Relationship between conformation traits and gait characteristics in Pura Raza Español horses

In the breeding program of breeds such as the Pura Raza Español horse, selection by gait quality is of great interest because of their use for dressage performance. However, biokinematic analyses are expensive and data processing is time consuming. So, indirect measurements related to movement quality are alternatively used for a precocious selection of the animals. The aim of this study is to estimate the genetic correlations between 13 conformation measurements and 16 biokinematic variables at trot (4 linear, 6 temporal and 6 angular) in order to identify objective selection criteria for locomotion ability. A total of 130 Pura Raza Español horses from 24 studs, aged between 4-7 years old, were measured and their biokinematic variables were obtained in experimental conditions on a treadmill. There were 155 significant genetic correlations between conformation and biokinematic traits. Croup length was the most correlated trait with biokinematic variables at trot (16), and croup width was the least correlated one (7). Forelimb length and forelimb duration were the most correlated with conformation measurements (12), whereas minimal angle of carpus was the least correlated one (5). All the conformation measurements were genetically correlated with biokinematic variables, and through these relationships when trotting, a total of 6 body measurements were selected for the indirect and precocious selection of gait quality, which could be included directly or combined in body indices.


Introduction
The sport horse is an athletic animal whose value depends mainly on its performance in competitions.However, performance is the result of a complex combination of conformational, physiological and behavioural traits, which are heritable (Giulotto et al. 2001).
Traits included in breeding programs have to show good correlations with competition performance and they should also be possible to measure accurately early in life (Holmström et al. 1994).Conformation assesses the unalterable structure of an animal in relation to its function and it is of primary interest to breeders and owners, since overall body shape defines the limits for range of movement, the function of the horse and its ability to perform (Bakhtiari & Heshmat 2009, Rustin et al. 2009, Schroderus & Ojala 2010).Such results support the common practice of indirect performance selection via selection for functional conformation (Schröder et al. 2010).Therefore, it plays an important role in horse breeding and almost all breeding objectives for sport horses include functional-conformation and movements (Koenen et al. 2004), as an aid to improve performance in sport.In fact, gait traits have moderate to high positive correlations to dressage (Ducro et al. 2007).
Although talent can be considered as a very complex combination of more or less substitutive traits (Borowska et al. 2011), conformation traits in sport horses are not difficult to define and evaluate (Posta et al. 2007), and are used in indirect selection for performance traits, since most performance variables have low levels of heritability and can be measured only late in life (Koenen et al. 1995).The efficiency of indirect selection for performance depends on the genetic variation of conformation traits and on the genetic correlations between conformation and performance variables (Koenen et al. 1995).
In the Pura Raza Español horse breeding program, although selection by gait quality is of great interest, biokinematic analyses are very expensive.Therefore, indirect measurements related to gait quality would allow some cost saving, and it is used instead for a precocious selection of the animals.Thus, the genetic correlations between conformation measurements and biokinematic variables when trotting were estimated, in order to identify objective selection criteria for locomotive ability.

Material and methods
A total of 130 Pura Raza Español horse males from 24 different studs, chosen randomly, registered in the official stud-book, were evaluated at the Laboratory of Equine Performance Control (Veterinary Faculty of Cordoba, Spain) for conformation and biokinematic variables.Their age ranged from 4-7 years old (4.6 ± 1.5) and they were selected in order to be representative of most of the genetic lines of the Pura Raza Español horse population, with an average inbreeding of 0.9 and coancestry of 0.05.Because of the complex methodology applied for the estimation of biokinematic variables in this work (with high spend of time and money), the reduced number of animals used for the estimation of genetic parameters is justified as in other published equine papers, with a number of animals ranging between 100 and 362 (Rivero et al. 1996, Barrey et al. 1999, Rivero & Barrey 2001, Górecka et al. 2006).

Conformation measurements
Conformation analysis was carried out through quantification of the main body mea surements (Figure 1), following the methodology described by Cervantes et al. (2009).A total of 13 conformation measurements were included, instead of subjective evaluations, because these would improve the traditional judgement procedure, increasing the accuracy of the prediction of performance potential (Holmström et al. 1994), since they could be used as a descriptive tool (Barrey et al. 2002) and are more repeatable.

Biokinematic variables
All variables were recorded using a camcorder while horses were trotting on a treadmill at the constant speed of 4 m/s, following the methodology described by Valera et al. (2008).Adhesive markers were attached at pre-defined skeletal reference points which were easily identifiable and representative of the joints and radii under investigation (Figure 1).
A total of 16 biokinematic variables at trot (4 linear, 6 temporal and 6 angular) were analysed.All of them were selected because of their relationship with dressage ability.
The linear variables were: forelimb and hindlimb length (because of their importance in the »overtracking« or »overreach« length in the trot, which is a desirable feature in Dressage), longer forelimb and hindlimb stride lengths are associated with lower stride frequencies, which are desirable in Dressage, according to Merz & Knopfhart (1996), and forelimb and hindlimb maximum height of hoof (because the Pura Raza Español horses exhibited elevated movements rather than extended movements of the limbs).
The temporal variables were: forelimb and hindlimb duration, forelimb and hindlimb stance phase duration (according to Holmström et al. [1994], horses judged as good at trot had longer stance phase duration compared to the poor horses and the stance phase duration increased with increased collection) and forelimb and hindlimb swing phase duration (because elite horses typically exhibited shorter stance durations in both fore and hindlimbs, which results according to Drevemo et al. [1980] on a longer swing phase duration).
Finally, the angular variables were: minimal angle of carpus, stifle and tarsus, minimal retraction-protraction angle of hindlimb, maximal retraction-protraction angle of forelimb, (because of their importance in dressage performance they could be characteristics to be included in the breeding program of the Pura Raza Español horse), and maximal angle of pelvis (which represents the maximum angle of the croup with respect to the horizontal when the horse is moving).All this angles are important to get the movement described in the official breed standard: agile, high, extensive, harmonic and rhythmic, with a particular predisposition for collection and turns on haunches (Valera et al. 2009).Markers are placed: 1: withers, 2: tuber of the spine of the scapula, 3: greater tubercle of the humerus (caudal part), 4: lateral collateral ligament of the elbow joint, 5: lateral styloid process of the radius, 6: base of the 4th metacarpal bone, 7: lateral collateral ligament of the fore fetlock joint, 8: coronet of the fore hoof (over the pastern axis), 9: tuber coxae, 10: greater trochanter of the femur (caudal part), 11: lateral collateral ligament of the stifle joint, 12: lateral malleolus of the tibia, 13: base of the 4th metatarsal bone, 14: lateral collateral ligament of the hind fetlock joint, 15: coronet of the hind hoof (over the pastern axis) Figure 1 Graphical representation of the conformation measurements taken in the Pura Raza Español horse

Genetic and statistical analysis
A preliminary study of the phenotypic relationships between the analysed traits was carried out by making a factor analysis, using Statistica for Windows (StatSoft, Inc.Tulsa, OK, USA, version 8.0).The genetic correlations were estimated by VCE v 6.1 (Groeneveld et al. 2010), using a mutivariate mixed animal model.For the genetic analysis, the general model for this analysis was: where Y is a vector of animal observations, β is the fixed effects vector (stud-season of evaluation, age of the animal), associated with the incidence matrix X, α is the vector of direct genetic effects, associated with the incidence matrix Z 1 , and e is the random error effects matrix.
To complete the pedigree for the calculation of the inverse of the relationship matrix, the Pura Raza Español horse stud-book was used, and all the registered ancestors of the recorded animals were added until the fourth generation, making a total figure of 1 704 animals.
The additive genetic variance and covariance of the traits were estimated according to a Restricted Maximum Likelihood procedure (REML), using a Quasi-Newton algorithm with exact derivatives to maximise the log likelihood.An approximate standard error of the genetic correlations was estimated from the inverse of the approximation of the Hessian matrix when convergence was reached (Groeneveld et al. 2010).

Results and discussion
The assessment of a horse's merits by virtue of its conformation is as ancient as man's usage of the species.Conformation traits remain an interesting subject, because they are linked to desirable characteristics for breeders of performance and soundness (Bakhtiari & Heshmat 2009).Efficiency of horses is the main demand in all breeds with whatever purpose of use (Halo et al. 2008).Therefore, in sport horses, the objective evaluation of conformation and its relation to performance is of great importance (Moore 2010), and insufficient knowledge of the influence of conformation on performance and health can result in inaccurate selection.The breeding objective must be focused on the conformation traits (Jakubec et al. 2009), after all, the final aim of breeding programmes is a horse with certain conformation characteristics which stands out for its performance in sport (Belloy & Bathe 1996).
Nevertheless, the ideal conformation does not exist, because one conformation trait could be both advantageous for a certain locomotion characteristic and detrimental for others (Back et al. 1996).
When conformation measurements were analysed in the Pura Raza Español horse popula tion (Table 1), the means obtained were similar to those reported in the same breed in previous analyses (Molina et al. 1999, Gómez et al. 2009) or in other breeds used for dressage performance, such as Lipizzan (Zechner et al. 2001) and Lusitano horses (Güedes 2008).
Descriptive statistics of the 13 conformation measurements are shown in Table 1.In general, their level of variation was medium to low, with coefficients of variation ranging between 2.3 % (croup height) and 9.3 % (croup width).
The coefficients of variation obtained were of a medium to low level (all of them lower than 10 %).Similar results were shown in the same breed and in other selected breeds (Molina et al. 1999, Zechner et al. 2001, Güedes 2008, Bakhtiari & Heshmat 2009).Therefore, we concluded that the analysed population is sufficiently homogeneous for these characters.
Consequently, we aim to detect conformation measurements that are good indicators of locomotive and gaits quality.Specific characteristics of trotting and canter are required for dressage, and so could be selected genetically and contribute to performance.Although the gait and conformation tests could be applied in breeding programs to detect more accurately young horses with good dressage performance (Barrey et al. 2002), its routine application is very expensive and the data processing takes a long time.
The present work aimed to estimate correlations between conformation measurements and biokinematic variables at trot (Table 2).Trotting quality (for dressage) is determined mainly by the amplitude of limb movements, the elasticity and a marked phase of suspension (Moore 2010).So, it is not a surprise that horse conformation conditioned locomotion ability (Güedes 2008).This relationship between conformation and function is a constant in physical issues.The measurements are expressed in cm and the coefficients of variation in %.
The 155 significant genetic correlations between conformation measurements and bio kinematic variables at trot (74.5 % of the total number of estimated correlations), are shown in Table 2 -43.2 % of these were negative, most of them (49.2 %) with angular variables.The highest genetic correlation was 0.70 (maximal angle of pelvis with neck-body perimeter and with thorax perimeter), and the lowest (absolute value) was between hindlimb stance phase duration and knee perimeter (0.02).Only 10.32 % of them were higher than or equal to 0.50 (absolute value).
Croup length was the most correlated measurement with biokinematic variables at trot (16 genetic correlations), and croup width was the least correlated one (7).Forelimb length and forelimb duration were the variables most correlated with conformation measurements in this analysis (12), whereas the minimal angle of carpus was the least correlated one (5).
Croup length has been the trait which correlated most closely with all the biokinematic variables analysed.Previous papers have shown the importance of the croup, for example in the Spanish Arab horse, for »size« analysis, and the most significant differences between morphological and endurance aptitude were observed in the posterior triangle (Cervantes ).According to Koenen et al. (1995), a long, steep croup shows a very close correlation with trotting characteristics.Güedes (2008) showed that the croup length is a trait which correlates very closely with biokinematic variables at trot in Lusitano horses, with an important negative correlation with the maximum retraction angle of hindlimb and maximum protraction angle of hindlimb.In Pura Raza Español horse, croup length is associated with angles and temporal traits at trot.A total of 43.7 % of the genetic correlations obtained for this trait have been negative, mainly with angular traits.Back et al. (1996) reported that as this angle was smaller, more of it was tucked under the trunk of the hindlimb, which is conducive to concentration of gait.Clayton (2001) also considered that the pelvis should be nearer to the horizontal in dressage horses.Croup width was the lowest correlated trait with biokinematic variables at trot (6: 1 linear, 1 temporal and 4 angular traits).
As regards withers height, different results have been shown in previous papers.Some analyses have shown a close correlation between withers height and performance or loco mo tion problems (Magnusson & Thafvelin 1985, Baban et al. 2009), whereas Dusek et al. (1970) affirmed that withers height was not correlated to stride length for different gaits.Galisteo et al. (1998) obtained positive correlations between withers height and stride length and overtracking; and they also record a moderate influence of withers height on angular parameters while trotting, without there being any temporal ones.Our results differ from those of the previous authors, because withers height correlated with most of the biokinematic variables (14: 6 temporal, 4 linear and 4 angular variables), including all the temporal and linear ones.
Croup height has similar genetic correlations with the biokinematic variables to withers height (same sign and similar values), except for those in the hindlimb (hindlimb duration and length; stance and swing phase duration).This could be caused by the close phenotypic correlation between both traits, the highest between all the conformation measurements included in this study (0.80, results not shown).The differences between them are related to the changes in hindlimb function due to changes in the relative measurements and angles (more influenced by croup height).
The head and neck determine athletic ability (Lawrence 2001), back movement and stride characteristics at trot, as well as stride length (Rhodin et al. 2005).In this regard, Holmström et al. (2001) suggested that good head-neck and neck-body insertion are more important than neck length for dressage ability.Lawrence (2001) also affirmed that the head-neck connection must be favourable to achieve free movement and flexion.Two conformation variables were analysed to illustrate these two insertions: head-neck perimeter and neck-body perimeter, both of which are correlated with biokinematic variables at trot (11 and 9 genetic correlations, respectively), ranging between 0.07 and 0.70 (1 correlation equal to or above 0.50, for both traits).
If the neck acts as a lever, head length acts as a counterweight.A total of 9 genetic cor rela tions were significant, 3 of them are equal to or over 0.50 in absolute values.The closest cor re la tions were with: forelimb stance phase duration (0.65), forelimb swing phase duration (0.61) and minimal angle of tarsus (−0.54).
Thorax perimeter, thorax depth and chest width have shown a large number of mediumrange genetic correlations with biokinematic variables at trot (13, 11 and 13, respectively), all of which had similar values and signs, except for the hindlimb variables at trot.This could be explained because the horse's forelimbs are attached to the trunk by a strong muscular belt (no joints) and therefore the impact that traits like trunk width or trunk perimeter can have over the biokinematic traits of the forelimb.Finally, conformation measurements analysed in the limbs, knee and hock perimeter, correlated with most of the biokinematic variables at trot (15 and 14, respectively).Both of them correlated with all the linear and temporal variables analysed, and some differences were observed in angular traits.Although they have similar signs and values, the very close correlation between knee perimeter and maximal retractionprotraction angle of forelimb (0.53), and between hock perimeter and minimal angle of tarsus (−0.61) is remarkable.
In addition to this, stifle angle has been considered as an important variable for gait quality.A large opening stifle angle causes a significant constriction of the quadriceps in the thigh, which is probably the most overworked muscle in collected gaits.The inability of the quadriceps to support the maximum weight makes the horse shift the burden onto the forelimb, thus altering the balance (Holmström 2001).Magnusson & Thafvelin (1985) found a positive correlation between stifle angle and sports results.This could be caused by changes in maximum retraction-protraction range with a more upright pelvis and lower angles of the knee joint (Back et al. 1996).In this work, the stifle angle correlates negatively with most of the analysed traits, except croup width, neck-body perimeter and chest width.
The analysis of body measurements allows us to describe an animal or breed's conformation and to detect conformation traits that identify locomotion quality.Barrey et al. (2002) affirmed that, although conformation by itself can not explain the ability for dressage performance, differences in conformation can be responsible for some locomotion characteristics.The importance of the locomotor pattern is related to the fact that for each type of exercise, the horse uses a specific type of locomotion, where its individual characteristics determine the level of performance it can achieve (Leleu et al. 2005).
The factor analysis for the 13 body measurements and the 16 biokinematic variables at trot (Figure 2) showed that Factor 1 separates durations and limb length from the others.It also separates minimal retraction-protraction angle of the hindlimb and maximal retractionpro trac tion angle of the forelimb, whereas Factor 2 separates the temporal and linear variables (including the conformation measurements) from the other traits.
Temporal and linear traits were related between them (Factors 1 and 2), and with the conformation traits analysed (Factor 2), whereas angular traits measured in the distal area of the limbs were not related with conformation measurements (Factor 2).Therefore, the factorial analysis indicated that the length of body regions influence linear and temporal parameters for trotting more than angular parameters in the Pura Raza Español horses.
In conclusion, most of the analysed body measurements are genetically correlated with some biokinematic variables at trot.Therefore, their inclusion in the breeding programme of a breed, such as the Pura Raza Español horse, is recommended.This ensures the implementation of an indirect and precocious selection of the animals based on the objective conformation measurements proposed in this study, thus producing a suitable response.According to our results, the relationships of conformation traits between each other and with biokinematic variables while trotting show that it is important to study withers height, croup length, croup width, knee perimeter, hock perimeter and thorax perimeter in order to make an indirect and precocious selection of gait quality in Pura Raza Español horses.These could be included directly or combined in body indices.Factor 1 Figure 2 Factor loadings for 29 selected variables (13 body measurements and 16 variables biokinematic at trot) analysed in 130 Pura Raza Español horses

Table 1
Descriptive statistics of 13 body measurements in 130 representative Pura Raza Español horses 1 Significant genetic correlations (with standard errors) between 13 body measurements and 16 biokinematic variables at trot in 130 representative Pura Raza Español horses 2009