The Deep History of Dogs, Part 2: Palaeolithic Partners

← Back to Blog Archive

A sequel to The Deep History of Dogs: A Story Written in Ancient DNA.

The Story Just Got 5,000 Years Older

In October, we explored the deep history of dogs through ancient DNA—from the proposed domestication during the Last Glacial Maximum roughly 23,000 years ago, through the debate over single versus dual origins, to the genetic legacy carried by modern breeds. At the time, the earliest genetically confirmed dogs came from the Mesolithic site of Veretye in Karelia, Russia, dated to about 10,900 years ago. Everything older was tentative—morphological identifications that could just as easily be unusual wolves.

That just changed—twice, in the same week.

Two companion studies published together in Nature on March 25, 2026—Marsh et al. and Bergström et al.—have pushed back the definitive genomic identification of dogs by nearly 5,000 years.

The first, by Marsh, Scarsbrook, Frantz and colleagues, confirmed that a canid from the Epipalaeolithic site of Pınarbaşı in central Türkiye, dated to approximately 15,800 years ago, was unequivocally a dog. The specimen was a female puppy, identified from her teeth, with bone so degraded it resembled freeze-dried coffee, according to co-author Lachie Scarsbrook. A canid from Gough’s Cave in the UK (~14,300 years ago) was also confirmed as a dog. The second study, by Bergström, Skoglund and colleagues, screened 216 canid remains—181 from Palaeolithic and Mesolithic Europe—using a newly developed genome-wide capture approach, enabling confident dog-versus-wolf classification for 141 specimens. Their oldest confirmed dog: a 14,200-year-old individual from Kesslerloch Cave in Switzerland.

Together, these papers tell us how dogs spread across the ancient world, how they moved between human cultures, what happened when farming arrived, and how their ancestry flows into modern breeds.

Dogs Were Everywhere

Previous claims of Palaeolithic dogs were always contested. Morphological analyses had identified potential dogs at sites across Europe—Goyet in Belgium (~34,000 years ago), Předmostí in Czechia (~28,500 years ago), Razboinichya Cave in the Altai Mountains (~33,000 years ago)—but when nuclear genomes were generated, they turned out to be wolves. As William Marsh put it, “A lot of very, very early putative dogs, when you run the DNA on them, they actually come out as wolves.”

The Bergström team’s screening revealed just how careful you have to be. A canid from Goyet Cave, dated to 13,700 years ago and previously proposed as a dog based on its small size and human modification, turned out to have fully wolf-like ancestry. At the Gnirshöhle site in Germany (~15,000 years ago), several specimens showed dog DNA—but with suspiciously low ancient DNA damage and a specific affinity to modern Bernese Mountain Dogs, strongly suggesting modern contamination. Morphology, context, and even radiocarbon dates can mislead without genomic data to back them up.

With rigorous genetic confirmation applied, the picture is striking. By at least 14,300 years ago, dogs were distributed across an enormous geographic range:

• Pınarbaşı, Türkiye (~15,800 years ago) — Anatolian hunter-gatherers
• Gough’s Cave, UK (~14,300 years ago) — Magdalenian culture
• Bonn-Oberkassel, Germany (~14,300 years ago) — Epigravettian culture (dog status based on morphology and mitochondrial DNA; nuclear genome not yet available)
• Kesslerloch Cave, Switzerland (~14,200 years ago) — Magdalenian material culture; Epigravettian-associated human ancestry
• Grotta Paglicci, Italy (~14,100 years ago) — Epigravettian context

That’s a span from central Anatolia to the British Isles—thousands of kilometers—all within a narrow window. And the dogs at all of these sites were genetically very similar, grouped together in a newly defined mitochondrial clade called C5. An unpublished preliminary analysis by Marsh found that the Kesslerloch, Gough’s Cave, and Pınarbaşı dogs share nearly identical mitochondrial DNA. Three dogs spanning from Türkiye to Switzerland to England were probably part of a single ice-age population. As palaeogeneticist Ludovic Orlando of the University of Toulouse put it: “That was unknown. That is a big discovery.”

Dogs as Cultural Currency

Here’s where the story gets really compelling.

Those five Palaeolithic dog sites span three genetically distinct human populations: the Magdalenian (Gough’s Cave), the Epigravettian (Bonn-Oberkassel, Kesslerloch, Grotta Paglicci), and Anatolian hunter-gatherers (Pınarbaşı). These were populations with different genetic ancestries, different toolmaking traditions, and different ways of life. Yet their dogs were essentially the same population. As Cornell canine geneticist Adam Boyko told Nature, “The dogs don’t seem to care too much which culture they’re attached to.” Scarsbrook has a term for this adaptability: “We call it the Swiss army dog.”

The Marsh team tested this formally, comparing dog and human genetic similarity from the same sites across 35 locations spanning the Late Upper Palaeolithic to the Medieval period. Human and dog population histories are generally correlated—but the Palaeolithic dogs from Pınarbaşı and Gough’s Cave were far more similar to each other than their associated human populations were. Genetically distinct human groups were sharing genetically similar dogs.

Both research teams converge on the same explanation.

This dog population likely spread across Europe alongside the expansion of Epigravettian-associated ancestry roughly 16,000 years ago. The Bergström team notes that the earliest observation of this expanding human ancestry north of the Alps is at Bonn-Oberkassel—where humans were buried alongside one of the earliest dogs. When Epigravettian people encountered Magdalenian groups, dogs were exchanged. But this exchange didn’t leave a detectable genetic signature in the human populations. The Magdalenian humans at Gough’s Cave show no Epigravettian ancestry in their DNA.

Dogs were crossing cultural boundaries that humans themselves maintained.

Living Together, Eating Together, Dying Together

The evidence for close human-dog association in the Palaeolithic goes beyond genetics.

At Gough’s Cave, the human remains showed evidence of funerary cannibalism—skulls shaped into cups, bones engraved—hallmarks of Magdalenian ritual. The dog remains show similar postmortem modification, most notably a deliberate perforation of the masseteric fossa on the mandible. Humans and dogs were receiving the same ritual treatment in death.

At Pınarbaşı, neonatal and juvenile dogs were buried in the same area as contemporaneous human burials. As Marsh noted, “Four thousand kilometres apart, we see these dogs being treated in very similar ways.”

The isotopic evidence tells us about their lives, not just their deaths. At Gough’s Cave, dogs and humans occupied similar trophic positions—eating similar diets. At Pınarbaşı, the perinatal dogs showed isotopic signatures indicating an aquatic dietary component—consistent with the abundant small freshwater fish found in the human-occupied layers. The dogs were being provisioned with fish by their human companions.

Fifteen thousand eight hundred years ago, on the Anatolian Plateau, people were feeding their dogs fish.

The Ancestry That Built Modern European Dogs

In Part 1, we discussed how dogs can be broadly divided into eastern and western Eurasian lineages. The new studies refine this picture substantially.

The Palaeolithic dogs from Pınarbaşı and Gough’s Cave belong to the western Eurasian dog lineage—the same lineage that gives rise to modern European breeds. But these western dogs derive from an eastern wolf progenitor, the same ancestral source that gave rise to dogs in Siberia, East Asia, and Australasia. This allows us to reject the hypothesis that Palaeolithic dogs in Europe arose from an independent domestication—a major update to the dual-origin debate we covered in Part 1.

The Bergström team also found that the diversity reduction typical of dogs—roughly one-third lower than in contemporary wolves—was already in place by 14,200 years ago. The domestication bottleneck happened early. This is the baseline the ancient inbreeding studies we discussed in Part 1 built upon—dogs have been working with reduced diversity from the very beginning, making the subsequent 10,000 years of stable, low inbreeding all the more remarkable, and the modern breed-era spike a sharp statistical contrast.

But when did eastern Eurasian dog ancestry arrive in Europe? Previous work attributed this to the Late Neolithic, when Steppe pastoralists brought their dogs west around 5,000 years ago. The Marsh team shows it happened much earlier. Mesolithic dogs from Serbia and Russia (~10,900 years ago) already carried roughly 44% eastern Eurasian dog ancestry—coinciding precisely with the documented spread of eastern hunter-gatherer human populations into Europe. Dogs and people were moving together.

Once established, that mixed ancestry persisted: ~30% eastern Eurasian in Neolithic dogs, ~33% in the Bronze Age, ~22% in the Iron Age, ~18% in the Medieval period, and roughly 20% in modern breeds. When you look at your German Shepherd, your Saint Bernard, or your Great Dane, one-fifth of their ancestry traces to eastern Eurasian dogs that arrived with Mesolithic hunter-gatherers thousands of years before the first farmers. The Bergström team’s complementary analysis tells a consistent story: modern European dogs trace roughly half their ancestry to the pre-agricultural dog population of Europe.

For those interested in exploring modern breed-level genetic data, we’ve built a Canine Genome Browser that provides access to Dog10K project data across over 300 breeds.

Dogs Survived the Neolithic Revolution (Even When Humans Didn’t)

Perhaps the most striking finding concerns what happened when farming arrived in Europe.

Beginning around 8,000 years ago, Neolithic farmers migrated from Southwest Asia into Europe. Ancient human genomes show these farmers replaced roughly 80–90% of the genetic ancestry of hunter-gatherer populations already living there.

But the dogs tell a different story. The Bergström team found that Southwest Asian ancestry contributions in Neolithic European dogs were far lower—ranging from possibly zero in Germany to 25% in Scotland to 66% in Greece. Compare that to Neolithic humans, who typically carried at least 70–80% Southwest Asian ancestry. The dogs were being kept; the people were being replaced.

In Scandinavia, Mesolithic dog ancestry persisted for over three millennia, through the Pitted Ware Culture dogs at 4,800 years ago. Farmers arriving in a new landscape apparently recognized the value of the dogs already there. The Neolithic transition played out very differently for dogs than for humans—heterozygosity levels in dogs didn’t even increase, unlike the 20–40% boost seen in humans from farmer-forager mixing.

Dogs crossed the cultural boundary between Magdalenian and Epigravettian peoples. Then they survived the even more dramatic boundary between hunter-gatherer and farming societies. The genetic threads in your Great Dane—or any European breed—have survived the Neolithic, the Bronze Age, and everything since.

As Hennelly and Sinding note in their Nature commentary, this European pattern of dog continuity is actually unusual—in Asia and the Americas, when people were replaced, their dogs were too. Europe is the exception: a place where the oldest ancestry of modern dogs reaches all the way back to the Ice Age.

Wolves and Dogs: A Surprisingly Clean Break

Despite more than 15,000 years of dogs and wolves sharing the same European landscapes, European dogs acquired virtually no wolf ancestry. The Marsh team found limited admixture between dogs and Near Eastern wolves—peaking at ~19% in a 7,000-year-old Israeli dog before declining to less than 5% by 2,300 years ago—but in Europe, almost nothing. This stands in sharp contrast to pigs and cattle, which interbred extensively with their wild relatives after being introduced to Europe.

By the Late Upper Palaeolithic, something was already keeping dog and wolf populations reproductively separate. Dogs were already “dogs” in a meaningful biological sense by 15,800 years ago.

What This Means for Breeders

For those of us who think carefully about genetics and breeding, these studies reinforce and extend several themes from Part 1:

The ancestry in your dogs is ancient. The western Eurasian dog lineage was established by 15,800 years ago. The eastern Eurasian contribution—roughly 20% of modern European breed ancestry—arrived during the Mesolithic. These are foundational components maintained through more than 10,000 years of population history.

The domestication bottleneck was real—and early. Genetic diversity was already reduced by one-third compared to wolves by 14,200 years ago—making the subsequent 10,000 years of stable, low inbreeding remarkable, and the modern breed-era spike a sharp statistical contrast.

Dogs have always been shaped by human networks. Palaeolithic exchange, Mesolithic migration, Neolithic adoption. Modern breeding is the latest chapter in a very long story.

Dog populations are resilient. Dogs survived the Neolithic replacement when 80–90% of human ancestry was replaced. The genetic diversity in a breed today has often survived far more than we realize.

Reproductive isolation from wolves is deep. The genetic identity of “dog” is far older and more robust than the identity of any individual breed.

Dustin's Remaining Questions

These two studies answer long-standing questions, but they also open new ones. Here are three I find especially compelling:

The Vavilov paradox. In crop domestication, there’s a foundational principle attributed to Nikolai Vavilov: the center of genetic diversity is the center of origin. This is how we traced wheat to the Fertile Crescent and maize to Mexico. For decades, the same logic was applied to dogs—modern East Asian dogs have the greatest genetic diversity, so dogs must have originated in East Asia. Both teams in these new studies found that the earliest European dogs were related to Asian wolves, not European ones—as Bergström told Nature, “Probably we are instead looking towards Asia, but Asia is a very big place.” Yet the oldest genetically confirmed dogs are all in western Eurasia, between 14,000 and 16,000 years old, while no genetically confirmed dogs older than ~10,000 years exist from East Asia. Is the East Asian diversity simply a product of larger long-term effective population sizes—village dogs maintaining diversity that breed dogs lost—rather than indicating geographic origin? This is the same kind of question that arises in crop genetics when wild relatives in secondary centers of diversity get mistaken for progenitor populations. The Vavilov principle may not apply here, and that’s a fascinating departure.

What maintained the wolf-dog reproductive barrier? European dogs maintained reproductive isolation from wolves for 15,000 years despite living in the same landscapes—that’s extraordinary, and it’s unlike anything we see in other domesticated species. Pigs and cattle interbred extensively with their wild relatives after being introduced to Europe. So what was different about dogs? Was it behavioral—dogs simply wouldn’t mate with wolves, or humans actively prevented it? Was it ecological—niche separation reducing encounters? The fact that the Near East shows some wolf-dog admixture while Europe shows almost none suggests the barrier wasn’t purely biological. It may have been mediated by the intensity of human management: dogs living in closer human association may have had less opportunity for wolf contact. Understanding this mechanism matters for modern breed management, because it tells us something fundamental about how long and how effectively reproductive isolation can be maintained in a domestic population.

What happened during the bottleneck? The Bergström team showed that genetic diversity in dogs was already reduced by one-third compared to wolves by 14,200 years ago. But was that loss a sharp crash followed by rapid recovery—a classic founder event—or a prolonged squeeze over thousands of years? Higher-coverage ancient genomes from the 14,000–10,000 year gap could answer this through demographic reconstruction methods. The shape of the bottleneck matters: a sharp crash and fast recovery preserves more diversity than a prolonged small population, because genetic drift has less time to erode variation. This connects directly to what we discussed in Part 1 about ancient inbreeding patterns—if dogs recovered quickly from the domestication bottleneck, that helps explain how they maintained stable, low inbreeding for the next 10,000 years despite starting from a reduced-diversity baseline.

The Story Keeps Getting Deeper

Six months ago, I wrote that “the story written in their DNA is far from complete.” These two companion studies prove that emphatically. In just half a year, the oldest confirmed dogs went from 10,900 to 15,800 years old, and the picture of how dogs spread across the ancient world transformed from a vague outline into a detailed narrative of cultural exchange, shared meals, parallel migrations, and remarkable genetic continuity.

The DNA-capture approach developed by the Bergström team has palaeogeneticists especially hopeful—Ludovic Orlando noted that poorly preserved samples predating these finds may now become amenable to sequencing. The ancestral grey wolf population that gave rise to all dogs has still never been identified. Every new sample has the potential to reshape the story.

What’s becoming increasingly clear is that the human-dog partnership was already deep, widespread, and consequential by the end of the last ice age. Dogs were embedded in human societies from the British Isles to Anatolia, shared between cultures, fed by human hands, and treated with the same ritual practices as humans in death. And when the world changed—when farming replaced foraging, when new peoples arrived, when civilizations rose and fell—the dogs endured.

When you look into your dog’s eyes, you’re not just looking at 200 years of breed history, or even 10,000 years of agricultural partnership. You’re looking at a relationship that was already ancient when the ice sheets retreated.

Sources

Primary Research:

Marsh, W. A., Scarsbrook, L., Yüncü, E., et al. (2026). Dogs were widely distributed across western Eurasia during the Palaeolithic. Nature, 651, 995–1003. DOI: 10.1038/s41586-026-10170-x

Bergström, A., Furtwängler, A., Johnston, S., et al. (2026). Genomic history of early dogs in Europe. Nature, 651, 986–994. DOI: 10.1038/s41586-026-10112-7

Commentary & Coverage:

Hennelly, L. M. & Sinding, M.-H. S. (2026). Dogs have deep genetic roots in ice-age Europe. Nature, 651, 890–891. DOI: 10.1038/d41586-026-00378-2

Callaway, E. (2026). Who let the wolves in? Nature News, 25 March 2026. DOI: 10.1038/d41586-026-00900-6

Studies Referenced from Part 1:

Bergström, A., et al. (2020). Origins and genetic legacy of prehistoric dogs. Science, 370, 557–564. DOI: 10.1126/science.aba9572

Frantz, L. A. F., et al. (2016). Genomic and archaeological evidence suggest a dual origin of domestic dogs. Science, 352, 1228–1231. DOI: 10.1126/science.aaf3161

Bergström, A., et al. (2022). Grey wolf genomic history reveals a dual ancestry of dogs. Nature, 607, 313–320. DOI: 10.1038/s41586-022-04824-9

Posth, C., et al. (2023). Palaeogenomics of upper Palaeolithic to neolithic European hunter-gatherers. Nature, 615, 117–126. DOI: 10.1038/s41586-023-05726-0

Charlton, S., et al. (2022). Dual ancestries and ecologies of the Late Glacial Palaeolithic in Britain. Nature Ecology & Evolution, 6, 1658–1668. DOI: 10.1038/s41559-022-01883-z