Fierer Lab

Exploring the structure and function of microbial communities

Do all animals need microbes?

By Tobin Hammer, Jon Sanders, & Noah Fierer

May 18th, 2018


Have you—like us—ever written a paper or grant proposal with a statement along the lines of “All animals host microbial symbionts that play critical roles in many aspects of host ecology, behavior, and development”? If so, this blog post is for you. We argue that this statement is not a fact, but a hypothesis, and one that is not actually based on firm evidence. We suggest that this “universal microbiome paradigm” should be tested through careful analyses, using methods that go beyond conventional 16S rRNA gene amplicon surveys, as well as sampling of a more diverse array of potential host taxa.


It is widely assumed that all animals are hosts to resident microbiomes. As just one example, consider the first principle of hologenome theory, as formulated in a recent review: “All animals and plants harbor abundant and diverse microbiota and are thus considered holobionts” (1). However, this statement is simply not well-supported by data. We don’t yet have robust measurements of microbial abundances for a diverse enough suite of macrobes to even get close to approximating the diversity of all animals and plants. For example, there are more species of jumping spiders than mammals, yet when was the last time you read a paper about jumping spider microbiomes? Even a more conservative statement like “most animals harbor abundant microbiota…” must reckon with the enormous abundance and diversity of groups like caterpillars and ants, many (if not most) of which have few resident microbes (2, 3).


We already have examples of animal taxa that lack a resident microbiome in certain tissues (e.g., guts of aphids (4)). Some animals lack a resident microbiome in certain life stages (e.g., larvae of honeybees (5)). Likewise, and in contrast to the implications of some claims (see this quote [i] from ref. 6), not all microbial taxa are abundant and active in a given animal. For example, fungi are only transiently present in the human gut (7). We already know that central biological functions, like digestion, can be mediated by microbes in some animals, but not in others. Is it then that much of a stretch for there to be animal taxa that simply do not harbor a resident, functional microbiome in any tissues or life stages? There is already evidence that such animals are common if nature, if not yet widely reported or recognized as such in the microbiome research community (2, 3, 8, 9).


Because microbial contaminants, transients, and parasites/pathogens are ubiquitous, any animal could be said to “have a microbiome”. Therefore, it may be more productive to instead ask, for said animal: how strongly does it depend on microbes, and which aspects of its biology (if any) are influenced by microbial activities? Although here we are emphasizing the extreme hypothetical case of an animal fully independent of microbes, the degree of microbial dependence—as might, in an ideal world, be measured by fitness reduction upon microbiome removal under a range of conditions—will vary as a continuum across animal taxa (e.g. (10)).


Below we highlight a few reasons why we think the paradigm that all animals have abundant and necessary microbiomes remains so strongly entrenched in the microbiome research community:


Contaminants. Prior to sequencing, we typically dunk our samples in a series of reagents that contain microbial DNA. Thus, you’ll always be able to sequence something in any tissue of any animal you choose. The lower the microbial biomass in your sample, the more easily it will be contaminated by Propionibacterium on your skin or Acinetobacter in your DNA extraction kit (11). This wouldn’t be a problem if all animals hosted abundant microbes, but they don’t. Often, these contaminant taxa are common in water, soil, and plants, and so they can masquerade as real symbionts, and it’s easy to come up with stories about what they might do for the animal of interest.


Transient microbes. Let’s say that the DNA you sequenced was derived from live microbes that were actually present in the sample, and not from your skin bacteria, or from lab reagents. The question then becomes: were these microbes transiently present (e.g., ingested with food and then excreted), or resident? By “resident” microbes, we mean those populations that are reasonably stable [ii] and capable of growing on or in the host (i.e., symbionts). In some cases, like the bean bug Riptortus pedestris, ingested microbes may colonize at high densities and provide beneficial functions (12), but in other cases they may simply pass through inactive, or be digested themselves.


Parasites and pathogens. Even a thriving assemblage of symbionts, consisting of taxa common among host individuals, isn’t necessarily beneficial. To use an example from another host-symbiont system, take plants and their folivorous caterpillars. In Guanacaste, Costa Rica, over 70% of caterpillars recorded from the tree Enterolobium cyclocarpum belong to a single species, Coenipeta bibitrix (data from But regardless of its core status, C. bibitrix is a bona fide parasite, not a mutualist. Likewise, in animal microbiome surveys, core microbes could simply be common parasites, or microbes commonly ingested with food or water (3, 7). This is why observational data can only take you so far—it is often necessary to conduct experiments that explicitly test the effect of a given microbe, or whole microbiome, on host fitness (keeping in mind that these effects can be context-dependent).


Literature bias. Researchers naturally gravitate to model systems, and in general, to hosts that have interesting symbioses with microbes. So right off the bat, animals lacking resident microbiomes will be neglected. And when they are studied, those data are likely to get shelved — maybe the PCRs “didn’t work” or the data looked contaminated — and never published. For example, several years ago one of us (Noah) worked on a project with other collaborators investigating the microbiome of Timema stick insects. Even after weeks of lab work and protocol tweaking, the samples failed to yield any usable microbial data—an outcome that now makes sense in light of stick insects’ lack of resident gut microbiome (8). The project was shelved and ultimately forgotten because no one wants to write or read a paper with unusable data. We were working under the assumption that all animals have abundant microbiomes, so clearly we must just have been doing something wrong. There are likely many other studies that fell into this dustbin of history. The outcome of all this is that the published literature gives an unrepresentative view of the extent of abundant and functional microbiomes across animals.


How, then, to recognize whether an animal has, and needs, resident microbes? Here we list some suggestions, with an eye to researchers studying non-model systems or animal taxa that have not yet had their microbiomes characterized:


  • View microbiome data skeptically: your animal may not work the same way as aphids, humans, cows, or bobtail squid. Be open to the diversity of ways in which animals interact with microbes. And please publish “negative” results!
  • Take precautions to reduce potential sources of contamination, like sterile sampling techniques and sequencing of your negative controls. Most importantly, keep a copy of Salter and colleagues’ paper (11) on your desk at all times; this alone would prevent a lot of mistakes from entering the literature (such as the human placental microbiome (13)).
  • Measure the absolute abundance of microbes, e.g., with qPCR, microscopy, and/or culturing (if most of the microbes are culturable under your conditions); see (2, 3) for two recent examples. This does not have to add that much extra time and expense, yet it is strangely uncommon among host-microbiome surveys. Comparisons of microbial composition and total counts between the environment (e.g., food) and the animal (e.g., guts) are especially useful in distinguishing between transient and resident microbes (3, 7, 14).
  • Do experiments to test for microbial effects on host fitness – an abundant core symbiont may actually be a garden-variety pathogen. Keep in mind that microbial function in vitro or in silico may not apply in vivo: for example, a cellulose-degrading or nitrogen-fixing microbe could be isolated from an animal (or its genes detected in a metagenome), but it may not be performing those functions—or even doing anything at all—in the animal itself.


The concept of microbial symbionts as a universal force mediating the ecology and evolution of all larger organisms—and backlash against it—is not new. Over 60 years ago, Paul Buchner, a principal founder of symbiosis research, complained that “again and again there have been authors who insist that endosymbiosis is an elementary principle of all organisms…” (p. 69, (15)). Following in the tradition of Buchner, we’d like to encourage the microbiome research community to question this paradigm and to start a productive discussion about how we define and study microbiomes across the diversity of macrobes. Yes, there have been loads of fascinating, surprising, ground-breaking discoveries made on the roles of microbiomes in animal biology. And these findings have led to an awareness that microbiomes *could* be involved in virtually any biological process of virtually any animal. But let’s hit the brakes a bit so we don’t cross over into the tacit assumption that microbial symbionts actually *are* necessary to all of animal life.



[i] “Today we realize that any multicellular organism must be considered a metaorganism comprising the macroscopic host and its synergistic interdependence with bacteria, archaea, fungi, and numerous other microbial and eukaryotic species including algal symbionts.” (6)

[ii] In more precise terms, a stably associated microbial population’s replication rate inside the host equals or exceeds the rate of loss due to cell death or excretion. We also note that macrobial community ecologists have long placed an important distinction on transient versus resident (or “persistent”, “breeding”, “self-maintaining”, etc.) populations (16).



  1. Rosenberg E, Zilber-Rosenberg I (2018) The hologenome concept of evolution after 10 years. Microbiome 6(1):78.
  2. Sanders JG, et al. (2017) Dramatic differences in gut bacterial densities correlate with diet and habitat in rainforest ants. Integr Comp Biol 57(4):705–722.
  3. Hammer TJ, Janzen DH, Hallwachs W, Jaffe SL, Fierer N (2017) Caterpillars lack a resident gut microbiome. Proc Natl Acad Sci 114(36):9641–9646.
  4. Engel P, Moran NA (2013) The gut microbiota of insects – diversity in structure and function. FEMS Microbiol Rev 37(5):699–735.
  5. Martinson VG, Moy J, Moran NA (2012) Establishment of characteristic gut bacteria during development of the honey bee worker. Appl Environ Microbiol 78(8):2830-2840.
  6. Bosch TCG, McFall-Ngai MJ (2011) Metaorganisms as the new frontier. Zoology 114(4):185–190.
  7. Auchtung TA, et al. (2018) Investigating Colonization of the Healthy Adult Gastrointestinal Tract by Fungi. mSphere 3(2):1–16.
  8. Shelomi M, Lo W-S, Kimsey LS, Kuo C-H (2013) Analysis of the gut microbiota of walking sticks (Phasmatodea). BMC Res Notes 6(1):368.
  9. Hudson AJ, Floate KD (2009) Further evidence for the absence of bacteria in horsehair worms (Nematomorpha: Gordiidae). J Parasitol 95(6):1545–1547.
  10. Fisher RM, Henry LM, Cornwallis CK, Kiers ET, West SA (2017) The evolution of host-symbiont dependence. Nat Commun 8:1–8.
  11. Salter SJ, et al. (2014) Reagent and laboratory contamination can critically impact sequence-based microbiome analyses. BMC Biol 12(1):87.
  12. Kikuchi Y, Yumoto I (2013) Efficient colonization of the bean bug Riptortus pedestris by an environmentally transmitted Burkholderia symbiont. Appl Environ Microbiol 79(6):2088–2091.
  13. Perez-Muñoz ME, et al. (2017) A critical assessment of the “sterile womb” and “in utero colonization” hypotheses: implications for research on the pioneer infant microbiome. Microbiome 5(1):48.
  14. David LA, et al. (2014) Diet rapidly and reproducibly alters the human gut microbiome. Nature 505:559–63.
  15. Buchner P (1965) Endosymbiosis of animals with plant microorganisms (John Wiley & Sons).
  16. Snell S, Evans B, White E, Hurlbert A (2017) The prevalence and impact of transient species in ecological communities. bioRxiv doi:10.1101/163816

10 thoughts on “Do all animals need microbes?

  1. I guess you can apply a lot of these ways of thinking to sites in the human with low microbial abundance, such as the lower airway of healthy individuals as well?

    On the idea of transient passengers rather than residents. I have a hypothesis that Bacillus subtilis has become a component in probiotics because of its ability to form spores and therefore being detected in gut samples. Is this just because the spores can tolerate the harsh environment in the stomach. Also B. subtilis must be pretty ubiquitous on plant material/food in general.

    • Yes, I agree — some sites within the human body have much higher microbial abundances than others, and microbes will be more important to the biological processes occurring there. I think there has been some overestimation of the ubiquity of microbes within the body, but we know that blood, the cornea (doi: 10.3389/fmicb.2018.01117), and other tissues are typically pretty sterile, in a non-disease state. Even within the human gut there are huge differences in host-derived enzymatic activity and microbial densities, e.g., between the small and large intestines, that can mirror differences between animal species.

  2. Thank you for this great post. Judging from recent pubs, I think many authors and reviewers are not aware of lots of the issues raised here. Maybe you should think about publishing this as a small opinion/review piece somewhere to make it more visible.

    The underlying problem is that people oversell their results, which is of course not only a problem in microbiome research. It should never be a valid conclusion that something is true for all organism just because one finds it to be important in the few groups that have been studied.

    • Thanks for your comment Michael! We are definitely considering submitting a version of this post to a journal, to reach people who aren’t on twitter or reading blogs. Overselling and overgeneralization are huge drivers. One should always be careful to infer general laws in biology… Also, my beef with this whole thing is not just that the paradigm is wrong, but that it blinds us to interspecific variation in microbial dependence, which is interesting and which we should try to understand.

      If you have any other suggestions for things you’d like to see included in a journal version please feel free to contact us directly!

  3. Wow, interesting post. I study the biological control in aquaculture systems using phages and bacterial probiotics, and from my perspective, one problem is the selection of microorganisms to biological control. How you said in the post, in vitro and in silico evaluations not necessarily show us the in vivo effect during not controlled conditions. In fact, I am searching papers focused on microbiome in general, and how these conceptualize microbial communities. I agree with you because I found papers that said that microbial communities are important and abundant, and the presence of one type is an indication of illness or symbiosis, but they did not reflect the “real function” of the complete community and how microorganisms interact (since the presence of one microorganism not necessary mean illness or a symbiont).

    I still looking for a paper that can guide me for future trials, considering that I have only experiments in vitro and in vivo during controlled conditions.

    Thanks a lot for your post,

  4. Thank you Hammer et al. for a great article — I’ll be sharing this with my lab group. Overgeneralisation and overselling are one of my pet peeves and I agree with Michael that the field would benefit from having an article such as this published.

    I’m glad you touched on DNA contamination, as I think it is currently a major issue (especially for low-biomass samples) and we have a paper in review at the moment about this.

    Thanks for your post!

  5. Dear Fierer Lab folks-
    I have been working on animal/microbe-interactions and symbiosis for 2+ decades, most of the time following the paradigm you question in this blog…
    …until we worked on the comb jelly Mnemiopsis – we found no evidence for bacteria or archaea on the umbrelly surface, and tried to publish this finding in high-ranking journals: without success (admittedly, partly because the study does have some ambiguities…). In the end, we proceeded with an irrelevant journal, but at least, the paper is out;-)

    Hammann S, Moss A, Zimmer M. 2015. Sterile surfaces in Mnemiopsis leidyi (Ctenophora) in bacterial suspension – a key to invasion success? Open Journal of Marine Science 5: 237.

    We don’t work on comb jellies anymore but are now back to crustaceans (with which all began): mangrove crabs and their intearctions with free-living and symbiotic bacteria…

    • Hi Martin, thank you for bringing that work to my attention. Very interesting that they can have microbe-free surfaces despite inhabiting an obviously microbe-rich medium! I’ll keep that reference on hand as an additional example of animals that can (actively) maintain basically sterile tissues. It would be interesting to know more about what’s going on in its gut. Anyway, I also look forward to reading about your work on mangrove crabs!

  6. Really thought provocative piece!! I would also like to add that another way to understand a host´s microbime is through review or meta-analysis. For example, consider serveral labs working on the microbiome of a given animal of plant, while it is true that contamination or transient microbes would be a problem for individual studies, I think it would extremely unlikely that these microbes would be consistently found across studies. One could not say what they do, but at least one gets pretty reliable evidence about the who is “core microbiome”.

    • Hi Carlos, thanks for your comment. In general I agree that independent studies on the same system help shore up the evidence quite a bit. However, some of the contaminants are surprisingly widely distributed across reagents and kits (see the Salter paper) and across human skin (Corynebacterium, Propionibacterium, Staph, etc.) — at least at the level of phylogenetic resolution that we typically get by sequencing only a short region of the 16S rRNA gene. Likewise, “core” in terms of consistent presence doesn’t necessarily indicate it’s a resident symbiont. For example, if three labs all sequenced the gut microbes of a bunch of insect herbivores, they might find a “core microbiome” of Methylobacterium, Acinetobacter and Pseudomonas. But those bacterial genera could have just been “core” among the plants themselves, and then passed transiently though the guts of all of those insects. There is a brief mention of this in the caterpillar paper, ref. #3 above. This is one reason why I am a strong proponent of sequencing the diet (and other potential environmental inputs, in some cases).

Leave a comment

Your email address will not be published. Required fields are marked *