and public health.