RIP performance recording

By Nicola Dennis

I would like a dollar for every time I hear someone give a eulogy for good ol’ performance recording. It seems many people believe that genomic testing means that we can soon take a break from recording and scoring animals. It’s a great dream, but here are a couple of reasons why performance recording is here to stay.

Firstly, phenotyping is relatively cheap and it works.

Multi-trait selection has been around for nearly forever. Over thousands of years, humans domesticated and adapted over 700 species. The majority of this happened at a time when we had very little inkling about how offspring came to inherit the traits of their parents. One leading theory of the time was that tiny pre-formed babies rode on the backs of sperm (microscopes were not great back then). Yet, armed with very poor genetic theory, humans contorted wild canines into at least 350 breeds of dogs.

You don’t need to understand much to run an effective breeding programme; just which traits you would like to improve and a method for choosing between the animals you already have on hand.

Scientists REALLY like to know how things work, but the sad fact (for scientists) is that you don’t need a genome wide association study to create a Chihuahua, all you need is some dogs and an unwavering passion for tiny violent animals.

Secondly, the genome is a big, complex place.

At its most basic level, a gene is merely the assembly instructions for a protein. When a gene is activated inside a cell, the cell keeps pumping out copies of the desired protein until something asks it to stop. Proteins can stay in their home cell, move to a neighbouring cell or pass into the bloodstream and affect a lot of cells at once. Furthermore, it is rare to find a protein that does “just one thing”. Proteins bounce all over the show interacting with other molecules and turning genes on and off all over the place. Some even ride on the backs of sperm.

The whole thing is complex, and no one truly understands it all. The genome is kind of like a phone book for tiny workers. If you have a problem, then you can use the phone book to find someone to help you with your problem. If you have a small, simple problem (i.e. “I need to ask my consultant a question”), then you only need to find one number to ring. If it is a large problem (i.e. “World peace”), you are going to have to coordinate a lot of different people in that phonebook.

Some traits in animal breeding are simple problems because they are usually only affected by variations in a small number of genes (e.g. horns, double muscling etc). If these qualitative traits are all you care about, then you only need to monitor one or two places in the genome to get what you need. Basic genotyping will get this job done.

However, most traits are affected by a lot of different genes (e.g. growth rate, milk production). In fact, the models we use to analyse these traits assume that there are an infinite number of genes involved (i.e. we expect to ring everyone in the phonebook to improve the trait). It is difficult to pick one gene at random and determine its worth to the whole system just by looking at it. Whether or not a variant of a gene is bad or good depends very much on context. Performance recording (i.e. phenotyping) is really the only way to get that context.