The holobiont is not a useful model for most host-microbiome interactions

Comment #312 Gavin M Douglas @ 2026-06-16 19:13

Thank you for your detailed comment, Dr. Gilbert. I [Drew] have found your previous work on this and other topics inspiring and remarkably multidisciplinary, and I [Gavin] also appreciate it, although I am less-well versed in your broader work. We very much appreciate your engagement with our manuscript, which is currently under review at Proceedings B. We will take your arguments into consideration as we revise the manuscript. If we are published there we encourage you and any colleagues to publish a reply to our work. We would be happy to keep you informed on the status of our manuscript if you are interested.

We agree with your points about the importance of the microbiome for host physiology, development, and even the potential for important evolutionary dynamics. The point we are focusing on is whether it’s useful to reify the host and microbiome as an entity (or as you say, an organism), denoted by the term holobiont. In other words, we are enthusiastic about further exploration of the biological examples you raised and supportive of highlighting the importance of how microbes impact and interact with hosts. However, we see the holobiont concept as an oversimplification that is misleading and leads to conceptual confusion for practicing researchers (and also likely to the public, although that is harder to track).

The history of the holobiont concept certainly deserves to be explored and discussed, but that is not the focus of our article; instead, we focus on how the concept is being used today. The examples of obligate mutualists would be a much more defensible usage of the term holobiont, as we mention in our manuscript (but this raises the question of how it specifically differs from obligate symbiosis more generally). However, the term now generally encompasses all microbiota with a host, including microbes that do not affect host phenotypes. These clearly are at odds with your examples, yet would be considered as part of the same holobiont entity. Similarly, we do not find your arguments for why the holobiont boundary should be limited to a host and microbiota convincing. Cross-species interactions that have major impacts on physiology and development occur between non-microbes as well, as we note, including obligate interactions for the continued propagation of a species.

Our main focus in the manuscript is on what could be called the “broad” holobiont definition, which includes neutral microbes. We think this is the least defensible version and leads to clear conceptual confusion. As we discuss in our manuscript, others have argued for different conceptions of the holobiont, which could include the “evolutionary holobiont” or “developmental holobiont”, which might correspond to different subsets of microbes in each case, and would be approaching something more similar to the metaorganism concept. This would be an improvement, but we still do not think that this approach is desirable. In addition to the issue of interest-relativity that we raise in our manuscript (if different interests pick out compositionally different metaorganisms, there seems little reason to refer to *a* metaorganism at all), it will be very easy for researchers to still use these terms ambiguously and for intended meanings to be unclear. Terminological pluralism of this kind seems more appropriate for embedded terminology that would be hard to remove (like “gene” or “species”). More generally, if one needs to do a lot of work to define in what precise sense the term holobiont is being used, it raises the question why we are using the term at all. As Nancy Moran said in a previous interview (https://doi.org/10.1073/pnas.1908139116): “Being able to say what a term means in one short sentence, not a long paper—that’s kind of the test of whether a word is useful”.

It was not our intention to imply that developmental biology is not relevant to understanding host-microbe interactions, and it is unfortunate that previous critiques of the holobiont idea focused largely on the “unit of selection” issue. We see development as another important area where microbes are phenotypically important for hosts, which can include obligate mutualistic interactions. This is beyond the scope of our preprint, but, more generally, we at least do not see any incompatibility between population genetics and developmental biology: they simply have different focuses. It is regrettable that developmental biology was underappreciated by major evolutionary biologists throughout the 20th century, but regardless, there is a distinction between treating mechanisms as an unknown black-box to be filled in and treating mechanisms as irrelevant. For instance, it is hugely helpful to know why certain mutations are strongly deleterious (if they negatively impact development for instance), but this is entirely compatible with the model of a distribution of mutational fitness effects (caused by developmental constraints among other factors).

This above point should not be misinterpreted as us feeling the need to forcefully debate what biological concepts are the purview of which biological discipline (e.g., developmental biology or microbial ecology). Any sufficiently complex biological system will have some components that are more relevant to one discipline than another: we do not think the holobiont concept is unique in this regard. Although it is often practically required, we idealize systems when we treat them from the perspective of one discipline only. Our guiding approach in the manuscript is to evaluate the utility of concepts on the basis of what they causally track and what unique predictions they make, rather than to what biology subdisciplines they relate.

In addition, we did not intend to imply that some members of the microbiome cannot be vertically transmitted, coevolve, and even be part of multi-level selection with the host. The key question we focus on is whether treating all members of the microbiome and the host as a single entity, the holobiont, is helpful. We believe this framing oversimplifies the complexity of individual microbe-host interactions. This is particularly an issue when hosts with complex microbiomes that include transiently present and environmentally acquired microbes are treated as a unit of selection. We are highly supportive of work that investigates previously unappreciated species interactions and multi-level selection. However, treating all microbes as a single pool, rather than modelling them separately, or not including important non-microbial organisms besides the host, would, we think, be misleading.

We suspect many proponents of the holobiont concept would counter-argue that they do not intend for the concept to mean that all members are modelled as a pool, and that they recognize each species can be modelled separately. However, we worry that this simplistic framing is precisely what is implied by the holobiont term. As we argue in our manuscript, it will be very difficult to avoid researchers confusing that a reified holobiont entity is not actually what is intended to be referred to, when such language is used.

Although this is beyond the current manuscript, we believe that we need to better recognize that the biological entities we typically name, such as everyday organisms, are relationally constituted, and for exactly the sorts of reasons that you discuss. But we don’t think it is helpful to reify a new entity, because this again just raises further questions about boundary drawing. More generally, we think it’s important to compare the relative utility of biological concepts from a pragmatic perspective. Even if the term “individual organism” is imperfect, it nonetheless provides clear predictions and tracks causal patterns in nature, whereas we believe our points regarding the broad holobiont concept make it clear that the same cannot be said for this concept.

Comment #311 Scott F. Gilbert @ 2026-06-12 08:25

The holobiont is alive and well and living everywhere

Scott F. Gilbert
Howard A. Schneiderman Professor, emeritus, Swarthmore College
Finland Distinguished Professor, emeritus, University of Helsinki
Visiting Professor, University of Oulu

The holobiont concept is alive, well, and flourishing. Despite the claims of Douglas and Inkpen (and bolstered by their first figure), the holobiont concept is both needed and useful. This is because the holobiont concept came from physiology and anatomy, not evolutionary theory. The term was coined at least four times, because there was a need to name the concept that most organisms were teams, consortia of many organisms and genomes working together as a unit. This need was obvious in the corals studied by Rosenberg and Zilber-Rosenberg. These cnidarians become functional coral when they acquired algal symbionts, and these symbionts had a special placement within the coral body. The algal symbiosis supplied carbon resources to the coral, thereby allowing the coral to survive in nutrient-deficient (but sun-lit) waters. When the symbionts leave, the coral dies. Similarly, cattle have no genes in their genomes that generate enzymes that can digest plant cell walls. But cattle are obligate herbivores. That digestion is performed by their symbiotic microbes. There is no such animal as the non-holobiont cow. Similarly, termites can't digest wood until they acquire their gut microbes. These are obvious cases of different species working together as physiological teams.

I've been a proponent of seeing development and community ecology working together to form the holobiont organism. Indeed, I call the holobiont organism both an organism and a biome--a collection of ecosystems--and I contend that island biogeography extends to the seeding of each host by its symbiotic microbes. But this doesn't mean that the holobiont is a proper subset of community ecology any more than it would make the holobiont a subset of developmental biology. One of the chief differences between a holobiont and an ecological community is that the proximate causes of the holobiont involve the symbiont's regulation of developmental and immunological pathways that ensure the continuity of the organism during its life cycle.

One of the best examples of this microbes acting with developmental pathway to construct the organism is the construction and maintenance of the mammalian gut by Bacteroides and some other symbiotic microbes. First, the microbes interact with the nascent murine gut to establish its polarity. (In the calf, the gut microbes even help convert part of the gut into the rumen that will house the symbiotic ecosystem). Second, the gut microbes help induce the formation of the gut stem cells. This allows the continued presence of a healthy gut in the adult. Third, one of the cells made by the gut stem cells is the intestinal Paneth cell. In the Paneth cells, gut microbes (especially Bacteroides thetaiotamicron) induce the activation of the gene encoding angiogenin-4. When this protein is secreted, it induces the mesodermal cells around the gut to form the intestinal blood vessels. Thus, bacteria help form the gut anatomy by inducing gene expression in the mammalian cells.

Fourth, this bacterial induced protein made by the mouse Paneth cells, Angiogenin-4, helps organize the gut ecosystem. It is a bactericidal factor against Listeria and Salmonella, major competitors of Bacteroides and important gut pathogens. And, fifth, there is reciprocity: the bile of the intestine induces Bacteroides thetaiotamicron to secrete enzymes that help form biofilms, the mats of bacteria that can stick to the host epithelial cells. The gut microbes also metabolizes tryptophan to begin its conversion into serotonin. This hormone is used to mature the enteric neurons that push food from the mouth to the anus. In addition, the gut microbes are responsible for inducing the lymphocytes that constitute the gut-associated lymphoid tissue that is crucial for organizing immune responses to food and other ingested material. Symbionts are involved in the actual construction of the mammal. To know how they do this involves knowing how they operate on developmental pathways, not just community ecology.

The symbiotic gut microbes also help linking generations together. And they do this by altering development and immunity. The microbiome of male mice is needed for proper sperm production, and the microbiome of female mice is required for the normal immunosuppression of the placental and uterine immune systems that would otherwise attack the fetus. Gut microbes metabolize tryptophan into indole derivatives that induce the development of suppressor lymphocytes that migrate to the placenta and uterus. Without the gut microbiome, the embryos are killed by the maternal immune system. Here, symbiotic bacteria are performing a function needed for the continuation of the species by manipulating the immune system (which is often seen as killing off foreign microbes.) Understanding the formation of the organism demand not only community ecology, but also developmental biology and immunology. The organism is a holobiont.

The above evidence shows that the holobiont condition is not merely a subset of community ecology. The holobiont idea does involve the population dynamics of community ecology, but its proximate causes are not the same. They involve immunology, physiology, and developmental biology in its three major modes of cell differentiation, morphogenesis, and reproduction.

This integration of bodily physiology, immunology, and development makes the holobiont different from other ecological communities. It is obvious that anatomically, physiologically, immunologically, and developmentally an organism is a multigenomic, multispecies, organism. It deserves the name holobiont, if only to differentiate it from that mythical organism, the monogenomic individual, so beloved of classical evolutionary biology. We need a word that describes the organism as community, and holobiont works well. We must remember, though, that the holobiont concept was not proposed as a model for host-microbiome interactions; it was proposed as a model for organisms. The symbionts are not an "add-on" to the host. They participate in host physiology and the help construct the host developmentally. The host isn't a host until the symbionts have modified it to be so.

Classical evolutionary biology claimed that all evolution was a subset of population genetics. In so doing, it jettisoned developmental biology and the proximate causation that helps drive evolution. The proposal made here by Douglas and Inkpen would similarly jettison development to place the holobiont under the categories of microbial ecology.

The holobiont is an evolutionary unit. The holobionts are partly responsible for the phenotype of the organism, the variation that nature selects. And what gets selected is the holobiont "team," not the individual players. The pitcher or goalie doesn't go into the playoffs. The team does. Holobiont principles can even explain cases of pre-and post-zygotic reproductive isolation. But classical evolutionary theory does not model holobionts well, and the individuals cited in the Douglas and Inkpen paper (among others) are just starting to realistically model multigenomic organisms. Not only are there multiple genomes within the same organism, there are multiple simultaneous modes of hereditary transmission going on. In some cases, transmission is through the egg (like Wolbachia in fruit flies). In other cases (such as mammals or dung beetles), transmission of symbiotic genes occurs by infection at or near birth. In other cases (such as in the Euprymna squid and its Vibrio symbiont), what's inherited is the ability of the immune system to accept one type of bacteria and to exclude all others. Evolution has made organisms holobiont consortia, and evolutionary theory will have to conform to what an organism actually is. Holobionts add a new dimension to evolutionary theory, providing a whole new dimension and several new axes for evolutionary inheritances. Integrating microbial ecology with development, physiology, and immunology is going to be difficult but very interesting. The holobiont organism is a fact and evolution is an established theory. As Thomas Huxley wrote, "When an established theory meets an awkward, contradictory fact, it forces science into a painful but necessary evolution." Vivat holobiont!