Behavioural correlations of the domestication syndrome are decoupled in modern dog breeds

Domestication is hypothesized to drive correlated responses in animal morphology, physiology and behaviour, a phenomenon known as the domestication syndrome. However, we currently lack quantitative confirmation that suites of behaviours are correlated during domestication. Here we evaluate the strength and direction of behavioural correlations among key prosocial (sociability, playfulness) and reactive (fearfulness, aggression) behaviours implicated in the domestication syndrome in 76,158 dogs representing 78 registered breeds. Consistent with the domestication syndrome hypothesis, behavioural correlations within prosocial and reactive categories demonstrated the expected direction-specificity across dogs. However, correlational strength varied between dog breeds representing early (ancient) and late (modern) stages of domestication, with ancient breeds exhibiting exaggerated correlations compared to modern breeds across prosocial and reactive behaviours. Our results suggest that suites of correlated behaviours have been temporally decoupled during dog domestication and that recent shifts in selection pressures in modern dog breeds affect the expression of domestication-related behaviours independently.


Supplementary Table 1. Behavioural assays.
Description of the behavioural assays in the DMA test and the elements in each behavioural category. For full description of the DMA test battery see Svartberg and Forkman (2002) 1 . Behaviour Description, context and scoring

Sociability
The dog is on leash led by its handler (a familiar person, usually the owner). The test leader, a stranger, approaches and greets the dog and handler. The test leader takes the leash and goes for a short walk with the dog, which includes a standstill during which the test leader pets the dog.
Returning the dog to the handler, the test leader performs a brief physical examination of the dog. The dog greeting reaction and reaction to the physical examination is scored on a scale from 1 (rejecting contact with growling and/or bite attempt) to 5 (escalated contact initiative, i.e. jumping, whimpering, barking etc.).

Playfulness
The dog is taken off the leash and the handler and the test leader start throwing a rag between them. The rag is thrown for a distance, and if the dog runs to fetch it, the test leader calls the dog back. The test leader then invites the dog to play tug of war. The dog is scored based on its interest, engagement and activity in playing on a scale from 1 (no interest, no playing) to 5 (fast engagement, very active in playing).

Aggression (distant threat)
A person dressed in a hooded cape appears from a hiding spot 40 meters away from the dog and the handler. The person is clapping his hands and moving sideways. After having moved 3 meters sideway, the person stops while widening the cape and crouching at the same time. This sequence is repeated three times while moving altogether 12 meters sideway, and then hiding again. The dog's aggressive response towards the person in the cape is scored while it is held next to its handler on a scale from 1 (no signs of aggression or threat display) to 5 (growling, snarling, raised hackles, raised tail etc.).

Fearfulness (sudden threat)
The handler is walking the dog on a leash when a human-like dummy is suddenly pulled up in front of the dog from a distance of two meters. The handler releases the dog upon appearance of the dummy. The dog's immediate response to the dummy is scored on a scale from 1 (no hesitation or only a brief stop) to 5 (fleeing more than 5 meters).

Aggression (sudden threat)
The context is the same as in the test for Fearfulnesssudden threat and the behavioural assessment follows immediately after the fear response has been assessed. The dog's aggressive behaviour towards the dummy is scored on a scale from 1 (no threatening or aggressive behaviour) to 5 (display of threatening behaviour, attacking, possibly biting the dummy).

Fearfulness (persistent threat)
The handler is standing still with the dog on a leash when two people dressed in white sheets with white hoods over their heads slowly approach the dog from 20 meters distance. The ghosts are 25 meters apart when starting the approach, creating a triangular shape with the dog. After three minutes the ghosts close in on the dog and the handler then releases the dog. The dog is scored based on the immediate behaviour towards the ghosts on a scale from 1 (stays in place in front or next to the handler) to 5 (attempting to move back further than the length of the leash and/or attempting to leave the scene. Possibly trying to flee).

Aggression (persistent threat)
The context is the same as in the test for Fearfulnesspersistent threat and the behavioural assessment follows immediately after the fear response has been assessed, while the ghosts are approaching the dog. The dog's aggressive behaviour towards the dummy is scored on a scale from 1 (no threatening or aggressive behaviour) to 5 (display of threatening behaviour, several attemps of attacking the ghosts).

Supplementary Table 2. Summary statistics.
Sample sizes (N = individual dogs in each breed category) and percentage are given for both males and females, as well as mean age (± standard error) upon testing and testing location for ancient and modern breeds. Each of the three geographic testing locaions (East, South and North) were divided into the categories City or Rural. The three geographical categories were based on the three regions of Sweden classified by the European Union and city-locations were assigned to owners living witin the city limit of the three large cities in Sweden (based on populations: Stockholm, Göteborg and Malmö). Dogs living outside the three cities were classified as rural. None of the three largest cities are located in the northern part of Sweden, and all dogs from this geographical area were therefore given the category rural. For the analyses the three geographic locations and the city or rural categories were combined to give five different combinations of locations in total. , Aggressiondistant threat (AggrD), Fearfulnesssudden threat (FearS), Aggressionsudden threat (AggrS), Fearfulnesspersistent threat (FearP) and Aggressionpersistent threat (AggrP) for ancient (below the diagonal) and modern (above the diagonal) dog breeds. The a priori direction of correlations between the behaviours based on the domestication syndrome hypothesis is presented in brackets, with (+) indicating expected positive correlations and (-) indicating expected negative correlations. We did not consider correlations using alternative measures of the same behavior (e.g. correlations between alternative measures of fear or aggression) as the domestication syndrome hypothesis does not make explicit predictions about within behaviour correlations; these cases are marked with NA in the table. Bold values indicate cases where the correlation coefficients were exaggerated in the direction-specific manner predicted by the domestication syndrome hypothesis. Note that of the 17 behavioural correlations, 15 are exaggerated in ancient breeds, one is exaggerated in modern breeds, and one goes in the opposite direction from the predictions of the domestication syndrome hypothesis for both ancient and modern breeds. Mean effect sizes (Zr) and 95% confidence intervals for 16 individual behavioural correlations between Sociability, Playfulness, Fearfulness -sudden threat (Fear (S)), Fearfulnesspersistent threat (Fear (P)), Aggression -distant threat (Aggression (D)), Aggressionsudden threat Aggression (S)) and Aggression -persistent threat (Aggression (P)).
Correlations are divided into expected positive (top seven) and negative (bottom nine) correlations. Note that the model for one of the expected negative correlations (sociability vs. aggression (P)) did not converge and was therefore not included. Source data are available as a Source Data file. Funnel plots for the correlations between sociability and playfulness, fearfulness -sudden threat (Fear S) and fearfulness -persistent threat (Fear (P)), aggression -distant threat (Aggression (D)), aggression -sudden threat (Aggression (S)) and aggressionpersistent threat (Aggression (P)) for all 78 dog breeds. The model on the correlation between Sociablity and Aggression -persistent threat (Aggression (P)) did not converge and was not included in the result section of the manuscript. Estimated mean (red line) for a given correlation overlaps completely with the fitted regression mean (blue line). N = number of dogs within each of the 78 breeds in our study. Source data are available as a Source Data file.

Supplementary Figure 3. Funnel plots for playfulness correlations.
Funnel plots for the correlations between Playfulness and Aggression -distant threat (Aggression (D)), Aggression -sudden threat (Aggression (S)), Aggression -persistent threat (Aggression (P)), Fearfulness -sudden threat (Fear (S)) and fearfulness -persistent threat (Fear (P)) for all 78 dog breeds. Estimated mean (red line) for a given correlation overlaps completely with the fitted regression mean (blue line). N = number of dogs within each of the 78 breeds in our study. Source data are available as a Source Data file.