1. Introduction
The main aim of this article is to point out that, according to the combined logic of the immunological account of physiological individuality courtesy of Thomas Pradeu (Reference Pradeu2012) and the evolutionary account of biological individuality from Ellen Clarke (Reference Clarke2012, Reference Clarke2013), all physiological individuals are evolutionary individuals. However, for Godfrey-Smith (Reference Godfrey-Smith, Bouchard and Huneman2013), Booth (Reference Booth2014), and many others,Footnote 1 the main objection to this conclusion is the holobiont, that is, an individual comprising a host and its physiologically integrated symbiotic microorganisms. This is generally taken to be a physiological individual that is not also an evolutionary individual (see figure 1).
Here I suggest that this assessment is too quick, the objection is unjustified, and the distinction between physiological and evolutionary individuality is not what many take it to be. It is worth noting, however, that I do not consider all sorts of physiological individuals hereFootnote 2 or every account of evolutionary individuality. The main aim of the article is simply to clarify the relationship between physiological and evolutionary individuality to the extent that we take Thomas Pradeu and Ellen Clarke to be our guides in these matters. This conclusion may well not follow from other accounts of physiological and evolutionary individuality.
I begin by examining Clarke’s (Reference Clarke2012, Reference Clarke2013) account of evolutionary individuality,Footnote 3 which aims to unify other accounts by proposing two basic mechanisms that determine units of natural selection. She calls them policing and demarcating mechanisms. Policing mechanisms reduce intraindividual variations in fitness, and demarcating mechanisms increase or maintain interindividual variation in fitness. In tandem, these mechanisms determine what counts as a unitary bearer of evolutionary fitness and the level at which natural selection occurs.
Policing and demarcating mechanisms are multiply realizable by different physiological systems in different kingdoms, phyla, and species. One system, proposed by Clarke (Reference Clarke2013), that is both a policing and a demarcating mechanism is the immune system. As such, any entity policed and demarcated by an immune system will be capable of undergoing natural selection. For Pradeu (Reference Pradeu2012), however, the immune system plays a different role. He argues that it is present in all living things and serves as the de facto arbiter of physiological individuality. That is, anything that interacts with the immune system and is not rejected by it counts as part of the physiological individual to which that immune system belongs.
So for Clarke, immune systems pick out evolutionary individuals, whereas for Pradeu, they pick out physiological individuals, including holobionts. If holobionts are not evolutionary individuals, as is generally supposed, then either Clarke or Pradeu must be wrong. However, I shall argue that they are both right, on the grounds that holobionts are perfectly respectable evolutionary individuals.
At first pass, it might seem wrong to count holobionts and similar entities as evolutionary individuals (though, for a dissenting view, see Gilbert et al. Reference Gilbert, Sapp and Tauber2012). Surely the selection pressures operating on the symbionts are different from those operating on their hosts; they reproduce independently and form separate lineages (Godfrey-Smith Reference Godfrey-Smith, Bouchard and Huneman2013). However, I will argue that holobionts meet Lewontin’s (Reference Lewontin1970) conditions for undergoing natural selection. Precisely because of the way the immune system polices and demarcates the whole, holobionts vary in their traits in ways that have a bearing on holobiont fitness, and this can be transmitted across generations by epigenetic means and the transmission of the immune phenotype.
2. Pluralism and biological individuality
A number of different authors (Clarke Reference Clarke2010; Godfrey-Smith Reference Godfrey-Smith, Bouchard and Huneman2013; Pradeu Reference Pradeu2016a; DiFrisco Reference DiFrisco2019; Wilson and Barker Reference Wilson, Barker, Edward and Nodelman2024) have pointed out that the term biological individual can refer to a variety of different sorts of entities, depending on which biological theory it is being used for or the aims of the biologists who use it. One particular distinction that has gained traction in the philosophical literature is the distinction between physiological and evolutionary individuals (Godfrey-Smith Reference Godfrey-Smith, Bouchard and Huneman2013; Pradeu Reference Pradeu2016a; Wilson and Barker Reference Wilson, Barker, Edward and Nodelman2024), with the term physiological individual usually being used interchangeably with organism (Pradeu Reference Pradeu2016a, Reference Pradeu2016b). In this section, I start by roughly outlining a prominent account of biological individuality in the context of evolutionary theory courtesy of Clarke (Reference Clarke2012, Reference Clarke2013, Reference Clarke2016a). I then contrast this with an equally influential account of the physiological individual according to Pradeu (Reference Pradeu2012, Reference Pradeu2016a, Reference Pradeu2016b, Reference Margulis and Fester2020). I then present Pradeu’s (Reference Pradeu2016a, Reference Pradeu2016b) argument that not all physiological individuals are evolutionary individuals (and vice versa). In particular, I focus on the case of holobionts.
2.1 Evolutionary or Darwinian individuals
In her 2010 paper, Clarke argues that “it is hard to overemphasize the importance of individuals within the modern synthesis. They are central to the inner logic of evolution by natural selection, according to which evolution occurs because of the differential survival and reproduction of individuals” (313). She also points out that there are a number of different ways biological individuals are defined and counted for the purposes of evolutionary theory. She lists thirteen different ways in which authors have proposed we might define and count biological individuals, and the most common criteria include a reproductive bottleneck, germ/soma differentiation, and spatial boundaries/contiguity.
The bottleneck view takes each individual to begin life either as a single cell or a small number of genetically homogenous cells. It also takes the individual to be the entire meiotic or mitotic product of the bottleneck stage in its life cycle. The bottleneck stage is also important in that it distinguishes parent from offspring and reproduction from growth. When a single cell divides to produce two new cells in an organism, that is usually taken to be growth of the same organism, so the number of organisms does not increase. However, if a cell peels off and starts dividing such that it produces a new organism, then that is considered reproduction, and so the number of organisms increases.
The germ/soma view takes it to be an essential property of a biological individual that there is reproductive division of labor such that some parts of the individual (somatic parts, e.g., epithelial skin cells) only carry out physiological functions that support the whole organism but are incapable of producing (by reproduction) a new organism by themselves. Germ cells, for example, sperm or ova cells, on the other hand, are those cells whose function it is to produce another organism.
The spatial boundaries/contiguity view takes biological individuals to be physically discrete and spatially localized. Usually this involves the individual being surrounded by a physical barrier, such as a skin or membrane, that isolates it from the environment and other individuals of the same or another kind.
However, Clarke (Reference Clarke2012) argues that the standard ways in which biological individuality is determined in evolutionary theories run into major difficulties when applied in the case of plants. She starts by pointing to the inadequacy of traditional criteria like germ/soma differentiation and reproductive bottlenecks in the case of individual plants. In organisms that have germ/soma differentiation, the somatic parts (such as skin cells in animals) can increase their inclusive fitness only by contributing to the success of the whole. It is in virtue of this feature that they are considered parts of that larger whole. As such, worker bees would be considered parts of a colony, and the colony would be considered the evolutionary individual on this view. However, in many plants, although there may be specific germ lines, all parts can reproduce independently. Though roses, for example, can reproduce sexually, a cutting of a stem from a rosebush can be used to produce a whole new plant. So all parts have some degree of reproductive independence, so germ/soma differentiation cannot be the grounds for the intuition that stems, for example, are parts of rosebushes.
Similarly, with respect to the reproductive bottleneck, plants that reproduce vegetatively can do so via multicellular propagules like runners or bulbs. Although some (Janzen Reference Janzen1977; Harper Reference Harper1977; Ariew and Lewontin Reference Ariew and Lewontin2004) would take vegetative reproduction to be mere growth, Dawkins (Reference Dawkins1982) points out that multicellular runners are efficient at transmitting mutations and so vegetative propagation is capable of producing the kind of heritable variation that drives evolutionary processes. Moreover, many plants (ferns being the classic example) have two single-cell stages in their life cycles—a spore that produces a sporophyte and a zygote that produces a gametophyte. Taking a bottleneck to define the boundaries of parents and offspring would imply that sporophytes and gametophytes, rather than being stages in the life cycle of a single individual, are distinct individuals, and this would create problems for the notion of parent–offspring similarity (Godfrey-Smith Reference Godfrey-Smith2009).
One response might simply be to discount plants as evolutionary individuals, but that seems like an extreme and unpalatable response. Instead, Clarke (Reference Clarke2012, 338) proposes that the standard accounts all point toward two generic mechanisms that are multiply realized in different taxa; that is, “the classical criteria achieve their success by homing in on mechanisms which constrain the hierarchical level at which selection is able to act.” That is to say, these criteria are really pointing toward general principles that succeed “in picking out the optimal unit for evolution tracking purposes in the same way: by identifying a mechanism which successfully manipulates heritable variance in fitness amongst … parts so that evolution by natural selection can only occur at one level” (342).
She calls these the principles of policing and demarcating, and these are multiply realized by different mechanisms in different species. Clarke (Reference Clarke2013, 421) defines the policing mechanism as “any mechanism that inhibits the capacity of an object to undergo within-object selection.” That is to say, policing mechanisms limit intraindividual variation, therefore reducing selection within evolutionary individuals. On the other hand, a demarcating mechanism is “any mechanism that increases or maintains the capacity of an object to undergo between-object selection” (424). Demarcating mechanisms promote, permit, or maintain interindividual variation, therefore generating selection between evolutionary individuals.
Examples of policing mechanisms include single-cell bottlenecks, germ/soma differentiation, and the immune system. For example, by having passed through the same bottleneck, the cells comprising an organism will have a high degree of genetic similarity, and this limits the amount of evolutionary selection that can occur between them. Another policing mechanism is the immune system. One function of the immune system is to identify and destroy cancerous cells, that is, cells with certain mutations that give rise to novel phenotypes. The immune system employs a number of different methods to identify cells that share the same genetic lineage (e.g., major histocompatibility complexes) and often will eliminate entities that fail to display them. The immune system thus constrains the phenotypical variation within the organism and therefore the degree of selection between its different parts.
Demarcating mechanisms, on the other hand, include things like sexual reproduction, physical barriers, and also the immune system. These mechanisms favor interindividual phenotypical variation by either creating it through sexual recombination events or maintaining it by preventing the migration of parts of one evolutionary individual into another. As Clarke (Reference Clarke2012, 340) argues, “sexual reproduction increases the capacity for populations of pigs to undergo evolution by natural selection, by increasing the extent to which those populations exhibit genetic variance.”
Similarly, “boundaries or barriers around a collection of parts can help to keep within-boundary variance lower than across boundary variance” (Clarke Reference Clarke2012, 340). Many of these barriers are also regulated by the immune system. Mucous membranes and interfaces with the environment are usually particularly rich in immune cells (Murphy et al. Reference Murphy, Weaver and Berg2022). These immune populations interact with the environment and will accept or reject entities with which they come into contact, depending on whether they are considered harmful or benign (Matzinger Reference Matzinger1994). So the immune system also has an important role in maintaining interindividual differences in fitness.
According to Clarke, biological individuals are units that undergo natural selection, that is, evolutionary individuals, to the extent that they possess both policing and demarcating mechanisms. Although, in principle, the more policing and demarcating mechanisms one possesses, the more of an evolutionary individual one is, only one of each will suffice to allow an individual to undergo natural selection. Clarke (Reference Clarke2013, 425) clearly recognizes that the immune system is both a policing and a demarcating mechanism, pointing out that “immunity can function as a policing mechanism, such as when the vertebrate ‘adaptive’ immune system polices the organism by eliminating mutant cells. In addition to this, there are clear cases in which immunity plays a demarcating role,” but she seems not to note its significance. Given that an immune system can serve as both a policing and a demarcating mechanism, any entity individuated by an immune system is capable of undergoing natural selection according to Clarke’s account. That means that according to Clarke (Reference Clarke2012, Reference Clarke2013), immunological individuals are, at least minimally, evolutionary individuals.
Now it is worth emphasizing that Clarke (Reference Clarke2013) takes individuality to be a property that objects might possess to a greater or lesser degree; that is, it is a continuous rather than a discrete variable. This leads her to a pragmatic thesis that “a population-biological model that omits objects with a weak capacity for participating in a selection process will make smaller errors than a model that omits objects with a stronger capacity” (429). It is then left open how one might quantify the capacity of objects with both policing and demarcating mechanisms to undergo natural selection. So, in section 3, I argue that the immune system is a sufficient policer and demarcator such that holobionts do seem to have the capacity to undergo natural selection and are plausible candidates for evolutionary individuals. However, I make no specific claim about the degree of evolutionary individuality they possess.
2.2 Physiological or immunological individuals
It is now widely supposed that not all biological individuals are evolutionary individuals in Clarke’s sense (Godfrey-Smith Reference Godfrey-Smith, Bouchard and Huneman2013; DiFrisco Reference DiFrisco2019; Wilson and Barker Reference Wilson, Barker, Edward and Nodelman2024). Pradeu (Reference Pradeu2016a) argues that there is at least one other kind of biological individual, the physiological individual or organism, and that physiological individuals are not always evolutionary individuals because some of them are holobionts. Godfrey-Smith (Reference Godfrey-Smith, Bouchard and Huneman2013) in particular cites the case of the Hawaiian bobtail squid and its symbiotic vibrio bacteria as a case of a holobiont that is not an evolutionary individual (see figure 1).
Figure 1. Physiological individuals and evolutionary individuals form overlapping sets, where some individuals are both physiological individuals and evolutionary individuals, such as fruit flies and aphids with their Buchnera symbionts (aphids holobionts), while some individuals are evolutionary individuals but not physiological individuals, such as chromosomes, and some individuals are physiological individuals but not evolutionary individuals, such as squid, with their symbiotic Vibrio species (squid holobionts). (Color online).
However, it is a long-standing objection among philosophers of biology that physiological accounts of biological individuality are too vague to settle substantial disputes (Hull Reference Hull, Evelyn and Elisabeth1992). In response, Thomas Pradeu (Reference Pradeu2012, Reference Pradeu2016b, Reference Pradeu2020) argues that the immune system can provide a precise and substantive account of biological individuality because it is the immune system that determines the constitution of an organism,Footnote 4 that is, what is and is not a part of it. Pradeu’s (Reference Pradeu2012, 240) argument is that
immunology strives to offer a criterion of immunogenicity, which is itself a criterion of individuality…. The immune system, with its surveillance activity, determines what is accepted or rejected by the organism. A criterion of immunogenicity thus constitutes a criterion of inclusion: the distinction between entities that are interconnected and form a whole as constituents of the organism and those that are rejected is carried out by the immune system.
Part of the appeal of the immunological account of biological individuality is that it relies on a very simple and widely accepted premise—everything that interacts with an immune system and is not rejected by it is part of the organism to which that immune system belongs. Together with a theory of immunology, that is, an account of how the immune system interacts with entities and determines what is tolerated and what is rejected, Pradeu constructs a precise and substantive physiological theory of biological individuality. He also argues that “all living things have an immune system, including prokaryotes, plants, invertebrates and vertebrates, so an immunity-based account of biological individuality applies to the whole living world” (Pradeu Reference Pradeu2016b, 804).
One significant upshot of Pradeu’s theory is that if anything that interacts with the immune system and is not rejected by it counts as part of the physiological individual, and the immune system interacts with and accepts the various symbiotic bacteria in the gut, skin, and respiratory and urogenital tracts, then those bacteria count as part of the physiological individual. This is a feature of immune interactions in physiological individuals that Pradeu (Reference Pradeu2012, Reference Pradeu2016a, Reference Pradeu2016b) is keen to point out. The immune systems of most, if not all, physiological individuals interact with a host of other living things, such as archaea, protists, bacteria, and fungi, but accept rather than reject them. Some of these regularly interact with the rich and active immune system in the lining of the gut and are accepted there, while others are rejected.
If we then accept the premise that anything that interacts with the immune system but is not rejected by it is part of the physiological individual (or organism) to which the immune system belongs, we should accept that these bacteria are part of the physiological individual and that therefore this individual is composed of a much more genetically diverse set of constituent parts than previously thought. As Pradeu (Reference Pradeu2016b, 784) argues, “the physiological individual, immunologically, is the unit made of the association of a host and many microbes…. If this view is correct, then all the criteria of the supposedly paradigmatic ‘unitary organisms’ … are problematic.”
2.3 Holobionts as physiological and evolutionary individuals
Margulis and Fester (Reference Margulis and Fester1991) is usually credited with introducing the term holobiont. Originally, this referred to a host and a single inherited symbiont (usually a microorganism). Now the term has been extended to refer to a host and a community of microorganisms that interact with the host in either mutualistic or parasitic symbiosis. Holobionts are typically thought of as physiological individuals (Pradeu Reference Pradeu2016a, Reference Pradeu2016b) because they are tightly physiologically integrated. However, as DiFrisco (Reference DiFrisco2019, 867) points out, “the component symbionts tend to have correlated mortality rates, but they are not transmitted vertically from the same source and they reproduce independently. That makes it difficult to meaningfully assign parent–offspring lineages between successive ‘generations’ of whole holobionts, putting the heritability of holobiont-level properties into question.” So, holobionts are thought not to be evolutionary individuals. In what follows, I clarify the role of heritability and lineage formation in the classical logic of evolution by natural selection to show that this view is unjustified.
Pradeu (Reference Pradeu2016b), picking up on Godfrey-Smith (Reference Godfrey-Smith2009, Reference Godfrey-Smith, Bouchard and Huneman2013), takes evolutionary individuals to necessarily be reproducing units. Furthermore, he points out that “a major result of recent biological research is precisely that very often a physiological individual is not as such a reproducing entity, but rather a local nexus of different lineages of reproducing entities. Indeed, work on symbiosis has shown that virtually all physiological individuals are multispecies units” (809). If, as Pradeu argues here, holobionts do not collectively reproduce but are groups of independently reproducing individuals, it is hard to see how they might be units of natural selection.
In some cases, as DiFrisco (Reference DiFrisco2019) and Pradeu (Reference Pradeu2016b) point out, the symbiotes that colonize a parent are not vertically or directly transmitted to their offspring. Offspring may acquire the same kinds of symbiotic microbes from the environment during the course of development, but “in the case of horizontal transmission, physiological individuals understood as host–microbe associations do not constitute lineages as associations. Rather, those associations are local concentrations of different lineages. For example, a physiologically-defined human being is the locus of one genetically ‘human’ lineage, and many microbial lineages” (Pradeu Reference Pradeu2016b, 810).
But this raises an important issue. According to Pradeu’s (Reference Pradeu2012, Reference Pradeu2016b) immune account of physiological individuality, holobionts are physiological individuals because the immune system of a host tolerates all its symbionts. Furthermore, as argued, according to Clarke’s (Reference Clarke2012, Reference Clarke2013) account of evolutionary individuality, the immune system serves as a policing and demarcating mechanism. If this is so, holobionts meet Pradeu’s (Reference Netea, Domínguez-Andrés, Barreiro, Triantafyllos Chavakis, Fuchs and Joosten2012, Reference Pradeu2016b) criteria for physiological individuality and meet Clarke’s (Reference Clarke2013, 427) definition of a biological (evolutionary) individual: all and only those objects that possess both kinds of individuating mechanisms, that is, policing and demarcating mechanisms.
So as far as Pradeu’s (Reference Netea, Domínguez-Andrés, Barreiro, Triantafyllos Chavakis, Fuchs and Joosten2012, Reference Pradeu2016b) account of physiological individuality and Clarke’s (Reference Clarke2012, Reference Clarke2013) account of evolutionary individuality are concerned, given that physiological individuals are definitively constituted according to the behavior of the immune system, and the immune system is a policing and demarcating mechanism, any such physiological individual is also an evolutionary individual, including holobionts. This is simply what follows from the logic of these two accounts. If holobionts are not evolutionary individuals, as Pradeu himself and others seem to suggest, then something has gone wrong with these accounts.
What this shows is that Pradeu’s (Reference Pradeu2012, Reference Pradeu2016b) account of physiological individuality and Clarke’s (Reference Clarke2012, Reference Clarke2013) account of evolutionary individuality, taken together, are in tension with the view that holobionts are not evolutionary individuals. To resolve this tension, in the next section, I argue that holobionts are plausible candidates for evolutionary individuals. Furthermore, given that holobionts are the main counterexample to the view that all physiological individuals are evolutionary individuals (Godfrey-Smith Reference Godfrey-Smith, Bouchard and Huneman2013; Pradeu Reference Pradeu2016a; Wilson and Barker Reference Wilson, Barker, Edward and Nodelman2024), a convincing argument that holobionts are evolutionary individuals should make us reconsider the distinction between physiological and evolutionary individuals.
3. Holobionts and evolutionary individuality revisited
Humans are multicellular organisms which seem to be paradigm evolutionary individuals (Godfrey-Smith Reference Godfrey-Smith2009). However, as Wilson and Sober (Reference Wilson and Sober1989) argued, if groups of cells can be units of natural selection in the case of multicellular organisms like humans, what is to stop other sorts of groups behaving as units undergoing evolution by natural selection as well? As Lewontin (Reference Lewontin1970) also pointed out, it is at least logically possible for evolution by natural selection to occur at any hierarchical level, so he proposed a set of necessary and sufficient conditions for evolution by natural selection that have become orthodox.
Those who advocate for natural selection occurring at multiple hierarchical levels generally try to stipulate conditions that must be met for evolution by natural selection to occur or define the properties any sort of group must possess for it to undergo such a process. The most famous formulation of conditions for evolution by natural selection was proposed by Lewontin (Reference Lewontin1970). These are that there must be a population with phenotypic variation,Footnote 5 that is, that individuals in the population have morphological differences; that their differences give rise to variation in fitness, that is, that different individuals produce more or fewer offspring in virtue of their phenotype; and finally, that this phenotypical variation is heritable such that parents pass on their phenotypical features to their offspring.
A number of refinements have been proposed (see Godfrey-Smith Reference Godfrey-Smith2007), but for the purposes of this article, what is important is that they all propose some principle of variance, differential fitness, and inheritance. More important, for these criteria to be met, we must first be able to divide up a population into individual units, each with a specific phenotype and evolutionary fitness, and we must be able to distinguish growth from reproduction. In that way, Clarke’s (Reference Clarke2012, Reference Clarke2013) and Lewontin’s (Reference Lewontin1970) accounts are complementary. To know whether Lewontin’s conditions obtain, we must first have a principled means by which to divide a population into units. Clarke (Reference Clarke2012, Reference Clarke2013) provides a method for doing that.
However, once we have divided the population by attending to policing and demarcating mechanisms, it is still a substantial question whether these population units meet Lewontin’s (Reference Lewontin1970) conditions for evolution by natural selection. For example, even if we divide a population into units based on the behavior of immune systems, anatomical boundaries, germ/soma differentiation, and such, it does not follow that these individuals will have differences in phenotype relevant to their fitness or that these differences in phenotype between individuals will be inherited by their offspring in the way required by Lewontin’s conditions. That is to say, phenotypical variation and heritability of differential fitness do not logically follow from Clarke’s (Reference Clarke2012, Reference Clarke2013) account. This is especially important when we think about holobionts because, as previously argued, they appear to meet Clarke’s criteria for counting as evolutionary individuals, but the main reason they are generally supposed not to be is that they are thought not to form appropriate parent–offspring relationships.
It is reasonably straightforward to show that holobionts meet Lewontin’s (Reference Lewontin1970) first two conditions, that is, phenotypical variation and differential fitness. In the case of humans, associations with certain microbial species over evolutionary time have led to mechanisms that actively facilitate colonization with some microbes in the gut but not others. for example, many Bacteroides species are tolerated by the immune system of the gut, but Listeria species are not. It is also worth emphasizing that this phenotypical variation occurs at multiple levels, including at the level of the holobiont.
That means that, for example, monozygotic (i.e., “identical”) twins can, and probably will, form nonidentical holobionts because the composition of their microbial communities will differ. Holobionts will only be phenotypically identical if the hosts are identical and are colonized with exactly the same proportions of exactly the same microbial species and physiologically interact with their symbionts in the same way. This is highly unlikely to occur, so phenotypical variation is practically guaranteed. Whether these phenotypical differences give rise to differences in fitness is, however, a substantial and important question, and I argue that it does.
The immune system, being the main way in which human hosts regulate their interaction with microbes, is thought to have developed its receptor morphologies partly in response to evolutionary and selection pressures (Mushegian and Medzhitov Reference Mushegian and Medzhitov2001; Nyholm and Graf Reference Nyholm and Graf2012; Devanesan Reference Devanesan2024). The ability of the immune system to delicately regulate the colonization of the host with symbiotic microbes is an active and evolved process whereby “the immune system can discriminate between pathogens and the microbiota through recognition of symbiotic bacterial molecules in a process that engenders commensal colonization” (Round et al. Reference Round, O’Connell and Mazmanian2010, 974).
These microbes also facilitate a number of other critical physiological processes in the human gut, including the production of essential vitamins. In this way, it is thought that associations with certain microbes may have relaxed selective pressures on the host to obtain foods with these vitamins and facilitated dietary transitions, and this enabled colonization of new environments (Moeller and Sanders Reference Moeller and Sanders2020). So it is clear that holobionts can show variations in phenotype depending at least in part on the kinds of bacterial symbionts that compose them, and these difference have an impact on fitness.
However, unless these variations in phenotype are also heritable, we will not have satisfied the conditions for evolution by natural selection. The most uncontroversial cases occur in species in which there is vertical transmission of obligate symbionts. In the case of aphids, Buchnera aphidicola are maternally inherited obligate symbionts. The aphids are nutritionally dependent on their Buchnera for providing essential amino acids. The bacterial symbionts are transmitted from mother to offspring by specialized cells called bacteriocytes, which ensures the continuation of a specific relationship between the descendants of a particular aphid and descendants of its symbionts (Koga et al. Reference Koga, Meng, Tsuchida and Fukatsu2012).
So, it is now generally accepted that such obligate symbiotic relationships are heritable and shape the evolution of the holobiont complex as a whole. As Bennett and Moran (Reference Bennett and Moran2015, 10170) point out,
acquiring a heritable symbiont is effectively a mutation of major effect, increasing host fitness at the population and clade level. In many, although not all, identified cases, these acquisitions have resulted in a proliferation of descendant lineages, usually comprised of species restricted to a particular dietary niche. Thus, long-term, heritable symbiosis underlies many dominant insect lifestyles and has shaped macroevolutionary and ecological patterns.
Although this is well documented in insects, and aphids in particular, what about the case with facultative (nonobligate) symbiosis and symbiosis in mammals?
One mechanism that facilitates transmission of the holobiont phenotype involving both obligate and facultative symbiosis, particularly in mammals, is our beloved immune system. Until very recently, the standard view of immunology divided the immune system into the innate and adaptive systems. The innate system is characterized by genetically encoded receptors that trigger specific immune responses without the need for prior exposure and without an augmented response given repeated exposure. The acquired system, on the other hand, is characterized by requiring exposure to an antigen to develop, and repeated exposure augments the response.Footnote 6
It is generally accepted that the receptor profile of the innate immune system is inherited and so the dispositions of the innate immune system of a parent will determine (to a significant extent) the dispositions of the innate immune system of offspring (Boraschi et al. Reference Boraschi, Elfi and Italiani2024). For example, while immune receptors in general demonstrate remarkable plasticity, natural killer cell receptors in humans are the result of convergent evolution and are thought to have coevolved with MHC-I receptors in mammals because certain combinations will lead to problems with mammalian pregnancy (Parham and Moffett Reference Parham and Moffett2013). Similarly, innate immune cells play a crucial role in regulating the colonization of the gut with microbes, and the immune system of the gut in turn does not properly develop and mature without microbial colonization (Khan et al. Reference Khan, Bai, Zha, Ullah, Ullah, Shah, Sun and Zhang2021).
However, even if the disposition of the innate immune system is inherited and the innate immune system influences the composition and therefore the phenotype of the holobiont, that does not yet entail that the phenotype of the holobiont is inherited. First, it is important to distinguish reproduction at the level of the host and reproduction at the level of the holobiont. In human holobionts, host reproduction is of the familiar sort—sexual reproduction, which produces a fertilized zygote, a unicellular bottleneck, which is gestated for nine months or so and then born. Once born, the new human host is colonized by an array of microbes, acquired primarily from its mother’s genital tract, skin, and gut but also from the environment. This then forms a new holobiont.
To show that this new holobiont is the offspring of some parent holobiont, we must show that the parent holobiont significantly (but not necessarily entirely) determines the existence and phenotype of the offspring. And to satisfy Lewontin’s (Reference Lewontin1970) criteria, the phenotype of the offspring must be similar to the parent to a degree higher than would be expected due to random chance. So far I have shown that the immune phenotype of a parent host influences the phenotype of the holobiont of which it is a part (the parent holobiont), and this immune phenotype is transmitted to offspring through the usual mechanism of sexual or asexual reproduction. In such offspring, the inherited immune phenotype will influence the constitution and phenotype of the holobiont of which it will be a part later in its development (the offspring holobiont).
However, there is one element left to demonstrate. Why should we think that the immune phenotype of the parent host, when transmitted to its offspring, should result in an offspring holobiont phenotype that is similar to the parent holobiont phenotype? As I show in what follows, this is because the interaction between the immune system and microbial communities results in covariation in parent and offspring holobiont phenotype.
Recent research, for example, has shown that what was classically thought of as innate immune systems have the ability to modulate their response to an antigen given repeated exposure in what is now being called trained immunity (Prigot-Maurice, Beltran-Bech, and Braquart-Varnier Reference Prigot-Maurice, Beltran-Bech and Braquart-Varnier2021). This is most clearly demonstrated in the gut: it is now well known that the gut immune system properly develops only in conjunction with certain bacteria (Khan et al. Reference Khan, Bai, Zha, Ullah, Ullah, Shah, Sun and Zhang2021; Boraschi et al. Reference Boraschi, Elfi and Italiani2024). Also, the composition of the gut flora required for proper maturation of the gut immune system is host specific (Chung et al. Reference Chung, Pamp, Hill, Surana, Edelman, Troy and Reading2012). That is to say, certain host species require colonization by certain microbial species in specific proportions to properly develop. So, the immune system determines the composition of gut microbial flora, which in turn influences the dispositions of the immune system in a mutually reinforcing feedback loop.
This trained immunity is thought to be the result of epigenetic reprogramming of immune cells in a way that can be inherited, possibly by gametic DNA methylation and chromatin remodeling (de Candia and Matarese Reference de Candia and Matarese2021). While this has been studied primarily in the context of immune responses to infections (Katzmarski et al. Reference Katzmarski, Jorge Domínguez-Andrés, Georgios Renieris and Delphine Le Roy2021), there is also emerging evidence that inheritance of immune traits has a bearing on immune tolerance of symbionts (Prigot-Maurice, Beltran-Bech, and Braquart-Varnier Reference Prigot-Maurice, Beltran-Bech and Braquart-Varnier2021). So both the innate and acquired immune phenotype is influenced, over time, by exposure to the antigens carried on symbiotic bacteria and fungi. And this can be passed on to offspring such that they then enter into symbiotic relationships with similar species of microbes in similar proportions.
The inheritance of immune traits ensures that the next generation enters into the same or similar symbiotic partnerships as its parents, even in the case when the symbionts are not directly transmitted from parent to offspring, such as in the case of aphids and Buchnera, but are acquired from the environment. Even in such cases of horizontal transmission of symbionts, the mechanisms by which the host immune system tolerates such symbionts are inherited (Nyholm and Graf Reference Nyholm and Graf2012), and this facilitates reproduction of the holobiont phenotype. What this demonstrates is that there is a mechanism, the transmission of immune phenotype through reproduction at the host level, that ensures that reproduction occurs at the level of the holobiont. This requires a broader view of reproduction than the one Godfrey-Smith and others appear to standardly endorse, but an increasing body of literature has put pressure on this notion (Laland et al. Reference Laland, Uller, Feldman, Sterelny, Müller, Moczek, Jablonka and Odling-Smee2015; Griesemer Reference Griesemer2016; Veigl Reference Veigl, Suárez and Stencel2022).
If the holobiont phenotype of one generation is partly determined by the holobiont phenotype of the previous generation, and these phenotypes show differences in fitness, this is sufficient to meet Lewontin’s (Reference Matzinger1970) criterion of heritability of differential fitness. As Godfrey-Smith (Reference Godfrey-Smith2007, 494) points out, “it is sufficient for [evolution by natural selection] (given other conditions) that parent and offspring be more similar than randomly chosen individuals of different generations.” This is a reasonably weak condition that appears to be adequately satisfied in the case of holobionts because of the transmission of immune phenotype. As such, given that holobionts show heritable variation in phenotype and fitness, we should be willing to grant that they are evolutionary individuals at least in a minimal but significant sense.
4. Objections
One objection one might raise at this point is that, appearances notwithstanding, unless holobionts can collectively reproduce, they cannot be evolutionary individuals. This point was raised by Godfrey-Smith (Reference Godfrey-Smith, Bouchard and Huneman2013) and is one of the main reasons that Pradeu (Reference Pradeu2016b) rejects the possibility that holobionts are evolutionary individuals, the other one being that holobionts do not form appropriate parent–offspring lineages. Godfrey-Smith (Reference Godfrey-Smith, Bouchard and Huneman2013) points to the case of the Hawaiian bobtail squid and the symbiotic vibrio bacteria that provide the squid with bioluminescence. The bobtail squid has specialized crypts that accommodate a specific bioluminescent vibrio bacteria (Vibrio fischeri). Its immune system prevents other bacteria from colonizing these crypts but tolerates Vibrio species. Every night, the bacteria light up and are thought to help camouflage the squid. At dawn, most of the bacteria are expelled into the surrounding water and are allowed to regrow from a small retained population during the day.
On the question of whether the squid–vibrio holobiont is an evolutionary individual, Godfrey-Smith (Reference Godfrey-Smith, Bouchard and Huneman2013, 29) objects that “if we accept that the [squid–vibrio] combination is an organism, then we find that the combination does not reproduce in the sense that is relevant to being a Darwinian individual. The combinations do not form parent–offspring lineages.” The reason he claims that they do not reproduce in the sense relevant to being a Darwinian individual, that is, an evolutionary individual, is because the bacteria are not passed directly from parent to offspring but are acquired from the environment. He argues further that
if you are a squid, there is no mechanism ensuring that the bacteria in you are the offspring of bacteria in your parents, or any other specific individuals. The bacteria in you might come from many sources, and some might have not been inside squid for many generations. Squid–Vibrio combinations “make more of themselves” in one sense, but not in the sense that gives rise to parent–offspring lineages. (29)
So Godfrey-Smith (Reference Godfrey-Smith, Bouchard and Huneman2013) accepts that squid–vibrio holobionts “make more of themselves” because one generation of holobionts “makes” a successive generation in the sense of being causally responsible for its existence and phenotype.Footnote 7 However, he rejects the idea that this counts as “reproduction” because “the parent–offspring lines connect only the parts—they connect bacteria with bacteria and squid with squid.” That is to say, the squid–vibrio holobiont does not seem to reproduce as a whole. Its parts reproduce independently and separately, coming together to form a new holobiont only at some later time.Footnote 8 More importantly, the squid–vibrio holobiont does not seem to form parent–offspring lineages in the way that Godfrey-Smith thinks is required of Darwinian individuals because
some of the bacteria that initiate a colony may have an ancestry that can be traced back to other colonies just a few bacterial generations back. Others may have not had ancestors inside squid–Vibrio complexes for a great many generations—perhaps ever. This is not a case where each squid–Vibrio collective has a definite and reasonably small number of parent collectives, even though each squid has exactly two parent squid and each colony-initiating bacterium has one parent bacterium. (Godfrey-Smith Reference Godfrey-Smith, Bouchard and Huneman2013, 30)
However, nothing in Lewontin’s (Reference Matzinger1970) criteria for evolution by natural selection requires that offspring derive all their parts from their parents, that offspring have any specific number of parents, or that all of an offspring’s parts from its parents arrive at the same time.Footnote 9 Bacteria are well known to exchange genetic material with other members of the same generation, and yet I cannot imagine anyone would deny that bacteria are evolutionary individuals. If bacteria can acquire parts from the environment or other bacteria and remain evolutionary individuals, then why not squid?
Strictly speaking, as far as Lewontin’s (Reference Matzinger1970) criteria are concerned, all that is required of “reproduction” is that parents be causally responsible for the existence and phenotype of their offspring.Footnote 10 And in the case of squid–vibrio holobionts, Godfrey-Smith (Reference Godfrey-Smith, Bouchard and Huneman2013) admits that they do in fact “make more of themselves” in this sense. Perhaps this “making more of” could be more precisely articulated as a case of scaffolded reproduction in the sense proposed by Griesemer (Reference Griesemer2016) and others. There is no space to fully articulate this idea here, but, insofar as we accept that there are successive generations of holobionts at all, we ought to grant that squid–vibrio and other holobionts that “make more of themselves” do so in a way that allows for evolution by natural selection.
It is also clear that holobionts stretch our understanding of lineages. Skillings (Reference Skillings2016) illustrates the point with a hypothetical case of a doctor who delivers a baby, and the baby acquires a bacterium from the doctor and is quickly colonized by it. Here Skillings argues that “we can now pick out a new parent–offspring relation between the doctor holobiont and the baby holobiont. From a lineage-neutral perspective at the holobiont level, this is no stranger than saying that the parent–offspring relation is between the mother holobiont and the baby holobiont” (883).
It may sound strange that the baby holobiont has, as its parents, the father holobiont, the mother holobiont, and the doctor holobiont, but I see no reason why this strangeness should undermine the view that the baby holobiont is an evolutionary individual. While Lewontin’s (Reference Matzinger1970) criteria require that evolutionary individuals form lineages, there is no stipulation or restriction on the number or complexity of the parent–offspring lineages any particular individual is allowed to be a part of. All that is required is that any individual has some parent or parents and is a part of some lineage—and in the case of holobionts, both of these requirements are met.
Skillings (Reference Skillings2016, 884) makes the further point that “high partner fidelity is a prerequisite for evolutionary individuality because the holobiont can only evolve as a unit if the host and its symbionts co-occur across multiple host generations.” He argues that partner fidelity is important because it aligns the fitness of the host and its associated symbionts, and without this, there is an
expectation of increased conflict between the members of the holobiont as they “pursue their own goals”; namely, selection for increased replication of one’s own lineage at the expense of the success of the multi-lineage holobiont. As conflicts of interests among partners increase (e.g., due to weak partner fidelity), then the holobiont is undermined as a higher-level unit of selection. (884)
Although this sounds reasonable, and some symbiont exchange is probably fairly commonplace, it is not obvious that this undermines holobiont evolutionary individuality in any particular case. Though the whole point of demarcating mechanisms is to minimize the extent to which different individuals exchange parts, it is not necessary for these mechanisms to prohibit any exchange whatsoever. Ultimately, what matters is whether that mechanism keeps interindividual variation in fitness higher than intraindividual variation in fitness.
So, while Skillings’s point is well taken, partner fidelity and material exchange depend on the strength and specificity of demarcating mechanisms, and the immune system is a particularly strong and specific example of one. As Clarke (Reference Clarke2016a, 907) argues in the case of herds, “the giraffe herds qualify as individuals, on this view, only if there are mechanisms enforcing the between-group variance and the within-group homogeneity.” Whether the effect of these mechanisms is enough in the case of any particular holobiont is a substantial question, but I see no reason why Skillings’s objection should undermine the view that, given sufficient policing and demarcation by the immune system, holobionts will undergo natural selection.
However, one might respond with the objection that in the case of holobionts, the immune system does not appear to be functioning as a demarcating mechanism at all. After all, by allowing microbes to colonize the host, the immune system of the host is increasing intraorganismal variation relative to interorganismal variation.Footnote 11 However, this objection arises from a misunderstanding about the different levels of immune systems in an organism or holobiont. There is no space here to fully describe the architecture of holobiont immune systems, but some basic distinctions are in order. In the same way that the human brain has a distinct immune system that is part of the immune system of a human organism, a human organism (minus its symbiotic microbes) has an immune system that is part of the immune system of a human holobiont.
That is to say, the holobiont immune system is larger than the host immune system (for a suggestion in this direction, see Schneider Reference Schneider2021). For example, in the human gut, though the composition of microbial species changes with diet and diseases, it also shows remarkable stability over time. This is thought to be due to a number of mechanisms, including interaction with the host immune system. This includes the host immune system actively tolerating or destroying certain species, as mentioned earlier, or the immune system being induced by one species of bacteria to destroy another. For example, Bacteroides species induce intestinal Paneth cells to produce angiogenin, which suppresses the growth of Listeria species (Cash et al. Reference Cash, Whitham, Behrendt and Hooper2006).
In addition, microbes show both competitive and commensal behaviors. Some species actively devour others or produce waste products that are toxic to competitors. Some species also actively foster other species by producing waste on which other species feed or by maintaining a certain pH that is conducive to some species of microbes but not others. This system results in a balanced microbial ecosystem characteristic of a healthy gut (Coyte and Rakoff-Nahoum Reference Coyte and Rakoff-Nahoum2019). So, the holobiont immune system extends beyond the immune system of the host to include the competitive and commensal behaviors of those microbes that constitute the holobiont as a whole. It is this immune system that maintains the composition of the holobiont and acts as a demarcating mechanism.
Finally, it is also worth emphasizing that both Clarke (Reference Clarke2013) and Godfrey-Smith (Reference Godfrey-Smith2009) view evolutionary individuality as a property of entities that comes in degrees. So an entity like a human holobiont might be less of an evolutionary individual than a human organism without its commensal microbes but still be an evolutionary individual nonetheless. In this article, I make no comparison of relative individuality in the case of holobionts. I argue only that they are individuals that are policed and demarcated by the immune system to a sufficient degree to allow them to undergo evolution by natural selection. Moreover, because the argument takes the immune system to generically ground evolutionary individuality, it would follow that all immunological individuals are capable of undergoing natural selection.
5. Conclusion
In this article, I started by examining Clarke’s (Reference Clarke2012, Reference Clarke2013) account of evolutionary individuality and argued that the immune system is both a policing and a demarcating mechanism. If this is so, then any entity individuated by an immune system must be an evolutionary individual. However, Pradeu (Reference Pradeu2012) argues that the immune system determines the constituent parts of a physiological individual, that is, an organism. If we take both accounts seriously, we are led to believe that all immunological individuals qua physiological individuals are evolutionary individuals.
Although this does not definitively prove that all physiological individuals, that is, organisms, are evolutionary individuals, it shows that on at least two important accounts of immunological and evolutionary individuality, this appears to be the case. I have not specifically entertained the notion of a metabolic individual and am open to the possibility that it may be a physiological individual that turns out not to be an evolutionary individual. At present, however, I know of no such cases. At a minimum, what I hope to have shown here is that the way we currently distinguish evolutionary and physiological individuality requires closer evaluation.
What this does not imply is that the concept of physiological individuality is obsolete. After all, I have not shown that all evolutionary individuals are physiological individuals, and indeed, at least at face value, they do not appear to be. Chromosomes and RNA can undergo natural selection, but they are not considered to be physiological individuals. So the concept of a physiological individual would still be important in picking out those evolutionary individuals that are individuated by physiological systems and those that are not. Physiological individuality can also serve as a different descriptive mode for certain evolutionary individuals for which the question of interest—how reproduction occurs in a species—might be better answered by appealing to physiology rather than evolutionary mechanics.
I hope that this article clarifies the nature of the relationship between these two categories of biological individuality. It also forms part of a growing body of literature that challenges traditional ideas about evolutionary individuality and invites us to critically evaluate our notions of inheritance and reproduction along the lines of an extended evolutionary synthesis (EES) (Laland et al. Reference Laland, Uller, Feldman, Sterelny, Müller, Moczek, Jablonka and Odling-Smee2015; Griesemer Reference Griesemer2016). I have not specifically mentioned EES here because doing so would be needlessly distracting and require exposition beyond the scope of the article. Instead, I claim that holobionts could be considered evolutionary individuals even according to the logic of the Modern Synthesis.
I argued that holobionts meet Lewontin’s (Reference Clarke1970) conditions for evolution by natural selection. I showed that holobionts show variations in phenotype that affect the reproductive success of the holobiont as whole. I also argued that there are a number of ways this variation can be inherited and therefore have important evolutionary consequences. As such, we should be willing to grant that holobionts are evolutionary individuals at least to a minimal but significant degree. This lends further support to the view that physiological individuals are also evolutionary individuals.
Acknowledgments
I sincerely thank David Papineau, Elselijn Kingma, and two anonymous reviewers for their encouragement and helpful feedback on previous drafts of this article.
Open access
