Artificial selection begins when a breeder selects what to him are the “best” animals to be parents and deselects those that aren’t. The expectation is that what makes animals “best” will be passed to their progeny, while what isn’t “best” is removed from future generations as much as possible. In some industries, artificial insemination or embryo transfer introduces “best” into a herd through genetic material rather than physical animals.

Animals selected for breeding may be deselected in turn as even better animals are brought in later. Replacement selection­ is when animals are chosen to be parents for the first time in a breeder’s herd (they may already have produced progeny elsewhere), while animals removed as parents are said to have been culled from that herd. (Culled animals may still be sought-after by other breeders to be first-time parents in their own herds.)

Phenotypic selection is when animals are chosen purely by how they look or perform as individuals. For example, mate biggest to biggest, or fastest to fastest, and repeat. After a few generations you may well have animals bigger, or faster, than the ones you began with. Or not, as anyone who’s mated the same two white alpacas of unknown ancestry year after year and had every colour but white may know!

Many breeders also take into consideration phenotypic records of relatives.

For example, a dog breeder may like the visible phenotype of a puppy, but draws further on pedigree data to assess what other qualities may have been acquired through its parents from their ancestors. This approach is subjective, as nothing can be measured definitively.

A sheep breeder on the other hand may go the other way and consider progeny data rather than ancestral data. Traits like birth weight, weaning weight, fleece weight and micron are routinely measured in commercial industries, and vast amounts of data accumulate this way. This is a more objective approach. A sire who is consistently producing quality, measurable, offspring is likely to keep doing so. But there is still a degree of subjectivity, in that gains may not be as strong if the new dams themselves are not as “best” as previously covered dams.

Whether subjective or objective, the end goal is the same: assessing the ability of a particular animal to pass on the “best” qualities, based on its and its relatives’ performance (phenotype). In other words, assessing that animal’s “breeding value”. Using relatives’ performance helps increase the accuracy of such predictions: if ancestors or progeny show very similar characteristics to each other then an individual animal closely related to those is also likely to be similar. And if closely-related animals have similar performances, it follows that the phenotype of interest is highly heritable. We’ll be covering heritability and the measurement of breeding value in much more detail in future posts.

Once animals have been selected the next step is mating, or deciding which of those selected animals to put to whom. We’ll cover this in the next post.