‘The importance of symbiosis in philosophy of biology: an analysis of the current debate on biological individuality and its historical roots’

1.1 Anton de Bary (1831–1888)

The introduction of the term “symbiosis” in biology is usually credited to Anton de Bary, who originally used it for the first time in the history of biology in his speech to the Association of German Naturalists and Physicians, “Die Erscheinung der Symbiose”, the “phenomenon of symbiosis” (de Bary 1878 [2016]; Oulhen et al. 2016).⁵ Nonetheless, one year before de Bary’s lecture, Albert B. Frank had introduced the term “Symbiotismus”, to designate those “cases where two different species live on or in one another” (Frank 1877, quoted in Sapp 1994: 6). When they first used the term, both Frank and de Bary were interested in the study of lichens, whose dual nature had been hypothesized ten years before by Simon Schwendener (Honegger 2000; Egerton 2015; Gontier 2016a). For Schwendener, the dual nature of lichens was understood as a relationship where the fungus is in control of the algae, that it uses to obtain its nutrients, but also, and most importantly, as a new biological individual: “the organisms are so intrinsically and reciprocally connected that through their penetration and merging, they constitute new plants with a clear individual character” (1868, quoted in Gontier 2016a; emphasis added). De Bary, drawing upon those observations plus the experimental results that demonstrated that the dual nature of lichens was not merely a fiction of Schwendener –the two elements that constitute the lichen were separated for the first time in 1876, and by 1877 it was already possible to synthetize lichens in the lab by merging algae with fungal spores (Stahl 1877, referred in de Bary 1878 [2016]; Sapp 1994; Egerton 2015: 104, 106; Gontier 2016a: 276)–decided to refer to “the living together of differently named organisms” by the term “symbiosis” (1878 [Oulhen et al. 2016: 133]).

In his original lecture, de Bary emphasizes two aspects of symbiotic relationships: first, the different degrees of dependency that the partners in a symbiotic relationship sometimes generate with respect to each other; second, the different kind of effects that can be generated as a consequence of the symbiotic association. With respect to the latter point, by the time when de Bary coined the concept of “symbiosis”, Pierre-Joseph van Beneden’s classification of the different types of associations between organisms in mutualism, commensalism and parasitism, had become very popular (van Beneden 1876), and de Bary would precisely use that classification to better capture the nature of symbiotic phenomenon, a phenomenon of which, in his words: “[p]arasitism, mutualism and lichenism are special cases” (1878 [Oulhen et al. 2016: 136]; emphasis added).⁶ With respect to the former, de Bary’s speech is predominantly dedicated, in almost its totality, in explaining the types of dependencies among partners, including, especially, the morphological and physiological effects that symbiosis can cause the individuals that are interacting symbiotically. Interestingly, he decided to include lichenism as a distinct type of symbiotic relationship together. Why, then, is lichenism different to mutualism and commensalism?

Most of de Bary’s paper is dedicated in explaining the association between Azolla and Anabaena, Nostoc and Cycas, and the fungi and algae that constitute lichens. In fact, de Bary seemed to perceive something particular in those associations, which is the reason why he asserted:

Furthermore, he also discards the hypothesis that those associations might be considered simply as cases of mutualism: “[i]t is however doubtful that there are mutual advantages to the partners. We can definitely say that they do not harm each other significantly (…). But presently, we have no evidence of the mutual benefits that they could afford each other” (1878 [Oulhen et al. 2016: 136]). What de Bary finds particularly noticeable about the cases of lichenism are precisely the morphological and non-pathological effects of these types of associations, as well as the sorts of physiological dependencies that emerge from the partners. Drawing directly upon the experiments carried out by Stahl (1877), he highlights the important morphological changes that accompany the synthesis of lichens: “right after their association with the fungus of the lichen, the cells of the algae become much larger, contain more chlorophyll, [and] are stronger in every way. Beyond doubt, according to data, that have been known for a long time regarding the structure of the lichen, all of these characteristics are retained for the entire life cycle of the lichen, sometimes for several dozen years” (1878 [Oulhen et al. 2016: 138]). It is precisely at this place of his discussion, when de Bary applies Darwin’s theory, explaining that symbiosis might work as an inducer of the morphological changes that are required for natural selection to generate adaptation (see also Sapp 1994: 9; Sapp 2003: 234–251).

What it is at stake in de Bary’s discussion of lichenism is also a debate about the nature of biological individuality. If the elements that conform the lichen are dissociated, the lichen does not exist anymore, and we would only have a fungus –that will eventually die– and an alga. However, if we put them together to generate new lichen, they become somehow de-naturalized, they lose their main morphological characteristics, adopting a new configuration that makes them different from their free-living counterparts. De Bary’s insistence in the non-parasitic, non-mutualistic nature of lichens is also noteworthy. On the one hand, he seems to have identified a new dimension of the “living together”, which was not reducible to van Beneden’s categories. On the other hand, he still wanted to keep the concept of “symbiosis” to name the associations that, in his words “we can group under the term sociability” (1878 [Oulhen et al. 2016: 136]), including some associations that do not question the individuality of the partners involved (e.g. pollination). Jan Sapp argues that “[t]his was a strategic argument that was designed to ensure that lichens were not discarded as exceptions” (1994: 9). I agree with him, and I think de Bary actually believed he was identifying a very distinct phenomenon, which questioned the conventionally accepted boundaries of biological individuals.⁷

1.2 Roscoe Pound (1870–1964)

After de Bary delivered his lecture, research on symbiosis started growing and new cases were discovered: Karl Bradt discovered the presence of the symbiotic alga Zooxanthella in the bodies of Hydra and sponges (1881, in Sapp 1994: 11) and Patrick Geddes discovered the presence of non-pathogenic alga in sea anemones (1882); some time later, Albert B. Frank discovered the presence of fungi in the root of legumes, which, he hypothesized, was a symbiont with important physiological functions, naming it “mycorrhiza” (1885 [Frank 2005]). In this period, symbiosis practically became identified with mutualism to a point where the two terms became interchangeable, while the general meaning of which de Bary had suggested became lost (Sapp 1994: 18–34; cf. Martin and Schwab 2012, who argue that the association between symbiosis and mutualism lasted until 1970).⁸

It is precisely in this context that Roscoe Pound lectured his: “Symbiosis and mutualism,” with the aim of disentangling the two concepts (Pound 1893). Pound started his paper by distinguishing three types of relationships between hosts and parasites: those where the host kills the parasite; those where the parasite kills the host; and those where “the host lives on side by side with the parasite indefinitely” and continues “[a] further development is attained in cases where the parasite and host not only live together, but are mutually beneficial, and, perhaps, even, in extreme cases, inter-dependent” (1893: 509; emphasis added). For him, following de Bary, symbiosis just meant “living together for a long time,” and mutualism is just one of the forms that this “living together” might take. Yet, in most circumstances, he claims, this “living together” does not take the form of mutualism. Furthermore, he provides some clarification that is especially important for the debate about biological individuality: acknowledging that mutualism can take forms other than “living together,” he says “it should be noted that the mutualism of which we are here speaking is mutualism of parasite and host –not mutualism of independent organisms” (1893: 509; emphasis added).Why that distinction between cases of inter-dependent organisms versus cases of independent organisms? What makes the former cases so special? We must recall that de Bary had explicitly said that he had no objection to using “symbiosis” to refer to those associations that can be grouped “under the term sociability.” My impression is that Pound had already perceived the qualitative difference between those two types of association: whereas the latter do not compromise the concept of biological individuality, as the organisms that interact can be clearly recognized as independent, the former do, in so far as the organisms (1) live in close association during all their life cycle and (2) might become inter-dependent to such a agree as to form a new entity.

Granted, there is a qualitative difference between the two types of associations (i.e. sociability vs. symbiosis) in so far as the latter, but not the former, compromise currently accepted ideas about biological individuality. In particular, it challenges the idea that one individual can belong to only one species classified according to the criteria of systematists. The rest of Pound’s paper is dedicated in arguing that the biologists of his time had tended to overemphasize the presence of mutualism, claiming to have identified it in many symbiotic associations, where it was not at all clear that the partners were acting mutually. First, he argues, mutualism does not occur in every lichen: it does not exist in homoeomerous lichens, or in what he calls “pseudo-heteromerous lichens,” although evidence suggests that it might exist in heteromerous lichens, as they exhibit a complex interdependence among the fungus and the algae that form the lichen (Pound 1893: 511–513). Second, he analyses Frank’s studies on mycorrhizas and, while recognizing part of Frank’s discoveries, he refers to some evidence by R. Hartig, “a more sober and trustworthy writer than Frank” (1893: 516). He argued that:

Finally, he considers the presence of Rhizobium in the root of Leguminosae.⁹ He says that the evidence is uncertain, and although it might sometimes seem as if the Rhizobium were mutualists, “[t]he bacteria (…) are parasites. They are there for their own purposes, and are incidentally beneficial to the plant” (1893: 518). Moreover, while admitting that in some cases the symbiosis might lead to a mutualism –as the plants infected do better than those uninfected–he continues diminishing the evolutionary importance of these symbioses by criticizing some of Frank’s observations:

And he concludes his paper saying:

Even despite Pound’s dismissal of the importance of mutualistic symbiosis, as well as its general importance, his example helpfully illustrates the general awareness of the phenomenon among biologists in the late nineteenth century. Especially remarkable is his insistence of distinguishing between those cases where interdependence is generated versus those where two (or more) individuals can be recognized as different. Second, and also remarkable, is his way of neglecting the individuality of the symbiotic aggregate. As he expresses here and there, even if in some rare cases the individuality of the symbiotic aggregate might occur, the organisms are there for their own benefit, and many of them would probably be better outside the symbiosis.¹¹ These claims have two important consequences. First, it suggests that the general rejection of symbiosis research by biologists writing at this time was for the reason that it seemed to negatively affect the traditional conception of biological individuality and “struggle for life” (see also Sapp 2003, 2004). Second, it paves the way for a new and important conceptual change in symbiosis, the important division between symbiosis and other forms of sociality, forms that de Bary had considered as manifestations of the same phenomenon.

1.3 Albert Schneider (1863–1928)

The next important step in the development of the concept was because of Albert Schneider, who in “The phenomena of symbiosis” proposed a new understanding of the symbiosis as “a continuous association of two or more morphologically distinct organisms, not of the same kind, resulting in a loss or acquisition of assimilated food-substances” (1897: 925). There were three purposes to his paper: first, to distinguish clearly between cases of associations of living thing and cases of real symbiosis; second, to suggest the possible evolutionary origin of symbiosis, accounting for the default behaviour of organisms, which he understood as a “struggle for life”; and third, to classify different types of symbiotic associations. Of course, the three questions are closely connected to one another: once symbiosis is distinguished from mere “association,” the classification of different types of symbiotic phenomena will be partially based on evolutionary criteria. Therefore, the different types of symbiotic relationships will be distinguished by degrees, from the forms that entail independent individuality of the organisms that interact, to those where the associated organisms lose their individuality and merge to form a higher level entity.¹²

Commenting on this classification Schneider makes two important remarks: first, the development of the different types of symbiotic associations and their particular character will depend largely on environmental opportunity; second, he puts the emphasis in studying “the phylogenetic relationship of the symbioses without any reference to the phylogeny of the organisms comprising them” (1897: 931). In this vein, as he indicates, it means that one does not need to study the phylogenetic evolution of the specific organisms that engage in the symbiotic relationship, but only the relative evolution of the physiological relationship. Figure 1, taken from Schneider (1897: 932) presents his phylogenetic schema of the physiological evolution of symbiotic relationships.

Open image in new window

Schneider acknowledges from the beginning of his paper the difficulty of determining the starting point of the symbiosis. Under the heading of “incipient accidental symbiosis” he includes those cases where the organisms are in close physical contact for a sufficiently prolonged time, understood ontogenetically, and irrespectively of whether or not morpho-physiological changes (either antagonistic or mutualistic) occur.¹³Moreover, he argues that, once an accidental symbiosis has been established, then the condition will immediately be subject to change, since the permanency of this symbiosis is in direct proportion to the degree of mutualistic specialization (1897: 934). In this sense, if an accidental symbiosis is not broken down, it will evolve towards a “contingent symbiosis,” where the organisms involved, despite not experiencing any morpho-physiological change, seem to manifest a sufficient degree of elective affinity. One case of contingent symbiosis, according to Schneider, is the bacterial flora of humans, which shows a certain degree of elective affinity but does not seem to show any kind of morpho-physiological relationship with the host.¹⁴

The second symbiotic phenomena considered are the cases of “Antagonistic Symbiosis.” According to Schneider, this category includes “mutual parasitism,” i.e. the situation where both organisms live together but their relationship is mutually damaging, “parasitism,” a situation where one of the organisms is damaged whereas the other obtains benefit from the relation, and, as a limiting non-symbiotic case, “saprophytism”. For Schneider, antagonistic forms of symbiosis can only give rise to very limited morpho-physiological specializations or adaptation, since the parasitic nature of the relationship causes it to be “a destructive association, [such that] [t]he morphological and physiological changes tend towards dissolution rather than evolution” (1897: 936). Therefore, he has a reason to believe that, even if antagonistic forms of symbiosis are conceivable (and, in principle, even expectable), a case of antagonistic symbiosis either evolves towards a case of mutualistic symbiosis or it will be driven towards extinction.¹⁵

The final kind of symbiotic phenomena, the cases of “mutualistic symbiosis,” can occur when two organisms interact with each other so that the relation is mutually beneficial. The mutual benefit might occur either because one organism benefits another without being damaged (“nutricism”), because both organisms “mutually benefit each other [while] are still capable of leading an independent existence” (“mutualism”) (1897: 941), or because “one or more of the symbionts is absolutely dependent upon the other for its existence” (“individualism”) (1897: 943). Schneider remarks, however, that it is very unlikely that something such as “absolute nutricism” really occurs in the biological world. He acknowledges that in some symbiotic associations one of the symbionts is clearly benefitted, whereas the material benefits for the other are not so clear. However, he thinks that in most cases nutricism will tend to evolve towards a relation of mutual benefit for both partners. This last type of relationships might happen either in cases where both symbionts can carry independent existence (he mentions insectivorous plants and their bacteria, Actinia prehensa and Melia tessellata or some species of ants and the branches of trees), or in cases where they are mutually dependent. About this last case he claims “[i]t (…) represents a higher form of mutualism, from which it is no doubt phylogenetically derived. (…) [In individualism] [t]he associations form an individual, a morphological unit, and the phenomena are frequently not recognized as symbiosis” (1897: 943, emphasis added).

It is important to realize, at this point, that Schneider’s work is conceptually revolutionary. First, he is the first to consider the possibility of studying the phylogenetic history of symbiotic associations (i.e. their evolution): (1) irrespectively of the evolution of the organisms that form the symbiosis; and (2) relative to the opportunities that the environment offers for their evolution (in this sense, symbiotic assemblages would be something conceptually similar to what we now call “units of selection”). Second, he realizes that symbiotic associations challenge the individuality of the organisms that interact, to the point that they might become a new independent emergent individual. Conceptually speaking, Schneider is the first author to recognise this last fact, thus opening the possibility of understanding symbiotic associations as genuine evolutionary individuals in their own right. Furthermore, he is conscious of the physiological importance of symbiosis, as well as why it is occasionally not possible to understand the physiology of the organisms in isolation from their symbionts.¹⁶ This fact has gained a lot of attention recently, especially after the hologenome concept of evolution was proposed.

1.4 Constantin Merezhkowsky (1855–1921)

A final and important step in the conceptual development of the association between symbiosis and biological individuality is because of the work of Constantin S. Merezhkowsky. In his “Über Natur und Ursprung der Chromatophoren im Pflanzenreiche,” (“On the nature and origin of chromatophores [plastids/chloroplasts] in the plant kingdom”), Merezhkowsky proposed, for the first time, the term “symbiogenesis,” further advancing the conception of symbiosis as an evolutionary mechanism (Merezhkowsky 1905, 1910) (see also Khakhina 1992; Sapp 1994: 47–59; Martin and Kowallik 1999; Sapp et al. 2002: 418–423; Gontier 2016b).

In his paper, Merezhkowsky aimed at discerning the origin of chloroplasts in plants. During his time, it was commonly believed that the chloroplasts, contained in the body of plants, appeared de novo every new generation and, furthermore, that they originated autogenously, as new organs which differentiated within the bodies of plant cells. Merezhkowsky strongly disagreed with that conception. Drawing upon Schimper’s discovery that chloroplasts do not appear de novo in plant cells, but are always present within their bodies since the beginning of the life of the plant (1885, referred in Merezhkowsky 1905), he proposed a revolutionary notion: chloroplasts should not be regarded as autogenous organs of plants, but as symbionts, i.e. as independent (foreign) organisms that live together with plant cells. Merezhkowsky offered two different types of arguments to support his theory. His first two arguments were theoretical. The first one was based on Schimper’s discovery: if chloroplasts do not arise de novo, though invaginations of the cytoplasm of the cell, but “rather, they always arise through division of pre-existing plastids, and since the latter in turn arise from pre-existing plastids, etc., we necessarily arrive at the logical conclusion that long ago the first chromatophore migrated into a colourless organism” (1905: 596 [Martin and Kowallik 1999: 289]).¹⁷ Secondly, Merezhkowsky argued that chloroplasts can be understood by analogy to Zooxanthella in the body of Amoeba viridis. In both cases, the structures (chloroplasts and Zooxanthella, respectively) can be said survive, divide and behave as independent organisms. If biologists do not have any issue in understanding Zooxanthella as independent symbiotic organisms within the bodies of their hosts, they should not have any prejudice in applying the same type of reasoning to chloroplasts, providing that the empirical evidence supported this conception.

The rest of the rationale to his hypothesis were of empirical observation. First, the discovery that chloroplasts, in contrast with other “organs” (“organelles”) in the body of plant cells, can survive and reproduce even after the nucleus of the cell has been removed, and also can do so outside the cell’s cytoplasm, which suggests that they behave like independent organisms. Second, the similarity between chloroplasts and free-living bacteria, concretely, with free-living forms of Cyanophyceae, which has been qualified by Martin & Kowallik as the “unquestionably most novel line of reasoning” (1999: 287). According to Merezhkowsky, chloroplasts and Cyanophyceae had a very similar physical appearance, both in form and colour, very similar biochemical (physiological) properties (with a similar type of nutrition), and analogous ways of proliferation and reproduction, which “makes it exceedingly likely that chromatophores are Cyanophyceae that invaded the plasma” (Merezhkowsky 1905: 600–601 [Martin and Kowallik 1999: 291]). Finally, he argued that, as it was empirically proven that Cyanophyceae can also engage in symbiotic relationships with other organisms (diatoms, rhizopods, etc.), even with cells that are protected by a cell wall, it was possible that at some point in their evolutionary history Cyanophyceae could have entered in contact with a plant cell so as to give rise to chloroplasts.

Merezhkowsky’s symbiogenetic hypothesis, as well as his arguments, gives symbiotic ideas a new meaning. Authors writing prior to him had discussed the importance of the symbiotic relationship, the nature of the symbiotic relationship, how symbiotic relationships could cause several morpho-physiological changes in biological individuals, etc. However, no one had considered the possibility that symbiosis might be a hereditary phenomenon, i.e. that symbiotic associations might be intergenerationally transmitted (e.g. like gametes passing between germ-line cells). Authors had assumed that genetically heterogeneous organisms reproduced independently, and later would form symbioses. Merezhkowsky, on the contrary, challenged the necessity of this assumption; and, by implication, questioned the boundaries of biological individuals, understood evolutionarily. For instance, it is not just that different organisms engage symbiotically and later their morpho-physiological independence is lost; in the case of plant cells, also their hereditary independence (evolutionary individuality) is lost, as the two previously independent organisms are now inherited exclusively together. In summary, Merezhkowsky includes the main element that was lacking in the symbiosis picture, conceiving, for the first time, the idea of hereditary symbiosis. In one sense, symbiotic assemblages had already been attributed all the necessary elements for being considered units of selection (variance, inheritance, fitness; e.g. Lewontin 1970; Godfrey-Smith 2009). The question afterwards changed this sense, since it now asked us to determine the real importance of heritable symbiosis. Was this just an isolated case special to a different phenomenon or was it general to symbiosis as such?

The notion of hereditary symbiosis was later supported by Hermann Reinheimer, Andrei S. Famintsyn and Boris M. Kozo-Polyansky (1924/2010) (Khakhina 1992; Sapp 1994: 47–59; Carrapiço 2015). Afterwards, Paul J. Portier (1918) and Ivan E. Wallin (1927) would apply symbiogenetic ideas to the origin of mitochondria, extending Merezhkowsky’s original application to another cellular organelle. And even later, Paul Buchner would explore the importance of hereditary symbiosis in insects, proposing a new field of application for the hypothesis (Boucher 1965; Sapp 2002). Symbiogenetic theories of the origin of the eukaryotic cell, however, were frequently rejected. This trend continued for almost 50 years, until Lynn Margulis provided new support and the symbiotic origin of the eukaryotic cell became almost universally accepted (Sagan 1967; Margulis 1970, 1991, 1993; see Sapp 2010). Nonetheless, it is important to remark that the conceptual basis for understanding the role of symbiosis in evolution, as well as the possibility of considering some symbiotic assemblages as what we would call “units of selection” in contemporary jargon, were already settled by Merezhkowsky in 1905. By then, all the conceptual connections between the notions of symbiosis and biological individuality were already present, as well as the conceptual challenges that the former presented for traditional conceptions of the later. In the next part of the paper, I will explore how those conceptual connections have been explored in recent times, especially after the proposal of the hologenome concept of evolution.

2.1 The origin of the concept – the importance of Lynn Margulis (1938–2011)

Lynn Margulis (born Alexander) was a pioneer in the field of symbiosis, to which she dedicated almost 50 years. She is especially known for giving new life to the hypothesis of the symbiotic origin of eukaryotic cells, as well as for her enthusiasm about the importance of symbiosis for life on Earth and evolution (Margulis 1990, 1991, 1998, 2010; Sagan & Margulis 2002; Díaz 2015; O’Malley 2017). Margulis is acknowledged as the first person to introduce the term “holobiont”, which was published in her paper “Words as battle cries – symbiogenesis and the new field of endocytobiology” (1990). In this work, she compares cyclical hereditary symbiosis with meiotic sex. In both of these compared cases of inheritance, she argues that two entities are present, which cyclically recognize each other and merge together for every generation. Moreover, in both of these cases she speculates the presence of mechanisms, which guarantee the integration of these two entities and, also, their subsequent dissociation, resulting in the formation of a new individual. An entity formed of two different gametes is what we call a “zygote,” whereas the entity that results from the merger of two symbionts is what Margulis refers to as “holobiont”, which she recognises as a new individual (1990: 676, Fig. 3). Margulis does not, however, specify which “bionts” should be regarded as part of the holobiont, nor does she explicitly define the term in the paper.

One year later, in “Symbiogenesis and symbioticism”, the first chapter of a book she edited with René Fester, Margulis defines the holobiont as a “symbiont composed of recognizable bionts”, and she defines symbiosis as the physical contact between organisms of different species occurring “throughout a significant proportion of the life history” (1991: 2, Table 1). Again, she does not explicitly specify which bionts should be included in the holobiont. If one follows her definition of life history strictly –“events throughout the development of an individual organism correlating environment with changes in external morphology, formation of propagules, and other observable aspects” (1991: 2, Table 1)– it might be argued that the holobiont would encompass all the bionts that share their lifetime together, irrespective of whether they are inherited or not. Clearly, this conception of the holobiont would be incoherent with the concept she had put forward in her previous (1990), where she seemed to suggest that the holobiont should exclusively include the cases of hereditary symbiosis, in her analogy between symbiogenesis and embryogenesis. This second formulation is reasonable if one takes into account the purpose of the chapter, namely, to vindicate the proposition of symbiogenesis as a way in which new species, kingdoms and taxa could evolve –for instance, she says that “the highest level taxa (…) have evolved by acquisitions of symbionts that have become hereditary” (1991: 11, emphasis added)–. This formulation is also coherent with claims she made in her later writings (Margulis and Fester 1991; Margulis 1998, 2010; Margulis & Sagan 2001, Margulis and Sagan 2002; and also see O’Malley 2017). For instance, in one of her latest paper, where she justifies the historical role of Kozo-Polyansky in introducing the idea of symbiogenesis to biology, she argues for the necessity of genetically distinct bionts reproducing together in order for symbiogenesis to occur. Analysing the association between eels and a specific species of shrimp (cleaning symbiosis), she argues:

In this vein, one might argue that, as “holobiont” was introduced in comparison to meiotic reproduction, and Margulis discusses it while reflecting the importance of symbiogenesis as an evolutionary mechanism (and evolution requires inheritance), the holobiont is thus the biological individual that includes all those symbionts that are inherited together (organelles in eukaryotes, obligatory endosymbionts in insects, etc.) (O’Malley 2017: 36, for a defence of this interpretation).

This interpretation of Margulis’ understanding of holobionts is not without contestation, though. In the same volume where Margulis published her paper, Maynard-Smith suggests “a Darwinian view of symbiosis” (Maynard-Smith 1991). There, he relates the problem of symbiosis to the problem of the units of selection¹⁸ and embeds it in the framework of the theory of evolutionary transitions in individuality that he was starting to develop. According to Maynard-Smith, symbiosis can be understood as an evolutionary mechanism and interpreted in a Darwinian fashion (i.e. with the entities that interact symbiotically being a unit of selection) only if the entities that interact symbiotically are transmitted directly, because “[w]ith direct transmission the genes of the symbionts will leave descendants only to the extent that the host survives and reproduces” (1991: 35). Therefore, as far as the two bionts have their fitness interests aligned, it is expected that those symbionts will tend to maintain a mutualistic relation that, eventually, might make it “reasonable to consider the association as a single unit” (1991: 38). However, in cases of indirect transmission, this possibility is much less likely, and he suggests that the interacting entities should be considered as independent units (of selection).

Maynard-Smith’s paper is relevant because he seems to be discussing Margulis’ liberal views about the power of symbiosis. For him, those cases where symbiosis might be considered to have evolutionary power, in the sense of affecting the role of natural selection, are very limited, and probably precluded only to cases such as cellular organelles, as he suggests at the end of his paper. If this is so, then Margulis’ notion of the holobiont might be interpreted not as constrained exclusively to the cases of the eukaryotic cell, but as including the associations of many different bionts. In fact, this view is endorsed in Guerrero et al. (2013), published two years after Margulis’ death. In that paper, holobionts, considered as autopoietic (self-sustaining) units, are defined as “integrated biont organisms, i.e., animals or plants, with all of their associated microbiota” (2013: 133, emphasis added). In the same place, they also coined the term “holobiome”, referring to “the assembly of genetic information contributed by the animal or plant and its associated microbiota” (2013: 134), and demanding a new look at evolution that would take into account the importance of the host genome plus the genome of its microbiota. They argued this to be a new entity, whose basic interacting elements that would give rise to new species and, in general, new biological variety. At some point of the paper, the authors even endorse the theses that: (1) holobionts are subjected to natural selection; and (2) holobiomes are entities that have been selected due to their selective advantages. Even if the authors do not mention the concept "units of selection", their paper might be interpreted as endorsing the hologenome concept of evolution, thus considering the holobiont, with its hologenome (holobiome), as a possible unit of selection in evolution.

Whether Margulis’ concept of the holobiont has to be interpreted as encompassing only hereditary symbiosis or, on the contrary, encompassing the whole collection of symbionts, and whether she was claiming that holobionts are units of selection or not, it seems clear that her conceptual heritage in the field of symbiosis is very important. She was one of the most vigorous defenders of the role of symbiosis for causing novelty in evolution (Margulis 1998; Margulis and Sagan 2002). Moreover, she coined the notion of the “holobiont”, which is one of the most discussed concepts in philosophy of biology at present. In the next section, I analyse the recent usage of the notion of the holobiont, as well as the criticisms that have been raised against it.

2.2 The hologenome concept of evolution and its critics: a review of current debates

The hologenome concept of evolution¹⁹ was originally proposed by Eugene Rosenberg and collaborators (Rosenberg et al. 2007), in their review paper: “The role of microorganisms in coral health, disease, and evolution”, as a generalization of the coral probiotic hypothesis (see Reshef et al. 2006).²⁰ Drawing upon their observations on coral disease, the authors suggested the existence of: “a dynamic relationship (…) between symbiotic microorganisms and corals at different environmental conditions that selects for the most advantageous coral holobiont in the context of the prevailing conditions. By altering the structure of its resident microbial community, the holobiont can adapt to changing environmental conditions more rapidly and with greater versatility than a process that is dependent on genetic mutation and selection of the coral host” (2007: 360).

Moreover, reasoning from the existence of this dynamic relation between the coral host and its microbiota, as well as the knowledge that the possibility such a relation offers for the adaptive evolution of a coral to changing environmental conditions, the authors inferred that the coral holobiont must be a unit of selection, i.e. that it is subjected to the process of evolution by natural selection. Drawing upon the observation that, as it happens in corals, all animals and plants harbour an abundant number of symbiotic microorganisms in their bodies, the authors suggest that we generalise the coral probiotic hypothesis to include every animal and plant. Thus, Reshef et al. proposed the hologenome concept of evolution, the notion that “the holobiont with its hologenome should be considered as the unit of natural selection in evolution, and microbial symbionts have an important role in adaptation and evolution in higher organisms” (2007: 360, Box 2).

Nevertheless, in the original paper, the authors do not specify: the meaning of "holobiont", the meaning of "hologenome", or how their hypothesis could be applied to other model organisms. Instead they briefly justify its appeal on four grounds: first, the universality of symbiosis between animals/plants and microorganisms; second, the existence of phenotypic variance between host species and their microbiota, i.e. the fact that hosts of the same species harbour different microbiotas; third, the different range of effects of the microorganisms on their hosts (parasitism, mutualism, commensalism); and fourth, the possible mechanisms of change for the holobiont (including microbial amplification, microbial acquisition, etc.) (Rosenberg et al. 2007: 360, Box 2). However, the authors acknowledged that their reasons were insufficient to support their generalization of the coral probiotic hypothesis. To overcome this difficulty Zilber-Rosenberg and Rosenberg (2008) would publish “Role of microorganisms in the evolution of animals and plants: the hologenome theory of evolution” one year later. Beginning with the acknowledgment that microorganisms have been discovered to play a fundamental role in the life of higher organisms (animals, plants), including humans, the authors introduced their hypothesis with a rhetorical question:. “[i]f microbial symbionts play such an important role in the lives of their eukaryotic hosts, why should they not also play a role in the evolution of these higher organisms?” (2008: 723, emphasis added). Zilber-Resenberg and Rosenberg hypothesized that holobionts (i.e. biological entities composed by a host plus all its microbial symbionts), with their hologenomes (i.e. the sum of all the genetic information of the host plus the genetic information of its symbionts) are units of selection. More specifically, concerning the notion of the holobiont, they explicated that:

This last point is particularly relevant because it frames the hologenome concept in a very distinctive way. It is not just that very particular host-microbe associations should be considered as units of selection (e.g. the eukaryotic cell, aphids and Buchnera aphidicola, squids and Vibrio fischeri, etc.). This last proposal would not be so revolutionary, after all. The hologenome concept suggests that one should consider the host, with all its microbes (i.e. the holobiont), as a unit of selection in evolution. Notice that this definition of the holobiont might be contrasted with Margulis’ understanding, which seemed to be limited to cases of hereditary symbiosis, at least according to some interpretaters (e.g. O’Malley 2017). What is the justification that Zilber-Rosenberg and Rosenberg believe to have found for their hypothesis? They claim the existence of four sources of evidence: the observation that all higher organisms associate with microorganisms; the fact that symbionts are reliably transmitted intergenerationally; the fact that symbionts affect the fitness of the holobiont; and, finally, the possibility of generating genotypic variation within the holobiont by changing their microbial composition.

It must be noted that the way in which Rosenberg and Zilber-Rosenberg present the hologenome concept is based on a particular interpretation of the units of selection, according to which two types of questions should be distinguished: first, the question about the interactor, or vehicle, the entity that interacts with the environment as a cohesive whole, in such a way that replication is differential²¹; second, the question about the replicator, the entity of which copies are made (Dawkins 1976; Hull 1980; Okasha 2006; Godfrey-Smith 2009; Lloyd 2017a). For Zilber-Rosenberg and Rosenberg, the holobiont would be an interactor, a cohesive physiological and metabolic entity, whereas the hologenome would be a replicator (see also Rosenberg et al. 2010; Rosenberg and Zilber-Rosenberg 2014, 2016; Author 2015; Bordenstein and Theis 2015; Shropshire and Bordenstein 2016; Theis et al. 2016).

After Rosenberg and Zilber-Rosenberg proposed their hypothesis, the notion that the holobiont with its hologenome constitutes a biological individual has been defended in different ways by different authors, some of which have interpreted it as a unit of selection. Dupré and O’Malley (2009), and John Dupré (2010, 2012) have defended the notion that the holobiont should be considered as the interactor in evolution, in so far as it is the entity responsible for the differential reproduction of the entities that compose it. The authors do not mention, however, the possibility of conceiving the hologenome as a replicator. Scott F. Gilbert, Jan Sapp and Alfred I. Tauber have suggested that we understand the holobiont as a biological individual anatomically, developmentally, immunologically, physiologically and genetically (Gilbert et al. 2012; also Gilbert et al. 2017; Roughgarden et al. 2017). Lynn Chiu and James Griesemer have separately proposed a concept of the holobiont as a developmental hybrid in which the microbes would act as scaffolds of the individuality of the host (Gilbert & Chiu 2015; Chiu and Eberl 2016; Griesemer 2016, 2017). Lisa Lloyd has suggested an understanding of the holobiont as an interactor, as a reproducer, and as a manifestor of adaptation (Lloyd 2017b; see also Griesemer 2017). Ford Doolittle and Austin Booth have proposed to conceive the hologenome as a functional replicator, i.e. as a network of genetic interaction patterns that can be instantiated across different generations of holobionts (Doolittle and Booth 2017; see also Lemanceau et al. 2017); Suárez (under review) has defended a group-selection interpretation of the holobiont, suggesting that we conceive of holobionts as intergenerationally inherited collections of traits associated to successive generations of a particular host. In so far as holobionts can be considered collections of traits, he argues that they can be conceived of as units of selection. Finally, Ehud Lamm has suggested that holobionts should be understood as “structures of evolution”: “constellation[s] of evolutionary factors and their relations […] [that] provide scientists with a common framework and terminology and [allow them] to elicit research questions and hypothesis that apply to many systems of interest” (2017: 372).

Furthermore, some evidence has been gathered in support of the hologenome hypothesis (e.g. Rosenberg and Zilber-Rosenberg 2014; Bosch & Miller 2016, for general summaries). In a pioneer study on Nasonia wasps, Robert M. Brucker and Seth R. Bordenstein have argued that hybrid lethality among different Nasonia species is caused by a disruption of the relation between their species-specific microbiomes and the host genome, which suggests that the different species represent a coevolved hologenome (2013; cf. Chandler & Turelli 2014 for a response; cf. Brucker & Bordenstein 2014).Their study has prompted an immediate interest in the study of the phenomenon of phylosymbiosis, “the eco-evolutionary pattern, whereby the ecological relatedness of host-associated microbial communities parallels the phylogeny of related host species” (Brooks et al. 2016: 1). Convergent host-microbe phylogenies that support the existence of phylosymbiosis have been found in hominids (Ochman et al. 2010; Moeller et al. 2016). Julia K. Goodrich and collaborators have found some evidence that suggests that the microbiome might be heritable and its composition could be partially determined by the host genome (Goodrich et al. 2014, 2016, 2017; see also Turpin et al. 2016). Finally, Thomas W. Cullen and collaborators have found some evidence that might suggest that the host’s immune system might control microbiota acquisition (Cullen et al. 2015). However, some evidence has also been found that suggests that there are no such tight host-microbiome intergenerational associations. For instance, Eric R. Hester and collaborators have not found evidence that supports inheritance of the microbiome among corals. Instead, they found that the microbiota that associates with a coral species are selected according to functional criteria, and thus there are no intergenerational phylogenetic convergences (Hester et al. 2016). The same results have been found in ruminal ecosystems: even if the hosts of the same species might share a functionally similar microbiota, the specific microbial taxa that they associate with are different. The authors explained the occurrence of this phenomenon with a metaphor: “the players might change but the game remains” (Taxis et al. 2015; Doolittle and Booth 2017 base their account of the holobiont on these results).

The hologenome concept, however, has also been contested by many, who propose that: (1) the holobiont is a sufficiently coherent biological entity for it to be considered an evolutionary interactor (Booth 2014; Queller and Strassmann 2016; Skillings 2016); there is no real empirical evidence supporting the claim that the hologenome can be a replicator or a reproducer, in so far as the fidelity of its intergenerational transmission is very low (Moran and Sloan 2015; Godfrey-Smith 2015; Stencel 2016; Douglas and Werren 2016; Hester et al. 2016; Hurst 2017; Stencel & Wloch-Salamon under review). Detractors of the holobiont concept tend to emphasize the lack of shared interests and unifying mechanisms between the entities that compose holobionts; and, on this basis, they are reluctant to accept the notion that holobionts are units of selection in any of the aforementioned senses.

The claim that holobionts are interactors has been recently disputed by Austin Booth who, emphasizing the fact that the different entities that compose a holobiont can reproduce independently, has argued that “the interactor perspective on holobionts, as currently endorsed, suffers from imprecision. More needs to be said about just what kinds of causal interactions among parts serve to bind independently reproducing populations into interactors” (2014: 670). This notion has also been criticized by David C. Queller and Joan E. Strassmann, who argued that holobiont defenders make an illegitimate inference from physical proximity (symbionts living together) to functional integration (symbionts constituting an interactor): “The holobiont is defined by spatial criteria. There is no reason to believe that spatial proximity necessarily leads to functional integration” (2016: 869). And also Derek J. Skillings has criticized this notion on the basis that the entities that compose the microbiome of a holobiont might change during the host’s lifetime. If this is so, he argues, then there are no criteria of identity to recognize a holobiont as a biological individual (sensu organism or interactor), because the microbial species that compose it are constantly and fluidly changing (Skillings 2016).

In relation to the claim that holobionts are replicators, Angela E. Douglas and John H. Werren have rejected the possibility on the basis that holobionts lack the proper type of intergenerational inheritance (Douglas and Werren 2016). For them, the holobiont can be considered a unit of selection if and only if there is sufficient partner fidelity –“stable association of host and symbiont genotypes across multiple generations” (2016: 2) – among the different species that constitute the holobiont. Otherwise, the entities that compose the holobiont would not have their fitness interests aligned; and thus, selection at the level of the holobiont would be disrupted by selection at lower levels. They concede that very specific and tight host-symbiont associations, under very special circumstances, may qualify as units of selection. However, they are sceptical that the same might be said about all the members of the microbiota: “We do not argue that selection cannot act on the host-microbiome as a unit. We simply argue that the evidence for this is weak, and the conditions necessary for it to occur are unlikely” (Douglas and Werren 2016: 5; see also Moran and Sloan 2015; Hurst 2017). Suárez (under review) has offered a specific reply to this criticism, arguing that their requirement of partner fidelity is unreasonable, since it relies on some assumptions about biological individuality that are disputable (the cooperation/conflict concept of biological individuality). Furthermore, he argues that the same type of assumptions are not applied to other levels of the biological hierarchy (e.g. transitions in evolutionary individuality), which creates a disparity of criteria. Finally, Peter Godfrey-Smith has also criticized the notion that holobionts are reproducers on the grounds of his concept of Darwinian populations (2015). He believes that host-microbe associations can only qualify as units of selection in the situations when the host is able to “kidnap” the reproduction of the microbe, i.e. when host and microbe can only reproduce together as a unit, but not independently from each other, since otherwise the system would be disrupted. He claims this to be true of eukaryotic cells generally. Godfrey-Smith also acknowledges the existence of intermediate reproductive stages (i.e. reproduction partially kidnapped, but with a high degree of independence). In any case, he does not believe that there is any evidence to qualify the holobiont, conceived as the host plus all its microbes, as a unit of selection, because the parts can still reproduce independently of the whole and thus will not have the same interests.

The debates between defenders of the hologenome concept of evolution and its detractors reflect diverging conceptions of biological individuality. Defenders of the hologenome concept tend to emphasize the collaborative nature of life, as well as the importance of symbiotic associations for maintaining life as we know it. They seem to share a commitment to a view of biological individuality according to which the existence of conflicts amongst the parts of a system does not rule out the possibility of the system evolving as a unit. Furthermore, they concentrate on studying symbiosis as an independent phenomenon, and try to understand the evolution of symbiotic relationships by partially abstracting away from the organisms that engage in symbiosis. Detractors of the concept, on the other hand, tend to emphasize the impossibility of having a biological individual if the parts of the systems are in conflict with one another, thus rejecting any claim about the individuality of holobionts. They are prone to consider holobionts as mere ecological communities of independent organisms that are together due to environmental convenience, not due to shared evolution. They put more emphasis on the study of the different species that engage in the symbiosis that in the study of the evolution of the symbiotic relationship itself. More research is needed to determine the empirical consequences of the hologenome concept of evolution, as well as to unravel the empirical consequences of the different conceptual assumptions made by defenders and detractors of the notion. Research on the historical roots on some of the recent debates will help to determine the origins of some of the present assumptions in current debates, as well as help with clarifying different issues raised by the hologenome concept of evolution, some of which were already present in the debates of prior literature.

2.3 The historical roots of the hologenome concept of evolution

Most of the debates about the hologenome concept of evolution explored in the previous section parallel some of the debates about symbiosis explored in Part I. I will explore four parallelisms between them, uncovering the similarities between recent research and the research conducted in the nineteenth century. Finally, I will explore the novelties introduced by the hologenome concept of evolution, exploring its differences to previous research.