E . Gerber Olsson et al. Livestock Production Science 65 2000 81 –89
85 Table 4
Y 5 m 1 year of birth 1 stallion 1 e Model 3
ij i
ij ij
2
Heritabilities h , standard errors S.E. and repeatabilities t of traits scored in the stallion performance test
2
where Y is the score of each trait for the ijth
ij
Trait h S.E.
t S.E.
stallion; m is the population mean; year of birth is
j
Walk 0.46 0.13
0.77 0.03
the fixed effect of ith year of birth i 5 82, 83 . . . ,
Trot 0.37 0.11
0.77 0.03
89; stallion is the random effect of the breeding
Gallop 0.39 0.11
0.75 0.03
ij
value of the ijth tested stallion, mean50 and addi-
Free jumping 0.47 0.13
0.58 0.04
a
2
Temp. free jumping 0.23 0.14
0.38 0.07
tive genetic variance5 As ; e is the random re-
a ij
2
Jumping under rider 0.32 0.14
0.47 0.06
sidual effect, mean50 and variance5s .
e
Temp. jumping under rider 0.33 0.23
0.43 0.09
In model 1 an age effect was considered, but not
a
Temp., temperament and general appearance.
in model 3 as the traits in the latter case consisted of average values of several observations measured at
different ages. It was assumed that the environmental correlation between the SPT and RHQT traits was
the heritability estimates for gaits and jumping zero, as the traits were recorded on different in-
estimated by Bruns et al. 1985, Huizinga et al. dividuals and at different occasions. Estimates of the
1991a,b, von Velsen-Zerweck 1998 and Friemel sampling variance was not obtainable by this version
et al. 1998. of the derivative free REML programme.
The heritability estimates for temperament and general appearance for free jumping were lower than
the corresponding estimate for free jumping ability. The explanation for this may be that the tempera-
4. Results and discussion
ment and general appearance traits are more difficult to assess, but also that they were scored on a smaller
4.1. Stallion performance test data number of horses, and thus subjected to larger errors.
It was not possible to estimate heritabilities, re- 4.1.1. Means and standard deviations
peatabilities and correlations for the trait tempera- Table 2 shows that the scores for individual gaits
ment for gaits, because of too few observations. generally had a relatively lower variation than the
The highest repeatabilities were found between the scores for jumping. Among these, free jumping
gaits, 0.75–0.77. Free jumping was the trait within scores showed the largest variation. For gaits the
jumping that showed the highest repeatability, 0.58. scale was not fully used. However, a score of 5
The temperament traits for free jumping and jumping means that the stallion has an acceptable ‘clean’ gait,
under rider had a lower repeatability than the corre- while lower values indicate lameness or some ir-
sponding jumping trait, but none of the traits showed regularities. For jumping the whole scale was better
a low repeatability. The high repeatabilities for gaits used and the scores showed a normal distribution.
show that the number of tests for gaits can be reduced without affecting the accuracy in the judge-
4.1.2. Heritabilities and repeatability ment. The lower repeatability for the temperament
The gaits generally had high heritabilities as well traits indicates that these are more difficult to assess
as repeatabilities Table 4. Jumping under rider had in a correct way at a single occasion. These traits had
a lower heritability than free jumping. The reason for less observations which may affect the accuracy of
this may be that the riders influence the horses so the results, though, the estimates of repeatability
that the natural variation among horses is not ex- cannot be compared to repeatabilities estimated in
pressed as much. Furthermore, some of the poorly other countries because the repeatabilities in this
free jumping horses did not participate in jumping study are repeatabilities between independent tests at
under rider. Thus, a certain amount of pre-selection different testing periods and not between repeated
took place which also may cause the lower value. tests within the same testing period, as has been
The heritabilities in this study are within the range of common in other studies.
86 E
. Gerber Olsson et al. Livestock Production Science 65 2000 81 –89
4.1.3. Genetic and phenotypic correlations variation. The average conformation score had the
Positive genetic correlations were found between highest mean and the lowest coefficient of variation.
all gaits, the highest between trot and gallop, 0.71 Table 5. These results are in agreement with Bruns
4.2.2. Heritabilities et al. 1985, Huizinga et al. 1991a and Christman
Table 6 contains the heritabilities for traits in the and Bruns 1997, who got genetic correlations in
SPT and RHQT data. The highest heritabilities were the range of 0.5 to 0.9 between the individual gaits.
estimated for the traits scored in the SPT. The main Gallop was quite naturally also the gait most associ-
reasons for this are that in the SPT generally the ated with jumping results and more highly correlated
same very experienced judges are used in this test for to jumping under rider than to free jumping. Huiz-
several years, and the stallions are judged at the inga et al. 1991a got similar results, although
same place. Furthermore the judges, as regards the somewhat lower correlations for gallop and show
gaits, have the opportunity to judge the stallions jumping and free jumping, respectively. Jumping
twice during consecutive days before the scores are under rider was highly correlated to free jumping
set. The heritabilities calculated for traits in the SPT 0.93, which indicates that they to a great extent are
in this analysis were in the same range 0.43 Table governed by the same genes. The genetic correla-
6 as when the traits were analysed as individual tions between the four jumping scores were in
traits Table 4. The explanation for this could be general high 0.44–0.99, but somewhat lower with
that the model is not accounting for the permanent temperament and general appearance for jumping
environmental effect that affects the results of the under rider. This may partly be explained by larger
stallion in the SPT, which may contribute to a standard errors as the number of observations for this
greater heritability.
On the
other hand
the trait was smaller. The phenotypic correlations among
heritabilities of means of repeated measures should the jumping traits 0.43–0.89 were slightly lower
than the genetic ones.
Table 6
4.2. Stallion performance test and riding horse
Heritabilities diagonal, genetic above the diagonal and pheno-
quality test data
typic correlations under the diagonal estimated for the mean of traits scored in the stallion performance test SPT and in the
riding horse quality test RHQT
4.2.1. Means and standard deviations Table 3 shows the means and standard deviations
Trait average score 1
2 3
4 5
for the traits in data from SPT and RHQT. For the
1 Gaits RHQT 0.31
1.00 0.35
stallions, jumping had a lower mean and a greater
2 Jumping RHQT 0.17
0.26 1.00
3 Conformation RHQT 0.27
0.75 0.07
coefficient of variation than the gaits. In the RHQT,
4 Gaits SPT 0.29
0.16 0.30
0.43
jumping had a higher mean than the gaits and, as in
5 Jumping SPT 0.13
0.20 0.08
0.43
the SPT-data, also showed a higher coefficient of
Table 5 Genetic correlations with standard errors above the diagonal and phenotypic correlations under the diagonal for stallion performance test
data. Standard errors within brackets Trait
1 2
3 4
5 6
7 1 Walk
0.40 0.08 0.30 0.06
0.27 0.07 0.21 0.06
2 Trot 0.40
0.71 0.16 0.24 0.06
0.14 0.05 3 Gallop
0.36 0.67
0.40 0.10 0.54 0.18
4 Free jumping 0.14
0.22 0.31
0.99.03 0.93 0.23
0.44 0.14
a
5 Temp. free jumping 0.89
0.79 0.19 0.50 0.26
6 Jumping under rider 0.11
0.29 0.37
0.57 0.51
0.90 0.06
a
7 Temp. jumping u.r. 0.43
0.44 0.84
a
Temp, temperament and general appearance; jumping u.r., jumping under rider.
E . Gerber Olsson et al. Livestock Production Science 65 2000 81 –89
87
be higher, though the trait was not always the same, competition for show jumping, 0.90. The corre-
e.g. gaits recorded as individual gaits or as means for sponding genetic correlation for dressage was 0.68.
the three scored traits. Also, between gaits in the SPT and conformation
The heritability for the traits in the RHQT are in the RHQT the genetic correlation was high 0.75.
´ similar to the heritabilities estimated by Arnason
This can partly be explained by the fact that walk 1993. The author estimated the heritabilities for
and trot at hand are two of the five traits scored in gaits and temperament at the gait test to 0.35 and
the conformation evaluation. It is also clear that the 0.27, respectively, and in this study the average
genetic correlation between conformation in the value of gaits and temperament and general appear-
RHQT and jumping in the SPT is low 0.07. Gaits ance for gaits was 0.31. The genetic correlation
are positively associated with jumping traits, though between the two former traits were 0.84 according to
at a modest level, 0.26–0.35. ´
Arnason 1993, which may explain the high In this analysis it was assumed that the environ-
heritability for the average trait gaits in this analysis. mental correlation between SPT and RHQT is zero.
For conformation and jumping the heritability was This assumption may not be completely correct as
also very similar between the results in this study performance test results of the stallions may affect
´ and the one by Arnason 1993 which were 0.29 for
how their offspring are trained and the judges in the conformation and for jumping and temperament for
RHQT may unconsciously consider this when judg- jumping, 0.20 and 0.14, respectively. The heritability
ing. This fact may cause the correlations in Table 6 for gaits in the RHQT are similar to the ones
to be somewhat overestimated. The judges in the ¨
estimated by Kuhl 1991 and von Velsen-Zerweck RHQT are all participating in a regular training
1998, but for jumping the authors got a higher programme where the judges of the SPT are the
heritability, 0.37 and 0.35, respectively. The explana- instructors, which also can be a part of an explana-
tion for this could be that the results are based on tion for the high correlations.
15–50 days stationary tests and 1-day field tests. 4.3. Aspects on testing and selection
4.2.3. Genetic and phenotypic correlations The genetic correlations were estimated to unity
Analysis of data from the Swedish stallion per- between gaits scored in the SPT and in RHQT and
formance testing scheme show moderately high between jumping scored in the SPT and in the
heritabilities for all recorded traits, and it can be RHQT Table 6. Therefore, it can be assumed that
concluded that the heritabilities of the traits are high the same gene complex affects the traits judged in
enough for an effective, selection. Then having the two different tests, although they are scored with
estimated the heritabilities it is possible to estimate different heritabilities. Correlations between sire and
breeding values for the stallions in the SPT and to progeny tests were also analysed in Germany by
use all available pedigree information in a BLUP- Schade 1996 and von Velsen-Zerweck 1998. The
evaluation. To make the breeding values more tests are primarily made at the age of 3.5 years. The
accurate, the goal should be to include the results of former author found high genetic correlations 0.83–
both SPT and RHQT data in the breeding value 0.96 between traits scored at stallion performance
estimations of stallions at the performance tests as tests and field tests for mares, while the latter came
the genetic correlations between the results of these to the conclusion that traits at performance tests for
tests are very high. A complicating factor is that a stallions and mares cannot be seen as exactly the
number of the stallions prospects at the tests are same. However, the genetic correlations were high
foreign bred and thus largely lack pedigree infor- and ranged from 0.74 to 0.90. Brockmann and Bruns
mation from the two types of Swedish tests. 1997 got a genetic correlation between results at
Jumping traits showed the largest variation, while stallion performance test and mare performance tests
individual gait scores showed the highest re- which was 0.68. van Veldhuizen 1997 found a high
peatabilities. For a given accuracy in assessing genetic correlation between results for jumping in
expected breeding values fewer tests at the age of stallion performance tests and offspring results in
four are required for evaluation of the gaits under
88 E
. Gerber Olsson et al. Livestock Production Science 65 2000 81 –89
rider than for jumping traits. The stallion perform- selection criterion. Positive genetic correlations were
ance tests could therefore be shortened and then also found between gaits and jumping traits, thus breed-
less costly. Free jumping test results, which can be ing for both characteristics is facilitated. The genetic
done at a lower age than jumping under rider, are correlations between the stallion performance test
highly correlated to results under rider, and because and the riding horse quality test are very high and
of their higher heritability very suitable for selection therefore similarly defined traits in the two different
purposes. However, an argument of ‘the riders and tests can be regarded as virtually the same genetic
the owners of the stallions’ is that the stallions must traits.
be jumped under rider because that is how they are going to be used later, and coincides with the
breeding objective. It is therefore difficult to get
References
across the advantages of free jumping from a selec-
´
tion point of view, and in the shortened stallion
Arnason, Th., 1993. In: Rapport om genetiska analyser av ¨
˚ ¨
kvalitetsbedomningar av svenska fyrariga ridhastar 1973–1993,
performance test introduced in 1997 free jumping
˚
IHBC AB, Knubbo, Morgongava, Sweden, p. 25.
was excluded. An important finding from both sets
´ Arnason, Th., Philipsson, G., Philipsson, J., 1997. In: Rapport om
of data is that there are positive genetic and pheno-
¨ ˚
¨ BLUP-avelsvardering baserad pa kvalitetsbedomningsresultat
typic correlations between gaits and jumping traits,
¨ ˚
¨ for fyraariga ridhastar 1973–1996, IHBC AB, Knubbo, Mor-
˚
although at moderate levels, but breeding for both
gongava, Sweden, p. 19. ¨
¨ Bade, B., 1974. Schatzung genetischer Parameter fur Leis-
characteristics within the population is facilitated.
¨ tungsmerkmale hannoverscher Reitpferde, Diss, Gottingen.
Although results from 1-day field tests of 4-year-
Brockmann, A., Bruns, E., 1997. Estimation of genetic parameters
old horses are less accurately assessed, the very high
with data from competitions for young dressage horses. In:
genetic correlations with the corresponding trait
48th Annual Meeting of European Association for Animal
evaluated at stallion performance tests, prove that the
Production, Vienna, Austria, 25–28 August, p. 6. Bruns, E., Rauls, B., Bade, B., 1985. Die Entwicklung von
field tests can be effectively used for early progeny
¨ ¨
Selektionskriterien fur die Reitpferdezucht. Zuchtungskunde 57
testing of the stallions, thereby further improve the
3, 172–182.
accuracy in selection of stallions for performance
Christman, L., Bruns, E., 1997. Estimating breeding values based
traits.
on evaluations of young Hanoverian mares. In: 48th Annual
It would be interesting in a future study to
Meeting of European Association for Animal Production, Vienna, Austria, 25–28 August, p. 4.
estimate the correlation between stallion performance
Darenius, A., Philipsson, J., Fredricson, I., Thafvelin, B., Berg-
test results and later competition performance, and to
˚
sten, G., Radberg, L., Elowson-Anda, E., 1982. Kvalitet-
include both results from the SPT and competitions
¨ ¨
¨ ˚
¨ sbedomning av unga ridhastars halsotillstand, exterior och
in the annual BLUP breeding evaluation, presently
¨ ¨ ridegenskaper. In: Sveriges Veterinarforbund, Forsellsymposiet,
based only on RHQT data.
Stockholm, 9–11 Sept. ¨
Friemel, G., Rohe, R., Kalm, E., 1998. Study to modifications of stallion performance test on station. In: 49th Annual Meeting
of European Association for Animal Production, Warsaw,
5. Conclusions
