Methodological approaches to assessing the genetic predisposition of osteochondrosis of the tarsocrural joint in the Pura Raza Española horses

Methodological approaches to assessing the genetic predisposition of osteochondrosis of the tarsocrural joint in the Pura Raza Española horses M. Ripollés-Lobo a, ⇑ , A. Molina b , M. Novales c , C. Ziadi b , E. Hernández c , M. Valera a a Departamento de Agronomía, Escuela Técnica Superior de Ingeniería Agromómica, Universidad de Sevilla, Ctra Utrera Km 1, 41013 Sevilla, Spain b Departamento de Genética, Universidad de Córdoba, Edificio Mendel, Campus de Rabanales, 14014 Córdoba, Spain c Departamento de Medicina y Cirugía Animal, Universidad de Córdoba, Edificio Hospital Clínico Veterinario, Campus de Rabanales,14014 Córdoba, Spain article info Article history: Received 27 October 2023 Revised 5 July 2024 Accepted 8 July 2024 Available online 25 July 2024 Keywords: Breeding genetics Equine Genetic disease Heritability Hock abstract Osteochondrosis (OC) is a developmental orthopaedic disease of significant concern in numerous sport horse breeds, with significant international relevance. Using digital radiographs, we assessed the occurrence of hock (tarsocrural joint) OC in 3 048 Pura Raza Española (PRE) horses which took part in a morpho-functional test, in three specific locations in the tarsus limbs: the Distal Intermediate Ridge of the Tibia (DIRT), the lateral trochlear ridges of the talus (LTT), and the medial trochlear ridges of the talus (MTT). An incidence rate of 13.3% was found for hock OC in the analysed sample, with the highest incidence rate observed in DIRT (10.0%) and the lowest in MTT (0.2%). Estimates of genetic predisposition to hock OC were carried out using three genetic approaches: 1a) a binomial threshold model based on the presence or absence of OC, 1b) a multinomial threshold model, on a scale from 0 (absence) to 3 (maximum), and 2) a linear model. The effects considered in the models included sex, genetic origin and stud class. All the analyses were based on the Bayesian inference methodology, using the THRGIBBS3F90 software. The binomial threshold model yielded the most suitable results, with an estimated heritability for Overall hock OC of 0.71 ± 0.055 on the underlying scale (0.53 on the observed scale), ranging in different locations from 0.48 ± 0.087 (LTT) to 0.66 ± 0.063 (DIRT) on the underlying scale (0.10 and 0.38 on the observed scale, respectively). The highest significative genetic correlation was observed between Overall and DIRT (0.97) for approach 1a, and the lowest significant genetic correlation was between Overall and LTT (0.49), for approach 2. This study contributes valuable insights into the genetic predisposition towards, as well as for the potential for selective breeding against, hock OC in PRE horses, and provides a basis for future research and breeding programmes aimed at minimising the occurrence of hock OC and promoting the overall health of this breed. Ó2024 The Authors. Published by Elsevier B.V. on behalf of The Animal Consortium. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Implications Research into hock osteochondrosis in Pura Raza Española horses has significant implications for breeding programmes. The study found a notable incidence rate in the analysed sample, pinpointing specific tarsocrural joint locations. Genetic parameters analysis using different models determines that measuring the disease on a binary scale is the most practical method for genetic selection to reduce disease transmission risk. This information is crucial for breeders to selectively breed against hock osteochondrosis and reduce its occurrence in future generations. By enhancing overall health through selective breeding, the research supports Pura Raza Española horse well-being and breed sustainability at an international level. Introduction Osteochondrosis (OC) is a joint disease of horses, which can be categorised under a generic group of diseases known as Developmental Orthopaedic Disease (Van Weeren, 2005). OC is characterised by a disturbance in the process of enchondral or endochondral ossification (Romvári et al., 2006). Endochondral ossification takes place within the growth cartilage, a temporary tissue destined to be replaced by bone (Hunziker et al., 2007). This growth cartilage is found at the metaphyseal growth plate or physis and in the epiphyseal growth cartilage, located immediately beneath the articular cartilage in immature joints (Ytrehus et al., 2007). The cartilage on the surface of an OC lesion can fracture, givhttps://doi.org/10.1016/j.animal.2024.101255 1751-7311/Ó2024 The Authors. Published by Elsevier B.V. on behalf of The Animal Consortium. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). ⇑ Corresponding author. E-mail address: [email protected] (M. Ripollés-Lobo). Animal 18 (2024) 101255 Contents lists available at ScienceDirect Animal The international journal of animal biosciences ing rise to fragments in joints known as osteochondrosis dissecans (OCD). If an osteochondral fragment is present, there may be a component of inflammation/synovitis leading to effusion and possibly mild lameness (Van Grevenhof et al., 2009a). Untreated OC/ OCD accompanied by inflammation may initiate a degenerative process. When OC is diagnosed in older animals, the inflammation and lameness are attributed to secondary degeneration. In older horses, osteoarthritis and arthropathies can play a significant role (Van Weeren and Back, 2016). OC can affect a wide range of joints in the horse, including the hock, fetlock, stifle, and shoulder (Van Weeren, 2006; Distl, 2013). Recently, it has been suggested that osteomyelitis, physitis and septic arthritis may also lead to OC (Wormstrand et al., 2018). The disease develops during the 1st year of life, although it can be diagnosed later in life, but it is essentially present from birth in affected animals (Van Weeren, 2006). The disease affects the majority of horse breeds, such as Standardbred Trotter horses (Lykkjen et al., 2014), Thoroughbred horses (Sloan et al., 2023), Warmblood horses (Pieramati et al., 2003; Hilla and Distl, 2014a; Lauper et al., 2017; Ramos et al., 2022; Zimmermann and Distl, 2023). Estimates of hock OC indicate a prevalence ranging from 5 to 40%, depending on the breed (Wittwer et al., 2007; Jönsson et al., 2011; Naccache et al., 2018). Consequently, this disease carries major economic implications for the worldwide equine industry (Holmbak-Petersen and López, 2018). Numerous studies have been conducted into the aetiology, diagnosis, and management of OC in horses (Ytrehus et al., 2007; Ortved, 2017; Hoey et al., 2022; Ramos et al., 2022). It is a multifactorial disease caused by a variety of factors (Lewczuk and Korwin- Kossakowska, 2012), including genetic predisposition (Douglas, 2003; Distl, 2013), nutrition (Gee et al., 2005; Lepeule et al., 2009), and environmental factors (Denoix et al., 2013). Diagnosis of OC is typically based on clinical diagnosis and digital radiography techniques (Wagner et al., 2005; Lykkjen et al., 2012; Gallo et al., 2013). If OC is left untreated, horses can develop a degenerative joint disease (McCoy et al., 2013); however, it is important to note that not all cases of OC show clinical symptoms, which makes diagnosis challenging. The Pura Raza Española (PRE) is an autochthonous Spanish equine population and is the most important equine breed in the country, in terms of both its census and its historical significance, with 275 018 active horses (Encina et al., 2023). PRE horses are distributed over sixty countries (Solé et al., 2018), carrying significant international weight, and are all managed by a single association, the Royal Association National of Purebred Spanish Horse Breeders (ANCCE). The PRE breed possesses exceptional functional temperamental qualities, making it suitable for a range of equestrian competitions. The breed has also had a significant influence on the formation of numerous European (Frederiksberg, Friesian, Kaldruby, Lipizzaner, and Lusitano) and creole American equine breeds (Ripolles et al., 2020). One of the objectives of the breed improvement programme is to achieve healthy animals by selecting disease-free horses, focusing particularly on conditions with a strong genetic influence, such as OC. Therefore, it is crucial to study OC in this breed, given that this disease can impact the horses’ performance and health, significantly affecting their participation in sport competitions. The objective of this article is to study, for the first time, the genetic predisposition of hock OC in PRE horses, in relation to the location within the hock joint. It also aims to estimate the genetic parameters using different genetic approaches by, on the one hand, treating the data as a dichotomous variable (presents hock OC/does not present hock OC) or taking into account the different classes determined by the radiological analysis; and, on the other, assuming a binomial, multinomial or linear distribution. Material and methods Description of traits and dataset The analysis for this study was based on the results of digital radiographic examinations of the hock (tarsocrural joint) OC of 3 048 PRE horses (1 429 males and 1 619 females). The study data come from three types of controls carried out in the PRE: (a) Qualified Breeding Stock Tribunals (ANCCE, 2024a); (b) morphofunctional competitions (MFC;ANCCE, 2024b); (c) animals with the genetic category of ‘‘recommended young breeders” (horses between 4 and 7 years old participating in young horse selection tests for Classical Dressage and with a high breeding value; ANCCE, 2024c). The qualified breeding stock tribunals and MFC are morpho-functional tests where the movements of PRE horses of different ages are evaluated with tests comparable to classical dressage reprisals. In all cases, each horse underwent one radiographic evaluation. These events, organised by ANCCE to assess a horse’s conformational and/or functional characteristics, as well as its biomechanics and movement, took place between the years 2007 and 2022. The median age of the radiographically examined horses was 4.4 years (between 1 year and 20 years old, with 70% of the animals evaluated being under 4–5 years old). A pedigree matrix of 12 032 animals was created from the horses studied, covering 5 known generations. The genealogical data were taken from the Stud Book ANCCE website (LgANCCE, 2023). The radiographs were analysed by the Clinical Veterinary Hospital of the University of Córdoba (Spain). All x-ray digital images were transferred to a Digital Imaging and Communication in Medicine format workstation and then were analysed with an open-source media image viewer HorosÓ(open-source code software (FOSS, v.3.3.6) program distributed free of charge under the LGPL licence at Horosproject.org and sponsored by Nimble CO LLC d/b/a Purview in Annapolis, MD, USA). The radiographic projections performed for the hock were the dorsolateralplantaromedial oblique view (D45°L-PlMO) and the dorsomedialplantarolateral oblique view (D45°M-PlLO). The data submitted included each animal’s individual registration number, sex, date of birth, and radiographic examination results. In this study, three different locations of the hock were studied: (a) distal intermediate ridge of the tibia (DIRT), (b) lateral trochlear ridges of the talus (LTT) (c) medial trochlear ridges of the talus (MTT)(Fig. 1). Osteochondrosis lesions at each location within the tarsus (DIRT, LTT, MTT) and Overall OC were classified according to the contour of the subchondral bone on a grading scale of 0–3 (Table 1). The presence or absence of an osteochondral fragment (OCD) did not affect the grading of the lesion, as the horses included may have had fragments surgically removed prior to evaluation. In practice, OC is commonly addressed surgically, typically through arthroscopic removal of loose fragments and/or cleaning affected areas, with animals undergoing these procedures for aesthetic and/or commercial reasons. The prognosis for these interventions is good to excellent for most affected joints (Van Weeren and Olstad, 2016). Given that this circumstance occurs in many breeds, horses with and without OCD fragments were included in the study cohort (Fig. 2). Statistical and genetic analysis The incidence rate of the presence / absence of hock OC was estimated based on its location. Two phenotypic definitions have been used to estimate the genetic propensity to suffer hock OC. Trait type 1: treating the data as a dichotomous variable (presents hock OC/absence hock OC) and Trait type 2: considering the different categories determined by the radiological analysis (grades 0, 1, M. Ripollés-Lobo, A. Molina, M. Novales et al. Animal 18 (2024) 101255 2 2, 3), for each of the three location traits and the overall hock OC trait. Furthermore, a Generalised Non-linear Model (GLZ) analysis was conducted to determine the factors influencing the occurrence of hock OC (with a binomial distribution using logit GLZ used for the dichotomic traits, while for discrete traits, we used a GLZ with an ordinal multinomial distribution), determining the incidence rate in relation to the different categories of these factors. All the analyses were carried out using Statistica for Windows v.10.0 software (StatSoft, 2007). The fixed effects included in all approaches (significant effects after the GLZ analysis) were as follows: sex (two levels: male and female), genetic origin (two levels: the ‘‘Ancestral line”: horses with genetic origin from Spain, phenotypically of grey coat colour, and the ‘‘Modern line”: horses with genetic origin from Central Europe and Arabic countries, phenotypically of non-grey coat colour, with the coat colour of the PRE horse serving as a pathway to understanding the various genetic origins), and stud class (three levels: (1) Intensive management in which horses are kept in stables and primarily fed with compound feed and straw; (2) Mixed management that combines stabling and feeding based on compound feed and natural forage, but with limited access to pasture; (3) Semi-extensive management in which the horse has unlimited access to natural pastures although the animals also are kept in stables, and the diet is primarily based on natural grass, with some supplemental compound feed). Genetic parameters of the three traits studied, DIRT, LTT, and OC hock estimated at an overall level (disregarding the specific location) were estimated using an animal model with three different approaches, as follows: Approach 1a: Binomial threshold model. The traits were measured as dichotomous traits: does not present hock OC (0) or presents hock OC (1). To ensure accurate estimation of heritability for binary data, it is necessary to utilise a threshold model to find the underlying distribution. Hock OC is a discrete trait which is not distributed normally. A linear variable underlies the discretely observed trait, defining the threshold in which the category is observed (Casellas and Piedrafita, 2006). The heritability estimates of the binomial threshold model analyses were transformed into the observed scale, following Dempster and Lerner (1950). Approach 1b: Multinomial threshold model. The traits were measured using OC grades (0, 1, 2, 3) and treated as variable with more than one threshold, assuming the ordinal scale in the original variable (Sorensen et al., 2002; Masuda, 2018). Approach 2: Linear model. The traits analysed using OC grades (0,1,2,3) were treated as one continuous variable and used as a ref- Fig. 1. X-rays hock of a 5-year-old male Pura Raza Española horse. Dorsomedial-plantarolateral oblique view (A) and dorsolateral-plantaromedial oblique view (B). References: (1) Distal intermediate ridge of the tibia; (2) Lateral trochlear ridges of the talus; (3) Medial trochlear ridge of the talus. Scale bar, 10 cm. Table 1 Description of the grades of hock osteochondrosis in the Pura Raza Español Horse. Location Grade Description Distal intermediate ridge of the tibia (DIRT) 0 Rounded bone contour, without osteochondrosis. 1 Flattened or slightly irregular bone contour. 2 Small subsidence (concavity) of the bone contour, less than or equal to 50% of the total area to be assessed. Osteochondral fragments may or may not be present.* 3 Large subsidence (concavity) in the bone contour, greater than 50% of the area to be assessed. Osteochondral fragments may or may not be present.* Lateral trochlear ridges of the talus (LTT) or Medial trochlear ridges of the talus (MTT) 0 Rounded bone contour, without osteochondrosis. 1 Flattened or slightly irregular bone contour. 2 Flattened or irregular bone contour. Osteochondral fragments may or may not be present.* 3 Very irregular bone contour, as well as cases with ossification defects inside the bone. Osteochondral fragments may or may not be present.* * In both grade 2 and grade 3, there may or may not be osteochondral fragments; the real difference lies in the concavity in the bone contour, either less than or equal to 50% (grade 2) or greater than 50% (grade 3) of the area to be assessed. M. Ripollés-Lobo, A. Molina, M. Novales et al. Animal 18 (2024) 101255 3 erence model. We also used a linear model using a Bayesian inference. The equation in matrix notation used to solve the mixed models was as follows: y¼Xb þZu þe; with :uN0;A r 2 u  ;eN0;1ðÞfor Approach 1a1b with :uNð0;A r 2 u Þ;eNð0;I r 2 e Þfor Approach 2 where yis a vector with n observations of the different traits, bis the vector of fixed effects: sex, genetic origin and the stud class, u is the vector of additive animal genetic effects, and erepresents the vector of residual effects. Xand Zare incidence matrices relating y with the new parameters in band u, respectively. r u 2 is the additive genetic variance, r e 2 is the residual variance, Iis an identity matrix and Ais the numerator relationship matrix. Multivariate analyses were performed for each of the genetic models (Approach 1a, 1b, 2) to estimate the genetic parameters of locationspecific traits (DIRT, LTT) and overall level of hock OC. The MTT variable has not been included as an individual trait in the genetic parameter estimation model due to the low incidence of OC in this location (it caused convergence issues in the models when it was included). However, the MTT variable has been taken into account in the overall set of OC in the hock joint. All these analyses were conducted for each of the genetic approach models through Bayesian inference methodology, using the THRGIBBS3F90 software from the BLUPF90 family programs (Misztal et al. 2016). A total of 250 000 iterations were considered, with the first 50 000 considered as burn-in, after which every 50th sample was saved for later analysis. Next, a post-Gibbs analysis was performed using the POSTGIBBSf90 program (Tsuruta and Misztal, 2006) to check convergence and to calculate posterior means of the additive and residual variances and high posterior density intervals. Convergence was tested using Monte Carlo sampling error, computed by time-series procedures (Geyer, 1992). Comparison of the different methodological approaches was based on the deviance information criterion parameters (Spiegelhalter et al., 2002) and the 95% Higher Posterior Density confidence intervals (Chen and Shao, 1999). Finally, estimated breeding values (EBVs) of the different approaches, using multivariate models, were calculated for each of the animals in the relationship matrix. Fig. 2. Dorsomedial-plantarolateral oblique view of the hock of a Pura Raza Española horse. Details of Distal intermediate ridge of the tibia. (A) Grade 0; (B) Grade 1; (C) Grade 2; (D) Grade 3. Details of Lateral trochlear ridges of the talus: (E) Grade 0; (F) Grade 1; (G) Grade 2; (H) Grade 3. Details of Medial trochlear ridges of the talus: (I) Grade 0; (J) Grade 1; (K) Grade 2. Images C, D and G show osteochondral fragments (OCD). Scale bar, 2 cm. In the Medial trochlear ridges of the talus location, osteochondrosis is very infrequent and often shows anatomical variations; there are no images for grade 3. M. Ripollés-Lobo, A. Molina, M. Novales et al. Animal 18 (2024) 101255 4 Results Comprehensive knowledge of OC prevalence in specific equine breeds is crucial for devising more effective management and breeding strategies, as well as for promoting the health and welfare of these animals. Table 2 shows the OC incidence rate in the different locations of the hock in PRE horses. A global incidence rate of 13.3% was measured for hock OC in sports PRE horses. Additionally, the cases of hock OC were classified by severity into three categories (1,2,3), which particularly affected the DIRT (10.0%) and to a lesser extent the MTT (0.3%) (Table 2). Considering the degrees of severity, it is notable that grade 2 is the most frequent in the three locations, accounting for 71.2% of animals with hock OC presenting grade 2 in the DIRT location, 61.5% in LTT, and 75.0% in MTT. Meanwhile, grade 3 is less common in the LTT (12.8%) and MTT (0.0%) locations, and grade 1 is the least abundant in the DIRT location (9.8%). Table 3 shows the incidence rate of OC hock in PRE horses according to the different factors analysed. Males had a higher incidence of overall OC and OC at the MTT than females. Horses belonging to the ancestral lineage had a higher incidence of OC at the DIRT (10.6%) than those of the non-ancestral lineage (9.3%). In addition, horses managed in semi-extensive studs had a higher incidence of OC overall (14.7%) and at all the individual locations (10.5% in DIRT; 4.4% in LTT) when compared to the other types of studs. Estimating the heritability of OC in horses is crucial for identifying the genetic contribution to the occurrence of this disease and implementing effective breeding strategies (Table 4). In the underlying scale, the heritability in the binomial (1a) methodology ranged from 0.71 ± 0.055 (Overall) to 0.48 ± 0.087 (LTT), while in the multinomial (1b) methodology, it ranged from 0.34 ± 0.066 (DIRT) to 0.25 ± 0.093 (LTT). In the observed scale, the heritability in the binomial (1a) methodology ranged from 0.53 (Overall) to 0.10 (LTT), while in the linear (2) methodology, the range was 0.17 ± 0.032 (Overall and DIRT) to 0.07 ± 0.024 (LTT) (Table 4). The results suggest that the heritability of OC in different locations of the hock can vary depending on the type of genetic analysis used. All these parameters are statistically significant at the Higher Posterior Density 0.95. The Monte Carlo SEs were small, ranging between 0.0002 for approach 2 and 0.0012 (LTT) for approach 1b. The minimum deviance information criterion value obtained across different approaches was –38 652 for methodology 1a, while the maximum value was 5 637 for methodology 2. The estimation of genetic correlations between distinct traits was carried out using the multivariate models for the three approaches, for the Overall-DIRT-LTT traits, as shown in Table 5. In general, all the genetic correlations were moderately high and positive, and statistically significant (as indicated by the 95% P interval). However, there was an exception in the multinomial and linear models for the correlation between Dirt-LTT, where no statistical significance was found. In all three models, all genetic correlations followed the same pattern: the highest correlation was observed in Overall-Dirt and the lowest in DIRT and LTT. In the binomial model, genetic correlations ranged from 0.97 ± 0.014 (Overall-DIRT) to 0.55 ± 0.164 (DIRT-LTT), while in the multinomial model, they ranged from 0.96 ± 0.016 (Overall- DIRT) to 0.36 ± 0.189 (DIRT-LTT), and in the linear model, the range was 0.96 ± 0.019(Overall-DIRT) to 0.24 ± 0.199(DIRT-LTT). The proportion of coincident animals was therefore determined for the top and bottom 5 and 20% percentiles of the EBV (Table 6). The proportion of coincident animals, based on EBV, between the overall hock OC variable of the binomial model (1a) and the multinomial (1b) and linear (2) models is high for both the 5th and 20th percentiles, with a higher match between binomial and multinomial models. The proportion of coincident animals between the overall binomial model and the multinomial model varied from 86.6% for the top 20 to 66.7% for the bottom 5%. Meanwhile, the comparison overall binomial model vs linear model varies from 84.5% for the top 20% to 65.9% for the bottom 5%. The correspondence in the binomial model varies depending on the location of OC in the hock joint (DIRT, LTT) with the overall trait. The comparison with the DIRT variable shows a very high correspondence, ranging between 93.5% and 84.4% of matching animals for the top 5% and bottom 5% of the EBV, respectively. On the other hand, the comparison with the LTT variable presents a moderate correspondence, with a range between 70.6 and 52.1% of coincident animals for the bottom 20% and bottom 5% of the EBV, respectively. Discussion Osteochondrosis is a developmental orthopaedic disease affecting various equine breeds worldwide. This condition is characterised by aberrant cartilage and underlying bone development, resulting in structural bone weakness. OC can present in different joints and skeletal regions of the horse, with different prevalence and varying across breeds, lineages, and environmental factors. Conformation, nutrition, biomechanical factors, stress response, intrauterine environment, hormonal interactions (Ralston et al., 2011; McCoy et al., 2013), and mitochondrial dysfunction (Desjardin et al., 2014) may play a role in the risk of developing OC. Periodic diagnostic evaluations have been a common practice in many livestock farms to ensure their suitability for genetic selection for several years, but it comes with high costs. However, investing in accurate diagnostics and the early detection of OC can yield substantial long-term benefits by avoiding more serious and costly health complications that may arise in the more advanced stages. Osteochondrosis is a disease of international relevance, and therefore, several studies have investigated the incidence rate of general hock OC in different horse breeds, with some horse breeds being more susceptible than others (Robert et al., 2013). Although incidence rates of hock OC are difficult to compare among studies due to different procedures and the preselection of horses and joints for investigation, previous studies have nevertheless reported a slightly lower incidence of OC in sport horse breeds, Table 2 Incidence rate of hock osteochondrosis according to the location and the degree of osteochondrosis in the Pura Raza Española horses evaluated. Hock OC Number of PRE horses evaluated (% Incidence rate) No OC OC G1 G2 G3 Overall*2 641 407 (13.3) 57 (14.0) 281 (69.0) 69 (16.9) DIRT 2 742 306 (10.0) 30 (9.8) 218 (71.2) 58 (19.0) LTT 2 939 109 (3.6) 28 (25.7) 67 (61.5) 14 (12.8) MTT 3 040 8 (0.3) 2 (25.0) 6 (75.0) 0 (0.0) Abbreviations: OC=Osteochondrosis; DIRT=Distal intermediate ridge of the tibia; LTT=Lateral trochlear ridges of the talus; MTT=Medial trochlear ridges of the talus; G1 = Grade 1; G2 = Grade 2; G3 = Grade 3. * In the event that the same degree of OC is present for more than one location, it is only counted once in the overall assessment. M. Ripollés-Lobo, A. Molina, M. Novales et al. Animal 18 (2024) 101255 5 such as Hanoverian horses, with an incidence rate of 9.7% (Hilla and Distl, 2014a), and very similar incidence rates in French trotters, with 11% (Teyssèdre et al., 2012), and Norwegian trotters, with 14.3% (Grøndahl and Dolvik, 1993). Conversely, lower incidence rates have been documented in breeds like Dutch Warmblood horses with an incidence rate of 1.4% (Van Grevenhof et al., 2009a) and Swedish Warmblood horses with an incidence rate of 6.2% (Jönsson et al., 2011), although these also displayed variability based on their genetic lineage and rearing environment. Additionally, a higher incidence rate of hock OC has been documented in certain horse breeds, with Standardbred horses showing a 39.9% incidence (Hendrickson et al., 2018), South German Coldblood horses a 40.1% incidence (Wittwer et al., 2007), and PRE horses an incidence rate of 33.3% (Boado and López-Sanromán, 2016). The high prevalence of hock OC in some sport breeds highlights the need for early detection and management to ensure optimal performance and welfare. In contrast, in non-sports equine breeds, such as Shetland or Exmoor ponies, Icelandic and Dülmener horses, OC is either absent or very rarely observed (Beckmann et al., 2012). In the PRE horses, OC is more frequent in the hock joint (tarsocrural) than in other locations, such as the fetlock or stifle (Novales et al., 1999). It is noteworthy that grade zero is the most frequent in the population of PRE horses analysed, indicating that Table 3 Result of Generalised Non-linear Model (GLZ) of the analysed traits of the tarsocrural joint, and incidence rate in each of the classes of the factors (%) of hock osteochondrosis, according to the effects, in the Pura Raza Española horses. Hock OC Sex Genetic origin Stud class Male Female P-value Modern line Ancestral line P-value Intensive Mixed Semi-extensive P-value Overall*14.1 12.6 0.042 13.1 13.5 0.756 10.2 11.7 14.7 0.034 DIRT 10.5 9.6 0.424 9.3 10.6 0.049 9.0 9.5 10.5 0.023 LTT 3.8 3.3 0.446 4.1 3.1 0.182 1.8 2.6 4.4 0.009 MTT 0.4 0.1 0.047 0.2 0.3 0.783 0.0 0.2 0.3 0.209 Abbreviations: OC=Osteochondrosis; DIRT=Distal intermediate ridge of the tibia; LTT=Lateral trochlear ridges of the talus; MTT=Medial trochlear ridges of the talus. * If OC is present in more than one location, it is only counted once in the overall assessment. Table 4 Estimation of heritability (SD) of the different traits of hock osteochondrosis using the binomial threshold (1a), multinomial threshold (1b) and linear model (2) using a multivariate analysis in Pura Raza Española horses. Genetic Parameters Hock Osteochondrosis Overall DIRT LTT D.I.C. 1a. Binomial h 2 liab (SD) 0.71 (0.055) 0.66 (0.063) 0.48 (0.087) 38 652 h 2 obs 0.53 0.38 0.10 HPD 0.95 0.601, 0.818 0.526, 0.777 0.305, 0.645 MCse 0.0003 0.0005 0.0009 1b. Multinomial h 2liab (SD) 0.34 (0.066) 0.34 (0.067) 0.25 (0.093) 7 679 HPD 0.95 0.203, 0.458 0.212, 0.471 0.077, 0.429 MCse 0.0004 0.0004 0.0012 2. Linear h 2 obs (SD) 0.17 (0.032) 0.17 (0.032) 0.07 (0.024) 5 637 HPD 0.95 0.108, 0.229 0.105, 0.228 0.024, 0.118 MCse 0.0002 0.0002 0.0002 Abbreviations: DIRT=Distal intermediate ridge of the tibia; LTT=Lateral trochlear ridges of the talus; h 2 liab = heritability on the underlying continuous scale; h 2 obs = heritability on the observed scale; HPD 0.95 = 95% higher posterior density; D.I.C. = deviance information criterion; MCse = Monte Carlo SE. Table 5 Genetic correlations (SD) between the different traits of hock osteochondrosis, using the multivariate binomial threshold, multinomial threshold and linear model in Pura Raza Española horses. Hock OC 1a. Binomial 1b. Multinomial 2. Linear Overall-DIRT HPD 0.95 0.97 (0.014) 0.947, 0.996 0.96 (0.016) 0.935, 0.994 0.96 (0.019) 0.926, 0.990 Overall-LTT HPD 0.95 0.72 (0.106) 0.517, 0.921 0.58 (0.145) 0.312, 0.844 0.49 (0.165) 0.120, 0.775 DIRT-LTT HPD 0.95 0.55 (0.164) 0.239, 0.870 0.36 (0.189) 0.020, 0.716 0.24 (0.199) 0.192, 0.609 Abbreviations: OC=Osteochondrosis; DIRT=Distal intermediate ridge of the tibia; LTT=Lateral trochlear ridges of the talus; HPD 0.95 = 95% higher posterior density. Table 6 Percentage of coincident Pura Raza Española horses within the upper 5 and 20 percentile of the estimated breeding values (EBV), with overall traits approach 1a and different models and traits. Hock Osteochondrosis EBV 20% EBV 5% Top Bottom Top Bottom Overall approach 1a vs Overall approach 1b 86.6 83.2 79.9 66.7 Overall approach 1a vs Overall approach 2 84.5 81.4 75.9 65.9 Overall approach 1a vs DIRT approach 1a 92.5 91.0 93.5 84.4 Overall approach 1a vs LTT approach 1a 68.4 70.6 66.9 52.1 Abbreviations: DIRT=Distal intermediate ridge of the tibia; LTT=Lateral trochlear ridges of the talus; EBV=estimated breeding values; 1a = Binomial threshold model; 1b = Multinomial threshold model; 2 = Linear model. M. Ripollés-Lobo, A. Molina, M. Novales et al. Animal 18 (2024) 101255 6 86.6% of the animals do not exhibit hock OC. The main location of OC in the PRE is the DIRT (10.0%), as shown in Table 2, as occurs in other breeds (Dik et al., 1999; Denoix and Valette, 2001). Up to 8% of the animals can be found with two or more simultaneous locations in this anatomical region (Novales et al., 2000). The lesions generally do not cause pronounced lameness; instead, they can lead to significant effusion. It is noteworthy that young equines with OC do not always exhibit lameness, and the postoperative prognosis of surgical intervention tends to show excellent outcomes in such situations (Laws et al., 1993; McIlwraith, 2013; Brink et al, 2015; Clarke et al., 2015). Lastly, the presence of the MTT, in the distal aspect, typically represents incidental findings which warrant cautious interpretation, and may even be regarded as an anatomical variation. Pieramati et al. (2003) reported a prevalence of 8.3% in DIRT, 0.6% in LTT, and 0.3% in MTT in Maremmano horses. In Dutch Warmblood horses, Van Grevenhof et al. (2009a) reported a prevalence of 11% in DIRT, 9.6% in LTT, and 7.3% in MTT. Lykkjen et al. (2014) reported a prevalence of 15.8% in DIRT and 17% in LTT in Norwegian Standardbred trotter horses. Similarly, Ramos et al. (2022) found a higher prevalence of hock OC in the DIRT location (30.9%) and a complete absence of OC in MTT in Lusitano horses, which followed a similar pattern to that of PRE horses. Osteochondrosis is a disease with a multifactorial origin (McIlwraith, 2005), and a combination of genetic and environmental factors may play a role in the development of the disease. In relation to sex, there are more males with OC in the hock joint than females in the PRE breed. Several authors have reported a higher prevalence of OC in the hock joint in males (Mohammed, 1990; Stock et al., 2005b; Philipsson et al., 1993; Sandgren et al., 1993a). A possible explanation for the influence of sex on the development of OC could be, in the case of PRE horses, the morphological competitions which are held for 1-year-old foals (ANCCE, 2024b), which means that males and females are handled differently before reaching that age. Generally, the aim is to produce males of a larger size and with a more robust conformation than females, so their handling and feeding are adapted accordingly. Repetitive exercise/training could also play a significant role in the development of lesions (Wittwer et al., 2007; Lepeule et al., 2009; McCoy et al., 2013). Although in most studies of genetic parameter estimation, sex has not proven to be significant, it is usually included as a factor, as its inclusion does not affect the estimation of genetic parameters (Yovich et al., 1985; Grøndahl, 1991; Van Weeren et al., 1999). Another important consideration is the regulation of management, nutrition, and exercise during the animal’s early years (Donabédian et al., 2006; Lepeule et al., 2009, Pieramati et al., 2003; Wittwer et al., 2007; Hilla and Distl, 2014b; Russell et al., 2017; Lewczuk et al., 2018). Exercise and nutrition play a key role in OC development, given that maintaining the quality of their cartilage, joints and bones is crucial, and this depends on the type of management (McIlwraith, 2005). Curiously, contrary to expectations, our results indicate that in semi-extensive studs, the prevalence of hock OC is higher. This may be due to the fact that in semiextensive studs, there is not a strong selection against this disease, as the proportion of animals competing is much lower. The different genetic origins of the individuals in the PRE population result in differences in conformation, functionality, reproductive aspects, and behaviour (Sánchez-Guerrero et al., 2019). In the PRE horse population, coat colour is an indication of the different genetic origin of those animals (Sánchez-Guerrero et al., 2019; Perdomo-González et al., 2021), as each coat colour has its origin in the influence of different breeds in the formation of the PRE breed: grey originated from animals in northern Africa, bay and black from European breeds; and chestnut from Arabian horse breeds. Although grey coat colour is the most common in the PRE horse breed, and the original population of PRE had a grey coat, the frequency of other dark colours has been increasing (Poyato-Bonilla et al., 2018), which has also contributed to the formation of the current breed. The existence of significant morphological variability between horses with different genetic origin, therefore, may be associated with a possible differential prevalence in these animals. In other populations, it was found that the body size of horses was associated with OC (Stromberg, 1979; Jeffcott, 1997, Stock et al., 2005b), as there is a predisposition to OC in rapidly growing young horses (Stromberg, 1979; Weeren et al., 1999) or heavy horses (Sandgren et al., 1993a; Sandgren et al., 1993b). Additionally, in several studies, it has been observed that horses of the ancestral line (which, due to its genetic origin, has a grey coat colour) have a higher incidence of presenting cresty neck (Sánchez et al., 2017) and obesity (Sánchez et al., 2017), which may be closely related to the insulin metabolic pathway and/or fat metabolism (Mendoza et al., 2009). Indeed, during growth, alterations in insulin metabolism may be related to the development of skeletal diseases in equines (Dobbs et al., 2011). It is important to note that heritability values can vary depending on the data selection and the genetic models analysed, especially in these types of characters which have been analysed with models with very different assumptions. Genetic models used in studies of equine OC include various methodologies such as the linear sire model (Philipsson et al., 1993); linear animal model (Winter et al., 1996; Willms et al., 1999; Stock et al., 2005a; Stock and Distl, 2006; Wittwer et al., 2007; Van Grevenhof et al., 2009b); sire threshold model (Schougaard et al., 1990; Grøndahl and Dolvik 1993; Lykkjen et al., 2014; Russell et al., 2017) and animal threshold model (Willms et al., 1999). Currently, there is no available literature that addresses all the diverse methodologies for estimating genetic parameters for OC, as presented in this study. Despite numerous studies conducted to determine the heritability of hock OC, it has never been researched with reference to its location. Moreover, very few articles have compared the different genetic models to find which one is optimal. In this study, we compared three different methodologies to find the most suitable approach for predicting the most realistic genetic parameters of OC in the hock joint (Table 4). Following the official protocol for radiograph evaluation, the distribution of the OC data can be multinomial or binomial, both of which are commonly used in parameter estimation studies and are the ones followed here. We have also added one more distribution, which is linear, considering that the severity level grows similarly between different levels and therefore can be measured linearly through a Bayesian approximation (which does not need, as in the case of the frequentist methodology, fitting to a specific distribution). The comparison of the three methodologies revealed that the binomial model exhibited a lower deviance information criterion (the best fit), and a lesser estimate of the variance components and derived genetic parameters. The Monte Carlo SEs are small, suggesting that the results obtained through this approach are consistent and reliable. This shows that reducing radiological information to ‘‘has” or ‘‘does not have” OC, from the perspective of selection against OC, is more efficient than using a seemingly more precise scale. In contrast, the linear model displayed the poorest fit, while the multinomial model occupied an intermediate position. The linear model might face difficulties due to the inherent nature of the traits being evaluated, which involve either the presence or absence, or a low number of degrees, whereas the estimates in a multinomial model are unexpectedly lower. It could also be related to the possibility that the degree to which the disease manifests itself may be influenced by non-genetic factors which are not considered in the model, such as diet or the intensity of exercise the animals undergo. This discrepancy could be also attributed to several factors, such as the small size of certain categories: with such M. Ripollés-Lobo, A. Molina, M. Novales et al. Animal 18 (2024) 101255 7 small frequencies per category, these models often tend to converge to the starting values, underestimate variances or not converge. The differentiation between classes typically applied in this type of phenotypic analysis does not align well with the genetic cause. In other words, what is clearly evident from a genetic standpoint is whether the individual has OC or not, and its severity tends to depend more on non-genetic effects, some of which are unknown and therefore not included in the model, than on differences attributable to the animal’s genetic profile. Consequently, opting for the binomial methodology for breeding evaluation provides more accurate EBVs, and therefore a higher response to selection if used to select against OC (or in favour of the most genetically resistant animals). Published heritability estimates for hock OC in other breeds with a distinctly sporty aptitude, such as the PRE population used in this study, have been lower than the heritability in PRE horses when the binomial methodology is applied. However, the heritability in PRE horses falls within the range of estimates obtained when employing the linear model. In Holsteiner horses, the heritability using Threshold models was 0.34 and 0.19 (Willms et al., 1999), while in Dutch Warmblood horses, the estimated heritability was 0.25 using a binomial model (observed scale) and 0.36 using a linear model (underlying scale) (Van Grevenhof et al., 2009b). For Hanoverian Warmblood horses, heritability estimates with a linear model ranged from 0.06 to 0.37 (Schober et al., 2003; Stock et al., 2005b; Stock and Distl, 2006). In Swedish Warmblood and South German Coldblood horses, using a linear model, heritability was estimated at 0.32 (Jönsson et al., 2011) and 0.04 (Wittwer et al., 2007), respectively. In similar studies conducted on Swedish trotters, the heritability estimates using a linear methodology ranged from 0.24 to 0.27 (Philipsson et al., 1993), while in French trotters, the heritability was 0.45 (Teyssèdre et al., 2012). Finally, recent papers have estimated genetic parameters through a linear animal model but using ssREML with genomic relationship matrices derived from genomic data. For instance, McCoy et al. (2019) studied North American Standardbred horses, yielding a heritability estimate of 0.24, while Zimmermann and Distl (2023) focused on Hanoverian Warmblood Horses, obtaining an estimate of 0.62. The specific joint studied, sample size of animals, breed, and OC calculation method (including statistical models and software) all contribute to the variation in genetic parameter estimates (Van Grevenhof et al., 2009b; Jönsson et al., 2011). It is therefore important to analyse thoroughly the external factors that can influence the occurrence of these types of developmental orthopaedic disease. The estimation of genetic correlations between different traits was carried out using multivariate models for each of the approaches (binomial, multinomial, and linear). As can be seen in Table 5, there is a very high correlation between the presence of OC in the DIRT location and the overall presence of OC in the hock joint. Additionally, the moderate positive genetic correlation between DIRT and LTT may reflect the influence of common genetic factors or genetic interactions affecting the predisposition to OC in different locations of the hock joint. However, it could be that the different locations of the hock should not be considered the same trait from a genetic standpoint. Some studies also use different locations for the study of OC (Pieramati et al.,2003; Van Grevenhof et al., 2009a; Ramos et al., 2022). Although they share influencing genes, there are genes that affect one location differentially compared to another. For instance, at the level of gene expression or even in interaction with the environment, genetic expression can vary depending on the type of exercise or stress to which the joint is subjected during movements. This matter can be highly complex, as it may also be related to the type of training or competition to which the animal is exposed, and addressing this issue without a performance control specifically designed to delve into these aspects is practically impossible. To date, there is a lack of scientific studies investigating genetic parameters at the specific locations (DIRT, LTT, MTT) of the hock joint in horses. These findings emphasise the complex genetic interactions and potential trade-offs that exist between different hock locations. While research at a similar level of depth has been conducted in Lusitano horses, genetic parameters have not been studied, possibly due to a lack of available data (Ramos et al., 2022). A small simulation was carried out, comparing the percentage of coincident animals within a specific percentile, according to their EBV, to find the repercussions of the selection of the breeders based on the result of the OC assessment in the different locations (Table 6). For practical purposes of animal selection, the most appropriate of these is the use of the overall trait of the binomial model, as it represents the best approach. At the overall level of hock OC, there is a high concurrence of animals in the binomial threshold model (1a), together with the multinomial threshold model (1b) and the linear model (2), with the percentage of coincident animals being higher in the comparison of approach 1a with 1b. These findings indicate a close agreement between the binomial and multinomial methodologies, in capturing practical information to relate to hock OC from the point of view of genetic selection. This means that using one approach or the other would have major consequences for animal selection. The selection of breeding horses with lower estimated breeding values (EBVs) for hock OC can be useful to reduce the frequency and transmission of the disease at the population level. Over time, this can lead to a decrease in cases of OC and a healthier equine population in terms of OC-related issues in the hock joint. The intensity of selection will depend on the objectives of the stud and the OC prevalence, either to select breeders or to discard them. Population-level improvement is a long-term and gradual process that requires the careful selection of breeding horses. For this reason, it is imperative to have accurate, up-to-date information regarding the genetic values of horses, as well as to conduct regular evaluations to monitor advances in reducing OC within the population. However, it is important to note that other factors not included in the models may also affect the development and severity of this disease, highlighting the need for further research, in which it is imperative to consider the type of management and feeding practices applied in livestock, with moderating caloric and energy intake in horses being crucial factors. Conclusions We found a moderate incidence (13.35%) of hock OC in the PRE horses which participated in morpho-functional tests. The highest incidence rate in specific zones of the hocks was observed in the distal intermediate ridge of the tibia (10.0%), while the lowest incidence rate was found in the medial trochlear ridges of the talus (0.3%). Genetic analyses using different approaches demonstrated a heritable component of hock OC, with heritability estimates ranging from 0.07 to 0.37 (observed scale) depending on the location and the genetic approach employed. The binomial threshold model provided the most accurate results, yielding a heritability estimate of 0.53 on the observed scale in overall OC hock. These findings strongly suggest that genetic factors play a significant role in the development of hock OC in PRE horses. Selection for OC hock, treating it as a binomial overall trait, instead of by location, would therefore produce a better response. These findings support the feasibility of reducing the genetic risk of hock OC in PRE horses through the implementation of selective breeding strategies. By selectively breeding against hock OC, M. Ripollés-Lobo, A. Molina, M. Novales et al. Animal 18 (2024) 101255 8 breeders can make significant progress in reducing the prevalence of this condition, thus improving the overall health and welfare of the breed. Ethics approval Not applicable. Data and model availability statement None of the data were deposited in an official repository. The data supporting the findings of this study are available from Real Asociación Nacional de Criadores de Caballos de Pura Raza Español (ANCCE). Restrictions apply to the availability of these data, which were used under licence for this study. The data are available from the authors with the permission of ANCCE. Declaration of Generative AI and AI-assisted technologies in the writing process During the preparation of this work the author(s) did not use any AI and AI-assisted technologies. Author ORCIDs Maria Ripollés-Lobo: https://orcid.org/0000-0002-5079-7551. Antonio Molina: https://orcid.org/0000-0002-9566-6600. Manuel Novales: https://orcid.org/0000-0002-4609-8694. Chiraz Ziadi: https://orcid.org/0000-0002-1866-7028. Eduardo Hernández: https://orcid.org/0000-0001-6420-5361. Mercedes Valera: https://orcid.org/0000-0003-1742-550X. CRediT authorship contribution statement M. Ripollés-Lobo: Writing – review & editing, Writing – original draft, Visualization, Validation, Software, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. A. Molina: Writing – review & editing, Validation, Supervision, Software, Methodology, Investigation, Conceptualization. M. Novales: Writing – review & editing, Validation, Resources, Investigation, Funding acquisition, Formal analysis, Data curation, Methodology. C. Ziadi: Validation, Software, Methodology, Investigation, Formal analysis, Data curation. E. Hernández: Investigation. M. Valera: Writing – review & editing, Validation, Supervision, Resources, Methodology, Investigation, Funding acquisition, Conceptualization. Declaration of interest The authors declare no conflict of interest. Acknowledgements We wish to thank the Pura Raza Española Horse Breeders Association for their collaboration. Financial support statement Not applicable. References ANCCE, 2024a.The Tribunals for Qualified Breeding Stock (TQB) TQB. Retrieved on 4 November 2023 from https://www.ancce.es/en/concursos/trc. ANCCE, 2024b. Rules & regulations conformation-functional competition purebred Spanish horses. Retrieved on 2 October 2023 from: https://www.ancce.es/ _docs/documentos/2022/MORFOLOGICO/reglamento_concursos22_ing.pdf. ANCCE, 2024c. Purebred Spanish Horse. Breeding Program. Retrieved on 2 October 2023 from https://www.lgancce.com/Documentacion/Normativa/Nacional/ programa_cria_en.pdf. Beckmann, S., Baltus, V., Hertsch, B., 2012. Vorkommen und Häufigkeit der Osteochondrosis dissecans bei Dülmener Wildpferden und Islandpferden. Der Praktische Tierarzt 93, 36–42. Boado, A., López-Sanromán, F.J., 2016. Prevalence and characteristics of osteochondrosis in 309 Spanish Purebred horses. The Veterinary Journal 207, 112–117. Brink, P., Smith, R.K., Tverdal, A., Dolvik, N.I., 2015. Changes in synovial fluid biomarker concentrations following arthroscopic surgery in horses with osteochondritis dissecans of the distal intermediate ridge of the tibia. American Journal of Veterinary Research 76, 599–607. Casellas, J., Piedrafita, J., 2006. Bayes factor for testing the genetic background of quantitative threshold traits. Journal of Animal Breeding and Genetics 123, 301–306. Chen, M.H., Shao, Q.M., 1999. Monte carlo estimation of Bayesian credible and HPD intervals. Journal of Computational and Graphical Statistics 8, 69–92. Clarke, K.L., Reardon, R., Russell, T., 2015. Treatment of osteochondrosis dissecans in the stifle and tarsus of juvenile thoroughbred horses. Veterinary Surgery 44, 297–303. Dempster, E.R., Lerner, I.M., 1950. Heritability of threshold characters. Genetics 35, 212–236. Denoix, J. M., Valette, J. P. 2001. Pathologie ostéo-articulaire chez le jeune cheval (incidence, évaluation clinique, facteurs de risque et conséquences). In: Proceedings of the 27 ème Journée d’étude des Haras Nationaux, 7 March 2001, Paris, France, pp. 101–113. Denoix, J.M., Jeffcott, L.B., McIlwraith, C.W., Van Weeren, P.R., 2013. A review of terminology for equine juvenile osteochondral conditions (JOCC) based on anatomical and functional considerations. The Veterinary Journal 197, 29–35. Desjardin, C., Chat, S., Gilles, M., Legendre, R., Riviere, J., Mata, X., Balliau, T., Esquerré, D., Betch, J., Schibler, L., 2014. Involvement of mitochondrial dysfunction and ER-stress in the physiopathology of equine osteochondritis dissecans (OCD). Experimental and Molecular Pathology 96, 328–338. Dik, K.J., Enzerink, E., van Weeren, P.V., 1999. Radiographic development of osteochondral abnormalities, in the hock and stifle of Dutch Warmblood foals, from age 1 to 11 months. Equine Veterinary Journal 31, 9–15. Distl, O., 2013. The genetics of equine osteochondrosis. Veterinary Journal 197, 13– 18. Dobbs, T.N., Foote, C.E., Cawdell-Smith, A.J., Anderson, S.T., Bryden, W.L., 2011. Fasting insulin status and osteochondritis dissecans in Thoroughbred yearlings. In: Cronjé, P.B. (Ed.), Recent Advances in Animal Nutrition - Australia. University of Queensland, Queensland, Australia, pp. 95–96. Donabédian, M., Fleurance, G., Perona, G., Robert, C., Lepage, O., Trillaud-Geyl, C., Leger, S., Ricard, A., Bergero, D., Martin-Rosset, W., 2006. Effect of fast vs. moderate growth rate related to nutrient intake on developmental orthopedic disease in the horse. Animal Research 55, 471–486. Douglas, J., 2003. Pathogenesis of osteochondrosis. In: Ross, M., Dysson, S. (Eds.), Diagnosis and Management of Lameness in the Horse. Saunders, Philadelphia, PA, pp. 534–543. Encina, A., Ligero, M., Sánchez-Guerrero, M.J., Rodríguez-Sainz de los Terreros, A., Bartolomé, E., Valera, M., 2023. Phenotypic and genetic study of the presence of hair whorls in Pura Raza Español horses. Animals 13, 2943. Gallo, M.A., Pimentel, L.D.O., Zoppa, A.D.V., 2013. Occurrence of osteochondritis dissecans in tibiotarsal joints of equines ‘‘Brasileiro de Hipismo” using digital radiography. Revista Brasileira De Ciência Veterinária 20, 204–207. Gee, E., Firth, E., Morel, P., Fennessy, P., Grace, N., Mogg, T., 2005. Articular/ epiphyseal osteochondrosis in Thoroughbred foals at 5 months of age: influences of growth of the foal and prenatal copper supplementation of the dam. New Zealand Veterinary Journal 53, 448–456. Geyer, C.M., 1992. Practical markov chain monte carlo. Statistical Science 7, 467– 511. Grøndahl, A.M., 1991. The incidence of osteochondrosis in the tibiotarsal joint of Norwegian Standardbred trotters. Journal of Equine Veterinary Science 11, 272– 274. Grøndahl, A.M., Dolvik, N.I., 1993. Heritability estimations of osteochondrosis in the tibiotarsal joint and of bony fragments in the palmar/plantar portion of the metacarpo- and metatarsophalangeal joints of horses. Journal of the American Veterinary Medical Association 203, 101–104. Hendrickson, E.H., Lykkjen, S., Dolvik, N.I., Olstad, K., 2018. Prevalence of osteochondral lesions in the fetlock and hock joints of Standardbred horses that survived bacterial infection before 6 months of age. BMC Veterinary Research 14, 11. Hilla, D., Distl, O., 2014a. Genetic parameters for osteoarthrosis, radiographic changes of the navicular bone and sidebone, and their correlation with osteochondrosis and osteochondral fragments in Hanoverian warmblood horses. Livestock Science 169, 19–26. Hilla, D., Distl, O., 2014b. Heritabilities and genetic correlations between fetlock, hock and stifle osteochondrosis and fetlock osteochondral fragments in Hanoverian Warmblood horses. Journal of Animal Breeding and Genetics 131, 71–81. M. Ripollés-Lobo, A. Molina, M. Novales et al. Animal 18 (2024) 101255 9