This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.
Chapter 7 - Discussion
The Cleveland Bay - A Breed with a History and a Future
The project reported in this thesis set out to examine the genetic status of the Cleveland Bay Horse through pedigree and molecular methods as well as to evaluate the effectiveness of a breed management scheme. Each of the previous four chapters has examined one aspect of this analysis. This chapter presents a summary, appraisal and evaluation of the various findings of this study and examines the implications for the breed in both the short and long term.
By its very nature this project has needed to adopt established mathematical and scientific methods to analyse the available data, to explore within and between breed diversity and to uncover the true status of the genetic health of the breed. In 2007 the second International Workshop on Population Genetics for Animal Conservation took place at the Centro di Ecologia Alpina in Italy. At that time the various guest speakers took the opportunity to evaluate recent developments in the discipline, including new statistical approaches, the growth of the use of Bayesian algorithms and the influence of molecular data such as single nucleotide polymorphisms (SNPs). Despite much progress since their first workshop, the most frustrating conclusion was that very few of the speakers used their data and quantitative methods to propose practical conservation solutions. Dr. M Bruford, of Cardiff University, also pointed out that despite the increasing number of high quality peer reviewed publications on conservation genetics, conservation biologists were failing to make full use of extensive data for actual management . Whilst those comments were addressed primarily at wildlife biologists, the same can easily be said of the field of livestock biology. It is hoped that this project bucks that trend in providing the analysis, the toolkit and a clear demonstration that simple, on the ground breed management strategies can make a real difference to endangered breeds of livestock. In doing so one cannot lose sight of the fact that many rare or endangered breeds of domestic livestock only continue through the enthusiastic support of a small number of breeders. To this end, (and not least because the author considers himself first and foremost to be a horseman and breeder with a keen interest in conservation biology and not vice versa!) this chapter is written with the breeder in mind. No doubt many will have mulled over the introduction, got lost in the mathematics of the first equations of the pedigree analysis, flicked through the subsequent chapters and found themselves looking for readily assimilated information, here in the discussion and conclusions. With the intention that this can be a practical document rather than a scientific tome, to sit gathering dust on library shelves, unashamedly this chapter is aimed at the very people that can make a real difference to the survival of the Cleveland Bay horse.
Demographics, Genetic Diversity and Breed Structure assessed by Pedigree Analysis
The Cleveland Bay is recognized as being the only true British "warmblood" breed of horse. It is amongst the few breeds that have maintained a closed studbook, with records going back nearly 300 years. As such we have a valuable resource documenting many of the founders of the breed and from which we can deduce much of the current demographic and genetic status of the breed. Many breeders will have attended the previous four Cleveland Bay breed conferences, and listened to presentation from well-respected studbook editors and geneticists (including Grant Walling and Dr. Ian Gill), In answer to the question what is different about this project and what new contribution does it make, we need to understand the nature of the work that has gone before.
Early research by the likes of Scarth Dixon and Pease relied very much on anecdotal evidence and in their analysis of the studbooks were restricted by the absence of personal computers. Indeed even the work of Emmerson and others on female ancestry lines, which appeared in the centenary studbook in 1985, was based on manual analysis of paper records. It was not until the late 1980's and early 1990's that the use of computers became more widespread and software was being developed within the academic community for the study of livestock pedigrees and wildlife conservation. Walling's undergraduate project - "An Analysis of the Breed Structure in the Cleveland Bay Horse and a plan for the maximal Maintenance of its Genome" marked a turning point for the breed in that it highlighted the issues of loss of founder representation and the accumulation of inbreeding in a restricted population. The project write up appeared in volume 33 of the studbook, making it physically accessible to breeders, even if many were lost by the mathematics. The true limitation of the study was the restricted dataset that it had access to - going back in electronic form only to the 1930's (Walling pers. Communication). This project set out to remedy that situation with the first digitisation of the complete Cleveland Bay Studbook, upon not just this study, but the current Society electronic database is now based. The process of digitising the records for in excess of six thousand animals dating back to the 18th century was bound to reveal pedigree errors. The process of integration of the studbook data into the SPARKS (single population animal record keeping system) software, originally designed for use in managing zoological collections around the world, highlighted many such errors, which needed to be amended to fit chronological and geographical evidence. At the time this was done the Cleveland Bay SPARKS database was the first livestock studbook to be managed in this way, and was the fourth largest SPARKS dataset of any animal species, being surpassed only by the likes of the world giraffe studbook!
As at February 2010 the Cleveland Bay pure studbook contained 5757 horses, of which 2763 were male and the remaining 2552 were female. In addition 230 animals were listed on the grading register and 212 held overseas registrations. For statistical purposes a reference population of 402 horses was identified, spanning one generation (10 years) from 1997 to 2006. Microsatellite parentage testing reports were available for all of these horses, which meant that the pedigree could be verified and that data was available for molecular analysis.
MacCluer was the first to recognise that the completeness of pedigree records had a significant effect on data analysis, especially when calculating the inbreeding coefficient .The pedigree completeness for the Cleveland Bay reference population was 100% at 2 generations, dropping to 92.6% at 5 generations and to 83.7% at 6 generations. The maximum number of generation traced was 39 making the Cleveland Bay dataset a considerably deeper pedigree than most other domestic equines. However, the pattern is confused by the presence of overlapping, rather than discrete generations, meaning that year on year analysis needs to be treated with some caution. When looked at by maximum trace generations completeness drops from 89% at 1 generation to under 10% at 17 generations.
Another issue that is well known for causing difficulties in restricted breeds is the unequal contributions of males and females to the population. In the Cleveland Bay reference population whilst 219 separate mares had foals registered only 83 stallions were involved in their production. Over the three breeding seasons 2005 to 2007 153 foals were registered to 120 separate mares, although only 52 different stallions were used. This pattern of unbalanced contributions of the sexes is common to almost all domestic horse breeds, with only a limited number of stallions being kept, primarily because of the management implications of keeping them entire rather than castrating them. This can have a direct effect on the effective population size and for rare breeds it is important to balance contributions as far as possible by maximising the number of stallions being kept entire. In 2008 88 male animals held society stallion licences. Of these only 40 had progeny registered in the studbook. To be of maximum genetic benefit not only must colts (young males) be kept entire, a system needs to be in place that does not encourage a biased use of a limited number of the registered stallions.
The pattern of number of foals registered per annum has fluctuated throughout the 125 years the Cleveland Bay Horse Society has been in existence. From a peak when the society was formed there was a substantial decline in the years immediately preceding the First World War. In the inter-war years the level of registrations seldom reached double figures in any one year, and this reduction continued in the immediate post-war years, such that the breed is said to have gone through a genetic bottleneck during the 1950's . Since the early 1970's the overall trend in number of foals registered has shown a significant increase, with annual registrations now about 35 animals per annum. However of importance is that only 33% of all females registered go on to produce foals themselves. Whilst there is some disagreement about the way the figure is calculated, all of the governmental and non-governmental organisations tasked with maintaining livestock biodiversity are in agreement that the number of breeding females in a breed is of great importance and that the level of "endangerment" is directly related to this. The Rare Breeds Survival Trust currently advises the number of breeding females should be maintained above 300, below which a breed is said to be "Critical". The Cleveland Bay remains one of only three endangered horse breeds on the RBST Critical list for 2010, with an estimate of 162 breeding females. With this being the case a doubling of the number of breeding females would be needed for the status of the breed to be downgraded from "Critical".
By far the majority of breeding females only have one registered pure bred foal. There are no available records to illustrate how many pure females are currently being used for part bred breeding. However it is interesting to note that well over 200 pure mares have successfully bred 5 pure foals and a significant number have produced 10 or more. One mare is recorded as having produced 15 pure foals in her lifetime. A similar analysis of male animals reveals the very unbalanced contribution of a small number of stallions, with a restricted pool having sired over 60 pure bred foals each and the most prolific in excess of 250.
The pattern of generations is concealed because they overlap considerably in the studbook. However they have been analysed by four discrete pathways - sire to son and daughter and dam to son and daughter. The average generation interval was at a minimum of 5.5 years in the period 1946 to 1950, which corresponds to the previously identified bottleneck period. The maximum was 13 years as recently as 1994. The largest generation interval is found on the sire - son pathway, with a peak of 25 years in 1904 and a figure of 19.9 years as recently as 2003. This is could be associated with the very limited number of colts being retained entire and being brought back into production, with a distinct possibility that the trend has been for the original stallion owner to do this as his horse becomes more aged, with a view to future replacement. It could equally be that breeders have a preference for older and more proven stallions. The average generation interval for both the whole studbook and for the reference population is 10 years , which is commensurate with that found in many other breeds of horse .
The accumulation of inbreeding in the Cleveland Bay horse was first highlighted by Walling in Volume 33 of the studbook. Analysis of the full studbook shows a mean inbreeding across the whole population of 7.8% with an associated Average Relatedness of 8.3%. Whilst these figures in themselves are high in relation to other breeds of horse it is now widely accepted that the rate of accumulation of inbreeding is of more importance in endangered breeds than the absolute levels .From near 0% at the time of formation of the studbook in 1885, inbreeding accumulated in a near linear trend until 1985, by which time it was approximately 20%. Over the last 20 years the rate of accumulation has slowed, such that the average inbreeding in the reference population is approximately 21%. However, when corrected for Pedigree Completeness we find an average inbreeding in the 1990s of 24%, which is approaching the equivalent of full brother - sister mating. Whilst absolute inbreeding values continue to increase beyond the year 2000, both the values corrected for pedigree completeness and absolute values after 2004 show a decline, which may be associated with the implementation of the breed management programme.
The overall inbreeding reported in other breeds of horses were 8.48% for Andalusians , 8.99 in the North American Standardbred ,7.0% in Spanish Arab horses , 12.5% in Thoroughbreds , 10.81% in Lipizzan horses . Even in more restricted populations such as Fresian horses 15.7% , and South German coldblood horses 2.28% or Black Forest horses 5.21% , which are maintained as closed populations, we do not see levels of inbreeding approaching those seen in the Cleveland Bay horse. The Portuguese Sorraia horse is recognised as critical maintained status by the FAO . It is composed of only about 150 individuals, with only about 10 founders represented in the living population. It has a low effective population size and high levels of inbreeding, estimated at 0.363 from genealogical data . This is one of the very few breeds of horse that demonstrate an even greater level of inbreeding than the Cleveland Bay. There is very limited data on inbreeding assessed by non-molecular methods in native British Horses. The Exmoor Pony was found to have an average inbreeding coefficient between 9.3% and 14.8% and the Suffolk Punch 9% in 2007 . Whilst the Suffolk shares category 1 Critical status with the Cleveland Bay horse on the Rare Breeds Survival Trust watchlist, and in real terms is a numerically more restricted population, the levels of inbreeding determined over the last 50 years are significantly less. The rate of increase in inbreeding (1.4% per generation) in the Suffolk is less than the 1.61% average for the reference population of the Cleveland bay horse. This has a direct bearing on the effective population size of both populations.
Effective Population Size
The Effective Population Size Ne was calculated using both the rate of change of inbreeding and from the number of parents methods. When calculated from the rate of inbreeding the figures vary considerably, with peaks of over 700 in the 1880s at the time the studbook was being established, and at a minimum of 7 in the early 1960s immediately after the recorded bottleneck. When Ne is calculated from the number of parents the figures are more stable, with smaller annual fluctuations. These peak at 112 in 1890 and again at 103 in 2006. However it must be noted that when there is very unbalanced genetic contributions from male and females to the population this methodology is less valid than calculation from the rate of increase in inbreeding . In the reference population the effective population size from rate of increase in inbreeding is at a minimum of 20 in 2002 and peaks at 85 in 2006, with the increase explained by pro-active breed management. When calculated from the number of parents the minimum occurs a year earlier in 2001 but the peak is in the same year at 105. Demographic stochasticity and genetic drift can negatively affect small populations, and in the worst case scenario can interact in an extinction vortex that leads to the loss of an entire population . The 50/500 rule has been proposed to identify populations at risk where the "50" numerator states that populations with an inbreeding effective population size under 50 are at immediate risk of extinction. This is because in such restricted populations the joint processes of demographic stochasticity and inbreeding can cause a very rapid extinction vortex. The 500 denominator represents the minimum effective population size from the number of parents (variance effective size) where populations <500 are at long term risk of extinction. In these populations genetic drift is said to be a strong influence, leading to eventual loss of genetic variation . Once variation is lost a population may not be able to respond to environmental changes and thus be reduced in size or become extinct. There is however not universal agreement over the 50/500 rule, with criticism over its generality and some authors claiming the figures are too small , suggesting that effective population sizes in the thousands are necessary to avoid strong genetic drift. What is clear is that the Cleveland Bay horse does not fulfil even the 50/500 rule and has not done so since the formation of the studbook. Whilst small, highly inbred populations can remain viable, as illustrated in the case of Chillingham Cattle , there is as yet no way of predicting which populations will be able to cope with such pressures , and so maximising effective population size remains a high priority in protecting endangered breeds.
Whilst there is limited published data on other native British equines, in the Suffolk Punch the effective population size ( from rate of increase in inbreeding) has been estimated to be about 40 . Over the period 1997 to 2003 the Cleveland Bay reference population had an effective population size less than this. In 2002 it was as low as 20. This is depite the fact that the physical population size is substantially greater. Since 2004 the Cleveland Bay effective population size has increased, and surpassed that of the Suffolk Punch, no doubt influenced by the implementation of proactive management of inbreeding. The Food and Agriculture Organisation of the United Nations recommends a minimum effective population size of 50 animals, which corresponds to a rate of accumulation of inbreeding of 1% per generation. We shall explore this further when we consider the effectiveness of the breed management programme that has been implemented for the Cleveland Bay horse.
Gene drop analysis identified 182 founders (f) for the Cleveland Bay studbook (both parents unknown), with a mean retention of 0.033. The number of founder genomes surviving was 6.015 and the number of founder genome equivalents was 2.219. The number of ancestors contributing to this base population was 614. The balancing of contributions of founders animals to the population is expressed by the Effective Number of Founders (fe) which was found to be 77. To account for population bottlenecks the Effective Number of Ancestors (fa) is used . This was determined to be 24 for the whole population. 8 ancestors explain 50% of the genetic variability in the whole population.
In the reference population the corresponding analysis reveals a total of only 30 ancestors, with an Effective Number of Founders (fe)= 61 and the Effective Number of Ancestors (fa) = 9. Only 3 ancestors explain 50% of the genetic variability in the whole population.
The effective number of founders in the reference population is larger than the 39.6 reported in Andalusian horses , 39.5 in Spanish Arabs , 28 in Thoroughbreds or 48.2 in Lipizzan horses . It is however much smaller than the 245 reported in Hannoverian Horses , or Anglo Arab (129) Arab (135) Thoroughbred (236) and Selle Francais (333) . The same authors did find a lower figure of 70 in the French Trotter. Unfortunately there is no published data of equivalent figures for native British horse or pony breeds.
The ratios of fe/f and fa/fe are of significance. Ignoring the effect of any bottleneck in the population, the effective number of founders will approach the actual number of founders as the population becomes balanced. In the Cleveland Bay studbook the effective number of founders/number of founders = 42.3% whilst in the reference population fe/f = 33.5% . This is indicative of the excessive and unbalanced contributions of some individuals as parent animals, with a consequent loss of genetic diversity. In comparison, for the DØle horse fe/f has been determined as 6.23% and in the Nordland/Lyngen 28.57% , 8.30% in the Swiss bred Frances-Montagnes and 5.8% in the Austrian Noriker draught horse , Anglo Arab 8.6%, Hispano Arab 10.11%, Spanish Sporthorse 10.64% , illustrating greater proportional loss of founder diversity in these breeds.
The ratio fa/fe reveals the decrease in genetic variation caused by a population passing through a genetic bottleneck. If the ratio is close to 1 then the population is stable in terms of the number of effectively contributing animals. However, of the effective number of founders is larger than the effective number of ancestors, then there are ancestors that have played a large part in population formation . For the whole population this figure was calculated at 31.2% whilst in the reference population it was only 14.8%. By way of comparison fe/f ratios in other horse breeds have been determines as Anglo Arab = 46.66%, Hispano Arab = 47.12%, Spanish Sporthorse = 72.42% , 54.35% in Lipizzans = 54.35% , Andalusians = 41.66% , Frances - Montagnes = 26.98% and Hanoverians = 31.74% . This is indicative of the Cleveland Bay having experienced more intense bottleneck events than any of the other reported breeds, with the effect being most evident in the reference population.
The effective number of founder genomes accounts for both unequal contributions of founders and random loss of alleles caused by genetic drift during bottlenecks and is closely related to the inbreeding in a population It is equivalent to the number of equally contributing founders with no loss of alleles that would be expected to produce the same amount of diversity as in the reference population . In total 6.015 effective founder genomes are maintained in the Cleveland Bay studbook. In comparison 10.63 are reported in the Noriker and 6.0 in Lipizzan horses .
Taken with the understanding of the inbreeding status of the breed, the study of gene origin highlights the appreciable loss of genetic diversity that the Cleveland Bay horse has undergone, and confirms the existence of one or more influential bottleneck events in the breed's history. Previous studies have highlighted that the breed is now restricted to one surviving male tail line and nine female lines . This study has traced the surviving male tail line back to its Cleveland bay origins in the founder Skyrocket 280 and back though the thoroughbred General Studbook to the Byerley Turk. On the female side of the breed lines 1, 5 and 6 are the most common , making up 70% of the present female population. There have been losses since the previous studies, with very few surviving members of lines 2, 4 or 9 present in the reference population. However the influence of line 2 in the reference population has been determined to be far greater than would at first appear, with a 27.44% contribution, which is greater than any other line. Later in this chapter we shall discuss the implications of mitochondrial DNA testing and revisit the relationship of Line 2 to the other more common lines.
In order to reveal within breed relationships genetic distance analysis was conducted based on Wrights F statistics. And on Nei Standard distance. The resulting matrices were used to construct phylogenetic trees, based on various methods including Neighbour Joining, Proportional Neighbour Joining and Unweighted Neighbour Joining and Weighted least Squares. Whilst there was inconsistency between the resulting trees, the pattern of the breed dividing into three main clades was revealed, with lines 1, 3 and 5 appearing in separate clusters. Previous studies have advised caution in interpreting the use of gentic distance and relationship trees constructed for these purposes advising that it is important to balance findings with a good knowledge of breed history .
In Chapter 4 we explored past breeding practices in the Cleveland Bay horse, to gain an understanding of the loss of genetic diversity that has been identified by this and other studies. We then went on to examine breed management theory and to determine the most appropriate manner in which to manage endangered horse breeds. It is recognised that the best conservation strategy based on pedigrees is to maximize genetic diversity and thus to minimize average mean kinship .
Three principal strategies exist to maximise genetic diversity - Equalising Founder Contributions , Mean Kinship and OPTIMAL Contribution Selection . Mean kinship is a widely used as a management and conservation strategy in zoos worldwide . The method has its limitations in that the system does not indicate how many offspring an individual should have, nor whether an animal should be selected for reproduction and it does not optimize genetic diversity . In contrast the system of optimum contribution selection does find the minimal average mean kinship and select s parents for their contributions to future generations . In practice, because no one authority has absolute control over all of the animals in the population and because of irregularities in pedigree data it is not the perfect system, and for these reasons a mating advisory scheme based on minimising inbreeding through management of mean kinship was considered the most appropriate method for the management of the Cleveland Bay horse.
The breed management plan for the Cleveland Bay Horse consisted of three elements:-
A breed advisory scheme applicable to all breeders, based on minimising the rate of inbreeding by minimizing coancestry and mating animals of similar mean kinship.
The prioritising and promotion of highly desirable matings through sponsorship of a selected matings scheme.
The collection and cryopreservation of semen as part of the RBST ReGENEration scheme.
Whilst absolute evaluation of the scheme is problematic, the breed advisory scheme appears to have experienced an uptake amongst breeders in excess of 40% throughout its six year existence, with a peak in 2005 of over 50% compliance. This is determined based on the fit of foals registered to the advice given and published objectives. The scheme has led to a consistent decrease in the rate of increase in inbreeding (Î”F) and a corresponding increase in the Effective Population Size (Ne)., taking it over the FAO minimum recommendation of 50 over the period 2004 to 2008. In the most recent year of the scheme there has been no proactive central support or promotion from the Cleveland bay Horse Society, and this may have played a significant part in the observed increase in rate of increase in inbreeding and diminution of the Effective Population Size.
One effect of the advisory scheme that was not anticipated was the reduction in the number of undesirable, highly inbred matings. Analysis of studbook records showed that these have occurred throughout the Society's existence, in what appears to be a random pattern. This is probably a reflection of the random pattern of mating, the lack of readily available information to breeders (in part a reflection of the recent impact of computers and their ability to analyse complex pedigrees). Since the implementation of the advisory scheme in 2004 the number of highly inbred matings has dropped dramatically, and those that have occurred have been explained by the breeders concerned. This reduction would seem to be at least in part through breeders being able to identify potentially harmful matings and to avoid them, as opposed to proactive compliance with recommendations on desirable matings for their particular animals.
Whilst complying with Society regulations regarding eligibility of progeny for registration, by not listing unlicensed colts, the datasheets produced have not identified stallions by premium status. This was a conscious decision in order not to advantage any group of males and to balance the contribution of stallions as far as possible. The pedigree analysis has already revealed the loss of diversity brought about by the unbalanced contribution of parents, and the design of the breed advisory scheme provided some, albeit limited, opportunity to address this issue by promoting stallions based on their genetic suitability for individual matings rather than on phenotypic preference. The existence of the stallion premium scheme is somewhat counter-productive in the maintenance of genetic diversity in a restricted population and will be discussed further later in this chapter.
The selected matings scheme was funded through sponsorship by the Horse Race Betting Levy Board. It provided financial incentives over and above those normally available to Cleveland Bay breeders, in order to encourage the most highly desirable matings and to ensure that animals of reduced mean kinship would be available as parents in future generations. The scheme ran for 3 breeding seasons, sponsored 26 matings and resulted in 5 live colts and 2 live fillies. The existence of the scheme raised a number of issues - some of them social and political as much as genetic.
Of particular concern were the facts that the same animals were being selected in the list for sponsorship year on year, and this was considered unfair to other breeders who owned animals of higher mean kinship. The number of owners signing up for the scheme was reducing year on year, because of the geographic difficulties imposed by restricted stallion choice. The scheme was only available to breeders in the UK, where live cover is the norm. Had the scheme been extended to North America where artificial insemination is more prevalent then ongoing uptake may have been better. He final issue of concern was the lack of ongoing support for breeders producing colts who lacked the facilities necessary to keep them entire until an age when they could be used for breeding themselves. In other breeds such as the Highland Pony, there is an arrangement for young colts to be farmed out to approved facilities that are capable of dealing with the livestock management issues involved. In the Irish Draught Horse whilst there is no central register of approved facilities, there is a financial support mechanism available.
Interestingly, an analysis of the effect of the selected matings scheme on the rate of increase in inbreeding revealed how little beneficial effect the progeny of the selected matings scheme had across the population. In terms of maintenance of effective population size the scheme had little positive benefit and thus was suspended in 2007 on a cost benefit basis. The funds have now been targeted at more general support for breeders with increased foal premiums across the breed.
Running parallel to the breed advisory and selected matings schemes was a cooperative cryopreservation scheme, financed and implemented by the Rare Breeds Survival Trust. Stallions were put forward for collection based on mean kinship analysis, and to date 11 stallions have been collected from, with arrangements in place for a further one going for collection in late 2010. A total of 481 straws of semen are now being held, with 50% (265) of these ring fenced to be maintained in perpetuity by the RBST in case of a genetic disaster, such as that closely avoided for sheep breeds during the 2001 foot and mouth disease outbreak. A futher 38% ( 145) is earmarked for conservation breeding and 12% maintained by the RBST on behalf of the stallion owner. At present there are no plans to use the portion allocated for conservation breeding, until such a time as sufficient stallions have been collected and a positive strategy has been agreed between the RBST and the breed society.
In the short term there is uncertainty over breed society support for the management scheme. What is clear from the findings of this study is that if genetic considerations are balanced with social and political ones, it is possible to manage a dispersed population of endangered horses and to bring about a sustained increase in effective population size. There is little documentary evidence of this having been carried out with such beneficial effect in any other endangered breed of horses, distributed around the world, as is the Cleveland Bay. The conservation of livestock biodiversity is very much in the hands of individual breeders rather than under the control of one studbook keeper in charge of a zoological collection, and the management of "hearts and minds" is of as great if not more importance than management of genetic diversity alone.
The use of molecular methods for parentage testing and verification in domestic horse breeds is now accepted and common practice. The FAO now recognises a panel of over 30 markers that are deemed appropriate for such work and the International Society for Animal Genetics Equine Genetics and Thoroughbred Parentage Testing workshop recently run at Edinburgh met to discuss additional markers.
The suitability of microsatellite markers for studying not just parentage, but population genetics of animals has led to an ever increasing number of studies on a wide variety of livestock including cattle, pigs, sheep, camelids and horses. In the late 1990s the Rare Breeds Survival Trust sponsored research to characterise the molecular diversity of native equine breeds . In 1997 the Cleveland Bay Horse Society implemented a system of compulsory genotyping for parentage verification of animals at time of registration. This testing has been carried out on their behalf by the Animal Health Trust at Newmarket and continues to date. Over 13 years in excess of 500 horses have now been tested, but apart from parentage verification the microsatellite data accumulated has not been thoroughly analysed. This study has been the first to do so for a sample that represents a complete generation of Cleveland Bay horse.
The 535 Cleveland Bay microsatellite reports examined all proved to be polymorphic with a mean observed heterozygosity (Ho) of .578 in the reference population of 402 samples. This is slightly less than that found in an earlier study crew , which found Ho = 0.65 in a sample of 30 individuals tested for 10 microsatellite loci using blood proteins. Very recent unpublished work by Dr. EG Cothran at Texas A&M University (pers.communication) has found an Observed heterozygosity for a sample of 47 North American Cleveland Bay horses to be 0.610. Because this work has been conducted using hair samples it probably provides a better direct comparison. However, the sample is only one tenth of the size of the records analysed by the present study, and represents only part of one geographic sub- population of the breed. Whilst observed heterozygosity is the proportion of the genome that varies in an individual, the expected herterozygosity (He) is representative of the average proportion of the genome that varies based upon population genetic theory, or the predicted number of heterozygous loci based upon gene frequencies. This study determined He = 0.5920 for the reference population, wheras Cothran has found He = 0.627. Both values are of the same order of magnitude higher than the corresponding observed heterozygosity. At sub- population level (assessed by female ancestry line) there is polymorphism at a number of loci. This will be influenced by the very small sample size tested in some of the rarer ancestry lines. The Cleveland Bay demonstrates a lower observed heterozygosity than other British breeds such as the Suffolk Punch (.683), the Fell Pony (.822), the Shire Horse ( .677),the Thoroughbred (0.734) . Of other domestic horses the Ho of the Cleveland Bay was also less than that observed in Andalusian (.722), Arabian (.660), Irish Draught (0.802), Morgan (0.715) and American Saddlebred Horses (.740) cothran . Higher levels have also been found in the Brazillian Criollo (.621), Campolina (.650), Chillean Criollo (.638), Columbian Paso-Fino (.636),Pantaneiro (.723) Garrano (.747) and Lusitano (.679) . Of the native breeds, the Exmoor was found to have an observed heterozygosity lower than that found in the Cleveland Bay ( .535), as did the Fresian (.545) and Corolla horse (0.555) and the Sorraia horse (.535). In these four breeds variation is more restricted than in the Cleveland Bay. However, it is clear than when compared to the majority of domestic horse breeds variation in the Cleveland Bay horse is considerably restriced. In addition the difference between observed and expected heteroxygosity, with the expected figure being greater, indicates that inbreeding is an important factor in the breed history and will continue to be for future management. Of the three breeds that the Rare Breeds Survival Trust presently lists as "Critical" , the Cleveland Bay appears to have the most restricted genetic diversity, based on analysis of microsatellite genotype data.
The existence of a genetic bottleneck in the breed's history has been previously reported . Various model of analysis of the molecular data have been implemented to search for evidence of this. Under the Infinite Allele Model there is broad excess of observed heterozygosity, being an positive indicator of a population bottleneck. The other two models (Stepwise Mutation and Two-phase Mutation) failed to find such evidence. There is however some deviation from the normal L-shaped distribution of allele frequencies under the mode shift indicator, suggesting the presence of a bottleneck. Whilst the microsatellite evidence for the existence of a bottleneck event is contradictory, when the theory behind the various models is examined it is evident that gene divestity excess has only been demonstrated for loci that have evolved under the IAM model. Under this model there is strong evidence to support the presence of a bottleneck. Subsequent analysis of mitochondrial sequence data also supports this.
The used of genetic distance and various tree joining methods is now common practice in the determining relationships within and between breeds, and has been successful documented in horses , donkeys , sheep and cattle . Various methods of distance matrix and tree drawing have been employed in this study, with variable success at defining sub-population structure in the Cleveland bay breed. As had been previously highlighted , the results of such analyses need to be interpreted with caution and with some background knowledge of breed history. As with trees drawn from similar distance analysis of pedigree data, a variety of patterns of relationships within the breed have been found. Whilst there is variation in the ancestry lines allocated to them there is an underlying pattern, in common with the pedigree analysis for 3 main groups or clades within the breed. However, from the pedigree and microsatellite data It is not possible to define the process or processes involved in this division. Further analysis in addition to that already conducted was clearly warranted, and sequencing and analysis of mitochondrial DNA is reported in chapter 7 . The phylogenetic trees drawn from microsatellite data need to be viewed and interpreted in conjunction with those resulting from mitochondrial DNA analysis before a true pattern of within breed relationships can be revealed.
The use of Bayesian analysis has been hailed as one of the most important advances in the field of conservation genetics the last decade . The use of Bayesian algorithms to assess the likelihood of the number of clusters, the assignment of individuals to populations, and the influence of admixture in a plethora of livestock species is now widespread . This study used two methods of Bayesian analysis - STRUCTURE and BAPS . Analysis with STRUCTURE suggest that the most likely number of clusters or subdivisions within the beed, based on plateauing of the log likelihood is at K = 11. However, visual scrutiny of the results is suggestive of K= 3 being an equally good if not better fit. The best fit of K is very subjective and needs to be carefully interpreted corrander . BAPS analysis requires less subjective interpretation of the results, and gives K=17. The allocation of individuals to clusters in this model produced a particularly uneven distribution, with 2 clusters each containing >11% of the population whilst 3 others each contained only 2%. Distance matrix analysis or the results of the BAPS analysis once again produced trees showing three main clades within the population.
Molecular Estimates of Kinship
The use of marker estimated kinships is of growing interest in the fields of livestock conservation and in the management of wildlife populations. It is particularly relevant in the absence of accurate studbook and pedigree records . We examined the relationship between the kinship data derived from very extensive pedigree records and that derived from microsatellite data using MOLKIN . We found a strong linear relationship between the two dataset, demonstrating that there is consistency between the two methods. However, it has been shown that where there is extensive and accurate pedigree information this remains the preferred method for assessing relationships and kinship between individuals in a population . In the absence of such detailed records it would be possible to use molecular markers as a method of determining kinships, with potential for advising breed management by molecular methods alone. This would certainly be the case in wild or feral horse populations such as those in Mongolia and in the Americas . In the case of the Cleveland Bay there would currently seem to be no advantage in incorporating molecular mean kinships into the breed management scheme.