Wednesday, September 2, 2009

Silver fox domestication

I recently linked to an extract from Richard Dawkins’ new book in which he mentions a fascinating long-term experiment on silver foxes. The short version: starting in the late 1950s, the Russian geneticist Dmitry Belyaev selectively bred a population of silver foxes for tameness, and, surprisingly, they acquired a dog-like morphology as a by-product (floppy ears, turned-up tails, and so on). In other words, determining which foxes got to breed based solely on how tame and friendly they were produced not only successively tamer foxes, but dog-like physical traits as well. Belyaev believed (and Dawkins concurs) that the reason for this link is pleiotropy, the phenomenon of a single gene having multiple and seemingly unconnected phenotypic effects. As Lyudmila Trut, Belyaev’s successor as head of the Institute of Cytology and Genetics, explains (pdf):
Behavioral responses, [Belyaev] reasoned, are regulated by a fine balance between neurotransmitters and hormones at the level of the whole organism. The genes that control that balance occupy a high level in the hierarchical system of the genome. Even slight alterations in those regulatory genes can give rise to a wide network of changes in the developmental processes they govern. Thus, selecting animals for behavior may lead to other, far-reaching changes in the animals’ development. Because mammals from widely different taxonomic groups share similar regulatory mechanisms for hormones and neurochemistry, it is reasonable to believe that selecting them for similar behavior—tameness—should alter those mechanisms, and the developmental pathways they govern, in similar ways.
Now, this may be entirely correct but I can think of a fairly obvious alternative explanation: subtle biases in the researchers that meant the foxes were not really selected based purely on tameness. (A bit like Clever Hans in reverse). There is an Olympus Mons-sized literature on how human decision-making is influenced, entirely subconsciously, by a dizzying array of crazy things. To take one random example (also previously linked to), holding a heavier clipboard affects judgments of value and importance. Given the ubiquity of such latent biases, are we really to believe that some mutation (unconnected behavior) that merely made the affected fox look tame – made it look a bit more like a dog, say – didn't influenced judgments of tameness? To flesh this thought out a bit more, consider how the foxes were classified. Trut again:
At seven or eight months, when the foxes reach sexual maturity, they are scored for tameness and assigned to one of three classes. The least domesticated foxes, those that flee from experimenters or bite when stroked or handled, are assigned to Class III… Foxes in Class II let themselves be petted and handled but show no emotionally friendly response to experimenters. Foxes in Class I are friendly toward experimenters, wagging their tails and whining. In the sixth generation bred for tameness we had to add an even higher-scoring category. Members of Class IE, the “domesticated elite,” are eager to establish human contact, whimpering to attract attention and sniffing and licking experimenters like dogs. 
Class III seems unambiguously defined and it’s likely pretty straightforward to spot animals that belong to this category. The differences between the other classes, though, are significantly more subjective, and thus liable to all sorts of subtle biases. What, exactly, is an ‘emotional or friendly response to an experimenter’? What, exactly, is ‘eagerness to establish human contact’? It seems entirely possible – indeed likely – that animals that just looked tamer, had stereotypically domesticated features, were more likely to be assigned to Class I than to class II. If so, the foxes were not really selectively bred for “tameness and tameness alone”. No matter how scrupulous and honest the experimenters tried to be, I find it very hard to believe that they succeeded, continuously and without fail, to assign animals objectively to categories. Indeed, the researchers working on the foxes (including Trut) outlined a new scoring method in a 2007 paper, in which they admitted that a cross-breeding experiment “clearly demonstrates that the traditional scoring systems established for selection of foxes for behavior has limited resolution for measuring behavior as a continuous variable”. Assuming, as seems likely, that tameness-aggressiveness forms a continuous behavioral axis, we cannot be confident that Belyaev and his colleagues invariably selected for tameness alone. If this is correct, the pleiotropy story is somewhat undermined, though by no means refuted, of course. It seems significant, however, that the alternative explanation is more parsimonious: it need not posit nearly infallible experimenters, nor a priori unlikely pleiotropic linkages.

Of course, I’m no expert on this topic, so maybe I’ve misunderstood the protocols, or perhaps the alternative I sketch was been refuted somewhere in the literature. I would, however, be very interested to find out how the researchers ruled out this alternative hypothesis...

Trut, L. (1999). Early Canid Domestication: The Farm-Fox Experiment American Scientist, 87 (2) DOI: 10.1511/1999.2.160

Kukekova, A., Trut, L., Chase, K., Shepeleva, D., Vladimirova, A., Kharlamova, A., Oskina, I., Stepika, A., Klebanov, S., Erb, H., & Acland, G. (2007). Measurement of Segregating Behaviors in Experimental Silver Fox Pedigrees Behavior Genetics, 38 (2), 185-194 DOI: 10.1007/s10519-007-9180-1


  1. Great Post. I hope you'll include it in the next Carnival of Evolution.

    One point I don't think you made: if the researchers perceived the "doglike" foxes as more friendly, their reactions (or expectations) might have stimulated a more friendly behavior in the foxes involved (and vice versa).

    There are many accounts (admittedly anecdotal) of exceptional people who could interact in a friendly mode with even the wildest animals, perhaps this is almost universal, at a much lower level.

    This would be much easier to test: use many researchers and many animals, and then do an analysis of the results by researcher.

    An interesting aspect to include would be to use a number of researchers from different cultural backgrounds with different attitudes towards domestic dogs, and (if possible) different standards for what constitutes a domestic dog. Then analyze for correspondence between the cultural standards and both the overall level of friendly interaction and the specific differences (if any) based on correspondence of physical characteristics with those of "friendly" dogs in the culture.

    Another thought: there could be genetically determined differences in posture that interfered with the wild ability to communicate hostility or desire for privacy, and it could be those that are linked with spotted coats and floppy ears, etc. rather than the actual behavioral traits.

  2. Thanks AK... Those are certainly possibilities as well, I agree.

  3. where can you get a domesticated russian silver tame fox?