Merging pedigree databases to describe and compare mating practices and gene flow between pedigree dogs in France, Sweden and the UK

Methods

After merging the pedigree databases for each of the four breeds, dogs registered in France, Sweden and the UK, born between 1980 and 2014 with both parents known, were defined as the reference population. The reference population was divided into cohorts based on the country of registration (the initial country to assign the dog ID) and year of birth period (intervals of 5 years, from the period 1980–84 to the period 2010–14). To simplify the discussion of the reference populations belonging to different registration countries, we have shortened the label hereafter; for instance, we used ‘French population’ to refer to the reference population of dogs first registered in France.

The effective number of sires (NeffS) indicates the number of sires used that would be equivalent to the actual population in terms of the loss of diversity due to unbalanced use of sires (James 1972) and was computed as follows:

where is the relative frequency of use of individual sire i among all sires in a specific reference population. With the exception of NeffS, the other genealogical parameters were computed using PEDIG, a FORTRAN package (Boichard 2002). Coefficients of inbreeding (F) were computed according to the method of Meuwissen & Luo (1992), and coefficients of coancestry (C) were computed following the method of Reynolds et al. (1983). Equivalent complete generations (EqG) were computed as the sum over all known ancestors of the term (1/2^t), where t is the ancestor's generation number (Boichard et al. 1997). For a given breed, the effective population size (Ne) was estimated by averaging the increase in coancestry for all pairs of individuals in a reference population () (Cervantes et al. 2011):

where is the inbreeding of a descendent from individuals j and k, and and are the discrete equivalent generations for the individuals j and k. Ne was estimated by averaging over each generation, using the following formula:

The proportion of inbred individuals was computed as the percentage of individuals with an F higher than 0 when taking into account only two generations of a pedigree according to the method of VanRaden (1992). On the basis of the breed origin of each founder (ancestors without parents known), the founder contributions of different origins were calculated (Boichard et al. 1997). All of the calculated parameters are shown for each cohort of the reference population.

Demographic parameters

The four breeds exhibited different reference population sizes over time in the three registration countries (Table 1). France showed the largest reference populations for ESE (157 186) and BMD (59 848). The UK showed the largest reference populations for BMA (44 377) and LBR (1 037 174). For each breed, the smallest reference population was found in Sweden. The sizes of the French populations increased over time for BMA, ESE and BMD, while decreased beginning in 1995–99 for LBR. The sizes of the Swedish populations started to decrease after 1990–94 for BMA, ESE and BMD, while fluctuated for LBR. The sizes of the UK populations first increased and then decreased for all four breeds.

Both the actual number of sires and NeffS followed the evolution of the size of the reference population for the four breeds in each country (Table S1). The evolution of the ratio of effective number of sires to actual number of sires (NeffS/Number of sires) is presented in Figure S1, where a stable or slightly increasing tendency of more balanced use of breeding animals over time was demonstrated in both the breeds and registration countries.

Genealogical parameters

The amount of pedigree knowledge increased over the period, regardless of the breed and the country of registration (Figure 1). As shown by EqG, pedigree knowledge increased after merging the databases (Table S2). For instance, before merging, the EqG values for BMA in France, Sweden and the UK were 4.15, 3.94 and 8.60, respectively. After merging, the EqG values for BMA in France, Sweden and the UK increased to 4.39, 5.92 and 8.70.

Depending on the period, breed and country, the average F and average C values varied between 0.01 and 0.16 but were below 0.1 in the majority of periods (Figure 2). In general, both the average F and average C values were higher for the UK population than for the French and Swedish populations for all four breeds. In particular, the average C value was below the average F value for the UK population, while these parameters were at the same level in the French and Swedish populations. Additionally, both the average F and average C values slightly decreased or remained stable in all breeds and all countries with the exception of an increase for ESE in the UK and LBR in France and the UK.

The effective population size based on the coefficient of coancestry (Ne) varied from 26 to 425 (Figure 3). Ne decreased for ESE in the UK and LBR in France and the UK over the entire study period, while it increased for the other populations. Throughout the investigated period and for all of the breeds, when we considered the populations from the three registration countries as a single population, Ne was always lower than the sum of Ne for the populations in the three registration countries, indicating a detectable amount of gene flow across countries. However, the global Ne value was always higher than the highest country-specific Ne (considering the three breeds independently), except for LBR. In LBR, global Ne was equivalent to the UK value, probably because the UK subpopulation was demographically much more important than the two others.

Considering two generations, the proportion of inbred individuals ranged from 0.00 to 0.12 (Figure 4). In comparison between the three countries, the proportion of inbred individuals was often highest in the UK populations and lowest in the Swedish populations. Moreover, the proportion of inbred individuals fluctuated for BMA and was stable for other breeds in the French populations, while it tended to decrease in the UK and Swedish populations. Overall, the proportions of inbred individuals, considering the three countries, decreased during the period.

Founder contribution

In Sweden, the origins of populations became more diverse and the founder contributions (from the various origins) became more equal (Figure 5). In France, the population became increasingly diverse over time for BMA; the influence of the dominating origins of founders in 1980–84 reduced considerably for ESE and BMD, and the French LBR population in 1980–84 seems to be substituted through a large influx from the UK. For the native British breeds BMA, ESE and LBR, the founder contributions in the UK populations were dominated by the UK founders, while for native Swiss breed BMD, the founder contributions were dominated by Swiss founders (average 0.49). From 2000–04 onwards, the UK populations showed greater founder contributions from foreign countries than in previous periods. Considering populations from the three countries together, for BMA and LBR, the global founder contributions were still dominated by the UK founders, largely due to the demographic weight of the UK populations. For ESE and BMD, the founder contributions of France and other European countries shifted, but the contribution of the main founder remained fairly stable.

Merging pedigrees

Merging pedigree databases across countries is complex and time-consuming work that is not confined to dog populations but is also carried out for cattle (Fikse 2004) and horse populations (Viklund et al. 2015). As a prerequisite to explore within-breed genetic variability through international collaborations, an operational method to merge different pedigree databases was needed. We have developed a solution to merge databases from different countries. We observed a substantial improvement in pedigree knowledge, especially for the Swedish population (Table S2).

Two points are worth highlighting based on our experience in handling the pedigree records from four breeds across these three countries. First, the success of detecting duplicates within and between countries relies strongly on the discovery of potential duplicate records (for the same individuals) based on attributes (e.g. name of individual, birthdate of individual) recorded in the databases. However, these attributes may be recorded with missing information, or they may not be exactly the same due to typographical errors, or they may be inconsistently formatted during registration. Second, if a standard format for international dog IDs was proposed and adopted for the registration of pedigree dogs across countries, similar to the Universal Equine Life Number (UELN) system used in horses, the merging of pedigrees would become more efficient and precise.

Comparing genealogical parameters

The results of our study presented herein are similar to other published results for the examined breeds (Leroy et al. 2009; Jansson & Laikre 2014; Lewis et al. 2015). In the French populations, the EqG values for ESE, BMD and LBR born from 2000 to 04 were 6.3, 4.8 and 6.9, respectively, in the present study, similar to results of Leroy et al. (2009), who reported EqG values for ESE, BMD and LBR born from 2001 to 05 of 6.4, 5.0 and 5.5. The average F and average C values for French dogs born from 2000 to 04 differed by 0.012 at most compared with the results for the birth year period of 2001–05 reported by Leroy et al. (2009). For the Swedish populations, the average F values for BMA born from 1990 to 94 and 2005 to 09 were 0.06 and 0.03, respectively, while those for BMD were 0.05 and 0.04 in the present study. These results are similar to those of Jansson & Laikre (2014) for the Swedish population, where the average F values for BMA alive at the end of 1995 and 2010 were 0.06 and 0.03, respectively, while for BMD, they were 0.05 and 0.02. For the UK LBR population, the trends in the average F and average C values observed in our study were the same as those reported by Lewis et al. (2015).

Evolution of breeding practices and gene flow

Mating between close relatives and popular sire effect are considered to be the major causes of overexpression and dissemination of inherited disorders (Leroy & Baumung 2011). When considering mating between close relatives, evaluation of the proportion of inbred individuals, considering two generations of pedigrees in different subpopulations (dogs from the same breed, but registered in different countries), led to interesting results.

In France, a sudden increase in the level of inbreeding occurred in 1995–99 for BMA. This was likely a result of the highest proportion of inbred individuals occurring in 1990–94 (Figure 4). In Sweden, the level of inbreeding and the proportion of inbred individuals has decreased since 1980–85 for BMA, ESE and BMD, while it has been stable or low for LBR (less than 0.02). Moreover, the genetic variability of LBR was not improved according to the level of inbreeding, which may be related to the stability of proportion of inbred individuals in LBR (Figure 2). In the UK, proportions of inbred individuals were the highest among the three countries. Explaining the difference we found between the subpopulations studied is difficult; however, we can link the decrease of this practice in Sweden with the implementation of recommendations aiming at avoiding mating between close relatives due to its potential impact on dog health. Additionally, the average F value was higher than the average C value for the four breeds in the UK, while this was not the case in France and Sweden. In particular, the level of inbreeding and the proportion of inbred individuals has decreased in the UK population from since the 2010s, indicating that breeding practices may have changed. This finding is similar to results published for both vulnerable native breeds and the 20 most common breeds in the UK showing that the mean rate of inbreeding per generation has significantly decreased from 2000 onwards (Lewis et al. 2015). The decreasing proportion of mating between close relatives in Sweden may reflect a more important concern of Swedish citizens towards animal welfare, as illustrated by different surveys (European Commission 2016).

When considering the popular sire effect, it is difficult to clearly determine to what extent the practice has changed over the time period we studied. The actual number of sires and NeffS was mainly related to the evolution of subpopulation size (Table 1). Similarly, the NeffS/Number of Sires ratio was observed to be higher in large subpopulations, indicating less intensive use of sires in large populations. On the other hand, the ratio showed either a stable or increasing trend, even for subpopulations with an increasing population size, which could indicate a tendency towards a more balanced use of sires.

The results of the present study clearly showed that gene flow across countries has become more frequent in all of the countries and all of the breeds we studied. The Swedish populations exhibited the most diverse origins of founder contributions among the three countries based on the four breeds we studied. The UK populations have received a greater founder contribution from overseas, even among native breeds (BMA, ESE and LBR) since 2000–04, which may be related to the relaxation of the UK quarantine law for the importation of dogs (Lewis et al. 2015). Additionally, when gene flow was compared between any two of the countries, Sweden and the UK were found to exhibit the most frequent exchange of breeding animals.

Management of genetic variability

Overall, Ne has increased over the last several years in all of our subpopulations, except for the ESE subpopulation in the UK (which is related to a large decrease in the number of births). This could be related to favourable demographic trends (especially in France), but it could also be related to diversification of origins (through increased gene flow), and possibly to better management of the popular sire effect, although this should be confirmed through more extensive analysis.

The choice of adequate strategies for the management of genetic variability must be made on a breed by breed case in pedigree dogs because each breed faces a unique situation of inbreeding (especially influenced by the evolution of population size). For breeds with a small or decreasing population size, cross-breeding (or introgression) can be an interesting solution to reduce inbreeding or ensure genetic improvement (Pirault et al. 2013; Leroy et al. 2014). However, cross-breeding is rarely practised in pedigree dogs, which makes the management of genetic variability within a breed more challenging. Improvement of within subpopulation genetic variability could be achieved by exchanging breeding animals between countries. This is illustrated by the observation that, in the breeds with a small or medium population size that we studied (BMA, ESE and BMD), Ne was always higher when considering the dogs registered in all three countries together than when considering only a single country. This result illustrates the value of genetic exchanges between subpopulations, as is theoretically well established (Wright 1931). Conversely, for breeds with a large population size, such as LBR, Ne was not higher when considering the dogs registered in all three countries together than in the French population alone and was only a slightly higher than in the UK population. It appears that the effect of gene flow was ‘diluted’ within this breed with a large actual population size. At a more global level, an equal use of sires, but restricted number of litters, could be a solution to improve the effective population size (Windig & Oldenbroek 2015).

Considering the registration of dogs without pedigree but with appropriate phenotype (or even genotype), could constitute another interesting option to increase the breeding population, and it opens up for the possibility to register dogs with different genetic background, which could be assessed through adequate genetic testing. As an example, registries in France are open for most breeds (only 15 breeds have their pedigree closed), meaning that in theory, a dog without pedigree can be registered if it is judged to have the adequate phenotype. However, in practice, those registrations represented in 2014 were only 0.4% of the total number of registration for the corresponding breeds. Therefore, this approach could be efficient only if breed clubs were much less strict in the acceptance of dogs without pedigree. Optimal contribution selection is another solution for managing genetic variability that is potentially more efficient but also results in greater constraints for breeders. Future research could focus on how to fairly compare the genealogical parameters of potential breeding animals across countries, and monitor the consequence on genetic variability and management of inherited diseases, simple or more complex (e.g. hip dysplasia), at an international scale.