Abstract
Environmental pathogen reservoirs exist for many globally important diseases and can fuel disease outbreaks, affect pathogen evolution, and increase the threat of host extinction. Differences in pathogen shedding among hosts can create mosaics of infection risk across landscapes by increasing pathogen contamination in high use areas. However, how the environmental reservoir establishes in multi-host communities and the importance of factors like host-specific infection and abundance in environmental contamination and transmission remain important outstanding questions. Here we examine how Pseudogymnoascus destructans, the fungal pathogen that causes white-nose syndrome in bats, invades and establishes in the environment. We quantified dynamic changes in pathogen shedding, infection intensities, host abundance, and the subsequent propagule pressure imposed by each species within the community. We find that the initial establishment of the pathogen reservoir is driven by different species within the community than those that are responsible for maintaining the reservoir over time. Our results also show that highly shedding species do not always contribute the most to pathogen reservoirs. More broadly, we demonstrate how individual host shedding rates scale to influence landscape-level pathogen contamination.
Open Research statement Data will be made available through the Dryad Digital Repository before publication or upon reviewer request.
Introduction
Emerging infectious diseases threaten efforts to conserve global biodiversity (Daszak et al. 2000, Taylor et al. 2001, Jones et al. 2008, Fisher et al. 2012). In some disease systems, pathogens may survive for long periods of time in the environment in the absence of a living host (Turner et al. 2016, Plummer et al. 2018, Islam et al. 2020). Pathogen persistence in the environment allows for transmission independent of infected hosts, can exacerbate disease impacts, and increase the risk of host extinction (de Castro and Bolker 2005, Mitchell et al. 2008, Almberg et al. 2011, Hoyt et al. 2020). However, pathogen contamination in the environment is not homogenous, rather, variation in the extent of the environmental reservoir is likely driven by a complex process of pathogen shedding from hosts within the community that lead to subsequent transmission events.
Infected hosts can vary in the amount of pathogen they shed into the environment with some hosts producing disproportionately high amounts of pathogen, independent of host contacts (Sheth et al. 2006, Chase-Topping et al. 2008, Lawley et al. 2008, Direnzo et al. 2014). Research has shown that inherent variation in host shedding can be driven by differences in behavior (Godfrey 2013, Rushmore et al. 2013, VanderWaal and Ezenwa 2016), innate susceptibility (Searle et al. 2011, Gervasi et al. 2013), space use (Brooks-Pollock et al. 2014), and infection severity (Lloyd-Smith et al. 2005, Munywoki et al. 2015). In multi-host disease systems, the variation produced through community composition and species abundance play key roles that determine transmission dynamics (Kilpatrick et al. 2006, Paull et al. 2012).
Host abundance within a community can interact with host shedding to moderate transmission from a particular species (Lloyd-Smith et al. 2005, Paull et al. 2012, Kilonzo et al. 2013). Species that have low rates of shedding, but are highly abundant, may contribute more to transmission than might be expected based on the amount of pathogen they shed (Peterson and McKenzie 2014, Scheele et al. 2017). Conversely, a species that has high rates of shedding or is highly infectious, but at low abundance, may contribute less to transmission than other species (Lloyd-Smith et al. 2005, Kilpatrick et al. 2006). In addition, for some wildlife infectious diseases, infectiousness, shedding, and impacts are positively correlated (Langwig et al. 2016, Brannelly et al. 2020) such that hosts that are initially important in transmission suffer from high impacts and become less important contributors to the environmental reservoir over time. Heterogeneity in pathogen shedding and how it influences disease dynamics is important for many disease systems (Sheth et al. 2006, Chase-Topping et al. 2008, Henaux and Samuel 2011, Brooks-Pollock et al. 2014, Direnzo et al. 2014, Slater et al. 2016), but how differences in host shedding scale to a community-level and influence the environmental reservoir is rarely linked together.
White-nose syndrome (WNS) is an emerging infectious disease caused by the fungal pathogen Pseudogymnoascus destructans (Lorch et al. 2011, Warnecke et al. 2012), that has had devastating effects on bat populations (Langwig et al. 2012, Frick et al. 2015, Langwig et al. 2016). White-nose syndrome exhibits seasonal infection dynamics that are driven by the environmental reservoir and host-pathogen ecology (Langwig et al. 2015a, Hoyt et al. 2021, Langwig et al. 2021). Pseudogymnoascus destructans can persist for long periods of time in the environment, which results in widespread infection when hosts return to hibernacula (subterranean sites where bats hibernate in the winter) in the fall (Lorch et al. 2013, Hoyt et al. 2015, Langwig et al. 2015a, Campbell et al. 2019, Hoyt et al. 2020, Hicks et al. 2021). During this time, susceptible bats become infected or reinfected by P. destructans when they come into contact with the environmental reservoir upon entering hibernacula (Langwig et al. 2015a). Over the winter hibernation period, P. destructans grows into the skin tissue, causing deleterious physiological changes, including increased arousals from hibernation, weight loss, dehydration, and ultimately death (Warnecke et al. 2013, Verant et al. 2014, McGuire et al. 2017, Hoyt et al. 2021). During hibernation, susceptible bat species vary greatly in their infection intensities and three species have suffered declines that exceed 95% (Langwig et al. 2012, Langwig et al. 2016, Hoyt et al. 2020, Hoyt et al. 2021). While species vary in infection intensities, species abundance also varies greatly within bat communities (Langwig et al. 2012, Frick et al. 2015), which may also influence pathogen contamination in the environment.
Contamination of P. destructans in the environment increases after the first year of invasion (Hoyt et al. 2020) and the extent of the environmental reservoir has been linked to increased pathogen prevalence and increased pathogen loads for bats across the globe (Hoyt et al. 2020, Hoyt et al. 2021). As a result, bat mortality also increases with higher levels of environmental contamination (Hoyt et al. 2018, Hoyt et al. 2020, Hicks et al. 2021). However, how the environmental pathogen reservoir becomes established in these multi-host communities has not been investigated.
Environmental transmission is an important driver of infectious disease dynamics and understanding the factors that lead to pathogen establishment in the environment is crucial for disease control and prevention. Here we use a unique dataset that encompasses the stages of P. destructans invasion and establishment to capture pathogen accumulation in the environmental reservoir across 21 sites in the Midwestern, United States. Using these data, we explored potential differences in pathogen shedding among species. We also assessed the relationship between bat infections and the amount of pathogen shed into the environment by each species present in the community. Because we hypothesized that bat species abundance would play a key role in site-level contamination and pathogen establishment of the environmental reservoir, we also explored how differential pathogen shedding among species and their abundance influences the pathogen pressure within the community.
Methods
Sample collection and quantification
We quantified P. destructans fungal loads on bats and from hibernacula substrate throughout bat hibernation sites in the Midwestern, United States. Samples were collected from 21 sites in Wisconsin, Illinois, and Michigan over seven years from pathogen invasion to establishment. Hibernacula were visited twice yearly, once during early hibernation (November to December) and once during late hibernation (March to April) to capture differences in infection and environmental contamination at the beginning and end of hibernation. During each visit, we counted the total number of bats within each site by species (Eptesicus fuscus, Myotis lucifugus, Myotis septentrionalis and Perimyotis subflavus). We collected epidermal swab samples from bats within sites to quantify bat infections (fungal loads) and determine infection prevalence. Samples were collected using previously established protocols that consisted of rubbing a polyester swab dipped in sterile water over the muzzle and forearm of the bat five times (Langwig et al. 2015a, Hoyt et al. 2016).
We also collected environmental substrate swabs from hibernacula walls and ceilings directly under hibernating bats to measure the amount of P. destructans shed into the environment. To capture site-level environmental contamination, P. destructans was collected from the environment in locations more than two meters away from any roosting bats, but in areas that bats might roost. These environmental substrate samples were taken by swabbing an area of substrate equal to the length of a bat’s forearm (36-40 mm) five times back and forth, as described previously (Langwig et al. 2015b). We preserved P. destructans DNA samples by storing all swabs in salt preservation buffer (RNAlater; Thermo Fisher Scientific) directly after collection. DNA was extracted from all samples with a modified Qiagen DNeasy Blood & Tissue Kit (Langwig et al. 2015b). The presence and quantity of P. destructans was determined by quantitative Polymerase Chain Reaction (qPCR) (Muller et al. 2013).
To verify that fungal loads measured using qPCR accurately reflected viable fungal spores in the environment, that would be capable of infecting a host, we collected additional substrate swab samples from a subset of locations that were paired with swabs used for qPCR. These samples were cultured by streaking the substrate swab across a plate containing Sabouraud Dextrose Agar treated with chloramphenicol and gentamicin. The plates were stored at 4 °C and colony forming units (CFU’s) of P. destructans were quantified within six weeks of initial inoculation. We paired substrate samples analyzed using qPCR to determine log10 P. destructans loads for comparison with colony forming units obtained from culture samples to validate viability. There was a significant relationship between quantity of DNA measured through qPCR and the number of P. destructans CFU’s (Figure S1). This suggests that qPCR was a valid method to estimate the extent of P. destructans in the environment and supports that qPCR results are reflective of the number of infectious propagules in the environment and not relic DNA (Figure S1).
All research was approved through Institutional Animal Care and Use Committee protocols: Virginia Polytechnic Institute: 17-180; University of California, Santa Cruz: Kilpm1705; Wisconsin Endangered/Threatened Species Permit 882 & 886; Michigan Department of Natural Resources permit SC-1651; Illinois Endangered/ Threatened Species Permit 5015, 2582 and Scientific Collections permit NH20.5888; US Fish and Wildlife Service Threatened & Endangered Species Permit TE64081B-1.
Data analysis
We separated invasion stage into two distinct categories: “invasion” which included the first year the pathogen arrived, as described previously (Langwig et al. 2015b, Hoyt et al. 2020), and “establishment” which included the second and third years of P. destructans presence in a site when species declines begin to occur, which corresponds to the epidemic phase as has been previously noted (Langwig et al. 2015b, Hoyt et al. 2020). We used these stages to capture pathogen shed into the environment before and after pathogen accumulation in the environment occurred (invasion and establishment, respectively) and to examine the dynamic changes between stages of pathogen invasion and establishment.
We first examined the presence and quantity of P. destructans on each species and the amount each species shed into the environment. We used mixed effect models with log10 environmental fungal load collected under each bat as our response variable, bat species interacting with pathogen invasion stage as our predictor and site as a random effect. We similarly compared infection intensity among species using the same model as described above, but with log10 fungal loads on bats as the response variable. We examined differences in infection prevalence by species using a generalized linear mixed effects model with a binomial distribution and a logit link with species as our predictor and bat infection prevalence (0|1) as our response and site as a random effect.
To examine how differences in infection on each bat contributed to contamination of the environment under each individual, we used a linear mixed effects model to explore the relationship between infection on each individual bat and pathogen shed into the environment under each individual. In this analysis, we used paired log10 environmental P. destructans loads under a bat as our response variable with log10 bat fungal loads interacting with species as our predictor and site as a random effect. We combined the invasion and established stages since the amount of pathogen shed into the environment was hypothesized to be a product of how infected the host was, and therefore, comparable across years.
We investigated the role of species abundance in environmental contamination of P. destructans. We tested for differences among species abundance within each site by using a linear mixed effects model with species as our predictor and log10 average abundance during early hibernation (before over-winter declines occur) as our response and included site as a random effect for both invasion stages. To explore how species abundance influenced the degree of pathogen contamination within sites, we used a linear mixed effect model with log10 environmental fungal loads collected > 2m from any bat during late hibernation as our response variable and log10 average abundance interacting with species identity as our predictors with site as a random effect.
Finally, we calculated differences in propagule pressure among species by multiplying infection prevalence by the species abundance within each site to get the number of infected individuals. We then multiplied the number of infected individuals by the average amount of fungal spores (conidia) shed into the environment by each species in each site. We used a generalized linear mixed effect model with a negative binomial distribution to account for dispersed counts with an interaction between invasion stage and species as our predictor and propagule pressure as our response. We performed this analysis for both invasion and establishment during late hibernation to investigate how pathogen pressure may differ across stages of pathogen invasion when bats are heavily shedding into the environment. To assess if propagule pressure was a suitable metric for pathogen invasion of the environmental reservoir, we examined the relationship between site-level environmental contamination, as described above, and sum propagule pressure for each site using a linear mixed effects model with site included as a random effect. All analyses were conducted in R v.4.2.1 (R Core Team 2022). Mixed-effects models were run using the package “lme4” (Bates et al. 2015) except for analyses using a negative binomial distribution which were ran using “glmmTMB” (Brooks et al. 2017)
Results
We found that pathogen shedding (the amount of pathogen under bats) varied among the four species (Figure 1). During initial pathogen invasion into bat communities, we found that on average individual M. septentrionalis (Figure 1A; Table S1; intercept =-3.79 ± 0.23) contributed more pathogen into the environment than E. fuscus (coeff = -0.77 ± 0.37, P = 0.04) and M. lucifugus (coeff= -0.58 ± 0.27, P = 0.03), but did not differ statistically from individual P. subflavus (coeff =-0.21 ± 0.38, P = 0.58). In the established stage (Figure 1B), we found support for higher shedding in M. lucifugus (intercept = -3.38 ± 0.11) than E. fuscus (coeff = -0.80 ± 0.14, P < 0.0001) and P. subflavus (coeff = -0.62 ± 0.13, P < 0.0001), but similar shedding to M. septentrionalis (coeff = -0.38 ± 0.24, P = 0.11).
Our results showed consistent support that host infection intensity predicted the amount of pathogen shed into the environment for all species (Figure 2C; Table S4; host infection and shedding relationship: M. lucifugus slope = 0.38 ± 0.05, P < 0.001; E. fuscus slope = 0.33 ± 0.07, P < 0.00001; M. septentrionalis slope = 0.41 ± 0.10, P = 0.00003; P. subflavus slope = 0.33 ± 0.08, P = 0.00006). Importantly, we found no support for differences among species in the slope between environmental pathogen shedding and host infection level (Figure 2C; Table S4; all P > 0.05), suggesting that differences in shedding were primarily driven by the extent of infection on individual bats. Eptesicus fuscus (intercept = -3.37 ± 0.16) had significantly lower infection levels than all other species within the community (Figure 2B; Table S3; M. lucifugus coeff = 1.38 ± 0.15, P < 0.0001; M. septentrionalis coeff = 1.08 ± 0.21, P < 0.0001; P. subflavus coeff = 1.30 ± 0.17, P < 0.0001) In addition, we found that M. lucifugus (intercept = 2.19 ± 0.27) had the highest infection prevalence, with 87% of hosts infected on average across the invasion and established stages (Figure 2A; Table S2; E. fuscus coeff = -0.79 ± 0.31, P < 0.01; M. septentrionalis coeff = -0.83 ± 0.35, P < 0.05; P. subflavus coeff = -1.10 ± 0.24, P < 0.0001).
In addition to differences in infection intensity, species also varied in abundance, with M. lucifugus (intercept = 1.63 ± 0.19) being the most abundant species across sites, with an average population size of 39.81 ± 7.76 before declines from WNS (invasion) (Figure 3; Table S5; E. fuscus coeff = -0.65 ± 0.24, P < 0.01; M. septentrionalis coeff = -0.60 ± 0.21, P < 0.01; P. subflavus coeff = -0.41 ± 0.23, P < 0.08). During the established stage, while population declines were occurring, M. lucifugus (intercept = 1.28 ± 0.13) remained the most abundant species within the community with an average population size of 18.20 ± 7.41 (Figure 3; Table S6; E. fuscus coeff = -0.54 ± 0.16, P < 0.001; M. septentrionalis coeff = -0.76 ± 0.14, P < 0.0001; P. subflavus coeff = -0.38 ± 0.15, P < 0.01). M. septentrionalis had the lowest abundance in the community with an average colony size of 11.22 ± 5.50 individuals before disease impacts and declined to an average of 3.52 ± 4.12 individuals in a site during the established stage (Figure 3; Table S5-7). Furthermore, host abundance was important in influencing the degree of site-level environmental pathogen contamination (Figure S2; Table S7; M. lucifugus slope = 0.40 ± 0.14, P = 0.03).
Finally, when we combined infection prevalence, species abundance, and pathogen shedding into the metric of propagule pressure, we found that during the first year of invasion, M. septentrionalis (Figure4A; Table S8; intercept = 7.85 ± 0.66) had higher propagule pressure than P. subflavus (coeff = -2.81 ± 1.01, P < 0.01) and E. fuscus (−4.28 ± 0.85, P < 0.0001), but not M. lucifugus (coeff = 0.43 ± 0.73, P = 0.55). In years following invasion, once P. destructans is established, M. lucifugus (Figure 4B; intercept = 11.08 ± 0.56) had consistently higher propagule pressure and contributed more pathogen to the environmental reservoir than all other species (Figure 4B; M. septentrionalis coeff = -4.14 ± 0.96, P < 0.0001; E. fuscus coeff = -4.18 ± 0.68, P < 0.0001; P. subflavus coeff = -2.53 ± 0.68, P = 0.002). Myotis septentrionalis (Figure 4B; intercept = 6.94 ± 0.95) no longer contributed more than P. subflavus (coeff = 1.62 ± 1.03, P = 0.11) and E. fuscus (coeff = -0.04 ± 1.02, P = 0.97) due to their rarity following disease-induced declines (Figure 4). Mean environmental contamination in areas > 2m from bats increased with total propagule pressure at a site (summed propagule pressure among species) (Figure 4C; relationship between environmental contamination and propagule pressure intercept = -5.48 ± 0.14, slope = 0.12, P = 0.002).
Discussion
Host heterogeneity in pathogen shedding and subsequent environmental transmission can have dramatic impacts on community-level disease outcomes. Our findings demonstrate how host heterogeneity in infection intensity can lead to increased pathogen shedding, which is consistent with research in other systems (Lloyd-Smith et al. 2005, Matthews et al. 2006, Chase-Topping et al. 2008, Direnzo et al. 2014, Munywoki et al. 2015, Maguire et al. 2016, VanderWaal and Ezenwa 2016). In addition, our study demonstrates that host heterogeneities in abundance and infection interact to influence the extent of the environmental reservoir and the contribution of species, which varied over time. We identified a shift in the species responsible for the greatest environmental reservoir contamination between disease stages, suggesting that reservoir contamination is a dynamic process that varies over time.
Our findings suggest that the observed heterogeneities in pathogen shedding are primarily due to variation in infection intensities among species within the community. The relationship between individual host shedding into the environment and host infection intensity did not differ among species, suggesting that individuals of different species with the same infection intensity are equally efficient at depositing pathogen into the environment. Instead, the severity of infection, which varied among species, determined how much pathogen is shed per individual. Some species such as E. fuscus had low level infections, while the other species such as M. septentrionalis and M. lucifugus were much more infected and contributed more pathogen to the environmental reservoir (Figure 2B).
Disease-caused declines in several species (Figure 3) resulted in large changes in species abundance and community composition. We found evidence that species abundance predicted site-level contamination (Figure S2), and this supported the need to account for both species’ infection and abundance when determining the contribution to the environmental reservoir. To combine these factors, we used propagule pressure or “introduction effort” which is a fundamental ecological metric that determines whether a biological invasion will be successful in reaching establishment (Lockwood et al. 2005), and is especially applicable to emerging infectious diseases. Our results indicated that during the initial invasion of P. destructans, both M. lucifugus and M. septentrionalis had higher propagule pressure than the other two species (Figure 4A), which was not apparent through examination of only pathogen shedding. The high infection levels of M. septentrionalis resulted in a large contribution to the initial establishment of the environmental reservoir despite their low relative abundance (Figure 2B; Figure 3). However, in subsequent years, as this species declined precipitously (Figure 3), their contribution was greatly reduced and equivalent to less infected species (e.g. E. fuscus and P. subflavus, Figure 4B).
Our results suggest that host abundance, shedding, and infection prevalence are collectively required to describe pathogen contamination in the environment. We found that environmental contamination in the sampled locations that were not associated with bats increased as propagule pressure increased (Figure 4C), suggesting that propagule pressure is an accurate metric for measuring species contributions to the environmental reservoir. For example, while E. fuscus, M. septentrionalis, and P. subflavus eventually contributed similar propagule pressures during disease establishment, this was likely driven by different factors that need to be considered when evaluating their influence on the reservoir. Eptesicus fuscus had low infection intensity but had moderate prevalence (Figure 2A-B) and moderate abundance (Figure 3) which elevated its contribution. Perimyotis subflavus had high infection levels, but reduced prevalence (Figure 2A-B) compared to the other highly infected species and was only present in moderate abundances (Figure 3). Finally, M. septentrionalis was highly infected (Figure2B), but at low abundance across communities (Figure 3), which reduced its overall importance in years following initial pathogen invasion. Myotis lucifugus maintained high propagule pressure through the invasion and establishment of P. destructans (Figure 4A-B), suggesting that the presence of M. lucifugus within a community will result in rapid establishment of P. destructans in the environment and maintenance of environmental contamination within a site over time. Our results highlight how both species infection intensity and abundance must be considered to evaluate the influence of pathogen shedding on environmental reservoir establishment, subsequent environmental transmission and how this can change dynamically across disease stages.
Understanding environmental reservoir dynamics is crucial for many globally important disease systems. Discerning how variation across species contributes to maintenance of the environmental reservoir is important in determining the extent of epidemics, predicting long-term impacts on host communities, and developing control strategies. We demonstrate that the establishment and maintenance of the environmental reservoir is strongly influenced by both heterogeneity in pathogen infection and species abundance. Evaluating these effects together through the metric of propagule pressure allowed us to capture which species within the community contributed to pathogen invasion success and ultimately the maintenance of indirect transmission which is an important driver of infection and mortality (Hoyt et al. 2018). Broadly, our results demonstrate that multiple factors of species heterogeneity can scale to influence environmental reservoir dynamics within communities.
Author Contributions
NAL and JRH wrote the original draft of the manuscript. JRH, KEL, and AMK designed methodology; NAL, JRH, KEL, AMK, JPW, HMK, JAR, JED, WHS, and JK collected the data; NAL conducted the laboratory experiment; JTF and KLP supervised and performed sample testing; NAL analyzed the data with assistance from KEL and JRH; All authors contributed critically to draft revision.
Conflict of Interest
The authors declare no conflicts of interest
Acknowledgements
The research was funded by the joint NSF-NIH-NIFA Ecology and Evolution of Infectious Disease award DEB-1911853, NSF DEB-1115895, USFWS (F17AP00591), and by the NSF GRFP. We would like to acknowledge landowners for access to sites, and M.J. Kailing for assisting culture collection and B. Heeringa and D. Kirk for field assistance.