How the wolf became the dog

Two accounts of the oldest domestication, and why neither can be confirmed

A Every domesticated animal was domesticated by farmers, with one exception. Sheep, goats, cattle and pigs were taken into human control after agriculture began, by people with settlements, surpluses and a use for a captive herd. Dogs preceded all of that by a wide margin, and were acquired by hunter-gatherers who had no fields, no fences and nothing obvious to gain from a large carnivore. The exception is why the question of how it happened has attracted so much argument and why it remains open. It is also why the answers proposed tend to say as much about their authors' assumptions as about the evidence: an account of the origin of the dog is necessarily an account of what early people were like, and that is a subject on which everyone has a prior view.

B The genetic evidence has narrowed some questions and complicated others. All dogs descend from wolves, but not from any wolf population now living: the closest relatives of the ancestral group are extinct, which is why comparisons with modern wolves mislead. Ancient genomes indicate that by fifteen thousand years ago there were already several distinct dog lineages spread across Eurasia, which pushes the origin considerably earlier than that date and implies a substantial period of divergence before the earliest secure archaeological remains. Whether domestication happened once or twice in different regions has been argued from the same datasets in both directions, and the honest position is that the data do not yet resolve it.

C Two accounts of the process compete. The first holds that people took wolf pups, raised them and bred the most tractable, which is how one would set about domesticating an animal deliberately. The second holds that no one intended anything: wolves that tolerated proximity to human camps gained access to discarded food, tolerance was heritable, the tolerant animals bred with each other more than with the wary, and a population diverged without anybody selecting for anything. The second account has become the more popular, partly because it requires no foresight from people who had no reason to expect a useful outcome.

D Both accounts have difficulties that their advocates tend to underplay. Deliberate capture requires that a group with an uncertain food supply divert meat to an animal that will be dangerous for years before it is useful, which is a considerable investment on speculation. Self-domestication requires a reliable accumulation of edible waste, and mobile hunter-gatherers moving every few weeks do not obviously generate one; the objection is usually met by supposing seasonal aggregation sites, which is plausible and not demonstrated. Neither account is refuted. Neither is established.

E The experiment most often cited in support of the second account has itself been questioned. A long-running programme in Siberia selected silver foxes for tolerance of human handling and reported that, within relatively few generations, the animals developed floppy ears, patched coats, curled tails and shortened snouts — the constellation of traits seen in domesticated species generally. The result was taken to show that selecting for temperament alone drags a suite of physical changes with it. Later scrutiny established that the founding animals came from a Canadian fur farm where they had been bred in captivity for decades and already showed some of the traits in question, which does not invalidate the experiment but does weaken its use as a demonstration that the changes appeared from nothing.

F What is not in dispute is that the animals that emerged are strange. Dogs attend to human faces, follow a pointing finger to a hidden object at a level wolves raised identically do not reach, and look to a person for guidance when a task becomes impossible, where a wolf continues working at it. Some of these abilities appear in puppies before training, and hand-reared wolves given identical upbringing do not acquire them. The comparison is the important one, because it removes the obvious objection that dogs simply have more practice at watching people. The differences are not simply the effect of familiarity with people; they are specific to the kind of information dogs take from people, and they constitute the clearest evidence that what was selected was not merely tameness but a capacity for a particular sort of attention.

G The wider significance of the argument lies outside the species. If a large predator can be converted into a cooperative partner without anyone intending it, then the boundary between natural and artificial selection is considerably less distinct than the terms suggest, and the same reasoning has been applied — controversially — to our own species, which shows several of the same physical changes relative to its extinct relatives. The dog is therefore not merely the first domesticated animal but the test case for a general claim about how domestication works, which is why an argument about events forty thousand years old continues to be conducted with such energy.