Is "genetic breeding-back" possible?

It has become en vogue in contemporary “breeding-back” to claim that their project also involves “genetics” or is based on genetic information, executing “breeding-back” not only on a “phenotypic” basis but on a “genotypic” level, which should lead to a more authentic result. The argumentation is that the original “genes” (simplified language, actually we are talking about alleles) of the aurochs have not been lost during domestication, but split up and mixed with domestic mutations among the modern domestic cattle populations. The goal of “genetic breeding-back” is therefore to trace down these genes among modern cattle and to unite them by crossbreeding and selective breeding in one population. Basically the same as with morphological traits, but more in-depth and concerning the whole organism as genetics are the key determinant for all biological aspects. Therefore, this method should lead to an approximation of the original aurochs and not a mere morphological double. This is the theory, and what some projects claim or claimed to do. 

But this is a scenario communicated in press releases. In order to see how much of that is in line with what is to be expected in a fact-based approach to the subject requires to have a look at the following questions: Where do aurochs and cattle differ genetically? Is it likely that domestic cattle preserve all the genetic material of the aurochs? Has the aurochs gene material been traced down in modern day cattle? Do any projects actively execute selective breeding on this genetic material? 

This is not a discussion where we can rely much on “hard facts” given in peer-reviewed publications but much of it depends on interpretation, predictions and deductive reasoning. This post presents my personal take-on to this subject so it is merely my personal opinion, but I give reasons for every statement (and, if necessary, also literature references). In short, I think the answer to the questions above is rather disillusioning concerning the possibility of executing “breeding-back” on a genetic level. 

Problem 1: Where do aurochs and cattle differ genetically and are all aurochs genes still alive?

First off, we have to elaborate where aurochs and cattle will differ genetically. The complete genome of a British aurochs has already been resolved, and a comparison to modern cattle has shown that apparently selection on genes evolved with milk production has taken place₁. This is not surprising, but only the tip of an iceberg. Domestication alters the whole organism and these changes are morphological, developmental, behavioural, endocrinologic and genomic. Not only some new mutated alleles such as for new colour variants popped up during domestication, it altered the development of the whole organism. The timing of the ontogeny has been affected, some developmental changes are stopped in the process (see paedomorphy), others speeded up (see earlier maturity). These changes are caused by genes regulating endocrinologic system (in essence, much of the ontogeny of an animal is regulated by endocrinology) and the nervous system and there is a mutual interplay of both. An organism is one functional entity and you cannot look at just one aspect only. These genes probably cause most of the fundamental differences between a domestic animal and its wildtype in development, morphology and behaviour. New mutations directly affecting morphology, such as new colour alleles, play a role as well of course. How many genes are we talking about in total? It is hard to give a precise number or only an approximation as those particular genes involved in all these developmental, endocrinologic, neurological, and morphological and their individual function have not been identified yet. But the number might be very or even very, very high. For comparison, body size in humans is suspected to be influenced by about 50 loci (genes) alone that have a more or less significant impact on this one quantitative trait according to  Visscher 2008₂. Other sources give even way higher numbers like 6000 loci influencing the metric dimensions in mice₃. Dobney and Larson 2006 write that the thyroid production, which affects many morphological aspects and produces typically domestic behaviour and morphology traits, is regulated by a “myriad of genes”₅. So my personal estimate that the number of genes regulating all the developmental, endocrinologic, neurologic, and morphologic differences between domestic and wildtype, cattle and aurochs, might be hundreds, or even thousands of loci (even considering that pleiotropy and cascades probably play an significant role as well). 

As I wrote above, we are in a speculative area here as the individual genes and their exact functions in this complex interplay of factors that shape a living organism have not been identified. This is why publications only give approximations of numbers of genes that might be involved in examples such as body size. In short: we do not know on how many loci (genes) aurochs and cattle differ but they might or should be of considerable number. Furthermore, it has not been identified yet which those particular genes are. 

Aurochs versus domestic cattle: the defining key genes 

What are those studies allegedly comparing aurochs and cattle on genetic basis focusing on, then? For once, phylogenetic markers such as haplotypes. These are used for comparative studies to resolve phylogenetic relationships among species or groups of species and work well for this purpose. Phylogenetic markers are regions in the genome that are barely effected by selection and mutate on a certain rate so that they can be used for resolving phylogenetic relationships. Those include haplotypes on the Y-chromosome, for example, or mitochondrial DNA. In the case of cattle and aurochs, markers are useful when we want to determinate phylogenetic relationships of regional/chronological subgroups and clear up cases of introgression. But when we want to resolve how much and where aurochs and cattle differ on a genetic (and consequently organismic) basis, phylogenetic markers are not suitable because they do not involve the key genetic regions listed above. I think this is the actual misunderstanding in the debate. 

In the most recent publication on this issue, the SNP comparison between the British aurochs and a number of cattle breeds calculated after the Nei distance method, the study focused on about half a million Single Nucleotid Polymorphisms. SNPs do affect the phenotype, as the study itself writes, but considering that mammal genomes can have about 3 billion base pairs, you can do the math how much. Not only is the actual influence of these SNPs on the organism not that considerable, the very small sample size might also result in completely different results when another, and another, and another half a million SNPs are investigated. Breeds that score high in the one result might score low in the other results and so on. That goes as much for Pajuna as for Fleckvieh and even the outgroup Nelore. Furthermore, the Nei distance method is, in my opinion as a layman, not a suitable tool to describe the difference between aurochs and cattle or to resolve the population genetic evolution that has taken place during cattle domestication. I give my reasons for that thoroughly in this post. So the fact that Pajuna scores highest in this comparison does not justify the claim that it is closest to the aurochs in any way. Even though it is morphologically aurochs-like in a number of respects – but the chart obviously does not show a correlation between aurochs-likeness and high scoring anyway, which might also say something. 

Is it likely that domestic cattle preserve all the genetic material of the aurochs? 

No matter how many loci we are talking about, a quintessential question is whether or not we can expect that much or all of the original aurochs alleles are present among domestic cattle. To explain why I write of alleles now instead of loci: in the above section, the question was on how much gene locations (loci) aurochs and cattle differ. Cattle have two variants of the gene, called alleles, per locus. In domestic cattle, new alleles turned up due to mutations and replaced wildtype alleles. The question is to what extent and if, despite this fact, still all or at least most of the defining, influential wildtypealleles can be found within the modern cattle gene pool. By defining alleles, or key alleles, I am referring only to those alleles that were responsible for the organismic differences we see between cattle and aurochs. Cattle and aurochs probably also differed on a number of loci that did not have a considerable influence on the genotype but were just more or less neutral diversity or variation. These loci are not relevant for us in this discussion. Furthermore, it is not necessarily the case that all alleles that form the domestic phenotype are of novel origin. For example, some genes associated with large size in horses are of predomestic origin₆. So it is possible that some domestic conditions on quantitative traits can partly also be the result of cumulative effects due to selection. This, on the other hand, does not increase the chance for all wildtype alleles to have survived in cattle. 

The evolution of domestic cattle from a population genetic perspective 

The scenario proposed by those who claim that indeed all or most defining aurochs alleles are to be found within modern cattle is one of a classic population fragmentation and genetic drift: the population becomes fragmented and split-up as it is spread, new alleles intermingle with wildtype alleles, genetic drift and fragmentation produce an uneven distribution of wildtype alleles among the new lines or clades that have been formed. Population genetics know this as the classic scheme of a fragmented and expanded population. While one population or breed might still have the alleles A, B, C, D, E, F, G, H, I, the other one might have J, K, L, M, N, O, P, and yet another one might have the alleles Q, R, S, T, U, V, W, X, Y, and Z among a quantity of new, domestic alleles. And genetic breeding-back should take these unevenly distributed aurochs-alleles and re-unite them in one lineage, or at least what is left of it. 

But this is not the evolutional process that domestic cattle underwent during domestication. At first, there was a dramatic genetic bottleneck event as the founding population of domestic cattle apparently was very small₃. Then there was massive directive selective pressure that certainly altered the genotype as it altered pretty much all aspects of the organism (development, behaviour, immunology, morphology). And as the farm fox experiment has shown, these changes can happen quite rapidly, even within few generations₄. It has to be assumed that the starting population of domestic cattle was more or less uniform for defining traits of domestic animals; all of them must have been domesticated in a similar manner, including modifications in behaviour (docility, agreeableness, trainability etc.) in morphology (meat, milk etc.) and those that go hand in hand with these changes as a result of pleiotropic effects and developmental cascades (for more, see the dedomestication series). I do not believe that the starting population of domestic cattle was heterogeneous on these traits in the sense that some had the right behaviour while others had a wildtype behaviour, some had an unaltered morphology and so on. Clearly the first domestic cattle were only half-domesticated at some point, but probably all individuals in a similar way. Some basic changes must have been universal to all of them. This is my main point: in the early phase, when wild aurochs were transformed to a domestic state, some key wildtype alleles must have necessarily been lost completely from the population as a result of the directive selective pressure and the more or less uniform result – more or less uniform only in the sense that all individuals of the populations are domestic in the same manner. Certainly some lineages were modified more drastically than others (and this is what we see very clearly in modern day cattle). But the basic alterations that made the animals domestic must necessarily have been universal among those early domestic cattle and still are today, otherwise not all of them would show the same kind of modifications. Consequently it is very likely that the process of domestication irreversibly eradicated some essential wildtype alleles. What happened after this early point were 8000 years of further evolution in human custody: reproductive isolation (more or less), selective pressure on certain phenotypic traits plus genetic drift. It is very likely that another number of key wildtype alleles has been lost or actively eliminated in this process. Or to be more precisely, I think it is highly unlikely that it would not have happened. 

It is clear that domestication is not some discrete step that is completed at some point. Actually, domestication is still an on-going, fluid process: we have numerous cattle lineages and breeds that are way more modified than others, each after a different purpose. In a sense, Fleckvieh is more domesticated than Maronesa, and Maronesa is more domesticated than the cattle of the first hundred generations in human custody (and probably way, way more domesticated despite its aurochs-like looks and hardiness). You could even say that by putting Holstein-Frisian bulls on a Sayaguesa herd or overbreeding highly derived breeds we are still continuing to domesticate the domestic cattle gene pool even further today. But that is not to say that there is a vivid chance of having all defining aurochs alleles still present in domestic cattle to an extent that allows to more or less restore the wildtype. I think that the population genetic processes this population underwent during the last 8000 years do not allow it, or at least make it very unlikely. 

The domestic cattle gene pool was not entirely reproductively isolated as there apparently was occasional regional introgression from wild aurochs. It seems that these introgression events contributed traits helpful for establishing the newly introduced domestic cattle in other regions (immunological and other alleles) and it would, theoretically, be possible that these introgression events reintroduced wildtype alleles previously lost during domestication. However, I think this is not plausible. Local aurochs might have contributed a few advantageous alleles, but it seemingly did not alter the integrity of the cattle as domestic animals. This is not surprising considering the selection policy the ancient farmers must have had: most likely they continued breeding with hybrids that showed increased robustness or whatever trait they considered advantageous, but probably not continued breeding with hybrids that were a throwback from the economic perspective (morphologic and behavioural adaptions shaped after man’s purpose). 

I am convinced that the 8000 years of domestication eradicated a lot of alleles that made the aurochs the wildtype animal that it was and were replaced by new alleles that make cattle the domestic animals that they are. All cattle on this world are domesticated in the same way, just not to the same extent. They underwent the same morphological changes (paedomorphy, reduced sexual dimorphism, reduced brain volume, altered body shape/muscling, proportions, intestine size) and the same behavioural changes (docility, agreeableness, trainability) and consequently also the accompanying neurological and developmental alterations. Not to the same extent in all cattle, obviously, but they are universally present among all domestic cattle on this world and so we have to assume that the genetic basis for these changes is so as well, and therefore that these have replaced the wildtype alleles on the involved loci. This fact plus the population genetic history of the 8000 years of domestic cattle evolution makes it very likely to me that many of the key wildtype alleles that shaped the wildtype organism called aurochs are lost entirely among domestic cattle. This is an assumption, and this assumption could be tested by a rigorous comparison of the now fully resolved aurochs genome with that of domestic cattle. 

Problem 2: Has the genetic material of the aurochs been traced down in modern day cattle? 

Even if the proposal that “all aurochs genes” are still present in domestic cattle and that no original alleles were lost bears any truth (which I am convinced is very, very unlikely), “genetic breeding-back” would require these alleles to be identified and to be located within modern day cattle. First of all, these alleles are not identified. Only some are, like colour alleles, and not even in this case it has been tested if the E⁺allele causing the aurochs colour scheme is really identical to the allele the British aurochs individual had (what if, for example, if such a test would reveal a surprise?). As the example with body size in mice and human shows (see the literature cited above) it is not even known in these well-studied model organisms how many loci are involved in body size. And this is just one complex trait. We cannot even dream of knowing the specific alleles responsible for all the developmental changes resulting in the domestic phenotype we see, we only speculate that they are there. And having those loci identified and alleles identified in aurochs and tracking them down in modern cattle, distinguishing them from domestic variants and even measuring the quantity to which they are present in which cattle clade or breed is just a few (illusive) steps further. Based on the literature we have, and even counting scientifically worthless press releases, apparently nothing has been done in this direction, not even a little bit. 

A number of cattle breeds have been investigated (but not for the important genetic factors explained above), but an important group of taurine cattle have not, probably for their difficult accessibility: Near-Eastern and North African rural cattle. Those cattle, although mostly short-horned and small, have a quite aurochs-like physique and might be less derived than European ones, perhaps also on a genetic basis. Furthermore, turano-mongolian cattle should not be ignored either as they apparently represent a distinct gene pool within the taurine branch on its own (see here). 

Do any projects select on this material?

Even if all necessary key aurochs alleles were present in domestic cattle (which is very unlikely and neither proven nor investigated) and even if all these alleles had been identified (which is not the case) and even if these alleles had been traced down and quantified in a representative sample of domestic cattle (which is not the case either), how would genetic selective breeding look like? 

At first, a set of cattle breeds would have to be chosen that has all includes all the desired wildtype alleles in sum, no matter how it is distributed among the breeds. It might be that you need only three breeds because one breed already has 50% of the alleles, another breed those 50% pls another 10%, and the last one the missing 40%, or maybe as much as twelve or 30 breeds (again, I think that such a scenario is as much as impossible anyway). You would have to know which breed has which alleles. Then you execute crossbreeding. The first generation is heterozygous in any case, so you neither gain nor loose something. In the second generation, however, you would have to check the genotype of the offspring and make sure that you continue breeding only with those that unite a higher portion of wildtype alleles than the individuals of the parental generation do. And so forth, until you have re-united all wildtype aurochs alleles in one genotype. Depending on how many loci you have to deal with you can do the math how long you have to carry out the breeding. I cannot give any number because we would need to know two things: 1) how many loci are relevant in total (as explained above, the number might be very high) 2) how much of the relevant loci with wildtype alleles are already shared in the set of breeds chosen. For example, if the breeds chosen share already 90% of the wildtype alleles, and you only have to unite the remaining 10% that are split-up among the breeds the work is less. This is all speculation, but just to give an idea about the selective breeding work that would have to be done let us do a simple calculation. Let us say the key genes where aurochs and cattle differ compromise 1000 loci (which might be a conservative estimation), and let us say that 90% of the wildtype alleles are already shared in the set of breeds chosen (which is high), the breeding would have to focus on 100 loci. This is a lot.

Are any of the current projects doing something anything like this? There is no hint in any press release of any project that they are doing something that would qualify as a precise genotypic selective breeding as described above. I also see no way how they could, considering that the necessary research work has not been done and that a considerable number of wildtype alleles are probably not present in domestic cattle anyway. If they are not doing that, what do they do?

The only project that is claiming that “genetics” are influencing or will influence their breeding choices and that has presented a little glimpse on what those “genetics” are is the Tauros Project. So far, the Tauros Project has presented a Nei analysis of a number of breeds comparing half a million SNPs to that of the aurochs. You see the resulting chart here. The project claims their chosen breeds (bold) score high – actually it looks like they are distributed quite evenly among the list, and the derived breed Fleckvieh scores high as well. The project writes further that the fact(?) that their chosen breeds score high is a confirmation that their choice criteria were right as they initially chose their breeds based on the “phenotype” and not genotype (which is funny since the project actually claimed to choose their breeds based on genetics in early years). 

As I wrote above, SNPs only have a minor influence on the organism as a whole, and so do haplotypic variations on phylogenetic markers. They are useful to resolve relationships between phylogenetic clades but not of much use when determining the organismic differences between a domestic animal and its wildtype. Furthermore, the Nei distance method is not the appropriate tool resolving the complex population genetic evolution that domestic cattle underwent but rather for a textbook example of simple population fragmentation. Thus it does not surprise me at all that there is no correlation between neither morphological primitiveness nor hardiness/robustness in the chart. Apart from that, a number of important or at least interesting types of taurine cattle have not been tested (Near-Eastern cattle, Turano-Mongolian cattle). To conclude that “genetics obviously do not matter” based on this fact would be the worst assumption to make and would show a lack of biologic understanding. Genetics of course determinate what the organism is going to be like, and genetics are of highest importance in animal breeding and in determining the differences between a domestic animal and its wildtype. 

What you see is an analysis of genetic material that is not of high relevance of the organismic aspects we are talking about, analysed with a tool that is not really appropriate for the evolutional process we are looking at. If you say “well, but it is the best we currently have so we have to go with that”, I can only counter that you have to be aware of the fact that the “best we have” in this case does not say much if anything. Thus, if the Tauros Project is indeed giving any importance to the “genetic similarity” in the form of SNPs or phylogenetic markers in their breeding policy and if indeed future generations of Tauros cattle by coincidence* score a little higher in this respect I honestly have to say that it would not impress me that much. I go even further: I predict that a domestic cattle population that is even identicalon all of the half a million SNPs and phylogenetic markers (that is, neutral variations on gonosomal or mitochondrial sequences) will not be any closer to the aurochs on an organismical basis that is recognizable in the phenotype(including not only morphology but also all other aspects of the organism). It is only my personal opinion but I am rather confident with my statement for the reasons I gave here. 

* I found nothing so far that hints how the SNPs will influence their selection scheme. So if future Tauros cattle indeed show a higher match, which would surprise me, it would probably be coincidence. 

I would say it would be honest and fact-based if you do not claim that you execute breeding-back  on a genetic level unless you: 

- have identified the loci responsible for the organismic differences between aurochs and domestic cattle

- have identified the individual wildtype alleles on these loci and have traced them down in modern cattle 

- have collected a set of cattle breeds that evidently contains those wildtype alleles in sufficient quantities 

- breed on a genotypic and not phenotypic basis by screening the genome of each individual and select in a way that more progressed generations contain a higher number of wildtype alleles than the parental generation

Unless you do not do that, I would say that you do not “breed-back” on a genetic level. Thus, what has been presented so far does not really work for “genetic breeding-back” in any way that is relevant. But it does work as a fig-leaf in press releases to back-up the statement that you work on a genetic level and to give the whole a scientific touch. It apparently works well, as the claim that “Pajuna is genetically closest to the aurochs” is already quite widespread in the internet.*  

* I am not saying that Pajuna is not aurochs-like, far from it, I am just saying that the work that has been done does not sufficiently endorse this statement. 

What can be expected from current projects then?

If no current project is doing a true selection on the key genes of the aurochs, which are not even identified yet let alone located in living cattle and possibly a large number of them is not retained in domestic cattle anyway, what can we expect from “breeding-back” then? Actually, the same as we did before some projects started to include “genetics” in their public relations in 2009: a population of robust cattle that shares a number of aurochs-like traits in external features (such as body shape, coat colour, horns, size etc.) that is also hardy and robust at the same time so that it fulfils the ecological niche of the aurochs in its natural environment but still they will remain domestic animals which will be recognizable in traits shared by all domestic cattle on this world (reduced brain volume and sexual dimorphism, modified morphology and altered developmental biology et cetera). Basically, they will be bovine Tamaskans (what that is supposed to mean will be explained in a future post) but they will do the ecological job sufficiently and look/work authentic at the same time. Which is actually pretty much and a desirable goal which I am looking forward to. 

This will not be the last post analysing the differences between aurochs and cattle on an organismic basis. I am planning to do some more and they are in preparation. 

Literature 

₁Orlando, L.: The first aurochs genome reveals the breeding history of British and European cattle. 2015. 

₁Visscher, P.: Sizing up human height variation. Nature publishing Group. 2008.

₂Kemper, K., Visscher, P., Goddard, M.: Genetic architecture of body size in mammals. Genome Biology. 2012.

₃Bollongino et al.: Modern taurine cattle descended from small number of near-eastern founders. 2012. 

₄Dominic Wright: The genetic architecture of domestication in animals. 2015. 

₅Dobney & Larson: Genetics and animal domestication: new windows on an elusive process. 2006. 

₆Schubert et al. 2014: Prehistoric genomes reveal the genetic foundation and cost of horse domestication.