Genes + Selection + Behaviour + Learning = Possibly A Dog
By Dr Ed Bailey
Permisson to reprint
First published Gun Dog Magazine 1986
Among the wild members of the dog family, Canidae, of which several species of foxes, the coyote and the wolf still occur in North America, there are some very species-specific hunting strategies as well as some common-to-all-dogs strategies. Hunting strategies are simply how the animal gets food. A hunting strategy of a read fox hunting mice is a stalk, poise, pounce sequence. A pack of coyotes on a deer hunt rundown the deer, pull out the gut, get the deer down and dismember it quickly. Each grabs his chunk of venison and goes off to dine privately. Wolves working as a pack on a moose will bring it to bay and harry it to exhaustion, then bring it down.
Besides these relatively regal methods of foraging, all will try to steal from other predators and will pick up carrion whenever they can. Canids also store food by burying it. Burying behaviour is so strong in foxes that the movements of digging, placing a morsel and covering it, go on in a fixed action pattern even if the animal is caged in an area with a cement floor. Performance of the motions of burying apparently satisfies the animal even though the morsel is in a corner in plain view.
These behaviour patterns used in acquiring food have evolved through natural selection over millions of years. Survival value of these hunting strategies for wild canids is obvious. However, today’s hunting dogs are expected to behave in ways quite different, even opposite of what natural selection has brought about in wild members of the dog family.
Our hunting dogs and the existing wild canids probably evolved from some common ancestor. However, by selective breeding we have altered the hunting strategies in our domestic dogs to suit our use. Our pointing breeds point instead of pounce because we have extended the pause and depressed the pounce seen in the naturally selected strategy of a fox hunting mice. Retrievers run or swim down a crippled bird and bring it to be put into our game bag rather than into theirs. But, on occasion, our pointers do pounce, our retrievers do eat or buy game just as in some ancestral form of hunting.
However, to say selective breeding has altered the behaviour to satisfy our wants is to imply a strong genetic determinism and so ascribe to genes the overriding control of everything. It implies there is a gene for pointing so we must then assume there is, or at least some of the time in the historical past has been, the allele for not pointing. To carry this to the extreme then, retrievers have a gene for not pointing and a gene for retrieving. Hounds have a gene for not pointing and a gene for not retrieving. This is overly simplified and somewhat ridiculous. However, the point is genetic control of behaviour does not work in this deterministic way.
A behaviour that we can see such as pointing is in reality the way in which a genetic characteristic, the genotype, manifests itself, how it appears. The behaviour we see is the phenotype. If we know the genotype of a dog, we know it possesses some particular (genetic) characteristics. Knowing this enables us to predict with some statistical accuracy the phenotype of the animal as it will appear in some specified population (a line within a breed, or a breed itself) and in some specified environment. However, knowing more about the upbringing and education of the dog would lead to a more or less revised statistical prediction of phenotype. So when trying to predict phenotype, a behaviour such as pointing, genetic control is not the begin all and end all inviolate determinator of behaviour. It is not like a program in a computer which munches up the entered data the same way each time. Genetic control really means that a dog possessing a genotype for the expression of some specific behaviour is more apt to show that behaviour than would a dog without the genotype for that behaviour. It does not mean the dog possessing the genotype must behave in some certain locked-in way as specified by the genotype for that certain behaviour.
Genes really control probabilities; they do not control the behaviours themselves. A gene for a given behaviour, say pointing, is really a gene for the morphological, physiological and psychological states which tend to produce or are required for that behaviour. But because we always talk in terms of selection for a certain behaviour pattern, we are bound to postulate, both implicitly and explicitly, genes for the behaviour we are selecting. In reality there is no such thing as a gene for a specific behaviour.
So, if all the generations of selective breeding for specific behaviour has been based on non-existent genes and we arrived a the behaviour we wanted anyway, what have we selected for to give us point instead of pounce or retrieve instead of eat or bury as well as other behavioural traits we wanted in our hunting dogs? We have selected for the only things we could select, morphological, physiological and psychological characteristics.
These are the genetically controlled characteristics which appear outwardly as certain phenotypes, those certain behaviours we want.
Morphology of a pointing dog, its shape, size, bone structure, its coat colour and density, hair type, feet shape, the development and distribution of muscle mass, chest shape and size, are all selected for and can easily be contrasted and compared to any of the retriever breeds and the differences and similarities are obvious.
Physiological attributes such as scenting ability or quality of nose, fat deposition, type of and production level of sebaceous (oil) glands, general metabolic rate, tolerance of temperature extremes are also selected. Though not so obvious as morphological differences, these physiological characteristics are more easily seen than are the psychological ones.
By selection of psychological characteristics we are able to make point instead of pounce, retrieve instead of eat or bury. In both cases selection has been to increase the characteristic in the dog which we could call cooperation. There is no gene for cooperation as such, but the neural and neurophysiological “hooking up” – the wiring in the dog’s brain – is such that the dog does it for you rather than for himself. He is being altruistic.
The neural, neurophysiological aspects are transmitted genetically. Because the dog possesses the behavioural characteristic we want, that is, he has the correct phenotype as a result of correct genotype, he is used for breeding more than the dog which does not possess the correct phenotype. By selecting for phenotype, that is, cooperation manifested as pointing or retrieving (or both), we increase the probability of genes necessary for this phenotype and decrease the probability of alleles for non-cooperation. There are many other behavioural characteristics we can select for, pointing and retrieving are only two. But all would work the same way. We select for alleles on the basis of their phenotypic effects. What the genes control, physical, physiological or psychological properties
Either alone or in combination determine the behaviour of the dog.
What natural selection took millions of years to establish in the wild canids, selective breeding has turned to our advantage in a relatively short time. In most breeds, development is only a few hundred years old or less. This development has let us take the “natural” behaviours like the pause before pounce and the eat or buy what is caught and to bend or mold them through selective breeding to the behaviours we now have. The fact that pointers still pounce and retrievers still eat or chew up without swallowing, or bury, only means we still have alleles present for non-cooperation and under certain conditions they reappear as a reminder that we haven’t yet reached our behaviourally perfect dog. But the genotype also makes possible the training with which we can push the phenotype toward its limit, toward the more perfect dog. So with learning (training) the phenotype can exceed the apparent limit imposed by genotype because we can, to some extent at least, mask the phenotype.
Return to Main Page
Updated 4th January 2005