A Genealogical Survey of Australian Registered Dog Breeds ⇑ Mohammad R

The Veterinary Journal xxx (2011) xxx–xxx

Contents lists available at ScienceDirect

The Veterinary Journal

journal homepage: www.elsevier.com/locate/tvjl

A genealogical survey of Australian registered dog breeds ⇑ Mohammad R. Shariﬂou, John W. James, Frank W. Nicholas, Claire M. Wade

Faculty of Veterinary Science, University of Sydney, NSW 2006, Australia article info abstract

Article history: Breeding practices were analysed for 32 registered dog breeds representing very small registries (120 Available online xxxx Central Asian shepherd dogs) through to very large registries (252,521 German shepherd dogs) in Australia. The vast majority (91%) of registered kennels in Australia that were sampled did not regularly Keywords: employ either close breeding or popular sire usage in their kennels and the weighted mean inbreeding Canine coefficient of Australian pedigree dogs was <5%. Australian breed mean inbreeding coefficients ranged Inherited disorders from 0% (Central Asian shepherd dog) to 10.1% (Bichon Frise). Breed effective population sizes ranged Inbreeding from 26 (Ibizan hound) to 1090 (Golden retriever), comparable with other species of domesticated Effective population size animals. Popular sire Pedigree analysis The relatively low levels of inbreeding suggest that pedigree dog disorders are unlikely to arise frequently from the use of popular sires or close breeding in Australian registered dog breeds. It is pos- sible that deleterious allele fixation might be driven by founder effects, genetic drift or adverse selection practices, which were not assessed in this analysis. European popular sire definitions should be revisited for rare breeds. Ó 2011 Published by Elsevier Ltd.

Introduction We also assessed the relative application of risk practices between kennels with different levels of genetic contribution to The pedigree dog community recently has been criticised in the these breeds. Kennels were separated into those that contribute media for using breeding practices unfavourable for population more than 50 offspring to the breed (major kennels) and those health, most notably selection for traits with adverse affects on with fewer than 50 registered offspring (minor kennels). The health, close inbreeding and heavy use of popular sires. The actual objective was to separate kennel behaviour according to com- application of these practices within breed registries currently is mitment to breeding and genetic influence within the breeds not well understood and allegations frequently are based on hear- and population-size cohorts of breeds examined. Understanding say and presumption rather than on scientific analysis. To better the behaviour of breeds and kennels according to simple dis- understand the actual nature of pedigree dog population structure criminators (such as relative contributions to registrations) will on a whole-population basis, we examined data records for a rep- enable better targeting of education to improve animal welfare resentative sample of recognised pedigree breed populations in practices. Australia. Using complete national pedigree data supplied by the Austra- Materials and methods lian National Kennel Council (ANKC), this study examined the levels of inbreeding and popular sire use in 32 Australian dog Pedigree information for 32 Australian dog breeds (Table 1) was provided by the breeding populations with different registry sizes. Dogs registered ANKC. The breeds were chosen as representative of the entire range of the number of registrations of Australian breeds, from very small registries to very large regis- with the 205 breeds in the ANKC are estimated to account for 20% tries. Data were filtered to remove duplicate individuals and those with missing of puppies homed in Australia annually (ANKC, 2011). By studying information. Animals that were recorded as their own ancestor or that were re- both large and small populations of dogs, we can observe the rela- corded as both dam and sire were omitted from the analysis. The pedigree data tionships between base population size and inbreeding risk in real for analysed breeds commenced in years 1943 (Labrador retriever) to 1990 (Central populations. The evaluation of data over the entire registration his- Asian shepherd dog). The final represented year was 2008 or 2009 for all breeds. After filtering, the total number of registered dogs represented in the data was tories of breeds shows inbreeding trends over time. 1,234,475. The genders of animals were determined only from their positions in the pedigree and hence animals not used for breeding were recorded as having unknown gender. Data for each of the 32 observed breeds, including the year from which data ⇑ Corresponding author. Tel.: +61 2 93518097. were available, the breed size grouping and the number of observed animals in each E-mail address: [email protected] (C.M. Wade). breed, are presented in Table 1. Breeds were divided into those with registration

1090-0233/$ - see front matter Ó 2011 Published by Elsevier Ltd. doi:10.1016/j.tvjl.2011.06.020

Table 1 List of dog breeds in alphabetical order, the year from which data were available, analysis group and the total number of animals in each breed.

Breed Data available from Analysis group Registered animals Australian cattle dog 1954 Large registry 74,290 Australian terrier 1955 Medium registry 22,460 Bichon Frise 1971 Medium registry 13,790 Border collie 1954 Large registry 68,535 Bouvier des Flandres 1975 Small registry 1769 Boxer 1955 Large registry 75,316 British bulldog 1956 Medium registry 16,207 Cavalier King Charles spaniel 1958 Large registry 77,108 Central Asian shepherd dog 1990 Small registry 120 English cocker spaniel 1952 Large registry 78,902 Collie (smooth) 1972 Small registry 1534 Dachshund (miniature wire-haired) 1962 Small registry 1991 Dandie Dinmont terrier 1963 Small registry 445 Fox terrier (smooth) 1955 Medium registry 19,302 German shepherd dog 1959 Very large registry 252,521 Golden retriever 1952 Very large registry 98,542 Harrier 1971 Small registry 134 Ibizan hound 1983 Small registry 130 Italian Spinone 1985 Small registry 190 Labrador retriever 1943 Very large registry 135,550 Lakeland terrier 1957 Small registry 1644 Leonberger 1988 Small registry 512 Nova Scotia duck tolling retriever 1989 Small registry 901 Papillon 1956 Medium registry 17,157 Pekingese 1958 Large registry 34,325 Polish lowland sheepdog 1984 Small registry 166 Poodle (miniature) 1954 Large registry 31,441 Rottweiler 1960 Very large registry 98,727 Samoyed 1955 Medium registry 20,290 Skye terrier 1958 Small registry 1739 Staffordshire bull terrier 1957 Very large registry 88,589 Sussex spaniel 1979 Small registry 148 Total 1,234,475

numbers <2000, 2001–25,000, 25,001–80,000 and >80,000, here named as ‘small >50% of progeny in the kennel were sired by dogs that had in turn sired >5% of prog- registries’, ‘medium registries’, ‘large registries’ and ‘very large registries’, eny registered in the breed (over all kennels of that breed) in the previous 5 years respectively. (Fédération Cynologique Internationale [FCI], 2011). Restricted maximum In the genealogical analysis, animals with known parents born from 2000 to likelihood (Genstat; Payne et al., 2008) was used to assess the significance of fixed 2009 inclusive were used as the reference population. The completeness of pedigree effects (breed, breed size, kennel type, close-breeding and popular sire usage) in the information was determined by complete equivalent generations (EqG) in the ref- kennel behaviour analysis. erence population. Founders (f) were defined as animals with unknown parents (typically imported dogs with known parentage but outside of the Australian registry) contributing to the gene pool in the reference population. To account for differ- Results ing relative contributions of each true founder, the effective number of founders (fe) was determined based on the concept of ‘equivalent founders’ (Lacy, 1989) using Pedigree information, determined by the number of generation the algorithm of Boichard et al. (1997), as instituted in the programme prob_orig equivalents (EqG), varied among breeds, ranging from almost no in the Pedig package (Boichard, 2002). information content to highly informative pedigrees (Table 2). The effective population size (Ne) was estimated from the rate of inbreeding per generation, using the formula Ne = 1/2DF (Wright, 1923; Falconer and Mackay, Among the breeds studied, the Central Asian shepherd dog, Polish 1996). DF was calculated following Maki (2010) as DF =(b Ã l)/(1 À (Fly À b Ã l)), lowland sheepdog and Sussex spaniel had values of EqG < 3 and so where ‘b’ is the regression coefficient of the average inbreeding coefficient on year provided the poorest pedigree information. The Australian cattle of birth, ‘l’ is the average generation interval and ‘Fly’ is the average inbreeding coefficient in the last year of birth available for the study. An inbreeding coefficient was dog, Australian terrier and Bichon Frise breeds had EqG values >9 calculated for each animal, including those in the reference population, by the and so provided the richest pedigrees for analysis. Two of the most method of Meuwissen and Luo (1992), using CFC software (Sargolzaei et al., informative breeds (Australian cattle dog and Australian terrier) 2006). The purpose of including the reference population was to see if there were were derived in Australia. recent changes in inbreeding trend. Estimated effective population sizes (N ) ranged from 26 (Ibizan Pedig programmes ngen, segreg and intgen were used to calculate complete e hound) to 1090 (Golden retriever). The number of observed foun- equivalent generations (EqG), the number of founder genome equivalents (Ng) and generation interval (l), respectively. The Pedig programme prob_orig was used ders (f) ranged from 12 (Central Asian shepherd dog and Harrier) to calculate the total number of founders (f), the effective number of founders (fe) to 1893 (German shepherd dog), while the number of effective and the effective number of ancestors (fa), the latter being taken as the lower founders (fe) ranged from 8 (Harrier and Ibizan hound) to 258 boundary of the marginally contributing ancestors after 100 iterations. Ng was calculated using 500 iterations. (Cavalier King Charles spaniel) (Table 2). Results comparable to Kennels were divided into major kennels (registering 50 or more progeny) or those of previous studies were observed for effective number of minor kennels (registering <50 progeny). Kennels were defined in the data by the ancestors (fa) and the effective number of founder genomes (Ng) use of a kennel prefix. The purpose of this comparison was to differentiate breeders (Leroy et al., 2006; Maki, 2010). demonstrating a longer-term influence on breed genetics from those breeding few As expected, the number of observed founders (f) was greater litters. We were interested to know whether the major and minor kennels behave differently with respect to breeding practices, including the use of close breeding than the number of effective founders (fe) in all breeds, with the and popular sires. For this purpose, we defined kennels as being ‘close-breeding’ latter term indicating the number of founders that could account kennels when more than 50% of the progeny born to the kennel had inbreeding for the observed genetic diversity if each founder contributed coefficients >10%. We defined a kennel as a regular user of popular sires when equally. The ratio (fe/f) is relatively high (mean 0.5) in ‘small regis-

Table 2 Genealogical parameters for each breed and analysis group.

Breed Reference Generation Effective Mean Proportion Generation Observed Founder Effective Founder fe/ffa/fe Ng/fe animals equivalents population F F 6 10% interval (l) founders equivalents ancestors genomes

(EqG) size (Ne) (f) (fe) (fa) (Ng) Bouvier des Flandres 218 3.9 NAa 0.070 0.72 4.1 111 22 11 7.5 0.20 0.50 0.34 Central Asian shepherd dog 106 2.1 NAa 0.000 1.00 4.8 12 9 6 3.8 0.75 0.67 0.42 Collie (smooth) 429 3.5 88 0.026 0.88 3.7 135 48 36 17.4 0.36 0.75 0.36 Dachshund (miniature 479 5.4 84 0.048 0.80 3.7 161 50 21 11.7 0.31 0.42 0.23 wire-haired) Dandie Dinmont terrier 116 4.4 75 0.037 0.89 3.9 36 15 7 3.9 0.42 0.47 0.26 Harrier 84 3.1 59 0.039 0.78 3.8 12 8 4 3.2 0.67 0.50 0.40 Ibizan hound 54 3.4 26 0.098 0.51 5 15 8 4 2.3 0.53 0.50 0.29 Italian Spinone 52 3.3 NAa 0.061 0.66 4.2 16 12 7 4 0.75 0.58 0.33 Lakeland terrier 410 4.8 58 0.061 0.73 3.7 71 31 16 8.2 0.44 0.52 0.26 Leonberger 144 4.1 508 0.002 1.00 3.3 149 63 17 10.7 0.42 0.27 0.17 Nova Scotia duck tolling 589 4.1 47 0.029 0.90 4.6 84 26 12 9.1 0.31 0.46 0.35 retriever Polish lowland sheepdog 12 1.7 69 0.039 0.80 5.3 17 14 10 5.2 0.82 0.71 0.37 Skye terrier 209 7.6 40 0.061 0.76 4.5 126 62 23 7.6 0.49 0.37 0.12 Sussex spaniel 61 2.6 55 0.023 0.82 4.9 23 12 8 4.4 0.52 0.67 0.37 Australian terrier 3352 10.1 49 0.075 0.69 3.9 298 42 25 8.7 0.14 0.60 0.21 Bichon fries 4273 9.4 53 0.101 0.61 3.8 256 52 25 9.8 0.20 0.48 0.19 British bulldog 5536 7.7 101 0.056 0.79 3.4 343 83 31 15.5 0.24 0.37 0.19 Fox terrier (smooth) 3140 8.3 40 0.079 0.67 4.3 329 19 12 7.4 0.06 0.63 0.39 Papillon 4959 8.1 120 0.045 0.84 3.9 570 138 47 23.9 0.24 0.34 0.17 Samoyed 3598 7.6 60 0.048 0.81 4.2 391 79 33 14.8 0.20 0.42 0.19 Australian cattle dog 10,956 9.7 57 0.057 0.78 4.1 546 87 59 22.5 0.16 0.68 0.26 Border collie 20,173 7.6 129 0.041 0.88 4.4 511 87 52 25.2 0.17 0.60 0.29 Boxer 16,267 6 113 0.043 0.82 3.9 910 171 93 56.8 0.19 0.54 0.33 Cavalier King Charles 29,855 7.5 204 0.035 0.89 3.5 836 258 113 55.9 0.31 0.44 0.22 spaniel English cocker spaniel 14,237 7.3 89 0.044 0.83 4.2 837 173 66 31.3 0.21 0.38 0.18 Pekingese 2352 8.7 118 0.033 0.87 3.8 1053 209 54 24.6 0.20 0.26 0.12 Poodle (miniature) 4279 6.6 80 0.052 0.80 3.9 561 99 56 27.4 0.18 0.57 0.28 German shepherd dog 34,870 4.8 250 0.012 0.96 4.3 1893 91 71 49.2 0.05 0.78 0.54 Golden retriever 31,834 6.7 1090 0.051 0.82 4.3 727 109 61 33.3 0.15 0.56 0.31 Labrador retriever 44,396 6.6 153 0.034 0.89 4.1 1134 146 73 35 0.13 0.50 0.24 Rottweiler 15,405 4.9 191 0.025 0.91 4.4 883 195 107 70.5 0.22 0.55 0.36 Staffordshire bull terrier 36,412 8 102 0.052 0.81 3.4 625 81 47 27.2 0.13 0.58 0.34 Breed size mean Small 212 3.9 101 0.042 0.80 4.3 69 27 13 7.1 0.50 0.53 0.31 registry Medium 4143 8.5 71 0.067 0.74 3.9 365 69 29 13.4 0.18 0.47 0.22 registry Large 14,017 7.6 113 0.044 0.84 4.0 751 155 70 34.8 0.20 0.49 0.24 registry Very 32,583 6.2 357 0.035 0.88 4.1 1052 124 72 43.0 0.14 0.59 0.36 large registry

a Not available, cannot be computed because of null or negative DF.

try’ breeds, but roughly half this value in the other three groups showed at least some levels of inbreeding (Table 2). The breed (mean 0.18, 0.20 and 0.14 in ‘medium registries’, ‘large registries’ mean inbreeding coefﬁcients over all years ranged from 0 (Central and ‘very large registries’, respectively) (Table 2). It is evident that Asian shepherd dog) to 0.101 (Bichon Frise). founder contribution has been more uneven in breeds with larger The mean breed inbreeding coefﬁcients within birth year were registries than those with smaller registries. used to determine the inbreeding trends over time (Fig. 1). For breed

The ratio (fa/fe) describes the impact of genetic bottlenecks on size groups other than ‘small registries’ (for which meaningful breed populations (Boichard et al., 1997). The geographical isola- trends could not be detected due to large annual ﬂuctuations), the tion of Australia from the originating countries of the majority of general observed trend was for inbreeding to increase as breeds be- dog breeds predicts that all breeds of non-local origin should be come established. bottlenecked; this is observed in the data, with the mean ratio After establishment, the majority of breeds demonstrated at

(fa/fe) being in the vicinity of 0.5, irrespective of the numerical size least some level of inbreeding in the majority of dogs. Here, inbred of the breed. Breeds of Australian derivation (Australian terrier and dogs are regarded as those with any known co-ancestry between Australian cattle dog) still appeared to have breed bottlenecks and the parents; this co-ancestry may be quite distant, although limited their ratios of (fa/fe) were exceeded by some imported breeds, by the depth of the pedigree (EqG). In 19/32 breeds, the trend is for a including the Polish lowland sheepdog, the Smooth collie and the decline in mean inbreeding coefficient after the year 2000 (the German shepherd dog (Table 2). breeds are labelled as ‘decrease’ or ‘faster decrease’ in Table 3). This can be seen in negative values for DF (change in inbreeding coeffi- Inbreeding cient on an annual basis) over that period. The highest mean inbreeding coefficients are generated by ‘medium registry’ breeds. Other than the Central Asian shepherd dog, for which no inbred One way to prevent excessive inbreeding is to prevent matings individuals were observed in the data, all remaining 31 breeds among close relatives. Fig. 2 shows the percentage of matings

(a) 0.15

0.1

0.05

Breed mean inbreeding coefficient (F) coefficient inbreeding Breed mean 0 1971 1979 1985 1989 1995 1999 2005 1973 1975 1977 1981 1983 1987 1991 1993 1997 2001 2003 2007 2009 Year of birth Australian terrier Bichon Frise British Bulldog Fox terrier (smooth) Papillon Samoyed

(b) 0.15

0.1

0.05

0 Breed mean inbreeding coefficient (F) coefficient Breed mean inbreeding 1975 1985 1987 1989 1997 1999 2001 2009 1971 1973 1977 1979 1981 1983 1991 1993 1995 2003 2005 2007 Year of birth

Australian cattle dog Border collie Boxer Cavalier King Charles spaniel English cocker spaniel Pekingese Poodle (miniature)

0.1

0.05

Breed mean inbreeding coefficient (F) coefficient inbreeding Breed mean 0 1971 1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1995 1997 1999 2001 2003 2005 2007 2009 1993 Year of birth German shepherd dog Golden retriever Labrador retriever Rottweiler Staffordshire bull terrier

Fig. 1. Breed mean inbreeding coefﬁcients by year of birth since 1971 for breeds with >2000 registrations. (a) ‘Medium registry’ breeds (registered dog numbers between 2001 and 25,000). (b) ‘Large registry’ breeds (registered numbers between 25,001 and 80,000). (c) ‘Very large registry’ breeds (registered numbers >80,000).

among ﬁrst-degree relatives (sire to daughter, dam to son, full- Australian data, organised by breed size. Across breed size groups, brother to sister) and second-degree relatives (grandparent to the mean frequency of ﬁrst-degree and second-degree matings was grand-offspring, half-brother to sister, uncle to niece) in the 2.6% and 9.8%, respectively.

Table 3 Rate of inbreeding over all years and in reference population (2000–2009).

Breed DF in entire population DF in reference population Change of DF in reference population compared to entire population Australian cattle dog 0.009 0.006 Slower increase Australian terrier 0.010 À0.006 Decrease Bichon Frise 0.010 À0.012 Decrease Border collie 0.004 0.004 No change Bouvier des Flandres À0.001 À0.057 Faster decrease Boxer 0.004 À0.002 Decrease British bulldog 0.005 À0.009 Decrease Cavalier King Charles spaniel 0.002 0.002 No change Central Asian shepherd dog 0.000 0.000 No change English cocker spaniel 0.006 À0.002 Decrease Collie (smooth) 0.006 À0.007 Decrease Dachshund (miniature wire-haired) 0.006 0.002 Slower increase Dandie Dinmont terrier 0.007 0.004 Slower increase Fox terrier (smooth) 0.012 À0.012 Decrease German shepherd dog 0.002 À0.001 Decrease Golden retriever 0.000 À0.004 Decrease Harrier 0.008 0.045 Faster increase Ibizan hound 0.019 À0.107 Decrease Italian Spinone À0.009 À0.018 Faster decrease Labrador retriever 0.003 À0.001 Decrease Lakeland terrier 0.009 0.030 Faster increase Leonberger 0.001 À0.004 Decrease Nova Scotia duck tolling retriever 0.011 0.007 Slower increase Papillon 0.004 À0.001 Decrease Pekingese 0.006 0.001 Slower increase Polish lowland sheepdog 0.007 À0.055 Decrease Poodle (miniature) 0.006 À0.010 Decrease Rottweiler 0.003 À0.006 Decrease Samoyed 0.008 À0.007 Decrease Skye terrier 0.013 0.006 Slower increase Staffordshire bull terrier 0.005 0.005 No change Sussex spaniel 0.009 0.059 Faster increase

Progeny contributions by individuals (popular sires/dams) dams produce >10 breeding progeny. The maximum observed number of breeding progeny for each breed size group is shown. Observing sire proliﬁcacy without regard to breed registrations The most proliﬁc animal in the data set produced 62 breeding (Table 4) shows that 6% of sires in the data contribute >30 regis- progeny. tered progeny. The general trend is for the proportion of individuals with >30 progeny to increase with the number of registrations Kennel breeding behaviour in the breed. Rather than considering popular sires or dams as those leaving There are four categories of kennel breeding behaviour in Ta- many progeny regardless of future breeding status, from the point ble 6 based on the extent of use of popular sires and the extent of view of breed health it is rational to consider animals (sires or of close breeding. The higher risk kennels (category PF) have dams) that leave many breeding progeny. Considering the data P50% progeny from popular sires and P50% progeny with from this viewpoint allows us to assess potential impacts from rare F > 0.10; the lower risk kennels (category pf) have <50% progeny recessive alleles carried by popular parents and the impact that from popular sires and <50% progeny with F > 0.10; the other these might have on the population (Table 5). While there are two categories are intermediate. Because imported dogs in this 2220 animals in the data with >50 progeny, only 271 have >10% analysis are treated as founders, there is a very high correlation of their progeny going on to breed and only 240 of all sires and (0.86) between the proportion of progeny with F > 0.10 and the

Very large

Large Size of registry Medium

Small

0.0 5.0 10.0 15.0 Percentage of matings

Fig. 2. Matings between ﬁrst- and second-degree relatives as a percentage of all matings within breed size groups.

Table 4 Proportions of sires within breed size groups producing offspring numbers in the ranges listed.

Breed size Number of progeny per sire 1–10 11–20 21–30 31–40 41–50 51–200 >200 Small registry 0.872 0.093 0.025 0.005 0.002 0.003 0.000 Medium registry 0.777 0.147 0.042 0.014 0.007 0.012 0.000 Large registry 0.725 0.173 0.051 0.020 0.010 0.020 0.001 Very large registry 0.627 0.212 0.083 0.033 0.014 0.027 0.004 Weighted mean 0.690 0.186 0.064 0.025 0.011 0.022 0.002

Table 5 Analysis of animal popularity (sires and dams) by analysis group comparing the total number of progeny and the number of progeny that goes on to become a sire or a dam.

Breed size >10 progeny >50 >50 progeny and >10 >50 progeny and >10 Maximum number of observed breeding breeding progeny progeny breeding progeny breeding progeny for any single individual (sire or dam) Small registry 0 24 3 0 8 Medium registry 41 326 50 36 31 Large registry 88 718 116 70 34 Very large registry 91 1066 71 85 62 Total 240 2220 271 196

proportion of first-degree and second-degree matings. Table 6 The observed risk behaviours are reflected in the higher ob- shows the percentage of kennels in each of the four breeding served mean inbreeding coefficients for breeds in this group. In behaviour categories, categorised according to the number of prog- ‘small registry’ breeds, popular sire usage is prevalent, but this is eny registered, i.e. ‘major’ (>50) or ‘minor’ (650), and the total most likely to be an analysis artefact related to the breed mean lit- number of registrations in a breed, from ‘small registries’ (<2000) ter size exceeding 5% of registrations over 5 years for certain low- to ‘very large registries’ (P80,000). registration number breeds. This artefact in turn highlights that, in Lower risk breeding strategies were followed by 91% of 64,748 rare breeds, FCI guidelines on sire use need to be revisited, since a kennels (column ‘pf’). Of the remaining 9%, 6% made use of popular single litter may define a sire as ‘popular’. sires for breeding, 2% employed close breeding practices and <1% Major kennels (those producing at least 50 offspring) represent used both close breeding and popular sires. The overall 6% of kennel prefixes in the data. Surprisingly, if we observe the observations of kennel behaviour are heavily influenced by analy- behaviour of kennels on a proportionate basis according to the sis of the ‘large registry’ and ‘very large registry’ groups because magnitude of their contribution of progeny to the breeds over all these two groups comprise the vast majority (>89%) of kennels in breed size groups, major and minor kennels do not behave very dif- the data. ‘Small registry’ and ‘medium registry’ breeds (total regis- ferently. At the same time, within ‘small registry’ breeds, major trations <25,000 dogs) demonstrated a greater tendency to make kennels made exclusive use of popular sires, whereas 22% of minor use of both popular sires and close breeding than kennels in other kennels in ‘small registry’ breeds used neither inbreeding nor pop- breed size groups. In ‘medium registry’ breeds, 22% of kennels used ular sires. In ‘medium registry’ breeds, >4% of major kennels used popular sires and >10% used close breeding. both popular sires and close breeding. This identifies major kennels

Table 6 Risk behaviour by kennel type and breed size.

Kennel type Breed size Proportion of kennels Number of kennels PF Pf pF pf Major kennel (6% of kennels) Small registry 0.000 1.000 0.000 0.000 43 Medium registry 0.047 0.284 0.100 0.568 408 Large registry 0.004 0.043 0.020 0.933 1547 Very large registry 0.000 0.019 0.001 0.980 2275 All Groups 0.012 0.055 0.014 0.919 4273 Minor kennel (94% of kennels) Small registry 0.049 0.722 0.007 0.223 952 Medium registry 0.015 0.219 0.106 0.660 5602 Large registry 0.001 0.040 0.038 0.921 20,178 Very large registry 0.000 0.030 0.008 0.962 33,743 All Groups 0.002 0.062 0.024 0.912 60,475 All kennel types Small registry 0.046 0.738 0.007 0.209 995 Medium registry 0.018 0.224 0.105 0.652 6010 Large registry 0.001 0.041 0.036 0.922 21,725 Very large registry 0.000 0.029 0.008 0.963 36,018 All 0.003 0.061 0.024 0.913 64,748

P, P50% progeny in kennel from popular sires (‘popular sire’ kennels). p, <50% progeny in kennel from popular sires. F, P50% progeny in kennel have F > 0.10 (‘close’-breeding kennels). f, <50% progeny in kennel have F > 0.10. Major kennel: Kennel responsible for >50 progeny in ANKC database for this breed. Minor kennel: Kennel responsible for <50 progeny in ANKC database for its breed.

Please cite this article in press as: Shariflou, M.R., et al. A genealogical survey of Australian registered dog breeds. The Veterinary Journal (2011), doi:10.1016/j.tvjl.2011.06.020 M.R. Shariflou et al. / The Veterinary Journal xxx (2011) xxx–xxx 7 in ‘medium registry’ breeds as good targets for programmes offer- 91 had contributed equally. By contrast, the Skye terrier began ing breeder education. Fitting a restricted maximum-likelihood with fewer founders (126), but their contribution was more evenly model accounting for the fixed effects of breed size group, kennel spread (fe = 62). type, popular sire usage and use of close breeding, all effects and Observed values for breed mean inbreeding coefficients are con- their interactions were highly significant. siderably lower than those reported by Maki (2010), who analysed international pedigrees of rare breeds, but are in line with those reported by others based on national data (Calboli et al., 2008; Leroy Discussion et al., 2006). For example, it is likely that the observed disagreement with the Maki analysis for the Nova Scotia duck tolling retriever An alternative basis for defining breed-size groups would have breed (Australian F = 0.03 vs. International F = 0.26) is related to been the numbers registered in the reference population: only four the treatment of imported individuals in our data and, as part of this, breeds change groups under this definition (Pekingese, Poodle, the lack of pedigree information from ancestors of the imported Rottweiler and Cavalier King Charles spaniel), and the conclusions dogs. The correspondence of our breed mean results with other na- reached are, in essence, the same. Completeness of pedigree data is tional data collections suggests that the majority of the inbreeding reported by the number of generation equivalents (EqG). Incom- identified by Maki (2010) for this breed is probably derived from plete pedigree information can result in underestimated inbreed- distant rather than recent co-ancestry. Recent inbreeding is ex- ing levels (Boichard et al., 1997). Observed Australian values of pected to be more deleterious to population health (Hinrichs EqG are within the ranges reported in other national data sets (Ler- et al., 2007) and so inbreeding coefficients calculated over more re- oy et al., 2006, 2009). The most complete pedigree data in our cent population history (such as are presented in this analysis) may study was observed for the Australian terrier, with an average have more relevance to breed health than total inbreeding. EqG of 10.1. From eight breeds with EqG values <4 (Table 2), three The ‘small registry’ breeds with very low inbreeding coefficients breeds (Central Asian shepherd dog, Italian Spinone and Polish are likely to be affected by the importation of breeding dogs. If the lowland sheepdog) had pedigree depths <25 years. Although we current Australian populations of these breeds were closed to have provided population parameters for these breeds, the data importation, then the low inbreeding coefficients would be ex- should be interpreted and used cautiously. Low values of EqG ob- pected to rise rapidly. In our data, there appears to be a strong served in at least some breeds may be due to recent breed estab- trend for inbreeding coefficients to rise during the establishment lishment. For example, the Central Asian shepherd dog of breeds. Encouragingly, for many established breeds, it appears (EqG = 2.1) was first registered in Australia in 2000. Fluctuations that the rate of inbreeding over the past 10 years is stabilising or in registrations by year of birth are profound in ‘small registry’ declining, due to either better breeding management or increased breeds (<2000 registrations in total). In some breeds, there were use of importations. years with no recorded registrations. Such fluctuations influence The mean percentage of first-degree matings (2.6%) is of inter- the reliability of results. est in the context of this type of mating having been banned in

Effective population size (Ne) predicts the equivalent number of 2009 by the Kennel Club in the UK, with consequent pressure for animals equally participating in the breeding cohort and so ac- a similar ban in Australia. While the well-known undesirable con- counts for differences in the usage of sires and dams. Large values sequences of such matings are sufficient to justify a similar ban in for Ne predict greater genetic diversity. Founders are defined as Australia, the relatively low incidence of first-degree matings indi- animals with unknown parents that contribute directly to the gene cates that, in general, Australian breeders have been avoiding such pool of the reference population. Effective population size predicts matings. Interestingly, the average percentage of dogs with F < 10% genetic variation in future generations, while effective number of in ‘very large registry’ Australian breeds is 88%, which is exactly the founders indicates the past genetic variation that has contributed same as reported for 10 of the most populous UK breeds by Calboli to the breed. If the founders contribute equally to the gene pool, et al. (2008). Even for the ‘small registry’ Australian breeds, this the effective number of founders (fe) is equal to the actual number percentage is 80%. of the founders (Lacy, 1989). As the discrepancy in contribution be- The increase in progeny per sire with breed registry size sug- tween founders increases (i.e. some dogs contribute much more to gests that Australian breeders self-regulate with respect to the future generations than others), fe decreases and fe/f is reduced usage of sires and that they appear to adhere informally to the from its maximum value of one. The greater the effective popula- FCI international breeding strategy, which states that sires should tion size and the effective number of founders, the lower the not be bred to produce >5% of the progeny registered in a breed chance of inbreeding in random mating populations. over the preceding 5 years (FCI, 2011). It should be recognised that, The estimated effective population sizes and effective founder in rare breeds, one litter may exceed the FCI recommendation and, numbers in our data fall within the ranges reported by other breed by this criterion, it is difficult to adhere to recommendations on the analyses for 61 French breeds (29–2136) (Leroy et al., 2006, 2009) use of sires in ‘small registry’ breeds. Most analyses for popular and Bavarian hounds (Voges and Distl, 2009). The range of effective sires report numbers of registered progeny but it is perhaps more population sizes also compares favourably with the estimates of 45 important to assess the relative contribution of their genes to the for Australian Thoroughbred horses from 1940 to 1970 and 260 future breeding cohort. For a sire to have 300 progeny is significant from 1970 to 2000, reflecting increased use of imported sires in the current population, but the legacy of this sire in the breed as (Castle, 2007; K. Castle, personal communication) and 100 for a whole will only be realised if his progeny go onto have progeny of Australian Holstein–Friesian cattle from 2002 to 2008 (calculated their own. from Haile-Mariam et al., 2007). The median of effective population sizes across all 32 breeds analysed in this paper is 101, which compares favourably with the medians of 105 for 19 UK cattle Conclusions breeds and 90 for 17 UK sheep breeds in 2007 (Villanueva et al., 2010). Analysis of full pedigree data for 32 Australian dog breeds is in

The ratio fe/f for the 24 breeds with EqG > 4 ranged from 0.05 broad agreement with other national analyses of pedigree dog (German shepherd dog) to 0.49 (Skye terrier). This result suggests structure, showing that the mean inbreeding coefﬁcient is typically that, although the German shepherd dog breed has 1893 founders, <10% for individual breeds. Effective population sizes are often their contribution to genetic diversity in the breed is the same as if lower than ideal and uneven genetic contributions by founders

Please cite this article in press as: Shariflou, M.R., et al. A genealogical survey of Australian registered dog breeds. The Veterinary Journal (2011), doi:10.1016/j.tvjl.2011.06.020 8 M.R. Shariflou et al. / The Veterinary Journal xxx (2011) xxx–xxx are frequently demonstrated. Analysis of kennel behaviour indi- Calboli, F.C.F., Sampson, J., Fretwell, N., Balding, D.J., 2008. Population structure and cates that 91% of all kennels in the data are following relatively inbreeding from pedigree analysis of purebred dogs. Genetics 179, 593–601. Castle, K., 2007. An update on inbreeding in the Australian Thoroughbred horse low-risk breeding practices by avoiding excess popular sire use population. In: Genetic Improvement: Making it Happen. Proceedings of the and close breeding. Breeds with registrations of <25,000 dogs are 17th Conference of the Association for the Advancement of Animal Breeding at risk of increased inbreeding and popular sire use compared with and Genetics, Armidale, New South Wales, Australia, 23–26 September, 2007, pp. 525–528. more common breeds. Within ‘medium registry’ breeds, major Falconer, D.S., Mackay, T.F.C., 1996. Introduction to Quantitative Genetics, Fourth kennels (contributing >50 progeny to the breed) are more likely Ed. Longman Scientific and Technical, Harlow, UK, 480 pp. to use close breeding and may represent a group that would ben- Fédération Cynologique Internationale, 2011. FCI International Breeding Strategies. www.fci.be/uploaded_files/29-2010-annex-en.pdf (accessed 10 June 2011). efit from access to improved education in animal breeding and Haile-Mariam, M., Bowman, P.J., Goddard, M.E., 2007. A practical approach for genetics. Popular sire use by FCI (2011) definition is nearly impos- minimising inbreeding and maximising genetic gain in dairy cattle. Genetics, sible to avoid for ‘small registry’ breeds (with fewer than 2000 reg- Selection, Evolution 39, 369–389. Hinrichs, D., Meuwissen, T.H., Odegard, J., Holt, M., Vangen, O., Woolliams, J.A., istered dogs). In general, rates of inbreeding appear to be in decline 2007. Analysis of inbreeding depression in the first litter size of mice in a long- over recent years; perhaps this is attributable to an increasing term selection experiment with respect to the age of the inbreeding. Heredity awareness of genetic principles and improved access to genealog- 99, 81–88. ical resources, such as web-based dog pedigree browsers. Lacy, R.C., 1989. Analysis of founder representation in pedigrees: Founder equivalents and founder genome equivalents. Zoo Biology 8, 111–123. Leroy, G., Rognon, X., Varlet, A., Joffrin, C., Verrier, E., 2006. Genetic variability in Conflict of interest statement French dog breeds assessed by pedigree data. Journal of Animal Breeding and Genetics 123, 1–9. Leroy, G., Verrier, E., Meriaux, J.C., Rognon, X., 2009. Genetic diversity of dog breeds: None of the authors of this paper has a financial or personal Within-breed diversity comparing genealogical and molecular data. Animal relationship with other people or organisations that could inappro- Genetics 40, 323–332. Maki, K., 2010. Population structure and genetic diversity of worldwide Nova Scotia priately influence or bias the content of the paper. duck tolling retriever and Lancashire heeler dog populations. Journal of Animal Breeding and Genetics 127, 318–326. Acknowledgement Meuwissen, T., Luo, Z., 1992. Computing inbreeding coefficients in large populations. Genetics Selection Evolution 24, 305–313. Payne, R.W., Harding, S.A., Murray, D.A., Soutar, D.M., Baird, D.B., Welham, S.J., Kane, We gratefully acknowledge the Australian National Kennel A.F., Gilmour, A.R., Thompson, R., Webster, R., Tunnicliffe Wilson, G., 2008. Council for providing full pedigree data. GenStat Release 11 Reference Manual. VSN International, Hemel Hempstead, UK. Sargolzaei, M., Iwaisaki, H., Colleau, J., 2006. CFC: A tool for monitoring genetic References diversity. In: Proceedings of the 8th World Congress on Genetics Applied to Livestock Production. Paper 27-28. Australian National Kennel Council, 2011, http://www.ankc.org.au/ National- Villanueva, B., Sawalha, R.M., Roughsedge, T., Rius-Vilarrasa, E., Woolliams, J.A., Registration-Statistics.aspx (accessed 10 June 2011). 2010. Development of a genetic indicator of biodiversity for farm animals. Boichard, D., 2002. Pedig: A Fortran package for pedigree analysis suited to large Livestock Science 129, 200–207. populations. In: Proceedings of the 7th World Congress on Genetics Applied to Voges, S., Distl, O., 2009. Inbreeding trends and pedigree analysis of Bavarian Livestock Production. Paper 28-13. mountain hounds, Hanoverian hounds and Tyrolean hounds. Journal of Animal Boichard, D., Maignel, L., Verrier, E., 1997. The value of using probabilities of gene Breeding and Genetics 126, 357–365. origin to measure genetic variability in a population. Genetics, Selection, Wright, S., 1923. Mendelian analysis of the pure breeds of livestock – The Evolution 29, 5–23. measurement of inbreeding and relationship. Journal of Heredity 14, 339–348.

Please cite this article in press as: Shariﬂou, M.R., et al. A genealogical survey of Australian registered dog breeds. The Veterinary Journal (2011), doi:10.1016/j.tvjl.2011.06.020