Pre/post-mating isolation and the hologenome concept of evolution A

The experiments conducted by Brucker and Bordenstein, the theoretical apparatus they introduced and the new concepts they have used to explore the consequences of HCE are probably some of the most elegant and influential steps that have been taken to gather wide support for the heavily controversial

36 _{The same type of response is given to refute Chandler and Turelli’s suggestion about the}

cross-inoculation experiment: while Brucker and Bordenstein accept that this would be additional evidence to support the hologenomic basic of speciation in Nasonia, they think that demanding that kind of experiment is unreasonable, since the same experiment could be demanded for every speciation gene in animals, and that type of experiments is usually not carried out.

hypothesis that natural selection can act on the multispecies consortia that hologenomes represent. However, their interpretation of the experimental results that they obtained, as well as their application of the principles of HCE to speciation is not without contestation. Specifically, their general explanation of the microbial-assisted BDM model of reproductive isolation (Figure 6) allows a

completely different reading, closely allied to Chandler and Turelli’s

interpretation of their results about the hybrid lethality in Nasonia. The reading would be as follows: every bacterium that interacts with a host is part of its environment. As such, some interactions will be beneficial, increasing its fitness, whereas others will be detrimental, decreasing it, or even killing the host in the most extreme cases. When, then, is it legitimate to consider the holobiont as the unit that speciates/gets selected, and when it is not? Brucker and Bordenstein need to be very specific about their way of answering this question, since their hypothesis about the BDM model does not clearly screen-off environmental factors from intrinsic factors. And this distinction is essential if the dispute is about the unit of selection qua replicator, which is the role that hologenomes are

assumed to play in the context of HCE.37

The existence of phylosymbiotic patterns in nature, though, could be argued to fill in this gap. The hologenome, thus, would be the ensemble of phylosymbiotic species. But notice that this could drastically reduce the ‘dimension’ of the holobiont, since the phylosymbiotic species that interact with a host are just a subset (variable in size) of the species that interact with a host during its lifetime. And, more importantly, proving the existence of phylosymbiotic patterns might yet not be enough to prove that holobionts are replicators because some of the phylosymbiotic species might interact within the holobiont for environmental—i.e. not intrinsic—reasons. It seems thus necessary to distinguish which elements are environmental and which elements are intrinsic if the position that will be defended is that the hologenome is a unit of selection qua replicator (for a long discussion of these issues, see chapter II).

4. ‘Getting the hologenome concept right’. Clarifying the eco-evolutionary principles of the hologenome concept of evolution

The formulation of HCE soon gave rise to some serious theoretical criticisms,

which I will review in chapter II.38 _{Those criticisms moved the authors working}

under the postulates of HCE to clarify the specific meaning of their hypothesis, which gave rise to two very important pieces in the literature about HCE: Bordenstein and Theis (2015), and Theis et al. (2016).

Bordenstein and Theis’s (2015) paper summarized ten principles about the holobiont and the hologenome with the purpose of making HCE clear for future research, including possible ways of evaluating the empirical validity of the hypothesis. I think what makes their contribution more valuable is that they make clear how not to conceive holobionts, more than they directly clarify how positively to conceive them. This is reasonable because, as the authors have expressed: ‘Holobionts and their hologenomes are less entities that elucidate something per se than they are entities that need elucidation’ (Theis et al. 2016: 2). Let me briefly summarize what I take to be the key elements that Bordenstein and Theis elucidate in their paper. First of all, the authors make clear that the holobiont must be considered as a unit of biological

organization,39 _{meaning basically that host-microbiome associations}

(hologenome) should not be considered as genotype-by-environment (G x E) interactions in any of its possible interpretations (i.e. the microbiome being an environment for the host, or the host being an environment for the microbiome), nor as a phenotype encoded by the host. In their view, host-microbiome associations must be conceived as genotype-by-genotype-by-environment (G x G x E) interactions. This last point should be interpreted as stating that the holobiont acts as a single unit with its environment so that certain G x G interactions can be selected for. But, what would be the conditions that G x G

38 _{The reason for presenting the clarificatory papers first, and the criticisms later, is that none of}

the clarificatory papers directly tackled any of the criticism ‘in a philosophical way’, so to speak. Rather, they just stated HCE in more specific and operationalizable terms, which help to perceive the historical development of the hypothesis.

39 _{Bordenstein and Theis never explicate what they mean by ‘unit of organization’, but from how}

they continue the section it might well be assumed that their definition has nothing to do with the meaning given by defenders of the organizational account of individuality (Moreno and Mossio 2015).

should satisfy so that evolution can operate on them? Basically, that their degree of temporal persistence is enough so that evolution can operate on them, and this would happen only if there is a sufficient degree of co- inheritance. Bordenstein and Theis argue this to be so based on the evidence about horizontal transmission of the microbiome reviewed in Bright and Bulgheresi (2010), Funkhouser and Bordenstein (2013), and Rosenberg and Zilber-Rosenberg (2013) (Bordenstein and Theis: Principles I and IV).

Second, following a previous statement by Rosenberg and Zilber- Rosenberg (2013), Bordenstein and Theis clarify that the holobiont is neither an organ system, nor a superorganism, nor a metagenome. These clarifications are far from trivial: Firstly, organs are conventionally individuated because they perform one function in a system; by saying that the holobiont is not an organ system, they are neglecting the possibility of considering the microbiome as an organ, thus clarifying that it can play more than one function. And, because of this, they are simultaneously indirectly clarifying that transgenerational changes in the composition of the microbiome are not commensurable with transgenerational changes in a quantitative trait. Therefore, they cannot be studied under the framework of ‘standard heritability’ h2_{; a different way of}

measuring it is required. Secondly, by clarifying that the holobiont is not a superorganism, the authors are clarifying that it is a polygenomic entity, i.e. an entity necessarily composed by organisms of different species. Thirdly, by clarifying that the holobiont is not a metagenome, they are clarifying that it is not constituted by a host plus all its ‘environmental microbes’, but only by a subset of those. Unfortunately, they fail to define precisely which is this subset of the totality of microbes that interact with a host that should be included in the holobiont (Bordenstein and Theis: Principle II).

A clarification of this last problem is however provided in Theis et al. (2016). According to them, the holobiont is composed of the individual host and its microbial community, and the members of this microbial community ‘can be constant or inconstant, can be vertically or horizontally transmitted, and can act in a context-dependent manner as harmful, harmless, or helpful’ (2016: 1).

holobiont at a given point of time’ (2016: 1) (Figure 12).40 _{Notice that the}

definition encompasses three elements. First, a temporal timescale: the holobiont includes every symbiont that interacts with a host, no matter how long their interaction lasts (‘constant or inconstant’). Second, information about the modes of transmission/acquisition of the entities that compose it. Third, information about the possible effects of the microbes of the microbiome on the holobiont. Furthermore, the definition puts the emphasis on the fact that the hologenome is a spatio-temporally located entity which experiences/goes through (and results from) a series of eco-evolutionary processes, including selection, genetic drift, genetic conflict, epistasis, etc. As such, selection on the hologenome can lead to different evolutionary outcomes, including, but not restricted to, coevolution (see also Bordenstein and Theis (2015): Principles VII and VIII).

Figure 12. Schematic representation of the hologenome. The hologenome includes the total

sum of genes of the members of holobiont, including the genome of the host and part of the genetic material that is contained in the microbiome (the other part would constitute an environmental metagenome). The complete genetic material of the microbiome includes a) host and symbiont genes that affect the phenotype of the holobiont but have not coevolved; b) host and symbiont genes that affect the phenotype of the holobiont and have coevolved; c) host and symbiont genes that do not affect the phenotype of the holobiont. (From Theis et al. 2016: 2, Fig. 1)

40 _{This definition is different to the definition presented in Roughgarden et al. (2017), where the}

holobiont is defined as a host plus its symbionts, i.e. including long-term pathogens, but excluding pathogens ‘which kill their host or depart in a few days’.

From the definition of the holobiont presented above follows another consequence, which is the pluralistic view of the holobiont that HCE defenders embrace. In some of the original formulations of Brucker and Bordenstein, the authors put the emphasis on the cooperative (beneficial) microbiome as the entity that, together with the host, constituted the hologenome. However, in the definition presented in Theis et al. (2016), the authors put the emphasis both on cooperation and on conflict, without restriction. In fact, that is not surprising, since defenders of the hologenome concept have always emphasized their embracing of multilevel selection theory (MLS, hereafter), the notion that natural selection targets different levels of the biological hierarchy simultaneously. As such, it is expected to result in contradictory effects on each level of the biological hierarchy, promoting cooperation (low conflict) in some levels, and competition in others. In principle, selection at the level of the hologenome will tend to reduce the level of conflict among the host and its microbiome, but this is not necessary to argue that the hologenome is a unit of selection, as Theis et al. emphasize, since HCE presupposes MLS. Thus, they conclude: ‘Hologenomes then exist as hierarchically nested, although not necessarily integrated, levels of genomes in which all levels of selection are in play’ (Theis et al. 2016: 4). This statement, as well as the rest of the clarifications about the multilevel nature of the holobiont, will be reviewed more carefully in chapter V.

5. The hologenome concept of evolution is a story about songs, not about