Abstract
In recent years, emerging sequencing technologies and computational tools have driven a tidal wave of research on host-associated microbiomes, particularly the gut microbiome. These studies demonstrate numerous connections between the gut microbiome and vital host functions, primarily in humans, model organisms, and domestic animals. As the adaptive importance of the gut microbiome becomes clearer, interest in studying the gut microbiomes of wild populations has increased, in part due to the potential for discovering conservation applications. The study of wildlife gut microbiomes holds many new challenges and opportunities due to the complex genetic, spatial, and environmental structure of wild host populations, and the potential for these factors to interact with the microbiome. The emerging picture of adaptive coevolution in host–microbiome relationships highlights the importance of understanding microbiome variation in the context of host population genetics and landscape heterogeneity across a wide range of host populations. We propose a conceptual framework for understanding wildlife gut microbiomes in relation to landscape variables and host population genetics, including the potential of approaches derived from landscape genetics. We use this framework to review current research, synthesize important trends, highlight implications for conservation, and recommend future directions for research. Specifically, we focus on how spatial structure and environmental variation interact with host population genetics and microbiome variation in natural populations, and what we can learn from how these patterns of covariation differ depending on host ecological and evolutionary traits.
A generation of research on host-associated microbiomes has revealed key roles of microbes in animal physiology, ecology, and evolution. The vertebrate gut microbiome has been a subject of particularly intense research, as it is often composed of highly dense and diverse microbial communities that perform a critical set of functions, and it is easily sampled via fecal collection. Gut microbes contribute to vital host functions such as nutrient extraction (Krajmalnik-Brown et al. 2012), metabolism, and synthesis (Rowland et al. 2018), hormone and neurotransmitter regulation (Strandwitz 2018; Martin et al. 2019), and immune development (Levy et al. 2017). In controlled laboratory studies, the gut microbiome has been shown to influence multiple factors associated with host fitness, including development, longevity, and reproduction (Gould et al. 2018). Accumulating evidence from field studies suggests that gut microbiome variation between and within populations may affect host fitness in wild animals (Suzuki 2017; Lim and Bordenstein 2020). The gut microbiome may play a role in genetic divergence among host populations on a short time scale (Rudman et al. 2019), and conversely, genetic divergence among host populations may drive microbiome divergence (Garud and Pollard 2020). This emerging picture of adaptive coevolution within host–microbiome relationships highlights the importance of understanding microbiome variation in the context of host population genetic structure.
Most microbiome studies to date have focused on humans, model organisms, and domestic animals, with the hosts in each study comprising a relatively narrow range of host genetic variation. Until recently, studies of the gut microbiomes of nondomestic animals were largely descriptive, limited by the inherent challenges of collecting and interpreting microbiome data from wild animals. These challenges include collection, preservation, and transportation of field-collected samples, sample processing discrepancies that hamper the comparability of data collected across studies, and limitations to microbial taxonomic classification tools. However, as microbiome sequencing has become more widespread and accessible, and standardized sample processing and data management procedures have begun to be adopted (Gilbert et al. 2014), a growing body of research has begun to address wildlife gut microbiome variation in the context of host demographics (Liu et al. 2020; Rojas et al. 2020; Stoffel et al. 2020; Janiak et al. 2021), environmental change (Greenspan et al. 2020; Murray et al. 2020; Moustafa et al. 2021), health and disease status (Williams et al. 2016; McKenney et al. 2017; Knutie 2018; Couch, Stagaman, et al. 2021), phylogeny (Phillips et al. 2012; Gaulke et al. 2018; Youngblut et al. 2019), and other factors. In addition to clarifying ecological and evolutionary patterns obscured by narrowly focusing on humans and domestic species, broadening microbiome research to include wildlife hosts has the potential to contribute to conservation efforts, for example by informing the development of noninvasive monitoring (Bahrndorff et al. 2016; Trevelline et al. 2019; Couch, Wise, et al. 2021) and disease mitigation tools (Williams et al. 2018). However, because host-associated microbial communities are inextricably linked to landscape and host factors that vary across populations, the ultimate success of microbiome research as a conservation tool depends on understanding the microbiome in the context of host population structure. Current evidence suggests that the impact of host genomic variation on the microbiome differs across host species and environments, therefore the success of microbiome surveillance as a monitoring tool can be improved by understanding and accounting for genetic factors that influence the microbiome. For example, dissociation of microbiome structure and host genetic structure can be indicative of disturbance in oysters (Wegner et al. 2013) with potential implications for their response to climate change (Scanes et al. 2021) and disease (King et al. 2019). Genetic and spatial structure associate with the composition of the mouth microbiome in amphibians, which could have important implications for resistance to fungal pathogens (Griffiths et al. 2018). The gut microbiome can also reflect the geographic and genetic origin of individual hosts, as demonstrated in a large arctic ungulate (Bird et al. 2019). Moreover, differentiating the spatial, environmental, and genetic drivers of microbiome structure can elucidate the adaptive significance of intraspecific microbiome variation (Kurilshikov et al. 2017).
In addition to providing context for microbiome variation in natural populations, clarifying interactions between the microbiome and host genome in wild populations could enhance our understanding of host population adaptation across varied landscapes. Gut microbes have the ability to evolve more quickly than vertebrates due to rapid generation time, huge population sizes, and easy exchange of genetic information (Giraud et al. 2001). Additionally, due to the many close connections between the microbiome and host immunity, nutrition, reproduction, and other vital host functions, examining microbial divergence across host populations may contribute information that is complementary to population genetic studies. It has been suggested that the microbiome could partly explain the “missing heritability” problem of genetics, which states that genetic variants in genome-wide association studies cannot completely explain the heritability of complex traits (Sandoval-Motta et al. 2017); therefore, including microbiome data in studies of landscape and population genetics could contribute to clarifying host adaptations within and across populations.
Here, we propose a framework for understanding wildlife gut microbiomes in relation to landscape variables and host population genetics. We use this framework to review current research, synthesize important trends, highlight implications for conservation, and recommend future directions for research. Specifically, we focus on 1) how spatial structure and environmental variation interact with host population genetics and microbiome variation in natural populations and 2) what we can learn from how these patterns of covariation differ depending on host ecological and evolutionary traits.
Understanding the Gut Microbiome in the Context of Population Genetic and Spatial Structure
The gut microbiomes of natural host populations exhibit complex relationships with environmental, spatial, and host genetic factors (Figure 1). Because these variables are often highly correlated in natural populations, disentangling how they impact the microbiome is challenging, but is nonetheless key to understanding gut microbial ecology, addressing evolutionary questions, and identifying conservation and monitoring applications. This literature review and synthesis is framed around 3 questions with the goal of interpreting current knowledge and identifying gaps and future directions.
Environmental and spatial structure mediate the apparent relationships between host genetic variation and microbiome variation through a variety of mechanisms. Environmental variation can influence selection on the host genome as well as indirect transmission of microbes. Spatial structure can influence gene flow, genetic drift, and microbial transmission between hosts.
1. How Do Spatial Structure and Environmental Variation Mediate Interactions Between Host Population Genetics and Microbiome Communities?
Several mechanisms could potentially link host genetic variation with the gut microbiome. Spatial and environmental variables that influence host population genetics may indirectly influence microbiome communities via the selective forces they exert on host genomes, which in turn modulate host factors such as physiology, immunity, and behavior, and ultimately drive changes to the microbiome community. Conversely, the environment may directly impact microbiome variation, which could shape genomic adaptation of hosts, causing divergence in allele frequencies between populations.
2. What Are the Relative Effects of Host Genomic Variation, Spatial Structure, and External Environment on the Microbiome?
In addition to indirectly influencing the microbiome via selection on the host genome, spatial and environmental variables may also directly influence microbe diversity, composition, and dynamics, leading to microbiome communities that closely associate with host genotype even in the absence of causal interactions between the host genome and the gut microbiome. Understanding the relative roles of host genomic variation, spatial structure, and environmental variation is key to developing a robust framework to explain and predict natural variation in the gut microbiomes of natural host populations.
3. How Do the Answers to (1) and (2) Differ Among Host and Environment Categories?
The mechanisms by which environmental, spatial, and host genetic factors influence the microbiome are likely to vary in accordance with host lineages and ecological traits, as are the relative importance of these factors. For example, drivers and consequences of microbiome variation may differ across dietary types (e.g., carnivore, herbivore, frugivore) and between aquatic versus terrestrial hosts. Understanding how host biology and ecology determines the relative importance of microbiome covariates is key for understanding how translatable results are between study systems.
Synthesis of Current Research
Reduced sequencing costs and improved analytical pipelines are leading to rapid accumulation of data on wild animal microbiomes. Many studies are now going beyond taxonomic descriptions and are beginning to address questions that are central to understanding the ecological underpinnings of microbiome variation in natural host populations. However, the field is still early in its development, and paradigms are rapidly shifting to accommodate new information. Here we synthesize important findings from studies that are relevant to the 3 central questions described above, but we expect rapidly improving and accumulating research will continue to clarify these questions, especially as whole-genome shotgun sequencing and other “omics” technologies become more accessible and widely used.
1. How Do Spatial Structure and Environmental Variation Mediate Interactions Between Host Population Genetics and Microbiome Communities?
Spatial structure and environmental variation can mediate host population genetic divergence and isolation via dispersal limitation, local adaptation, or both (Orsini et al. 2013). Host genetic divergence is intimately linked with microbiome community variation, a concept that is synthesized within the “holobiome” paradigm, which considers the host and its microbes as a single biomolecular network, and their collective genomes as a single “hologenome” (Bordenstein and Theis 2015). Although there is ongoing debate regarding the validity and utility of the holobiome concept (Moran and Sloan 2015; Douglas and Werren 2016; Lloyd and Wade 2019), substantial evidence exists for relationships between microbial communities and host evolutionary lineages at deep evolutionary timescales (Brooks et al. 2016; Gaulke et al. 2018; Lim and Bordenstein 2020). For example, human and nonhuman primates share similar gut microbiomes relative to more distantly related vertebrate clades (Davenport et al. 2017). Evidence of phylosymbiosis is also detectable at finer evolutionary scales, for example, weak signatures of phylosymbiosis are present among all 15 species of cranes (Trevelline et al. 2020). However, the relationship between host genetic divergence and gut microbiome differentiation is less clear among conspecific populations. Within host species, host ecology (e.g., dietary differences) can obscure the effects of host genomic variation on microbiome communities (Groussin et al. 2017; Gaulke et al. 2018). However, several recent studies in natural animal populations suggest that intraspecific host genomic variation associates with gut microbiome variation at the host population level independent of ecological factors. For example, Steury et al. (2019) demonstrated that population genetic divergence among threespine sticklebacks explained significant variation in microbiome communities when controlling for environment and spatial distribution. The genetic divergence observed among these fish populations associates with geographic barriers that limit dispersal (Steury et al. 2019); therefore, it can be inferred that spatial structuring limits microbial transmission via dispersing animals. Similar patterns have been observed in widely divergent systems, including a bighorn sheep metapopulation in which host genetic lineage was shown to correlate explicitly with microbial phylogeny (Couch et al. 2020). Examples of studies that address intraspecific variation of wildlife microbiomes in the context of genetic variation (i.e., kinship, population genetic structure, adaptive genetic variation) are listed in Table 1. These studies likewise demonstrate that to disentangle the effects of distance, other barriers to host dispersal and gene flow, and environmental differences between sites, spatially explicit approaches stemming from the fields of landscape ecology and landscape genetics are needed (see Box 1 for an exploration of these concepts).
Examples of studies that address intraspecific variation of wildlife microbiomes in the context of host genetic variation and heritability
| Reference | Study species | Covariates included |
|---|---|---|
| Bird et al. (2019) | Muskoxen (Ovibos moschatus) | Population genetic cluster, seasonality, bacterial pathogenicity |
| Bolnick et al. (2014) | Threespine stickleback (Gasterosteus aculeatus) | Major histocompatibility class II genotype |
| Couch et al. (2020) | Bighorn sheep (Ovis canadensis) | Landscape connectivity, environmental variation, population genetic structure |
| DeCandia et al. (2021) | Gray wolves (Canis lupus) | Pedigree relationships, generations, and social group |
| Fleischer et al. (2020) | Galápagos mockingbirds (Mimusspp.) | Genetic similarity of neutral and major histocompatibility alleles across 9 islands |
| Grieneisen et al. (2021) | Yellow baboons (Papio cynocephalus) with some admixture from anubis baboons (Papio anubis) | Kinship, year, season, host age, social group, and diet |
| Griffiths et al. (2018) | Phofung river frogs (Amietia hymenopus) | Batrachochytrium dendrobatidis infection, spatial structure, population genetic structure |
| Montero et al. (2021) | Reddish-gray mouse lemurs (Microcebus griseorufus) | Major histocompatibility class I and II genotype, adenovirus infection status, helminth infection status |
| Ren et al. (2017) | North American red squirrels (Tamiasciurus hudsonicus) | Spatial distribution, season, diet availability, age, sex, maternal effects, familial effects |
| Rudman et al. (2019) | Fruit fly (Drosophila melanogaster) | Host genomic adaptation to an experimentally manipulated microbiome |
| Steury et al. (2019) | Threespine stickleback (Gasterosteus aculeatus) | Host population genetic divergence, environment, and geography |
| Tung et al. (2015) | Yellow baboons (Papio cynocephalus) | Social group, social network, diet, kinship, shared environment |
| Wasimuddin et al. (2017) | Namibian cheetahs (Acinonyx jubatus) | Kinship, spatial behavior, sex, age, environment |
| Wegner et al. (2013) | Pacific oysters (Crassostrea gigas) | Population genetic structure, heat stress |
| Yuan et al. (2015) | Gopher tortoises (Gopherus polyphemus) | Age, genetic diversity, spatial structure, and kinship |
| Reference | Study species | Covariates included |
|---|---|---|
| Bird et al. (2019) | Muskoxen (Ovibos moschatus) | Population genetic cluster, seasonality, bacterial pathogenicity |
| Bolnick et al. (2014) | Threespine stickleback (Gasterosteus aculeatus) | Major histocompatibility class II genotype |
| Couch et al. (2020) | Bighorn sheep (Ovis canadensis) | Landscape connectivity, environmental variation, population genetic structure |
| DeCandia et al. (2021) | Gray wolves (Canis lupus) | Pedigree relationships, generations, and social group |
| Fleischer et al. (2020) | Galápagos mockingbirds (Mimusspp.) | Genetic similarity of neutral and major histocompatibility alleles across 9 islands |
| Grieneisen et al. (2021) | Yellow baboons (Papio cynocephalus) with some admixture from anubis baboons (Papio anubis) | Kinship, year, season, host age, social group, and diet |
| Griffiths et al. (2018) | Phofung river frogs (Amietia hymenopus) | Batrachochytrium dendrobatidis infection, spatial structure, population genetic structure |
| Montero et al. (2021) | Reddish-gray mouse lemurs (Microcebus griseorufus) | Major histocompatibility class I and II genotype, adenovirus infection status, helminth infection status |
| Ren et al. (2017) | North American red squirrels (Tamiasciurus hudsonicus) | Spatial distribution, season, diet availability, age, sex, maternal effects, familial effects |
| Rudman et al. (2019) | Fruit fly (Drosophila melanogaster) | Host genomic adaptation to an experimentally manipulated microbiome |
| Steury et al. (2019) | Threespine stickleback (Gasterosteus aculeatus) | Host population genetic divergence, environment, and geography |
| Tung et al. (2015) | Yellow baboons (Papio cynocephalus) | Social group, social network, diet, kinship, shared environment |
| Wasimuddin et al. (2017) | Namibian cheetahs (Acinonyx jubatus) | Kinship, spatial behavior, sex, age, environment |
| Wegner et al. (2013) | Pacific oysters (Crassostrea gigas) | Population genetic structure, heat stress |
| Yuan et al. (2015) | Gopher tortoises (Gopherus polyphemus) | Age, genetic diversity, spatial structure, and kinship |
Examples of studies that address intraspecific variation of wildlife microbiomes in the context of host genetic variation and heritability
| Reference | Study species | Covariates included |
|---|---|---|
| Bird et al. (2019) | Muskoxen (Ovibos moschatus) | Population genetic cluster, seasonality, bacterial pathogenicity |
| Bolnick et al. (2014) | Threespine stickleback (Gasterosteus aculeatus) | Major histocompatibility class II genotype |
| Couch et al. (2020) | Bighorn sheep (Ovis canadensis) | Landscape connectivity, environmental variation, population genetic structure |
| DeCandia et al. (2021) | Gray wolves (Canis lupus) | Pedigree relationships, generations, and social group |
| Fleischer et al. (2020) | Galápagos mockingbirds (Mimusspp.) | Genetic similarity of neutral and major histocompatibility alleles across 9 islands |
| Grieneisen et al. (2021) | Yellow baboons (Papio cynocephalus) with some admixture from anubis baboons (Papio anubis) | Kinship, year, season, host age, social group, and diet |
| Griffiths et al. (2018) | Phofung river frogs (Amietia hymenopus) | Batrachochytrium dendrobatidis infection, spatial structure, population genetic structure |
| Montero et al. (2021) | Reddish-gray mouse lemurs (Microcebus griseorufus) | Major histocompatibility class I and II genotype, adenovirus infection status, helminth infection status |
| Ren et al. (2017) | North American red squirrels (Tamiasciurus hudsonicus) | Spatial distribution, season, diet availability, age, sex, maternal effects, familial effects |
| Rudman et al. (2019) | Fruit fly (Drosophila melanogaster) | Host genomic adaptation to an experimentally manipulated microbiome |
| Steury et al. (2019) | Threespine stickleback (Gasterosteus aculeatus) | Host population genetic divergence, environment, and geography |
| Tung et al. (2015) | Yellow baboons (Papio cynocephalus) | Social group, social network, diet, kinship, shared environment |
| Wasimuddin et al. (2017) | Namibian cheetahs (Acinonyx jubatus) | Kinship, spatial behavior, sex, age, environment |
| Wegner et al. (2013) | Pacific oysters (Crassostrea gigas) | Population genetic structure, heat stress |
| Yuan et al. (2015) | Gopher tortoises (Gopherus polyphemus) | Age, genetic diversity, spatial structure, and kinship |
| Reference | Study species | Covariates included |
|---|---|---|
| Bird et al. (2019) | Muskoxen (Ovibos moschatus) | Population genetic cluster, seasonality, bacterial pathogenicity |
| Bolnick et al. (2014) | Threespine stickleback (Gasterosteus aculeatus) | Major histocompatibility class II genotype |
| Couch et al. (2020) | Bighorn sheep (Ovis canadensis) | Landscape connectivity, environmental variation, population genetic structure |
| DeCandia et al. (2021) | Gray wolves (Canis lupus) | Pedigree relationships, generations, and social group |
| Fleischer et al. (2020) | Galápagos mockingbirds (Mimusspp.) | Genetic similarity of neutral and major histocompatibility alleles across 9 islands |
| Grieneisen et al. (2021) | Yellow baboons (Papio cynocephalus) with some admixture from anubis baboons (Papio anubis) | Kinship, year, season, host age, social group, and diet |
| Griffiths et al. (2018) | Phofung river frogs (Amietia hymenopus) | Batrachochytrium dendrobatidis infection, spatial structure, population genetic structure |
| Montero et al. (2021) | Reddish-gray mouse lemurs (Microcebus griseorufus) | Major histocompatibility class I and II genotype, adenovirus infection status, helminth infection status |
| Ren et al. (2017) | North American red squirrels (Tamiasciurus hudsonicus) | Spatial distribution, season, diet availability, age, sex, maternal effects, familial effects |
| Rudman et al. (2019) | Fruit fly (Drosophila melanogaster) | Host genomic adaptation to an experimentally manipulated microbiome |
| Steury et al. (2019) | Threespine stickleback (Gasterosteus aculeatus) | Host population genetic divergence, environment, and geography |
| Tung et al. (2015) | Yellow baboons (Papio cynocephalus) | Social group, social network, diet, kinship, shared environment |
| Wasimuddin et al. (2017) | Namibian cheetahs (Acinonyx jubatus) | Kinship, spatial behavior, sex, age, environment |
| Wegner et al. (2013) | Pacific oysters (Crassostrea gigas) | Population genetic structure, heat stress |
| Yuan et al. (2015) | Gopher tortoises (Gopherus polyphemus) | Age, genetic diversity, spatial structure, and kinship |
In multicellular, sexually reproducing organisms, genetic differentiation in a population arises primarily from 4 processes. The first of these processes is genetic drift, the random loss of alleles from a population over time due to death of individuals or stochastic variation in the inheritance of alleles during reproduction. As this process is stochastic, any separation of a population that prevents random mating among the members of each generation leads to genetic differentiation. Natural selection (hereafter, selection) can also drive genetic differentiation if operating differently between subpopulations. The rise of new mutations can also contribute to genetic divergence, although this is often treated as a negligible process in multicellular organisms on ecological time scales. Gene flow (exchange of genetic material through dispersal, typically referred to as migration in population genetics) between subpopulations can move new variation into any given subpopulations, and overall, reduces differentiation from genetic drift or other processes. Thus, genetic structure (i.e., the degree of differentiation between individuals or populations) can signal the degree of interaction between populations or individuals at different locations, mediated by the strength of genetic drift and in some cases selection. The rate of genetic divergence due to genetic drift between 2 subpopulations is highest when the number of individuals is small and generation times are short. In very large populations or when generation times are long, even when a complete barrier to movement arises, genetic differentiation may not be detectable for many generations (Epps and Keyghobadi 2015).
In microbes, however, genetic differentiation is driven by occasional mutations during asexual reproduction, as well as occasional horizontal gene transfer by other processes, such as conjugation (Dudaniec and Tesson 2016). Thus, in host-associated microbial populations, separation between hosts that restricts microbial transfer can result in differentiation of microbial communities that might otherwise remain similar due to frequent sharing of taxa or strains. This can occur through direct host–host transmission or indirect environmental transmission. Microbes exhibit extremely large population sizes and rapid generation times, such that mutations accumulate rapidly (Pepper 2014). Changes in host physiology could result in differential selection and microbial evolution within a single host lifetime (Scanlan 2019).
The term landscape genetics is used to denote detailed, spatially explicit investigations of genetic structure (Manel et al. 2003; Storfer et al. 2007). Landscape genetic studies draw on both population genetics and landscape ecology and can be conducted on individuals or populations. Broadly, 2 approaches are typically employed: one seeks to identify discontinuities in frequencies of alleles (genetic variants) on a landscape (Guillot et al. 2009), whereas the other evaluates competing models of how genetic differentiation accumulates on landscapes (Cushman et al. 2006; Epps et al. 2007). Although the former could be analogous to mapping presence of particular strains or shifts in host-associated microbial communities in hosts across a landscape, the latter approach may offer the clearest applications to microbiome research. Competing models of genetic differentiation on a landscape include:
Isolation by distance (Wright 1943): where genetic differentiation increases with geographic distance. This typically is used as a null hypothesis in landscape genetic studies.
Isolation by resistance (McRae 2006): genetic differentiation increases according to the relative cost of moving across habitats that are less favorable for movement or dispersal by the study organism.
Isolation by barriers (Blair et al. 2012): genetic differentiation increases sharply when locations are separated by a dispersal barrier.
Isolation by environment (Wang and Bradburd 2014): genetic differentiation increases according to the degree of environmental differentiation between locations, such as temperature or elevation.
Isolation by adaptation (Orsini et al. 2013): genetic differentiation occurs because of adaptive differences between 2 populations, which may or may not be associated with environmental variation.
In a typical landscape genetics study based on the landscape resistance model concept, different habitat types (e.g., forests, grasslands) or landscape features (e.g., roads, rivers) within a study area are assigned different resistance values, where resistance value represents the hypothesized difficulty of movement for the study species across a habitat type or feature. Figure 2 depicts a simple example, where each host sampled allows estimation of genetic differences among individuals (as well as microbiome, as we describe subsequently); alternatively, individuals in each habitat patch could be grouped into populations for analysis at that scale. After resistance values are assigned to all habitats, creating a resistance surface (Spear et al. 2010), the cumulative cost of moving between each pair of individuals or populations can be estimated using procedures such as least-cost path (Adriaensen et al. 2003) or circuit theory(McRae 2006; McRae and Beier 2007) that evaluate all possible movement routes between locations. Least-cost path analyses determine the cost (distance times resistance value in each map cell traversed) of the most efficient route, whereas circuit theoretic approaches estimate the cumulative resistance of all possible paths between 2 locations. Either approach generates a matrix of resistance distances between sample locations that can be compared to the matrix of genetic differences between those locations (e.g., Figure 2) to determine whether any given resistance map better predicts genetic distance than a null model (e.g., a random model or isolation by distance). In a process sometimes described as optimization, different parameterizations of resistance surfaces can likewise be tested (e.g., in Figure 2, grassland could be weighted at 5, 10, or 20 times the cost of forested habitat, and the cumulative costs or resistances of each scheme tested against genetic differences), and genetic differences characterized at the individual or population level (Cushman et al. 2006; Epps et al. 2007). Appropriate genetic metrics and statistical techniques have been debated, but a variety of options exist (Shirk et al. 2017; Shirk et al. 2018).
A hypothetical model combining landscape genetics and microbiome. Here, individuals of a forest-loving, terrestrial species are sampled for host genetics and gut microbiota. The landscape is described with a resistance model, reflecting the hypothesized relative cost of movement by the host. In this example, cells in grassland are assigned a resistance value 10 times higher than cells in forest, and rivers are assigned a cost of 100 to cross. The least-cost paths between sample locations are depicted, with examples of cumulative resistance costs arrayed above the diagonal (italics) and pairwise comparisons of individuals below the diagonal (bold). Those pairwise comparisons could be individual host genetic distances or relatedness, as when building and optimizing a landscape genetics model (i.e., testing what resistance values best predict genetic differences), or measures of difference in microbiome between individual hosts, as when testing whether beta diversity between host individuals reflects host movement as inferred from an optimized landscape genetic model. In this case, host samples (HS) 1 and 2 are closer geographically than HS 2 and HS 3, but due to the wide expanse of grassland and a river between HS 1 and HS 2, this model predicts that resistance distance and thus genetic distance are highest between these locations. Gut microbiome differentiation might likewise be predicted to be highest between those locations if host contacts are a primary driver of microbiome communities. A circuit theoretic analysis that integrates all possible paths and thus multiple dispersal pathways, rather than least-cost paths, could also be applied here.
When selection acts differently across subpopulations, it typically leads to higher levels of genetic structure than would be expected solely from genetic drift, but that higher genetic differentiation is usually only detectable when investigating markers in or near the gene under strong selection. Conversely, balancing selection, such as selection for maintenance of diverse alleles in genes associated with immune function (Aguilar et al. 2004), may result in lower differentiation than expected; other models of selection can have more complicated effects (Charlesworth et al. 1997). Selection can be evaluated using outlier tests (Foll and Gaggiotti 2008), in which variable sites are screened to determine whether allele frequencies at any given site diverge more than expected given variation at other sites, or other approaches such as environmental association analyses (Jones et al. 2013; Ahrens et al. 2018). In turn, differences in allele frequencies apparently resulting from selection can be tested for correlation with landscape features or environmental differences across the sampling sites (Schoville et al. 2012). Approaches to detect selection on landscapes are often referred to as landscape genomics (Manel and Holderegger 2013). As metagenomic datasets become more available in natural host populations, facilitating characterization of the collective genome of all the microorganisms in a given host, landscape genomics approaches could be used to explore the effects of selection on the microbiome across host individuals or host populations.
Landscape genetics and genomics offer a rich tool set to investigate relationships with microbiomes. For instance:
Landscape genetic approaches such as resistance models could be used to test how landscape influences host genetic structure across space and host contacts between populations and individuals (e.g., Figure 2), which in turn would facilitate the study of how these factors affect microbiome differences (Couch et al. 2020); such approaches have been applied to particular infectious diseases (Biek and Real 2010; Côté et al. 2012; Lee et al. 2012).
Landscape genomic approaches could be used to link evidence of selection in, for example, markers linked to host immune function to variation in host gut microbiomes at the landscape level. For instance, if links between gut microbiota and host immune function were posited, genetic differentiation at host immune-linked genes might be predicted to be more closely associated with variation in beta diversity of the gut microbiome than genetic differentiation at neutral genetic markers.
Statistical approaches from landscape studies for comparing multivariate datasets based on differences between individuals, populations, or sites could likewise be employed when evaluating how differences in microbiomes correlate with a variety of potential predictive factors such as distance, landscape resistance, or environmental conditions. For instance, DeCandia et al. (2021) used such an approach to assess associations between relatedness and microbiome composition while controlling for social structure. We suggest 3 basic ways to develop response variables for describing differences among microbes or microbial communities that could, in turn, be modeled with spatially explicit hypotheses:
Statistics to describe the similarity of gut biota communities among individual hosts or host populations. Such an approach necessitates defining individual biota, such as operational taxonomic units (OTUs) or amplicon sequent variants (ASVs) (Callahan et al. 2016). OTUs rely on a somewhat arbitrary level of differentiation (e.g., 97% sequence similarity), whereas ASVs can be resolved exactly, down to the level of single-nucleotide differences over the sequenced gene region (Callahan et al. 2017). Using OTUs may therefore obscure some fine-scale differentiation of interest, whereas ASVs may obscure deeper level differentiation amidst large numbers of variants with relatively few differences. Thus, a statistic describing beta diversity of ASVs between host individuals or populations, based on presence/absence (e.g., Jacard distance) or abundance (e.g., Bray–Curtis) of ASVs, could be used as a response variable in a landscape-genetic type analysis. A phylogenetically informed estimate of beta diversity, such as unifrac distance or weighted unifrac distance (Lozupone et al. 2011), could offer a useful alternative that integrates all known variants but incorporates microbial phylogenetic information.
In a hypothetical example, as depicted in Figure 2, Bray–Curtis distances could be estimated from gut microbe ASVs sampled from individual hosts across a landscape. In turn, to determine how geographical distance, host dispersal, or environmental differences shaped those gut microbiota, those Bray–Curtis distances could be modeled as a function of 1) resistance distance for hosts that incorporate influences of landscape variation on host dispersal (derived in turn from landscape genetic analyses, satellite telemetry, or other data on movement), 2) geographical distance among hosts, 3) barriers to host movement, 4) differences in precipitation or other environmental variables between sampling sites, or 5) any other pairwise measure of interest, such as a binary matrix indicating whether both sampling sites fell within the same vegetation type (1) or in different types (0) if researchers wished to test hypotheses about the effects of vegetation type on herbivore gut microbiome.
2. Evaluation of functional diversity. Similar to amplicon-based approaches describing similarity of gut biota communities (see #1), technologies such as whole-genome metagenomics, transcriptomics, and metabolomics can be used to evaluating the presence and abundance of gene products associated with particular functions, such as metabolic pathways or antibiotic resistance (Miller et al. 2019). This approach, particularly in combination with analyses of compositional beta diversity, could help shed light on whether host interactions or environmental similarity more strongly shape gut biota. For instance, one might predict environmental similarity better predicts functional similarity in gut microbiota than spatial distance or host contacts, even when abundance-weighted or phylogenetically weighted measures of microbiome compositional differentiation showed that distance or contacts were influential.
3. Phylogenetic differentiation of particular strains of a given taxon. Here, rather than focusing on many taxa, strain diversity or phylogenetic differentiation of particular taxa could be evaluated, for instance, focusing on a microbe that is particularly representative of the study question of interest (e.g., disease transmission), or particularly easy to culture or otherwise characterize. This approach has been employed with E. coli (VanderWaal et al. 2014), evaluating whether strains were shared among individuals with social associations or spatial overlap, or more commonly with viruses. For instance, Lee et al. (2012) developed landscape genetic models of bobcats and paired that analysis with phylogenetic analyses of feline immunodeficiency virus to make inferences about virus transmission.
In addition to mechanisms that isolate populations through dispersal limitation, there is some evidence that isolation by adaptation can also drive microbiome variation across intraspecific populations. In laboratory systems, immune-associated genes such as those involved in the major histocompatibility complex (MHC) have been shown to regulate microbiome composition (Toivanen et al. 2001). Associations between specific adaptive (e.g., immune-associated) alleles and microbiome variation are only beginning to be explored in wildlife, and a clear pattern has yet to emerge. In Galapagos mockingbirds, genetic dissimilarity based on neutral and MHC regions exhibited little relationship with gut microbiome distributions (Fleischer et al. 2020). Given the challenges of accounting for the numerous internal and external influences on the gut microbiomes of natural animal populations; however, it is perhaps unsurprising that many studies have failed to identify clear relationships between immune loci and microbiota. However, a few studies in wild populations have identified correlation with MHC genes and microbiome variation (Bolnick et al. 2014; Pearce et al. 2017; Montero et al. 2021). Notably, in a recent study of the gut microbiomes of wild mouse lemurs, MHC I and MHC II variability explained shifts in microbiome composition, especially in a few select microbial taxa. These taxa were in turn linked to adenovirus and helminth infection status, suggesting a coupled role of MHC variability and microbial flora as contributing factors of parasite infection (Montero et al. 2021). In addition to linking adaptive host loci with microbiome variation, this study highlights the potential role of microbiome-driven selection as a contributor to evolutionary dynamics. The skin microbiome of a seabird have also been shown to correlate with MHC genotype, indicating that host adaptation may interact with microbial communities at multiple body sites (Bolnick et al. 2014).
If spatial structure or environmental variation alters the native gut microbiota, the microbiome could mediate selective pressure on host phenotype. A recent study on Drophila melanogaster used experimental and field data to demonstrate that microbiomes may be an agent of selection that shapes the patterns and processes of ecological and evolutionary adaptation, potentially driving rapid host evolution (Rudman et al. 2019), and similar patterns could exist in vertebrate hosts.
2. What Are the Relative Effects of Host Genomic Variation, Spatial Structure, and External Environment on the Microbiome?
In addition to mediating indirect effects on the microbiome via host genomic variation, spatial and environmental factors directly impact the gut microbiome communities of wild animal populations. Understanding the direct impacts of environment and spatial structure is key to contextualizing host genome-mediated gut microbiome variation.
Environmental variation could impact the gut microbiome either by modifying host dietary intake, and hence the selective pressures exerted on microbial communities by the internal gut environment, or by altering the composition of environmental microbes available to colonize the host. Several recent studies suggest that the importance of the internal environment (i.e., host diet) supersedes the composition of environmental microbes in modulating gut microbiome dynamics in wild species. A recent study of African buffalo in a South African national park demonstrated that the dramatic changes observed in the gut microbiome during the wet season could also be induced by dietary supplementation, even when dry season conditions persisted (Couch, Stagaman, et al. 2021). Similarly, an aviary experiment showed that dietary manipulation could alter the gut microbiome to reflect the differences observed in urban versus rural house sparrows (Teyssier et al. 2020). Interestingly, the strong effect of diet may even be evident across widely divergent host taxa as well as within species. A recent review of teleost fishes found that environmental conditions (i.e., marine versus freshwater) were a strong predictor of microbiome structure, but that herbivorous fish harbored gut microbiota that were more similar to terrestrial vertebrates than to other fishes, indicating that the internal environment played a strong selective role (Sullam et al. 2012).
Spatial structure can also influence the gut microbiota independently of host genomic variation. Spatial overlap allows exposure to the same microbial sources (e.g., shared water sources), and increased opportunities for horizontal transmission between hosts. Studies in primates demonstrate that horizontal transmission is an important driver of microbiome convergence between hosts. In wild baboons, gut microbial composition correlated with host social structure when controlling for genetic relatedness and shared environment (Tung et al. 2015). Spatial or social transmission of microbes also appears in social nonprimate species, such as house mice (Goertz et al. 2019) and feral horses (Stothart et al. 2021); spatial distribution impacted some metrics of gut microbiome beta diversity independent of environmental heterogeneity, but these studies did not control for direct social contacts. Spatial proximity associates with shared microbiota even in species with limited social contact, presumably due to environmental transmission or access to shared microbial sources. Solitary red squirrels exhibited spatially structured microbiome variation, while genetic relatedness had little effect (Ren et al. 2017). In gopher tortoises, another species with limited social interactions, spatial distribution associated with microbiome composition independent of the effects of genetic structure (Yuan et al. 2015).
In general, diet and environmental variation and spatial/social structure are considered to exert greater influence on gut microbiome variation than within-species host genomic variation (Dong and Gupta 2019). Yet, this is not always the case. In free-ranging cheetahs, kinship outweighs the effects of individual movement behaviors (in this case, defending a small home range versus ranging widely) and is the predominant predictor of gut microbiome variation, suggesting that genetically determined host factors outweigh spatial factors in the study population (Wasimuddin et al. 2017). In that study, however, highly related cheetahs were in the same social groups, and the authors did not attempt to disentangle genetic relatedness, social contacts, and spatial distance between individuals. In threespine sticklebacks, population genetic structure also outweighs the influence of geography and environmental variation (Steury et al. 2019). One potential explanation of these deviations is that free-living animals experience spatial scales, environmental gradients, and genetic constraints that vary enormously between species and populations. Wildlife microbiome studies, often constrained by cost and logistics, may unintentionally obscure the relative importance of alternative microbiome drivers simply based on the scales across which each variable is measured. Additionally, host and environmental factors are not the only contributors to microbiome variation. Many studies assessing the influence of spatial structure fail to consider interspecific interactions between microbes (commensal and pathogenic), which can comprise important mechanisms structuring gut microbial communities (Fountain-Jones et al. 2020).
In addition to considering the effects of spatial and genetic structure of host populations on microbiome communities, it may be informative to consider the genetic structure of individual microbial taxa. Understanding the genetic variation of individual microbial taxa and strains within and among hosts can further contribute to our understanding of the selective pressures, transmissibility, rates of evolution, and other properties of microbial populations. Garud and Pollard (2020) reviewed the genetics of host-associated microbial populations and identified outstanding research questions, including several that are especially pertinent to wild populations: 1) What factors contribute to genetic structure within microbiome species across hosts and geographic regions? 2) What is the tempo and mode of adaptation within and across hosts at different timescales? 3) How common are various mechanisms of recombination? 4) Given the rapid evolution within host-associated microbiomes, does evolution influence ecological processes and vice versa? 5) Which host and microbial traits are associated with genetic variation in microbes? Related questions have been addressed in several wildlife studies, providing useful insights for epidemiology. For example, a study in wild giraffe used different genetic variants of commensal Escherichia coli to evaluate the spatial and social drivers of microbial transmission between hosts (VanderWaal et al. 2014). They compared spatial overlap and social contacts with the distribution of genetic subtypes of E. coli and found that transmission was more likely to occur between individuals that were strongly linked in the social network. A second study in wild giraffe explored the distribution of antibiotic resistance genes in fecal E. coli within a host social network and discovered that antibiotic resistance was associated with host traits (age), but not with host connectedness (Miller et al. 2019). This study also highlighted the potential of horizontal gene transfer between microbial strains, which complicates the study of genetic structure in microbial populations. These studies illustrate the utility as well as the complexity of considering microbial genetic structure in wildlife studies, but further research is needed to fully leverage the potential of this approach for epidemiology and other applications.
3. How Do the Answers to (1) and (2) Differ Among Host and Environment Categories?
Host Diet
The symbiotic roles of gut microbes vary markedly among carnivores, omnivores, and herbivores. As such, we can expect that the contributions of environmental, spatial, and heritable host genetic effects on the microbiome differ based on dietary type. Indeed, trophic level was an important predictor of microbiome community in teleost fishes, and diet appeared to have a stronger influence on the microbiomes of herbivorous fishes than in carnivorous or omnivorous species (Sullam et al. 2012). In carnivores, gut microbial alpha diversity is low and functionally adapted toward degrading protein as an energy source, while herbivore gut microbiome diversity is relatively high and contains functional modules specialized in amino acid synthesis (Muegge et al. 2011). Omnivorous hosts tend toward gut microbiomes that are intermediate between these 2 phenotypes (e.g., Sullam et al. 2012). Given the important role of microbes for extracting and synthesizing nutrients from plants, it is unsurprising that spatial and seasonal dietary variation drives pronounced microbiome changes in herbivores (Couch, Stagaman, et al. 2021) and omnivores (Hicks et al. 2018). Similarly, microbiome variation between dietary groups (Bragg et al. 2020) and across spatial dietary gradients (Colborn et al. 2020) associates with microbiome variation in carnivores, but there is a little evidence of seasonal adaptation of the carnivore gut microbiome to dietary change except in cases of hibernation (Sommer et al. 2016) or seasonal fasting (Tang et al. 2019). Because carnivores tend to undergo less seasonal dietary change than herbivores (Humphries et al. 2017), we might expect that the effects of individual traits, such as host genetic background, would be more apparent in the gut microbiomes of carnivores than in herbivores. Indeed, current research suggests that intraspecific variation in the gut microbiomes of wild carnivores may be strongly influenced by relatedness, and thus, perhaps, heritability. In wild Namibian cheetahs, there is profound taxonomic and functional similarity in the gut microbiomes of kin versus nonkin (Wasimuddin et al. 2017). In North American gray wolves, genetic relatedness is one of the primary determinants of gut and skin microbiome community variation, even after controlling for social group membership (DeCandia et al. 2021). Interestingly, however, due to the paucity of large, longitudinal studies of wild carnivore gut microbiomes, this conclusion remains speculative, particularly when relatedness and social contacts are highly correlated (e.g., Wasimuddin et al. 2017). In contrast, many studies in wild herbivores and omnivores (including humans (Kurilshikov et al. 2017) show that diet is a more prominent driver of gut microbiome community variation than heritability (Ren et al. 2017). Due to the excessive influence of diet on the herbivore gut microbiome, large longitudinal datasets may be needed to detect heritable components of the microbiome in herbivore populations (Grieneisen et al. 2021).
Environment Types
The nature of the environmental matrix determines the ease with which microbes are transmitted between hosts, as well as determining the pool of potential environmental colonizers. In environments that are favorable to host–host transmission and are rich in potential colonizers, we would expect amplified effects of host spatial environmental structure on the gut microbiome relative to intrinsic, genetically determined host traits. Aquatic animal habitats are typically rich in environmental microbes and facilitate easy host–host transmission of waterborne microbes; therefore, we would expect environmental and spatial factors to play an outsized role in shaping the gut microbiome. In oysters, the gut microbiome corresponds to population genetic structure, but this correlation is easily erased by environmental perturbation (Wegner et al. 2013). A recent meta-analysis of teleost fishes in comparison with other taxa demonstrated that environmental conditions (salinity) outweigh host phylogeny in determining gut microbiome composition, both within teleosts and across other non-fish taxa (Sullam et al. 2012). This contrasts with studies in terrestrial vertebrates in which phylogeny and evolutionary history of hosts and microbiomes generally correlate closely (Nishida and Ochman 2018; Weinstein et al. 2021), though birds and bats show convergence of the microbiome with little association to host diet or phylogeny (Song et al. 2020). Research on the impacts of host–host interactions on wild aquatic animal gut microbiomes is limited, but experimental studies in laboratory fish suggest a strong role for social transmission of microbes that may overwhelm the influence of population genetic structure. In zebrafish, interhost dispersal of cohoused fish was strong enough to overwhelm the effects of host factors, largely eliminating the differences between wild-type and immune-deficient fish (Burns et al. 2017). Although host–host transmission of commensal microbiomes has been demonstrated in terrestrial vertebrates (e.g., Tung et al. 2015), there is no evidence that host–host transmission can overwhelm the effects of host diet, environmental variation, or phylogeny on the gut microbiome. Taken together, these studies suggest differing effects of environmental, spatial, and genetic structure on the gut microbiomes of aquatic versus terrestrial species, highlighting the need to consider host-associated microbiomes in the context of the environmental matrix.
Numerous studies in terrestrial vertebrates have demonstrated differences in the gut microbiomes of wild versus captive conspecific hosts. For example, a study examining the effects of captivity on 41 mammalian species (McKenzie et al. 2017) found that captivity significantly altered the gut microbiomes of 12 of 15 mammalian genera. However, the relative roles of colonization versus selection in mediating the differences between captive and wild hosts is unclear. It is likely that captive diets exert strong selective pressure on the gut microbiome (Martínez-Mota et al. 2020), but more research is needed to understand the contribution of colonization by diet- and environment-derived microbes to the gut microbiomes of captive wildlife. Additionally, because captive breeding programs may result in genetic divergence between captive and wild populations, as well as among captive populations, population genetic structure warrants consideration when comparing captive and wild populations. The greater environmental control that is possible in captive wildlife may allow researchers to assess the effects of population genetic structure more robustly than in wild populations.
Potential Applications for Conservation
Establishing a framework for understanding wildlife gut microbiomes in the context of environmental variation, spatial structure, and host population genetic structure will lay the groundwork for emerging conservation applications. Microbiome surveillance has great promise as a noninvasive monitoring tool for tracking host demographics, population structure, and spatial/social structure. In free-ranging elk, a microbiome-based machine learning classifier based on bacterial taxa successfully predicted host population, sex, age, and body condition with promising cross validation results (Pannoni et al. 2022). Community disruption (dysbiosis) in wildlife gut microbiomes has been linked with anthropogenic disturbance and disease, suggesting a potential roll for noninvasive monitoring of population health. For example, Asian elephants experienced gut dysbiosis due to translocation, captivity, and deworming (Moustafa et al. 2021).
Landscape genetic approaches (Box 1) likewise offer great potential for linking genetics of hosts and microbes and improving inference about host population dynamics in the context of microbiome research, which in turn would facilitate addressing many conservation-related questions. Landscape genetics, in turn, draws on landscape ecology, which offers many tools that could facilitate examining how configuration of human-altered landscapes interacts with microbiomes, host genomes, and their interactions. For instance, sophisticated approaches for measuring habitat patch configuration, such as the ratio of edge habitat or the degree of habitat fragmentation on any landscape (McGarigal and Marks 1995), could be employed as predictors for aspects of gut microbiomes such as species richness, functional diversity, or microbiome community similarity among individuals. Such measures also allow investigating aspects of the host genome such as genetic diversity that are measured at specific locations in space, rather than comparisons between locations in space as in resistance models (Box 1). Indeed, because of dietary, energetic, and other changes associated with habitat changes, microbiomes could serve as an indicator of the actual impacts of different types of habitat fragmentation or degradation. In a compelling example of this approach, black howler monkey microbiome structure was found to reflect habitat fragmentation and degradation, including functional changes in the microbiome in more disturbed habitats (Amato et al. 2013).
Beyond surveillance, a tantalizing question arises: can the microbiome be leveraged to proactively enhance health? Disease has been shown to disrupt the microbiome at multiple body sites and across numerous wild populations. Examples include the disruption of the skin microbiome in amphibians infected by the pathogen Batrachochytrium dendrobatis (Jani and Briggs 2014; Jani et al. 2021), as well as the decreased gut microbial diversity in Coquerel’s sifikas (Propithecus coquereli) infected with the protozoan Cryptosporidium (McKenney et al. 2017). Conversely, certain microbial phenotypes can enhance resistance to infectious diseases, for example, certain skin microbial phenotypes associate with resistance to B. dendrobatis in amphibians (Harris et al. 2009; Jani et al. 2017) and white-nose syndrome in bats (Lemieux-Labonté et al. 2017). Probiotic treatment to prevent infection and mortality has shown promise in captive wildlife, including farmed salmon (Klakegg et al. 2020). However, the utility of microbiome-based probiotic development requires that they function in the context of broad genomic variation among hosts and pathogens (Harrison, Sewell, et al. 2019), as well as within the natural range of spatial, social, and environmental contexts. To this end, metagenomic analysis can be used to examine the complete genetic makeup of a large number of strains of potentially beneficial bacteria and elucidate the traits that contribute to probiotic function (Ventura et al. 2012). This information could then be used to identify the necessary set of functional traits to solve a specific problem. Using population genetic theory to characterize the taxa that express these traits could then contribute to the selection of appropriate strains for probiotics. For example, landscape genetic models of gut microbe dispersal could predict the likelihood of probiotic transmission between hosts (see Box 1).
Recommendations for Future Research
Studying wildlife microbiomes in the context of natural environmental variation and population genetic, spatial, and social structure presents numerous challenges and opportunities. Much more descriptive and exploratory work is needed to understand host-microbiome ecology in wild species, particularly at additional body sites beyond the gut (e.g., skin, mucosa, reproductive tract). However, even as this exploratory work is ongoing, existing and emerging tools from related fields can help accelerate the expansion of wildlife microbiome research from descriptive and exploratory to hypothesis-driven and highly informative.
One of the most prominent challenges of wildlife microbiome research is that environmental variation, population genetic structure, spatial structure, and social structure are often highly correlated in natural populations. One way this can be addressed is by designing studies with multiple population replicates along the gradient of interest. Hosts that exist as metapopulations can provide opportunities to evaluate the effects of population-level variables on the gut microbiome. For example, studying gut microbiome variation across multiple populations in a metapopulation of bighorn sheep demonstrated that geographic distance and gene flow explained different aspects of microbiome beta diversity (Couch et al. 2020). Non-mammalian species that are found in metapopulations, including butterflies (Hanski et al. 2017) and coral reef fishes (Bay et al. 2008), would provide interesting opportunities for evaluating how spatial, genetic, and environmental variables interact to shape the microbiome under vastly different environmental and host conditions. Additionally, studying the population genetics of the taxa that comprise the microbiomes of wild hosts (as described in the previous section and in Box 1) could contribute to differentiating host–host transmission dynamics from host-microbe coevolution, particularly if both neutral markers and markers associated with genes of interest were employed: neutral markers should largely reflect host interactions and population history, whereas markers linked to genes under selection may diverge at different rates.
Longitudinal data from individuals can also contribute to clarifying the relative inputs of environmental, spatial, genetic, and social determinants of microbiome variation, as these variables can change with season and host life stage. In wild baboons, a large, longitudinal study that followed 585 wild baboons over 14 years demonstrated that heritability of the gut microbiome was nearly universal, but contingent on diet, age, and socioecological variation (Grieneisen et al. 2021). This study demonstrated the power of longitudinal profiles and large sample sizes for quantifying microbiome drivers in natural host populations. Though collecting longitudinal data from wild animals can be challenging, state-of-the-art animal tracking technologies are quickly reducing this barrier (Katzner and Arlettaz 2020). As tracking technologies continue their current trajectory to becoming mainstream in research and applied conservation, low-cost opportunities for collecting fecal samples from tracked individuals over time should be leveraged for longitudinal microbiome research.
Innovative experimental and analytical approaches can complement information gained from field studies. Experimental manipulation of wild species, both in field-based experiments and laboratory studies, can be powerful tools for disentangling the drivers of microbiome variation. Experimental approaches have been widely used to evaluate the relationships between skin microbiome variation and pathogen resistance in amphibians (Harris et al. 2009; Kueneman et al. 2016; Harrison, Price, et al. 2019), and similar approaches could be leveraged for experimental study of the gut microbiome in wild animals. Such experiments need not be limited to species that are tractable in laboratory settings, as there are many opportunities to study large, highly mobile, or otherwise intractable species that are subject to manipulation. For example, translocation (Chong et al. 2019; Moustafa et al. 2021), supplemental feeding (Couch, Wise, et al. 2021), and therapeutics (Moustafa et al. 2021) can all provide opportunities to isolate host and landscape variables associated with gut microbiome variation. Field studies can also inform and benefit from simulation modeling to test alternative ecological and evolutionary frameworks for microbiome assembly and dynamics. Simulation modeling can help explain the evolutionary (Zeng et al. 2015) and ecological (Zeng and Rodrigo 2018) patterns of microbiome communities observed in natural systems. For example, computational agent-based neutral frameworks have been used to explore microbial dynamics within host metapopulations (Zeng and Rodrigo 2018) and over evolutionary time (Zeng et al. 2015). Additionally, applying established analytical tools from landscape genetics can contribute to understanding microbiome variation in the context of highly correlated host and landscape variables (Box 1).
As we continue to clarify the drivers of microbiome variation in wild populations, understanding the functional significance of this variation will become increasingly relevant. Currently, many studies focus on microbial taxonomic composition and diversity. However, functional redundancy is a feature of many microbial communities (Barnes et al. 2020); therefore, taxonomic variation among animal microbiomes does not necessarily correlate with functional variation, and this carries important implications for interpreting results (Barnes et al. 2020). By expanding microbiome studies to include metagenomics, transcriptomics, proteomics, and metabolomics in addition to amplicon-based DNA sequencing, we can begin to fully leverage the information contained in gut microbial communities for wildlife research and conservation.
Conclusions
The burgeoning study of wildlife microbiome research has much to gain from integrating host genetic and microbiome datasets across a wide range of host species, body sites, and ecosystems. The conceptual framework we propose in this work has the potential to contribute to our understanding of the microbiome in the context of the complex spatial and host genetic structure that characterize natural wildlife populations. The framework we propose is structured around the following questions: 1) How do spatial structure and environmental variation mediate interactions between host population genetics and microbiome communities? 2) What are the relative effects of host genomic variation, spatial structure, and external environment on the microbiome? 3) How do the answers to (1) and (2) differ among host and environment categories? Examining current research through the lens of these questions, we find that spatial, environmental, and host genetic structure can significantly impact gut microbiome structure, with potential consequences for wildlife health and conservation. Moreover, the relative influences of spatial, environmental, and host genomic variation, as well as the mechanisms by which they impact the microbiome, appear to differ among host species and environments. However, the paucity of wildlife microbiome studies that integrate host genetic data alongside spatial and environmental covariates currently limits our understanding of the underlying processes that shape host-associated microbiomes across complex space. Consequently, we lack theoretical bases for understanding and predicting how the microbiomes of wild populations may change in response to ecological, evolutionary, and anthropogenic change. To address these fundamental gaps, we recommend that researchers studying the microbiomes of wildlife populations consider incorporating information on host population genetic structure, in addition to spatial and environmental covariates. We identify several key approaches for untangling the complex interactions between highly correlated variables, including by studying hosts that exist as metapopulations, conducting longitudinal microbiome sampling of individuals, taking advantage of opportunities for experimental manipulation, and adopting innovative analytical approaches from related fields.
Funding
Funding from the American Genetic Association Evolutionary, Ecological, and Conservation Genomics Award supported Dr. Couch’s previous work that was instrumental in the conceptualization of this work.
Acknowledgments
We gratefully acknowledge Dr. Keaton Stagaman for reviewing this paper to ensure its accuracy and completeness.
