Tuesday, 1 August 2017

The genetic and developmental background of visible traits

I have been planning to do such a post for quite some time now, and have gathering information for several months. Finally I had the time to put it all together. I think it is a really fascinating subject that can tell us a lot about how to achieve the traits we want our “breeding-back” cattle to display.
Genetics are the most important key to animal breeding. In fact, animal breeding is nothing but applied genetics. Genetics and developmental biology determine everything that is possible or not possible in breeding (leaving aside environmental factors) and therefore it is important to look at the genetic and developmental backgrounds of the traits that are of importance for the breeding goals that are to be achieved. In this post, I want to give an overview on how genetics and developmental biology influence breeding and how visible traits come into shape. As genetics are not my major in biology (but zoology and soon also ecology),  I am glad to be corrected if I got something wrong here in this post.

Yes, environmental factors, but…

Of course environment is essential for the becoming of an organism, but I will leave it aside in this post. Surely, if animals are constantly malnourished from their birth to death, you cannot expect them to develop in the same way as their genetic potential enables them to. But I am going to neglect environment here, because for this post I assume that all of the animals kept in breeding-back are kept under appropriate, sufficient conditions, so that this factor is eliminated (feral populations, however, might be an exception).

How genes are inherited

It should be common knowledge to every school kid that genes are organised on chromosomes, and that chromosomes are the unit of inheritance. Cattle have a diploid karyotype, at least taurine cattle have 29 chromosomal pairs. Each set of chromosomes is inherited from one parent respectively, and passed on by chance to the offspring. For single loci, the Mendelian rules apply. For the whole genome, one would have to imagine a kind of bell curve. I.e. that means that when you breed two F1 individuals made of breed A and B to each other, it is possible that you would receive a pure individual of either breed A or B for the second generation – but statistically most unlikely. However, it is also usually not the case that F2 are an even mix of both parental breeds, but usually happen to lie somewhere on the bell curve.
The goal of breeding is to unite all desired alleles homozygous in one population and get rid of the undesired alleles. So you have to keep in mind chromosomal genetics when choosing animals for breeding. For details, go here or here. Chromosomes can also put some challenges for the desired goal, f.e. if the loci of two desired traits happen to be on the same chromosome, and breed A has the desired allele for one locus, and breed B has the desired allele for the other. In this case, you will never be able to unite the desired traits homozygously and therefore stabilize both traits in the population, unless you have luck with recombination (and then again luck in picking the right individual).

Qualitative vs. quantitative traits

There is a traditional difference between the so-called qualitative traits and quantitative traits. Qualitative traits are regulated by only a handful of genes or even just one, what makes the Mendelian rules easily discernible. A prime example for qualitative traits are the colours of animals, which are controlled by comparably few genes. That is why every animal breeder will confirm to you that it is easiest to breed for colour characteristics. For example, there is the famous Extension locus that has three known alleles in domestic cattle: Ed, E+ and e, in the following dominance hierarchy Ed > E+ > e. The first allele, just to pick one, produces an excess in black pigment, and the whole animal ends up black. If you see a completely black bull, you can be quite sure that it has at least one Ed allele. And so on. Therefore it is quite easy to breed for qualitative traits as long as the loci and alleles are known (in cattle, the colour genes are unfortunately less resolved than in other domestic animals, which is probably because there is more interest in other domestic animals which are not considered the mooing pre-stadium of hamburgers).
Quantitative traits, on the other hand, are different. They are based on many different loci which all have a cumulative or additive effect on the quantity of the trait. Classical examples are body size, horn size and others. No matter how you define body size (measured in length, height or weight), the number of genes influencing it is way larger than in the case of coat colour. Human height seems to be influenced by 54 loci according to Visscher 2008, each locus contributing a bit to the phenotypic variance (0,3-0,5%)1. Height is a two-dimensional measure, for total size (I assume mass) Kemper et al. speculate that even more than 6000 genes might be involved2 (many of them would have an only minor contribution of course). Assuming that the genetic architecture of human height/size is comparable to cattle height/size (not completely, of course), we can take 50 as a number to work with. If you have such a high number of loci that have to be right, breeding takes rather long when you have a mosaic population. So if we have a bull that has the right colour, which involves about a dozen of loci, but not the right size, and a bull with the right size but the wrong colour, which bull should be preferred as sire? From the genetic perspective, definitely the bull with the right size but wrong colour because he has 50 genes right and only about a dozen wrong, while the bull with the wrong size has only a dozens of genes right while 50 are wrong. Therefore, the bull with the right size is about five times as valuable from the genetic point of view (that’s why I opted for keeping this well-built half Chianina bull in the Lippeaue herd despite its diluted colour in 2015). Of course there are also single genes that have a dramatic impact on body size, such as those causing achondroplasia in humans or numerous other forms of disproportionate or proportionate dwarfisms and gigantisms in humans and animals.
The genetics of horns is badly researched in cattle, because there is only one big commercial interest in cattle horns: present or not? Two loci that regulate the presence of horns have been resolved, called Polled and Scurred. There are no studies on how many different loci might be involved in horn volume (I usually divide horn size in two different factors: length and diameter), but I would be surprised if they are less then all the colour genes in sum (for details, see this website). So if you have a population of cattle with undesirably small horns, it would be wiser to cross-in suitable cattle with really large horns instead of selecting for several generations (and therefore decades) just to slowly increase this trait, especially because other traits would get neglected. Apart from that, there is a limit when the cumulative effects at maximum until new mutations appear.
I was unable to find any literature or remarks on the genetic background of horn curvature in cattle, but the way it is inherited in heterogeneous cross populations that show a fluent transition without following the Mendelian rules implicates to me that it is a polygenic trait (see below).

Even though quantitative traits are not comparable to qualitative traits in the way they get inherited and respond to selection, and more importantly, the Mendelian rules do not apply for the quantitative trait as a whole, the individual chromosomes with the involved loci do follow the Mendelian rules of course. This thought was the basis when I set up my “F”-based breeding plan.

The role of individual genes

The old one-gene-one-trait axiom is not used in modern genetics anymore, and it also creates a wrong impression. Genes do not work parallel and un-affected from each other, and it is not just one gene that creates one trait. Many genes actually are responsible for more than one trait, called pleiotropy. Selecting on pleiotropic genes will therefore also have an effect on other traits, what happened in domestication. Colour genes, for example, a mutation on the Agouti locus in mice (which regulates the expression of red pigment in mammals) is found to be associated with obesity3. A deviant brown colour variant found in some Lippeaue Taurus cows seems to be associated with brittle horns and fur, therefore maybe correlated with some metabolic disease as I was told (see here). For details go here: The Dedomestication Series.  
The reverse, a trait that is caused by many genes, called polygenic trait, is no rarity either. Most traits are actually more or less polygenic. Furthermore, genes also influence the expression of each other. When a gene inhibits the expression of other genes on other loci, it is called an epistatic gene.

How an organism takes shape

It is basic biological knowledge that (most) biological cells carry (almost) all of the genetic material of an organism. However, if from the start of fertilization all cells would start to express all genes, an amorphous mass of cells and not a viable organism would be the result. Therefore, the genetic plan needs a plan itself how to be expressed properly so that a living, differentiated organism takes shape. This role is taken by epigenetics on the one side and regulator genes on the other side that determine the timing and where genes are expressed. Genes regulate each other; how long, where and how much gene product is produced. I do not want to go much more into details of developmental biology here, but if you are interested in this subject I suggest you to dig into the literature on masterregulator genes, the genetic toolkit, EvoDevo etc. – it is a fascinating subject.
Transcription factors regulate how long, where and how much of genes are expressed, and this developmental regulation plays a crucial rule in how an organism will be shaped in the end. Different phenotypes can be produced with the same genetic background just by extending or shortening the expression of certain genes. A typical example is developmental delay as a result of domestication: the development stops earlier than in the wild type, and as a result, adult animals retain juvenile conditions f.e. in skull anatomy or behaviour (called paedomorphy or neoteny). For more details, see the Dedomestication Series. Many developmental genes work by the feedback loop principle: if the level of a signal molecule is high enough, the gene starts to express its product, if it is below, the gene will not be produced. Hormones, as signal molecules, play a significant rule in the developmental cascades that shape an organism. Hormonal disorders and castration are known to have dramatic physiological impact that also manifests in an altered morphology, and many of the morphological changes we see in domesticated animals are seemingly linked to developmental changes caused by hormonal pathways (again, see the Dedomestication Series).
Developmental cascades affect morphological aspects such as length of the limbs, skull shape, brain volume, muscle expression, size of internal organs et cetera.
Horn length, curvature and size are probably also influenced by developmental factors. It is logical: if the transcription factors that say it is time to grow horns are never produced, horns are not going to be produced. If the transcription factors never stop initiating expression, horns will continue to grow. If the development of the horn curve is stopped earlier than in the wild type, it will never reach its full shape. I came up with this idea when playing around with the horn sheaths of the skull of the Taurus bull Latino, and discovered that when pushing the sheaths towards almost the end of the bony core while following its curve, and therefore imitating continuing growth of the horn in length, suddenly the banana-shaped horns of Latino turn into the curled horns of an adult aurochs bull that even have the right size. I made two sketches illustrating that interesting observation.
Left: the skull with its horn sheaths in place. Right: horn sheaths moved outwards following the curvature, and an
aurochs-like horn shape and length appears
Based on that, I inferred that perhaps if the development of the horn growth was not stopped prematurely but transcription factors would have continued to induce horn length growing following the – only seemingly weak – curvature, the horns would have developed the curvature and length of that of an aurochs. If this idea is correct, it would also imply that this Sayaguesa x (Heck x Lidia) crossbreed has the right genetic make-up for horn curvature and length, and that developmental factors are the reason why the horns do not look like those of the wild type. This would then apply also to a lot of other domestic bulls and also cows.

[I could actually imagine that the genetic architecture of the aurochs horn curvature is not that complex – perhaps it is only one protein produced by only one locus that causes the horn to curve in a spiral, the so-called primigenius spiral, instead of growing straight out of the skull like a pen. There could be several mutations on this locus and also several epistatic genes to this gene that causes the variety of horn shapes we see in domestic cattle. But as long as nobody who has the possibility to get the funding is interested in studying this question, this resembles speculation.]

The most interesting note in this respect is that steers show a totally different morphology than bulls, but with the same genetic make-up. Steers tend to grow larger, have longer legs, horns and snouts and therefore, are superficially more aurochs-like in these respects. This is probably because the removal of the gonads alters the timing of the development. The gonads do not stop the production of certain transcription factors because they are absent and therefore, the development gets extended – the opposite of what has happened in domestication, and as a result, steers are superficially more aurochs-like in certain morphological respects than functional bulls of the same genotype.
Holstein steer displaying inwards-facing horns, much more so than any functional bull of this breed
Very high-legged Hungarian Grey steer; functional bulls of the same breeds are built way more longish

Sexual dimorphism

Cattle have the XX:XY gonosome system and therefore have the easy standard mammal scheme (there are also mammals that lack Y chromosoma, or have multiple X). Basically, all mammal embryos start as females and then the male-specific genes on the male-specific regions on the Y chromosome start to work that are responsible for the differences between male and female. That is not to say that male traits are found exclusively on the Y chromosome. For a lot of traits, the blueprint is on the autosomes, and the Y chromosomes just provides the switch to make them work, such as a high testosterone level. Sexual hormones, especially steroid hormones, are the most important factors in sexual differentiation and not surprisingly they are mostly produced in the gonads but also in female kidneys. For example, the amount of melanisation on E+ cattle is said to be regulated by the testosterone level, therefore most bulls of wildtype coloured cattle are darker than the cows. But cows are capable of expressing a “bull colour”, as we see in many breeds. In the aurochs, this difference in testosterone level was much more well-expressed, in colour as much as in morphology and size. Domestication reduces sexual dimorphism. This was probably not achieved by actively selecting against sexually dimorphic traits, but was the results of pleiotropic effects and developmental cascades as the result of selection on tameness alone, as the farm fox experiment suggests (again, for more on that see the Dedomestication series).
As for selecting on sexually dimorphic traits, I think the conventional method won’t work here. Always simply choosing red cows and dark bulls, or small cows and large bulls, or whatever sexually dimorphic trait, probably will work not here. You would actually have to choose individuals where the sexual dimorphism is laid down in the genome. A phenotypically red cow might either have a more or less strong sexual dichromatism (s. dimorphism in colour), or simply be red and have no sexual dichromatism at all, depending on the bulls that she would produce. The same is the case with dark bulls. How to know that? Well, at first the genetic background of the dimorphic trait has to be resolved, and then the individuals would have to be tested. But since this is rather impractical on a large scale, one could also look at the ancestors, siblings and offspring of the individual in question. But for that, you would have to have a significant number of siblings and offspring to get an idea of the genotype, and what makes it especially challenging is that crossbreeds are genetic patchworks that are rather hard to judge.
A good example why the conventional method of simply always taking red cows and dark bulls is not a proper way to achieve true sexual dimorphism are the Sayaguesa of Peter van Geneijgen. Sayaguesa usually have very reduced sexual dimorphism: bulls are black, cows either bull-coloured or very dark coloured. This particular Dutch herd is slightly influenced by Alistana-Sanabresa, a related and lighter-coloured breed with red cows and bulls with a colour saddle. Some of these Sayaguesa, as a consequence, show the reddish brown coat colour desired for aurochs-like cows. I was happy at first. However, as a consequence, some Sayaguesa bulls started to show a colour saddle. This implies that the sexual dichromatism did not truly increase, the change was merely cosmetic as there is now more variation in this respect.

Therefore, I think there are only two possibilities regarding breeding a good sexual dimorphism: either relying heavily on a breed where a well-marked sexual dimorphism is still retained (such as Maronesa, for example), or putting up with the fact that there will always be a certain number of cow-coloured bulls and bull-coloured cows (the latter are, however, historically confirmed to have existed in aurochs populations on occasion) and that the size difference between the sexes is not that big as in the aurochs.

Implications for “breeding-back”

Now I am going to give a quick summary of what is written above on which traits relevant for “breeding-back” belong to which category.

Qualitative traits:
Coat colour

Quantitative traits:
Body size
Horn size
Horn curvature (more or less?)


Body shape (including muscling, size of organs, dorsal processes et cetera)
Skull shape (also including brain size)
Proportions i.e. length of the limbs
Sexual dimorphism
Horn length and curvature

As I wrote in the beginning, breeding for qualitative traits, and therefore colour characteristics is comparably easy. However, it is no secret that recessive genes are rather tricky to breed out. Especially because it becomes more difficult the less frequent a recessive allele is, because it shows up less frequently. The most effective way to clear a recessive allele from a population is to genetically screen all the individuals for it, and take all the carriers out of the population, but that would be costly. But it is not impossible to get rid of recessive alleles on the conventional way; in Heck cattle, it seems that the grey Agouti dilutions inherited from Steppe cattle have been purged out in the Neandertal-Wörth lineage at least.
For quantitative traits, it is probably best to try to compensate the extremes. If you work with a number of breeds that have too small horns, using at least one that has very large or “too large” horns would be advantageous, as the results will have intermediate horns and you would have to purge out offspring showing the extremes in the future. Otherwise you would have to start a long phase of selecting only for the quantity of that one trait and neglect all the other traits, which probably not practical as “breeding-back” focuses on so many phenotypic traits.
(That is why I worry a bit about the average body size and horn size of the cattle used in the Tauros Project, but it is probably too early to judge the situation in this case)

The most complicated traits to breed for are those that are influenced by development in the list above. Domestication dramatically altered the developmental biology of cattle, resulting in the gross morphological differences between domestic cattle and aurochs. How can we revert this by selective breeding? Is it possible at all? One could say that it is of course possible to select on all traits that are quantifiable. However, fur breeders tried unsuccessfully to breed foxes for earlier maturity by selecting on this trait directly. In contrast, selecting on tameness resulted in foxes reaching earlier maturity as they slowly became more and more domestic4. It were pleiotropic effects that caused developmental changes to reach exactly that. And most importantly, we do not know all the involved gene loci for those developmental mechanisms, and do not know the developmental cascades in particular. In turn, I think it will not work to achieve long, aurochs-like snouts if you work with a population that exclusively has shortened snouts, even if it shows a bit of variation in snout length. As with sexual dimorphism, it would probably be necessary to get individuals into the mix that show the adult, aurochs-like skull shape that is desired because there obviously the developmental processes are still like in the wild-type in this respect. Luckily, there are a number of cattle on this world where we can find an aurochs-like skull shape, including primitive breeds and also derived ones (some derived breeds, like Holstein, even more so than most primitive breeds). The same also goes for body shape and other morphological factors. There are some breeds, luckily, that still show a rather aurochs-like body shape in having a muscular body with a slender waist resembling wild bovines. These include Lidia, Corriente and Camargue (and it is probably not a coincidence that these breeds are not bred for docility and mass, but for agility and “fighting spirit”). As with sexual dimorphism, it might be useful to rely heavily on those breeds where the developmental set is obviously still right to produce an aurochs-like body shape.

Even more efficient to get rid of typical domestic traits in the morphology of cattle might be to do the exact reverse of domestication: to breed with animals that mature later and select on a slow individual development, and to also select on “wild” behaviour (note: not necessarily aggressive). Breeding for an extended development might really work wonders for achieving a wild-type morphology. But the problem is that it will take really, really long (not to mention that the breeding itself takes longer then). Perhaps a century and more. We cannot effort endless patience since there currently is a time frame where a lot of suited areas become free for extensive grazing and the reintroduction of large herbivores, therefore we have to find a compromise between breeding the “perfect aurochs mimic” and taking the chances of filling reserves with cattle that are fit for the job. Of course selective breeding and releasing the cattle in large nature reserves do not exclude each other, but to the point where the cattle get to live more naturally and less controlled, the breeding influence of man is becoming less. The second aspect, breeding the cattle for wild behaviour and hoping that the same pleiotropic effects will reverse domestic traits that caused them, is even more impractical. Actually, extremely impractical. Let us be honest, nobody, really nobody, who has to work with the cattle wants them to behave like a wild animal. It is dangerous, costly, exhausting, and risky for the project, people and cattle. It is surely not a coincidence that most grazing projects work with cattle instead of wisent (and surely not a matter of the habitat), and if aurochs could finally be recreated genetically, they would probably not be used as often as cattle either.

What does this all mean for breeding-back? It is probably not all that problematic to breed an accurate aurochs-like colour (as some Heck herds have shown), with recessive variants showing up occasionally for quite some time. It is also well possible to breed cattle with the right body and horn size. As for the horn curvature, patience is needed, as truly aurochs-shaped horns are not that frequent among modern cattle, even in primitive breeds and breeding-back populations. For the proportions, and to a lesser extent also the body shape, there are a number of breeds that are very suitable, and some of these are currently used in breeding-back. The same goes for the skull shape. But I assume that, no matter which project or breed in particular, that horn shape, body shape and proportions will not end up perfectly aurochs-like and universally distributed among the individuals of the populations in the breeding-back results except after a very long period of selection because of the complicated genetic and developmental background of these traits and the very large number of genes involved. As for sexual dimorphism, I assume that the degree of sexual dimorphism will always remain below the extent of the aurochs for the reasons outlined above.
Selective breeding alone, especially when many of the key genes (developmental genes) are not known or visible, will probably not enable us to fully reverse domestication and turn domestic cattle into a morphological wild type. And we also have to assume that a lot of the specific gene material of the aurochs was lost during domestication. However, we are probably able to mimic it to a large degree, and then releasing the product into nature and letting natural selection work will probably retrieve or refine a lot of wild type characteristics. For more on that, see the Dedomestication Series.


1 Visscher, P.: Sizing up human height variation. Nature publishing Group. 2008.
2 Kemper, K., Visscher, P., Goddard, M.: Genetic architecture of body size in mammals. Genome Biology. 2012.
3 Corva, P., Medrano, J.: Quantitative trait loci (QTLs) mapping for growth traits in the mouse: A review. 2001.

4 Trut, L. N.: Early canid domestication: The farm-fox experiment. 1999.


  1. "As for selecting on sexually dimorphic traits, I think the conventional method won’t work here. Always simply choosing red cows and dark bulls, or small cows and large bulls, or whatever sexually dimorphic trait, probably will work not here."
    So is it that both cows and bulls can pass on the sexual dimorphism, are they weighted equally ? Or may it be mostly the bulls that carry pass these on, could it be bound to the Y-chromosomes ? In the first case you would be rigth, otherwise it could work for cows...but one should be more picky with the bulls. Then the cows should be red, and the bull-lines should be from breeds with cows that show little or no black color.

    "These include Lidia, Corriente and Camargue (and it is probably not a coincidence that these breeds are not bred for docility and mass, but for agility and “fighting spirit”). As with sexual dimorphism, it might be useful to rely heavily on those breeds where the developmental set is obviously still right to produce an aurochs-like body shape."
    Is sexual dimorphism especially high in this breeds ? Or could the cows here have a larger influence on the appereance of their male offsprings ? This breeds also have athletic cows, so maybe it's just that the bulls look auroch-like because the extra-muscles are build on the right frames ?

    1. To your first question: No, both males and females pass on dimorphic traits and also the extent of it. Of course in cattle a Y chromosome (or the necessary signal molecules) are needed to express male traits, but the traits themselves and the code for the intensity of these differences can also be found on the X chromosome or autosomes. So both cows and bulls pass on dimorphic traits.
      To your second question: If you are talking about sexual dichromatism, in Lidia the situation is often unclear because there are many lightly coloured bulls and dark cows, but there seem to be some herds where the dimorphism is ok but not perfect (so basically similar to what we see in Heck cattle). Corriente seems to be quite good regarding sexual dichromatism based on what I have seen so far. In Camargue, the sexual dichromatism cannot be judged because it is masked under the black colour allele E^d.
      The only aurochs-like breed that I am aware of that almost always has a clearly marked sexual dimorphism is Maronesa; to a varying extent.

    2. Thanks for reply.
      You mean Maronesa is the only breed with a clearly expressed sexual diChromatism, don't you ?
      From the Lidia-strains at least some individuals of the the Casta Navarra have some degree of dichromatism. This breed is on the smaller side, but maybe due to the colour it would be less tricky to cross with this than with Camargue...

      To that XY-dimorhism-question, may it be that some breeds are better in passing on the male part of it than others ? Could it make sense to focus on certain bull-lines, Maremmana and Pajuna for example ? Because there is at least one very good cross of these...

    3. The black allele E^d carried by Camargue is not problematic, as it is dominant and therefore easy to select out. I would not worry about that.

      Why should some breeds be "better at passing on the male part"? What's that supposed to mean, are you talking about the Y chromosome?
      I think the looks of Manolo Uno are ok, but the looks of an F1 are completely irrelevant as it is solely the product of heterozygosity - it does not tell you what to expect for later generations (except for dominant-recessive traits). F1 do not give an accurate idea of the potential for a breed combination. So if you cross two breeds and the result seems to be disappointing in F1, it does not mean anything for further breeding.

    4. Well, just wanted to say that in Navarra the dichromatism is already unmasked...
      Yes, i think i basically mean the Y-chromosome.
      If this is the case maybe it could make a breeding-strategy to build two pools, and finally take only the cows from the one and the bulls from the other and mix these up.

      Well, if it would be completly unpredictable it would hardly be possible to make any decisions about selection.
      I think if two breeds are combined that are much different then one could expect more surprises in future generations, and if there are certain similarities then it's likely that these could be kept up.
      So in general i would expect that the result of an Pajuna-Maremmana-cross is more predictable than the result of an Watussi-Chianina-cross, for example.
      (my comments tend to disappear, is thias due to moderation ?)