Heritabilities of health traits in different species are usually found to be low, which might reflect a large environmental influence or suboptimal recording of the traits. Moreover, many traits are expressed categorically, e.g. healthy vs. sick or clinical findings vs. no clinical findings, which together with the level of prevalence contribute to lower heritabilities on the observed scale of recording the traits. The data in this study showed that sum of clinical findings within type of examination produced heritabilities of up to 0.12 for PALP and 0.10 for HOOF findings. When defining PALP findings further into separate clinical signs and systemic locations, respectively, the heritability increased to 0.14 for effusions and to 0.13 for findings related to joints that were the dominantly occurring PALP findings
. For comparison, health and functional traits of dairy cattle, such as somatic cell count as indicator of mastitis, and female fertility, have mostly shown heritabilities of 0.01–0.14
. Despite the low heritabilities the genetic variation per se has shown to be sizeable and has led to distinctly improved breeding programs
. This suggests good opportunities for similar population improvement in PALP and HOOF health of horses if including these health traits in the genetic evaluations and selection programs. However, due to the weak and non-significant genetic correlation between PALP and HOOF, both traits need to be selected for in order to improve both traits.
The estimated heritability of joint findings is equal to the previously estimated underlying quantitative heritability of OC overall, including fetlock-, tarsal- and patellar joint findings, in the SWB
. Possibly, affected horses might coincide to some extent in these two studies, where joint effusion could be an early symptom of OC. It might be speculated that the low heritabilities of other clinical signs and systemic locations in some cases might be influenced by relatively low prevalences. However, chronic joint lesions may also have a stronger genetic background compared to e.g. acute lesions that expectably are due to environmental effects to a larger extent. The estimated heritabilities of separate hoof characteristics in the SWB (0.02–0.11) are lower than the estimates from the Dutch warmblood studbook admission inspection of hoof related traits at between 0.12–0.27
 and of hoof characteristics in Icelandic horses at 0.45
. Variations may be due to population differences in hoof variation or included hoof traits in each estimate. Possibly, Swedish veterinarians might need to be encouraged to focus more on the hoof examination than has previously been practiced. There is also a possibility that hoof characteristics are influenced by limb conformation and thus limb loading, as well as environmental factors regarding shoeing intervals etc. which may or may not be different in different populations.
The correlation between HOOF and LOCO/PALP were positive but non-significant. The possible influence of hoof quality on health status may need to be studied for more specific traits to show any significance. Furthermore, development of secondary effects of poor hoof status may increase at an older age when horses have been exposed to higher intensity of training for a longer period. It could be speculated whether the high (but non-significant) genetic correlation of MED and LOCO might be a result of an overall predisposition to impaired health that affects both results. Most phenotypic correlations were weaker than the corresponding genetic correlations, which is expected, however, this was not seen for the correlation between MED and H1, suggesting environmentally influenced factors within MED to largely influence the H1 score.
Generally, summed findings for type of examination produced higher heritabilities compared to overall scores given by the examining veterinarian. In case of the H2 score the heritability was identical to the earlier estimate of the same trait by Wallin et al.. The higher heritabilities for type of examination might be due to a higher specificity of recording, fairly standardised practice of scoring, and a less subjective evaluation, compared to assessed overall scores where several types of examinations were combined. The low heritability estimate of MED may be due to inclusion of a wide spread of non-related clinical findings, commonly represented of e.g. scars and skin wounds that are only temporarily present and highly dependent on environmental exposure. Such traits may be present in all horses at some point in life but are merely assigned to chance if affected at a particular day or not.
The recording of LOCO was almost as specific as PALP regarding recording regime and with an identical grading scale used. Still LOCO produced distinctly lower heritability estimates. Thus, LOCO results might be more influenced by environmental factors, e.g. training intensity as a cause of unsoundness. Further, differences in regime of examination between veterinarians might contribute to the environmental variation. However, R2 values revealed that fixed effects of event and gender explained only half the variation of LOCO, compared to other types of examination, suggesting LOCO to be less influenced by fixed effects than other types of examination. Possibly, efforts towards harmonisation of LOCO assessments have been more successful compared to other health examinations. This would imply that non-systematic factors, such as training intensity, stand for a substantial part of the variation in LOCO results. However, other studies have shown that the agreement between experienced veterinarians examining the same horses with mild lameness is low and that the rate of provocation during flexion tests, i.e. applied force and time, varies between veterinarians and has a significant effect on the results
[17, 18]. Accordingly, whether flexion tests are valid in the absence of standardised regimes has been debated. The heritability of lameness in 265 Norwegian Standardbred trotters was estimated to 0.25 (s.e. 0.21), when all horses were examined by the same veterinarian including flexion tests
. The estimate was based on a small number of horses and had a high standard error but might still suggest that the heritability of lameness in horses may be distinctly larger than those generated in the present study. Differences might be breed dependent, but could also reflect the importance of standardised examinations and recording regimes, perhaps aided by modern lameness diagnostics equipment, to obtain comparable information within a population. The lower heritability of hindlimb flexion test results compared to forelimbs is not due to differences in prevalence or effects on overall health. On the contrary hindlimb lesions are more common, and the effect on H2 scores was similar for flexion tests of both fore- and hindlimbs
. However, R2 values indicate that effects of event and gender had a somewhat larger impact on hindlimb flexion test results (0.16) compared to forelimbs (0.12), suggesting a larger variation among examiners for hindlimb flexion test results compared to forelimb results. Possibly, it is easier to correctly diagnose forelimb health status after flexion tests, due to differences in anatomy where present forelimb lesions are harder to compensate for compared to hindlimb lesions.
Genetic vs. environmental effects on health status
The 4–5 times larger effect of event compared to the genetic variation, indicates that further harmonisation of examination regimes would be advantageous to the RHQT as well as for practicing veterinarians. The horse industry relies on similar examination results for e.g. sales and insurance purposes, and the results should be independent of choice of examiner. The event effect also accounts for time differences regarding development of veterinary care and use of limb correction regimes, and also, horse owner awareness and thus choice of horses entered. In the present study the effect of event was corrected for, which adjusts for the present veterinary and time differences.
The large genetic variation was demonstrated by calculated additive genetic coefficients of variation and the large differences between top and bottom stallions regarding offspring health status of PALP and HOOF.
Factors contributing to genetic improvement of health traits
Prior to use in breeding, SWB stallions are examined for health status, including radiographic examination for OC and bone spavin, in connection to the stallion performance test. However, no routine genetic evaluation of health traits in the SWB has previously been practiced. Despite this absence, a consistent genetic improvement of health status, corresponding to one third of a genetic standard deviation, has occurred in the studied population between 1983 and 2001 for HOOF and PALP, respectively (Figure
3). Influencing factors for the genetic improvements may be the phenotypic stallion health inspections, and an indirect selection for longevity as life time performance is selected for. The trait for competition performance, included in the genetic evaluation, is based on accumulated competition results
. Thus, stallions with the highest performance results throughout several years obtain the best values. Another factor influencing genetic improvement for health might be that the RHQT health recordings, which have been conducted since 1973, to some extent have contributed with an informative description of individual horses’ health status to breeders. For further improvements, specific breeding values for orthopaedic health in stallions may be introduced as a complement to present breeding values for performance. Indexes could be produced either as separate evaluations of PALP and HOOF, or as a combined orthopaedic health index of PALP, HOOF and LOCO where the information from the H2 score also may be included. The ability for mares to produce healthy offspring is harder to correctly estimate as they produce less progeny at a slow rate in combination with a low heritability, resulting in low accuracy of breeding values. Nevertheless the role of mare health status should not be disregarded and in a BLUP animal model all information about relatives is also included, thus raising the accuracy in mare evaluations.