The Domestication Syndrome: What the Evidence Supports in Dogs

1. What the Claim Is

1.1 The Trait Package

The domestication syndrome refers to a set of morphological, physiological and behavioral characteristics said to distinguish domesticated animals from their wild ancestors. The classic morphological markers are white pigmentation, floppy ears and curly tails; the behavioral side is usually given as reduced aggression and increased sociability, often with increased playfulness (a trait with its own contested evidence base) and retained juvenile traits.

Physiological changes are included too — reduced stress reactivity (with the canine stress systems described separately), altered reproductive cycling, and smaller brain size relative to body mass.

1.2 The Structural Claim

The interesting part is not the list but the assertion attached to it. Because these traits are observed to co-occur across many domesticated species, they are assumed to covary within species, and a single developmental mechanism has been hypothesized to cause the co-occurrence (Hansen Wheat, van der Bijl & Wheat, 2020).

That is a testable prediction, and it went untested for a remarkably long time.

1.3 Where the Term Came From

The concept has a borrowed history worth stating early. It originated in plant biology, where it described the recurring changes crops undergo under cultivation, and according to its critics it was popularized for animals by Price in the 1980s (Johnsson, Henriksen & Wright, 2021).

Even that handover is disputed. The authors of the neural crest hypothesis report that they could find no use of the term in the Price sources cited, which speak of the "domesticated phenotype" instead, and they credit its first use for animals to an earlier paper co-written by one of them (Wilkins, Wrangham & Fitch, 2021). For a concept whose definition is the central point of contention, it is fitting that even its origin story is not agreed.

In plants the package is reasonably well specified — loss of seed shattering, loss of dormancy, larger seeds, reduced defensive compounds — and the mechanism is often traceable to selection on the harvested part. Carrying that framework to animals meant importing a term whose precision did not travel with it.

1.4 Domestication Is Not Breed Formation

One distinction has to be fixed before anything else, because the two processes are routinely run together. Domestication is what separated dogs from wolves over thousands of years. Breed formation is what separated a Border Collie from a Chihuahua, mostly within the last two centuries, under selection for specific work and specific appearance.

Modern breeds are therefore not variants of a single domesticated baseline, and a finding about one process does not transfer to the other. Where this article discusses partial decoupling of behavioral correlations in modern breeds, that finding concerns present-day breed comparisons, not a direct test of early domestication (which is where the breed evidence belongs).

1.5 Why It Matters for Dogs

If the syndrome is real and mechanistically unified, then selecting for temperament should drag morphology along with it, and morphology might correlate with behavior in particular populations. Yet a shared developmental mechanism would not, by itself, allow reliable predictions about the temperament of individual dogs. If the mechanism is not established, then the widespread assumption that a dog's appearance says something about its temperament loses one of its scientific supports.

2. Darwin's Observation and Its Afterlife

2.1 The Original Puzzle

Darwin, working from the reports of animal and plant breeders, noticed that domesticated mammals possess a distinctive suite of heritable traits absent in their wild progenitors, some of which also appear in domesticated birds and fish. The origin of that pattern remained unexplained for more than 140 years (Wilkins, Wrangham & Fitch, 2014).

2.2 Why the Observation Was Compelling

The pattern Darwin described has a specific property that makes it demand explanation: the traits are not obviously useful to the people doing the selecting.

Nobody bred cattle for white patches or pigs for floppy ears. If those features appeared reliably alongside traits that were selected, then something links them — which is why the observation generated unified explanations rather than being filed as coincidence (much as behavioral co-occurrence invites single-cause accounts).

That inference is sound in structure. What it requires is that the co-occurrence be real, consistent and not explained by simpler routes such as relaxed selection or shared ancestry — which is precisely what sections 5 and 6 put in question.

2.3 The Competing Explanations

Several unifying accounts have been proposed. Neoteny holds that domestication alters developmental timing so that adults retain juvenile characteristics, and the fox researchers themselves noted that many researchers regard this tendency as a mechanism of domestication, while its causes remain unclear (Trut, Oskina & Kharlamova, 2009) (a claim that touches how development is read generally). A thyroid-based account, associated with Crockford, assigns thyroid hormone physiology a general role in domestication changes; Wilkins and colleagues point to it as the one general alternative to their own proposal (Wilkins, Wrangham & Fitch, 2014).

Both assume the syndrome exists. As the critics later noted, the defining characteristics vary widely between accounts and have not been observed in most domesticated species (Lord, Larson, Coppinger & Karlsson, 2020).

2.4 The Shape of the Problem

An explanation that assumes its explanandum is vulnerable in a specific way: if the pattern turns out to be weaker or more variable than assumed, the mechanism explains something that may not need explaining (a failure mode worth recognizing generally).

3. How the Trait List Was Assembled

3.1 Nobody Agreed the List

The traits said to constitute the syndrome — floppy ears, curly tails, depigmentation, shorter muzzles, smaller brains, extended juvenile behavior — were never fixed by any definitional paper. Different authors include different items.

3.2 Why That Matters for Testing

A hypothesis about covariation among traits requires an agreed list of the traits. Where the list varies between authors, a result showing covariation among one set does not address another.

Wright and colleagues (2020) made exactly this point: the concept needs defining before it can be evaluated, and the attempt to define it exposed how little agreement there was.

3.3 The Selection Problem in the List Itself

Traits that support the hypothesis are more likely to be remembered as part of the syndrome, and traits that do not are more likely to be treated as exceptions. That is a hard bias to avoid without a list fixed in advance.

3.4 What a Specified Version Would Look Like

A named set of traits, operational definitions for each, a specified set of species, and a prediction about which should covary and how strongly. None of that is difficult and none of it was done before the hypothesis became widely cited.

That ordering — hypothesis first, specification later, testing later still — explains much of the argument that followed.

4. The Neural Crest Hypothesis

4.1 The Proposal

The most influential modern account proposes that the domestication syndrome results predominantly from mild neural crest cell deficits during embryonic development. Most of the modified traits, morphological and physiological, are presented as direct consequences of such deficits, with others explicable as indirect consequences (Wilkins et al., 2014).

4.2 Why It Is Attractive

Neural crest cells contribute to an unusually wide range of vertebrate structures: pigment cells, parts of the skull and jaws, some craniofacial cartilage and tooth tissues, and the chromaffin cells of the adrenal medulla. A mild reduction in their proliferation or migration would plausibly produce depigmentation, floppy ears, shortened muzzles and a dampened stress response simultaneously — which is exactly the observed list (and the stress axis is central to the behavioral claim).

The authors proposed that dozens of genes influencing neural crest development underlie this hypofunction, rather than a single locus (Wilkins et al., 2014).

4.3 Where the Trait List Stops Being Consistent

The core alterations — tameness, body size and composition, brain size and composition, color — have been altered in almost every domesticate. The type and direction of change, however, are not consistent. Relative telencephalon size is increased in pigs, llamas and alpacas but not in sheep, and in dogs brain composition is highly variable between breeds (Johnsson et al., 2021).

A package in which the same trait moves in different directions in different species is doing less explanatory work than the name suggests.

4.4 What the Hypothesis Would Predict

Stated as a testable proposal rather than as a description, the neural crest account makes specific predictions.

Variation in neural crest development should correspond to variation in the trait package within a species. Genes affecting neural crest proliferation and migration should be over-represented among those differing between domesticates and their wild ancestors. And the traits should co-vary, because they share an upstream cause.

Section 6 tests aspects of that third prediction in modern breed data; the morphological associations were not supported, while some behavioral correlations persisted. The first two have not been tested in a way that would settle them, which is what the critics mean by calling the evidence indirect.

4.5 Its Status

By 2021 the neural crest cell hypothesis had become arguably the most popular explanation for the genetics of the initial phase of domestication (Johnsson et al., 2021). Popularity is not evidence, and the hypothesis had by then attracted two distinct lines of attack.

5. The Farm-Fox Experiment

5.1 The Standard Account

Beginning in 1959, Dmitry Belyaev selected silver foxes for tameness alone. The behavioral response came first: pups wagging their tails at humans appeared in the fourth selected generation, and the first members of a "domesticated elite" that actively sought human contact in the sixth. Morphological changes followed more slowly. Altered coat color was the earliest, from the eighth to tenth selected generation, with floppy ears, curly tails and skeletal changes later still (Trut et al., 2009). Together with altered reproductive timing, this was read as the syndrome appearing as a package, from selection on behavior alone. It became the central experimental support for the whole framework — and, in the retrospective framing of its later critics, the first study to test whether selection for tameness causes the syndrome (Lord et al., 2020). This distinction matters because the experiment was designed differently: Belyaev selected on behavior and recorded associated changes, rather than setting out to test that specific causal claim.

5.2 What the Experiment Actually Did

The design is worth stating in detail, because both the claims made for it and the criticisms of it depend on specifics that rarely get reported.

Selection was on tameness alone, scored on a purpose-built scale, and the score was the sole criterion for choosing breeders. Selection was severe: in the years Trut described, typically no more than four or five percent of male offspring and about twenty percent of female offspring were allowed to breed. Foxes were not trained, specifically so that tameness could be attributed to selection rather than handling (Trut, 1999).

The response was rapid and quantified. Pups in the highest behavioral category — the "domesticated elite" — made up 18 percent of the population by the tenth generation, 35 percent by the twentieth, and 70 to 80 percent at the time of Trut's report, forty years and roughly 45,000 foxes into the program (Trut, 1999).

5.3 The Control Line

One methodological feature matters for how the results are read. Alongside the tame line, foxes were selected in the opposite direction, for the retention and enhancement of aggressive behavior, which produced a separate aggressive population (Trut et al., 2009).

That line functions as a comparison. Whatever the starting population was, the two lines diverged from it under opposite selection pressures, which makes selection, rather than handling or husbandry, the more likely source of the behavioral difference between them.

5.4 What the Historical Record Shows

Historical records and genetic analysis indicate that the foxes used in the experiment originated from fur farms in eastern Canada, and that most traits attributed to behavioral selection for tameness predated the experiment (Lord et al., 2020).

The Canadian fur farms had already been breeding foxes for docility and for coat characteristics, including depigmentation, for decades. Animals imported into the Soviet program were therefore not a wild baseline — which matters because the entire inference depends on the starting point being unselected.

On this point the critics and the program's own account are closer than the debate suggests. Trut and colleagues described the founders, 100 females and 30 males, as foxes that had been farm-bred for about fifty years and had "undoubtedly" been under natural and artificial selection for tameability throughout that time, while still showing the standard morphology and wild-type responses to humans (Trut et al., 2009). It was the later neural crest paper that described them as "effectively wild" (Wilkins et al., 2014). What is actually in dispute is how far that prior selection had already gone, and which traits it had already produced.

5.5 The Authors' Conclusion

Lord and colleagues put it in two registers. Cautiously: the conclusions of the Farm-Fox Experiment and the ubiquity of the domestication syndrome have both been overstated. More directly: the Farm-Fox Experiment does not validate the domestication syndrome (Lord et al., 2020).

Their broader assessment across domesticated species was that the overall weight of evidence does not unambiguously support the existence of the syndrome in animals.

5.6 What This Is and Is Not

An important qualification that popular coverage of this paper drops: it is an Opinion article combining historical research with a critical assessment of the existing literature. It is not a new experiment, and it did not falsify anything in the laboratory sense. What it did was remove a pillar of support.

5.7 The Counter-Critique

The response from within domestication research is more precise than the popular reading, and it draws a distinction that matters. Zeder argued that the pre-existence of syndrome traits in the Russian foxes does not undermine the existence of the domestication syndrome — what it weakens is the claim that Belyaev's selection for tameness was the catalyst for their emergence (Zeder, 2020).

Those are two different propositions, and Lord et al. attacked both while the evidence they assembled bears mainly on the second. Trut and colleagues, from the fox program itself, published their own reply (Trut, Kharlamova & Herbeck, 2020). Lord and colleagues answered Zeder and Trut together under the title "An attractive hypothesis in need of evidence" (Lord, Larson & Karlsson, 2020a), and after a further comment by Wright and colleagues they replied again in the same issue (Wright, Henriksen & Johnsson, 2020; Lord, Larson & Karlsson, 2020b). The exchange in Trends in Ecology & Evolution thus ran through several rounds during 2020 before it moved to Genetics.

5.8 What the Program's Own Data Show

Beyond the general distinction, the program's published data bear directly on the critique's central examples.

On white spotting: the "Star" spotting pattern was not confined to the experiment. In the program's 2009 review it also occurred in farm-bred populations not selected for behavior, at a frequency of 0.71 percent, compared with 12.4 percent in the tame population (Trut et al., 2009). That is consistent with the critique's point that the trait was present before the experiment, and it also shows that its frequency rose sharply under selection for tameness.

On curly tails: tail shape was never a selection criterion, since selection was on tameness alone (Trut, 1999). Coat changes were the first morphological novelties to appear, from the eighth to tenth selected generation, and curly tails came later. In Table 2 of Trut et al. (2009), the frequencies are 9.4 percent in the tame population and 0.8 percent in farm-bred foxes not specifically selected for tameness (Trut et al., 2009).

What these figures can and cannot carry needs stating. They come from the research group itself, they compare the tame line with contemporary farm populations rather than with the original founders, and they are a summary table rather than a re-analysis designed to answer the critique. They show that the data closest to the material do not simply confirm the critique. They do not settle which traits were present in the founders, and at what frequency.

5.9 Where That Leaves the Fox Experiment

Three things can be held at once. The founding population was not wild, which weakens the strongest version of the claim. The selection was genuine, severe and single-criterion, with a line selected in the opposite direction, which is more than most domestication evidence offers. And several traits became markedly more frequent under selection for tameness than in farm populations not selected for behavior. That is what a correlated response to selection looks like, but a correlated response does not by itself demonstrate a shared developmental mechanism (Johnsson et al., 2021).

What the experiment cannot do is what it has been asked to do for sixty years: serve as the demonstration that selection for tameness causes a fixed package of traits in any mammal.

6. Testing Covariation in Dogs

6.1 The Behavioral Test

The prediction that syndrome traits covary within a species was finally tested directly. The data came from 76,158 dogs of 78 breeds that had completed the Swedish Dog Mentality Assessment, a standardized behavioral test scored by trained observers rather than by owners. Across all dogs, the correlations among fear, aggression, sociability and playfulness pointed in the directions the syndrome predicts. Their strength, however, differed by breed group: the correlations were more pronounced in the seven ancient breeds, while in the 71 modern breeds the expected negative correlations between prosocial and reactive behavior did not differ significantly from zero (Hansen Wheat, Fitzpatrick, Rogell & Temrin, 2019).

The result is therefore not a flat failure of the prediction. Positive correlations within prosocial traits and within reactive traits were also detectable in modern breeds. What was particularly reduced in modern breeds were the predicted negative correlations between prosocial and reactive traits. Whether that counts against a single binding mechanism depends on how the decoupling is read (section 6.4) — and fear and aggression in particular turn out to be far less unitary than the syndrome framing assumes (as the fear literature shows).

One qualification belongs here rather than only in the appraisal below: the ancient-versus-modern classification compares present-day genetic breed groups, not dogs sampled from successive stages of ancient domestication. It does not exclude that traits were more tightly linked during earlier stages of domestication and have since been decoupled by more recent directional breeding.

6.2 The Morphological Test

The follow-up tested whether the classic morphological markers — white pigmentation, floppy ears, curly tails — predict the strength of behavioral correlations across 78 dog breeds. They did not. The morphological traits did not covary among themselves, and neither their presence nor their number predicted the strength of the behavioral correlations, contrary to the prediction that a single mechanism produces them together (Hansen Wheat, van der Bijl & Wheat, 2020).

The authors attach a qualification of their own to that result. Floppy ears, white markings and curly tails in today's breeds may not be domestication traits at all, but traits shaped later, during breed formation, in which case no covariation would be expected in the first place. They regard that as likely for dogs and conclude that present-day breeds have limited bearing on the selection pressures of initial domestication (Hansen Wheat, van der Bijl & Wheat, 2020).

6.3 What Covariation Would Have to Look Like

The prediction being tested is specific, and worth stating in measurable terms before the result is read.

If a shared genetic or developmental mechanism affects several traits, statistical associations may arise under suitable conditions. That does not by itself imply that an individual dog with more white markings must be calmer or that one with floppy ears must be less fearful. The studies considered here tested within-breed correlations between behavioral traits and whether the strength of these correlations was associated with morphological characteristics assigned at breed level.

The first study tested the behavioral half — whether fear, aggression, sociability and playfulness hold together. The second tested the crossover — whether the morphological markers relate to how tightly the behavioral traits cluster (Hansen Wheat et al., 2019; 2020).

6.4 What the Decoupling Suggests

That behavioral correlations were stronger in ancient breeds and weaker in modern ones is the more interesting half of the first result, and it admits two readings.

Either the traits were linked and two centuries of intense directional breeding pulled them apart — in which case the syndrome may have been real early and is not now. Or the correlation in ancient breeds reflects something other than a shared mechanism: less intensive selection, more shared ancestry, smaller effective population sizes.

The authors lean toward the first reading and interpret the pattern as a temporal decoupling, driven by recent shifts in selection pressure in modern breeds (Hansen Wheat et al., 2019). The design cannot exclude the second. The ancient group comprised seven breeds and 251 dogs against more than 75,000 dogs in the modern group, and a cross-sectional comparison of breed groups shows a difference without identifying its cause. The morphological study adds a third possibility for its own half of the question: that the markers being tested are breed traits rather than domestication traits (Hansen Wheat et al., 2020).

6.5 What That Result Means

Few other domesticates offer behavioral data on this scale: tens of thousands of animals assessed in one standardized test (Hansen Wheat et al., 2019). Some predicted positive behavioral correlations were detectable in modern breeds, whereas the anticipated negative correlations between prosocial and reactive behaviors were not statistically supported in this group. This challenges the expectation of an intact unified package in modern breeds, but does not rule out a shared mechanism in early domestication. The follow-up authors also caution that present-day breeds may offer limited insight into early domestication (Hansen Wheat et al., 2020).

It also bears directly on a widespread assumption in practice: that a dog's morphology carries information about its temperament (which the aggression evidence independently contradicts). On the authors' reading, coat pattern and ear carriage in modern breeds were shaped largely during breed formation (Hansen Wheat et al., 2020) (as merle genetics illustrates in detail).

7. The Exchange in Genetics

7.1 The Critique

A formal critique of the neural crest hypothesis appeared in the same journal that had published it. Its two central points are worth stating precisely: to the extent that a domestication syndrome exists, that does not imply a universal genetic mechanism; and even granting a universal mechanism, the evidence that it is neural crest cell proliferation and migration is indirect and weak (Johnsson et al., 2021).

The authors also noted that direct evidence for pleiotropic alleles underlying the syndrome is sparse.

7.2 The Reply

The response is instructive, because it identifies a genuine misreading. Wilkins and colleagues stated that they never believed nor claimed that the domestication syndrome involves an identical set of altered traits across species, and pointed to their own 2014 table listing which traits are associated with which species — the clear implication being that other species do not exhibit them (Wilkins, Wrangham & Fitch, 2021).

The criticism that the syndrome is not universal therefore lands on a position the original authors say they did not hold.

7.3 A Point About What Kind of Claim This Is

Wilkins and colleagues made one further argument that reframes the dispute: the domestication syndrome is not a hypothesis at all but a generalization from observations. Their illustration is difficult to argue with — smaller teeth and floppy ears cannot appear in chickens, which have neither teeth nor external ears, so a demand that every trait appear in every domesticate was never coherent (Wilkins et al., 2021).

The opposing camp had itself argued that the syndrome needs a definition before it can be evaluated (Wright, Henriksen & Johnsson, 2020). Both sides therefore agree that the concept is underspecified — and disagree about who bears the cost of that.

7.4 Why Both Sides Can Be Partly Right

The exchange is unusual in that neither party is making an obviously weak argument, and the reason is that two separate questions are in play.

Is there an observable pattern? Probably yes. Domesticates differ from their ancestors in recurring ways, the observation goes back to Darwin, and neither camp denies it. What is contested is how consistent the pattern is across species and whether it is tight enough to require a unified explanation.

Is there a universal mechanism? Not established. Johnsson and colleagues ask whether the evidence supports neural crest involvement as the mechanism, and it does not — which both sides accept, since Wilkins et al. presented a hypothesis rather than a demonstration. Direct evidence for pleiotropic alleles underlying the package is sparse.

Those two answers are not symmetrical, and reading the debate as evenly balanced misses that. The pattern is likely real; the mechanism is unproven; and a hypothesis can be worth pursuing, poorly evidenced and correctly criticized at the same time.

7.5 What Remains Contested

Strip away the misreading and the substantive disagreement persists: whether the co-occurrence is strong and consistent enough to require a unified explanation, and whether neural crest involvement has direct evidential support or only mechanistic plausibility. Neither question is settled, and both are live in the primary literature.

Nothing in the critique establishes that neural crest cells play no part. What is contested is whether they constitute the unified mechanism, and a role that is partial, species-variable or one contributor among several remains entirely compatible with the evidence as it stands. The critics themselves conceded a point worth recording — that the hypothesis increased interest in development among domestication researchers (Johnsson et al., 2021).

8. What Survives

8.1 The Descriptive Observation

Domesticated animals do differ from their wild ancestors, and some traits recur across domesticates often enough to be worth explaining. Darwin's observation was not an artifact.

8.2 What Does Not Survive Intact

Three claims are in worse shape than their circulation suggests. That the trait package is universal across domesticates. That the fox experiment demonstrated selection for tameness producing the package from a wild baseline. And that the traits covary within species. In dogs, the behavioral correlations appeared in the predicted direction but held mainly in a small group of ancient breeds and not in modern ones, and the morphological markers neither covaried nor predicted the behavioral pattern (Hansen Wheat et al., 2019; Hansen Wheat et al., 2020).

8.3 Competing Accounts of Early Domestication

Several explanations circulate, but they are competing historical hypotheses, not independently established findings of the dog studies reviewed here. Serpell (2021) critically compares the commensal-scavenging account, which he describes as the dominant narrative in recent literature, with a human-initiated route involving the adoption and rearing of wolf pups. He himself concludes that the scavenging account is untenable and favors early socialization of wolf pups, which makes his review one side of an open debate rather than its resolution.

Reduced fear of humans — one proposed early step is a higher threshold for fleeing from people, which is narrower than a general increase in sociability. This is a component of the contested scavenging account, not a verified first step (Serpell, 2021).

So-called self-domestication — under the commensal hypothesis, wolves better able to tolerate humans might have gained food around settlements and left more descendants without initially being deliberately bred. Whether this scenario describes the actual origin of dogs is disputed (Serpell, 2021).

Multiple routes and selection pressures — domestication need not have followed a single uniform pathway across animal species. That possibility alone does not establish a particular alternative mechanism or a universal suite of traits.

8.4 A Third Position

The debate has not stalled at neural crest versus no mechanism. A recent proposal accepts that a domestication syndrome exists while treating it as a variable rather than fixed condition, and locates the cause in shared reproductive disruption — changes to male competition, maternal stress and related selective pressures — rather than in tameness or neural crest deficits (Gleeson & Wilson, 2023).

Whether that account fares better is an open question. Its relevance here is structural: the field is now generating alternative unified explanations rather than abandoning the search for one, which suggests the underlying pattern is considered real enough to require explaining.

8.5 What This Means for Dogs Today

The syndrome debate reaches practice through one inference: that appearance carries information about temperament because both come from the same developmental source. Four consequences follow from the weakening of that inference.

Floppy ears are no sign of a gentler dog. Across 78 breeds, the presence of floppy ears did not predict how strongly the behavioral traits clustered (Hansen Wheat et al., 2020). These are breed-level data, not measurements of individual dogs. They do not establish that ear shape can never correlate with any behavior, but they offer no reliable basis for predicting an individual dog's temperament from ear shape.

Appearance is a poor guide to how "original" a dog is. Some northern breeds, such as the Siberian Husky and Alaskan Malamute, belong to the genetically defined ancient-breed group used in the study, characterized by an origin more than 500 years ago and detectable wolf admixture (Hansen Wheat et al., 2019). This category must not be mistaken for a direct measure of recent wolf admixture or an unchanged early domestic dog. But the coat, ear set and muzzle that define breeds today were shaped largely during breed formation (Hansen Wheat et al., 2020), so a wolf-like look by itself says little about where a particular dog stands.

Head shape is not a reliable window onto social behavior. The domestication literature gives no basis for reading temperament from skull shape. Where skull conformation does matter for welfare, it does so through mechanics rather than temperament (as the brachycephalic evidence shows).

Behavior has to be assessed in the individual. That was already the conclusion from the breed research, and the domestication literature removes one of the remaining scientific-sounding arguments against it.

8.6 Why the Practical Conclusion Barely Moves

Little changes at the level of practice, and that is itself informative. A plausible explanation for the missing links between the external markers tested and the behavioral correlations is that appearance and behavior in modern breeds were shaped under different breeding goals (Hansen Wheat et al., 2020) (with the genomic evidence pointing the same way). Behavioral tendencies remain partly heritable and poorly predicted by breed or by looks — and the wolf-derived narratives that often travel alongside the domestication story have their own evidential problems (the dominance construct being the clearest case).

What the debate does supply is a reason to distrust the tidy narrative — the version in which one selective pressure produces a coherent package, and in which the dog's whole nature falls out of a single mechanism.

9. Why the Argument Persists

9.1 The Concept Was Never Specified

Both camps agree the domestication syndrome lacks an operational definition (Wright, Henriksen & Johnsson, 2020; Wilkins, Wrangham & Fitch, 2021), and they disagree about what follows.

For the critics, an unspecified concept cannot be evaluated and should not be treated as established. For the proponents, it is a generalization from observation rather than a hypothesis, and demanding a fixed trait list misunderstands what kind of claim it is.

Neither position is unreasonable, and no amount of new data resolves a disagreement about what the claim is (the same problem that afflicts practitioner constructs).

9.2 The Story Is Better Than the Evidence

The narrative — select for tameness, get floppy ears free, all through one developmental switch — is memorable, mechanistically satisfying and easy to teach. That combination reliably outperforms a qualified account, independently of which is correct.

It is also the version that reached textbooks, popular science and dog training material, where it now functions as background knowledge rather than as a contested claim.

9.3 Both Sides Have Been Over-Read

Wilkins and colleagues proposed a mechanism and were reported as having established one. Lord and colleagues published an opinion article and were reported as having disproved a syndrome. Hansen Wheat found breed-level decoupling and was reported as having shown the syndrome does not exist.

In each case the reporting is stronger than the paper, which is the standing hazard when a field's findings are interesting enough to travel (a pattern documented across canine research).

9.4 The Cost of the Simplified Version

The narrative has practical consequences beyond the seminar room. It licenses the inference that appearance and temperament are linked at a mechanistic level — which is the assumption behind reading a dog's character from its ears, coat or muzzle.

That inference was never supported by the syndrome even in its strong form, since a shared developmental origin across species says nothing about individual variation within one. But it is the version that reaches breeders and buyers, and it is the reason a debate about neural crest cells matters to anyone choosing a puppy.

9.5 What Would Actually Settle It

Three strands of evidence that are not yet available in the combination required. An operational definition specifying which traits count and how they are measured. Within-species covariation data from populations that have not been through two centuries of directional breeding. And a genetic test of the neural crest proposal — showing that variation in neural crest development corresponds to variation in the trait package within a domesticate.

Until then the disagreement is about interpretation of a shared and incomplete body of evidence, which is why the published exchange between 2014 and 2021 left it open, and why a further unified account was still being proposed in 2023 (Gleeson & Wilson, 2023).

10. What a Reader Should Take From the Dispute

10.1 Both Sides Are Arguing Carefully

This is not a case of good science against bad. The proposal was serious, the critiques are serious, and the replies to the critiques are serious. What makes it hard to resolve is that the original claim was never specified tightly enough to be decided.

10.2 The Honest Summary of the Evidence

Domesticated animals do show some recurring changes. Whether those changes form a coherent package with a single developmental cause is unresolved, and the canine tests of covariation did not find in modern breeds what the hypothesis predicts (Hansen Wheat et al., 2019, 2020).

10.3 What Changes for Dogs Specifically

Very little in practice. Whether floppy ears and tameness share a developmental origin does not alter how a dog learns, what it needs or how it should be trained.

What it does affect is the confidence with which certain stories about dogs are told, and those stories do reach practice.

10.4 Why the Article Is Worth Its Length

Because the simplified version is taught widely and presented as settled. Where a contested hypothesis circulates as a finding, setting out the dispute accurately is the useful contribution.

11. Summary at a Glance

The syndrome is a package claim — White pigmentation, floppy ears, curly tails, reduced aggression, increased sociability, retained juvenile traits, altered stress and reproductive physiology.

The central hypothesis concerns covariation — Traits are assumed to covary within species because they co-occur across species, with a single developmental mechanism proposed as the cause (Hansen Wheat et al., 2020).

The mechanism proposed is neural crest deficit — Mild reductions in neural crest cell development during embryogenesis, explaining pigment, cartilage, jaw and adrenal changes together (Wilkins et al., 2014).

The fox experiment did not start from a wild population — The animals originated from eastern Canadian fur farms, where they had already been bred for docility and coat traits, and most attributed characteristics predated the experiment (Lord et al., 2020).

The full predicted trait package is not supported in modern dog breeds — Among 76,158 dogs, some positive behavioral correlations were evident in both breed groups, while the expected negative correlations between prosocial and reactive traits were pronounced mainly in seven ancient breeds (Hansen Wheat et al., 2019). Across 78 breeds, the classic morphological markers neither covaried nor predicted the strength of behavioral correlations, possibly because they are breed traits rather than domestication traits (Hansen Wheat et al., 2020).

The mechanism critique is about evidence, not plausibility — Even granting a universal mechanism, evidence for neural crest involvement was judged indirect and weak (Johnsson et al., 2021).

The critics partly attacked a position not held — The original authors state they never claimed an identical trait set across species (Wilkins et al., 2021).

A key point in the critique was qualified — Pre-existing traits in the foxes weaken the claim that tameness selection caused them, not the existence of the syndrome (Zeder, 2020).

The selection was severe and single-criterion — Typically four to five percent of males and about twenty percent of females were allowed to breed, on tameness score alone, with no training (Trut, 1999).

A comparison line existed — A separate population was selected in the opposite direction, for aggressive behavior (Trut et al., 2009); the archival evidence on the founders does not bear on that part of the design.

Nothing here is settled — The exchange ran through several rounds in 2020–2021 and continues, with a third mechanism proposed as recently as 2023 (Gleeson & Wilson, 2023).

12. Research Gaps and Critical Appraisal

The central critique is an Opinion article. Lord et al. combine historical archival work with a literature assessment. That is legitimate scholarship and it is not an experiment; describing it as having "disproved" the syndrome misrepresents what it is.

The canine covariation tests have specific limits. The behavioral data come from a standardized test scored by trained observers, which is a strength over owner questionnaires, but from a single test battery applied to purebred dogs in one country (Hansen Wheat et al., 2019). The ancient breed group, on which the contrast rests, comprised only seven breeds and 251 dogs. The morphological traits were scored from breed standards as present or absent per breed (Hansen Wheat et al., 2020), so that part of the analysis works at breed level and does not strictly exclude mechanisms operating at the level of the individual.

Absence of covariation now does not settle covariation then. Modern breeds have been under intense directional selection for two centuries. Traits could have been mechanistically linked during early domestication and subsequently decoupled — which the ancient-versus-modern breed difference is at least consistent with.

The syndrome has never been operationally defined. Different authors list different traits, which is precisely why "does the syndrome exist" is hard to answer: there is no agreed specification to test.

Direct genetic evidence is sparse on both sides. Evidence for pleiotropic alleles underlying the syndrome is thin (Johnsson et al., 2021), and no study has demonstrated that neural crest variation produces the trait package in any domesticate.

The fox critique is historical, not experimental. Lord et al. established where the founding animals came from and what had been bred into them. That is archival work. The selection design, the line selected for aggression and the frequency changes under selection are separate questions, and the archival evidence does not settle them.

Wild baselines are largely unavailable. For most domesticates the ancestral population is extinct, geographically displaced or itself altered by contact with humans, which makes before-and-after comparison structurally difficult.

The debate is partly terminological. Whether a weakened, species-variable version of the syndrome counts as "the syndrome existing" depends on a definition nobody has fixed — and both camps have said so, while disagreeing about the consequence (Wright et al., 2020; Wilkins et al., 2021).

The concept was imported from plant biology, and even the handover is disputed. "Domestication syndrome" was developed to describe crops (Johnsson et al., 2021). Who first applied it to animals is contested between the two camps (Wilkins et al., 2021), and the term arrived in animal biology without an agreed definition (Wright et al., 2020) — which is part of why the trait list has never stabilized.

13. Conclusion

The domestication syndrome is a good case study in how a compelling story can outrun its evidence and then prove hard to dislodge. The narrative — select for tameness, get floppy ears and curly tails for free, all through one developmental switch — is memorable, mechanistically satisfying, and was supported by an experiment whose founding population turned out to have already been shaped by prior breeding for docility and coat characteristics on Canadian fur farms. When the covariation prediction was finally tested directly, in more than 76,000 dogs, some predicted behavioral associations remained in modern breeds, while the expected negative links between prosocial and reactive behavior were stronger in a small group of ancient breeds, and the classic morphological markers showed no link to it. That is a serious problem for the unified-mechanism account. It is not, however, the demolition it is often reported as: the central critique is an opinion piece rather than an experiment, the ancient-breed contrast rests on seven breeds, the morphological test works at breed level, modern breeds have been under two centuries of directional selection that could have decoupled anything, and the original authors have pointed out that the universality they are accused of claiming was never their claim. What is left is more modest and more useful than either headline. Domesticated animals really do differ from their ancestors in patterned ways that deserve explanation; the explanation is not currently in hand; the disagreement is partly about what kind of claim the syndrome is, which no further data will settle; and for dogs specifically, the tested classic morphological markers do not reliably predict the measured behavioral correlations, and while that does not settle every possible link between conformation and behavior, it leaves no basis for reliable character predictions from appearance — which was worth knowing regardless of how the mechanism question resolves (as the evidence on individual variation already indicated).

Key Insights (Takeaways)

• The foxes were not a wild baseline. Historical records and genetic analysis show that the animals used in the Russian Farm-Fox Experiment came from fur farms in eastern Canada, where they had already been bred for docility and coat traits, and that most characteristics attributed to selection for tameness predated the experiment (Lord et al., 2020). On that basis the experiment does not establish the strong causal version of the claim — that selection for tameness produced the package from an unselected starting point. The selection design itself, single-criterion selection on tameness (Trut, 1999) with a separate line selected for aggression (Trut et al., 2009), is not what the archival evidence calls into question.

• The full proposed trait package is not supported in modern dog breeds. Among 76,158 dogs, some predicted behavioral correlations remained detectable in both breed groups. The expected negative correlations between prosocial and reactive traits were stronger in seven ancient breeds but not statistically supported in modern breeds (Hansen Wheat et al., 2019); across 78 breeds, the classic morphological markers did not predict the behavioral correlations (Hansen Wheat et al., 2020). The prediction that a single mechanism binds these traits together is not supported in modern dog breeds — which, as section 6.4 sets out, is compatible with the traits never having been linked, with their having been decoupled by two centuries of directional selection, and, for the morphological markers, with their being breed traits rather than domestication traits.

• The mechanism critique concerns evidence, not plausibility. Neural crest deficits would elegantly explain pigment, cartilage, jaw and stress-axis changes at once — but the evidence that this is what actually happened was judged indirect and weak, and direct evidence for pleiotropic alleles underlying the syndrome is sparse (Johnsson et al., 2021).

• Part of the critique targets a claim the authors deny making. Wilkins and colleagues responded that they never held the syndrome to involve an identical trait set across species, citing their own 2014 table of species-specific traits (Wilkins et al., 2021). The dispute over universality is therefore partly a dispute about what was claimed.

• The fox experiment is stronger methodologically than the critique suggests, and weaker evidentially than the story requires. Selection was on tameness alone, severe — four to five percent of males and about twenty percent of females breeding — with no training (Trut, 1999), and a separate line was selected in the opposite direction (Trut et al., 2009). The founding population was nonetheless not wild (Lord et al., 2020), which weakens the causal claim without touching the selection design.

• For dogs, the practical upshot is what it already was. The classic domestication markers tested do not reliably predict the measured behavioral correlations, plausibly because appearance and behavior in modern breeds were shaped under different breeding goals (Hansen Wheat et al., 2020). That does not rule out every possible link between conformation and behavior, but it offers no basis for reading an individual dog's character from its appearance. The weakening of the syndrome narrative removes a scientific-sounding support for reading a dog's character off its ears, coat or tail — a support that was never load-bearing.

References

Gleeson, B. T., & Wilson, L. A. B. (2023). Shared reproductive disruption, not neural crest or tameness, explains the domestication syndrome. Proceedings of the Royal Society B: Biological Sciences, 290(1995), 20222464. https://doi.org/10.1098/rspb.2022.2464

Hansen Wheat, C., Fitzpatrick, J. L., Rogell, B., & Temrin, H. (2019). Behavioural correlations of the domestication syndrome are decoupled in modern dog breeds. Nature Communications, 10(1), 2422. https://doi.org/10.1038/s41467-019-10426-3

Hansen Wheat, C., van der Bijl, W., & Wheat, C. W. (2020). Morphology does not covary with predicted behavioral correlations of the domestication syndrome in dogs. Evolution Letters, 4(3), 189–199. https://doi.org/10.1002/evl3.168

Johnsson, M., Henriksen, R., & Wright, D. (2021). The neural crest cell hypothesis: No unified explanation for domestication. Genetics, 219(1), iyab097. https://doi.org/10.1093/genetics/iyab097

Lord, K. A., Larson, G., Coppinger, R. P., & Karlsson, E. K. (2020). The history of farm foxes undermines the animal domestication syndrome. Trends in Ecology & Evolution, 35(2), 125–136. https://doi.org/10.1016/j.tree.2019.10.011

Lord, K. A., Larson, G., & Karlsson, E. K. (2020a). Reply to Zeder and Trut et al.: An attractive hypothesis in need of evidence. Trends in Ecology & Evolution, 35(8), 651–652. https://doi.org/10.1016/j.tree.2020.05.008

Lord, K. A., Larson, G., & Karlsson, E. K. (2020b). Brain size does not rescue domestication syndrome. Trends in Ecology & Evolution, 35(12), 1061–1062. https://doi.org/10.1016/j.tree.2020.10.004

Serpell, J. A. (2021). Commensalism or cross-species adoption? A critical review of theories of wolf domestication. Frontiers in Veterinary Science, 8, 662370. https://doi.org/10.3389/fvets.2021.662370

Trut, L. N. (1999). Early canid domestication: The farm-fox experiment. American Scientist, 87(2), 160–169. https://doi.org/10.1511/1999.2.160

Trut, L. N., Kharlamova, A. V., & Herbeck, Y. E. (2020). Belyaev's and PEI's foxes: A far cry. Trends in Ecology & Evolution, 35(8), 649–651. https://doi.org/10.1016/j.tree.2020.03.010

Trut, L., Oskina, I., & Kharlamova, A. (2009). Animal evolution during domestication: The domesticated fox as a model. BioEssays, 31(3), 349–360. https://doi.org/10.1002/bies.200800070

Wilkins, A. S., Wrangham, R. W., & Fitch, W. T. (2014). The "domestication syndrome" in mammals: A unified explanation based on neural crest cell behavior and genetics. Genetics, 197(3), 795–808. https://doi.org/10.1534/genetics.114.165423

Wilkins, A. S., Wrangham, R., & Fitch, W. T. (2021). The neural crest/domestication syndrome hypothesis, explained: Reply to Johnsson, Henriksen, and Wright. Genetics, 219(1), iyab098. https://doi.org/10.1093/genetics/iyab098

Wright, D., Henriksen, R., & Johnsson, M. (2020). Defining the domestication syndrome: Comment on Lord et al. 2020. Trends in Ecology & Evolution, 35(12), 1059–1060. https://doi.org/10.1016/j.tree.2020.08.009

Zeder, M. A. (2020). Straw foxes: Domestication syndrome evaluation comes up short. Trends in Ecology & Evolution, 35(8), 647–649. https://doi.org/10.1016/j.tree.2020.03.001