The mouth is its own ecosystem, and it barely overlaps with ours.
For most of the last century, what was known about bacteria in the dog's mouth came from culture, and culture sees only the fraction of species that will grow on a plate. The first deep, culture-independent description of the canine oral community, by Dewhirst and colleagues in 2012, changed the picture. Working from subgingival plaque taken from 51 dogs and sequencing bacterial genes directly, they placed 353 distinct bacterial taxa across 14 phyla. Around 80 per cent of those taxa had no formal species name, and only 16.4 per cent were shared with the human mouth at the sequence threshold used. To handle so many unnamed organisms, the authors introduced a provisional labelling scheme, the Canine Oral Taxon or COT number, which later canine studies have used ever since.
How much culture alone had been missing is clear from the work of Riggio and colleagues in 2011, the first study to apply culture-independent sequencing to canine periodontal disease. Sampling nine dogs across healthy, gingivitis and periodontitis groups, they found that between 35 and 38 per cent of the sequences they recovered represented potentially novel species, organisms that had never been formally described. Culture and sequencing also disagreed on which bacteria dominated: culture pointed to Actinomyces canis, at 19.4 per cent of periodontitis isolates, while sequencing put Pseudomonas species at 30.9 per cent of healthy clones, Porphyromonas cangingivalis at 16.1 per cent of gingivitis clones, and Desulfomicrobium orale at 12.0 per cent of periodontitis clones.
The practical message from both studies is that the canine mouth is a distinct microbial world, most of it invisible to older methods and much of it still unnamed. Any account of oral health in dogs has to start from that community rather than from a short list of familiar pathogens.
What a healthy dog's mouth looks like.
The composition of a healthy canine mouth is well characterised and, at first glance, surprising to anyone used to human dentistry. In the human mouth, Gram-positive Streptococcus species dominate the healthy state and drive the plaque and caries familiar from our own dentists. The healthy dog's mouth is the reverse. In the cross-sectional survey by Davis and colleagues in 2013, which examined subgingival plaque from 223 client-owned dogs across healthy, gingivitis and mild periodontitis groups, the healthy plaque was dominated by aerobic Gram-negative genera: Moraxella, Bergeyella and Neisseria, alongside Capnocytophaga and members of the Pasteurellaceae.
The specific organisms bear this out. In Davis's 2013 data the most abundant single taxon overall was Porphyromonas cangingivalis, at 7.4 per cent of sequences, followed by a Moraxella species at 3.47 per cent and Actinomyces canis at 3.23 per cent. The health-associated aerobes were not rare curiosities; they were among the numerically dominant members of the healthy mouth, which is why their later decline is so telling.
Davis's later comparative review in 2016 set the three companion hosts side by side. Six bacterial phyla dominate the oral community of humans, dogs and cats alike: Firmicutes, Bacteroidetes, Proteobacteria, Actinobacteria, Spirochaetes and Fusobacteria. Within that shared frame, the healthy state looks different in each host. The healthy human mouth is led by Streptococcus; the healthy dog by Moraxella, Bergeyella and Neisseria; and the healthy cat by Porphyromonas, Moraxella and Fusobacterium. The review also restated a point that limits every study in this field: roughly 97 per cent of oral bacteria will not grow in culture, which is why sequencing rather than swab-and-plate is now the standard tool.
One more feature of the canine mouth helps explain why it diverges from ours. Canine saliva sits at around pH 8.5, distinctly alkaline, where human saliva is close to neutral or mildly acidic. That alkaline environment does not favour the acid-producing streptococci that cause human cavities, which is part of why dogs rarely develop caries but readily develop periodontal disease.
How periodontal disease changes the community.
When periodontal disease develops, the community does not acquire a single new pathogen so much as reorganise itself. Davis's 2013 survey captured the shift in Gram-stain terms: the mean proportion of Gram-positive organisms rose from 0.29 in healthy plaque to 0.41 in gingivitis and 0.58 in mild periodontitis, a steady inversion of the healthy, Gram-negative-led state. Across the 274 operational taxonomic units they analysed, 90 showed a statistically supported association with health status.
The same reorganisation shows up at the phylum and genus level. Santibáñez and colleagues in 2021 compared 12 healthy dogs with 12 dogs with periodontal disease using gingival-margin swabs, and found clear separation between the two groups by community structure, confirmed by both ANOSIM and PERMANOVA testing. In disease, the phylum Bacteroidetes rose while Actinobacteria and Proteobacteria fell. The single clearest change was in the genus Porphyromonas, already the most abundant genus in both groups, which rose 2.7-fold, from 12.9 per cent of sequences in health to 34.7 per cent in disease.
Niemiec and colleagues in 2022 added an important layer by measuring not just proportions but absolute bacterial load. Across 51 dogs spanning healthy through severe periodontal disease, they recovered 714 bacterial species and showed that the total bacterial biomass was significantly higher in severe disease than in health. They framed the result as a pathobiome: disease is a property of the whole community and its total mass, rather than of one organism acting alone.
From the same data the authors drew up a list of 42 species that distinguished healthy from diseased mouths, a set of candidate markers rather than a single culprit. One organism in that list is worth naming for the opposite reason to Porphyromonas: Corynebacterium felinum was the only species significantly enriched in the healthy group and entirely absent from severe disease. Organisms like this, present when a mouth is well and gone when it is not, are as much a part of the disease signature as the pathogens that replace them.
The earliest signal is the loss of health-associated species.
The most useful finding for anyone thinking about early detection comes from the one truly longitudinal study in the field. Wallis and colleagues in 2015 followed 52 dogs over 60 weeks, sampling subgingival plaque from the same teeth every six weeks as some progressed from health to gingivitis to mild periodontitis. Because the same animals were tracked over time, the study could ask which changes came first.
The answer was that the earliest and most consistent signal was a decline in health-associated species rather than a bloom of disease-associated ones. As teeth progressed toward periodontitis, the aerobic Gram-negative organisms that mark the healthy mouth fell away first: Bergeyella zoohelcum, a Moraxella species, a member of the Pasteurellaceae and Neisseria shayeganii. The authors concluded that the reduction in these health-associated species was a clearer early indicator of approaching periodontitis than the appearance of the organisms usually blamed for it.
Wallis's study also drew a line under the human comparison with a practical warning: because canine and human plaque communities differ so much, oral-care products designed around human oral bacteria may not be suitable for dogs. That single-breed cohort, all miniature Schnauzers, is a limitation worth naming, and the dogs were not age-matched, but the longitudinal design makes this the strongest evidence available on the order in which the community changes.
Porphyromonas gulae and the disease-associated set.
If any organism deserves to be named in canine periodontal disease, it is Porphyromonas gulae, the canine counterpart of the human periodontal pathogen Porphyromonas gingivalis. In Niemiec's 2022 study, P. gulae was one of the two most abundant species overall, at a mean of 5.2 per cent, and it was significantly enriched in severe periodontal disease. It formed part of the core community in every disease group but not in the healthy group, which is the profile expected of a disease-associated organism.
It is worth keeping P. gulae in proportion, though, alongside its close relative Porphyromonas cangingivalis. In the same study P. cangingivalis was the single most abundant species, at 6.2 per cent, and it was one of only two organisms present as a core member across all four health groups, from healthy to severe. It thrives in health and disease alike. Comparative genomic work referenced across these papers attributes that flexibility to its ability to synthesise its own haem, where the more pathogenic Porphyromonas species must scavenge it. Wallis's longitudinal study makes the same point from the other direction: P. cangingivalis was present in 100 per cent of samples across all health states over the full 60 weeks, the most stable resident in the dataset. The lesson is that a Porphyromonas signal on its own is not a verdict; the genus is a permanent resident of the canine mouth, and it is the rise of specific members such as P. gulae, against a falling background of health-associated aerobes, that tracks disease.
Why species counts barely move while the balance shifts.
A recurring and slightly counter-intuitive result across these studies is that the number of species in the mouth does not change much between health and disease. Alpha diversity, the ecological measure of how many species are present and how evenly, is a poor separator of healthy from diseased dogs. Niemiec's 2022 study found no significant difference in the Shannon diversity index across health groups, at a p-value of 0.744, nor in the observed number of species, at 0.385. Santibáñez's 2021 study likewise found no significant Shannon difference between healthy and diseased dogs, at a p-value of 0.18.
What changes is composition, the identity and relative abundance of the members, which is captured by beta diversity rather than alpha diversity. Santibáñez's groups separated cleanly on that measure while overlapping on species count. The distinction matters for interpretation: a mouth heading toward disease is not becoming barren or overgrown in the sense of gaining or losing large numbers of species, but is being rebalanced, with the same broad cast of organisms present in very different proportions.
The predicted metabolic reading of that rebalancing is consistent with the ecology. Santibáñez's functional analysis found that healthy communities were enriched for aerobic respiration and fatty-acid synthesis, while diseased communities were enriched for lipopolysaccharide synthesis, anaerobic glycolysis and fermentation. The mouth shifts from an aerobic, Gram-negative-led community to an anaerobic, fermentative one, which fits both the fall in aerobic species and the inflammation that defines periodontal disease.
What Kislik 2024 found in 96 dogs: the mouth ages toward disease.
Age is a thread running quietly through the older studies, and one recent paper made it the subject. Kislik and colleagues in 2024 used shotgun sequencing to profile the mouths of 96 companion dogs and asked which organisms tracked with age. Of the 102 species present above a 0.1 per cent threshold, 19 correlated significantly with age. Six of those 19 were Porphyromonas species, and all six rose with age. Among the organisms that fell with age, the strongest was Conchiformibius steedae, a genus that also appears among the health-associated organisms in the diet and plaque studies.
The overall pattern was that the mouths of older dogs came to resemble the profiles previously attributed to periodontal disease: more Porphyromonas, fewer of the health-associated aerobes. The authors could even build a model that estimated a dog's age from its oral community, with a correlation of 0.563 and an average error of a little over three years. This study also provides the one partial counterpoint to the alpha-diversity story above: the Simpson diversity index rose modestly with age, at a p-value of 0.0013, so diversity is not entirely static, it drifts slowly with time. The reasonable reading is that ageing nudges the community in the same direction that disease does, which is part of why periodontal disease becomes more common in older animals.
What bad breath is telling you.
For most owners, the first noticeable sign that something has changed in a pet's mouth is the smell. Oral malodour is worth taking seriously because it is a direct readout of microbial metabolism. In a review of the subject, Culham and Rawlings in 1998 set out the chemistry: the odour comes from bacteria breaking down proteins in saliva, plaque and shed cells, and the organisms mainly responsible are Gram-negative, the same functional group that expands in periodontal disease. The specific molecules are volatile sulfur compounds, chiefly hydrogen sulfide and methyl mercaptan, and these have been implicated not just in the smell but in the tissue damage of periodontal disease itself. The review noted that in one United Kingdom survey around half of dog owners reported an odour problem, so this is far from a fringe complaint.
Because malodour originates in the microbial load, it responds to anything that reduces that load: tooth-brushing, appropriate diets and dental chews. That gives bad breath a dual role. It is a genuine early sign that the community may be shifting toward the anaerobic, protein-fermenting state of disease, and it is one of the few signs an owner can monitor at home between veterinary visits.
Diet and texture shape the oral community.
Diet is often blamed for poor oral health, and one recent study examined the question directly. Oba and colleagues in 2022 fed 12 adult dogs either a dry extruded food or a wet canned food for six weeks and profiled supragingival and subgingival plaque with shotgun sequencing, alongside breath scoring and salivary pH. Dogs on the dry food had a higher salivary pH, at 7.46 against 7.17 for the wet food, and lower breath odour, both statistically supported. Plaque coverage tended to be higher on the wet food, though as a trend rather than a firm difference.
The composition tracked the same health axis seen in the disease studies. Dogs on the dry food carried more of the health-associated organisms, including Capnocytophaga, Corynebacterium, Neisseria weaveri, Conchiformibius steedae and Porphyromonas cangingivalis, while dogs on the wet food carried more of the organisms associated with poor oral health, including Filifactor alocis, Fretibacterium fastidiosum, Treponema medium, Porphyromonas gingivalis and Tannerella forsythia. In the supragingival plaque, P. cangingivalis sat at 16.8 per cent on the dry food against 9.44 per cent on the wet.
The diet also moved the shape of the community, not just individual species. Above the gumline, the supragingival plaque of dogs on the wet food carried higher bacterial diversity than that of dogs on the dry food, and the two diets separated cleanly when the whole community was compared. Higher diversity is sometimes assumed to be better, but here it accompanied the wet-food, poorer-oral-health profile, a reminder that in the mouth it is the identity of the members rather than the count that matters.
This is a single study, on 12 dogs of one breed over six weeks, so it points a direction rather than settling the question, and the authors say as much in calling for longer intervention trials. Read carefully, it aligns with the broader picture: the mechanical action of a firmer food, and its effect on salivary pH, appears to favour the aerobic, health-associated community over the anaerobic one. Diet shapes the oral microbiome much as it shapes the gut microbiome.
What we know about cats, and what we do not.
The feline side of this story is thinner than the canine one, and it is worth being plain about that. Far fewer studies have characterised the cat's oral microbiome, and the comparative review by Davis in 2016 remains one of the main reference points, placing the healthy cat's mouth under Porphyromonas, Moraxella and Fusobacterium.
The most detailed feline data come from Anderson and colleagues in 2023, who sampled 32 cats across ten oral sites, comparing healthy cats with those affected by chronic gingivostomatitis, periodontal disease and tooth resorption. The direction of change mirrored the dog studies closely. In feline chronic gingivostomatitis, Porphyromonas gulae rose to 13.42 per cent alongside other Porphyromonas and Bacteroides species, while the health-associated commensals fell: Moraxella dropped from 6.02 per cent in healthy cats to 0.75 per cent in disease, and Conchiformibius roughly halved. The authors characterised the feline dysbiosis as a loss of commensal organisms rather than an overgrowth of a single pathogen, which is the same conclusion the longitudinal dog work reached. One methodological point from that study carries beyond cats: sampling all ten sites with swabs gave results that approximated the plaque samples collected by more invasive means, support for the idea that a well-placed swab can represent the oral community without sedation or scaling.
So the honest position is this. The cat's oral microbiome has been studied far less than the dog's, and much of what can be said rests on a small number of studies. What those studies show is that the underlying pattern, health defined by aerobic commensals and disease marked by their loss and a Porphyromonas rise, appears to hold across both species. That shared pattern is why a profiling approach built for dogs can be extended to cats, provided it is anchored to cat-specific reference data rather than borrowed wholesale from the dog.
How the mouth is measured, and why method matters.
The findings above only make sense against the methods that produced them, and those methods have changed a great deal. The earliest culture-based work saw only the minority of organisms that grow on a plate, which is why Riggio's 2011 study found that more than a third of what it sequenced was potentially novel. Culture-independent sequencing of a single marker gene, the 16S ribosomal RNA gene, opened up the full community and underlies most of the studies here, from Dewhirst and Davis through to Niemiec and Santibáñez. It resolves communities well but stops short of confident species-level identification, and different marker regions can give different answers.
Two newer approaches push further in different directions. Targeted quantitative PCR measures chosen species one at a time with high precision; Ruparell and colleagues in 2023 built and validated a panel of 41 single-species assays for canine oral organisms and showed they tracked sequencing results closely, with 30 of the 41 agreeing strongly with the matching sequencing measurement. Its strength is precision on a handful of chosen organisms; its limit is that it can only see the species it was designed to look for. Shotgun metagenomic sequencing goes the other way, reading all the DNA in a sample rather than one marker gene, which allows deeper species-level resolution across the whole community. The diet and ageing studies, Oba 2022 and Kislik 2024, both used shotgun sequencing, and it is the method BAARK uses.
Method matters for a practical reason that Ruparell's team put in stark terms: periodontal disease affects a large majority of adult dogs, yet only around 9 to 20 per cent are picked up in general practice, because diagnosis usually depends on a visual examination that cannot see below the gumline. A sequencing-based reading of the oral community offers a different kind of information, one grounded in the composition of the community rather than in what is visible on the tooth surface.
Where BAARK fits.
The gut sequenced, the mouth quantified: shotgun metagenomics on the faecal sample and a targeted qPCR panel on the oral swab, both read against dog-specific and cat-specific reference cohorts.
BAARK uses a different method on each sample, because the two sample types pose different questions. The gut is profiled by shotgun metagenomic sequencing, since the useful organisms there are not all known in advance and an open read is the only way to find them. The mouth is measured by a targeted quantitative PCR panel of 15 species plus a universal bacterial normaliser, because the organisms that separate healthy gingiva from periodontitis are already named in the literature above, and what is missing is not the species list but an accurate quantity for each one. Ruparell and colleagues made the same argument when they built and validated 41 single-species qPCR assays for canine oral organisms and showed they tracked sequencing. A quantity measured against a standard curve does not move simply because another organism moved, which a relative abundance does. That is a better measurement, though it is not the same as saying any difference between two tests is meaningful: in healthy dogs followed for six months the quantity of one target rose roughly a hundredfold, and no study has yet reported within-dog variation in an untreated animal. The value of that approach follows directly from the science above. Because oral health is defined by the balance of the community rather than by the presence or absence of one organism, a method that reads the whole community and reports relative abundances is well matched to the biology. Because the canine and feline mouths differ from the human mouth and from each other, the reference cohorts are species-specific rather than borrowed. And because the earliest signal of change is the quiet loss of health-associated aerobes rather than a dramatic bloom, a compositional profile can describe a community that is drifting before that drift is obvious on inspection.
What BAARK provides is a wellness profile, a description of the community as it stands. It is not a diagnostic test. It does not detect, diagnose, stage or predict the progression of periodontal disease or any other disease, and it is not a substitute for a veterinary oral examination, which remains the only way to assess what is happening below the gumline. The profile is best used as one input alongside that examination, a way to see the community that a visual check cannot.
The bottom line.
The mouth of a dog or cat holds a structured microbial community that is largely its own, sharing only about 16 per cent of its species with ours and running on an alkaline saliva that keeps human-style cavities rare. Health in that community is defined by aerobic, Gram-negative residents such as Moraxella, Bergeyella and Neisseria. Periodontal disease is the rebalancing of that community: those residents fall away, Porphyromonas and its anaerobic neighbours rise, the total mass of bacteria climbs, and the metabolism shifts toward fermentation and inflammation, all while the raw number of species stays much the same. The longitudinal evidence shows the loss of the healthy residents comes first, ageing pushes the community in the same direction, and the feline data, though thinner, point the same way. Reading that community directly, by sequencing rather than by eye, is a way to describe where a mouth sits on that path.