We found strong links between habitat disturbance, microbiome composition, and Bd infection in A. crepitans. Habitat disturbance was associated with higher dispersion of the skin microbiome, but also lower Bd prevalence. We originally predicted that habitat disturbance would either directly reduce Bd prevalence and loads as expected based on shifts in microclimate, or indirectly increase Bd prevalence through disruption to the skin microbiome. Our results indicate that despite the strong influence of habitat disturbance on microbiome dispersion, Bd prevalence was primarily driven by direct effects of disturbance on abiotic conditions, consistent with previous studies [48, 49, 62, 63]. We also found lower Bd prevalence at sites with warmer water temperatures for L. catesbeianus. As water temperatures at our sites ranged from 20–28°C, which exceeds the known thermal maximum of Bd in culture (25°C; [64, 65]), this is further support of a strong role of abiotic habitat in determining Bd infection dynamics.
Generally, Bd exhibits low pathogenicity in amphibians of the southeastern US [66, 67], indicating that frogs in this temperate climate are tolerant to infection or Bd in this system may have low virulence. This is further supported by high pathogen loads recorded in Acris sp. with no apparent signs of disease in this and past studies [68, 69]. Even in the absence of Bd-associated disease, however, microbiome disruption through proxy of dispersion can be seen as an indicator of negative health in situations of environmental stress like habitat degradation. This is in line with the Anna Karenina principle put forth by Zaneveld et al., [31] proposing that microbiomes under stress will exhibit higher variability than microbiomes of healthy individuals. A recent study on endangered frogs from Costa Rica shows support for this hypothesis [37], showing that skin microbiomes of frogs from sites with greater human disturbance had higher dispersion than those of frogs from natural sites. Jiménez et al. [37] also found higher numbers of Bd-inhibitory OTUs on frogs at more disturbed sites, possibly indicating stress-associated selection on the microbiome to increase protective effects. Another study [21] found that temperature stress increased dispersion in the gut microbiome of tadpoles, which negatively impacted growth. Our findings contribute further support for the Anna Karenina principle, revealing the potential for habitat disturbance to increase microbiome variability.
Our results were inconsistent between host species, likely due to morphological and life history differences between A. crepitans and L. catesbeianus. These two host species use the same habitat and are active during the same time of day, although Acris sp. are known to avoid forest cover [70]. Additionally, L. catesbeianus are large-bodied and known to be resistant and tolerant to Bd infection [71–73]. Although not significant, we found a positive trend between water quality PC2, associated with nutrient levels, and skin microbiome dispersion for L. catesbeianus. While our metric of habitat disturbance was not directly correlated with nutrient levels, elevated nutrient levels in aquatic systems are often linked with disturbance-associated runoff [19]. Therefore, elevated nutrient levels in our study could indicate indirect habitat disturbance beyond our direct measurement of habitat composition. Past findings show that elevated nutrient runoff can lead to pathogenic bacterial blooms and subsequently elevated disease in frogs [33]. As dispersion is often used as a metric of dysbiosis [31], elevated microbiome dispersion in our study could point to similar health concerns.
Disturbance was also linked with distinct bacterial community assembly, indicating that in this instance higher dispersion signifies a shift in community composition in addition to increased stochasticity. We found strong support for microbiome differentiation based on habitat connectivity for both A. crepitans and L. catesbeianus, even after controlling for the relationship between microbiome dissimilarity and Euclidean distance between sites. These species are closely tied to bodies of water during the breeding season [74–76], which indicates that aquatic reservoirs of bacteria are the primary determinants of skin microbial composition. The link between microbiome similarity and habitat connectivity could therefore indicate that natural habitats are allowing greater connectivity between environmental reservoirs of bacteria [77], leading to greater similarity on frogs. Our analysis comparing aquatic bacterial similarity to habitat connectivity did not support this hypothesis, which indicates that both reservoir connectivity and host movement between habitats may be influential in microbiome assembly.
The metric for habitat disturbance used in this study included any agricultural land, recently disturbed land, and developed land. The land within a 500 m radius of a given pond makes up the majority of habitat for frogs dwelling in that pond [74], particularly during the breeding season [75, 78]. For this reason, we can infer that more disturbed habitat within this radius will provide lower quality habitat for frogs at a given pond, even if the pond itself seems pristine. In addition, any agricultural or urban landscape would provide run-off into these aquatic systems [23, 79]. It is possible that some chemicals may directly disrupt the microbiome through antibiotic effects, indirectly disrupt the microbiome through changes in the environmental pool of bacteria or affect the amphibian immune system [80]. Our metric of habitat disturbance was not correlated with any environmental variables measured including basic water chemistry and temperature. Therefore, we cannot infer which aspect of habitat degradation is directly influencing microbiome assembly in this system, and it is more likely driven by a number of factors, including those beyond the scope of our study.
As we only sampled the aquatic bacteria at three aquatic points within the 500 m radius, it is possible that other sources of bacteria, such as soil reservoirs, follow the expected trend with disturbance. Amphibians in the temperate southeast breed in late April through May, months when temperatures are still relatively mild. Based on temperature logging a week before sampling, average water temperatures at all sites (20–24°C) were within the optimal range for Bd (15 and 25°C; [64, 65]). Through cyclic seasonal patterns in infection [81], with higher infections in the fall and spring months, amphibians in this area likely have immune responses primed for Bd infection [82]. The high prevalence in our two aquatic-associated species without signs of disease indicates a commensal role of Bd in this system.
With exponentially increasing development of natural habitats, understanding the impacts of habitat disturbance on wildlife populations is vital. Past studies have shown that Bd infection risk is generally lower for generalist frogs that use disturbed habitats [48, 49], but amphibians may still be negatively affected despite pathogen reduction in warmer open environments [83]. Our results suggest that even in the absence of disease, microbiome disruption can occur in conjunction with habitat disturbance. Other factors not measured in this study like immune function and stress response are likely influenced by habitat integrity [68, 84] and may be mediating links between habitat disturbance and the skin microbiome. Additional experimental work tracking recruitment of environmental bacteria to the microbiome would be valuable in disentangling which habitat parameters are most likely connected to changes in the skin microbiome. Studies focusing on immune modulation by habitat change would also be valuable in determining how much of the microbiome response is mediated by the immune system and vice versa. Our work adds support to a growing body of literature showing that disturbance in many forms can influence the assembly of animal microbiomes in a stochastic manner [21, 31, 36–39]. With so many contemporary threats facing amphibian populations, understanding the influences of habitat disturbance on all aspects of amphibian health is vital in detecting cryptic threats to seemingly stable populations.