Feasibility
The majority of dogs stuck around throughout all subtests even though they could have walked off at any time.
➞ The test was successfully implementable on the street!
Coding reliability
The majority of the variables had a good coding reliability that is comparable or even better than similar standardized pet dog tests.
➞ The ethogram was well defined and the chosen variables can be coded reliably despite the difficult street conditions.
However, the attitude variables seem to have been too complex and dependent on raters’ experience with dogs.
➞ Suggestion for the test going forward: code more distinct behaviours (e.g., fearfulness → ducking, tail between legs, ears back) and sum them as composite variables.
Test-retest reliability
The test-retest reliability was poor at an average ICC of 0.22 on the streets and thereby below common pet dog tests. This is problematic since the behavioural domestication syndrome traits are personality traits that are expected to be consistent in the dog over time. The question is whether the test was not able to capture those traits or whether outside influences altered the dogs’ expression of the traits between test and retest.
Since the test was modelled after reliable tests conducted in pet dog tests, and especially the first three tests represented common daily situations for the street dogs and they showed appropriate behaviours, it is worth focusing on the second explanation.
Indeed, there might have been several influential factors:
1. Learning effects
A commonly discussed problem in test-retest situations is that the time between test and retest might have been too short to create similar novelty. Concoordantly, the dogs might have remembered that the fake dog was fake and the novel object was not scary and thus behaved differently the second time. Since the dogs showed appropriate conspecific and explorative behaviour in the initial test, these subtests are likely still reliable in capturing the target traits and can be used as such going forward.
2. Disturbances
Compared to standardized lab tests, the streets are inherently more disturbance prone with other dogs, humans, and cars frequently passing by. In addition, the dogs were rarely re-tested in the same place as the initial test. This likely had a strong impact on the behavioural consistency between the test and retest. Since the test is to be used on the streets, managing disturbances and standardization well will be pivotal going forward.
3. Less genetically fixed behaviours
Finally, it could be speculated that free-ranging dogs are inherently more behaviourally flexible than our modern breeds. During breed selection, dogs were selected for specific functions and traits which decreased their genetic diversity by 35% compared to the initial ancestor. Meanwhile, free-ranging dogs pre-date modern breed selection and live in a more variable environment, which might have allowed them to be less genetically fixed in certain traits. However, this needs to be further explored.
Nevertheless, let’s not forget that there was a range of reliable variables, namely the interactive human-directed ones! While this is in line with sociability and biddability being the most reliable in pet dog studies, this finding is curious seeing as the tested free-ranging dogs’ lives are not as interwoven with human proximity as our modern pet dogs. This might suggest that a high level of human-directed sociability, tractability, and communication are an inherent trait in domesticated dogs, which is not only an assumption within the domestication syndrome but has also been central in dog-specific domestication hypotheses. Whether this is actually the case, however, will need more investigation into the correlations that can be drawn from this test, as well as comparisons of this test battery conducted with pet dogs and wolves.
Conclusion
The herein developed Domestication Syndrome test battery showed to be feasible and reliably codable with free-ranging dogs on the street. While the test-retest reliability was low, the suboptimal test-retest design and high environmental variability between the first and retest testing occasion were likely the more influential reason rather than the test’s inability to capture the traits reliably. Several recommendations are made in the thesis, including better management of disturbances and redefinition of the attitude variables. Since the subtests were modelled after validated tests with pet dogs and the dogs showed appropriate behaviour in the first test occasion, I argue cautiously that the test battery can go forward to test the domestication syndrome in free-ranging dogs if these recommendations are adhered to.
More importantly, the project showed that testing free-ranging dogs in long behavioural test batteries with complex ethograms is indeed possible, which is amazing news for such an understudied but highly valuable population!