Evidence of inbreeding depression
– a prospective study in the population of the European Dobermann –
by Professor Sonia Garzia (2011)
Modern purebred dogs are plagued by many kinds of heritable diseases. The high prevalence of diseases and defects in pure dog breeds are due to high selection intensities in genetically closed populations (Ubbink et al. 1998,1). Inbreeding has been used extensively as a mating system in dogs during the time the breeds were developed to be of uniform appearance. Even after that, popular sires have been widely used in breeding, which has decreased the genetic diversity in the breeds. Inbreeding leads to a relative loss in heterozygous loci, which causes a decline in the mean performance, genetic phenomenon known as inbreeding depression (Falconer and Mackay, 1996).
This practice of inbreeding on the popular dogs, preferably males (top sires), has been maintained over time with varying intensity during the hundred year history of the Dobermann breed. Often we speculate about the possible existence of the inbreeding depression in the Dobermann breed but no data are available.
1 – What is inbreeding depression?
Inbreeding depression is reduced fitness in a given population as a result of breeding of related individuals. It is often the result of a population bottleneck and allows the expression of deleterious recessive alleles from both parents, rather than a few genes of large affect, or caused by a few key genes that affect the expression of many other genes (epi static and pleiotropic effects on gene expression); and is often caused by cross breeding between close relatives. Inbreeding affects mostly fitness traits and traits of monogenic qualitative inheritance but it has an effect also on quantitative traits (Mäki et al. 2001).
Fitness is a central concept in evolutionary biology, and is defined as the relative ability of an individual to leave descendants and hence transmit its genes to future generations. This has two components. The ultimate component, sometimes called fecundity, is the production of viable offspring which, in turn, go on to produce their own offspring. However, before an individual can produce offspring it must survive to sexual maturity. Survival is the other component of fitness. Throughout an organism’s life, the probability of dying may differ from one life-history stage to the next.
Inbreeding depression includes a wide variety of physics and health defects, among which we find as most common: high incidence of genetic diseases, reduction in female fertility and sperm viability, high prenatal and newborn mortality and immune system deficiencies.
Although scientists have known for more than a century that small populations of closely related animals are likely to suffer from low reproductive success, the exact mechanism by which this “inbreeding depression” occurs is still the subject of debate. Usually the heterozygote is fitter than the two homozygotes in a normal relation of dominance. This can give rise to an increase of the hybrid vigour obtained by crossing parents differing in a single specified pair of allelic genes. However, in a part of genes expressed differentially when inbreeding depression is showed, heterozygote are expressed more markedly in the phenotype than in either homozygotes. In this case we speak of over-dominance in the genetic expression of these genes.
A recent paper by (Ayroles et al. 2009) applying a genomics scan with oligonucleotide microarrays in fruit flies – Drosophila melanogaster – shows that genes associated with inbreeding depression, could be grouped into three broad categories of function: those involved in metabolism, environmental stress (oxidative stress, temperature, etc), and defence against diseases and external agents. Results indicated that approximately 75% of all genes involved in inbreeding depression were additive, partially additive, or dominant, but about 25% of all genes expressed patterns of over-dominance.
2 – How to detect the possible inbreeding depression?
Only a few studies have investigated the relationship between litter size (O/L) and the body weight (BW). In a study based on 76 breeds provided a correlation estimate of 0.71 between dam’s BW and O/L and predicted an average increase in O/L of 0.09 pups/kg (Robinson, 1973). Data from Swiss dog breeding organizations on 17,106 pups from 2875 litters of nine breeds used in a breed comparison showed a clear positive relationship between shoulder height (SH) and average and maximal O/L. The correlation between SH and BW was 0.91, which allows SH to be used as a good predictor for BW (Kaiser, 1971).
More recently Gill (2001) has related a strong correlation between litter size and average birth weight for different breeds and groups (table 1 and figure 1)
Fetal losses and still borns may be caused by genetic traits and are reflected by lack of pregnancy or small litter sizes. The actual litter size for certain breeds: English Bulldog, Pomeranian, Cavalier King Charles and Newfoundland, it is smaller than predicted based on body weight, indicating a genetic basis for reduced litter size (Giger, 2006). Most fading puppies and puppy losses are seen during the first week of life, soon after the maternal homeostatic system no longer can compensate of any serious endogenous defect. Accordingly a simple method to detect inbreeding depression in a pure breed as Dobermann, is to follow the annual evolution of fertility through the time computing the phenotypic average of offsprings/litter (O/L score) in the population over time.
3 – Sample strategies, sizes and reliability
Materials have been obtained through gradually introducing the data from the most comprehensive source available: the “Dobermannpedigrees.nl” database.
The sample for the “population study” (2,307 litters) is selected by the technique of annual random stratification with proportional affixation to registered dogs by year in the data base.
Another separated sample for the “top sire progenies” calculations (1,777 litters) is selected by the technique of clusters between the sires with more than 300 offsprings registered in the data base. Obviously the available records do not include the early deaths (neonatal and perinatal) and litter sizes are all subject to this bias. Sample size overcomes the 95% level of significance and 3% of precision required for a good statistical reliability of the conclusions (critical value of reliability for 1970-2009 population with a variance measured of 5.67 is 20,934 individuals 2,974 litters). Litters with only one puppy (very frequents) are excluded from the samples in both cases. Operating in this way, decrease the incidence of lower tail frequency in the mean value.
Sample for “average life study” include 4120 data of dogs born between 1980 and 1997 years. For the purposes of calculating the average age in years for each record, the months of birth has not been taking into account, but only the year of birth and death.
All samples show a normal distribution for the mainly variables: litter size and average life through time.
For the “generation calculus” a simple method has been applied: rolled mean for eight years fixed period. A more precise calculus of “generation number”, but out of the scope of this study and less comprehensible to the no expert reader, can be made considering available methods for population genetics studies, particularity proposed by (Brinks et al. 1962), best adapted to interbreed comparisons. Anyway, in our case the same historical trend has been found, with rare exceptions, that is described by (Calboli et al. 2008) for different pure breeds of dogs applying this method: dams significantly younger than sires at the time of mating.
4 – Evolutions impact on the fertility of the breed over time?
Graphic (figure 2) bellow shows the evolution of the evolution of the phenotipic O/L score in the period 1970-2009. The curve shows the annual generational (eight year’s fixed period) average of O/L in the population. This mean for example, that O/L value shows for the generation year 2005 corresponds to the average O/L of all individuals in the sample born between 2001 and 2009, both inclusive. Tree clustering statistical analysis of the data shows four nodes corresponding to four periods of time: before 1984, 1984-2000, 2000-2004, and above 2004. The beginning of the inbreeding depression, whelped by the second derivative of the distribution, is placed in the 1985 year. Significant difference is confirmed by a Z test of all files before and after this year 1985 [P(Z<0.001)].
5 – O/L as detector of inbreeding depression
According with a Royal Canin information, relative to Average Weight at Birth (AWB) for different breeds, AWB for the Dobermann varies between 394g and 423 g. This made a mean of 425 g that interpolate in the correlation of Gillis (2001) for all breeds (figure 3) gives the value O/L=6.94 for the Dobermann breed, higher than the mean value for the actual population O/L=6.59 which confirms the genetic origin of this decrease.
Another possible cause of variation in the litter size, is the selection of conformational traits, but correlation of Kaiser (1971) and Robinson (1973) between BW, SH and O/L. Modern European Dobermann selection tends to a more robust (molosoid type) and higher shoulder height than old type (70-80’s), before the endogamic depression. In consequence of this cause an increase of litter size would be expected, just in the contrary sense of the experimental O/L data evolution in this study.
6 – Is the inbreeding depression associated with the breeding practices?
Figure 4 shows the evolution of the inbreeding coefficient F (or COI) in the population over time. Curve is made in the same way: annual generational average of F in the population during an eight years period.
The maximum in the average F curve of the population over time, shows an eloquent coincidence between the maximum of the F distribution and the beginning of inbreeding depression. It is interesting to note that from this time gradually moderates inbreeding practices which is probably associated with the perception by the breeders of the genetic problems associated with inbreeding depression.
Curve also shows the average F for progeny of some influential top sires, characterized for a high inbreed progeny, particularly for three of the most influential ancestors in the current population: Hertog Alpha v. Le Dobry, Graaf Quirinus v. Neerlands Stam and Jivago v.h. Wantij, progenies which exceed the inbreeding coefficient 14%.
Figure 5 shows the evolution of the population (number of individuals) over time. The beginning of the inbreeding depression coincides with a period in which the Dobermann greatly increases its popularity as can be seen in the incidence curve (second derivative).
Normally experiment of mating between relatives to show the impact of inbreeding depression are designed on experimental animals (basically fruit flies and mice) with many offsprings with high inbreeding coefficient: 25% and more (high inbreed classes). History of pure breed dogs is qualitatively different: more reduced inbreeding rates but maintained over time, generation after generation for a long period of time.
8 – Is inbreeding depression related to average life span (ALS)?
High incidence of genetic diseases and immune system deficiencies associated with inbreeding depression, resulting in a decrease in the average life of the population.
Average life curve (figure 6) is obtained by the same method than others for comparison purposes: rolling mean of eight years period for annual average life (ALS) in the population. For example, ALS value shown for the year 2000 corresponds to the average life ALS of all individuals of the sample born between 1996 and 2004, both inclusive. We can observe a change in trend around the year 1986 (calculated 1986.4), which at the same time begin to show signs of inbreeding depression. From this moment, average life shows a worrying and sustained tendency to decrease in successive generations. No correlation (statistical significant) exists between the average inbreeding coefficient F and the average life ALS.
Following, a calibration of the mean values has been created, comparing obtained data with the only scientific data available (Mandiguers et al., 2006). Study of this author does not evaluate the average life time for a sample of 340 Dobermann born between 1993-1999, but the survival curve from Kaplan–Meier estimator does. However, we can obtain a value for mean ALS from this curve, and at about 9 years. Data obtained here for the same period (mean of 752 data) is 8.8 +/- 2.6 years, which is in good agreement with that obtained by these authors.
For the actual population the calculated average from the data in an eight years period for dogs born between 1992 and 2000 (2395 data). The estimated mean value for actual average life span is 7.4 +/- 2.3 years, below the value for 1993-1999, which indicate a downward trend.
The historical value for the breed has been obtained by the average of data (326 data) in the 1980-1988 periods. This value is 9.0 +/-2.7 years. Difference of (-1.6 years) is significant at a statistical level of 95% [P(Z)<0.000001]. Kaplan-Meier survival curves for the actual population (dogs born 1992-2000) and for the previous period (1986-1992) (table 2 & figure 7). It is clear, that from an early age, the survival factor is lower for the most recent period.
We can observe (Figure 7) that from an early age the survival factor is lower for the most recent period. In this period the average life expectancy is reduced 1.6 years, and the percent of long lived class dogs (life expectancy >10 years) is reduced from 24.5% (1986-1992) to 7.85 % (1992-2000).
9 – Top sires litter size O/L score
Figure 8 shows the O/L score for the progeny of the most influent top sires in the population of the European Dobermann in the period 1970-2005 – other results for previous top sires not showed – compared with the evolution of O/L in the population.
Values of studs with significant differences of O/L score respect population are highlighted on the graph. Fertility score of studs depends of sire, but also depends on the status of the population through the participation of the dam.
Low values statistically significant at a level of 95% [P(Z)<0.001] are relevant for three influential dogs of the current population: Jivago v.h. Wantij, Prinz v. Norden Stamm, and Baron Nike Renewal. Jivago is a son of Graaf Quirinus v. Neerlands Stam and grandson of Hertog Alpha v. Le Dobry, two dogs with normal O/L score. Otherwise Prinz v. Norden Stamm, and Baron Nike Renewal are related by kinship – father and son -, but Baron is the father of a dog with a high O/L score: Pako Daker. This illustrates the case of a characteristic of the inbreeding depression: the variable influence in the population, even within a family or within littermates, making it difficult to find general trends.
10 – The actual population
Figure 9 shows the average values for the actual population, calculated from the values for the litters born between 2001 and 2009, both inclusive (675 litters). Historical values are obtained by means of 1976-1984 populations. For average life see section 8.
11 – Analysis of the population born in 2009
Having performed a detailed study of the lineages of a sample of the population born in 2009 (148 litters) in connection with the litter size O/L score and inbreeding coefficient F, and obtained the following analysis results.
11A – Structure of the 2009 population
Population born in 2009 shows two clusters differentiated by lineage origin (figure 10): Cluster 1 from Bingo v. Ellendonk lineage mostly working dogs, and a Cluster 2 including dogs related mainly with Graaf Quirinus v. Neerlands Stam, Hertog Alpha v. Le Dobry and Prinz v. Norden Stamm. We must note the presence in the second cluster of two popular sires characterized by a statistically significant low O/L score: Prinz v. Norden Stamm and Baron Nike Renewal.
11B – Relationships between the structure of the 2009 population and the litter size O/L
Interestingly, both groups show (Figures 11) significant differences in means O/L [P(Z)=0.00029<0.05], but not in means of inbreeding coefficient F [P(Z)=0.276>0.05]. Otherwise cluster 1 population (working dogs) does not show significant differences [P(Z)=0.224>0.05] in the O/L with the population of a selected sample (1980-1984) of the populations in the node before the start of inbreeding depression.
In view of full current population 2001-2009, the weight of the most influential dogs varies, highlighting a dog on the other: Graaf Quirinus v. Neerlands Stam (figure 12).
In this sense it is conceivable that many of the problems of genetic diseases that have manifested themselves in the population of the European Dobermann from the eighties are associated with the pass trough the “genetic funnel” of these three top sires: Graaf Quirinus v. Neerlands Stam, Hertog Alpha v. Le Dobry, Prinz v. Norden Stamm, and their ascendants. Remember that Hertog Alpha v. Le Dobry, Graaf Quirinus v. Neerlands Stam are characterized by high inbreed progenies, which exceed the average inbreeding coefficient 14% (see section 6 and figure 4). Influence of Prinz v. Norden Stamm is quantitatively less important than two others, but no qualitatively because two modern top sires: Baron Nike Renewal and Gino Gomez del Citone are strong related with Prinz by ascendance.
11C – Relationships between the structure of the 2009 population, litter size O/L and inbreeding coefficient F.
The statistical analysis shows no significant differences inter or intra clusters between both litters size O/L and inbreeding coefficient F of the progenies. This result shows that different episodes of inbreeding can have different effects on the O/L mean, so it is possible to get inbred lines which show no inbreeding depression.
12- Is average life span dependent of lineage origin?
The conclusions of the analysis of 2009 population, suggests the possibility that some phenomena associated with inbreeding depression, such as average life, may differentially affect dogs of both clusters.
The inability to perform the analysis of average life on the same sample of dogs born in 2009, determines that a statistical analysis on a separate sample, while respecting the same categorical approach: two separate groups corresponding to the pedigrees shows in figure 10, born in the same period of time.
The spreadsheet includes 282 records of dogs born between 1995 and 2005 belonging to both pedigree groups. Each record includes: name of the dog, dates of birth and death, life span ALS, group pedigree (B=cluster 2 lineage –Quirinus, etc-; T= cluster 1 lineage –Bingo, etc-, and inbreeding coefficient F.
Two stage cluster statistical analysis for the variables: Group (B or T) and average life ALS, shows three cluster statistical significant at 95% level of confidence. Two clusters with higher average life (about 8 years): one of the total cases of the working dog group (T3) and another of the 55.7% of B group (B1); and one cluster (26.6%) with low average life (3.3 years) of B group (B2) (figure 13). The average life of these two clusters (B1 & T3) with higher life expectancy (about 8 years) is still around the average life span of the breed (7.4 years). No statistically significant correlation between average life ALS and inbreeding coefficient F.
There is no statistically significant cluster characterized by average life significantly higher (ALS>10 years) than the average life for the breed. Basically, the high prevalence of DCM in the breed (about 50%) and the incidence of cancer at an early age, determines that the highest average life clusters differ by approximately -1 year for the historical average values in the eighties (9.0 years), despite remarkable advances in Veterinary science for thirty years. Inbreeding can have profound effects on the immune system, predisposing to increased immunodeficiency, autoimmune disease and cancer (Angles et al. 2005-p.174). Otherwise high longevity class dogs (ALS>12 years) is associated with the cancer resistance and delayed onset of major diseases (Waters et al. 2003), and age at risk for DCM is around 7 years (Wess et al. 2010).
Introducing a third variable O/L in the analysis (figure 14): litter size O/L, group B or T, and average life ALS, curiously only highlights a significant cluster that includes all dogs of group T and characterized by a higher average life and greater size litter.
13 – Is the litter size (O/L) influenced by the inbreeding coefficient F in progenies of the top sires?
Here a two stage cluster statistical analysis for the progenies have been created (705 litters) from the most relevant top sires born between 1995-2005, including: Jivago v.h. Wantij, Wanja Wandor v. Stevinhage, Baron Nike Renewal, Gino Gomez del Citone, Nitro del Rio Bianco, Pimm’s Number 1 iz Doma Domeni, Urbano del Diamante Nero, F‘Hiram-Abif Royal Bell, Fedor del Nasi and Pako Daker progenies (figure 15).
The results show two separated groups characterized by different O/L score and F level statistically significant at a level of confidence of 95% for all categorical variables –top sire- and for the two numerical variables: O/L –“n” in the figure- and F. The first cluster with a mean O/L=5.96 and F=12.38%, and a second cluster with O/L=6.84 and F=9.73. It’s clear that O/L score is influenced by level of F. This means that mating sired by dogs of cluster 2 and F<10% produce in average a gain of about 1 puppy per litter.
14 – Is the litter size of top sires smaller than the litter size in the general population?
Comparing in the last sample the top sires progenies born in 2000 with the general at random 2000 population (112 litters), regarding litter size (O/L). Top sires progenies including 52 litters born in 2000 from Wanja Wandor v. Stevinhage, Jivago v.h. Wantij, Gino Gomez del Citone, Baron Nike Renewal and Nitro del Rio Bianco. Results show significant differences (figure 16) at a statistical level of 95%. It confirms the same downward trend.
15 – Is inbreeding on popular sires correlate with the beginning of inbreeding depression?
Figures 2 and 4 demonstrates that a period of pronounced and sustained inbreeding during the period 1960-1970 is associated with the onset of inbreeding depression in the European population of Dobermann.
Genetic drift is the change in frequency of particular alleles over time because of random sampling or chance of the genes during the formation of gametes. This genetic phenomenon is especially accentuated in isolated populations (as pure breed dogs) with few founders, historical bottlenecks, and those where there is a disproportion in the reproductive population between males and females (popular sire effect). Over time there is a tendency to complete loss of some alleles, and concentration of others. Inbreeding practices on popular sires increases the probabilities of homozygous of these concentrated genes. Even when they were healthy, popular sires are heterozygous carriers of deleterious genes. Inbreeding is what makes them homozygous (see figure 17).
As Professor Frank Nicholas (2009) underlines: “A genetic contribution to the aetiology of a disorder is indicated if incidence is higher in some families than in others within a breed, and also incidence is higher in some breeds than in others”. Nearly half of genetic diseases reported in dogs occur predominantly or exclusively in one or few breeds. Susceptibility of some breeds to particular diseases coupled with a near absence in other breeds indicates that a subset of dog breeds is strongly enriched for particular disease alleles propitiated by origination from a small group of founders, population bottlenecks and popular sire effects (Sutter and Ostrander, 2004).
Last evaluation of impact of inherited disorders on the 50 most popular breeds, placed the Dobermann in the seventh worst place with predisposition to 53 genetic disorders (Asher et al. 2010).
The popular sire practice leads to a dissemination of genetic disorders. Simulations shows that, under a realistic scenario regarding the imbalance in the use of sires, the dissemination of the risk was indeed 4.4 times higher than under random mating conditions (Leroy. and Baumung, 2010).
Dog breed populations can go through a permanent reduction of genetic diversity due to three factors: (1) only a small fraction of all purebred males and females actually reproduce (Ubbink et al.,. 1998,2); (2) there is an unequal number of litters among reproductive males (Nielsen et al. 2001); and (3) dog breeds are often fragmented (Bjornerfeldt et al. 2008), as we have been observed in the European 2009 population (see section 11).
Practical recommendations for the maintenance of genetic diversity in relation with mating systems show minimizing co-ancestries is more effective in maintaining genetic variation than other strategies: the lowest average co-ancestry between sire and dam maximizes genetic diversity (Lacy, 1995).
Reduced heterozygosis in Dobermann genome has been demonstrated by some authors. In the Major Histocompatibility Complex (MHC) Class II genes (Kennedy et al. 2002), authors compare the number of alleles at the alleles frequency and diversity for the locus DLA II -DRB1, DLA II -DQA1 For locus DLA II -DRB1, the Dobermann is the breed with a smaller number of alleles (only 2 of 30 possible alleles) and frequency of one allele (DLA-DRB1-006) is nearly homozygous (97.5%), the bigger allele frequency of any allele in the nine analyzed pure breeds. In another study of MHC Class II: DRB1, DQA1 and DQB1 genes in 40 Dutch Dobermanns, results show that 89% of Dutch Dobermanns had the same haplotype (Mandinguers 2005).
The Y chromosome diversity and haplotypes might be particularly informative with reference to the limited number of sires that participates in the reproductive population of the breed.
The level of “top sire effect” in Dobermann is so high, that in a study of the diversity of the Y chromosome in different pure breeds, Dobermann has a diversity of 0: a single haplotype H7 for all unrelated 17 Dobermanns of the sample (Bannasch et al. 2005).
Different genetic parameters have been used to monitor genetic diversity or heterozygosis, such as “founder genome equivalents” and “effective population size”. The homozygosis or long-term inbreeding in the European Dobermann subpopulation is associated with two characteristics: (1) a small fraction of males and females actually reproduce, and (2) only few males participate in the genetic structure of the new generation (top sire effect).
A simple balance in the generation 1999-2007 reveals that only 34 studs (0.44% of the males population) have produced the 23.4% of the population, but in the period 1988-1996 only 20 studs have produced the 42.8% of the population (0.11% of the males population at this time).
Defining a simple parameter to estimate approximately the impact of the top sire effect on the population, calculated from the ratios of top sires (with more than 100 offsprings) progenies (Nt), versus total population (No). Figure 18 shows the evolution through generations of top sire impact Nt/No.
Two stage cluster statistical analysis for continuous variables: year generation, impact of the top sire effect (%), and litter size (O/L) (table 3); it shows two clusters highly significant at confidence level 95% for all variables. Clusters have different litter size O/L means: 7.54 versus 6.74 associated to different levels of top sire impact (Nt/No=48.19% versus 30.96%) and indicates that the top sire effect has a critical incidence in the inbreeding depression on the population of the European Dobermann. Mean time of two clusters (1988) is in good coincidence with the beginning of the inbreeding depression (~ generation 1985).
Angles J.M., Kennedy L.J., Pedersen N.C. (2005). Frequency and distribution of alleles of canine MHC-II DLA-DQB1, DLA-DQA1 and DLA-DRB1 in 25 representative American Kennel Club breeds. Tissue Antigens: 66: 173–184.
Asher L., Diesel G., Summers J.F., McGreevy P.D., Collins L.M (2009). Inherited defects in pedigree dogs. Part 1: Disorders related to breed standards. The Veterinary Journal 182, 402–411.
Ayroles JF, Hughes KA, Rowe KC, Reedy MM, Rodriguez-Zas SL, Drnevich JM, Cáceres CE, Paige KN. (2009). A Genomewide Assessment of Inbreeding Depression: Gene Number, Function, and Mode of Action. Conservation Biology: Volume 23, Issue 4, 920–930.
Bannasch D.L., Bannasch M.J., Ryun J.R., Famula T.R., Pedersen N.C. (2005). Y chromosome haplotype analysis in purebred dogs. Mammalian Genome, Volume 16, 273–280.
Bjornerfeldt S., Hailer F., Nord M., Vila C. (2008). Assortative mating and fragmentation within dog breeds. BMC Evol Biol, 8: 28.
Brinks, J. S., R. T. Clark and F. J. Rice, 1962. Estimation of genetic trends in beef cattle. J. Anim. Sci. 20 903.
Calboli F., Sampson J., Fretwell N., Balding D. (2008). Population Structure and Inbreeding From Pedigree Analysis of Purebred Dogs. Genetics; 179(1): 593–601.
Falconer, D.S. and Mackay, T.F.C. (1996). Introduction to quantitative genetics, 4th edition. Longman, Harlow, Essex.
Giger at al., (2006). Chapter 15 of The Dog and Its Genome, Ostrander Edt., CHS Monograph, V44, p.257.
Gill M.A. (2001). Perinatal and late neonatal mortality in the dog. Doctoral Thesis, University of Sydney.
Kaiser, G. 1971 Die Reproduktionsleistung der Haushunde in ihrer Beziehung zur Körpergrobe und zum Gewicht der Rassen. Teile 1–3. J. Anim. Breed. Genet. 88, 118.
Kennedy LJ, Barnes A, Happ GM (2002). Evidence for extensive DLA polymorphism in different dog populations. Tissue Antigens, 59: 194–204.
Lacy, R. (1995). Clarification of genetic terms and their use in the management of captive populations. Zoo Biology 14, 565-578.
Leroy G. and Baumung R. (2010) Mating practices and the dissemination of genetic disorders in domestic animals, based on the example of dog breeding Anim. Genet. 2010, Journal compilation. Stichting International Foundation for Animal Genetics.
Mäki, K., Groen, A.F., Liinamo, A.-E. and Ojala, M. (2001). Population structure, inbreeding trend and their association with hip and elbow dysplasia in dogs. Animal Science, 73: 217-228.
Mandinguers P.J. (2005). Insights in the pathogenesis of Dobermann hepatitis, Thesis Univ. Utrech, p.188.
Mandigers PJ, Senders T, Rothuizen J. (2006). Morbidity and mortality in 928 Dobermanns born in the Netherlands between 1993 and 1999. Vet. Rec., 18; 58 (7): 226-229.
Nicholas F. (2009). An Introduction to Veterinary Genetics. Blackwell’s Science. 3er edition, p. 156.
Nielen AL, Van Der Beek S, Ubbink GJ, Knol BW (2001). Population parameters to compare dog breeds: Differences between five Dutch purebred populations. Vet Q, 23: 43-49.
Robinson, R. (1973). Relationship between litter size and weight of dam in the dog. Vet. Rec. 92, 221–223.
Sargan D., Rooney N. et al., 2008. Pedigree dog breeding in the UK: A major welfare concern? Independent report of a expert group commissioned by the Royal Society for the Prevention of Cruelty to Animals (RSPCA),
Sutter N.B., Ostrander E.A. (2009). Dog star rising: The canine genetic system. Nat. Rev. Genet. 2004;5: 900–910.
Ubbink GJ, Van de Broek J, Hazewinkel HA, Rothuizen J. (19981).Risk estimates for dichotomous genetic disease traits based on a cohort study of relatedness in purebred dog populations. Vet. Rec.; 142:328-331.
Ubbink GJ, Van de Broek J, Hazewinkel HA, Rothuizen J. (19982). Cluster analysis of the genetic heterogeneity and disease distributions in purebred dog populations. Vet. Rec., 142:209-213.
Waters D., Kengeri S., Clever B., Hayek M. (2003). Exceptional Longevity in Pet Dogs Is Accompanied by Cancer Resistance and Delayed Onset of Major Diseases. J. Gerontol. A. Biol. Sci. Med. Sci. 58 (12): B1078-B1084.
Wess G, Schulze A, Butz V, Simak J, Killich M, Keller LJ, Maeurer J, Hartmann K: Prevalence of dilated cardiomyopathy in doberman pinschers in various age groups.
J Vet Intern Med. 2010; 24(3):533-538.
O/L = litter size or offsprings per litter
BW = body weight
SH = shoulder height
O/L score = average of litter size
P(Z )= probability for a Z test.
AWB = average weight at birth
F = inbreeding coefficient or COI
ALS = average life span
Cluster = statistical group of cases
Nt/No = impact of the top sire effect on the population
MHC = Major Histocompatibility Complex