Polygenic characters tend to follow a pattern in which most individuals under study fall around the center or mean, with fewer at the two extremes. Taking wither height as an example, there are few really small animals of a particular breed and few really large ones, with most around the middle height. The number of genes which, with environmental influences, control height may be unknown, but selection can still be undertaken; many breeds have been altered or 'improved' without breeders knowing the number of genes involved.
A dog breeder seeking to improve or increase a particular trait must first identify those members of the dog breed which excel in that trait and then mate extreme examples. Progress depends on two features. One, called the selection differential, is the extent to which the selected group (parents) exceeds their population mean; the other, called the heritability, is the extent to which the character under examination is inherited in an additive fashion. If a trait is not highl y inherited, even intense selection will give little response; in contrast a highl y inherited trait will not be altered if minimal selection is made for it.
Most dog breeds have been altered over the years, some out of all recognition, such as the Bulldog, and at the same time new dog breeds,like the Rottweiler and the Dobermann, have been created. Selection for particular features under genetic control has been instrumental in both instances, and new breeds have involved cross-breeding followed by selection for a specific type. Not all selection has been desirable, and the show ring has encouraged exaggeration in some aspects. There is, for example, little doubt that selection for particular eye shapes has resulted in inturned eyelids (entropion) in some dog breeds like the Chow Chow and Bloodhound, and selection for broad heads in Boxers and Newfoundlands has resulted in additional incisor teeth in the upper jaws. Extra teeth may be a minor issue, but entropion is painful and necessitates surgery. Exaggerated shapes have led to spinal abnormalities in some dog breeds and to respiratory problems, heart disease, bone problems, etc., in others.
Altering one part of a dog may result in correlated changes in another part. Reducing wither height and body weight, for example, is likely to bring about a reduction in litter size. In general, but excluding giant breeds, there is a tendency for larger taller breeds to produce large litters, but for litter size to be associated positively with mortality.
Breeding of dogs should be done only after considerable study, and before attempting to change a breed by selection, the dog should be considered as a whole. Improving hip status is laudable, but if done in isolation it might result in retardation in other aspects.