The problem with fixing traits through conventional breeding

Today comes a rather theoretic post on the obstacles of breeding, which is relevant for “breeding-back”. One of the biggest challenges for “breeding-back” is to genetically fix the desired traits. By fixing traits I mean achieving that all the individuals of the population have the desired trait. For example, the E+ allele is fixed in Taurus cattle because all individuals are homozygous for this allele. Fixing a trait requires consequent selection and to consider the difference between genotype and phenotype. 

 

As an example: Heck cattle of the Wörth/Steinberg lineage have very large thick horns that match those of the aurochs in absolute and relative dimensions. However, their curvature does not curl inwards enough in most individuals (see this individual). Maronesa sometimes have a very narrow, aurochs-like horn curvature, but lack the desired volume (see this individual). Something intermediate would be ideal. And considering that those traits are very likely regulated by more than one gene, the results of crossing Wörth/Steinberg Heck cattle with Maronesa would likely be intermediate, and hence show the desired phenotype: horns with the right volume and right curvature. Is the work done with that? Not even nearly – the desired phenotype would have to be genetically stabilized in the individual and to be fixed in the population. Simply backcrossing the offspring with the individual with the desired phenotype again and again will not be able to do that, and here is why: 

The cross individual might have the desired phenotype, but that phenotype is the result of maximum heterozygosity between the parental breeds. It has the alleles for the horns that are not curled enough and the alleles for the narrowly curled horns on the same loci, and it has the alleles for the very large horns and the not so large horns on the same loci, resulting in a phenotype that is intermediate and hence “perfect”. When the animal produces offspring it either passes on the alleles for the very large horns, the not so large horns, the not very curved horns or the narrowly curled horns, but it will never pass on the desired phenotype because it is the result of a heterozygous state. Thus backcrossing with the F1 individual with the desired phenotype will be pointless because it is genetically unable to stabilize the phenotype. By stabilizing in this case I mean that the individual always passes on the desired phenotype to its offspring, which is the case when the alleles producing the desired phenotype are present homozygous. When the phenotype is the result of heterozygosity, stabilizing it is impossible. However, since both horn volume and perhaps (or: hopefully, for this example) horn curvature are quantitative traits that are regulated by more than one locus, there is the chance to genetically stabilize the desired phenotype. As a simplified example, let us assume that horn size is controlled by only two loci, A and B, (unlikely, but only an example) and that Wörth/Steinberg Hecks are homozygous for alleles producing very large horns on both locus A and B, and that Maronesa are homozygous for alleles producing not so large horns on A and B. Thus, they only pass on very large horns or not so large horns, it is stabilized in their genome. What we want is a stabilized intermediate phenotype by crossing those breeds. The F1 individual will have the right phenotype because it is heterozygous for both very large horn alleles and not so large horn alleles, producing the intermediary phenotype desired. But, as mentioned above, it is impossible to stabilize this phenotype with this genotype, the genotype is not right yet. The right phenotype in this individual is actually only an “illusion”. The only way to stabilize the intermediary phenotype is to produce offspring that is homozygous for alleles producing very large horns on either locus A or B, and that is homozygous for not so large horns on the other locus. This way, the intermediate (= desired) phenotype can be passed on in a stable fashion. If horn curvature is a polygenic trait too, it works the same with horn curvature. It is impossible to have that in a first-generation hybrid. Breeding this F1 individual to another breed or one of the parental breeds will not stabilize the desired phenotype, even if you backcross it with the F1 again and again and again. This is why “breeding-back” herds are still heterogeneous and not genetically stable even if the same breeding bull has been used for multiple generations. The most efficient way to stabilize the desired phenotype is to breed the F1 individual to another F1 individual of the same combination (if you have only breeds A and B, in this example Wörth/Steinberg Heck and Maronesa). This way it is possible that an individual is born that has only alleles from Heck on one locus and only alleles from Maronesa on the other locus, if the goal is to stabilize the intermediary phenotype of a quantitative trait (f.e. body size or horn size). If you have such a true F2 individual with the right genotype (the more individual of this combinations you produce the less luck you need, the less individuals the more lucky you have to get for that), it is possible to fixate the trait in the population by backcrossing its offspring with it. 

Summa summarum: If you cross two breeds because a phenotype intermediary between both breeds is the phenotype you desire, the F1 individual will not have the right genotype for stabilizing the trait in the population, no matter how long you use it as a sire. If you produce a true F2 of this combination, the offspring has at least the chance to have the right genotype. 

If you are looking at a monogenetic trait, and you want to stabilize the intermediary phenotype that is the result of a heterozygous state, you will never ever be able to stabilize the intermediary phenotype in the population. To use Wörth/Steinberg Heck cattle and Maronesa as an example again: the former have horns that often are too upright for what is average in the European aurochs, while the angle of the horns is too narrow in most Maronesa. Let’s say the parental individuals have horns of an angle of 90° (Heck, too upright) and 35° (Maronesa, too narrow angle) and you want something intermediary to produce phenotype like in the European aurochs. If horn orientation relative to the snout is a monogenetic (Mendelian) trait, and the heterozygous state produces something intermediary like 60° as in the aurochs, it is impossible to stabilize this in the population, because the F2 will either have the 90° phenotype, the 35° phenotype or the desired 60° phenotype. You may be happy about the individuals having the right phenotype, but will never be able to fixate this trait in the population using these two breeds if horn orientation was regulated only by one locus. You would have to find a breed that has the right allele, as long as the right allele was not lost during domestication. I do not think that horn orientation is a Mendelian trait, but just to illustrate what I am saying, it would be impossible to fix a desired phenotype using two parental breeds that are homozygous for a phenotype that is not the desired phenotype. 

It gets more complicated if there are more than two alleles, f.e. three: let us say we have only one locus with three alleles, allele A producing the right phenotype when homozygous (60°), allele B producing very upright horns (120°) when homozygous, and allele C producing a very narrow angle when present homozygous (20°). Now let us assume that we have two parental individuals, one with horns having a 90° orientation and a genotype A//B (intermediary between the phenotypes A//A, the desired one, and B//B, the very upright horns), and one with a horn orientation of 35° with a phenotype A//C. By crossing these two animals, which both do not have the right phenotype, you would produce the following genotypes and phenotypes: 50% would be A//A, the right phenotype of 60° and right genotype, B//C would have a phenotype intermediary between 120° and 20° so perhaps the desired 60° but the wrong genotype, A//B with 90° as in one of the parents and A//C with 35° as in the other parent. If you produce only one offspring individual, the genotype will come about by chance. Either you have an individual that has the wrong phenotype and will be discarded, or you have the right phenotype but do not know if the genotype is right (A//A) or wrong producing the same or a similar phenotype (B//C). You can only know by producing further cross generations and analysing the offspring, when it is already too late if you picked the wrong individual. Thus, either the genetic background of traits is resolved (the individual alleles and the phenotypes produced are known, which is currently only the case in colour traits), or you have to accept that you also need a lot of luck or in some cases, stabilizing the trait is impossible with the breeds chosen because it is the result of a heterozygous state in a monogenetic trait.

Summa summarum #2: In a monogenetic trait, you need to have the one right allele in order to stabilize the fix phenotype. If the allele has been lost during domestication, it is not possible to stabilize the desired phenotype. 

 

Another problem for fixing desired phenotypes is genetic linkage. It is very likely that some domestic (= undesired) alleles are genetically linked with wildtype alleles (= desired) on the same chromosome. This can be a problem for “breeding-back” in two ways. For once, it might be the case that some desired phenotypes are genetically linked with domestic phenotypes in some breeds, f.e. the right colour might be linked to a wrong horn shape or other examples. In this case, a breed would have to be found in which this is not the case or you must hope to get very lucky with recombination. Another problem could be if a number of wildtype alleles for wildtype traits that are not morphologically visible (f.e. physiological traits) are genetically linked with visible mutations such as “wrong” colour variants or deviant horn shapes, the selection for aurochs-like morphology or looks might eradicate other wildtype traits for physiology, genetic fitness or immunology. This would be a case of bad luck, but cannot be ruled out completely. 

 

This is why it is a huge pity that there is currently no project that is trying to execute “genetic breeding-back” (go here for a post on that issue). For “genetic breeding-back”, the resolved full genome of that one British aurochs bull could be used as a template in order to accumulate wildtype alleles responsible for all aspects of the organism (physiology, morphology and life appearance, immunology, development, genetic fitness etc.), and produce something that is as aurochs-like as possible with living animals (not only in morphology, but also all other organismic traits). This would require a lot of extra research that has not been done yet, and it is problematic that we only have the genome of only one aurochs, which bears the danger of selecting against other wildtype alleles not found in that one particular genome.