Partnerships between host and microbes within an environmental setting exemplify a network of biotic relationships that are common across the tree of life (Rosenberg and Zilber-Rosenberg, 2013; Bosch and Miller, 2016; Hurst, 2016). The phenomena underlying these relationships complement more than a century of biological research that has focused on the evolution and ecology of individual species. To conceptualize the functional importance between host and microbiota, Zilber-Rosenberg and Rosenberg (2008) proposed the hologenome theory of evolution, stating that animals and plants, along with their microbiome serve as a unit of selection. This paradigm shift has led to advancements in our understanding of the spectrum of organismal symbioses in the life sciences (Bordenstein and Theis, 2015; Theis et al., 2016), with particular attention to developmental (McFall-Ngai and Ruby, 2000; McFall-Ngai, 2002), evolutionary (Brucker and Bordenstein, 2012, 2013a,b), and genetic modifications (Husnik et al., 2013) to the host.

The hologenome theory emphasizes the role microbes play in animal and plant evolution as integrated units of biological organization that intertwine Darwinian and Lamarckian principles (Zilber-Rosenberg and Rosenberg, 2008; Rosenberg et al., 2009; Bordenstein and Theis, 2015). The hologenome theory provides functional explanations for the role of the microbiome in a Darwinian framework as it relates to speciation (Brucker and Bordenstein, 2012, 2013b) and potentially host fitness (e.g., Callens et al., 2016). A Lamarckian framework (Rosenberg et al., 2009), on the other hand, is complementary to this, as it details the mechanisms whereby microbes are acquired or lost during an organism's lifetime, and that the acquisition of a novel species or strain of microorganism may be integrated into the hologenome. Therefore, hologenomes (and as a direct extension, the holobiont) integrate principles from multiple disciplines (Rosenberg et al., 2009) spanning the diverse fields of evolutionary genetics (Brucker and Bordenstein, 2012, 2013b) and evolutionary ecology (Macke et al., 2016; Theis et al., 2016).

One major challenge the field currently faces is melding insights from the evolutionary, genetic, and molecular underpinnings of host-microbe partnerships with the ecological conditions in which they formed and evolved. Recent work has begun addressing these disciplines as an integrative discipline (Gilbert et al., 2015; Theis et al., 2016); however, they remain largely as separate conceptual entities. Since the hologenome concept was proposed nearly a decade ago, other multi-disciplinary fields, such as evolutionary developmental biology (Moczek et al., 2015), have recognized and, in a conceptual as well as empirical manner, addressed this challenge in two successive steps. The first of these emphasizes the value of applying hypotheses and/or mechanisms derived from laboratory studies to complementary experiments in nature (Gilbert, 2016), while the second tests evolutionary principles across different environments. Based on our current interpretation of animal-microbe partnerships, we provide an initial assessment of these two steps, whereby using published data we quantify the degree that host-associated microbiomes differ between laboratory- and field-based studies (Table 1), as well as when the hologenome faces an environmental stress (Table 2).

Traditional animal models (e.g., Hydra: Fraune and Bosch, 2007; Drosophila: Shin et al., 2011; Mus: Sonnenburg et al., 2016) are powerful systems for laboratory experiments, particularly when dissecting the molecular mechanisms of organism-level processes under controlled settings. Substantial research with these species has led to fundamental discoveries in the formation, regulation, and diversity of microbial symbioses (e.g., Ley et al., 2005; Turnbaugh et al., 2006, 2009). However, these species, or any other for that matter, studied in the laboratory only partially represent the full spectrum of associations they may have with microbes in the natural setting where they have evolved (Chandler et al., 2011). In addition, species reared in the laboratory for many generations may have undergone artificial selection, intended or not, wherein phenotypic and genotypic traits having been selected for in response to abiotic and biotic environmental pressures may have been modified or lost (Kiers et al., 2007). As others have acknowledged before (e.g., Chandler et al., 2011; Har et al., 2015; Dirksen et al., 2016; Table 1) and as we do here, to what extent does the microbiome of laboratory animals reflect that of wild counterparts?

Marked declines in diversity and shifts in composition, and likely function, of the microbiome follow the onset of captivity, domestication, or other anthropogenically modified environments. Decreased microbial diversity and shifts in the species present have been reported for a number of the taxa, including but not limited to insects (Figure 1A), cnidarians (Figure 1B), lizards, fish, woodrats, and other mammals (Ley et al., 2008; Fraune et al., 2010; Wang et al., 2014; Kohl et al., 2014b, 2017; Mortzfeld et al., 2015; Clayton et al., 2016; Table 1). Therefore, defining the full diversity and functional traits of the microbiome with respect to the host would benefit from considering individuals in their native ecological niches to quantify if and how these differences shape the hologenome.

Recent studies using traditional model animals (e.g., Drosophila: Chandler et al., 2011; Caenorhabditis elegans: Berg et al., 2016; Dirksen et al., 2016; Samuel et al., 2016) have addressed the limitations of the laboratory and have begun performing complementary studies with wild populations. Drosophila, for example, has shown to be a tractable system for characterizing the role of the host genome in mutualistic partnerships with microbes. Using 14 Drosophila species spanning a broad geographical distribution and in direct comparison with common laboratory strains, Chandler et al. (2011) showed that the assembly of the microbiome is primarily bound by host diet and physiology, and the degree (i.e., diversity) of this assembly is a function of the environment (laboratory vs. wild) in which the individual resides. As such, laboratory-reared Drosophilids associate with a microbiome much lower in richness and diversity than wild conspecifics. Of the 139 Operational Taxonomic Units (OTUs) found to associate with Drosophila spp. in these settings, 90 (64.7%) were specific to wild individuals, six (4.4%) were unique to laboratory strains, and 43 (30.9%) were shared between environments (Chandler et al., 2011; Figure 1A).

Aquatic organisms, both marine and freshwater, inhabit fluids that are particularly rich in environmental microbiota. Similar to traditional terrestrial models (e.g., Drosophila and Caenorhabditis elegans), aquatic animals are amendable to both field and laboratory experiments. A rising aquatic model species in the fields of evolutionary ecology and genomics is the estuarine cnidarian Nematostella vectensis (Darling et al., 2005; Tarrant et al., 2015). Recent studies using laboratory populations have shown that the microbial diversity associated with N. vectensis individuals is influenced significantly by developmental stage, temperature, and salinity, as well as the geographic location from which individuals were collected, despite ten years of laboratory culture (Mortzfeld et al., 2015). Like with the Drosophilids, N. vectensis collected from their natural habitat and subsequently cultured in the laboratory lose a significant portion of their microbial diversity and, in turn, associate with species of microbes not identified in the field (Figure 1B) (see, Figure 3 in Har et al., 2015). Comparing field and laboratory cultured individuals from the same geographical location showed that of the total 29 OTUs found to associate with N. vectensis, 19 (66.5%) were specific to wild individuals, nine (31.0%) were unique to laboratory strains, and one (3.4%) was shared between environments (Har et al., 2015).

Like the examples above, Caenorhabditis elegans taken directly from native habitats associates with a rich community of microbial symbionts (Dirksen et al., 2016), which has significant impacts on individual physiology and life-history (Cabreiro and Gems, 2013). Similarly, gut microbiota of wild mice (Mus musculus) are rich in diversity (Weldon et al., 2015) and dominated by Bacteroides and Robinsoniella enterotypes, a term coined by Arumugam et al. (2011) in reference to clusters of microbiota communities. Following the onset of captivity, mice were found to only maintain an association with the Robinsoniella-dominated enterotype (Wang et al., 2014). In another rodent species, Kohl and Dearing (2014) showed that individual desert woodrats (Neotoma lepida) transferred to the laboratory share 64% of their gut microbiota with individuals found in the wild. Therefore, although model organisms have historically served as exquisite laboratory-based systems for studying animal genomics and development, this approach remains more limited to study the role of the microbiome on host ecology and evolution.

In nature, holobionts face a diversity of biotic and abiotic stressors that may challenge the host to associate with a microbial community that performs an appropriate physiological response. What this implies is that when facing an ecological “task” a holobiont maximizes fitness through changes in the associated microbiota, as derived from a larger network of microbial partners, to form a complementary metabolic and physiologic profile. As stated in the hologenome theory (see, Zilber-Rosenberg and Rosenberg, 2008; Rosenberg and Zilber-Rosenberg, 2013, 2016; Bordenstein and Theis, 2015), and as presented as a conceptual extension here, a network of microbial partners would enable the holobiont, with its respective hologenome, to acclimate to short-term changes in the local environment as well as serve as a fitness landscape for adaptation (e.g., Soen, 2014). We next compare the composition of host-associated microbiota across the host's natural environments and describe implications of an associated microbial repertoire.

The environment is a selective filter where variation in microbial communities is sorted, resulting in the opportunity for evolutionary innovation, whether the origin of eukaryotic cells (Margulis, 1970) or the gut microbiota of invertebrates and vertebrates (Alberdi et al., 2016; Shapira, 2016). The microbial community in association with a eukaryotic host is generally considered to be a mixture of resident and transient species that vary over space and time. Regardless of the host-microbe interaction, the associated microbial community represents a small subset of the total species present in the surrounding environment (Ley R. E. et al., 2006; Lynch and Neufeld, 2015). The genomes of these microbes, which are found in the environment, a fraction of which associate with a particular host, constitute the environmental metagenome (Lynch and Neufeld, 2015). The combination of obligate and facultative microbial symbionts represents a complex set of potential interactions between host and each individual microbe as well as amongst members of the microbial community. This set of interactions is vast and discerning the type of association (i.e., mutualistic, pathogenic, neutral; Mushegian and Ebert, 2015) is an immense but important goal for determining the relative role of a microbe to the holobiont. These interactions should, in part, be dependent on the specific environment experienced by the holobiont (Steinert et al., 2000) or more indirectly due to the host's condition, which may also depend on the environment.

Recent investigations have begun resolving questions focused on how and why microbial communities shift under abiotic and/or biotic environmental stressors (e.g., Casey et al., 2015; Lokmer and Wegner, 2015; Webster and Thomas, 2016). These associations can be broadly viewed as the product of a host genome by microbial metagenome by environment interaction, or G_H × G_M × E (Bordenstein and Theis, 2015). Further determining the functional role of the microbiome in the face of environmental stressors would require manipulating E and measuring G_M in the context of G_H. This would mean that, in some or all cases, G_H + G_M1 in E₁ differs to some degree from G_H + G_M2 in E₂. The sum of unique microbial associates (G_Mtotal) in E₁₊₂ therefore exceeds that of G_M1 and/or G_M2. G_Mtotal thus represents the microbial associates as part of the “hologenome” that exceeds either individual environment.

For example, when exposed to stressful temperatures, larvae of the Great Barrier Reef sponge Rhopaloeides odorabile lose partnerships with microbiota formed under ambient temperature (in particular the Nitrospira, Chloroflexi and a Roseobacter lineage) while forming partnerships with other microbiota (e.g., γ-proteobacteria) not previously part of the hologenome (Webster et al., 2011a). Under these two temperature regimes, Rhopaloeides odorabile larvae associated with 56 unique OTUs; 27 OTUs (48.2%) being specific to ambient temperature, 19 OTUs (33.9%) when faced with temperature stress, and 10 OTUs (17.9%) being shared between conditions (Webster et al., 2011a; Figure 2A). On the other hand, individual aphids (Acyrthosiphon pisum) specialized to the pea Pisum sativum as opposed to the red clover Trifolium pratense have five and two unique facultative associates, respectively, with three associates being shared between these diets (Figure 2B; Gauthier et al., 2015). Therefore, in these as well as other instances that span much of the animal kingdom (e.g., see Webster et al., 2011a; Gilbert et al., 2015; Har et al., 2015; Mortzfeld et al., 2015; Alberdi et al., 2016; Gilbert, 2016; Kohl and Yahn, 2016; Shapira, 2016; Webster et al., 2016; Webster and Thomas, 2016, as well as references therein; Figure 2), the collection of unique microbes as part of the hologenome in just two environmental conditions exceed that of either hologenome alone, such that by considering two environments in these focal cases (Webster et al., 2011a; Gauthier et al., 2015), the microbe associated with one host can more than double.

The properties of the environment can continually change, such that there are a nearly infinite number of unique environments, some combination of which must be taken into consideration when describing the evolutionary and ecological history of holobionts and their hologenomes. It is instructive to consider this interaction between the hologenome and the environment as G_H × G_M × E_∞. This raises the question: what is the maximum number of microbial species a host may associate with its hologenome over the course of its life cycle in the presence and absence of all natural and anthropogenic biotic and abiotic factors? The composition of G_M changes with respect to E due to the acquisition and loss of “transient” microbial symbionts while G_H is nearly constant per generation but dynamic over evolutionary time. For simplicity, if we first consider G_H as a constant for a given host and that the composition of G_M differs with respect to E and is subsequently integrated across this continuum, then G_H plus the sum of G_M, or G_H+M, should represent the host genome plus the maximum number of microbial species a host may associate with over the course of its life-history under all encountered environmental circumstances. As such, we define this as the “host-associated microbial repertoire” (H):

Furthermore, for a given species, G_H is variable (e.g., Wegner et al., 2013; Mortzfeld et al., 2015; Chong and Moran, 2016) between individuals within a population and dynamic over evolutionary time, implying that the magnitude and composition of H may vary with respect to the interaction between G_M and E. Thus, each G_H would have a specific capacity to establish partnerships with a set number of microbial species due to the host genome and the environment (see, Sonnenburg et al., 2016).

The environmental factors contributing to the structure and composition of the hologenome is also influenced by time (t). In the context of the host-associated microbial repertoire, t can be represented in two primary ways: (i) absolute time of the host genome, microbial metagenome, and environment (“ecological time”) or (ii) accumulative time for co-evolution of the holobiont (“evolutionary time”). For a holobiont, absolute time is the duration of a specific cycle at each level of the G_H × G_M × E interaction, whereas the generation time for a bacterium is typically minutes to days, while that of the host could be decades. Environmental cycles, on the other hand, can encompass all of these time scales: the North Atlantic Oscillation, one of the most prominent and recurrent patterns of atmospheric variability, differs on a decadal scale (Hurrell et al., 2003), but temperature, oxygen, pH, and other abiotic factors may vary daily. Accumulative time, however, more explicitly integrates across timescales when describing chronologically distinct events over the course of development or selection of advantageous characteristics. Thus, this suggests that the repertoire of microbial species a host genome can associate with should plateau (i.e., reaching H) over evolutionary time based on environment-specific physiological requirements (Figure 3) (e.g., Reveillaud et al., 2014). We may now incorporate t into Equation 1 and, in doing so, create an equation that is conceptually analogous: (2)H=∫0t∑​GM(t)⋅GH(t)⋅Et(t)dt Here, instead of integrating over E specifically, we integrate over t; thus, demonstrating that both t and E are important components driving ecological and evolutionary change of holobionts and hologenomes.

The environment experienced by the host, whether laboratory vs. field or the presence/absence of a stressor(s), may largely define the composition and structure of the associated microbiome and corresponding physiological function (Turnbaugh et al., 2009; Chandler et al., 2011; Webster et al., 2011a; Gauthier et al., 2015; Har et al., 2015). Based on the conceptual framework above, of which we view as a further development of an existing component of the hologenome theory of evolution (Zilber-Rosenberg and Rosenberg, 2008; Rosenberg et al., 2009; Bordenstein and Theis, 2015), we identify three “signatures” for applying our hypothesis: diet, indirect life cycles, and seasonality (but see Kohl and Carey, 2016 for others). In doing so, we use examples from “model” and “non-model” animals across the animal kingdom with unique life-histories that may serve to answer sets of questions pertaining to evolutionary and ecological functions of the microbiome.

Nearly a decade ago, Zilber-Rosenberg and Rosenberg (2008) proposed the hologenome theory of evolution, which has been summarized, expanded (Figure 4), and clarified in recent years, providing novel hypotheses for the evolution of holobionts (Bosch and McFall-Ngai, 2011; Gilbert et al., 2012, 2015; Theis et al., 2016). Current theory and research on holobionts are now beginning to explicitly consider the variation in a natural environmental context, in part, because it is a necessary factor to understand the evolution of holobionts and hologenomes (Zilber-Rosenberg and Rosenberg, 2008; Theis et al., 2016). Studies of the host-associated microbial repertoire are useful in the effort to highlight the importance of microbes in host biology (e.g., physiology, development, immunology, behavior, population genetics; Jaenike, 2012; Eisthen and Theis, 2016; Kohl and Carey, 2016; Shropshire and Bordenstein, 2016) and ecology. Furthermore, this concept may aid in advancing our understanding of how these associations may have driven or directed the evolutionary trajectory of the holobiont, and their inheritable repertoire of symbiotic microbiota (van Opstal and Bordenstein, 2015; Gilbert, 2016).

Our Hypothesis and Theory article has primarily focused on animal-associated microbiota but numerous studies suggest that plants, fungi, and other eukaryotes would have similar associations. Plant-associated microbiota (primarily bacteria and fungi) clearly have important roles in nutrient acquisition and buffering against both abiotic and biotic stressors, such that the plants are associated with particular microbiota that aid in acclimating to the local environment (Vandenkoornhuyse et al., 2015). For example, the evergreen tree Metrosideros polymorpha inhabits environments that broadly range in annual precipitation as well as mean temperature. In profiling the fungal endophyte communities across such abiotic gradients, the composition of these communities is directly related to temperature and rainfall (Zimmerman and Vitousek, 2012), implying that like animal examples presented through this article, plants may also associate with a larger network of microbial partners as a product of environmental variation.

As microbial taxa have been linked to specific evolutionary processes, such as Wolbachia and reproductive compatibility (e.g., Bordenstein et al., 2001) or Vibrio fischeri and countershading (e.g., Nyholm and Mcfall-Ngai, 2004), clusters of microbial associates as part of the host-associated microbial repertoire may be linked to ecological functions (Krause et al., 2006; Langille et al., 2013) including tolerance to salinity (e.g., Schmidt et al., 2015) and/or temperature (e.g., Webster et al., 2011a; Kohl and Yahn, 2016). Characterizing the host-associated microbial repertoire and subsequent function for clusters of associated microbiota may, therefore, serve as an conceptual framework that links the ecological role of microbes to the evolution of the host with which they associate. Furthermore, although these microbial associates have an ecologically important function for the holobiont, the environment in which they have the potential to associate with a holobiont may, in turn, be physiologically stressful to the microbes; thus, it is equally important to study the physiological parameters of the members of the host-associated microbial repertoire.

All authors listed, have made substantial, direct and intellectual contribution to the work, and approved it for publication.