Genetic comparison of pure bred dogs in Sweden

Transcription

1 Genetic comparison of pure bred dogs in Sweden Maria Nord Degree project in biology, 007 Examensarbete i biologi, 0p, 007 Biology Education Centre and Departement of Evolutionary Biology Supervisor: Carles Vilá

2 Abstract The modern dog (Canis familiaris) has lived together with humans for thousands of years since the domestication of wolves, longer than any other animal. Although most of the modern breeds were established in the 19 th century, the dog has been transformed by humans since the time of domestication and its transformation into around 400 different breeds has been called the most complex genetic experiment in history. The dog is the most phenotypically diverse mammal and ranges in size from about 1kg for a small adult Chihuahua to around 100kg for a big Saint Bernard or Great Dane. In fact the difference in size and morphology is greater within the dog itself than between the different species in the family Canidae. This study is based on genetic data from 173 pure bred Swedish dogs, including German shepherds (16), Labrador retrievers (16), giant schnauzers (17), miniature schnauzers (14), Siberian huskies (16), fox terriers smooth (18), fox terriers wire (18), standard poodles (0), miniature poodles (0) and bull terriers (18). I investigated the inbreeding within the different breeds, the relationships between them and the accuracy of genetical assignment of one individual to a specific breed. The results show that the inbreeding varies for the different breeds, from the highest inbreeding with observed heterozygosity values of 0,7 for bull terriers to the lowest inbreeding with observed heterozygosity values 0,60 for Siberian huskies. The relationship between the breeds was well established and showed that the bull terrier was very different from all other breeds, probably as a result of the limited genetic diversity in this breed. It also revealed that the giant schnauzer was more closely related to German shepherds than to miniature schnauzers. Finally the assignment tests revealed an overall success rate of assigning a dog to its correct breed ranging from 94% to 100% depending on the breed. In average, the assignment success rate was 98%. This implies that breeds are very well separated from each other from the genetic point of view and genetic composition allows identification of an individual as belonging to one breed or another. 1

3 Table of Contents Introduction 3 Origin of dogs 3 Microsatellites and buccal swabs 3 Genetic variation 4 Aim of this study 4 Materials and Methods 5 Samples 5 DNA extraction 5 Microsatellite amplification 5 Genotyping 6 Data analysis 7 Results 8 Discussion 14 Inbreeding 14 Relationships 14 Assignment 15 Acknowledgements 16 References 17 Appendix 1 0

4 Introduction Origin of dogs The dog (Canis familiaris) was the first animal to be domesticated. In the past it was thought that they derived from several species such as wolves and jackals but today it is widely accepted that the dog has originated only from the domestication of wolves. Archaeological records place it in the company of humans more than years ago (Wayne and Ostrander 1999). This number however, is based on the identification of bone remains based on morphological differences. It s not unlikely that dogs were domesticated much earlier but kept morphologically similar to wild wolves for quite a long period of time. Small differences between early dogs and wild wolves would be very easy to miss in old bone remains. In fact, studies of differences between dog and wolf DNA indicate that the domestication of dogs could have happened as early as years ago (Vilà et al. 1997). However, the exact time is not known and different studies propose different dates. Numbers like 7 000, and years ago have also been suggested (Lindblad-Toh et al. 005, Savolainen et al. 00). Genetic results show at least four different origination or interbreeding events between dog and wolf. However evidence shows that some native North American dogs did not derive from the domestication of North American wolves. Instead, they are more closely related to Old World dogs (Leonard et al. 00). Interbreeding between wild wolves and modern dogs is also ongoing today (Vilà and Wayne 1999) and the separation between dogs and wolves may not always be clear. Whether this is a problem for wild wolf populations is debated. Crossing dogs and wolves is not commonly accepted and in some countries is prohibited. It is generally acknowledged that wolf-dog hybrids make poor pets. When humans domesticate animals these tend to decrease in body size, at least initially. This might feel a bit backwards since we protect them and provide them with food. But the food that humans provided was not necessarily the best or had the highest level of nutrients in it. It is also much easier to have a small animal, they eat less and are generally easier to control, so there is selection towards small animals. Today man has successfully bred different dog breeds that differ more in size than the difference in any other vertebrate, alive or extinct and the biggest breeds are much bigger than the wolf from which all breeds originated (Sutter et al. 007). Microsatellites and buccal swabs There are three types of genetic markers commonly used in domestication studies today. These are mitochondrial DNA (mtdna) sequences, Y-chromosome markers and autosomal microsatellites (Björnerfeldt 007). mtdna is inherited maternally and the Y-chromosome markers are inherited paternally and therefore they are useful complements to each other in genetic studies. Autosomal microsatellites, on the other hand, are biparentally inherited. Microsatellites are short DNA sequences composed of less than six bases tandemly repeated. These markers are widespread throughout the genome. Microsatellites seem to be fairly uniformly distributed over the genome of eukaryotes although under-represented in coding regions (Goldstein and Schlötterer 1999). Microsatellites mutate by adding or removing repeats, becoming shorter or longer. This happens through a process in the replication of the DNA where the polymerase slips and then misreads the template DNA strand (Levinson and 3

5 Gutman 1987, Schlötterer and Tautz 199). The mutation rate of microsatellites often exceeds 10-4 per generation, and it increases with the number of repeated units. As a result of this, microsatellites accumulate variation in a short amount of time and can be highly polymorphic in a population (Zajc et al 1996). This makes them highly suitable in the study of dog breeds despite the relatively recent evolutionary origin of dogs. Most studies on the DNA of dogs rely heavily on the willing participation of dog owners and organisations. The classic source for extracting DNA is blood. However, buccal swabs are one frequently used alternative. A buccal swab is done with a brush used for scraping the inside of the dog s cheek. Buccal swabs contain epithelial cells and provide a good amount of DNA for genetic analyses. Needless to say, it is much easier to get participation from dog owners when using buccal swabs than when taking blood samples. In one case the participation level increased almost by a factor of three (Oberbauer et al. 003). Genetic variation Genetic bottlenecks have been present during the evolution of the modern dog. First when the dog was domesticated from wolves and later during the establishment of modern breeds. Some studies show that, despite strict breeding programs and popular sire effects, dogs have medium to high levels of genetic variability, even within breeds: microsatellite heterozygosity often ranges between 36% and 55%, compared to wild wolf populations with an average value of 53%. This suggests a diverse founding stock and interbreeding between breeds (Ruvinsky and Sampson 001; Morera et al. 1999). However a number of studies have shown severe inbreeding within breeds with genetic homogeneity much higher than in humans or even purebred cats (Zajc et al. 1997). Between breeds genetic differentiation is very large. A study on Finnish dogs showed that genetic variation between different breeds was in average two times higher than the largest difference observed among breeds of sheep or human populations. In fact the largest divergence between dog breeds was only slightly lower than the lowest one reported between humans and chimpanzees (Koskinen and Bredbacka 000). Aim of this study The aim of this study was to investigate the genetic diversity of purebred dogs in Sweden. In specific three different questions are investigated. The first question refers to the level of inbreeding within the different breeds. As stated above some studies show quite low levels of inbreeding even within different breeds whereas others clearly show quite high levels of inbreeding. I compare very different dog breeds in Sweden to assess the range of observed inbreeding values. Since the inbreeding coefficient is a relative term that only indicates the mating preferences within the breed, I use expected heterozygosity as a measure of the diversity within each breed. Larger values are likely to be correlated with larger number of founders, and thus it implies lower overall inbreeding. Secondly the relationship between the different breeds was investigated, i.e. how closely related the different breeds were to each other. It seems fairly obvious that a giant- and miniature schnauzer should be more closely related to each other than to a bull terrier. But, what is their relative degree of differentiation? And the difference between giant schnauzer, Siberian husky and Labrador retriever? 4

6 Finally the question of the separation between dog breeds was addressed. That is, if we have a sample from one specific dog, to what degree of certainty can we say that it belongs to a certain breed? Materials and Methods Samples Buccal swabs were taken from 00 purebred dogs registered by the Swedish Kennel Club (SKK). The samples included German shepherds, Labrador retrievers, bull terriers, Siberian huskies, standard poodles, miniature poodles, giant schnauzers, miniature schnauzers, fox terrier smooth, and fox terrier wire. From each of these ten breeds, 0 samples were obtained. However, not all of them worked equally well in the genetic analyses and some samples had to be excluded (see below). The samples were collected at a dog show in December 004, and by direct correspondence with dog owners during 005. The registration numbers in the Swedish Kennel Club were recorded for all individuals to avoid sampling dogs that shared any parent. The buccal cells were collected from the dogs using nylon bristle cytology brushes (Medical Packaging Corp, Camarillo, CA). The inside of the dog s cheek was brushed for at least 0 seconds to ensure that enough DNA was collected. The brush with the sample was then put into a tube with 1 ml Laird s buffer (0.1M Tris-HCl, 5mM EDTA, 0.M NaCl, 7mM SDS, adjusted to ph 8.5). When the samples arrived at the laboratory they were kept at -0 C until they were processed. DNA extraction Genomic DNA was extracted from 400µl of the buffer containing the cytology brush with the sample, by digestion with 0.3mg of proteinase K. The samples were then incubated over-night at 37 C and DNA was extracted using a modified phenol/chloroform protocol (Sambrook et al. 1989). The concentration in each sample was measured using a NanoDrop.5.3 instrument (NanoDrop technologies, Delaware, USA) and were then diluted to a final concentration of 10ng/µl. Microsatellite amplification Twenty-eight biparentally inherited autosomal microsatellites, distributed throughout the canine genome, were typed for all of the dogs: Ren94K11, C17.40, Ren39K4, C18.460, Ren74F18, Ren181K04, C11.873, Ren73F08, Ren11I0, C0.894, Ren04K13, Ren160J0, Ren106I06 (Breen et al. 001), FH3109, FH887, FH914, FH785, FH759 (Guyon et al. 003), Ren37H09, Ren49F (Jouquand et al. 000), c017 (Francisco et al. 1996), u109, u5, u50, u53 (Ostrander et al. 1993), vwf (Shibuya et al. 1994), PEZ05 and PEZ1 (Perkin-Elmer, Zoogen; see NHGRI Dog Genome Project at nhgri.nih.gov/dog_genome/). The microsatellites were amplified by polymerase chain reaction (PCR). The amplifications for each sample were done in 14 reactions of 10µl each, which included 1 multiplexes of two 5

7 or three loci, and two loci amplified separately (PEZ05 and PEZ1) (see table 1). The PCR mix included 1x HotStar buffer (QIAGEN, Hilden, Germany), 0.5mM dntp, 0.3μM of each primer, 3.0mM MgCl, 0.05x Q solution, 0.45U HotStarTaq and µl DNA template. The PCR profile included an initial denaturation step at 95 C for 15 minutes followed by 10 touchdown cycles (30 s of denaturation at 95 C, 30 s annealing starting at 58 C and decreasing 0.5 C each cycle, followed by extension at 7 C for 45 s), followed by 0 additional cycles (denaturation at 95 C for 30 s, annealing at 53 C for 30 s and extension at 7 C for 45 s), and a final extension step at 7 C for 10 min in a PTC-05 DNA Engine Tetrad (Bio-Rad). Genotyping PCR products were pooled in seven different pools as described in table 1. For each sample, 1µl of the pooled samples were transferred to a 96 well ABI skirted plate which contained 10µl of a loading solution which also had 0.05µM ETRox 400 marker that would be used as size standard. PCR products were electrophoresed on a MegaBACE 1000 instrument (Amersham Biosciences). Genotypes were identified using the software Genetic Profiler v. (Amersham Biosciences). Table 1 Pooling of the microsatellite markers. Final Pool PCR Amount [µl] Pool A Multiplex: Ren94K11, C17.40 Multiplex: FH3109, Ren39K4 Water 4 Pool B Pool C Pool D Pool E Pool F Pool G Multiplex: C18.460, Ren74F18 Multiplex: Ren181K04, FH887 Water Multiplex: C11.873, FH914, FH785 Multiplex: Ren73F08, Ren11I0 Water Multiplex: C0.894, Ren37H09, Ren04K13 Water Multiplex: Ren160J0, FH759 Multiplex: Ren49F, Ren106I06 Water Multiplex: u53, c017 Multiplex: u109, u5 Water Multiplex: u50, vwf PEZ1 PEZ05 Water Only one replicate per sample was genotyped since the quality of the DNA was considered to be good enough to ensure reliable genotypes. Allelic dropouts and false alleles were thus expected to be uncommon and not pose a big problem. Sample quality is closely correlated with the likelihood of successful DNA amplification. In order to avoid samples that were not

8 working well and were more likely to lead to errors, only samples that were successfully genotyped for at least 16 out of the 8 microsatellites were considered for further analysis (Figure 1). The number of samples remaining for the study were: 16 German shepherds, 16 Labrador retrievers, 18 bull terriers, 16 Siberian huskies, 0 standard poodles, 0 miniature poodles, 17 giant schnauzers, 14 miniature schnauzers, 18 fox terrier smooth, and 18 fox terrier wire. Number of samples with different succes rates Number of Samples Succes Rate Figure 1 Success rate for different samples. The green bars indicate samples that were successfully typed for more than 16 of the microsatellites. Red bars represent samples typed for fewer loci and excluded from subsequent analyses. Data analysis In order to calculate the expected (H e ) and observed (H obs ) heterozygosity values (Nei 1978) and the inbreeding coefficient F IS for each of the 10 dog breeds, the program Microsatellite Toolkit 3.1 (Park 001) was used. The degree of differentiation between the breeds was quantified using GENETIX 4.05 (Belkhir et al ) by calculating pairwise F ST values (Weir and Cockerham 1984). Significance was assessed from 1000 permutations. This programme was also used to calculate the inbreeding coefficient, F IS. Significance was, once again, assessed from 1000 permutations. A Factorial Correspondence Analysis (FCA) plot was also generated to show the relationship of all individuals based on their genotypes. Using the matrix with the pairwise F ST values the programme MEGA 3.1 was used to construct a tree of breeds using a neighbour-joining (NJ) approach. An additional test to see how well separated the breeds were genetically was the proportion of self-assignment. The assignment program Doh (Paetkau et al. 1995; was used to evaluate if the individual genotypes could allow correct assignment of each dog to its breed. The programme estimates 7

9 the likelihood of finding that particular genotype within a specific breed and the individual is assigned to the breed for which it has the highest probability. The software STRUCTURE.1 (Pritchard et al. 000; Falush et al 003) was used to group individuals together based on their genetic composition only, without providing any population information. The dogs were subdivided by the program into an increasing number of populations (K=1-15 with a burn-in length of 100,000, run length of 1,000,000). Each one of the populations was defined by the program trying to put together the samples in groups as close to Hardy-Weinberg and linkage equilibrium as possible. Results For all of the 10 dog breeds we see a positive inbreeding coefficient as shown in table. The highest value is found in the German shepherd (F IS =0.168). This high value could indicate that there is fragmentation within this breed. The significant difference in expected and observed heterozygosity values for the German shepherds is also a good indication of nonrandom mating. The miniature poodle, fox terrier wire and giant schnauzer are three other breeds with high F IS -values and significant differences in expected and observed heterozygosity values. The lowest F IS -value that is significant is found in the Labrador retriever, a breed that also exhibits a small difference in expected and observed heterozygosity values. This is an indication that the Labrador retrievers have some sort of structure within them and relatively high genetic diversity. For the standard poodle, miniature schnauzer and fox terrier smooth F IS values are not significant. For the standard poodle the F IS -value is lowest (0.008) and there is almost no difference in expected and observed heterozygosity. Table The sample size, inbreeding coefficient (F IS ), expected and observed heterozygosity values (H E and H O ) for each population (± standard deviation). n.s p 0.05; * p<0.05; ** p<0.01; *** p< Population Sample size F IS H E (SD) H O (SD) Miniature poodle *** (± 0.089) (± 0.011) Standard poodle n.s (± 0.036) (± 0.011) Miniature schnauzer n.s (± ) (± 0.095) Giant schnauzer ** (± ) (± 0.055) Fox terrier smooth n.s (± ) (± 0.041) Fox terrier wire ** (± ) (± 0.034) Bull terrier * (± ) (± 0.016) German shepherd * (± ) (± 0.075) Labrador retriever * (± ) (± 0.064) Siberian husky ** (± ) (± 0.053) Table 3 shows the F ST values for the pairwise comparison between different breeds. This is a measure of the genetic distance between them. For example the F ST value between the miniature poodle and the standard poodle is 0.19, but the F ST value between the miniature 8

10 poodle and the miniature schnauzer is also This means that the genetic difference or distance is the same between the miniature poodle and the standard poodle as between the miniature poodle and the miniature schnauzer even though this is not obvious just from looking at them. The highest F ST values (ranging from 0.37 to 0.54) can be found for the bull terrier. It is extremely well separated from the German shepherds with a F ST -value of This is probably due to the reduced diversity in bull terriers, the breed with the lowest diversity (Figure ). The lowest values are shown for the miniature poodle with values ranging from to Comparatively low values are also present for miniature schnauzer, giant schnauzer, standard poodle, Labrador retriever and Siberian husky whereas higher values are observed for German shepherd, fox terrier smooth and wire. Generally the F ST -values are high and indicate that the breeds are well differentiated from each other. Figure corresponds to the FCA analysis generated using GENETIX. The plot illustrates the relative similarity between individuals. Figure 3 is generated using Mega 3.1 which constructs a phylogenetic tree over the different breeds. As expected both plots display the same pattern. The FCA analysis shows that all bull terriers cluster closely together (Figure ), well separated from all other breeds. In the tree of breeds (Figure 3) the bull terrier is separated from the other breeds with a very long branch. These two observations indicate that all bull terriers are similar to each other due to the low diversity mentioned above, and different from all other breeds. In figure 3 fox terrier smooth and wire cluster together and are well separated from the rest by a long branch. Figure also shows the similarity between the two fox terriers. It is also clear that they are not clustered as closely together as the bull terriers, for example, indicating more diversity. The long branches that separate them in figure 3 show that although they are more closely related to each other than to any other breed, they are actually very well separated from each other. German shepherd has a relatively long branch separating it from the rest of the breeds (Figure 3). In Figure the individuals cluster separated from the other breeds. In both figures the breed that is closest to the German shepherds is the giant Schnauzer. Labrador retriever, giant schnauzer, miniature schnauzer, standard poodle, miniature poodle and Siberian husky all have short branches in the population tree (Figure 3) and are not very well separated from each other in the FCA analysis (Figure ). The assignment test uses the genetic makeup for each individual dog to determine which breed that particular individual belongs to. The results from the assignment test are shown in table 4. This shows a high number of correct assignments. 170 out of 173, or 98.3% of all individuals were assigned to the correct breed. The three that were not correctly assigned were a giant schnauzer that was assigned to miniature poodle, a fox terrier wire assigned to miniature poodle and a German shepherd assigned to giant schnauzer. 9

11 Table 3 Genetic differentiation between breeds measured as pairwise F ST over 8 autosomal microsatellites. All values are highly significant (p<0.01). F ST Standard poodle Miniature Schnauzer Giant schnauzer Fox terrier smooth Fox terrier wire Bull terrier German shepherd Labrador retriever Siberian husky Miniature poodle Standard poodle Miniature schnauzer Giant schnauzer Fox terrier smooth Fox terrier wire Bull terrier German shepherd Labrador retriever

12 German shepherd Giant schnauzer Miniature schnauzer Minature poodle Fox terrier smooth Fox terrier wire Labrador retriever Siberian husky Standard poodle Bull terrier Figure FCA analysis summarizing the similarity between all individual dogs included in the study.. GermanShepherd GiantSchnauzer MiniatureSchnauzer SiberianHusky FoxTerrierSmooth FoxTerrierWire LabradorRetriever MiniaturePoodle StandardPoodle BullTerrier 0.05 Figure 3 A neighbour-joining tree, based on F ST, showing the relationship among the breeds. 11

13 Table 4 Assignment of the dogs in the breeds indicated in the rows to the breeds indicated in columns. Miniature poodle Standard poodle Miniature schnauzer Giant schnauzer Fox terrier smooth Fox terrier wire Miniature poodle Standard poodle Miniature Schnauzer Giant schnauzer Fox terrier smooth Fox terrier wire Bull terrier German shepherd Labrador retriever Siberian husky (100%) (100%) (100%) (5,9%) (94,1%) (100%) 1 (5,6%) (94,4%) Bull terrier (100%) German shepherd (6,%) (93,8%) 0 0 Labrador retriever Siberian husky (100%) (100%) 1

14 To choose the best number of populations (K) in which the program Structure can separate the data, we plotted the likelihood value lnp(d) for each K (Figure 4). The maximum value provides information on which K is the most probable for a given dataset. We tested K values between 1 and 15. The best K value was K=10, which corresponds to the number of breeds included in the study. K=11 has a slightly higher probability, but the increase was very small and the 11 th group that Structure defined consisted of just one individual and had such weak support that K=10 better reflects the structure in the data. lnp(d) K Figure 4: Plot of lnp (D) values for different numbers of populations (K). Since the maximum is obtained at K= 10, this indicates that the genetic diversity across all studied dogs is best explained dividing them in 10 groups (as many as the number of breeds studied). As can be seen from figure 5, when 10 groups are formed they correspond very well with the breeds and almost all individuals have a high probability for only one of the groups, indicating that the breeds represent well separated entities. Figure 5: Populations defined by STRUCTURE when K=10. Each column indicates, with different colours, the probability of each individual of belonging to each one of the 10 groups defined. All breeds appear very well separated from each other, and only individuals (in breed 4 and 6) do not seem to fit with the remaining individuals from the breed. 13

15 Discussion Inbreeding All breeds exhibits a positive inbreeding coefficient, suggesting that the mating pattern is nonrandom. F IS is defined as F IS = 1 - H o /H E. In the case of random mating F IS = 0 (and H o = H E ). If there is inbreeding F IS should be above zero and the closer to one the higher the level of inbreeding. If F IS is below zero the result shows outbreeding. Inbreeding indicates that members of one breed tend to mate with relatives more often than would be expected at random. This is often the case when there is population fragmentation (Wahlund effect). Our results show that some breeds might be constituted by separate groups of individuals (fragmentation). Out of 10 breeds we see that three of the F IS results are non significant: standard poodle, miniature schnauzer and fox terrier smooth (Table ). This means that these breeds may be more uniform than the rest. On the other hand, the German shepherd showed the highest F IS value of all breeds indicating that the group is fragmented and non-homogenous. The fragmentation for the German shepherd is interesting as it could support the idea that there are actually two different populations within the breed, one for show dogs and one for working dogs. A similar pattern was shown in an earlier study on long- and short-haired Weimaraners (Schrameyer et al. 005). Also the miniature poodle has a relatively high F IS value and a big difference in expected and observed heterozygosity supporting the idea that they are actually more than one group. A previous study has in fact showed that the miniature poodles should not be regarded as one uniform group but as four different groups depending on a combination of the dogs coat colour and their size (Björnerfeldt in press). In the beginning of this project it was expected that also the Labrador retrievers should show quite a high level of fragmentation as the breeding lines are quite well separated into working dogs and show dogs. However, our analysis only showed small differentiation. Since the majority of the samples were collected at dog shows it is not unreasonable to believe that the working dogs may not have been well represented in the samples. Relationships All the breeds are well separated genetically. This can be clearly seen from the F ST values which, although they differ between different breeds, all are quite high. The relationships between the different breeds are quite clearly shown in figures and 3. Figure shows all the individuals and how they are distributed in relation to each other, whereas figure 3 shows a tree of the breeds. Similarly, the Structure analysis confirms the difference between the breeds. The results show that the fox terrier smooth and the fox terrier wire are two groups well separated from the rest and from each other, although they have rather large intra-breed variation. The bull terrier and the German shepherd are very well separated from each other and from all other breeds, probably due to the fact that both breeds have limited genetic diversity. In figure the bull terriers cluster closely together in the bottom right corner of the FCA-plot and the German shepherds cluster together at the top right part of the plot. In figure 3 the relationship between all the breeds are clearly shown and the result confirms the result from the FCA-plot regarding the bull terrier, German shepherd, fox terrier smooth and fox terrier wire. All four breeds have long branches which mean that the breed is well 14

16 separated. It is important to notice that since there is no outgroup in the tree, it shows relations and differentiation but not evolution. Also the standard poodle has quite a long branch and the separation of the breed is also clearly visible in the FCA-plot. Assignment Assignment tests using the online computer program Doh demonstrated that the dogs could be correctly assigned to their breed of origin in 98,3% of the cases. Out of 173 individuals only 3 were assigned to the wrong breed which. The assignment probability tells us that the markers are good and provide ample information which allows to separate the different breeds. One of the missassigned individuals was a giant schnauzer that was assigned to miniature poodle. In figure we can see that some individuals from giant schnauzers are overlapping with miniature poodles. Neither of the breeds have very long branches in the tree, suggesting that they are not very well separated compared to some of the other breeds. It has also been shown in a previous study that miniature poodles should be considered as four separate groups (Björnerfeldt 007). This too can be a contributing cause for the missassignment. One other individual, this time a fox terrier wire, was also assigned to the miniature poodles. Looking at the clustering results from Structure in figure 5, these two individuals are showing up as quite different compared to the other individuals in the same breed. The giant schnauzer has mostly orange and some yellow and pink suggesting that the genetic structure is most similar to miniature poodle, miniature schnauzer and Siberian husky respectively. The fox terrier wire has mostly orange and yellow, suggesting that it is a mix between miniature poodle and Siberian husky. Since the results are so clear in suggesting that these two should not be a part of their breed of origin it could be due to a mix up of the samples, although this is unlikely. These missassignments are thus probably the result of some breeds having a much larger diversity than others, and some of them being fragmented. This implies that these breeds do not represent cohesive groups and dogs from other breeds could be wrongly assigned to them because the genetic data do not allow their identification as a single group. The third individual that was assigned to the wrong breed was a German shepherd assigned to giant schnauzer, which is the breed closest to German shepherds according to the Fst analysis. According to Structure German shepherds from a fairly uniform group (group 8). Looking at figures and 3 we see that German shepherd and giant schnauzer are closest relatives according to the phylogenetic tree and that giant schnauzer is the breed clustering closest to German shepherd. Therefore it might not be such a coincidence that one of the German shepherds is assigned to giant schnauzer. These results support what previous studies have shown. Koskinen (003) sampled 50 individual dogs from five different breeds and used only 10 microsatellite loci. The assignment success rate was 100% of individuals into their correct breeds of origin, and 100% exclusion success of individuals from reference populations. Parker et al. (004) carried out an even larger study where they genotyped five unrelated dogs from each of the 85 breeds with 96 microsatellite markers. The dogs could be correctly assigned to their breed of origin 99% of the time. These high success rates indicate that dog breeds are really differentiated from each other despite looking very much alike in some cases. 15

17 Acknowledgements There are many persons who have been a great help during the time of this project. First I would like to thank my supervisor Carles Vilà, without him this project would never even have started. His enthusiasm and commitment has been a great source of inspiration and he also offered lots of ideas on different analysis possible on the data. I would also like to thank Susanne Björnerfeldt who has been my number one source of help during the entire project. She taught me, more in detail, every aspect of the work involved in this study. And she never seemed to grow tired of answering my numerous questions. Last but not least I would like to thank all the people at my department, Evolutionary Biology, at EBC, Uppsala University for providing a warm and welcoming environment. It s been really inspiring to get to know all the dedicated people working there. And of course I d like to thank my family and my boyfriend for all the support, especially during the end of this project. 16

18 References Belkhir K, Borsa P, Chikhi L, Raufaste N, Bonhomme F ( ) GENETIX 4.05, logiciel sous Windows TM pour la génétique des populations. Laboratoire Génome, Populations, Interactions, CNRS UMR 5000, Université de Montpellier II, Montpellier (France). Björnerfeldt S, Hailer F, Nord M, Vilà C (007) Disruptive selection within dog breeds. (in press) Breen M, Jouquand S, Renier C, Mellersh CS, Hitte C, Holmes NG, Chéron A, Suter N, Vignaux F, Bristow AE, Priat C, McCann E, André C, Boundy S, Gitsham P, Thomas R, Bridge WL, Spriggs HF, Ryder EJ, Curson A, Sampson J, Ostrander EA, Binns MM, Galibert F (001) Chromosome-specific single-locus FISH probes allow anchorage of an 1800-marker integrated radiation-hybrid/linkage map of the domestic dog genome to all chromosomes. Genome Research 11: Falush D, Stephens M, Pritchard JK (003). Inference of population structure using multilocus genotype data: linked loci and correlated allele frequencies. Genetics 164: Francisco LV, Langston AA, Mellersh CS, Neal CL, Ostrander EA (1996) A class of highly polymorphic tetranucleotide repeats for canine genetic mapping. Mammalian Genome 7: Goldstein DB, Schlötterer C (1999) Microsatellites Evolution and Applications. Oxford University Press, Suffolk, UK. Guyon R, Lorentzen TD, Hitte C, Kim L, Cadieu E, Parker HG, Quignon P, Lowe JK, Renier C, Gelfenbeyn B, Vignaux F, DeFrance HB, Gloux S, Mahairas GG, André C, Galibert F, Ostrander EA (003) A 1-Mb resolution radiation hybrid map of the canine genome. Proceedings of the National Academy of Sciences 100: Jouquand S, Priat C, Hitte C, Lachaume P, André C, Galibert F (000) Identification and characterization of a set of 100 tri- and dinucleotide microsatellites in the canine genome. Animal Genetics 31: Koskinen MT (003) Individual assignment using Microsatellite DNA reveals unambiguous breed identification in the domestic dog. Animal Genetics 34: Koskinen MT, Bredbacka P (000) Assessment of the population structure of five Finnish dog breeds with microsatellites. Animal Genetics 31: Leonard JA, Wayne RK, Wheeler J, Valadez R, Guillén S, Vilà C (00) Ancient DNA Evidence for Old World Origin of New World Dogs. Science 98: Levinson G, Gutman GA (1987) High frequencies of short frameshifts in poly-ca/tg tandem repeats borne by bacteriophage M13 in Escherichia coli K-1. Nucleic Acids Research 15:

19 Lindblad-Toh K, Wade CM, Mikkelsen TS, Karlsson EK, Jaffe DB, Kamal M, Clamp M, Chang JL, Kulbokas III EJ, Zody MC, Mauceli E, Xie X, Breen M, Wayne RK, Ostrander EA, Ponting CP, Galibert F, Smith DR, dejong PJ, Kirkness E, Alvarez P, Biagi T, Brockman W, Butler J, Chin C-W, Cook A, Cuff J, Daly MJ, DeCaprio D, Gnerre S, Grabherr M, Kellis M, Kleber M, Bardeleben C, Goodstadt L, Heger A, Hitte C, Kim L, Koepfli K-P, Parker HG, Pollinger JP, Searle SMJ, Sutter NB, Thomas R, Webber C, Lander ES (005) Genome sequence, comparative analysis and haplotype structure of the domestic dog. Nature 438: Morera L, Barba CJ, Garrido JJ, Barbancho M, de Andrés DF (1999) Genetic Variation Detected by Microsatellites in Five Spanish Dog Breeds. The Journal of Heredity 90: Nei M (1978) Estimation of average heterozygosity and genetic distance from a small number of individuals. Genetics 89: Oberbauer AM, Grossman DI, Eggleston ML, Irion DN, Schaffer AL, Pedersen NC, Belanger JM (003) Alternatives to Blood as a Source of DNA for Large-scale Scanning Studies of Canine Genome Linkage. Veterinary Research Communications 7:7-38 Ostrander EA, Sprague GF, Rine J (1993) Identification and characterization of dinucleotide repeat (CA)n markers for genetic-mapping in dog. Genomics 16: Park SDE (001) Trypanotolerance in West African Cattle and the Population Genetic Effects of Selection. Ph.D. thesis, University of Dublin. ( Parker HG, Kim LV, Sutter NB, Carlson S, Lorentzen TD, Malek TB, Johnson GS, DeFrance HB, Ostrander EA, Kruglyak L (004) Genetic Structure of the Purebred Domestic Dog. Science 304: Pritchard JK, Stephens M, Donnelly P (000) Inference of population structure using multilocus genotype data. Genetics 155: Ruvinsky A, Sampson JJ (001) The Genetics of the Dog. CABI Publishing, Wallingford, Oxon (UK) Sambrook J, Fritsch EF, Maniatis T (1989) Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Savolainen P, Zhang Y, Luo J, Lundeberg J, Leitner T (00) Genetic evidence for an East Asia origin of domestic dogs. Science 98: Schlötterer C, Tautz D (199) Slippage synthesis of simple sequence DNA. Nucleic Acids Research 0: Schrameyer T, Dekomien G, Pasternack SM, Reinartz BS, Santos EJM, Epplen JT (005) Long- and short-haired Weimaraner dogs represent two populations of one breed. Electrophoresis 6:

20 Shibuya H, Collins BK, Huang TH, Johnson GS (1994) A polymorphic (AGGAAT)n tandem repeat in an intron of canine von Willebrand factor gene. Animal Genetics 5: 1. Sutter NB, Bustamante CD, Chase K, Gray MM, Zhao K, Zhu L, Padhukasahasram B, Karlins E, Davis S, Jones PG, Quignon P, Johnson GS, Parker HG, Fretwell N, Mosher DS, Lawler DF, Satyaraj E, Nordborg M, Lark KG, Wayne RK, Ostrander EA (007) A single IGF1 allele is a major determinant for small size in dogs. Science 316: Wayne RK, Ostrander EA (1999) Origin, genetic diversity, and genome structure of the domestic dog. Genes and genomes 1: Weir BS, Cockerham CC (1984) Estimating F-statistics for the analysis of population structure. Evolution 38: Vilà C, Savolainen P, Maldonado JE, Amorim IR, Rice JE, Honeycutt RL, Crandall KA, Lundberg J, Wayne RK (1997) Multiple and ancient origins of the domestic dog. Nature 76: Vilà C, Maldonado JE, Wayne RK (1999) Phylogenetic relationships, evolution, and genetic diversity of the domestic dog. The Journal of Heredity 90: Zajc I, Mellersh CS, Sampson J (1997) Variability of canine microsatellites within and between different dog breeds. Mammalian Genome 8:

21 Appendix 1 Allele frequencies for all populations by locus Locus Populations Ren94K11 Mpoodle SPoodle MSchn GSchn FTSmo FTWire BullTerr GerShep Labbe SHusky C17.40 Mpoodle SPoodle MSchn GSchn FTSmo FTWire BullTerr GerShep Labbe SHusky FH3109 Mpoodle SPoodle MSchn GSchn FTSmo FTWire BullTerr GerShep Labbe SHusky

22 Ren39K4 Mpoodle SPoodle MSchn GSchn FTSmo FTWire BullTerr GerShep Labbe SHusky C Mpoodle SPoodle MSchn GSchn FTSmo FTWire BullTerr GerShep Labbe SHusky Ren74F18 Mpoodle SPoodle MSchn GSchn FTSmo FTWire BullTerr GerShep Labbe SHusky Ren181K04 Mpoodle SPoodle MSchn GSchn FTSmo FTWire BullTerr GerShep Labbe SHusky

23 FH887 Mpoodle SPoodle MSchn GSchn FTSmo FTWire BullTerr GerShep Labbe SHusky C Mpoodle SPoodle MSchn GSchn FTSmo FTWire BullTerr GerShep Labbe SHusky FH914 Mpoodle SPoodle MSchn GSchn FTSmo FTWire BullTerr GerShep Labbe SHusky

24 FH785 Mpoodle SPoodle MSchn GSchn FTSmo FTWire BullTerr GerShep Labbe SHusky Ren73F08 Mpoodle SPoodle MSchn GSchn FTSmo FTWire BullTerr GerShep Labbe SHusky Ren11I0 Mpoodle SPoodle MSchn GSchn FTSmo FTWire BullTerr GerShep Labbe SHusky

25 C0.894 Mpoodle SPoodle MSchn GSchn FTSmo FTWire BullTerr GerShep Labbe SHusky Ren37H09 Mpoodle SPoodle MSchn GSchn FTSmo FTWire BullTerr GerShep Labbe SHusky Ren04K13 Mpoodle SPoodle MSchn GSchn FTSmo FTWire BullTerr GerShep Labbe SHusky

26 Ren160J0 Mpoodle SPoodle MSchn GSchn FTSmo FTWire BullTerr GerShep Labbe SHusky FH759 Mpoodle SPoodle MSchn GSchn FTSmo FTWire BullTerr GerShep Labbe SHusky Ren49F Mpoodle SPoodle MSchn GSchn FTSmo FTWire BullTerr GerShep Labbe SHusky

27 Ren106I06 Mpoodle SPoodle MSchn GSchn FTSmo FTWire BullTerr GerShep Labbe SHusky Mpoodle SPoodle MSchn GSchn FTSmo FTWire BullTerr GerShep Labbe SHusky Mpoodle SPoodle MSchn GSchn FTSmo FTWire BullTerr GerShep Labbe SHusky Mpoodle SPoodle MSchn GSchn FTSmo FTWire BullTerr GerShep Labbe SHusky

28 5 Mpoodle SPoodle MSchn GSchn FTSmo FTWire BullTerr GerShep Labbe SHusky Mpoodle SPoodle MSchn GSchn FTSmo FTWire BullTerr GerShep Labbe SHusky vwf Mpoodle SPoodle MSchn GSchn FTSmo FTWire BullTerr GerShep Labbe SHusky

29 PEZ 1 Mpoodle SPoodle MSchn GSchn FTSmo FTWire BullTerr GerShep Labbe SHusky PEZ 5 Mpoodle SPoodle MSchn GSchn FTSmo FTWire BullTerr GerShep Labbe SHusky