Arabian horse race is one of the most important warm-blooded horse races. Arabian
horses get attention due to the harmony of morphological structure of their
body. They have small head, large and dry eyes and small and upright ears. Their
foreheads are broad and they have generally smooth profile. Hip heights of these
horses usually range from 145 to 160 cm. They have soft, thin skin and smooth,
shiny and short hairs. Their coat colors are usually Chestnuts (copper color),
grey or bay. Dark (horse) coat is rare. During gallop, their tails stand up,
making a curve which is called as hold queue. Arabian horses have solid bodies
and have different good abilities such as ride, freightage and racing. The heritabilitys
of these characters are high. These horses get mature by the age of four. Turk-Arabian
horse breeding in Turkey is conducted by the Turkish government in Bursa, Eskişehir
and Malatya. Those horses are grown and bred since, mid times of ottoman Empire
by a Anatolian farm located in Malatya. Horse breeding in the farm is an example
for the region and other parts of Turkey and is considered as a horse breeding
center of Turkey. Officially, horse races in Turkey are organized by Turkish
Jockey Club (TJK) in İstanbul, Ankara, Bursa, İzmir and Şanlıurfa
(Ekiz et al., 2005a).
The racing performance of horses is affected by environmental factors such
as hippodrome track (sand or turf), distance and year. Sand makes racing condition
hard, thus the races held in sand track are more difficult for horses than the
races held in turf track. But sand track requires more power and endurance (Koseman,
Determination of the environmental effects and the estimation of genetic parameter
effecting race performance is important in optimization of selection programs.
The heritability values of racing performance parameters usually range from
low to high and the repeatability are usually medium and high (Ekiz
et al., 2005a).
Restricted maximum likelihood (RELM) method has been extensively used to estimate
genetic parameters and variance components. The REML method has the most desirable
statistical properties for estimation of variance components (Ekiz
and Kocak, 2007; Ekiz et al., 2005a, b;
Bakhtiari and Kashan, 2009).
The reported heritability estimates in the literature for racing time generally
ranged from low to moderate (Ekiz and Kocak, 2007; Ekiz
et al., 2005a, b; Bakhtiari
and Kashan, 2009; Mota, 2006; Koseman,
2005). Ekiz et al. (2005b) reported moderate
to high heritability estimates of 0.465, 0.293 and 0.359 for racing time of
Thoroughbred horses at dirt, turf and combination of both tracks. Ekiz
and Kocak (2007) estimated heritability for Thoroughbred horses to be 0.177
to 0.353 and 0.289 to 0.404, respectively depending on the racing distances
1200 to 2400 m. Bakhtiari and Kashan (2009) estimated
heritability for Iranian Thoroughbred horses to be 0.13, 0.11 and 0.09 for distances
1000, 1400 and +1600 m, respectively.
These results indicated that a moderate level of genetic progress is possible
for racing time if selection is based on the phenotypic value of the horses.
Race finishing time is a direct measure of speed and is regarded as the best
method of the evaluation of race performance of horses (Ekiz
and Kocak, 2005b; Oki et al., 1995; Mota
et al., 2005).
Mota et al. (2005) reported the observed lowest
and highest heritability 0.04 and 0.29 at 1500 and 1000 m races for racing time,
respectively. And also lowest and highest repeatability were given 0.19 and
0.63 for the same distances, respectively. Bakhtiari and
Kashan (2009) pointed out the repeatability estimates of racing time for
1000, 1400 and +1600 m were 0.26, 0.19 and 0.17, respectively.
The objective of this study was to investigate environmental effects that influence
racing performance with using two statistical mathematics models and to estimate
genetic parameters over various distances.
MATERIALS AND METHODS
Open and group races finishing time data from 1200, 1300, 1400, 1500, 1600,
1800, 1900, 2000, 2100, 2200 and 2400 m races belonging to years at 1996-2008
of the Arabian horses, grown by Anatolian Agricultural Institution, were obtained
from Turkey Jokey Club records and used in this study.
The 3184 race records of 437 offspring of 32 horses were used in this study.
Factor affecting race finishing time (racing time) and race speed such as year,
the hippodrome (City), track, distance were evaluated. Two different models
were used in the evaluation of research data. In model-1, year, hippodrome,
track and race distance were included in the model affecting for racing time
and race speed. In model-2, race distance factor were dropped the model and
effects of other factors on racing time for each distance were investigated.
Mathematical models used in the analysis:
where, Yijkl(m) is racing time or race speed, μ is population
mean, ai is ith. effect of year, bj is jth. effect of
hippodrome, ck is kth. effect of track, dl is l. effect
of race distance and eijkl(m) is random experimental error (normally
distributed with mean zero and component of variance σ2e
For each model, to investigate the significance levels of factors on racing
time and race speed, Analysis of Variance (ANOVA) was conducted using the GLM
procedure of SAS package (Orhan et al., 2004).
Variance components and genetic parameters were estimated according to REML
method by DFREML package (Meyer, 1997).
RESULTS AND DISCUSSION
Descriptive statistics and significant levels of racing time and race speed
by factors included in the model-1 are given in Table 1. In
terms of racing time and race speed, the effects of year, hippodrome, distance
and track were found significantly (p<0.01) (Table 1).
There are 2 types of racing tracks. In competitions, racing time was shorter
and race speed was higher in turf tracks than dirt tracks. And also, racing
time and race speed were both affected by hippodromes. The highest race speeds
were observed in Istanbul and Ankara hippodrome, while the lowest race speeds
were obtained in Şanlıurfa hippodrome.
Descriptive statistics and significant levels of racing time by factors included
in the model-2 were shown in Table 2. In the overall evaluation
of 13 year period, it was observed that race speed increased and racing time
In the separate analysis for each distance, the effects of the year and hippodrome
were significant (p<0.01) for each distance and track effect was significant
for all distances (p<0.01) except for 1500 m. Due to insufficient data, evaluations
couldnt be conducted for 1700 and 2800 m records distances (Table
The variance components and genetic parameter estimates obtained for different
race distances were shown in Table 3. The highest heritability
for racing time was obtained from 1300 m (11.0%) while the lowest heritability
was observed at 2200 m (4.0%). For overall race distances, the heritability
was estimated as 5.0% which is considered to be low. Heritability and repeatability
estimates tended to decrease with increasing distance, this indicating that
selection based on racing time will be less successful.
In racing time, the highest repeatability (71.0%) was obtained in the 2200
m and the lowest repeatability (23.0%) was obtained in 1500 m. For overall race
distances, racing time repeatability was estimated as 35.0% and interpreted
as mid-high level.
In horse races, 800-1600 m races were classified as short distance and 1601-2400
m races were classified as medium distance races. The correlations coefficients
calculated for breeding values of stallions between raced in short and medium
races found 0.18, 0.26 and 0.18 by Pearson, Spearman and Kendals methods,
||Descriptive statistics, significance levels of race speed
and racing time for Model-1
|*: Significance level of ANOVA results, a-h: Means with the
same letter(s) are not significantly different for race speed and racing
time for each factor
||Racing time descriptive statistics and significance levels
resulted for each distance by Model-2
|N: No. of records, SE: Standard Error
||Estimated genotypic and phenotypic variances and parameter
estimates for race speed for each racing distance (m)
|σ2A: Additive genetic variance,
σ2PE: Permanent environmental variance, σ2E:
Error variance, σ2P: Phenotypic variance, h2:
Heritability, P2e = σ2PE/σ2P,
Correlations between the estimated breeding values of stallions, which raced
short and medium distance races, were found insignificantly (p>0.05).
The effect of the race track was significant both for race speed and racing
time (p<0.01). Racing time on grass track was found shorter than in the sand
track (Table 1, 2). Kocak
and Ekiz (2005), Moritsu et al. (1994), Oki
et al. (1995) and Mota et al. (1998)
have reported similar results for Brazilian Thoroughbreds. Sand makes racing
condition hard, thus the races held in sand track are more difficult for horses
than the races held in turf track.
On the other hand, race finishing time showed significant differences according
to the hippodromes where the races were held (p<0.01). Similar results have
been reported by Bakhtiari and Kashan (2009).
The effect of year was found to be significant on the race performance (p<0.01).
Over the years, race speed increased and race finishing time decreased. In terms
of years, both environmental improvements and progress in the breeding selection
improved race performance of horses. Ekiz and Kocak (2005)
have reported that race performance affected by the year effect in Arabian horses.
Also, similar findings have been mentioned by Kocak and
Ekiz (2005) and Ekiz et al. (2005b) for thoroughbred
Heritabilities of racing time varied according to race distances from 0.044
(2200 m) to 0.106 (1300 m). In general, heritability for racing times was low.
In the Anatolian farm where the experimental data was obtained the number of
breeding stock per year was about 8-9. The low number of breed-stocks for the
reproduction purpose may lead decreased genetic variance, leading to low heritability
values for racing times. On the other hand, the estimates of heritability and
repeatability of racing time tended to decrease with increasing racing distance.
Decreasing tendency in heritability could be due to the fact the racing speed
could be lowered to the extend where the environmental conditions are changed
by increased racing distance.
In the model-2 analysis, heritability for racing times was estimated as 0.0538.
According to this, heritability belonging to the racing times was found to be
higher in the shorter distances and lower for long distances. Similar result
was expressed by Sobczynska (2006) and Mota
et al. (2005).
Estimated heritabilities in this study were in agreement with the reports of
Sobczynska (2006), Moritsu et
al. (1994), Bakhtiari and Kashan (2009), Mota
et al. (2005), Taveria et al. (2004).
On the other hand, obtained heritability estimates in this study were lower
than the values found by Ekiz and Kocak (2005), Lee
et al. (1992), Park and Lee (1999), Oki
et al. (1995) and Villela et al. (2002).
Observing Table 3, repeatabilities belonging to racing time
varied according to race distances and ranged 0.233 (1500 m) to 0.626 (1300
m). In the analysis, without taking into account of the distance (Model-2),
repeatability of racing times was estimated as 0.351. Estimated repeatabilities
in this study were in agreement with the values of Park
and Lee (1999) and Villela et al. (2002).
On the other hand, obtained repeatability estimates in this study were lower
than the values found by Oki et al. (1995) and
were higher than the values found by Ekiz et al.
(2005a, b), Bakhtiari and Kasan
(2009) and Ekiz and Kocak (2007).
In this study, heritabilities were found to be low, while repeatabilities were
found to be high in relation with race performance. Therefore, possibility of
improvement by selection seemed limited. However, because of the high repeatabilities,
selection could be done in the early ages.
Minimum and maximum race speeds were observed at 1800 (13:55 m sec-1)
and 1200 m (14:29 m sec-1) races, respectively. Range of observed
race speeds was 0.74 m sec-1. This range value was higher than 0.24
m sec-1 reported by Mota et al. (1998)
and lower than 0.85 m sec-1 reported by Taveira
et al. (2004).
Between breeding values of short and medium distances, in both categories which
had 27 horse cubs, Pearson and Kendall correlation coefficients were found the
same 0.18 and Spearman correlation coefficient was found 0.26. These correlation
coefficients were lower than 0.35 reported by Moritsu et
al. (1994) for between 1200 and 1800 m. Interpretations of these results,
the short and medium distance running abilities of horses were independent of
each other; therefore, this situation should be taken into account in the selection
of breeding. Race distance and runway type should be taken into consideration
for estimation of the breeding values. Model-1 and model-2 had given similar
results for racing time. Model-1 (full model) may be preferred to Model-2. Because,
interpretation of racing time abilities for all the factors (included in model-1)
available via model-1.