Marker Assisted Selection: a Fast Track to Increase Genetic Gain in Plant and Animal Breeding? Session Ii: Mas in Animals

Selection under the infinitesimal model Standard animal breeding techniques have been largely successful in improving the performance of domestic animals in the last century. The success of these techniques finds its roots in the theoretical work of Fisher and Wright who showed how selection, migration and mutation could be derived from the simple Mendelian laws. All that theory was then transferred into the operational application in animal breeding by J. L. Lush (Animal Breeding Plans, 1945) and by his many “sons” and “grandsons”. In more recent times another landmark book, Introduction to Quantitative Genetics, (Falconer, 1960) probably represents the official birth of the quantitative genetics in plant and especially animal breeding. Plant and animal breeding have based their tools of investigation on this “central paradigm for the analysis of phenotypic variation” (Lynch and Walsh, 1998) even if their methods have radically diverged probably because of the different population structure in crop and domestic animals. Since the 1950s new techniques based on the mixed model methodology (C. Henderson) had a large impact in predicting the individual breeding value of the animals. But it was not before the 1970s that the development of high-speed computers ensured a widespread application to cattle populations and later to other species. The great achievements in cattle and pig selection during the last 50 years largely relied on quantitative genetic theory and infinitesimal genetic model. Most of the traits selected show an additive variation which justifies the effort to capitalize on the additive effects summing up, generation after generation, their infinitesimal small contributions. Under infinitesimal model, selection is therefore a sort of blind process where genotypes are progressively modified without any real knowledge about gene number, location, effect and frequencies of the favorable alleles. The whole process of predicting the breeding value of the animals to subsequently select the genetically superior parents of the next generation is entirely based on sophisticated computations. Complex statistical mixed models can adjust for all sort of environmental factors, maternal effects or repeated records and simultaneously release the requested individual additive genetic merit. MARKER ASSISTED SELECTION: A FAST TRACK TO INCREASE GENETIC GAIN IN PLANT AND ANIMAL BREEDING? SESSION II: MAS IN ANIMALS 36 Standard quantitative selection requires therefore field data collection on: • Individual phenotypic performance • Expected covariance among animals due to blood relationship between them Recorded traits so far have been mainly limited to production performances. But even for “simple” traits like milk yield, there may be regional or national differences which make comparisons difficult to manage. Phenotypes may also be sex-limited and recorded late in the life of an animal. Recording is also quite expensive and in many countries costs are largely covered by National Governments. In Italy, for example, the total budget for selection granted by the Ministry of Agriculture is around 93 million euro, 77 of which is field phenotype recording. Costs of phenotype recording are also quite variable in different species and the value of a phenotyped animal may not justify the expense. As an example, fat and protein content in sheep milk requires the collection of a milk sample which may represent more than 10% of the daily released milk. Recording requires also efficient extension services covering a relevant part of the farms, a structure which cannot be created quickly and cheaply. Parentage ascertaining is also extremely expensive. In cattle, where AI is largely used, the paternity of an animal is easily predictable. In species where natural mating is predominant, as in sheep, parentage definition is a crucial step in selection, but typing costs may be a relevant part of the certified animal. In recent years infinitesimal model has been also applied to other functional traits such as longevity, calving ease, milkability or udder health. In perspective, we can foresee an increasing interest in traits related to safe animal products and to food quality for human consumption. Many of these traits have a limited genetic variability and a high cost of recording, for example, pork meat quality for dry-cured ham, CLA, parasite resistance, immune response and so on, which justify the use of different genetic models. Selection under a finite locus model In the last two decades the development of molecular genetic techniques has allowed the mapping of some loci affecting the quantitative variation of traits of economic importance. Today their subsequent exploitation in selection is in some case a concrete reality. The primary interest of animal breeding and selection in molecular genetics is in finding all the genes (QTLs) which control a trait, being able to distinguish among their alleles and consequently exploit the variability for faster genetic gain in that trait. In such a scenario we would not need to phenotype the animals, nor wait until the phenotype is expressed. We presently are at the very beginning of this. In very few cases we have reached the target, hitting, in a single gene, the mutation responsible for the protein modification which causes the quantitative variation of some trait. In these cases prediction of breeding value takes advantage of both the polygenic (infinitesimal) component and of the discovered QTL (the finite locus) by fitting ad hoc models. More frequently we have some evidence that in a chromosomal region of some cM of length (where several hundreds of genes may be located), a certain animal is probably heterozygous at one or more genes (QTLs) involved in a quantitative trait. Traditionally two approaches are available to locate a QTL: