1. Introduction
Chronic progressive lymphedema (CPL) is an incurable disease, progressing throughout the life of horses with swelling, skin thickening and crusting, hyperkeratosis, fibrosis, skin folds, and nodules, often complicated with exudative wounds and ulcerations or even with verrucous lesions, of the distal limbs [
1,
2,
3,
4,
5,
6,
7,
8,
9,
10,
11,
12,
13,
14,
15,
16,
17,
18,
19,
20,
21]. Secondary bacterial, fungal, and recurrent parasitic infections worsen and complicate lesions [
8,
9,
10,
13,
14,
19,
21]. Finally, CPL leads to disability of limbs, loss of appetite, and condition deterioration [
4,
9,
10,
19,
21]. Draught horse breeds, particularly in breeds with genealogical relationships with Belgian draught horses [
2,
3,
4,
10,
19], are the only breeds that have been reported as susceptible to this disease. Reports on CPL include Ardennes [
6]; Belgian [
1,
4,
8,
15,
16,
19,
21,
22]; Breton [
6]; Boulonnais [
6]; Cheval de Trait Auxois [
21]; Clydesdale [
1,
4,
7]; Comtois [
19]; Danish draught [
1]; Friesian [
23]; German draught such as Black Forest, Mecklenburg, Rhenish German, Schleswig, Saxon-Thuringian, South German [
1,
2,
3,
4,
5,
10,
11,
12,
13,
20,
24], and Gypsy Cobs; Gypsy Vanners [
14]; Shire [
1,
4,
7]; Trait du Nord; and Trait Mulassier Poitevin [
18]. Anecdotal studies presumed large genetic effects on CPL based on progeny records of stallions [
1,
2,
3,
5]. In addition, breed disposition may be indicative for genetic involvement. Previous studies in Belgium and across all different breeds of German draught horses showed heritabilities in the range between 0.11 and 0.26 with standard errors of 0.05–0.06 [
10,
11,
12,
19,
22]. The restriction of data to older horses (>3 years) gave higher heritability estimates for Belgian draught horses compared to horses of all ages (0.26 versus 0.11) [
22]. However, within-breed heritabilities in German draught horse breeds varied considerably [
10]. In Black Forest and South German, heritability estimates for overall prevalence and prevalence per limb were 0.29–0.12 and 0.14–0.17, respectively. In Rhenish German and Schleswig, corresponding estimates were highest with values of 0.98–0.62 and 0.82–0.55, respectively, and in East German draught horses, heritabilities were at 0.25–0.39 [
10]. With the exception of South German (
n = 455), data sets for single breeds were small and included only 77–141 horses. A limitation of genetic studies in German draught horse breeds may be the small number of horses under study, resulting in higher standard errors for genetic parameter estimates [
10,
11,
12]. In South German, standard errors for heritabilities were at 0.07, but increased to 0.12–0.29 (Black Forest), 0.15 (East German draught), 0.24–0.25 (Schleswig), and 0.32 (Rhenish German) [
10]. The small population sizes of these horse breeds, ranging from 178 (Schleswig, N
e = 58) to 1085 (Black Forest, N
e = 337), 1189 (Rhenish German, N
e = 435), and 1974 (South German, N
e = 571), impose restrictions on sampling of large data sets (
https://www.genres.de/fachportale/nutztiere/rote-liste-nutztierrassen, accessed on 30 March 2024).
The Rhenish German draught horse was the most common horse breed in Germany in the first half of the 20th century. The main breeding areas were the Rhineland, Westphalia, and Saxony. The dramatic changes in agriculture and transportation in the 1960s led to an enormous reduction in demand for draught horses and thus to a sharp decline in the population size of draught horse breeds to just a few thousand (
https://www.genres.de/fachportale/nutztiere/rote-liste-nutztierrassen, accessed on 30 March 2024).
The Rhenish German draft horse was developed at the end of the 19th century from local populations and imported stallions of other draught horse breeds from England, Denmark, France, and the Netherlands and, from 1870 onwards, mainly from Belgian stallions. The Wickrath state stud in northern Rhine Prussia concentrated on the Belgian stallion type. The studbook was founded in 1892. As late as 1946, more than 26,000 mares were registered in the studbook (
https://www.g-e-h.de/rote-liste-menu/rote-liste, accessed on 30 March 2024) [
25]. A few photographs of Rhenish German draught horses are shown in
Supplementary Material Figure S1.
A multipoint linkage analysis across German draught horse breeds identified four-chromosome-wide significant loci on horse chromosome (ECA) 1, 9, 16, and 17, and three further breed-related loci on ECA 4, 7, and 10 [
24]. For three loci (ECA 1, 10, and 17), immune-response-associated genes were located within these loci. A vicious cycle of interconnected and reinforcing-each-other factors were claimed to be the key players in CLP pathogenesis [
21]. The main components may be the lymphatic elastic system with a failing elastic network and chronic inflammation leading to an overregulated inflammatory immune response [
26,
27,
28,
29] as well as autoimmune responses [
13,
21,
24]. The genetic disposition may be present in any of the components initiating and driving CPL [
10,
11,
12,
19,
21,
22].
The reason why males and stallions develop more severe CPL lesions at younger ages than geldings and females is still not fully understood [
10,
11,
12,
19,
22]. The progression of CPL in females is slower but continues to higher ages than in males and geldings among Rhenish German [
20]. An arrest of an increase in CPL scores occurred at a mean age of 16 years in males, 18 years in geldings, and 20 years in females and when the minimum age was ≥11, ≥14, and ≥16 years, respectively [
20].
In addition, CPL scores are influenced by non-genetic effects, which explain a significant proportion of the phenotypic variance of CPL scores in Rhenish German draught horses [
20]. Certain horse-farm-related factors trigger the prevalence and severity of lesions [
10,
11,
12,
13,
20,
30]. Keeping horses on pastures, in outside pens on rubber meadows or pastures, and in paddocks reduces the severity of CPL lesions [
13,
20]. Stable hygiene and cleanness are important for a lower prevalence [
10] and less severe CPL lesions [
10,
12,
20]. Restrictive feeding of concentrates and feeding hay with straw in winter instead of hay silage lowered the risk of CPL lesions [
10,
20,
26]. Nevertheless, the effects of farm-related variables did not explain more than 11.6% of the total phenotypic variance for the CPL score [
20]. Extending the model to include the random effect of the horse farm increased the explained phenotypic variance by a further 2.9% to 14.5%. Therefore, our objective of this follow-up study was to analyse the heritability of CPL in Rhenish German draught horses using linear and threshold animal models. Furthermore, we used age-structured subsamples and models for censored data to account for the progressive course of CPL. Relationship matrices were based either on pedigree data or on genomic relationship matrices. Models employed included a random animal effect and the fixed effects related to data structure and horse-farm-related factors, which were significant in our previous study in Rhenish German draught horses.
4. Discussion
In this study, we analysed the heritability of CPL scores using linear and threshold animal models and in addition, compared the heritability estimates with models using genomic relationship matrices and threshold models. The present analyses showed high heritability estimates for CPL scores when including all horses of any ages in the linear and threshold models and even higher estimates when employing age restrictions or censoring of data. Heritability estimates in the threshold models increased from 0.595 to 0.662–0.788 in censored data. We may assume that the progressive course of the disease [
1,
2,
3,
4,
5,
6,
7,
8,
9,
10,
11,
12,
13,
14,
15,
16,
17,
18,
19,
20,
21] may be responsible regarding that in younger horses, the lesions are not fully expressed and therefore, heritability estimates increase when younger horses are not regarded in the analyses or censoring is employed in the analysis. Similarly, the higher heritability estimates with lower standard errors for the CPL scores in the hind limbs may be interpreted in the same way, that expression of CPL is faster and can reach more severe scores than in the front limbs. In front limbs, standard errors may be larger because there is more random error variation due to differences in the progression of the expression of CPL within families. The results of our analyses let us propose that censoring may be useful for horses under an age of 8 to 9 years. In the subsamples of older horses, additive genetic variances and heritability estimates decrease because differences in the severity of the lesions due to the course of the progression of CPL decrease and even the selection of the most severely affected horses may take place [
20].
We found very high additive genetic correlations for the CPL scores between all four limbs as well as between the CPL scores of the limbs and the CPL score across all four limbs. Therefore, there is no loss of information when using the CPL score across all four limbs instead of the CPL scores for each limb. In addition, we found high phenotypic correlations between limbs and for the CPL score across all limbs with CPL scores of the limbs.
Employing a genomic relationship matrix increased heritability estimates for CPL scores to a slight extent. Due to the deep pedigree matrix, the additional increase was small but evident. However, an exception was the CPL trait defined as the sum of the dichotomous CPL-bin-scores over all four limbs with an increase from h
2 = 0.375 to h
2 = 0.433 for the genomic pedigree matrix. As our data included horses of both sexes, and even geldings, from more than 25 birth years with genotypes and phenotypes and as no selection by phenotypes or estimated breeding values was performed, we assume that the overestimation of heritabilities seems very unlikely [
37].
On the other hand, distinguishing only between unaffected and affected horses resulted in much smaller heritability estimates with values of 0.11–0.25 in linear and 0.18 to 0.20 in threshold models. This may indicate that a larger part of the additive genetic variation is due to differences in the severity of CPL lesions and a smaller part regarding whether a horse is affected by CPL or healthy. This offers breeders the option that breeding against CPL scores may first lead to milder signs of the CPL lesions and to a slower progression of CPL lesions with age and in the next stage also to horses being less susceptible to CPL at all.
The present results revealed much higher heritability estimates for CPL scores when compared to the previous study for Belgian draught horses [
22]. The previous estimates across all German draught horse breeds, based on linear animal models, were 0.211 for the prevalence as the 0/1-trait across all limbs and 0.235 for the number of CPL-affected limbs [
10]. These estimates are comparable to the results of the present study for the 0/1-traits in the linear and threshold model for CPL-bin-score, but higher estimates were obtained for the sum of the dichotomous CPL scores over all limbs in the threshold model, particularly with the genomic relationship matrix. It seems that the sum of the dichotomous CPL scores over limbs contains additional information on the degree of the severeness of the CPL lesions as the affection of CPL will not start in all limbs at the same age but progresses from the hind limbs to the front limbs [
10,
20]. In the present data, on average, more than two limbs were affected with 5 years of age and more than three limbs with the age of 8 years.
The rather high heritability estimates of 0.824–0.979 for the prevalence and 0.427–0.618 for the number of CPL-affected limbs in a small sample of Rhenish German in a previous study by Wallraf [
10] may either indicate that at this time there were even greater additive genetic differences in this population studied or non-random sampling may have contributed to these high estimates. Contrasting results for East German draught horses, which genealogically belong to Rhenish German [
38], were reported in the same previous study with heritability estimates of 0.248–0.359 (prevalence) and 0.392–0.501 (number of affected limbs) [
10]. The small data sets for the Rhenish German from West and East Germany in the previous study [
10] seem to have influenced the size of the heritability estimates. Whether horse lines that have been lost contributed to larger additive genetic differences cannot be excluded. The rather high prevalences of CPL for Rhenish German draught horses of 0.961 in West Germany and 0.812 in East Germany in this previous study [
10] let us assume that a very low number of horses has caused these rather high heritability estimates with linear models. Nevertheless, the additive genetic variation to be exploited for breeding programmes may be assumed as larger in the present German population than in Belgian draught horses. The reason for this result may be due to the influence of Belgian stallions in the former time, which were suspected to have sired progeny with a high prevalence of CPL [
2,
3,
4,
5]. This way, sire lines may have been selected with large differences for CPL susceptibility. The moderate-to-high positive additive genetic correlation of CPL scores and CPL-bin-scores with body size traits may also suggest that incrossings of large-framed Belgian stallions may have increased the risk of CPL and development of more severe CPL lesions. A limitation of the present study may be seen in the number of horses studied, and consequently the impact of families with extreme trait values may have a larger effect on heritability estimates in comparison to studies using large data sets. Furthermore, standard errors for additive genetic correlation estimates were for some body traits with the binary CPL traits rather high. Therefore, additive genetic correlation estimates with rather high standard errors need to be critically scrutinised. Inbreeding coefficients estimated from pedigree data were lowest in Rhenish German (1.80%) and even higher in South German (2.79%) [
38]. We propose therefore that positive assortative matings, which can increase inbreeding and inflate additive genetic variance [
39,
40], may only have a small influence on heritability estimates in Rhenish German. The small population sizes of the Rhenish German draught horse and stud farms with predominantly few horses, which are widely distributed across Germany, are limiting factors for increasing the number of horses studied to several thousand animals.
Body and hoof traits had moderate-to-high heritabilities, particularly, height at withers, circumference of the coronary band, and length of the dorsal wall of the hooves. The present study suggests that breeding for large-framed horses increases the risk for Rhenish German to be affected by CPL and to develop more severe lesions than smaller-framed horses [
2,
3,
30]. Additive genetic correlation estimates with body length and chest circumference have to be considered with caution due to their large standard errors, whereas standard errors for height at withers were much smaller. The residual and phenotypic correlations with the height at withers were smaller than the additive genetic correlations and therefore, selection on phenotypic traits may be less effective than the use of estimated breeding values for CPL scores and height at withers. Rather high additive genetic correlations were found for cannon bone circumference in the front and hind limb with CPL scores and even higher ones for CPL prevalence and CPL-bin-sum. These findings also confirm the observations of early reports that stallions exhibiting a heavy calibre, large frame, strong limbs, and bones with large articulations were more often affected by CPL [
2,
3,
5,
30]. In consequence, preferring those stallions in the decision for breeding would promote higher risk of CPL in their progeny [
2,
3,
5,
30]. In addition, we found that horses with longer lengths of the dorsal hoof walls may be genetically related with higher CPL scores, but not with a higher CPL prevalence. Shore D hardness seems to be negatively genetically correlated with the prevalence and severity of CPL. This may indicate that there could be a genetic relationship between hoof horn structure and CPL; this may support a relationship between an abnormal proliferation and differentiation of keratinocytes and CPL [
13]. This finding may also explain the increasing prominence of chestnuts, ergots, and bulges in the fetlocks with increasing CPL scores [
13,
20]. Therefore, we propose that the pathogenesis of CPL may also be influenced by the disturbance of the proliferation and differentiation of keratinocytes, resulting in epidermal hyperplasia and an excessive production of keratin [
13]. Thus, pathogenesis of CPL may be complicated by a further factor, which also has a genetic determination.
Skinfold thickness showed moderate positive additive genetic correlations of 0.440, 0.450, and 0.241 with CPL-bin-score, CPL-bin-sum, and the CPL score, respectively. Heritability of skinfold thickness and additive genetic correlation with CPL-bin-score agreed very well with the previous study across all German draught horse breeds and for South German draught horses [
10,
12]. On the other hand, the results for the Belgian draught horse were inconsistent between all horses and the > 3-year-old horses. In addition, a heritability estimate of 0.01 for skinfold thickness with a standard error of 0.02 can lead to uncertainty issues when estimating genetic correlations, as even small changes in the additive genetic variance have a large impact on the outcomes [
22]. The genetic association between skinfold thickness and prevalence of CPL in German draught horses may also indicate that a general epidermal hyperplasia may drive dermal thickness with hyperkeratosis and the overproduction of keratin in the distal limbs.
Infestation with
Chorioptes mites and bacterial colonisation are secondary complications in the distal limbs along with the progressing of CPL [
9,
10,
12,
13,
14,
21], but susceptibility to infestations with
Chorioptes mites showed high heritabilities in German draught horse breeds [
10,
12] and moderate additive genetic correlations with CPL prevalence in South German draught horses [
10,
12]. Therefore, we may assume that significant additive genetic differences between horses in skin immunity and food supply and environmental conditions for
Chorioptes mites, provided by skin scales, crusts, and excessive amounts of keratin, contribute to the increased susceptibility to these secondary complications with CPL [
13,
14,
21].
In summary, the present study corroborated the significant contribution of genetics to the prevalence and expression of CPL lesions with different degrees of severity in Rhenish German. In addition, the impact of breeding for heavy calibre, large frame, strong limbs, and bones with large articulations became obvious as additional genetically disposing factors through the additive genetic correlations shown in this study.