Probiotic therapy through bioaugmentation is a feasible disease mitigation strategy based on growing evidence that microbes contribute to host defences of plants and animals. Amphibians are currently threatened by the rapid global spread of the pathogen, Batrachochytrium dendrobatidis (Bd), which causes the disease chytridiomycosis. Bioaugmentation of locally occurring protective bacteria on amphibians has mitigated this disease effectively in laboratory trials and one recent field trial. Areas still naïve to Bd provide an opportunity for conservationists to proactively implement probiotic strategies to prevent further amphibian declines. In areas where Bd is endemic, bioaugmentation can facilitate repatriation of susceptible amphibians currently maintained in assurance colonies. Here, we synthesise the current research in amphibian microbial ecology and bioaugmentation to identify characteristics of effective probiotics in relation to their interactions with Bd, their host, other resident microbes and the environment. To target at-risk species and amphibian communities, we develop sampling strategies and filtering protocols that result in probiotics that inhibit Bd under ecologically relevant conditions and persist on susceptible amphibians. This filtering tool can be used proactively to guide amphibian disease mitigation and can be extended to other taxa threatened by emerging infectious diseases.
Microbial defences against pathogens are important for plants and animals (Berg 2009; Teplitski & Ritchie 2009; Gerritsen et al. 2011). Community structure is known to affect disease dynamics (Belden & Harris 2007; LoGiudice et al. 2008; Keesing et al. 2010), and evidence is accumulating that inter-specific interactions at the microbial level on individual hosts affect disease risk (Robinson et al. 2010; Grice & Segre 2011; Reid et al. 2011). Understanding interactions between pathogens and host microbes can enable us to manipulate microbial communities to improve health. Probiotic therapy through bioaugmentation is the augmentation of locally occurring protective bacteria to an individual or the environment with the purpose of altering the hosts' microbial community structure to mitigate disease (Haas & Défago 2005; Becker & Harris 2010; Gerritsen et al. 2011). Much research has targeted manipulation of microbiota in humans and aquacultural and agricultural species with positive and encouraging results (Fuller 1989; Verschuere et al. 2000; Kesarcodi-Watson et al. 2008). However, little research on disease mitigation using probiotics in nature has occurred despite the serious threats posed by emerging infectious diseases.
Amphibians are threatened by the fungal disease chytridiomycosis, which is associated with the dramatic declines or extinctions of over 200 amphibian species (Fisher et al. 2009; Kilpatrick et al. 2010). Chytridiomycosis is caused by the pathogen Batrachochytrium dendrobatidis (Bd) (Longcore et al. 1999) and is the largest disease threat to biodiversity at the present time (Wake & Vredenburg 2008; Crawford et al. 2010). Its devastating effects likely are amplified by interactions with other anthropogenic threats (Collins & Storfer 2003). There are some areas with diverse amphibian assemblages that are currently Bd-free, such as Madagascar, which provide an opportunity to proactively prevent further catastrophic amphibian declines and extinctions. Susceptible individuals in survival assurance colonies are also in need of repatriation. For these reasons, a feasible disease mitigation strategy is imperative. Accumulating evidence suggests that probiotic strategies can be effective for amphibians, perhaps because probiotics extend the hosts' innate immune system (Harris et al. 2009a,b; Vredenburg et al. 2011; Myers et al. 2012; Rollins-Smith & Woodhams 2012). Furthermore, probiotic therapy research is elucidating principles of microbial ecology including establishment, transmission and temporal dynamics of host-associated microbiota. Here, we review and synthesise current amphibian microbial ecology and bioaugmentation research and use this synthesis to define characteristics of effective probiotics. Past probiotic choices for laboratory and field trials have been based on incomplete information and were driven by the urgent need to protect amphibian populations. Therefore, we also develop sampling strategies and filtering protocols to guide the selection of amphibian probiotics, which is essential for a proactive rather than a reactive approach to disease mitigation in amphibians and other wildlife species.
Background and General Principles
Bd, a chytrid fungus, has two known life stages: a motile, flagellated zoospore and a sessile zoosporangium that resides in the amphibian epidermis (Berger et al. 2005). Released zoospores can either infect a new individual or re-infect the current host, meaning that infection and re-infection probability is a function of the hosts' defences, such as microbial defences, each generation. Periods of high host density, such as mating aggregations, can facilitate infection of new individuals (Kilpatrick et al. 2010). Several lineages of Bd have been identified. The hypervirulent global panzootic lineage is associated with massive declines and extinctions, and it tends to move in a wave (Lips et al. 2008; Farrer et al. 2011). Importantly, Bd persists in the environment due to some amphibians and perhaps other species acting as reservoirs (Reeder et al. 2012).
Amphibian defences against Bd include acquired immunity, innate immunity and cutaneous microbial communities, and these defences likely interact (Rollins-Smith & Woodhams 2012). The robustness of amphibians' acquired immune response to Bd is debated (Rosenblum et al. 2009; Ramsey et al. 2010; Savage & Zamudio 2011) and is mounted slowly if it occurs at all (Rollins-Smith 2009). Down-regulation of immune system genes (Rosenblum et al. 2009) and ineffective vaccination attempts (Stice & Briggs 2010) suggest a poor acquired immune response. Innate immune activity provides a non-specific defence against cutaneous pathogens (Rollins-Smith 2009). Cutaneous antimicrobial peptide (AMP) production is a main component of amphibian innate immunity (Rollins-Smith 2009), but lysozyme and small organic molecules, such as alkaloids, also may play a role (Rollins-Smith & Woodhams 2012). Innate and acquired immunity protect some amphibian species from Bd (Woodhams et al. 2007a; Savage & Zamudio 2011); however, they offer little hope to naïve, susceptible species unless natural selection increases the frequency of individuals with these genetically based defences against Bd. In some cases, Bd has caused extinctions (Fisher et al. 2009), which strongly suggests that evolution of adequate defences is not universal.
Amphibian skin harbours symbiotic resident microbes, which constitute the only line of defence that is not directly host produced and has been successfully manipulated to mitigate disease (Harris et al. 2009a; Vredenburg et al. 2011). Antifungal cutaneous microbes have been cultured from every host species sampled, suggesting they can play a role against various pathogens (Lauer et al. 2007, 2008). Growing evidence supports the hypothesis that antifungal skin microbes suppress chytridiomycosis (Becker et al. 2009; Harris et al. 2009a,b; Vredenburg et al. 2011; Muletz et al. 2012). Bacteria inhibit Bd directly through production of inhibitory metabolites and perhaps indirectly through immunomodulation where microbes regulate the production of host defences such as AMPs and lysozyme (Reid et al. 2011). A bacteria removal experiment with Plethodon cinereus demonstrated that individuals with reduced microbiota had higher morbidity than individuals with an unmanipulated microbiota when exposed to Bd (Becker & Harris 2010). In complementary probiotic experiments, amphibians inoculated with an anti-Bd bacterium had reduced morbidity and mortality from Bd (Harris et al. 2009a,b), which was associated with the presence of a bacterially produced anti-Bd metabolite (Harris et al. 2009a). Importantly, a field experiment involving bioaugmentation of an anti-Bd species, Janthinobacterium lividum, on Rana muscosa in the Sierra Nevadas showed that frogs treated with probiotic baths had lower peak infection loads than untreated controls (Vredenburg et al. 2011). One year after treatment, untreated controls were not recovered whereas 39% of probiotic-treated individuals were recovered (Vredenburg, pers. comm.), suggesting that probiotic treatment allowed individuals to persist by preventing Bd from reaching a lethal threshold. Although continued optimisation of probiotic selection and protocols are needed, this research demonstrates that bioaugmentation has tremendous potential to assist vulnerable amphibian populations.
Amphibian microbial community ecology and probiotics
Microbial community establishment begins at hatching and can be strongly influenced by parental microbes, the pool of microbes in the environment and the interaction between the host's immune system and mucous composition and colonising microbes (Fierer et al. 2012). Mucopolysaccharides secreted by mucous glands likely provide the resources needed for bacterial growth, which initiates microbial competition. A recent model suggests that the assembly of a beneficial microbiome is dependent on interference competition where threshold densities within and among taxa trigger anti-microbial metabolite production (Scheuring & Yu 2012). Therefore, both microbe–host and microbe–microbe interactions will dictate community establishment.
Microbes can be transmitted vertically, horizontally and environmentally, and probiotic bacteria can be transmitted by these mechanisms. Vertical transmission likely occurs in amphibians with parental care. The frog Hyalinobatrachium colymbiphyllum appears to transfer a defensive microbiota to embryos, which likely protect hatchlings from Bd (Walke et al. 2011). If vertical transmission occurs, it could lead to probiotic transfer and persistence between generations. Horizontal transmission has not been investigated in amphibians, but likely occurs during mating and during congregations in winter hibernacula. If horizontal transmission occurs at a high rate, fewer amphibians will need probiotic treatment because treated individuals could transfer the probiotic to untreated individuals. Environmental transmission has been demonstrated with Pl. cinereus, where the probiotic J. lividum was transmitted from soil to salamanders in a laboratory experiment (Muletz et al. 2012). This result suggests that environmental transfer occurs in nature. If this exchange is frequent then environmental probiotic inoculation could be effective, as it would allow numerous amphibians to acquire the probiotic without individual treatment. Pseudo-environmental transmission occurs when bacteria from parents or other individuals are transferred to the environment and then to offspring or other amphibians. These modes of transmission are not mutually exclusive and likely work in tandem to shape amphibian microbial communities.
Knowledge of amphibian cutaneous microbial community structure is increasing. Next generation sequencing of the 16S rRNA gene provides more complete estimates of community structure and diversity than culturing studies (Grice & Segre 2011; McKenzie et al. 2011). Using 454 pyrosequencing, McKenzie et al. (2011) found that amphibian microbial communities tended to be species-specific rather than environment-specific and that levels of microbial diversity differed among amphibian species. In addition, in several studies, a few microbial taxa were found across amphibian species and locations (Lauer et al. 2007, 2008; Woodhams et al. 2007b; Lam et al. 2010; McKenzie et al. 2011), suggesting some cutaneous symbionts have a broad host range.
Cutaneous microbial community structure and its stability are likely associated with disease outcome. In humans, it is not clear whether a certain community structure leads to disease or is a consequence of disease (Grice & Segre 2011); however, in amphibians, experimental studies show that altering their microbial community affects disease susceptibility (Harris et al. 2009b; Becker & Harris 2010). The stability of amphibian cutaneous microbial communities is linked to microbial maintenance. Microbes can be maintained after disturbances such as skin sloughing through environmental re-inoculation or from bacterial reservoirs on the host (Meyer et al. 2012; Muletz et al. 2012). In Pl. cinereus, concentrations of bacteria were found in gland openings (Lauer et al. 2007) that may provide a “seed bank” from which microbes can repopulate the skin. The rate of skin sloughing and microbial repopulation likely influences disease risk (Myers et al. 2012). At metamorphosis, the microbial community of aquatic larvae may shift to adjust to terrestrial conditions and changing host immunity, and this could cause a period of instability that affects disease susceptibility.
Microbial community structure likely fluctuates, but an important question is whether community function, including defensive function, will remain constant (Fierer et al. 2012; Huttenhower et al. 2012). Defensive function on amphibian skin can be assayed by determining the relative abundance of anti-Bd bacterial metabolites using HPLC-MS (Brucker et al. 2008a) or with total cutaneous molecule bioassays (Box 2). For defensive function, the specific bacterium may not be important, but rather the genes it carries. Horizontal gene transfer (HGT) plays a role in functional stability in the human gut microbiome (Smillie et al. 2011). Probiotic species that have anti-Bd metabolite genes on plasmids could readily pass these genes to other community members and contribute to functional stability (Robinson et al. 2010). The potential role of HGT in amphibian microbial communities and defensive function is unexplored. With a better understanding of community structure and functional stability and its relationship to chytridiomycosis, disease susceptibility can be predicted and interventions to establish protective microbial communities can be implemented.
Ecosystem function, including resistance to pathogen invasion, can improve as species diversity increases (Balvanera et al. 2006; van Elsas et al. 2012; Fig. 1), but it is also possible that the role of individual species trumps diversity (Lyons et al. 2005; Box 1). Few studies have used controlled experiments to relate microbial community diversity to defensive function. One experiment demonstrated that higher locust gut microbial diversity increased disease resistance (Dillon et al. 2005). Preliminary evidence from Australian Wet Tropics frogs indicates that Bd infection intensity is negatively correlated with anti-Bd bacterial richness, suggesting a possible role of diversity in defensive function. Alternatively, an individual species can provide a disproportionate share of the community's defensive function. To date, experimental evidence suggests that the addition of one probiotic species can increase defensive function (Becker et al. 2009; Harris et al. 2009a,b; Muletz et al. 2012). The degree to which diversity and key bacterial species provide defence against disease will dictate bioaugmentation strategies (Box 1).
Selection at the individual microbe, individual amphibian and amphibian population levels are expected following exposure to Bd. As Bd invades amphibian skin, selection for anti-Bd microbial genotypes should occur. Amphibian hosts in a population have different microbial communities that can vary in defence against Bd, and individuals with a more protective microbiota will have a higher fitness. In addition, Bd-naive populations are likely to differ in the proportion of individuals with protective microbial species (Lam et al. 2010). Populations with a high proportion of individuals with a protective microbiota appear to benefit from an analogue of herd immunity, in which an infectious disease dies out when the proportion of immunised or protected individuals is above a threshold value (Woodhams et al. 2007b; Lam et al. 2010). Therefore, one goal of probiotic therapy is to increase the proportion of individuals in a population that have protective microbes (Fig. 1). It is important to note that the pathogen is expected to evolve resistance to host and microbial defences and that evolution of host microbial defences is faster than evolution of host-produced defences because of the shorter generation time of microbes.
Microbial community structure may be heritable and therefore can respond to selection. Vertical and pseudo-environmental transmission are possible mechanisms of resident microbe heritability. In addition, amphibian AMPs and antibodies likely are important forces structuring amphibian microbial communities, as they are in humans (Grice & Segre 2011; Gallo & Hooper 2012). Genetically based differences in immune system characteristics can therefore lead to different and heritable community structures even without vertical transmission if environmental reservoirs are available. Evidence of a genetic component to microbial community structure in human intestinal tracts has been detected (Zoetendal et al. 2001), although recent studies have found low heritability (Yatsunenko et al. 2012). Knowledge of the heritability of amphibian microbial communities is lacking but will be important for understanding community dynamics and its evolution.
Box 1. Keystone-probiotic hypothesis
Recently, Hajishengallis et al. (2012) developed a keystone-pathogen hypothesis stating that microbial pathogens in low abundance can dictate outcomes of certain diseases, such as periodontal disease or inflammatory bowel diseases, by restructuring the normal microbiota into a dysbiotic state. Certain immunopathologies, chronic diseases, and cancers have also been posited with this trigger (Ewald 2010). This differs from the traditional understanding that the microbial pathogen causes disease, in part, by increasing in abundance (Hajishengallis et al. 2012). Our current hypothesis is that a probiotic protects an amphibian by becoming an abundant member of the cutaneous community and by producing anti-Bd metabolites (Brucker et al. 2008a,b; Becker et al. 2009). For example, Becker et al. (2009) found an association between the concentration of violacein, a bacterially produced anti-Bd metabolite, and survival of salamanders when exposed to Bd. The concentration of violacein was correlated with J. lividum abundance, which is the bacteria that produces it, suggesting that a more abundant probiotic will be more protective.
We use a similar framework to develop a keystone-probiotic hypothesis. A keystone-probiotic will be in low abundance, have a significant impact on community structure leading to a greater defensive function, and may not itself provide a health benefit but causes changes in the microbial community that benefit host health. The value of a keystone-probiotic is that it remodels the cutaneous microbial community into one that is stable and provides increased defensive function (Fig. 1). Community restructuring can be caused by the keystone-probiotic in several ways. The probiotic can stimulate host immune defences, such as AMP production, that can differentially affect microbial community members. Host-produced nutrients in mucosal secretions can also be manipulated by keystone-symbionts; for example, Bacteroides thetaiotaomicron help stabilise gut microbiota by inducing gut epithelial cells to produce fucosylated glycans (Bäckhed et al. 2005). Furthermore, the probiotic might differentially affect an important competitor in the microbial food web and either increase or decrease its population density and therefore lead to an altered community structure. If the resulting community inhibits pathogens by mechanisms such as increased anti-Bd metabolite production or spatial competition, then the probiotic has a keystone effect. The probiotic also might have a disproportionate effect by acting synergistically with resident members of the microbiota to inhibit Bd. For example, the probiotic might induce resident microbes to produce anti-Bd metabolites. Importantly, a keystone-probiotic must bring about a stable community, minimise pathogen colonisation, and prevent pathogen densities from increasing. At this point, it is not known whether the success of probiotic manipulations is due to direct effects of the bacterial species producing anti-Bd metabolites, indirect effects of community restructuring, synergistic interactions with resident microbes, induction of host responses or a combination of these mechanisms.
These two alternative models – a probiotic that becomes abundant and a keystone-probiotic that has large effects through community restructuring while remaining relatively rare – have implications for disease mitigation. Species that are relatively rare are subject to stochastic loss, especially during disturbances such as skin sloughing (Meyer et al. 2012). Thus, a keystone-probiotic may need continual replenishment via environmental transmission. Both models predict pathogen protection with at least some community restructuring, so an important criterion is stability of the new protective community.
Next generation sequencing data can lead to mechanistic insights. It is possible to correlate individual species' abundances with community structure, and this analysis will predict which species dictate characteristic community structures. These community structures also can be correlated with disease outcomes, and therefore a correlation between keystone species and disease outcome can be determined. Survey data can suggest which bacteria will be effective probiotic candidates regardless of whether they are relatively rare or relatively abundant on amphibians. If these species are culturable, they can be tested with the filtering protocols presented in the text. Of course, experimental manipulations, which are feasible with amphibian hosts, are necessary to determine cause and effect. For example, experiments can be conducted where a potential keystone-probiotic is added to a variety of resident microbial community structures and response to Bd infection can be measured. These experiments can determine the roles of keystone species, microbial diversity and their interaction leading to protection from Bd.
Characteristics for an Effective Probiotic against Chytridiomycosis
Understanding the ecological interactions that govern microbial community dynamics on amphibian skin is key to defining effective probiotic characteristics. The primary roles of a probiotic are to prevent Bd from colonising amphibian skin and to prevent minor infections from escalating to a lethal threshold (Vredenburg et al. 2010). These roles, along with the probiotic candidates' interactions with the host, the cutaneous microbial community and the environment, are important and must be considered when selecting bacteria for bioaugmentation strategies.
Interactions between Bd and the probiotic are a primary factor influencing a probiotic's ability to repel and inhibit Bd (Table 1). Bd requires space and nutrients to colonise, grow and develop. High densities of resident bacteria can limit attachment sites and available nutrients for Bd (Collado et al. 2008; Mohapatra et al. 2012), thereby decreasing zoospore colonisation and slowing development. Effective probiotic therapy requires that the probiotic bacteria be an effective competitor and colonise and persist on the skin, especially in regions that are typically infected by Bd, such as the ventral surface, limbs and feet (North & Alford 2008). Bacterial spatial distributions have not been investigated, but could be determined by sampling different body regions or visualising bacteria with fluorescence microscopy. With this knowledge, candidate probiotics could be graded on where and how densely they colonise and persist on amphibians.
Table 1. Summary of the characteristics of effective probiotics based on four interactions
Characteristics of effective probiotics
Repel and inhibit Bd
Maintain defensive function in presence of Bd
Probiotic-resident microbial community
Coexist with functionally important bacterial species
Positively interact with resident bacteria
Shift microbial community to a defensive state (Box 1)
Colonise and persist on host
Positively interact with host-produced defences
Do no harm to host
Inhibit Bd under appropriate ecological contexts
Have minimal non-target effects
Form a self-disseminating system between amphibian and the environment
Bacterially-produced metabolites are responsible for both repelling and inhibiting Bd. We have evidence that two metabolites (2,4-diacetylphloroglucinol [2,4-DAPG] and indole-3-carboxaldehyde [I3C], Fig. 2) repel Bd in laboratory assays (Lam et al. 2011). This negative chemotaxis may prevent colonisation or re-colonisation of zoospores discharged from zoosporangia. Bacterial metabolites (violacein, 2,4-DAPG, I3C) also inhibit Bd growth in in vitro bioassays (Brucker et al. 2008a,b). Importantly, metabolites were detected on amphibian skins in nature at concentrations that were inhibitory in laboratory assays (Brucker et al. 2008b).
Two bacterial addition experiments, where individuals were immersed in a probiotic bath, demonstrated a strong association between violacein concentration and survival after Bd exposure. In the first experiment, addition of the violacein producer, J. lividum, to R. muscosa led to significantly higher violacein concentrations on skins and higher survival compared with untreated controls (Harris et al. 2009a). In the second experiment, a threshold level of violacein on Pl. cinereus was associated with survival (Becker et al. 2009).
Microbial defences are likely achieved as a by-product of inter- and intra-specific microbial competition. Quorum sensing, a process of cell-to-cell communication where a threshold population density regulates gene expression, can trigger metabolite production. Bacterial species vary in their threshold densities (Mohapatra et al. 2012); therefore, an effective probiotic will produce anti-Bd metabolites at fairly low cell densities or will grow rapidly to reach a density where metabolites are produced. A fast-acting and stable defence is likely to be dependent on rapid metabolite production upon initial colonisation and re-colonisation after cutaneous disturbances. Furthermore, a probiotic must maintain its defensive function in Bd's presence. Studies in our laboratories have shown that some bacteria, when co-cultured with Bd, are later unable to inhibit Bd, suggesting Bd possesses a mechanism that down-regulates bacterial metabolite production. Such species would not be appropriate as probiotics. Research has focused on detection of small organic metabolites, but bacterially produced defensive peptides such as bacteriocins, documented in human skin microbiota (Gallo & Hooper 2012), also warrant investigation.
A probiotic's interactions with its host will determine its effectiveness at inhibiting Bd and decreasing Bd-associated morbidity and mortality (Table 1). In order for probiotic bacteria to colonise and persist successfully, they must use cutaneous resources provided by the host or resident microbes and not be inhibited by host immune defences. Strong evidence indicates that anti-Bd bacteria species, such as Pseudomonas fluorescens, are not inhibited by moderate to low concentrations of amphibian AMP mixtures (Myers et al. 2012), but this situation is not universal (Schadich & Cole 2010). Certain symbiotic bacteria can induce AMP production through immunomodulation (Grice & Segre 2011; Reid et al. 2011; Mohapatra et al. 2012); therefore, probiotic bioaugmentation may trigger production of Bd-inhibitory AMPs (Rollins-Smith & Woodhams 2012). Alternatively, microbes can reduce the need for costly peptide production. For example, Woodhams et al. (2012a) found that AMP production increased in response to Bd infection only when microbial defences were experimentally reduced.
A successful probiotic will not be pathogenic to the host or trigger a negative reaction from the innate or acquired immune system. The period after an amphibian hatches is likely a critical time of microbial community establishment when the acquired immune system adjusts to a resident microbiota. Treating larvae and hatchlings with a probiotic may increase the chances of acceptance by the host immune system and allow persistence. In aquaculture, treatment of larval stages often leads to increases in survival compared to control treatments (Nogami & Maeda 1992; Kesarcodi-Watson et al. 2008). For later stages, augmenting a probiotic already found on the host species will be less likely to trigger a negative host reaction.
An effective probiotic will work additively or synergistically with host AMPs. The metabolite 2,4-DAPG, produced by Ps. fluorescens, worked synergistically in vitro with the R. muscosa AMP mixtures to inhibit Bd, meaning that relatively low concentrations of each molecule were required for inhibition when they occurred together. Importantly, Ps. fluorescens was not killed by AMP concentrations needed for synergistic inhibition (Myers et al. 2012). This interaction with host immunity suggests the basis of a mutualism between amphibian hosts and their bacterial symbionts.
Probiotic-resident microbial community interactions
A probiotic's interaction with resident skin microbiota will influence its efficacy (Table 1, Box 1). A probiotic should not eliminate functionally important resident bacteria. In addition, it may be important to select a probiotic that synergises with resident microbes. We have preliminary evidence from in vitro assays that four inhibitory bacteria species collected from Pl. cinereus work additively or synergistically to inhibit Bd when their culture filtrates are mixed in pair-wise combinations. Further work is necessary to see if bacterial synergies occur in vivo.
The environmental context in which a probiotic will be used must be considered, as it will affect its defensive function (Table 1). For example, temperature affects the pathogen, host and cutaneous microbes (Rohr & Raffel 2010; Daskin & Alford 2012). Amphibian immune function has an optimal temperature range that may not correspond to the optimal temperature range for Bd growth (17–25 °C) (Piotrowski et al. 2004; Woodhams et al. 2008). Bd prevalence is greater in cooler seasons in temperate ecosystems and montane tropical regions (Raffel et al. 2006), which could be a function of reduced host defences. Ideally, an effective probiotic will compensate by functioning outside of the optimal temperature range of host immunity. Furthermore, it is important to eliminate a bacterial species that inhibits Bd at one temperature, but facilitates it at another. An ideal probiotic should maintain its defensive function over an ecologically relevant temperature range.
Preventing Bd-associated population collapses and successfully repatriating amphibians from assurance colonies can be aided if the probiotic forms a self-disseminating system between the amphibian and the environment (Muletz et al. 2012). Amphibians likely obtain their microbiota from the environment at some point during development. This transfer from the environment may occur continually, making persistence of microbial communities and probiotics in part dependent on environmental sources. If so, the need for bioaugmentation could be linked to changes in microbial communities in soil and water due to factors such as climate change and pesticide plumes (Belden & Harris 2007). Laboratory trials have demonstrated environmental transmission (Muletz et al. 2012). However, studies are urgently needed to determine whether broad-scale probiotic environmental inoculation can be transmitted to amphibians and confer defence against Bd, whether environmental inoculation can lead to a self-disseminating system, and whether non-target ecosystem effects occur. Some agricultural studies suggest that broad-scale inoculations can be safe and effective (Scherwinski et al. 2008), but careful testing to ensure biosafety is necessary.
The range of interactions among Bd, skin bacteria, host and environment leads us to propose sampling strategies and filtering protocols that are designed to guide selection of effective probiotics for protecting individual species and amphibian communities. The filtering protocol differs from listing effective probiotic characteristics as presented in other studies (Fuller 1989; Kesarcodi-Watson et al. 2008). Isolates are placed through a series of tests that progressively filter out ineffective ones, leaving the most promising candidates. A species-specific approach focuses on treating at-risk individuals with probiotic baths while a community-based approach targets amphibian assemblages by treating ponds or local areas with a broad-spectrum probiotic. We stress that bioaugmentation approaches must use microbes found in the local environment to improve success and minimise biosafety concerns.
Species-specific probiotics should target individuals being repatriated from survival assurance colonies (Becker et al. 2011) and individuals of critically endangered species in front of an advancing Bd wave (Woodhams et al. 2007b; Vredenburg et al. 2010). Assurance colonies have been implemented to rescue species when they are experiencing rapid declines across their range or when there is an imminent threat to amphibian populations due to the anticipated arrival of Bd. The goal of assurance colonies is to reintroduce threatened species to their native habitats and establish persisting populations; however, releasing susceptible individuals will be unsuccessful because Bd persists in the natural environment on reservoir species (Reeder et al. 2012). In addition, there are situations where susceptible species in front of an advancing Bd wave in the wild are not in assurance colonies (Vredenburg et al. 2010). In both cases, individuals can be treated with a probiotic derived from the appropriate sampling strategy and that successfully passes through the filtering protocol outlined below.
The sampling strategy for obtaining a species-specific probiotic for assurance colony species and endangered wild species will differ for species that have some populations in the wild coexisting with Bd (e.g. Anaxyrus boreas and R. muscosa), and species that are extirpated from the wild (e.g. Atelopus zeteki). If there are populations coexisting with Bd, it is essential to sample and culture microbes from members of these populations (Fig. 3a). For species that have been extirpated from the wild, it will be necessary to focus microbe sampling on related species that have a similar life-history, are found in similar habitats and locations, and are coexisting with Bd (Fig. 3b). Individuals in populations coexisting with Bd are surviving with Bd infection and are more likely to have anti-Bd bacteria.
Once microbes are collected from amphibians using standard methods (Harris et al. 2006), they must pass through the filtering criteria, which leaves a progressively smaller number of probiotic candidates. The candidate probiotics must inhibit Bd under ecologically relevant conditions of the intended host (Fig. 3 (SS1), Box 2). For example, it is essential that probiotics inhibit at temperatures at which the amphibian is most vulnerable to Bd infection (Daskin & Alford 2012). Preference should be given to inhibitory isolates that are present on a large proportion of sampled individuals since ubiquity suggests the isolate will persist on the target amphibians. The remaining candidate probiotics must colonise and persist on target amphibians at all life-history stages while not harming the host (Fig. 3 (SS2), Box 2). If bacteria are collected from surviving individuals of the intended host, the likelihood of bacterial persistence is high. If persistence is observed, it indicates that the host's immune system or resident microbes do not inhibit the isolate. It will be important to eliminate isolates that inhibit Bd in vitro, but do not persist or provide continual inhibition of Bd on amphibians (Box 2). The remaining candidates must inhibit Bd in clinical trials with all life-history stages to confirm in vivo effectiveness of the candidate probiotic to prevent disease (Fig. 3 (SS3), Box 2). Successful probiotics will decrease mortality and sub-lethal effects for all stages. Lastly, selected isolates must inhibit Bd in a small-scale field trial to assess effectiveness in the natural environment (Fig. 3(SS4), Box 2). At this point, remaining candidates have a high likelihood of being effective probiotics for the target amphibians.
Two amphibian species currently established in assurance colonies, the boreal toad, An. boreas, and the Panamanian golden frog (At. zeteki), are targets for pre-release probiotic treatment. The toad An. boreas is a species that has experienced population declines (Muths et al. 2003); however, there are wild populations persisting with Bd infection that should be sampled to obtain probiotic candidates (Fig. 3a). The collected microbes should be screened through the four-step filter discussed above. At. zeteki is a species that is likely extirpated from the wild. Becker et al. (2011) tested a probiotic candidate, J. lividum, on At. zeteki, which was isolated from North American salamanders. The probiotic treatment kept infection loads low initially, but the probiotic abundance declined and mortality occurred. Subsequently, 600 isolates were collected from related species coexisting with Bd in the same locations and habitats where At. zeteki was found and are currently being screened using the criteria listed above. Importantly, inhibition trials (step 1) removed 85% of isolates from consideration as a probiotic (Fig. 3b).
The susceptible frog in the Sierra Nevadas, R. muscosa, is an example of a species where populations in front of an advancing Bd wave are in need of protection. Probiotic candidates have been collected from populations that have persisted through the arrival of Bd and are therefore more likely to possess Bd-inhibitory bacteria (Fig. 3a) (Woodhams et al. 2007b; Lam et al. 2010). One R. muscosa population under threat from imminent Bd arrival, and predicted to be decimated, provided an opportunity for probiotic application. Due to the short lead-in time available, it was not possible to apply all elements of the filtering process. However, the probiotic J. lividum was chosen due to its success in previous experiments (Harris et al. 2009a) and its presence on a number of amphibian species across many locations, including the Sierra Nevadas. This trial was successful: greater survival was seen for treated individuals, and Bd loads remained low compared to untreated controls (Vredenburg et al. 2011). Therefore, when immediate treatment is necessary, and little time exists for a full filtering process, priority can be given to probiotics that have been successful in other studies, assuming that a strain of the probiotic can be found on amphibians in the intended application area.
Box 2. Methodologies of filtering protocol
To determine the inhibitory nature of the candidate probiotics, we advocate the following protocols. The bacterial isolate should be co-cultured with Bd, because it will induce the bacteria to produce anti-Bd metabolites. In addition, isolates that are inhibited by Bd will be excluded. The culture filtrate (cell-free supernatant) that includes bacterial metabolites from the co-culture is assayed for Bd inhibition in 96-well microtiter plates (Bell et al. 2013). A negative control (heat-killed Bd), positive control (Bd without culture filtrate but with the equivalent volume of medium) and a control for Bd-produced metabolites (culture filtrate from a Bd culture) should be included. Inhibition assays can also be carried out on agar plates (Harris et al. 2006). In this protocol, Bd is spread evenly across the tryptone-agar plate and bacteria are streaked across the Bd-covered plate. After 72–96 h of incubation, the inhibition zone is measured. Trials should be replicated to accurately estimate inhibition and allow for statistical tests.
Colonisation & persistence trials
To assess colonisation and persistence, candidate probiotics are inoculated onto amphibians of all life-history stages in laboratory trials. For species-specific treatment strategies, amphibians are bathed in probiotic baths; for community treatment strategies, the housing substrate is inoculated with the probiotic. Colonisation and persistence can be assessed using culture-based or molecular methods (Becker et al. 2011). If culture-based techniques are used, artificial selection of bacterial isolates for rifampicin resistance can facilitate tracking during experiments (Muletz et al. 2012). For molecular detection, polymerase chain reaction (PCR) can be used to confirm colonisation and persistence of the probiotic. This technique requires the use of species-specific primers, which have been developed for some species such as J. lividum (Harris et al. 2009a). In all experiments, control groups of untreated amphibians are required. Ideally, during these trials swabbing or bathing should be used to periodically collect amphibian skin secretions, which are a mixture of defensive products of amphibians and their microbial symbionts. This protocol is currently being optimised. These secretions are used in Bd inhibition assays to compare control treatments (no probiotic) to probiotic treatments as a measure of the probiotic's in vivo effectiveness against Bd. Because these bioassays assess in vivo effectiveness of potential probiotics, they reduce the possibility of unsuccessful clinical trials.
Environmental persistence trials
Probiotic persistence in the environment is determined through laboratory trials where an environmental substrate is inoculated with the probiotic candidate (Muletz et al. 2012) and monitored over time. Depending on the habitat of the intended hosts, trials are conducted with water or soil as the substrate. Probiotic transmission can also be assessed if amphibians are housed in the inoculated substrate. Transfer of the probiotic to the host and persistence in the environment can be measured using culture-based or molecular methods (Becker et al. 2009; Muletz et al. 2012). A similar protocol can be used for trials conducted in nature.
Laboratory-based clinical trials for species-specific probiotic treatment involve bathing amphibians in the probiotic and exposing both treated individuals and untreated controls to Bd in randomised, replicated trials (Harris et al. 2009a). Clinical trials for community-based probiotics involve inoculating the laboratory environment (water or soil) with the candidate probiotic and housing the selected host amphibians in these treated environments as well as housing a set of individuals in untreated control environments. Amphibians in both treatments should be exposed to Bd and monitored for survival and sublethal effects (i.e. growth rate, behaviours) (Harris et al. 2009a,b). Estimating Bd loads via qPCR (Hyatt et al. 2007) can be helpful in determining whether the probiotic kept Bd loads below a lethal threshold. These trials need to be replicated and conducted under ecologically relevant conditions. In addition, they should be conducted on all life-history stages, (i.e. larvae, juvenile, adult) to ensure the probiotic is effective across all stages.
Small-scale probiotic field trials should be completed at locations where appropriate regulatory approval has been obtained. For species-specific strategies, field trials involve treatment of individuals with and without a probiotic bath and release at the field location (Vredenburg et al. 2011). Monitoring of Bd infection, the establishment of the probiotic on amphibians and ultimately the survival of released individuals will determine the outcome of the experiment. Field trials for community-based environmental treatment involve inoculation of soil or water with a probiotic and release of amphibians to treated areas. Survival of amphibians at the treated sites, Bd loads and probiotic abundance on the hosts and in the environment should be monitored and compared to control sites to evaluate success.
There are large areas with diverse amphibian assemblages, such as Madagascar, that remain Bd-free; however, its arrival is inevitable (Fisher & Farrer 2011; Box 3). Ecological niche modelling predicts suitable habitat for Bd in regions of Madagascar where the amphibian diversity is the greatest (Andreone et al. 2005; Rödder et al. 2009; Fig. 4). In this situation, the opportunity exists to be proactive and prevent the loss of many amphibian species. Probiotic conservation approaches for naïve communities differ from species-specific conservation efforts. The goal for community treatment is not to have a probiotic specific to one amphibian species but to have one or more probiotics that are suitable for multiple amphibian hosts. Importantly, certain anti-Bd species can exist on a number of diverse host species. For example, the probiotic species, J. lividum, has been found on two species of plethodontid salamanders in Virginia, on one species of high-altitude pond-dwelling frogs in California, on one species of low elevation frogs in Switzerland and on three species of high-altitude rainforest frogs in Ecuador (Lauer et al. 2007, 2008; Woodhams et al. 2007b). Certain anti-Bd genera, such as Pseudomonas, also have been commonly found on amphibians (Lauer et al. 2007, 2008; Walke et al. 2011), suggesting that these taxa could be good community probiotic (Harris et al. 2009b). The genus Curvibacter has also been found on multiple amphibian species (McKenzie et al. 2011). If it inhibits Bd, it could act as a broad-spectrum antifungal probiotic, just as commercially available agricultural probiotics have a broad host range (Berg 2009). Community-based treatment ideally would occur through environmental bioaugmentation, where one or a few environmental inoculations would allow numerous amphibian species as well as both larval and adult stages to be treated without individual capture.
The sampling strategy for obtaining a community-based probiotic will differ for amphibian communities that have neighbouring communities persisting with Bd [e.g. areas in Panama (Fig. 3c) (M. Hughey, pers. comm.)] and those that do not [e.g. Madagascar (Fig. 3d)]. In the first case, a wave of Bd is moving forward, but there are areas that remain Bd-free in addition to areas behind the wave that are now persisting despite infection. Bacteria that have a high prevalence on species persisting with Bd are more likely to be inhibitory and should be sampled to find effective probiotics (Fig. 3c). Some large geographical areas, such as Madagascar, are Bd-free to date and there are no amphibians coexisting with Bd from which to collect bacteria (Box 3). Under this scenario, broad-scale microbial surveys of amphibians need to be completed proactively (Fig. 3d).
After culturing the collected bacteria, they must pass a series of filtering criteria, similar to that described above. Candidates must inhibit Bd under ecologically relevant conditions representative of the intended community (Fig. 3 (CB1), Box 2). Preference should be given to inhibitory isolates that are found on a high proportion of species. The remaining candidates must persist in environmental conditions, such as temperature and pH, representative of the target application area (Fig. 3 (CB2), Box 2). Those that do so must also colonise and persist on hosts via environmental transmission (Muletz et al. 2012) (Fig. 3 (CB3), Box 2). Candidates that remain must reduce infection and the effects of chytridiomycosis in randomised, replicated clinical trials (Fig. 3 (CB4), Box 2). Finally, remaining isolates must maintain their effectiveness on amphibians in the natural environment (Fig. 3 (CB5), Box 2). All trials involving host amphibians should be conducted on a sample of phylogenetically diverse host species and all life-history stages under ecologically relevant conditions. The isolates that make it through this filtering process will likely be strong probiotics for community-based treatment. Finding effective community-based probiotics will require a concerted effort; however, the potential benefits in terms of preventing amphibian extinctions are substantial. As we learn more about mechanisms of inhibition as a function of probiotic species, the host and community context, the filtering criteria can be optimised for both species-specific and community-based treatment modalities.
Box 3. Threatened amphibian species in Madagascar
Madagascar, a global biodiversity hotspot, has over 400 species of amphibians, 99% of which are endemic (Fisher & Farrer 2011; Lötters et al. 2011). Much of Madagascar's rich amphibian fauna inhabits regions predicted by environmental niche modelling to be climatically suitable for Bd (Andreone et al. 2005; Rödder et al. 2009). Pond-breeding and stream-dwelling species are also predicted to be at risk of decline or extinction based on these life-history traits, and indeed, susceptibility trials indicate that tested Malagasy frog species will succumb to chytridiomycosis (C. Weldon, pers. comm.; Vredenburg et al. 2012).
Currently, Bd has not yet been documented on the island but its arrival is imminent (Lötters et al. 2011). Surveys of over 50 species from 12 localities of differing altitudes and biogeographical regions did not find Bd (Vredenburg et al. 2012). However, Bd's dispersal ability is unquestionably high, considering its rapid spread around the globe, and therefore it is very likely to invade Madagascar. It has been proposed that human-mediated introduction of Bd via the amphibian trade is likely (Lötters et al. 2011). Once Bd is introduced, it has the potential to spread rapidly as seen in South America (Lips et al. 2008).
It is imperative to consider a prevention and mitigation strategy now in order to prevent catastrophic declines and extinctions in Madagascar like those seen in Central America and tropical Australia. Lötters et al. (2011) explain that effective responses for this potential threat include an increase in biosecurity, the development of breeding procedures for representatives of all major clades of Malagasy amphibians and the development of plans for ‘emergency response’. We suggest that development of probiotic disease mitigation strategies should be included in conservation planning, as they will allow for a proactive response. Currently, nothing is known about the microbiota of Malagasy amphibians, and therefore the identification of anti-Bd bacteria is urgently needed. Ideally, broad-spectrum probiotics that are effective within certain frog assemblages or within particular habitats will be identified. Tropical montane regions are often optimal habitats for Bd growth, and in general probiotics should be developed for amphibians in habitats predicted to be at high risk for decimation by Bd (Fig. 3). Using the sampling strategies and filtering protocols developed here, probiotics can be identified and an extinction crisis can be averted.
Application of probiotic bacteria
Optimisation of protocols for probiotic application is essential. There are two ways to apply probiotics: individual treatment and environmental treatment. For individual probiotic bath treatment, colonisation success could be increased by first reducing the resident microbiota. In some laboratory experiments to date, amphibians first had their existing microbiota reduced by treatment with 3% hydrogen peroxide, antibiotics or both (Harris et al. 2009a; Becker & Harris 2010; Vredenburg et al. 2011) to open an accessible niche for the probiotic (Reid et al. 2011). However, pre-treatment is not always necessary and could remove microbes that facilitate probiotic establishment or add defensive function. Ps. reactans was added successfully to Pl. cinereus without pre-treatment and led to lower morbidity effects in a laboratory experiment (Harris et al. 2009b). Similarly, in an agricultural study with probiotic treatment of wheat seeds, a probiotic's ability to increase yields in both laboratory and field trials was independent of pre-treatment disinfection (Pierson & Weller 1994). In the aquaculture literature, pre-treatment to reduce the existing microbiota typically is not done, and there have been many studies showing the efficacy of aquacultural probiotics (Verschuere et al. 2000; Kesarcodi-Watson et al. 2008). Applying a high density of the probiotic can be sufficient to ensure establishment, perhaps by giving the probiotic a competitive advantage. For individual treatment it also is necessary to determine the appropriate probiotic exposure time. In amphibian studies that showed a protective effect, a one-time probiotic bath between 2 and 48 h was used, suggesting that a bath within this time range is adequate for probiotic transmission (Harris et al. 2009a,b; Vredenburg et al. 2011).
Individual probiotic treatment has worked effectively in pond environments, where there is a high probability that pond-dwelling species can be captured (Vredenburg et al. 2011). In an aquacultural context, probiotic bath treatment of rainbow trout successfully reduced mortality (Gram et al. 1999). In agricultural contexts, seeds are often bacterised, which is analogous to probiotic bathing, and this treatment leads to improved survival (Haas & Défago 2005; Quagliotto et al. 2009).
In large-scale bioaugmentation field applications, hand-capturing amphibians and bathing frogs individually in probiotics is not possible in all situations, and environmental treatment may be a better option. When environmental treatment is feasible, it can be accomplished by soil or water inoculation. Studies in terrariums suggest that a probiotic can be established successfully in soil (Muletz et al. 2012). Suppressive soils, characterised by their ability to inhibit pathogens, have been added to agricultural environments to increase crop yield (Stutz et al. 1986), which suggests environmental inoculation is effective. The majority of amphibian species that have declined are aquatic breeders (Kriger & Hero 2007); therefore, inoculation of aquatic breeding sites could be a successful strategy. Studies to assess the efficacy of aquatic treatments are in progress. Environmental inoculation of aquacultural ponds can increase survival of farm-raised species (Moriarty 1998). For stream environments, it will be necessary for the probiotic to establish in the substrate such that bacterial reproduction is greater than emigration due to stream flow. Large-scale environmental inoculations are contingent on determining their efficacy and addressing safety concerns in small-scale field trials.
For both individual and environmental treatments, the optimal bacterial concentration to use and the optimal number of probiotic applications can be determined. For example, in aquacultural systems, probiotic concentration has been varied experimentally and effects on growth and survival have been assessed (Gatesoupe 1997). The time of appropriate within-year application for amphibians also needs to be determined and may be at the onset of breeding as treatment is likely to reduce pathogen transmission as well as increase probiotic transmission during mating aggregations. Many species with a larval stage experience mortality due to Bd spreading across the skin as keratinised epidermal tissues develops at metamorphosis; therefore, it is important to develop probiotics that are successful for larvae. Treating the larval stage can lower Bd transmission between larvae and post-larvae if both stages coexist in the same habitat.
Continued research is necessary to identify amphibian communities that are Bd-naïve so these communities can be prepared for Bd arrival. The optimal time of probiotic application in relation to Bd arrival also needs to be determined. In laboratory experiments, the probiotic control treatment (probiotic without Bd) has not caused any detectable morbidity or mortality (Harris et al. 2009a,b); therefore, areas in the path of an advancing Bd wave could be treated before Bd arrives. It is also possible to treat amphibians as Bd begins to emerge. The recent probiotic field trial in the Sierra Nevadas was effective in treating mildly infected frogs (Vredenburg et al. 2011).
Species-specific individual treatment and community-based environmental treatment need not be mutually exclusive. The most desirable protocol may be to individually treat as many members of a population as possible and also to inoculate the environment with the same probiotic to establish or re-establish a self-disseminating system of defensive microbes. It is possible that the best probiotic for a highly susceptible species is not suited for other amphibian species in an assemblage or vice versa. Under these circumstances, the highly susceptible species could be treated individually with a specific probiotic bath, and the environment could be treated with a broad-spectrum probiotic intended for the amphibian community. This could reduce infection in less susceptible reservoir species, therefore contributing to protection of the highly susceptible species. In addition, environmental application could reduce transmission by killing zoospores in the environment.
Ultimately, evaluation of success will be measured in terms of amphibian population survival and persistence. These measures can be estimated by visual encounter surveys and mark-recapture population size estimates (Heyer et al. 1994). Publication of all probiotic trial results is critical for effective protocol development and for avoiding repetition of unsuccessful experiments that are time-consuming and resource draining (Woodhams et al. 2012b).
The addition of bacteria to an ecosystem has the potential to affect non-target species and ecosystem processes (Simberloff & Stiling 1996). Importantly, a probiotic should not negatively impact human health. Bacterially-produced compounds can be toxic to aquatic organisms. For example, violacein is acutely toxic to bacterivorous nanoflagellates (Matz et al. 2004), which may lead to increases in the bacterial community. Ecosystem processes such as decomposition or primary productivity also could be affected. It is not known whether the addition of anti-fungal bacterial species will have negative impacts on other microbial species; however, agricultural studies suggest that probiotic additions have minimal and transient effects on the microbial community structure (Scherwinski et al. 2008; Edel-Hermann et al. 2009). Nonetheless, precautions to minimise non-target effects and maintain the integrity of the ecosystem are essential.
Continuing Research and Future Directions
An alternative probiotic selection approach is to begin with surveys of bacterially produced metabolites on amphibians in areas where populations are surviving with Bd. Currently, non-invasive, non-lethal screening techniques are being developed in collaboration with organic chemists that will allow researchers to detect and identify defensive metabolites on individuals. Bacteria that produce anti-Bd metabolites found in abundance on surviving amphibian hosts are likely to fulfil the most important effective probiotic criteria, that is, they inhibit Bd in vivo over a range of relevant environmental conditions and they persist on the host. In populations that are coexisting with Bd, common metabolites can be identified via HPLC-MS and linked to the bacteria that produce them either through known associations in the literature or through statistical methods that correlate metabolite presence with bacterial species' presence. The metabolites' defensive properties may be known or could be determined through in vitro Bd inhibition assays.
To date, most experimentation has used single-species probiotics; however, a mixture approach, where multiple bacterial species are used in synchrony, could be advantageous and should be explored (Gerritsen et al. 2011). In a field trial with wheat, some mixtures of fluorescent pseudomonads applied to seeds provided a significant increase in yield, whereas the use of individual strains was not effective (Pierson & Weller 1994). However, in a laboratory study of mussel larvae, a mixture of two probiotic strains did not improve survival over the effect provided by each strain individually (Kesarcodi-Watson et al. 2012). In one amphibian trial, a four-species probiotic mixture applied to infected R. muscosa did not persist on the host, but further research with other probiotic combinations is needed. Another approach is to transfer the microbial community from protected individuals to susceptible individuals as has been done successfully with faecal transplants in the human colon (Reid et al. 2011). A probiotic mixture could establish an anti-Bd community that works synergistically against Bd or include strains that inhibit pathogens through different modes of action. In addition, a mixture can allow treatment of multiple amphibian species and life-history stages that have different ideal probiotics in the same environmental inoculation.
Can probiotics offer a cure?
Research has concentrated on probiotics designed to prevent Bd infection. It is possible that probiotics will be able to cure or reduce established infections. Evidence suggests probiotic treatment can be effective if Bd infection is low at the time of treatment (Vredenburg et al. 2011). Additionally, highly infected individuals of R. muscosa have benefited temporarily from probiotic treatment in one study (Woodhams et al. 2012b). Treatment regimes involving conventional antifungal drugs and electrolyte treatments (Voyles et al. 2011; Woodhams et al. 2012b) followed by probiotic therapy can provide an additional way of treating established infections.
Probiotic strategies in other contexts
There are many papers in the agricultural and aquacultural literature that report improvements in growth, yield and survival from probiotic additions (Kesarcodi-Watson et al. 2008; Mohapatra et al. 2012). Protection from pathogens is one important function of probiotics (Berg 2009; Mohapatra et al. 2012). Selection of probiotics for disease mitigation typically involves in vitro inhibition trials with a pathogen and then application with baths, in feed, or by addition to the environment for clinical trials. It is encouraging that both bathing of animals and seeds and environmental treatment as well as the use of single and multi-strain probiotics have been successful in agricultural and aquacultural contexts. These results suggest that the positive effects of probiotics can be independent of treatment protocols. In addition, several probiotic strains, such as some pseudomonads, appear to have a broad host range in plants and animals (Berg 2009), suggesting that community-based probiotics can be effective for amphibians. There is likely a bias against publishing negative results, which makes it difficult to arrive at general conclusions of what protocols to avoid; however, the wealth of documented success is encouraging.
The success of probiotics in agriculture, aquaculture and with amphibians suggests that such treatment could be extended to other wildlife groups. Protection afforded by probiotics may be helpful in repatriation efforts, such as those involving hellbender salamanders (Cryptobranchus alleganiensis). Repatriated individuals are raised from eggs in the laboratory and are likely to have atypical and depauperate microbial communities and naïve immune defences. Released animals are likely to be stressed, which increases disease susceptibility. Probiotic inoculation of hellbenders prior to repatriation is likely to improve their survival rate and is under investigation.
To date, there are no bioaugmentation studies on endangered groups or on wildlife species other than amphibians. However, other wildlife such as corals and bats are decimated by disease, and probiotics may be a plausible conservation solution, as suggested by Teplitski & Ritchie (2009) for corals. Bats in North America are threatened by the pathogen, Geomyces destructans (Gd), which causes white-nose syndrome (Gargas et al. 2009). It is conceivable that bats' skin bacteria can provide protection from Gd infection and might also work synergistically with host-produced defences (T. Cheng, pers. comm.). Since bats in Europe are able to survive Gd infection, it will be advantageous to screen these populations for protective microbes. Gd becomes pathogenic during hibernation; therefore, one goal should be to find probiotics that are inhibitory at the bat's hibernation body temperature and also during breeding when disease transmission is high. Our framework can help direct probiotic research to aid bat conservation.
Manipulation of microbial defences through the use of probiotics to alter disease outcomes in humans, agricultural and aquacultural species, and amphibians is a promising disease mitigation strategy. For amphibians, effective probiotics are integrated with other aspects of host immunity and offer the most feasible approach to date for combating the devastating effects of chytridiomycosis. Bioaugmentation of individual amphibians and of amphibian habitats with carefully selected locally occurring, anti-Bd microbes can be implemented in areas under imminent threat of Bd arrival, thereby mitigating the threat of chytridiomycosis in wild amphibian populations (Vredenburg et al. 2011). In addition, individuals in survival assurance colonies can be treated and successfully repatriated. We have outlined sampling strategies and filtering protocols that will guide conservation professionals in identifying the most promising probiotics. Wildlife is under increasing threat from fungal diseases (Fisher et al. 2012); therefore, continued optimisation of protocols is urgently needed so that these disease threats can be lessened through the use of probiotics.
We thank E. Rebollar for helpful comments on the manuscript, and R. Alford for helpful discussions. This study was supported by NSF grant 1049699 to RNH and KPCM and by Swiss National Science Foundation grant 31-125099 to DCW, who was also supported by the Basler Stiftung für biologische Forschung.
MCB composed the first complete draft with substantial input from AHL, SCB and RNH. DCW provided information on amphibian immunology. KPCM and AHL provided information on bacterial metabolites. MHB provided insights on the filtering process to select a probiotic. All authors contributed substantially to the revisions.