Oral Bacteria in Komodo Dragon Mouths (Bull et al., 2010)

📖 18 min read~3249 words

In 2010, a team spanning evolutionary biology, ecology, and microbiology published the first formal mathematical treatment of a question that had haunted Komodo dragon research for decades: if pathogenic bacteria in dragon saliva are genuinely useful weapons, why do captive dragons — housed in clean enclosures and fed uncontaminated food — harbour almost none of them? Bull, Jessop, and Whiteley's answer reframed the bacterial story not as one of deliberate chemical warfare, but as an ecological cascade driven by communal feeding and escaping prey.

Quick Facts

Field	Detail
Authors	J.J. Bull, Tim S. Jessop, & Marvin Whiteley
Year	2010
Journal	PLoS ONE
Volume / Article	5(6): e11097
Published	21 June 2010
DOI	10.1371/journal.pone.0011097
Focus	Evolutionary and ecological basis for persistent oral bacteria in wild Varanus komodoensis
Key finding	A lizard-to-lizard epidemic model — spread via prey acting as vectors — best explains why dragons in the wild maintain septic oral flora that captive animals do not
Bacterial species surveyed	58 total; 93% classified as potentially pathogenic; Pasteurella multocida most prominent

Paper Overview

This page presents an original editorial summary prepared by the Komodo Guide team; it is not a reproduction of the authors' text. Readers wanting primary data should consult the open-access article directly at doi.org/10.1371/journal.pone.0011097.

The full verified citation is: Bull, J.J., Jessop, T.S., & Whiteley, M. (2010). "Deathly Drool: Evolutionary and Ecological Basis of Septic Bacteria in Komodo Dragon Mouths." PLoS ONE 5(6): e11097. doi:10.1371/journal.pone.0011097. Bull was affiliated with the Section of Integrative Biology and the Institute for Cellular and Molecular Biology at the University of Texas, Austin; Jessop with the Department of Zoology, University of Melbourne; and Whiteley with the Section of Molecular Genetics and Microbiology, also at the University of Texas.

Published just one year after Fry et al.'s landmark venom paper (see our Fry 2009 review), "Deathly Drool" arrived at a moment when the long-standing bacteria hypothesis was under serious attack. Rather than arguing for or against whether bacteria could harm prey, Bull and colleagues stepped back and asked a more tractable evolutionary question: given that wild dragons do carry abundant pathogenic bacteria while captive dragons typically do not, what ecological mechanism maintains this difference? Their answer involved a mathematical epidemiological model — a tool borrowed from infectious-disease biology and applied, for the first time, to a lizard predator's oral microbiome.

The Three Competing Models

At the heart of the paper is a taxonomy of three distinct hypotheses, each offering a different account of why Komodo dragon mouths harbour dangerous bacteria. Understanding these models is the key to understanding what the paper contributes.

Model 1 — Bacteria as Venom

Under this view, the pathogenic bacteria in dragon saliva function analogously to toxins: they are deliberately maintained because they provide the lizard with a hunting advantage by helping to kill or incapacitate bitten prey. The dragon is, in effect, a biological weapon system that has co-opted microbial agents to do part of its killing work. This is the model that had circulated most widely in popular media and earlier scientific commentary. Its appeal is intuitive — but it faces an awkward prediction: if bacteria are genuinely beneficial weapons, natural selection should maintain them consistently across all dragons, including those in captivity. Observation contradicts this.

Model 2 — Passive Acquisition

This alternative proposes that there is no evolutionary story to tell at all. Dragons simply pick up bacteria from the environment — from carrion they feed on, soil they contact, and prey they consume — and the oral flora is an incidental reflection of whatever microbes happen to be in the local habitat. Captive dragons lack the bacteria because captive environments are microbiologically impoverished relative to the wild island ecosystems of Komodo, Rinca, and Flores. The bacteria are passengers, not weapons. Under this model, asking "why do dragons maintain pathogenic bacteria" is the wrong question, because there is no active maintenance occurring.

Model 3 — Lizard-to-Lizard Epidemic (the Authors' Proposed Model)

Bull and colleagues propose a third route that draws on classical epidemiology. When a dragon bites large prey — a deer, pig, or water buffalo — and that animal escapes, it carries the inoculated bacteria in its wounds. If that wounded prey is subsequently killed and eaten communally by multiple dragons (a common occurrence in the field), each feeding lizard is re-exposed to the bacteria-laden flesh and saliva of its conspecifics. The prey animal acts as an indirect transmission vector between one dragon's mouth and another's, in the same way that a shared needle transmits bloodborne pathogens between human drug users. Crucially, the bacteria need not benefit the individual lizard hosting them; they simply need to spread efficiently enough through the dragon population to persist. This decouples the question of bacterial persistence from the question of bacterial virulence in prey.

The Epidemiological Model and Its Predictions

To test whether the lizard-to-lizard epidemic route was theoretically viable, Bull et al. constructed a susceptible-infected (SI) compartmental model — a standard epidemiological framework in which individuals move from a "susceptible" state (mouth currently lacking pathogenic bacteria) to an "infected" state (mouth harbouring pathogenic bacteria), with transmission rates governed by known or estimated biological parameters.

The critical transmission parameters the model required were drawn from field data compiled across multiple seasons. The authors documented that in a sample of 17 large-prey attacks observed between 2002 and 2009, roughly 30% of prey escaped the initial assault — four deer and one pig in the recorded cases. They also found that approximately 70% of 20 observed large-prey feeding events involved multiple dragons consuming the same carcass simultaneously. This communal feeding pattern is crucial: a single large carcass, particularly a buffalo, could expose more than a dozen lizards to each other's oral secretions during a single feeding bout.

When these rates were plugged into the SI model, the analysis showed that bacterial infection could persist in the wild dragon population under realistic life-history conditions. The model predicts an equilibrium at which a stable proportion of dragons carry pathogenic bacteria — not because those bacteria help individual dragons hunt, but because the communal feeding ecology creates reliable transmission pathways that keep the bacteria circulating through the population. Captive dragons fall outside this transmission network entirely, which explains their consistently clean oral environment without invoking any adaptation or deliberate maintenance by the animal.

The Bacterial Cast: What Lives in Dragon Saliva

Although Bull et al. did not conduct original bacterial culturing — that work was done by earlier researchers including those reviewed on our Goldstein 2013 page — the paper synthesises published surveys to characterise the oral flora that the model is explaining. The list is striking in its clinical weight.

Pasteurella multocida receives the most attention as the bacterium with the clearest documented capacity to cause severe systemic infection in mammalian hosts via wound inoculation. It is a common respiratory and wound pathogen across many mammal species and is fully capable of causing lethal bacteraemia in deer and buffalo given adequate inoculation. Staphylococcus aureus and Pseudomonas aeruginosa add layers of concern: S. aureus as a toxin producer and wound coloniser, and P. aeruginosa as an opportunistic pathogen capable of persisting in diverse microenvironments including the warm, protein-rich oral cavity of a large reptile.

In total, 58 bacterial taxa had been documented across published saliva surveys, with 93% assigned potential pathogenicity based on clinical literature. This diversity matters for the epidemic model: a multispecies community is more ecologically robust than a monoculture, making population-level persistence more likely even if individual bacterial species fluctuate seasonally or with prey availability.

Wild Versus Captive: The Pivotal Natural Experiment

Perhaps the cleanest empirical anchor in the paper is the contrast between wild and captive animals. Multiple independent studies had documented that Varanus komodoensis held in zoological institutions — fed clean, commercially prepared food and housed in sanitised enclosures — consistently lack the pathogenic bacterial species found in their wild counterparts. This is not a minor quantitative difference; it is a qualitative absence.

The bacteria-as-venom model struggles to account for this. If pathogenic bacteria were an evolved weapon actively maintained by the dragon's physiology or behaviour, captivity should not extinguish them so completely. The passive-acquisition model can explain the captive result (no environmental source, no bacteria) but then predicts that the bacterial community should be highly variable across wild individuals and locations, tracking local carrion availability — a prediction that is only partially supported by the field data.

The epidemic model handles both observations elegantly. In the wild, the transmission network is active: prey escapes, communal feeding occurs, bacteria circulate. In captivity, prey is not bitten and then shared; there are no wounded intermediaries; the transmission chain is severed. Bacteria disappear not because the animals are "clean" but because the ecological circuit that replenishes them no longer operates.

Context: How This Paper Relates to the Venom Discovery

It is essential to read Bull et al. (2010) in the shadow of Fry et al. (2009), which had already published evidence that V. komodoensis possesses functional venom glands delivering anticoagulant and hypotensive compounds (full discussion at our Fry 2009 review and our venom and bite overview). The venom findings did not make Bull et al. irrelevant — rather, they sharpened the question the 2010 paper was answering.

Once venom entered the picture, the ecological puzzle shifted: even if bacteria do not serve as the primary killing mechanism, they are still present in wild dragons and absent in captive ones. That distributional pattern requires an explanation independent of whether the bacteria are weapons. Bull et al. supply exactly that explanation. Their epidemic model is compatible with both the old bacteria-kills hypothesis and the newer venom-kills hypothesis — it answers a different question than either. The paper effectively argues that bacteria persist through ecological accident (communal feeding ecology and prey-mediated transmission) rather than through adaptive maintenance, regardless of whether those bacteria ever actually kill prey.

This position was later bolstered by the Goldstein et al. (2013) bacteriological study, which found that saliva bacterial loads were insufficient to serve as a reliable primary killing mechanism in healthy large mammals — broadly consistent with the Bull team's framing that bacteria are epidemiologically maintained passengers rather than evolved weapons. See our Goldstein 2013 page for full discussion of that finding.

Myths vs Facts

Common Assumption	What Bull et al. (2010) Actually Argue
Dragons deliberately maintain deadly bacteria as a hunting weapon.	Bacteria persist through ecological transmission dynamics (communal feeding, prey escape), not through any adaptive maintenance by individual dragons.
Captive dragons must have been treated with antibiotics to lack oral pathogens.	No treatment is needed. Breaking the transmission chain — by removing communal carcass feeding and prey escape events — is sufficient to explain the clean oral environment of captive animals.
The bacteria question was settled by the venom discovery.	The distributional pattern (wild vs. captive bacteria) remains a genuine ecological question regardless of which agent actually kills prey. The paper addresses that pattern.
There are only two possible explanations: bacteria as weapon, or purely random contamination.	The paper introduces a third, mechanistically distinct model — lizard-to-lizard epidemic via prey vectors — that better fits observed field data.
All 58 documented bacterial species actively harm prey.	The paper does not claim this. 93% are classified as potentially pathogenic, but potential pathogenicity in a clinical context does not equal reliable prey-killing in a field context with large, healthy ungulates.
This paper proves bacteria do kill prey.	It explicitly does not. The epidemic model is agnostic about whether bacteria contribute to prey death — it only explains how they persist in the wild population.

Key Takeaways

• The paper's core contribution is a model, not a measurement. Bull et al. synthesise existing data into a new theoretical framework — an SI epidemic model applied to a predator's oral microbiome. This is methodologically distinctive from the culturing and venom biochemistry studies it complements.
• Three competing hypotheses are clearly delineated. Bacteria as venom, passive acquisition, and lizard-to-lizard epidemic are evaluated against the same empirical constraints. Only the epidemic model predicts the wild-versus-captive contrast without additional ad hoc assumptions.
• Communal feeding is the ecological engine. The high frequency with which multiple dragons share a single large carcass is the primary driver of bacterial transmission. Remove communal feeding, and the model predicts bacterial populations collapse — exactly what is observed in captivity.
• The paper is compatible with the venom finding. Bull et al. do not claim bacteria kill prey; they claim bacteria persist ecologically. These two propositions can coexist comfortably with Fry et al.'s venom evidence.
• Prey escape rate is a key parameter. The observed approximately 30% escape rate is high enough to sustain bacterial transmission through the prey-vector route. A lower escape rate would reduce transmission and potentially allow bacteria to fade from the population.
• The model has conservation implications. Population fragmentation, reduced prey density, or changes in prey escape behaviour — all plausible under ongoing habitat pressure — could theoretically disrupt the transmission network and alter the oral microbiome of isolated dragon subpopulations.

Frequently Asked Questions

Is the paper's title "Deadly Drool" or "Deathly Drool"?

The verified published title is "Deathly Drool: Evolutionary and Ecological Basis of Septic Bacteria in Komodo Dragon Mouths." The variant "Deadly Drool" appears in some secondary sources and news coverage but does not match the journal record. The DOI-verified title uses "Deathly."

Did Bull et al. prove that bacteria help Komodo dragons kill prey?

No. The paper is explicit that it does not resolve whether bacteria cause prey death. Its goal is narrower: to explain why pathogenic bacteria are found in wild but not captive dragons. The epidemic model answers this distributional question without requiring bacteria to be lethal weapons. Whether they actually contribute to prey mortality is treated as a separate, and still open, empirical question.

What is a susceptible-infected (SI) model, and why use one here?

An SI model is an epidemiological framework that divides a population into two compartments — individuals who lack an infection (susceptible) and individuals who carry it (infected). Transition rates between compartments are governed by transmission and recovery parameters. The model predicts whether an infection can persist in a population at equilibrium. Bull et al. apply this framework to dragon oral bacteria because the ecological dynamics of bacterial spread through communal feeding closely parallel disease transmission dynamics in animal populations.

Why do captive Komodo dragons lack pathogenic oral bacteria?

Under the epidemic model, the answer is ecological rather than physiological. Captive animals are not fed live prey, do not participate in communal carcass consumption with conspecifics, and do not bite and lose prey that later return as transmission vehicles. All three elements of the wild transmission chain are absent, so bacteria are not re-inoculated and populations decline to undetectable levels. This does not require any immune difference between wild and captive animals.

How does this paper differ from the Montgomery saliva bacteria study?

The Montgomery study (reviewed at /research/montgomery-saliva-bacteria) was primarily empirical — culturing bacteria from saliva and identifying species. Bull et al. use those survey results as inputs to a theoretical model. The 2010 paper is doing evolutionary and ecological theory, not microbiology; it asks why those bacteria are there, not merely what they are.

Does the epidemic model predict all dragons carry bacteria equally?

No. The SI model predicts a stable equilibrium prevalence — a proportion of the population that carries pathogenic bacteria at any given time — rather than universal infection. The proportion depends on transmission rates (tied to communal feeding frequency and prey escape rates) and clearance rates (how quickly bacteria disappear from an individual mouth without reinfection). Dragons that feed less communally or in areas with lower prey escape rates would be predicted to have lower bacterial prevalence.

How do the 2010 findings hold up given Goldstein et al. 2013?

The Goldstein bacteriological study found bacterial loads and diversity insufficient to serve as a reliable primary killing mechanism in large healthy ungulates — see our Goldstein 2013 review. This is broadly consistent with Bull et al.'s framing: if bacteria were genuinely evolved weapons, one would expect strong selection for consistently high pathogenic loads, which Goldstein did not find. The two papers reinforce rather than contradict each other, with Bull providing the ecological mechanism and Goldstein providing the microbiological characterisation.

Is there any evidence for the prey-as-vector route beyond modelling?

Bull et al. present the field observation data on prey escape rates and communal feeding as circumstantial empirical support — they show the transmission parameters are within ranges that make the epidemic route viable. Direct experimental evidence (e.g., tracking bacterial strain identity from one dragon's bite wound through a prey animal and into a second dragon's mouth) had not been produced at the time of publication and, to this editorial team's knowledge, remains absent from the literature. The epidemic model is theoretically parsimonious and empirically consistent, but direct chain-of-transmission evidence would considerably strengthen it.

Sources & Further Reading

1. Bull, J.J., Jessop, T.S., & Whiteley, M. (2010). "Deathly Drool: Evolutionary and Ecological Basis of Septic Bacteria in Komodo Dragon Mouths." PLoS ONE 5(6): e11097. https://doi.org/10.1371/journal.pone.0011097 — Open-access primary source for this review.
2. Fry, B.G., et al. (2009). "A central role for venom in predation by Varanus komodoensis (Komodo dragon) and the extinct giant Varanus (Megalania) priscus." Proceedings of the National Academy of Sciences 106(22): 8969–8974. https://doi.org/10.1073/pnas.0810883106 — The venom paper that preceded and contextualises Bull et al.'s work.
3. Goldstein, E.J.C., et al. (2013). "Anaerobic and aerobic bacteriology of the saliva and gingiva of 16 captive Komodo dragons (Varanus komodoensis)." Journal of Zoo and Wildlife Medicine 44(2): 262–272. https://doi.org/10.1638/2012-0022R1.1 — Bacteriological survey finding insufficient pathogenicity to support bacteria-as-weapon hypothesis.
4. Auffenberg, W. (1981). The Behavioral Ecology of the Komodo Monitor. University Presses of Florida. — Foundational field observations on predation behaviour, communal feeding, and prey escape, which underpin the transmission-rate estimates used in the 2010 model.
5. Kermack, W.O., & McKendrick, A.G. (1927). "A contribution to the mathematical theory of epidemics." Proceedings of the Royal Society A 115(772): 700–721. — Original formulation of compartmental epidemic modelling; the theoretical ancestor of the SI model Bull et al. adapt.