Friday 30 October 2015

The In(breeding)s and Out(breeding)s of Managing Gene Pools Part 2

Read Part 1

Managing a Gene Pool

This blog post is going to look at some different methods of managing gene pools that have been used historically and in the present, and how effective they are at maintaining gene pools. All gene pools are closed, usually at the species level. Therefore, the diversity within the gene pool over the whole of the species (the 'megagenome' perhaps) is controlled by three different factors:

1. The amount of genetic diversity within the gene pool when it was formed (species evolve from other species when its gene pool is cut off, or closed, from the parent species).
2. Random mutations increase genetic diversity over time. It's estimated that each individual has a few completely new mutations never seen before in its genome. Therefore, the rate of increase of genetic diversity from mutation is dependent on the breeding population size: how many individuals there are in the species that pass on their genetics.
3. Selection and genetic drift (random changes in allele frequencies) cause loss of genetic diversity.

The gene pool is therefore stable, and the population has long-term viability, when the gene pool is of sufficient size and the loss of genetics is equal to or less than the rate of appearance of random mutations in the population.

I made a comparison table for this but it is too big to fit on a normal webpage. Click below to view:

Methods of managing gene pools -- table

Systematic inbreeding
The goal of this breeding model is to create a strain of animals that are both as genetically similar to each other as possible, and as homozygous as possible. Creation of such strains begins with mating together, for example, two mice, and taking one female and one male offspring and breeding these siblings together, and from this mating take one female and one male, and breed these siblings, and keep repeating this. This is a targeted inbreeding method to over several generations eliminate heterosis from the strain, which overcomes all natural defences against loss of heterosis, since once genes are lost there is no way of recovering them and the result can only go one way.

It is important to understand that many attempts to create these strains fail, because the strain self-destructs and becomes unbreedable or unviable through accumulations or combinations of harmful mutations that would not have been a problem in more heterozygous populations. The strains that do survive sufficient generations become 99.9% genetically identical. After this point they can be maintained in this near-identical state for many generations. The entire population and every animal in it contains only half the genetic diversity that would be expected to be seen in a single individual from unrelated parents, and the animals are effectively clones.

Graph on left showing loss of genetic diversity over several generations due to systematic inbreeding to create an inbred strain -- note the line never touches the x-axis; it's what's mathematically termed an asymptote, in that the strain never becomes 100% fully homozygous, but will be greater than 99% after a certain number of generations. On the right, how the population of the finished inbred strain looks on a graph rather like the ones used on the Genoscoper 'mydogdna' test. No matter how many individuals are in the population, they are genetically identical and so there appears to only ever be one data point since every individual is nearly identical.

Systematic inbreeding does not occur anywhere in nature and it is a manmade system specifically developed for a purpose. However, just because it is unnatural does not mean it doesn't serve a purpose, as inbred strains are very important for scientific research where it's necessary to test something on animals. The genetic similarity of the animals takes the unpredictability out and means small populations and control groups can produce reliable results, and fewer animals have to be used. The development of these strains can sometimes throw up interesting results, such as mice that are susceptible to certain kinds of cancer, or are resistant to particular diseases or medicines, which makes them useful as analogues for studying diseases.

Pen breeding
Here is another anecdote, and this time it is a true one. I wanted to breed French Marans, so I acquired eggs from two different sources. I took the best cockerel that resulted from one clutch and all the pullets from the other clutch and put them together. I also kept two pullets from the same clutch as the cockerel in a separate pen with some egg-laying birds. I sold or ate the rest of the birds. The next year I hatched eggs from the birds and sold some of the offspring. Then a fox got into the paddock and killed all the adult Marans. I put all the eggs that were left in the incubator and while the pullets were growing up, I took the best cockerel I had from them and the two hens who were related to the first cock (who were likely aunts or half-aunts of the cockerel) and bred them. This cock was then also killed by a fox. I took all the pullets and a cockerel from the earlier breeding and put them in together, and kept one cockerel of good type from the aunt chickens for use later.

Pen breeding tends to involve a lot of chaotic linebreeding of animals that aren't valuable enough to keep exchanging with other breeders, and environmental circumstances outside the breeders' control. Unsurprisingly, this isn't really the best method of breeding for keeping individuals as heterozygous as possible over many generations. But, perhaps surprisingly, it does work well at preserving genetic diversity across the entire breed, as different breeders will have different pens of genetically different birds, and there are ways of mitigating the loss of heterosis. Many breeds of chicken are hundreds of years old and some are even over a thousand years old. The Dorking is a breed that has existed in Britain since Roman times and is still in existence today. Considering chickens generally reproduce at 1 year of age, this is a lot of generations that these gene pools have remained viable despite being largely closed.

Illustration of pen breeding. On the right, appearance of pens as visualised on a graph showing their genetic similarity. On the left, diagram tracking genetic contributions of four founders in a pen system over six generations.

The illustration on the above left shows a pen system that starts with three (assumed) unrelated individuals from different bloodlines, male A and females B and C. The colours on the symbols show the genetic contribution from each founder, but don't take into account factors like heterosis. First of all, two daughters are kept from B and C, labelled D and E. On the next generation, B has died and another individual has been kept to replace her, F, who has in fact resulted from a mating of A to his daughter D. We still have founder C.

Sire A is then removed to prevent further inbreeding, and replaced with a new unrelated male from a different bloodline, H. In the 5th generation, we can see that A has mated with all the remaining females and a new pen has been restarted with the female offspring (possibly the parent stock is split into new pens or sold to a different breeder). H's offspring are then bred to a brother of an earlier ancestor, D2. D2 is the uncle of I and both the great uncle and half-brother of G, and the cousin of J, but is unrelated to H. This style of outcrossing to an unrelated individual followed by linebreeding and occasional closer inbreeding is what keeps the original foundation stock of the pen influential -- see below for how this compares to the stud system, where the starting material genetics tend to be swamped by outside sires. Because outcrosses are used every few generations and because the inbreeding is mild and involves haphazard combinations of second and third degree relatives rather than being targeted and systematic as in the first model, heterosis is not lost to the same extent.

There are various possibilities of where to go after the 6th generation. All the females are now related, but C's genetics are starting to look a bit underrepresented. A good idea might be to mate C to a new unrelated male and select a male offspring from this to be the sire for the 7th generation, who would be no closer than a first cousin to the existing females, and after this use a male who has even contributions of all four of A, B, C, and H.

On the right of the illustration are shown clusters. Each coloured cluster is a pen where the animals are related to each other and being bred. When an outcross is made, the outcrossed individuals will appear equidistant between the two clusters its parents come from, but when it is bred back into one of the parent clusters, the data points of the offspring will move more towards that cluster. When clusters are linebred for a long time, they will tend to shrink towards the centre of that cluster. If all the clusters were bred together, this might provide a short-term benefit for heterosis of individuals, but over time they would start to coalesce towards the centre of the clusters taken together as a whole. This would mean that eventually there would come a point where there were no outcrosses any more as the population on the global level would all be related. As pens exist, there is transfer of genetics between pens, but this transfer is not large and the pens (clusters) tend to develop mostly independently from each other. Breeders may trade animals with other breeders, and breeders might also keep several pens to transfer animals between. Importantly, they undergo loss of genetics independently of each other due to selection and genetic drift, whereas if they weren't separated to this degree, the selection and genetic drift would be much the same across the entire breed. It's actually possible for two different breeders to start with similar genetic foundations and end up with genetic clusters that are distinct from each other due to differences in what they select for and random genetic drift.

Sacrificing a degree of heterosis for the sake of maintaining genetic differences on a breed-wide level is why pen breeding has kept so many breeds of birds and other farm animals viable. Each pen has a lower effective population than the number of animals actually in it, but this duplication of genetics in several individuals is actually protective against those genetics being lost from generation to generation, particularly where it is not practical for numbers to keep expanding. If only one offspring is kept from each parent, only half the genetic potential of that parent will be passed on. Having related breeding stock means less genetic diversity is lost than if unrelated stock were constantly being brought in. Biological mechanisms that protect against excessive loss of heterosis, for example poor hatchability in very homozygous eggs, also comes into its own as a means to holding on to genetic diversity and indicating to the breeder when inbreeding depression might be developing and an outcross to an individual from another cluster is needed.

Mating systems similar to the pen model occur widely in nature with varying levels of genetic interchange between clusters. Animals that live in groups of females where one attendant male mates with the females and fights any other males who come and try to take this right are good examples. Fairly isolated gene pools exist in pools in caves and sometimes in trees, where transfer of individuals and genetics between the different environments is difficult but occasionally does happen

Kennel model
The kennel model is a variation on the pen model, but with more control and usually a larger foundation base. This was the traditional model used by dog breeders before it was possible to transport dogs easily to other breeders. Unlike pen systems, animals bred in kennel systems tend to have higher value and be more expensive to maintain. Whereas a breeder may keep several pens at once, breeders don't generally have several kennels, and in order to prevent either excessive inbreeding on one extreme or loss of genetic identity on the other, matings are carefully managed over several generations between acquiring breeding stock from outside.

The kennel would probably be founded on a number of males and females who were largely unrelated. Unlike in a pen where one male and several females are usually put together, and it's often difficult to tell which mother an offspring came from, the sexes are kept apart and only mated together when the breeder decides who should breed with whom. Accurate records can thus be kept and matings planned into the future using whatever level of linebreeding or outcrossing the breeder prefers, and more recently, genetic tests can be used.

Traditional kennel environments (at least in the case of dogs) have declined in current years mainly due to concerns that dogs are adapted to be companions to people and their welfare needs are better met when they live in a house with people rather than in outbuildings. Under these systems, however, many breeds of dogs were able to survive the World Wars with at least some diversity remaining. Kennel breeding systems can still be replicated by consortiums of breeders working together, or by breeders using co-own arrangements or retiring dogs as pets after breeding them.

Advantages, principles, and natural instances are similar to the ones for the pen model, although organised breeding strategies based on pedigrees or genetic analysis obviously do not occur in the wild. Because kennel systems are more expensive to manage than pen systems, breeds managed in this way usually have fewer subpopulations (kennels) globally across the whole breed than do pen systems as fewer people have the resources to maintain them.

Stud breeding model
High-value animals such as alpacas are mainly bred by the stud system. It is a modern model by which studs or females are able to travel extensively to mate, which is extended even further in some animals by the transport and preservation of semen for future use.

In the stud system, the majority of females are bred, but few males are. The males used for breeding are selected heavily. This can mean there is potential to greatly improve the desired characteristics in females quickly, but the genetic contribution of the females tends to be drowned out over time. Because offering a good stud service is a time-consuming responsibility and because of the competition and pressure on the owners of female animals to choose the best mates possible to improve their stock, there may be few studs to choose from, and popular sires may become rampant. The stud system does not necessarily result in inbreeding, but its results of the majority of offspring being produced from a limited number of sires can make it difficult to avoid it in subsequent generations. Ironically the sires that are most popular are often the ones that are seen to be most unrelated and different to the existing gene pool, and are used widely because this reason, only to cause a bottleneck in subsequent generations because of their popularity.
Illustration of alpacas bred using a stud system

The illustration above shows what happens when popular sires are used on a herd of alpaca hembras (females) owned by a breeder whose priority is improving the fibre quality of the animals. The breeder has started with four unrelated females and over four generations has used four different stud males (machos). Each of the males has been selected to have the best fibre quality by laboratory analysis as was available that year, and each year each hembra has produced a female cria.

There are 8 assumed unrelated animals imported from the wilds of Peru, so that is an effective population size of 8 to start with. By using effective conservation breeding methods, one would hope to hold on to most of this effective population size, but as can be seen from the diagram, this has not happened. The four offspring in the 5th generation don't even have an effective population size taken together as the four females the programme started with. All of them have 50% of their genetic material from the black macho who was the most recent popular sire. The lavender macho is the second most influential, contributing 1/4 of his DNA to this generation, and the original hembras, all of whom were unrelated and genetically different, have contributed only 1/16th to each of the females in the 5th generation. The genetic contribution of the hembras has been completely swamped by the popular sires. The only significant contribution these females can make is if they have a brother or a son who becomes a popular stud.

Note that in this illustration, no inbreeding has actually occurred. The alpacas in all generations are assumed to have high levels of heterosis and unrelated parents. But the gene pool is nevertheless reduced by excessive contributions of some animals and low contributions from others. If everyone is using the same few sires, it will become harder and harder in future to find unrelated studs to use. Another risk is if one of these studs carries a defective gene which he doesn't express, either because it is recessive, or it is dominant but only has an effect in particular situations. This can cause a gene to suddenly become widespread in a population, and near impossible to eradicate.
Illustration of effect of three popular sires (shown in red) on a population of largely unrelated females in a single generation. The genetic distribution collapses towards the males.

Mating systems similar to the stud model do occur in the wild. Most examples are when males fight amongst each other to defend territory in which only the successful get to mate with females, such as rutting deer and sea lion beachmasters. This is an evolutionary strategy in mammals, and to a lesser extent in birds, whereby one sex can be exposed to intense selection while the other is able to reproduce largely unimpeded. However, in the wild, such populations tend to be large and the effect of popular sires is restricted by geography and by the tremendous physiological stress on the male animals who are only able to maintain this position in the prime of life. Males who are successful in these wild systems tend to be mature and experienced, and have gone through a test of survival to be able to get to that position. Compared to this, stud males of domestic animals are often used young and can have a far-reaching impact on what is by comparison a small and at-risk population. While it is possible to manage stud systems effectively, in practice this rarely happens.

Outcross model
The outcross model is not a sustainable breeding system as such, but a strategy used to rescue a breed, subspecies, or other genetic resource that has reached a critical point already in terms of diversity loss and is no longer viable.

Illustration of 3+ generations of an outcross intended to re-establish new genetic diversity in a critical population (shown in black). Outcross animals are chosen from different breeds with as similar a phenotype as possible and mated to individuals from the endangered population. Their offspring are mated to different individuals from the same endangered population. Offspring of this mating are then mated to the original population and to each other widely to establish a broad genetic base that is 3/4 or more genetically derived from the original population.

Outcrossing in this system is not the same as the sort of outcrossing that occurs between pen breeding systems, where the animals are assumed to be largely unrelated, but are members of the same breed. Outcrossing animals from different breeds or subspecies is less predictable and can result in outbreeding depression if care is not taken in selecting outcrosses.

To select outcrosses, modern genetic tests may be beneficial to help establish which breeds that are phenotypically similar look most promising as outcross material. It is best to choose more than one similar breed to outcross to, and the outcrosses chosen should be genetically disparate both from each other and the endangered population. Using only one outcross source risks linebreeding on that outcross in the future, and makes it difficult to establish a broad genetic base. The outcrosses are mated to individuals from the population which are as different as possible. Usually, unless the population is extremely small, a second mating of the offspring of the first outcross to other members of the endangered population is done. This is because the genetic integrity of the breed is better preserved if the new individuals to be absorbed into the population are 3/4 rather than 1/2, and also because usually any serious problems due to outbreeding depression should become apparent by this second generation. If all is well, the second-generation outcrosses can then be bred to each other and to the main population. They have to be used widely to have an impact, and because of this outcrossing endeavours involving breeds usually require the cooperation and organisation of many stakeholders in the breed to be effective.

In nature, outcrosses from subspecies and distantly related populations sometimes occur. Usually the effects are not strong because of the geographic restrictions that caused the development of separate populations in the first place. Particularly when habitats are disrupted due to human or natural effects on the environment, subspecies or species genetically close enough to be able to breed and produce viable offspring, can be brought into contact and interbreed. These are often bottleneck situations in which the new environment subjects the combined populations to brutal natural selection. As a result of this, any unfortunate combinations of genetics and outbreeding depression that may occur is removed, and genes that don't combine well or work in the new environment are purged from the combined gene pool until a stable admixture is attained, which stabilises into a new species or subspecies

There is evidence that modern humans have low levels of DNA derived from genetic interchange with Neanderthals (a subspecies of Homo sapiens) that occurred over a long period of time.

Random model
The random model is an artificial system mainly used for computer modelling of populations to compare to other models. For every individual in the population, a mate is selected entirely at random. The mate could be very related or very unrelated, or somewhere in between.

Generally random models are not subject to selection (neither artificial nor natural). They tend to be relatively stable when compared to other models, as the only changes occur through genetic drift and random mutation.

Although they are used as a 'gold standard' to compare other models to, random models are actually completely unnatural. Wild populations are constrained by natural selection, geography, and instincts that discourage them from mating with closely related individuals. Some breeders of domestic animals use genetic tests or pedigree estimates such as COI to try to choose mates as genetically different as possible; it should be stressed that these are not random methods, as random means the matings are just that, and not chosen for the potential of heterosis!

Random systems and real-life systems that try to emulate them tend to treat the whole population as a group together in which there is unlimited genetic exchange. Because of this, genetic drift affects the whole population in the same way. This is not an issue when populations are numerically and effectively large, but with small populations that are at risk, more controlled breeding strategies where the population is split into groups with limited exchange may actually be more effective. In a small population, total, equal exchange of genetics throughout the entire population will ultimately result at some point in all animals being distantly related and sharing what's essentially the same pedigree arranged in a different order, and the point that this is reached is, unlike a much larger population, unlikely to be compensated for by random mutation.

Feral model
The feral model is the closest real-life equivalent to the random model. Large numbers of animals live in colonies or habitats with no human supervision. The individuals have to survive natural selection, and possibly also artificial selection by culls, etc, and choose who they mate with according to their own criteria and geographical limitations.

Terminal crosses
Terminal crosses are usually the province of farming industries. They are developed from two inbred strains that have been heavily selected for a particular characteristic, such as egg production. This selection has the function of improving this trait at the expense of a loss of fitness due to inbreeding depression. Crossing together two strains retains the genetics for the selected feature, as the inbred animals tend to be homozygous for it, but the heterosis of the outcross improves the performance of the strain.

Terminal crossing is not the same as an outcross chosen to bring diversity into an existing breed as described in the outcross model. Terminal crosses are 'terminal' because they are of no use for subsequent breeding and exist just to provide food. They are not chosen based on compatibility. The cross is predictable because of the level of inbreeding in the parent strain, but this predictability disintegrates if they are bred beyond this point to each other or to the parent strains, and both outbreeding and inbreeding depression can develop.

Illustration of terminal crosses derived from two different inbred strains.

Some terminal crosses are necessary because the traits needed for modern factory farming are mutually exclusive to functional breeding. Some cross-bred broiler chickens grow at such a monstrous rate they would never survive to adulthood, let alone breed naturally. There is a market demand for lean pork, whereas lean pigs do not make good mothers, so lean pigs are bred at a specialised level and sows of a non-lean breed are inseminated with semen from the males to produce a compromise. Animals that live in factory farm environments benefit from having traits that most livestock breeders in non-industrial conditions would view as unfavourable, for example, a stupid, listless animal that is content in overcrowded conditions and does not forage or squabble with its fellows.

Often terminal crosses are lucrative to those who develop them, as if the specific strains used to create them are kept secret, farmers cannot keep on offspring of animals from year to year, and have to return to the supplier in order to restock.

Terminal crossing does not occur in nature. It exists to make a product for human consumption. Ironically in its quest for predictability, it most resembles the systematic inbreeding of laboratory strains, although in this case the end result is a heterozygous near-clone rather than a homozygous one.

Hybrid breeding
Hybrids are another level up from outcrosses between different subspecies or breeds. Hybrids are crosses between different species. They usually suffer from outbreeding depression as a result of this, and generally are infertile and incapable of producing viable progeny. Despite normally suffering ill-effects of outbreeding, they can also benefit from heterosis and be hardier than either parent strain, particularly if the parent strains are adapted for complementary environments.

The area of human activity most known for producing hybrids is the breeding of ornamental (and occasionally food) plants. Like terminal cross-breeding in the farming industry, hybrids can prove lucrative if the key to a specific hybrid's creation is known only by a select few.

Epiphyllum hybrid (results from crosses between different species of cacti)

Domestic hybrids exist mainly as novelties. Because of their sterility, they are useless for breeding.

The hated 'Leylandii' a hybrid between two species of cypress tree adapted to different environments.

True hybrids (as opposed to subspecies outcrosses and interbreeding between species genetically similar enough to produce viable, fertile offspring) can and do occur naturally. Generally they are harmful when they do, as they are infertile and contribute nothing to the future of any species, and in the same habitat they often compete for resources.

To be continued.

Saturday 3 October 2015

Curse of the Fruit

It's that time of year when windfall fruit is going mouldy, and unfortunately the younger two poodles just will not leave the fruit alone. Not only does eating rotting fruit off the floor put dogs at risk of being stung in the mouth by wasps and potentially eating slugs that could carry lungworm, but either the fruit or the mould on it disagrees with them, and recently despite the best attempts to stop them from eating it, Hobsey consumed a large quantity of damsons and pebbledashed the hall with the stones (Hobsey lived outside for several months before she came here, and although she has learnt to be clean in the house, she has never really got the idea of asking to go out if she is caught out and has to rely on routine instead). Indi was later whining and indicating that she needed to go out, and when let out she had diarrhoea, but she trod in it and so it ended up back in the house.

So Hobsey and Indi are both having to wear muzzles while outside for the moment. This means they can still run about rather than having to be on a lead. Indi is a bit sensitive and doesn't like hers much, so I hope she will get used to it soon, but Hobsey doesn't seem to care and carries on trying to cram stuff in her mouth despite the muzzle.