This study examined genetic diversity among contemporary English bulldogs used for breeding purposes from the USA (n = 87) and several other countries (n = 15). Thirty-seven pet English bulldogs seen for health problems at UC Davis Veterinary Clinical Services were also included in the study. The populations that were tested evolved from one major and three minor paternal haplotypes and three major and two minor maternal haplotypes. One paternal haplotype, which was found only in a one dog, appeared to be distinct. The two remaining minor paternal haplotypes, also in single dogs, differed from the dominant paternal haplotype by a simple mutation. This pattern of a single dominant paternal haplotype and a small number of maternal haplotypes has been observed in most pure breeds and is therefore not unique to English bulldogs [21, 25].
The paternal and maternal haplotypes identified in these English bulldogs provides a window into the breed’s ancestry. The dominant paternal haplotype occurs as expected in several bracycephalic breeds that include “bull” in their breed names, as well as breeds such as the Beagle, Coton de Tulear, and Wire-haired fox terrier. The minor unrelated haplotype has been found in a related genetic form in Akita and Beagles. The two minor mutant haplotypes have not been seen in any other breed suggesting that these have arisen by mutations within the breed. The major maternal haplotype EBU-J occurs in several of the brachycephalic breeds (Boxer, Pug, Boston Terrier) as well as small Mastiff-type dogs (Bull Mastiff, English Bull Terrier, American Pit Bull Terrier), while EBU-K has been found in Asian Mastiffs such as the Shiba Inu and Shikoku. Therefore, the maternal haplotypes seen in English bulldogs support the general belief that English bulldogs evolved from Mastiff-type dogs crossed with brachycephalic breeds such as the Pug . It is tempting to associate paternal rather than maternal introgressions from smaller and much less aggressive non-brachycephalic breeds were used to make English bulldogs more acceptable as household pets.
Evidence obtained from the 33 genomic STR loci provides additional evidence for the small founder population and artificial genetic bottlenecks that led to the modern English bulldog. The average number of alleles per locus was similar to other pure breeds that have been diversity tested.Footnote 15 However, an examination of the average effective alleles per locus indicates that only one or two alleles are actually contributing to the overall genotype and phenotype diversity of the breed. Six of 33 loci had an allele with frequency >70 %, and allele 202 at locus REN162C04 was homozygous in 101/102 dogs studied. Loss of genetic diversity as a result of pure-breeding, especially when those breeds undergo selection for conformation, is well documented . The various conformational changes used in creating breed-specific phenotypes often results in large regions of extended homozygosity across the genome [31–35]. The larger and more numerous these regions, the more often they will be associated with an STR and the more likely that certain STR alleles will be at higher frequency and also exist in a homozygous state.
Measurements of observed and expected heterozygosity (Ho and He) and the inbreeding coefficient F are also useful in looking at the genetic makeup of a population. Observed and expected heterozygosity were similar for the English bulldogs tested, yielding an inbreeding coefficient F that was close to zero (0.001). This indicated that the population as a whole was in HWE despite a limited gene pool and that English bulldog breeders were doing a reasonable job of identifying more distantly related dogs for mating. However, IR values indicated that He, Ho and F values were misleading, as many individual dogs in the study were actually products of parents that were much more related to each other than assumed from the population-wide fixation indices.
Internal relatedness has been widely used as an indicator of population fitness [36–42], implying that closely related parents reflect a loss of genetic diversity in the total population under study. The average IR value is 0.25 for a litter of puppies born to full sibling parents from a genetically diverse and randomly bred population. The average IR value for English bulldogs was around 0.007, however there were a number of individuals with IR values around 0.20. However, IR does not take into account the degree of genetic diversity that has been lost as a result of breed development. Virtually all of modern pure-breeds can trace their origins to village-type dogs that proliferated during the Neolithic era in the Middle East and SE Asia and populated other regions of the world [22, 26]. Allele and allele frequencies found in pure breeds can be adjusted to the frequencies of those same alleles found in village dog populations and used to re-calculate or adjust IR values to approximate the expected diversity if no founder effects or artificial bottlenecks occurred during a breed’s evolution. When IR values were adjusted using allele frequencies in village dogs, the mean IRVD value for English bulldogs rose to 0.34, with 50 % of dogs having even greater values. A mean value of 0.34 indicates that the average English bulldog is genetically equivalent to offspring of full sibling parents that came from a highly inbred subpopulation of village-type progenitors. This shift to the right of the IRVD compared to IR curve was also seen in Standard Poodles, but to a much less degree, reflecting the greater amount of initial or retained genetic diversity in Standard Poodles.
The low number of different DLA class I and II haplotypes in English bulldogs was associated with an imbalance in the relative frequency of each haplotype. Four of 11 class I and 3/9 class II haplotypes were found in over 90 % of the individual English bulldogs tested. Moreover, there was a high level of homozygosity among the dominant DLA class I (19 %) and II (42 %) haplotypes. Although not balanced in frequency, genetic assessment of the seven STR alleles associated with the DLA class I and II regions showed them to be randomly segregating at this time. This indicated that the over-representation of certain haplotypes occurred at the onset of breed creation as a result of small founder numbers, and/or that it was associated with artificial genetic bottlenecks that were subsequently masked by a return to random selection. Although the DLA region is only a small part of the genome, the importance of these haplotype imbalances, small haplotype numbers, and increased homozygosity should not be underestimated [23, 42]. The breed suffers greatly from allergies, immunodeficiency, and a number of autoimmune disorders, which may be a reflection of loss of balanced selection and heterozygote advantage in the DLA region.
In order to gauge the extent to which humans shaped the phenotype of English bulldogs from typical dogs, we decided to compare the English bulldog with a breed that was similarly inbred based on the number of homozygous SNPs identified by Illumina 170 K canine SNP arrays and IR scores, but outwardly similar in appearance to the ancestral dog. The Standard Poodle, which has been similarly studied , met the desired criteria. One hundred nineteen ROH with consensus sizes ranging from 1.31 to 3.68 Mb and shared by at least 6/10 dogs tested were identified on 30 of the 38 autosomes. This was compared to 31 ROH with consensus sizes ranging from 0.11 to 1.96 MB on 19 chromosomes for the Standard Poodle. Although a proportion of the ROH observed in both English bulldog and Standard Poodle can be attributed to natural selection pressures occurring over thousands of years [33, 34], the differences in ROH size and number are better explained by the comparative changes in outward appearance. The physical traits of English bulldogs, such as extreme brachycephaly, chondrodysplasia, skin furrowing, differences in tail structure, size and behavior, are extreme compared to the physical changes seen in Standard Poodles. These specific and extreme phenotypic traits required strong positive selection (strong sweeps) in specific regions of the genome , while ROH required for the Standard Poodle phenotype were not nearly as strong or widespread. The variation in ROH observed in the genome of English bulldogs and Standard Poodles were in line with the findings of others. Vaysse and colleagues  identified 44 genomic regions among 49 pure breeds that had undergone intense selection and 22 blocks of SNPs in certain breeds that extended over one million bases. Lindblad-Toh and a large group of investigators  compared the haplotype structure of the genome of the Boxer and 10 other breeds and found regions of linkage disequilibrium extending over several megabases within a breed and tens of kilobases between breeds.
We did not associate runs of homozygosity in English bulldogs with characteristic phenotypic traits but are confident based on previous studies that the greater genomic variation in English bulldogs compared to Standard Poodles reflected stronger human-directed selection in the former breed than in the latter. Pollinger and colleagues  also concluded that strong artificial selection for breed-defining traits have reduced variation within many regions of the genome. Associations between ROH and species/breed traits such as those found in English bulldogs have been demonstrated for human-directed selection in many pure breeds [29, 32, 33, 45, 46]. Brachycephaly is a prominent phenotypic trait in the English bulldog and not a naturally selected phenotype of ancestral village- or landrace-type dogs. A ROH around 500 kb on CFA1 was associated with brachycephaly in a study that compared a number of brachycephalic breeds . A more recent study confirmed the existence of the brachycephaly-associated region on CFA1 in boxers, but also identified a >8 Mb ROH on CFA26 . Variation in CFA10 has been linked to ear morphology and body mass in a number of breeds . Genomic regions of positive selection in dog breeds have also been associated with adaption to a diet richer in starch . Vaysse and colleagues  also identified runs of homozygosity associated with breed-defining characteristics such as chondrodysplasia in Dachshund [590 Kb] and wrinkled skin in Sharpei [1.4 Mb], characteristics of English bulldogs but not Standard Poodles. They also concluded that artificial selection in domestic animals targeted different functional categories than natural selection. Pollinger et al.,  identified a 40 Mb selective sweep on CFA11 associated with black coat color in Large Munsterlander and a 10 MB region on CFA3 in Dachshund containing FGFR3, which is responsible for achondroplasia in humans and presumably linked with related genes responsible for canine chondrodysplasia.
It can be assumed from this and other studies that the small founder population of the English bulldog, estimated at 68 individuals,Footnote 16 coupled with human created artificial genetic bottlenecks have greatly diminished genetic diversity and fostered a wide range of health problems. Small founder numbers and artificial bottlenecks are a much more powerful cause of lost genetic diversity than inbreeding . Some bulldog breeders from the UK have already realized that the artificial selection process had gone too far, either on their own or bowed by public pressure, and have revised breed standards that discourage physical features “that might prevent a dog breathing, walking and seeing freely” . However, the brachycephalic syndrome in English bulldogs is much more complex than a “shortened head.” Breeds such as the Boxer are similarly brachycephalic but do not suffer to the same degree. In the case of the English bulldog, the nostrils are narrow, the base of the tongue is large and broad, the palate elongated and thickened, thus allowing for blockage of the pharynx especially during sleep [8–11]. The tracheas are usually hypoplastic (narrow), further compromising normal ventilation . The constant pressures on the upper airways created by these obstructions may also damage and weaken the laryngeal muscles, cause eversion of the lateral ventricles, and further decrease pharyngeal patency . The laryngeal collapse is in turn associated with bronchial collapse . Therefore, the breathing problems in the English bulldogs go beyond mere shortening of the face and require specific changes in the nares, rostral skull, tongue, oropharynx and trachea. Concentrating on the brachycephalic syndrome also ignores other serious problems in the breed, such as inability to breed and deliver normally, poor mothering, high puppy mortality, the accumulation of simple recessive deleterious traits, a number of orthopedic problems, certain cancers, allergies, immunodeficiency, and autoimmune disorders. The authors would agree with O’Neill and colleagues  that breeding reforms should target commonly-diagnosed complex disorders that are amenable to genetic improvement and should place special focus on at-risk breeds. Unfortunately, in the case of English bulldogs, this list is very long, but it may still be possible to target the most serious of these disorders for genetic correction.
Assuming that there is a will to improve the overall health of English bulldogs, the question raised by this study is whether or not there is sufficient genotypic diversity remaining in the breed to allow “reverse genetics” to correct phenotypic abnormalities that have major impacts on health. There are certainly phenotypic differences that still exist between various lines of English bulldogs and among individuals, and many English bulldogs enjoy much better health than others. The regions flanking consensus ROH vary greatly in size in individual English bulldogs, suggesting that some of these regions may contain “hidden” genetic diversity that may prove critical for reversing the degree of brachycephaly or chondrodysplasia. The existence of phenotypic variation within the breed is evident from photographs of modern English bulldogs.7,8 The one region of the genome of English bulldogs that is least amenable to reverse genetics may be the DLA, which has very low diversity based on DLA class I and II haplotypes and a high level of homozygosity. The nine DLA class II haplotypes recognized in this group of English bull dogs were only a fraction of the 88 haplotypes that were reported for dogs in 2007 . Genes within the DLA are important in regulating self/non-self recognition and immune responses and play a role in autoimmune disorders, allergies, and immun.
Populations that have lost genetic diversity through small founder numbers and artificial genetic bottlenecks are more likely to accumulate deleterious traits [33, 52]. A lack of genetic diversity also makes it harder to eliminate deleterious traits from a population once they are recognized. The mutation responsible for hyperuricosuria is carried by 25.5 % of English bulldogs and 3.1 % of the breed are homozygous and excrete uric acid . Elimination of this recessive mutation from the breed could lead to a significant loss of breed-wide genetic diversity.
A low level of breed-wide genetic diversity also limits the ability to rapidly introduce desired traits, usually in the form of simple recessive mutations. There is increased demand for smaller and more compact English bulldogs, dogs with wrinkled coats and rare coat colors. Such refinements in the breed create popular sire effects and yet more artificial genetic bottlenecks that will cause the loss of more genetic diversity if not properly managed. The negative effects of the rapid introduction of new genetic traits on the health have been best documented in the “The Rare Color Bulldog Craze”.Footnote 17 Indeed, English bulldog breeders appear to be more interested in adding recessive coat color mutations to increase puppy value than eliminating known deleterious mutations. English bulldog breeders around the world ordered 2482 tests from VGL UC Davis involving coat color between 2012 and 2016 compared to 62 tests for the hyperuricosuria mutation.Footnote 18
In conclusion, English bulldog breeders differ widely on their perception of health problems in their breed and what do about them. Some breeders blame disreputable or backyard “commercial” breeders for the unhealthy dogs that are being sold.Footnote 19
Footnote 20 However, genetic differences were not observed between pet English bulldogs seen at the UC Davis Veterinary Clinical Services and presumably healthy breeding dogs being genetically tested for certain traits. Healthy and unhealthy bulldogs shared the same alleles at genomic STR loci and the allele frequencies are virtually identical. The same was true for DLA class I and II haplotypes. Other English bulldog breeders believe that the health of the breed can be improved by breeding from within existing bloodlines, although there has been little movement by breeders to embrace this concept. This is fueled by purists that vigorously argue that any deviation from the original standards is no longer a British (English) bulldog , even though the breed has continued to evolve in appearance over centuries and even the last few decades. Still others believe that health cannot be restored from within the breed without resorting to outcrossing.12,13
The feelings of individual English bulldog breeders about the health of their breed and what if anything should be done about it may ultimately be taken out of their hands. English bulldog breeders across the world must take seriously constitutional amendments on the rights of animals. The European Union has recently updated their rules on animal welfare in 2015.Footnote 21 Although it was written specifically for farm animals; it holds that “animals” have rights of ‘freedom from discomfort” and ‘freedom from pain, injury and disease.” The EU rules on animal welfare have been restated in much greater detail by a 2013 constitutional amendment in Switzerland, which extended such rights to all animals . Although it has not been uniformly enforced, many Swiss breeders have proactively begun outcrossing English bulldogs with the Olde English Bulldogge to create what is known as the “Continental Bulldog”,Footnote 22 which will help bring the breed into compliance.