Copyright is owned by the Author of the thesis.  Permission is given for 
a copy to be downloaded by an individual for the purpose of research and 
private study only.  The thesis may not be reproduced elsewhere without 
the permission of the Author. 
 
Examination of New Zealand sport horse performance 
records and their suitability for the calculation of 
breeding values 
 
 
 
 
A thesis presented in partial fulfilment  
of the requirements for the Degree of Master of Science 
in 
Animal Science 
at 
Massey University 
 
 
 
 
 
 
 
Frances Emily Creagh 
2011  
 
 
i 
 
Declaration 
 
  
 
 
ii 
 
Abstract 
 
Currently, there is no system for genetic evaluation of sport horse sires in New Zealand; 
however, the implementation of such a system would be beneficial to the sport horse 
industry. Official performance data for the 2008/09 and 2009/10 competition seasons 
were obtained from Equestrian Sport New Zealand.  Initially data were examined using 
descriptive statistics. There were 1123 and 1472 horses registered for dressage, 902 and 
1255 horses registered for eventing and 1326 and 1331 horses registered for show 
jumping during the 2008/09 and 2009/10 seasons respectively.  13.2% and 14.2% of 
horses registered for dressage, 15% and 3.8% of horses registered for eventing and 
16.3% and 17.2% of horses registered for show jumping had no sire recorded. Between 
63.6% and 75% of sires had only 1 progeny record. For dressage and eventing points 
and number of starts were recorded. Zero points were recorded for 1.8% and 1.1% of 
horses in dressage and 64.6% and 13% of horses in eventing. For show jumping, prize 
money was recorded and records were usually only available for horses which placed in 
competition. 
Data on points, number of starts and prize money was skewed but approached normality 
under log10 transformation for all 3 disciplines. EBVs were calculated for these 
variables. 11.8%, 11.3 % and 14.2% of sires had 5 or more progeny records available 
for genetic analysis. Estimated breeding values for points per start ranged from -0.066 
to 0.158 and -0.076 to 0.182 for dressage and eventing. Estimated breeding values for 
number of starts ranged from -0.117 to 0.232 and -0.101 to 0.168 for dressage and 
eventing respectively and from -0.523 to 0.993 for prize money in show jumping. 
In conclusion, the use of estimated breeding values could lead to increased genetic gain 
and improved performance of New Zealand sport horses on the international stage. 
However, the current data recording has some limitations as records have not been kept 
for the purpose of genetic evaluation. Hence, there is a need for greater listing and 
reliability of pedigree data in order to effectively utilise estimated breeding values for 
selection of New Zealand sport horse sires.  
 
 
iii 
 
Acknowledgments 
 
This thesis would not have been possible without the guidance, time and patience of my 
supervisors Dr Chris Rogers and Dr Rebecca Hickson. Thank you so much for making 
the completion of this work possible. 
A big thank you is due to Equestrian Sport New Zealand for graciously allowing me 
access to their database. Without their support this project would not have been 
possible.  
Thanks also to Dr Nicolas Lopez-Villalobos for his help and guidance with using VCE 
and Dr Dave Johnson from Livestock Improvement Corporation for allowing me to use 
his REML programme for genetic analysis.  
I would also like to give a huge thank you to my family, my parents Phil and Lyn and 
siblings David and Katie. Without your support and encouragement I would never have 
been able to have come this far. 
  
 
 
iv 
 
List of Abbreviations 
BLP Best Linear Prediction 
BLUE Best Linear Unbiased Estimates 
BLUP Best Linear Unbiased Prediction  
dnst log(number of starts) for dressage 
dpps log(points per start+1) for dressage 
EBVs Estimated Breeding Values 
enst log(number of starts) for eventing 
epps log(points per start+1) for eventing 
ESNZ  Equestrian Sport New Zealand 
FEI Federation Equestre Internationale 
GLM General Linear Model 
HHANZ Holsteiner Horse Association of New Zealand 
LSE Least Squares Equations 
MME Mixed Model Equations 
MSI Missing Sire Information 
NZHS New Zealand Hanoverian Society 
NZWBA New Zealand Warmblood Breeders Association 
REML Restricted Maximum Likelihood 
Sib(s) Sibling(s) 
SPT Stallion Performance Test 
SQRT Square Root 
tpm log(total prize money+1) 
UELN Unique Equine Life Number 
VCE Variance Components Estimator 
WBFSH World Breeding Federation for Sport Horses 
WSI Wrong Sire Information  
 
 
v 
 
Table of Contents 
Declaration ......................................................................................................................... i 
Abstract ............................................................................................................................. ii 
Acknowledgments ............................................................................................................ iii 
List of Abbreviations........................................................................................................ iv 
List of tables ...................................................................................................................... x 
List of figures .................................................................................................................. xii 
 
Chapter 1: Introduction ..................................................................................................... 1 
 
Chapter 2: Literature Review ............................................................................................ 3 
 
2.1 SPORT HORSES ........................................................................................................ 3 
2.1.1 Definition of a sport horse ................................................................................ 3 
2.1.2 True age versus official age .............................................................................. 3 
 
2.2 SPORT HORSE COMPETITIONS IN NEW ZEALAND ......................................... 3 
2.2.1 Dressage ............................................................................................................ 3 
2.2.2 Eventing ............................................................................................................ 3 
2.2.3 Show jumping ................................................................................................... 4 
 
2.3 SPORT HORSE INDUSTRY ..................................................................................... 4 
2.3.1 European sport horse industries ........................................................................ 4 
2.3.2 New Zealand sport horse industry .................................................................... 5 
 
2.4 INTERNATIONAL EQUESTRIAN SPORT ............................................................. 6 
2.4.1 The International Equestrian Federation........................................................... 6 
2.4.2 World Breeding Federation for Sport Horses (WBFSH).................................. 7 
2.4.3 Interstallion ....................................................................................................... 7 
2.4.4 Unique Equine Life Number (UELN) .............................................................. 7 
2.4.5 Importance of the UELN for genetic evaluation .............................................. 8 
2.4.6 Genetic connectedness ...................................................................................... 8 
 
 
vi 
 
2.5 SPORT HORSE BREEDING ..................................................................................... 9 
2.5.1 Sport horse breeding in New Zealand............................................................... 9 
2.5.2 The potential impact of culture ....................................................................... 10 
 
2.6 CENTRAL PERFORMANCE TESTING ................................................................ 10 
2.6.1 Definition of central performance testing ....................................................... 10 
2.6.2 Use of central performance testing in sport horses ......................................... 11 
 
2.7 FIELD TESTING ...................................................................................................... 12 
2.7.1 Definition of field testing ................................................................................ 12 
2.7.2 Use of field testing in sport horses.................................................................. 12 
2.7.3 New Zealand ................................................................................................... 13 
2.7.4 Using competition results for genetic evaluation ........................................... 13 
2.7.5 Successful field testing schemes suitable for sport horse breeding ................ 15 
 
2.8 BREEDING OBJECTIVE ........................................................................................ 15 
2.8.1 Defining the breeding objective ...................................................................... 15 
2.8.2 Traits weightings............................................................................................. 16 
2.8.3 Targets of the breeding objective .................................................................... 16 
2.8.4 Profit satisfiers and profit maximisers ............................................................ 17 
2.8.5 Examples of breeding objectives for sport horse populations ........................ 17 
 
2.9 RELATIVES’ PERFORMANCE ............................................................................. 18 
 
2.10 THE IMPACT OF WRONG OR MISSING SIRE INFORMATION .................... 19 
2.10.1 Influence on variance .................................................................................... 19 
2.10.2 Reliability...................................................................................................... 19 
 
2.11 HERITABILITY ..................................................................................................... 19 
2.11.1 Definition of heritability ............................................................................... 19 
2.11.2 Heritability estimates in sport horses ............................................................ 20 
 
2.12 POTENTIAL SOURCES OF BIAS IN HERITABILITY ESTIMATES .............. 22 
 
 
vii 
 
2.12.1 Assortative mating ........................................................................................ 22 
2.12.2 Non random effect of environment ............................................................... 22 
 
2.13 METHODS OF DATA ANALYSIS ...................................................................... 22 
2.13.1 Methods ........................................................................................................ 22 
2.13.2 Best Linear Unbiased Prediction (BLUP) .................................................... 22 
2.13.3 Restricted Maximum Likelihood (REML) ................................................... 23 
2.13.4 Animal and Sire models ................................................................................ 23 
 
2.14 BREEDING VALUES ............................................................................................ 24 
2.14.1 Definition ...................................................................................................... 24 
2.14.2 Estimation of breeding values....................................................................... 24 
2.14.3 Advantages of the use of breeding values to the New Zealand sport horse 
industry .................................................................................................................... 28 
2.14.4 Transformations of sport horse data ............................................................. 28 
 
Chapter 3: Materials and Methods .................................................................................. 30 
 
3.1 DESCRIPTION OF THE SPORT HORSE POPULATION .................................... 30 
3.1.1 Dataset ............................................................................................................ 30 
3.1.2 Editing the datasets ......................................................................................... 30 
3.1.3 Investigation of opportunities for starts .......................................................... 31 
3.1.4 Performance measures .................................................................................... 31 
3.1.5 Statistical analysis ........................................................................................... 31 
 
3.2 ESTIMATED BREEDING VALUES ...................................................................... 32 
3.2.1 Definition of the estimated breeding value traits ............................................ 32 
3.2.2 Model and method .......................................................................................... 33 
 
Chapter 4: Results ........................................................................................................... 35 
 
4.1 DESCRIPTION OF THE SPORT HORSE DISCIPLINES BASED ON 
OPPORTUNITIES TO START ...................................................................................... 35 
4.1.1 Eventing .......................................................................................................... 35 
 
 
viii 
 
4.1.2 Dressage .......................................................................................................... 36 
4.1.3 Show jumping ................................................................................................. 37 
 
4.2 ANALYSIS OF THE 2008/09 AND 2009/10 COMPETITON SEASONS ............. 38 
4.2.1 Description of the population ......................................................................... 38 
4.2.2 Progeny numbers ............................................................................................ 39 
4.2.3 Normality tests ................................................................................................ 45 
4.2.4 Population age structure.................................................................................. 45 
4.2.5 Investigation of the variables for genetic analysis .......................................... 45 
 
4.3 ESTIMATED BREEDING VALUES ...................................................................... 48 
4.3.1 Description of the estimated breeding values calculated for New Zealand 
sport horse sires ....................................................................................................... 48 
4.3.2 Sire results....................................................................................................... 49 
4.3.3 The distribution of estimated breeding values for dressage sires ................... 55 
4.3.4 The distribution of estimated breeding values for eventing sires ................... 56 
4.3.5 The distribution of estimated breeding values for show jumping sires .......... 57 
 
Chapter 5: Discussion ..................................................................................................... 58 
 
5.1 SYSTEM OF EARNING POINTS AND PRIZE MONEY AND STARTS 
AVAILABILE IN NEW ZEALAND SPORT HORSE COMPETITIONS ................... 58 
5.1.1 Starts, points and prize money availability ..................................................... 58 
5.1.2 Comparisons across disciplines ...................................................................... 59 
 
5.2 THE UNIQUE SITUATION IN EQUESTRIAN SPORT........................................ 59 
5.2.1 Normality of the data ...................................................................................... 59 
5.2.2 Heterogeneity of the sport horse population ................................................... 60 
5.2.3 Missing records ............................................................................................... 61 
5.2.4 The effect of rider ........................................................................................... 61 
5.2.5 Stallions .......................................................................................................... 62 
5.2.6 Mares .............................................................................................................. 63 
 
5.3 INTERPRETATION OF ESTIMATED BREEDING VALUES ............................. 63 
 
 
ix 
 
5.3.1 Usefulness of estimated breeding values ........................................................ 63 
5.3.2 Heritability value assumptions........................................................................ 64 
5.3.3 Individual sire indicators ................................................................................ 64 
5.3.4 Multi-trait animal model ................................................................................. 65 
 
5.4 RECOMMENDATIONS .......................................................................................... 65 
5.4.1 A template for estimated breeding values....................................................... 65 
5.4.2 Conclusion ...................................................................................................... 66 
 
References ....................................................................................................................... 67 
Appendices (CD on inside back cover)  
 
 
x 
 
List of tables 
 
Table 2-1: Relative sizes of the sport horse industries in Germany and New Zealand 
(Rogers, 2008, Rogers and Firth, 2005). ........................................................................... 5 
 
Table 2-2: Grades of competition recognised by the FEI for show jumping, eventing 
and dressage (Federation Equeste Internationale, 2010)................................................... 6 
 
Table 2-3: The number of currently approved stallions and examples of influential 
stallions, measured by number of progeny, in the NZWBA and NZHS studbooks in 
New Zealand (The New Zealand Hanoverian Society, 2010, New Zealand Warmblood 
Breeders Association, 2010, Friedrich, 2010). ................................................................ 10 
 
Table 2-4: Breeding objectives of some of the German studbooks (Dutch Warmblood, 
Oldenburg, Holsteiner and Trakehner) (Holsteiner-verband, 2003, KWPN, 2003, 
Oldenburger-pferde, 2004, Trakehner-Verband, 2005). ................................................. 18 
 
Table 2-5: Heritability values of performance traits for sport horses based on field test 
data. ................................................................................................................................. 21 
 
Table 2-6: Previous transformations of sport horse and racing horse performance data.
 ......................................................................................................................................... 29 
 
Table 4-1: Description of the data set for the equestrian disciplines of dressage, 
eventing and show jumping during the 2008/09 and 2009/10 competition seasons. ...... 38 
 
Table 4-2: The ten stallions with the greatest number of progeny records (nprog) in 
dressage during the 2008/09 and 2009/10 seasons. ........................................................ 40 
 
Table 4-3: The ten stallions with the greatest number of progeny records (nprog) in 
eventing during the 2008/09 and 2009/10 seasons. ........................................................ 40 
 
Table 4-4: The ten stallions with the greatest number of progeny records (nprog) in 
show jumping during the 2008/09 and 2009/10 seasons. ............................................... 40 
 
Table 4-5: Percentile bands for the distribution of EBVs calculated for the New 
Zealand data set using data from the 2008/09 and 2009/10 seasons in dressage, eventing 
 
 
xi 
 
and show jumping. The variables are dpps and epps (log(points per start+1)), dnst and 
enst (log(number of starts)) and tpm (log(total prizemoney+1)). ................................... 48 
 
Table 4-6: Comparison of sires with the greatest number of progeny records (nprog) 
and sires with the highest EBVs (for the variable dpps) for dressage. Values are from 
data combined for the 2008/09 and 2009/10 seasons. .................................................... 50 
 
Table 4-7: Comparison of sires with the greatest number of progeny records (nprog) 
and sires with the highest EBVs (for the variable dnst) for dressage. Values are from 
data combined for the 2008/09 and 2009/10 seasons ..................................................... 51 
 
Table 4-8: Comparison of sires with the greatest number of progeny records (nprog) 
and sires with the highest EBVs (for the variable epps) for eventing. Values are from 
data combined for the 2008/09 and 2009/10 seasons. .................................................... 52 
 
Table 4-9: Comparison of sires with the greatest number of progeny records (nprog) 
and sires with the highest EBVs (for the variable enst) for eventing. Values are from 
data combined for the 2008/09 and 2009/10 seasons. .................................................... 53 
 
Table 4-10: Comparison of sires with the greatest number of progeny records (nprog) 
and sires with the highest EBVs (for the variable tpm) for show jumping. Values are 
from data combined for the 2008/09 and 2009/10 seasons. ............................................ 54 
  
 
 
xii 
 
List of figures 
 
Figure 4-1: Average number of points per start and number of starts available in 
eventing during the spring 2010/11 season, by level of competition. ............................. 35 
 
Figure 4-2: Number of tests available at each level in dressage during the spring 
2010/11 season. ............................................................................................................... 36 
 
Figure 4-3: Number of starts available and average money per start on offer for show 
jumping during the spring 2010/11 season. .................................................................... 37 
 
Figure 4-4: Histograms of raw (left) and transformed (log10 data plus 1) (right) data for 
the equestrian disciplines of (a) dressage (points/start); (b) eventing (points/start) and (c) 
show jumping (prize money). Data is combined for the 2008/09 and 2009/10 seasons. 
Normality curves show the ideally desired distribution for the given data..................... 41 
 
Figure 4-5: Histograms of log(points per start+1) with records valued at zero removed 
to show improved normality of excluding these records for horses in dressage (a) and 
eventing (b) and log(total money+1) for show jumping (c). Data is combined for the 
2008/09 and 2009/10 seasons. Normality curves show the ideally desired distribution 
for the given data. ............................................................................................................ 42 
 
Figure 4-6: Histograms of the age distribution of horses competing in dressage (a), 
eventing (b) and show jumping (c). Data is combined for the 2008/09 and 2009/10 
seasons. Normality curves show the ideally desired distribution for the given data. ..... 43 
 
Figure 4-7: Histograms for number of starts and log(number of starts) in dressage (a) 
and eventing (b). Data is combined for the 2008/09 and 2009/10 seasons. Normality 
curves show the ideally desired distribution for the given data. ..................................... 44 
 
Figure 4-8: Breakdown of the variables included in the model for genetic analysis 
showing the average points per start and number of starts for mares, geldings and 
stallions in dressage during the 2008/09 and 2009/10 seasons. ...................................... 46 
 
Figure 4-9: Breakdown of the variables included in the model for genetic analysis 
showing the average points per start and number of starts for mares, geldings and 
stallions in eventing during the 2008/09 and 2009/10 seasons. ...................................... 47 
 
 
xiii 
 
Figure 4-10: Breakdown of the variables included in the model for genetic analysis 
showing the average prize money won by mares, geldings and stallions in show 
jumping during the 2008/09 and 2009/10 seasons. ......................................................... 47 
 
Figure 4-11: Histogram of the distribution of EBVs calculated from New Zealand data 
over the 2008/09 and 2009/10 competition seasons for the variable log(points per 
start+1) in dressage (dpps). ............................................................................................. 55 
 
Figure 4-12: Histogram of the distribution of EBVs calculated from New Zealand data 
over the 2008/09 and 2009/10 competition seasons for the variable log(number of starts) 
in dressage (dnst)............................................................................................................. 55 
 
Figure 4-13: Histogram of the distribution of EBVs calculated from New Zealand data 
over the 2008/09 and 2009/10 competition seasons for the variable log(points per 
start+1) in eventing (epps). ............................................................................................. 56 
 
Figure 4-14: Histogram of the distribution of EBVs calculated from New Zealand data 
over the 2008/09 and 2009/10 competition seasons for the variable log(number of starts) 
in eventing(enst) . ............................................................................................................ 56 
 
Figure 4-15: Histogram of the distribution of EBVs calculated from New Zealand data 
over the 2008/09 and 2009/10 competition seasons for the variable log(total prize 
money+1) in show jumping (tpm). ................................................................................. 57 
 
 
 
 
1 
 
Chapter 1  
Introduction 
 
Equestrian Sport New Zealand (ESNZ) is responsible for registering horses and 
organising registered competitions in all disciplines within New Zealand. ESNZ has 
approximately 5,000 horses registered for competition each year and has an annual 
budget of $1,459,420 (Rogers and Firth, 2005, Equestrian Sports New Zealand, 2009a). 
In New Zealand, one of the major problems with utilising competition results for 
stallion selection is that there is very intense genetic selection at an early age due to the 
gelding1 of horses to be used in competition (Rogers, 1993a, Rogers and Firth, 2005). 
Sport horse competition data tends to be highly skewed by nature of the competition 
structure, as there are large numbers of horses which earn few or zero points and little or 
zero prize money during the season and a small number of horses that are exceptionally 
competitive (Whitaker and Hill, 2004). Therefore, sport horse data must be transformed 
in order to approach a normal distribution and be useful for genetic analysis (Langlois 
and Blouin, 1998). Data transformations which are commonly applied to sport horse 
performance data are logarithmic, power and square root transformations (Langlois and 
Blouin, 2004, Arnason, 1999, Luehrs-Behnke et al., 2002). 
An estimated breeding value (EBV) is a measure of phenotypic performance which can 
be used as a predictor of true breeding value (Lynch and Walsh, 1998). However, the 
reliability of an EBV as a predictor depends on the quality of the data. Reliability of 
EBVs calculated from datasets with wrong or missing sire information, small progeny 
groups or data with incomplete records for the model effects (such as age and gender) 
will be low (Woolliams, 2006, Kearsley et al., 2008). Utilising pedigree and progeny 
records in conjunction with the sire’s competition record increases the reliability2 of 
EBVs. A minimum of 5-10 progeny records is required to achieve sufficient reliability 
in EBV calculation. The greater the number of progeny records, the higher the 
reliability of the EBV (Sun et al., 2009, Phocas et al., 2008). In European sport horse 
                                                          
1 A neutered male horse 
2 How accurate the EBVs are as a measure of true genetic merit. Reliability is increased by collecting 
more phenotypic information on an animal or its relatives 
 
 
2 
 
populations, the average number of progeny records is approximately 50 progeny per 
stallion and 5 progeny per broodmare (Langlois, 1983). 
Estimated breeding values are calculated via the use of a general linear model (GLM), 
REML or best linear unbiased prediction (BLUP) method (Stewart et al., 2010, 
Arnason, 1999). EBV calculations are generally based on a single trait animal model, 
multi trait animal model or a sire model (Tavernier, 1991, Langlois and Blouin, 1998, 
Dietl et al., 2005, Klemetsdal, 1994). Many fixed and random effects have been 
included in sport horse models. However, the most common appear to be age (of the 
sire or progeny), gender and season (Langlois and Blouin, 1998, Langlois, 1980a, 
Huizinga and Vandermeij, 1989). The use of reasonably reliable EBVs (generally with a 
reliability of approximately 14% or higher based on Interstallion guidelines) 
complements the use of a detailed breeding objective in animal breeding programmes, 
as choices of animals selected for breeding can be made more effectively (Harris and 
Newman, 1994, Interstallion, 2005).  
EBVs have not previously been calculated for New Zealand sport horses; however 
EBVs are utilised in many European sport horse industries including in Germany, 
France and the Netherlands (KWPN, 2010, Tavernier, 1990a). In France, annual results 
(logtransformed earnings, number of starts and number of placings) and details from 
each event (rank in each competition) are utilised (Tavernier, 1990a, Langlois, 1980a). 
In Germany the variable earnings per start is analysed (Huizinga and Vandermeij, 
1989). 
In this study it was suggested that the New Zealand sport horse performance dataset 
would not lend itself well to genetic analysis due primarily to limited and inaccurate 
pedigree information. The aim of this thesis was to describe the current sport horse 
population in New Zealand by looking at ESNZ records for dressage, eventing and 
show jumping from the 2008/09 and 2009/10 competition seasons.  The accuracy and 
completeness of the data available for genetic analysis and the opportunities available to 
compete in the three disciplines of dressage, eventing and show jumping (during the 
2010/11 spring season) were also investigated. The purpose of these investigations was 
to analyse the suitability of New Zealand sport horse data for genetic evaluation and to 
calculate estimated breeding values in the New Zealand sport horse population. 
 
 
3 
 
Chapter 2  
Literature Review 
2.1 SPORT HORSES 
2.1.1 Definition of a sport horse 
A sport horse has been defined as a horse which competes in, or is suited to, the 
Olympic disciplines of dressage, eventing and show jumping (Tavernier, 1990b). The 
Federation Equestre Internationale (FEI) now also recognises endurance, vaulting, 
reining, driving and para-equestrian among the sport horse disciplines (Rogers and 
Wickham, 1993b). However, only the disciplines of dressage, eventing and show 
jumping were considered in this thesis. 
2.1.2 True age versus official age 
An official age is designated to all horses competing in sport horse disciplines in New 
Zealand. The cut off for assigning an official birth year is 1 August each year. 
Therefore, all horses born each season before 1 August each year will have an official 
birth year of one year earlier than their true birth year. Horses born each season on or 
after 1 August each year will have the same official birth year as their true birth year 
(Langlois and Blouin, 1998).  
2.2 SPORT HORSE COMPETITIONS IN NEW ZEALAND 
2.2.1 Dressage 
In dressage, points are awarded according to the horses final percentage score for each 
event, (points awarded being 7, 5, 3 or 1 irrespective of grade of competition). A horse 
can compete in up to three dressage tests per day of competition (Equestrian Sports 
New Zealand, 2009b). 
2.2.2 Eventing 
Eventing in New Zealand has three major grades in which horses can earn points. These 
are Advanced, Intermediate and Novice. Points are awarded in eventing on an ascending 
scale, with increasing points awarded with higher grades, such that a win earns 6, 9 or 
 
 
4 
 
12 points for Novice, Intermediate or Advanced grade (Equestrian Sports New Zealand, 
2009b). 
2.2.3 Show jumping 
Prize money won in show jumping is positively associated with height of jumps and 
difficulty of competition, and is usually allocated with each placing receiving 60% of 
the prize money given to the placing above it. Horses may compete in multiple show 
jumping rounds in a single day of competition (Equestrian Sports New Zealand, 2009b).  
2.3 SPORT HORSE INDUSTRY 
2.3.1 European sport horse industries 
In Europe, many horses are bred specifically to be competitive in the sport horse 
disciplines. European sport horse populations tend to be large and there is ample 
government support and funding for equestrian sport (Barneveld, 1996). There are 
strong links between the sport horse governing bodies and breeding industries, which 
allows national breeding objectives and breeding schemes to be developed (Aurich and 
Aurich, 2006).  
The development of national schemes includes the utilisation of stallion performance 
tests (SPT) and progeny tests. In order to gain entry into a stud book, stallions must 
undergo stallion performance testing to ensure they meet the standard required by that 
stud book (Ducro et al., 2007). The central performance testing procedure ranges from 
one week to 100 days dependent on the stallion age (generally 2.5 to 5 years), 
competition success and specific regulations of the different studbooks (Hellsten et al., 
2006, KWPN, 2003). Stallions are evaluated for soundness, conformation, gaits under 
rider, free jumping and jumping under rider. Temperament and general appearance 
scores and a pedigree evaluation, based on ancestors’ performance in dressage and show 
jumping, are also incorporated into the final score (Olsson et al., 2008). Progeny tests 
are used to provide information for genetic evaluation of mares and stallions. They are 
also useful to evaluate the quality of young sport horses, particularly their ability in both 
dressage and show jumping (Olsson et al., 2008). The highly integrated systems utilised 
in Europe lead to complete and accurate records on both pedigree and performance for 
 
 
5 
 
all horses registered for competition. Hence, it is possible to estimate breeding values on 
these horses, which can then be utilised in the selection of sport horses. 
2.3.2 New Zealand sport horse industry  
 Equestrian sport in New Zealand is managed by ESNZ which is responsible for 
registration of horses and organising registered competitions in all disciplines. ESNZ 
has approximately 5,000 horses registered for competition each year and has a current 
annual budget of $1,459,420 (Rogers and Firth, 2005, Equestrian Sports New Zealand, 
2009a). The comparative sizes of the German (largest) and New Zealand sport horse 
industries are shown in table 2-1. 
Table 2-1: Relative sizes of the sport horse industries in Germany and New Zealand (Rogers, 
2008, Rogers and Firth, 2005). 
 
 
 
 
In New Zealand, few horses are bred specifically for the sport horse disciplines. Horses 
bred for racing that were unsuccessful as race horses are commonly re-trained as sport 
horses. The New Zealand Thoroughbred has traditionally been very successful in 
eventing. However, race horses are not selected based on traits that are indicative of 
successful sport horses (Rogers and Wickham, 1993a). In contrast, leading European 
countries, including Germany and France, have instigated specific sport horse breeding 
programmes aimed at genetic improvement toward the sport horse objective. Hence, it 
is highly likely that the European bred sport horses will become superior to the 
Thoroughbred in sport horse competitions. The instigation of sport horse breeding 
programmes would ensure New Zealand remains competitive in international sport 
horse competition (Rogers and Wickham, 1993b). European societies and systems tend 
to lend themselves to the development of structured and integrated breeding 
programmes. The English system (similar to that in New Zealand) may not lend itself to 
an integrated breeding programme as there is limited breeding towards national 
 Germany New Zealand 
Annual budget $NZ12.6 million $NZ1.5 million 
Number of horses 87,000 5,000 
 
 
6 
 
objectives and greater separation exists between the breeding and sport sectors 
(Whitaker and Hill, 2004). 
2.4 INTERNATIONAL EQUESTRIAN SPORT 
2.4.1 The International Equestrian Federation 
The FEI was founded in 1921 as the international body responsible for governing 
equestrian sport. It is now the sole controlling authority for all international events in 
jumping, dressage, eventing, endurance and para-equestrian. There are 133 national 
federations affiliated with the FEI and 65 studbook members compiled by the World 
Breeding Federation for Sport Horses (WBFSH) (Federation Equeste Internationale, 
2010, World Breeding Federation for Sport Horses, 2011). The international grades 
recognised by the FEI are the same for all countries. These grades are listed in table 2-2. 
Table 2-2: Grades of competition recognised by the FEI for show jumping, eventing and 
dressage (Federation Equeste Internationale, 2010). 
Discipline FEI grades 
Show jumping 1*, 2*, 3*, 4* & 5* maximum heights are 1.20m, 
1.35m, 1.45m, 1.50m and 1.60m respectively 
Eventing CCN/CNC (national competitions), CIC/CICO 1-
3* (short format international competitions), 
CCI/CCIO 1-4* (official international 
competitions incorporating an official teams 
competition), CCIJ/CICJ (junior rider classes, 18 
years and under), CCIY/CICY (young rider 
classes, 21 years and under) 
Dressage CDI 1*-5* (international competitions), CDI-W 
(world cup competitions), CDIO 3*-5* 
(international competitions open to riders from 
any number of foreign nations), CDIY/CDIOY 
(riders 21 years and under), CDIJ/CDIOJ (riders 
18 years and under), CDI-YH (Young horse 
competitions) 
 
 
7 
 
Competition grades are relevant to genetic evaluation as different aptitudes tend to be 
measured in different grades. Therefore, using competition data from lower grades may 
not be representative of performance in elite sport. However, records from young 
progeny, competing in lower grades, will always be used for genetic analysis as a short 
generation interval3 is necessary to increase the rate of genetic gain4 (Aldridge et al., 
2000). 
2.4.2 World Breeding Federation for Sport Horses (WBFSH)  
The WBFSH is the only international federation of studbooks for sport horses. It 
registers approximately 130,000 foals each year, 80% of which are bred in Europe 
(Koenen and Aldridge, 2002). It provides a major connection between breed 
organisations and the FEI (UELN, 2010). 
2.4.3 Interstallion  
Sport horse breeding is becoming an increasingly international endeavour. Therefore, 
the correct interpretation of performance test results and genetic evaluations is 
important to ensure accurate selection of foreign genetic material. The Interstallion 
Committee was established in 1998 in order to develop understanding of these 
information sources (Interstallion, 2010). Interstallion has been working to compile a 
systematic overview of testing and genetic evaluation methods as practised by various 
organisations. They have also compiled recommendations on the publication of 
breeding values and collaborated with the FEI and WBFSH to increase the availability 
and quality of information on horses that compete internationally (Interstallion, 2010). 
2.4.4 Unique Equine Life Number (UELN) 
The Unique Equine Life Number (UELN) is an initiative started by the WBFSH and the 
FEI in 2000, with the aim to maintain horse identity in every database worldwide 
(Vandenplas, 2010, Koenen and Aldridge, 2002). At present, each organisation (stud 
book and national equestrian federation) has an independent identification system, 
making data exchange difficult.  The intention of the UELN initiative is to develop a 
common language between all organisations responsible for registering horses to ensure 
                                                          
3 Measures how quickly progeny will be able to replace parents in the breeding population. 
4 A measure of how much genetic improvement, towards a specific breeding goal, is seen in a given 
population. 
 
 
8 
 
that each horse can be uniquely identified. This is useful as stallions may now have 
progeny registered in many different countries (Hellsten et al., 2008). Hence, 
registration with a unique number will make exchanging data and progeny records 
between countries much easier, allowing for complete progeny records to be obtained 
on a sire (UELN, 2010, World Breeding Federation for Sport Horses, 2010, Federation 
Equeste Internationale, 2010, Koenen, 2002b). 
2.4.5 Importance of the UELN for genetic evaluation  
There have been significant improvements in reproductive techniques, such as increased 
use of chilled and frozen semen, in the equine industry (Koenen and Aldridge, 2002). 
Hence, breeding stallions can now be used in multiple countries and genetic 
connectedness among these countries is much higher than before. This has led to a 
greater need for transparency in breeding objectives for sport horses and a greater 
ability to track the progeny of a stallion which have been exported to a foreign country 
(Bruns et al., 2004, Koenen et al., 2003).  
At present, estimated breeding values (EBVs) are calculated on national data and results 
from FEI competitions are excluded due to a lack of complete information on 
internationally competing horses. A horse’s name and identification number are often 
changed after export; hence, it is difficult to trace pedigree details. This is one of the 
most serious drawbacks in international sport horse breeding, as many top sport horses 
are not properly identified and are therefore lost from genetic analysis, which introduces 
significant bias in EBV calculations  (Koenen and Philipsson, 2007).  
2.4.6 Genetic connectedness 
Genetic connectedness between any two countries is quantified as the average 
prediction error variance of differences in EBVs between both countries. UELN 
implementation would greatly improve the ability to estimate genetic connectedness 
among horse populations. However, the UELN system has yet to be implemented in all 
countries (Bruns et al., 2004). A UELN code exists for the New Zealand sport horse; 
however, it is not presently used in the identification of horses registered with ESNZ 
(UELN, 2010). In order for EBVs to be effectively used across countries, genetic 
connectedness must exist, as this makes it possible to accurately estimate differences in 
the average genetic merit of populations and determine the genetic correlation between 
 
 
9 
 
performance results obtained from any combination of two populations. The lack of an 
original identification number for each horse and the high potential for names to be 
changed during export, means underestimation of genetic connectedness will result 
(Ruhlmann et al., 2009). It is possible to estimate EBVs without a good degree of 
connectedness; however, this reduces the reliability of the estimate. Using populations 
with weak connectedness could introduce large bias in the estimation of genetic merit of 
populations across countries, because the differences among populations may not be 
estimable, but are still assumed to be zero (Hellsten et al., 2008, Koenen and Aldridge, 
2002). 
2.5 SPORT HORSE BREEDING 
2.5.1 Sport horse breeding in New Zealand 
Sport horse breeding in New Zealand is controlled by the major breed organisations 
including the New Zealand Warmblood Horse Breeders Association (NZWBA), the 
New Zealand Hanoverian Society (NZHS) and the Holsteiner Horse Association of 
New Zealand (HHANZ). These organisations are responsible for the registration of 
horses that meet the requirements of the stud book, defining the parameters for and 
instigating stallion testing procedures to develop a list of approved stallions. Integration 
of breed associations is very limited and they operate independently with no input from 
ESNZ (New Zealand Warmblood Breeders Association, 2010, The New Zealand 
Hanoverian Society, 2010, New Zealand Hanoverian Horse Association, 2009, 
Equestrian Sports New Zealand, 2009b).  
The major breeding evaluation programme in New Zealand is the New Zealand Sport 
Horse Breed Improvement Programme. This programme is based on genetic 
improvement without the inspection of mares and stallions before breeding. The 
emphasis is on voluntary publication of performance data (Rogers, 2010). Sport horse 
breeding organisations in New Zealand (NZWBA and NZHS) are small compared with 
large European sport horse breeding organisations, such as the Holsteiner Verband in 
Germany. The sizes of the NZWBA and NZHS are given in table 2-3. 
 
 
 
 
10 
 
Table 2-3: The number of currently approved stallions and examples of influential stallions, 
measured by number of progeny, in the NZWBA and NZHS studbooks in New Zealand (The 
New Zealand Hanoverian Society, 2010, New Zealand Warmblood Breeders Association, 2010, 
Friedrich, 2010). 
2.5.2 The potential impact of culture 
European horses are successful in international competition. This is largely due to the 
highly integrated breeding programmes the breed associations are utilising. The 
successful instigation of these programmes in Europe may in part be the result of the 
inland nature of many European countries. Due to this inland nature in many European 
countries, an integrated system, where a select group of people were responsible for 
decision making for the country as a whole developed. This was beneficial during times 
of war and conflict (Hamann and Distl, 2008). Today this mindset still holds, hence, 
European breeders tend to be open to the idea of an integrated breeding system 
operating toward a commonly defined goal. Anglo-Saxon cultures, such as that in New 
Zealand, tend to be less inclined to accept an overall authority and breeding is more 
segmented and independent. This mindset could create challenges with respect to sport 
horse breeding in New Zealand (Rogers, 2010). 
2.6 CENTRAL PERFORMANCE TESTING 
2.6.1 Definition of central performance testing 
There are three different systems of central performance testing utilised in Germany, 
Sweden and the Netherlands for sire selection. Central performance tests are performed 
over long time periods (typically 70-100 days) and are very extensive, covering both 
Breed organisation Number of approved 
stallions 
Influential stallions include: 
NZWBA 36 Dream Boy 
Anamour 
NZHS 7 Winnebago 
Witzbold 
Distelfink 
Aberlou 
 
 
11 
 
dressage and jumping. A major advantage of central performance testing is that all 
horses are tested in a consistent environment. Therefore, the environmental component 
of phenotype is reduced. As phenotype (P) = genotype (G) + environment (E), the test is 
more accurate at determining genotype when E is smaller (Hintz, 1980). However, due 
to the high cost associated with performance testing only a small number of candidates 
can be tested. Central performance testing is a multi stage selection system (selection 
occurs in many stages), with initial selection usually based on indirect selection criteria 
such as conformation and movement (Dubois et al., 2008, Hellsten et al., 2006, Wallin 
et al., 2003).  
2.6.2 Use of central performance testing in sport horses 
Central performance testing is labour intensive and expensive due to the length of the 
testing process and the facilities and personnel required. These tests may be more 
reliable (have higher reliability) than field tests, as the component traits of successful 
competition performance are measured as discrete traits over a period of time (Hellsten 
et al., 2006). However, central performance tests may not be the most effective method 
of selection for competition success, with accuracy of selection (rTI) of 60-70%. The 
generation interval is shorter than that of field testing, as stallions can be selected and 
licensed at an earlier age, with no need to wait for competition results. Selection 
intensity (ଓҧ)5 for entry into the central performance test via the Keuring6 is quite high. 
For example, in Germany, where approximately 40/4000 stallions are selected (ଓҧ= 2.66) 
(Holsteiner Verband, 2010). However, once a stallion has been selected to participate in 
the stallion performance test, it will almost certainly be approved during testing; hence, 
selection at the performance test itself tends to be very low. The age at selection for 
stallions is usually 3-4 years; hence, the generation interval would be approximately 4-5 
years. However, generation interval, selection intensity and overall rate of genetic gain 
will vary depending on the exact system implemented (Teegen et al., 2009). From the 
genetic gain equation: 
ΔG = 
పҧ௥೅಺ఙ೒
௅
 
                                                          
5 Determined by the proportion of available animals selected as replacements in the breeding population. 
6 A test performed at 2.5 years of age over 3-4 days. Young stallions are tested in walk, trot, canter and 
free jumping. 
 
 
12 
 
Where: 
ΔG = genetic gain 
ଓҧ = selection intensity 
rTI = accuracy of selection (the correlation between true and estimated genetic merit) 
σg = genetic standard deviation 
L = generation interval 
It can be seen that increasing selection intensity (ଓҧ) and reducing generation interval (L) 
will lead to a faster rate of genetic gain (Hazel, 1943).  
2.7 FIELD TESTING 
2.7.1 Definition of field testing 
Field testing is a method of genetic analysis which utilises competition data as a means 
of selecting sport horses for breeding. Field testing is utilised in sport horse selection in 
France, Finland, Norway and Belgium (and now also in Germany due to public 
pressure). It is useful for selecting horses for a single discipline, as competition results 
are separate for each discipline (Hellsten et al., 2006). However, a major problem in the 
use of field testing is the gelding of horses for competition. Field testing also has a 
greater environmental component as it is based on competition data. Hence, the 
accuracy of selection is lower (Dubois and Ricard, 2007). 
2.7.2 Use of field testing in sport horses 
Field testing is less accurate than performance testing, as it is based on the individual’s 
pedigree records and those of their progeny. However, it is cheaper and easier than 
performance testing (Thoren and Philipsson, 2010). The generation interval is slightly 
longer due to the use of progeny testing to increase reliability (Olsson, 2006). However, 
the use of early progeny performance records has greatly reduced this disadvantage. 
Early progeny performance records can be used, as high correlations between low 
grades and high grades and performance at a young age and later in life have been 
calculated (Aldridge et al., 2000, Huizinga and Vandermeij, 1989). The selection 
intensity is high; however, the precise value will depend on the specific system 
 
 
13 
 
implemented. The rate of genetic gain is much higher than if no system is implemented. 
The significantly lower cost of field testing often makes it the most realistic option. For 
this reason, field testing would be a suitable option for genetic analysis in New Zealand 
(Rogers and Firth, 2005). 
2.7.3 New Zealand 
 Central performance testing systems would not be implementable in New Zealand, as 
funding for equestrian sport is limited and the sport horse population is small 
(approximately 5,000 horses registered for sport each year) (Rogers and Firth, 2005). 
However, a simple field testing system could be utilised based on standard competition 
data for the calculation of estimated breeding values (EBVs). However, this would 
result in a longer generation interval than seen from central performance testing.  
At present, the New Zealand performance data set, which includes information on 
number of starts and points and prize money earnt during each competition season, is 
incomplete with regards to pedigree information. Therefore, to utilise a field testing 
scheme effectively, data captured and recorded by breed associations and ESNZ would 
need to be improved. To do this, breed associations and ESNZ would need to work 
together more closely and utilise unique identification numbers for all horses. Pedigree 
records could also be improved. For example, by charging breeders/owners more if they 
register a horse without a sire listed. In New Zealand, there is very high selection at an 
early age due to the gelding of horses to be used in competition. This is also a problem 
in utilising competition results for stallion selection as horses are gelded before 
competition results can be collected and analysed; however, records from related 
animals can also be used (Rogers and Firth, 2005, Rogers, 1993b). 
2.7.4 Using competition results for genetic evaluation 
Evaluating a stallion based on its phenotypic competition records alone leads to a biased 
outlook with respect to breeding potential, as it gives a less detailed indication of the 
stallion’s transmitting ability to his offspring than evaluation which also utilises 
pedigree and progeny records (Mrode and Thompson, 2005). Also, the stallion may 
only compete in one discipline, whereas his offspring may be competed in multiple or 
different disciplines (Hellsten et al., 2006). Therefore, when using competition results in 
genetic analysis it is important not only to look at the results of the stallion himself, but 
 
 
14 
 
also to utilise pedigree and progeny records. This significantly increases reliability; 
however, progeny testing will increase the generation interval. Hence, progeny testing 
must occur at the youngest age possible so that the generation interval remains 
reasonable (Huizinga and Vandermeij, 1989, Thoren and Philipsson, 2010).  
Moderate to high correlations (between 0.4 and 0.98) have been found between early 
competition results and performance later in life (Hassenstein, 1998, Reinhardt and 
Schmutz, 1997). However, it is still argued that the use of competition data from lower 
grades may not be representative of performance in elite sport due to different aptitudes 
being measured in different grades. For example, low grade jumping classes have lower 
fences and a lesser degree of technical difficulty than higher grades (Aldridge et al., 
2000). This has led to the suggestion by some horse breeders in Ireland that progeny 
testing should only include progeny performing in higher grades. However, this 
suggestion has been rejected by Aldridge et al. (2000), on the basis that it would lead to 
a biased sample of progeny from the parent being evaluated since at least 50% of 
performance records are from lower grades. Performance records from lower grades 
will also become available from progeny at a younger age. Therefore, utilising 
information from lower grades would result in a shorter generation interval than waiting 
for results from higher grades. Hence, the rate of genetic gain would increase (Aldridge 
et al., 2000). 
As these low grade results are utilised it is important to know the relationship between 
performance in low grades and high grades. These correlations have been calculated in 
two studies. Aldridge et al. (2000) found that genetic correlations (rg) for different levels 
varied between 0.07 and 0.78 depending on the amount of difference between the levels. 
Correlations were smaller as the difference between the levels increased. It was 
concluded that results from low and medium level competition are useful indicators of 
high level performance and can be used in genetic evaluation. Huizinga and Van Der 
Meij (1989) found high correlations between performance at different ages (rg = 0.75). 
This indicates that genetic evaluation at an early age will be effective in making genetic 
progress in performance. 
The age of stallions is also important. Older stallions will have been subjected to more 
selection than younger stallions and stallions that were considered better would have 
more progeny entered in competition. Therefore, the most unbiased estimates for 
 
 
15 
 
parameters relating to genetic analysis (variance and heritability) would be from young 
stallions, based on data from their oldest progeny when these progeny reach 4-6 years of 
age and can first contribute a competition record for genetic analysis (Huizinga and 
Vandermeij, 1989). 
2.7.5 Successful field testing schemes suitable for sport horse breeding 
Field testing schemes have been successfully implemented in European sport horse 
breeding industries including France and Germany. In France, annual results (log 
transformed earnings, number of starts and number of placings) and details from each 
event (rank in each competition) are utilised (Tavernier, 1990a, Langlois, 1980a). 
Ricard and Chanu (2001) evaluated eventing competition results over a 17-year period. 
The number of records available for analysis was large with 13,000 horses, 30,000 
annual results and 110,000 starts. Horses with no earnings during the year were 
removed from the analysis of annual results. All horses with performances had pedigree 
information (sire and dam) for at least two generations of ascendants. All pedigree 
information was utilised in the variance component estimation, however, most pedigree 
information was from paternal half sibs.  
In Germany the variable earnings per start has been analysed (Huizinga and 
Vandermeij, 1989). Bruns et al. (1981) analysed 136,000 German competition records 
(data from one season) for horses placed in dressage and jumping. Horses with zero 
placings recorded (60-70%) were not included in the analysis.  
Field testing schemes such as these would be implementable in the New Zealand sport 
horse breeding industry. However, the number of records available for analysis would 
be much smaller and pedigree information much sparser; hence, the reliability of 
genetic evaluation would be lower than in the French and German systems. 
2.8 BREEDING OBJECTIVE 
2.8.1 Defining the breeding objective  
The breeding objective is defined as a list of traits and their relative importance in 
achieving the desired breeding goals (Hazel and Lush, 1942). The most common 
breeding objective traits used in sport horses are jumping and dressage ability, 
conformation, movement and temperament. These traits, in contrast to those selected for 
 
 
16 
 
in other livestock species, are often subjective and difficult to measure (Koenen et al., 
2003, Koenen and Aldridge, 2002, Rogers and Wickham, 1993a). 
2.8.2 Traits weightings 
The practical value of an animal is always affected by several traits (Hazel and Lush, 
1942). Therefore, as several traits must be considered, the use of a selection index to 
properly weight each trait in the model is a more efficient method of selection than 
selection for a single trait or selection for several traits each with independent culling 
levels. Hence, the use of multi trait objectives has developed in animal breeding (Hazel, 
1943). 
Traits included in the model which determine profit are weighted according to their 
relative economic importance to the breeding objective. Economic weightings are a 
measure of how much profit is expected to increase for each unit of improvement in the 
trait. Traits with a higher weighting have a larger influence on the final value (Harris 
and Newman, 1994). Economic weightings can be calculated using the first derivatives 
of profit with respect to the traits of interest. They can also be derived by incrementing 
the mean of the trait of interest by a small amount and calculating the difference in 
profit (Hazel and Lush, 1942, Visscher et al., 1994). It is also possible to calculate 
economic weightings by translating inputs and outputs into monetary terms (Pearson 
and Miller, 1981). 
2.8.3 Targets of the breeding objective 
The definition of the breeding objective directs the genetic improvements in the target 
population. For a breeding objective to be well defined, expected future demands must 
be taken into account (Koenen, 2010). Hence, the issue of who will benefit from any 
genetic improvements achieved arises. There are two viewpoints from which profits 
from genetic gain are seen; the national viewpoint and the individual producer’s 
viewpoint. From a national perspective, when supply and demand are in equilibrium 
then decreases in expenses are worth more than increases in production income, as there 
is a limited market for any additional output. However, to the individual producer, 
increases in production are much more valuable as increased production can be 
marketed without materially affecting price per unit to that individual (Pearson and 
Miller, 1981). In the sport horse industry increased production would be measured as a 
 
 
17 
 
unit of performance. Maximising genetic gain can result in lower than expected 
economic returns. This is often due to the inadequate or incorrect description of the 
breeding objective (Miller, 1977, Harris and Newman, 1994). Breeders, riders and 
national breeding organisations will benefit differently from improvements in the 
genetic merit of populations. 
2.8.4 Profit satisfiers and profit maximisers 
Recent years have seen a shift from viewing farming as a lifestyle with income attached 
toward viewing farming as a business in which maximum economic returns are the 
desired outcome. Hence, a shift has occurred away from profit satisfiers toward profit 
maximisers. However, in the sport horse industry, breeders tend to be passionate about 
horses and are involved in sport horse breeding for the non monetary lifestyle values it 
offers. Hence, sport horse breeders often tend to be profit satisfiers and their motivation 
to increase production efficiency as individuals and to breed toward a specifically 
defined breeding objective is limited (Rogers, 1993b). 
2.8.5 Examples of breeding objectives for sport horse populations 
In sport horses, breeding objectives are developed with the intention of improving 
performance in elite competition; however, the majority of riders only require a fun, 
easy sport horse for amateur competition. The traits selected to achieve these breeding 
objectives are qualitative in sport horses; hence, it is difficult to quantify improvement 
in an economical sense. The choice of traits also differs between studbooks, as the traits 
most highly correlated with success vary between the sport horse disciplines. For 
example, the Holsteiner stud book primarily aims to produce show jumping horses. 
Hence, their objectives differ from those of the Trakehner stud book which primarily 
aims at producing dressage horses (Koenen et al., 2003). Examples of breeding 
objectives for the Dutch Warmblood, Oldenburg, Holsteiner and Trakehner stud books 
are shown in table 2-4. 
 
 
 
 
18 
 
Table 2-4: Breeding objectives of some of the German studbooks (Dutch Warmblood, 
Oldenburg, Holsteiner and Trakehner) (Holsteiner-verband, 2003, KWPN, 2003, Oldenburger-
pferde, 2004, Trakehner-Verband, 2005). 
Stud book Breeding Objectives 
Dutch Warmblood 
Sport horses at the highest level of international 
competition in dressage and show jumping, with good 
conformation. 
Oldenburg 
A noble, high performance horse with active impulsion and 
elastic movements, which is suitable for any type of sport. 
Holsteiner 
An athletic, expressive riding horse to perform at national 
and international level particularly in show jumping. Good 
control and intelligence. Must have typical high knee action 
of the Holstein horse. 
Trakehner 
Sound, Trakehner type horse, big frame, correct body 
proportions, easy to ride. Elastic, energetic movement. 
Spirited but kind, willing to perform, tremendous stamina. 
2.9 RELATIVES’ PERFORMANCE 
Pedigree information is valuable in predicting an animal’s worth and performance 
potential as progeny will on average express half of the genes of each parent 
(Henderson, 1974). However, this is no guarantee of ability, as Mendelian sampling 
determines that the potential performance of progeny may be significantly better or 
worse than that of either parent (Mrode and Thompson, 2005). The risk of selecting 
animals based on pedigree information is that this can place too great an emphasis on 
distant relatives (mathematically less genetically significant), which can reduce the 
accuracy of selection. The pedigree records may also be biased, as progeny from 
favoured individuals may have received greater management and training; hence, 
showing superior performance to equally talented individuals on this environmental 
basis (Rogers, 1993b, Mrode and Thompson, 2005). The attention paid to records from 
a particular ancestor should depend on: 
 How closely related the ancestor is to the individual: coefficient of relationship 
for parents is ͳൗʹ , grandparents ͳ Ͷൗ  and great grandparents ͳൗͺ . 
 
 
19 
 
 Heritability of the trait: the higher the heritability, the greater the accuracy of 
selection for that trait based on ancestry (Mrode and Thompson, 2005, Rogers, 
1993b). 
Information from the performance of full and half sibs can be useful when estimating 
breeding values. The accuracy of selection is higher when a greater amount of relevant 
information is utilised in the breeding value calculations (Mrode and Thompson, 2005). 
However, the coefficient of relationship and heritability must be considered (as for 
pedigree records above) when deciding to include sibling records in breeding value 
calculations (Rogers, 1993b). 
2.10 THE IMPACT OF WRONG OR MISSING SIRE INFORMATION 
2.10.1 Influence on variance 
Wrong sire information (WSI) and missing sire information (MSI) influence the 
variance of EBVs of the sire and reduce the response to selection (Sanders et al., 2006). 
WSI and MSI may have a greater effect on genetic gain of lowly heritable traits, such as 
those commonly seen in sport horse populations, as in this case the impact of pedigree 
information on the EBV under BLUP is higher (Sanders et al., 2006, Woolliams, 2006).   
2.10.2 Reliability 
The reliability of EBVs calculated will be lower as the amount of WSI and MSI 
increases (Woolliams, 2006). In dairy cattle the reliability of EBVs has been calculated 
as 0.17-0.9 for heritability values of 0.1-0.5 with 0-30% of sire information being wrong 
or missing (Sanders et al., 2006). Hence, there is likely to be low reliability of EBVs 
calculated for New Zealand sport horses as the dataset is incomplete (Rogers and Firth, 
2005). 
2.11 HERITABILITY 
2.11.1 Definition of heritability 
The proportion of observed phenotypic variance that results from differences in 
genotype between individuals is known as heritability (Lush, 1949, Lynch and Walsh, 
1998). Investigations into the genetic and phenotypic parameters of competitive sport 
 
 
20 
 
horse performance are limited, particularly for competition performance from field test 
data. Heritability values for sport horse competition traits tend to be low (Ducro et al., 
2007). 
2.11.2 Heritability estimates in sport horses 
Heritability values derived from field test data in table 2-5 ranged from 0.1-0.27 for 
ability in dressage or show jumping. No heritability values are shown for eventing, as 
each of the three phases may have a separate heritability value and need to be 
considered separately, hence, investigations of eventing data are relatively rare 
(Kearsley et al., 2008). Low estimates of heritability are expected for sport horse 
competition traits, as the effect of rider influence on horse performance cannot be 
accurately accounted for (Ducro et al., 2007, Koenen et al., 1995). The criteria 
measured when field test data is analysed varies between countries. Earnings per start 
and number of points are the criteria used in France and Germany respectively 
(Huizinga and Vandermeij, 1989, Luehrs-Behnke et al., 2002).  
Using information from the youngest stallions and their oldest progeny (when these 
progeny reach 4-6 years of age) has been shown to give the most accurate measure of 
heritability while maintaining a reasonable generation interval. Older stallions have 
been subjected to more selection. Therefore; some stallions would have more progeny 
than others in competition as decisions on their relative strengths as sires have been 
made based on the performance of older progeny. Hence, heritability values tend to be 
less consistent when data from older stallions is utilised (Huizinga and Vandermeij, 
1989). Heritability values which have been calculated from field test data in sport horses 
are shown in table 2-5. 
Assumptions of heritability values of 0.16-0.36 for conformation and movement traits 
have previously been made for genetic analysis of New Zealand data based on European 
values (Rogers, 1993a). However, when doing this it is essential to note that European 
horses are commonly from uniform environments, as horses tend to be boxed and 
training methods are uniform. New Zealand tends to have a highly heterogeneous 
environment as horses tend to be paddocked and training methods are highly variable 
depending on individual preferences. Hence, heritability values calculated on European 
data may be higher than those applicable to New Zealand’s more heterogeneous 
environment (Rogers, 1993a). 
 
 
21 
 
Table 2-5: Heritability values of performance traits for sport horses based on field test data. 
Author Breed Age of 
progeny 
Discipline Criteria Heritability 
Estimate 
(Bruns, 1981) German 
sport horse 
< 8 years show jumping Earnings per start 
(log) 
0.14 
Relative place 
number 
0.14 
Place value 0.13 
dressage Earnings per start 
(log) 
0.15 
Relative place 
number 
0.27 
Place value 0.19 
(Huizinga and 
Vandermeij, 
1989) 
Dutch 
Warmblood 
4-6 years show jumping Cumulative 
lifetime total of 
scores (SQRT) 
0.20 
dressage Cumulative 
lifetime total of 
scores (SQRT) 
0.10 
(Ducro et al., 
2007) 
Dutch 
Warmblood 
All ages show jumping Classification 
scores (SQRT) 
0.14 
dressage Classification 
scores (SQRT) 
0.14 
(Olsson et al., 
2008) 
Swedish 
Warmblood 
All ages show jumping Accumulated 
lifetime 
upgrading points 
(log10) 
0.27 
dressage Accumulated 
lifetime 
upgrading points 
(log10) 
0.17 
 
 
22 
 
2.12 POTENTIAL SOURCES OF BIAS IN HERITABILITY ESTIMATES 
2.12.1 Assortative mating 
In the case of equestrian sport where a large number of sires contribute only one 
progeny record, heritability tends to be overestimated when utilising methods for 
regression of offspring on one parent and for analyses of variance. The paternal 
component of variance is higher when the mean number of offspring per stallion is 
lower. The number of offspring per stallion is highly variable. Homogamy is mostly due 
to poor quality stallions as they only serve poor quality mares (high quality stallions 
serve both high and lower quality mares). However, as stallions with small progeny 
numbers are numerous and rapidly replaced, they must be included in genetic analysis 
in order to sufficiently describe the population (Langlois, 1980b). 
2.12.2 Non random effect of environment 
The non random effect of environment is an extremely large source of bias in sport 
horse populations which leads to the overestimation of heritability (O'Ferrall and 
Cunningham, 1974). A major aspect of environment which affects sport horse 
performance is management of the stud farm. However, it is generally agreed that stud 
owners and managers take equally good care of all horses and that problems which arise 
could occur anywhere (Langlois, 1980b). Hence, the effect of management of stud 
farms may not be as great as anticipated. The sale of young stock is another major 
aspect of environment. Young horses which are sold at a high price are likely to end up 
in competitive homes and receive more or better training giving them a greater chance 
of becoming superior performers (Langlois, 1980b).  
2.13 METHODS OF DATA ANALYSIS 
2.13.1 Methods 
EBVs can be calculated via many models including the use of a sire or animal model 
using the BLUP or REML procedures.  
2.13.2 Best Linear Unbiased Prediction (BLUP) 
The selection index (Best Linear Prediction or BLP) is a useful procedure for 
calculating breeding values and has been widely used for genetic analysis in animal 
 
 
23 
 
breeding (Tavernier, 1988, Árnason, 1987, Henderson and Quaas, 1976). However, 
there are two major problems with the selection index method: 
 Records must be adjusted for fixed or environmental factors before doing 
calculations. Hence, it is assumed these factors are known when most often they 
are not known (Henderson, 1953). 
 Index equation solutions require calculation of the inverse of the covariance 
matrix for observations.  Determination of the inverse matrix is difficult when 
the data set is large (Mrode and Thompson, 2005, Henderson, 1974). 
BLUP was developed by Henderson in 1949 in order to overcome the problems seen in 
the BLP. BLUP means that fixed effects and breeding values can be estimated at the 
same time. The methodology for BLUP is very similar to that of BLP (Mrode and 
Thompson, 2005). The properties of BLUP are: 
 It maximises the correlation between true (G) and predicted (X) breeding value 
or minimises the prediction error variance [Var(G-X)]. 
 Predictions are linear functions of the observations. 
 Estimations are unbiased, both for the breeding value and the estimation of fixed 
effects. 
 Is a method of predicting the true breeding value (Mrode and Thompson, 2005). 
2.13.3 Restricted Maximum Likelihood (REML) 
In order to utilise BLUP to calculate genetic merit, the covariance structure of the model 
elements must be known. However, in practical situations these are usually unknown 
and must be estimated. Hence, restricted maximum likelihood (REML) emerged as the 
method of choice for variance component estimation in animal breeding (Neumaier and 
Groeneveld, 1998, Gilmour et al., 1995). 
2.13.4 Animal and Sire models 
An animal model utilises all available pedigree information (sire, dam, dam sire, 
progeny and so on) in genetic evaluation. A sire model only utilises progeny 
information of each sire. The more information available on an animal, the more 
accurate the EBV for that animal will be; hence, an animal model is preferable when 
calculating EBVs (Groeneveld et al., 2008). 
 
 
24 
 
2.14 BREEDING VALUES 
2.14.1 Definition 
A true breeding value is the sum of the values of all of an animal’s alleles, summed over 
a pair of alleles at a locus and over all loci that affect the trait of interest. True breeding 
values are not measureable; therefore, a measure of phenotypic performance is taken to 
provide a predictor of the true breeding value. This is known as the Estimated Breeding 
Value (EBV) (Lynch and Walsh, 1998). For stallions a minimum of approximately 5-10 
progeny records are required to achieve sufficient reliability in EBV calculation. The 
greater the number of progeny records for a stallion, the higher the reliability of the 
EBV (Sun et al., 2009, Phocas et al., 2008). The average number of records in horse 
populations in Europe is approximately 50 progeny per stallion and 5 progeny per 
broodmare (Langlois, 1983). 
2.14.2 Estimation of breeding values 
Breeding values are generally estimated using a computer programme such as SAS, 
ASREML or PEST (Stewart et al., 2010, Lynch and Walsh, 1998). These programmes 
solve the mixed model equations (MME) based on the model created by the user. 
Examples of models used in sport horse populations are shown in table 2-6. The general 
form of the model when there is one random effect is: 
yij = pi + aj + eij 
where:  yij = is the trait measured in the objective 
 pi  = the fixed effect of the i
th gender 
 aj = the random effect of the j
th animal 
 eij = random error effect 
The statistical model includes both genetic and non-genetic effects. The main non-
genetic effects used in the analysis of sport horses are age and gender. The model is 
often also adjusted for the effects of location, rider and permanent environment 
(Koenen, 2002a). The construction of the mixed model equations requires a matrix X 
(relates records to fixed effects) and a Z matrix (relates records to all animals). Least 
 
 
25 
 
Square Equations (LSE) must be set up in order to construct the MME (Mrode and 
Thompson, 2005). The LSE are: 
ቂܺ′ܺ ܺ′ܼܼ′ܺ ܼ′ܼቃ ൤
෠ܾ
ොܽ
൨=൤
ܺ′ݕ
ܼ′ݕ൨ 
The addition of A-1α, to the Z’Z in the LSE yields the MME, where A-1 is the inverse of 
the relationship matrix and α = ߪ௘
ଶ
ߪ௚ଶ
൘ . Solving the MME by direct inversion of the 
coefficient matrix yields the estimated breeding value for the trait of interest for each 
animal, as well as the fixed effect averages (Mrode and Thompson, 2005). 
 
 
 
26
  
 
 
C
ou
nt
ry
 o
f 
po
pu
la
ti
on
 s
tu
di
ed
 
D
is
ci
pl
in
e(
s)
 
M
od
el
/p
ro
gr
am
m
e 
F
ix
ed
 a
nd
 r
an
do
m
 e
ff
ec
ts
 in
cl
ud
ed
 in
 t
he
 m
od
el
 
A
ut
ho
r 
F
ra
nc
e 
R
ac
e 
ho
rs
es
 
E
-K
M
P
 m
od
el
 
A
ge
, g
en
de
r,
 s
ea
so
n 
L
an
gl
oi
s 
(1
98
0a
) 
G
er
m
an
y 
D
re
ss
ag
e,
 s
ho
w
 
ju
m
pi
ng
 a
nd
 
ev
en
tin
g 
- 
B
ir
th
 y
ea
r 
of
 s
ir
e,
 g
en
de
r,
 s
ea
so
n,
 e
ff
ec
t o
f 
br
ee
d 
so
ci
et
y,
 c
at
eg
or
y 
of
 to
ur
na
m
en
t. 
A
ge
 w
as
 in
cl
ud
ed
 
in
di
re
ct
ly
 b
y 
sp
li
tt
in
g 
th
e 
da
ta
se
t i
n 
tw
o 
by
 a
ge
 
B
ru
ns
 (
19
81
) 
T
he
 N
et
he
rl
an
ds
 
D
re
ss
ag
e 
an
d 
sh
ow
 
ju
m
pi
ng
 
R
E
M
L
 
A
ge
 c
la
ss
, d
is
tr
ic
t, 
nu
m
be
r 
of
 y
ea
rs
 in
 
co
m
pe
tit
io
n,
 s
ea
so
n,
 g
en
de
r,
 s
ir
e 
H
ui
zi
ng
a 
an
d 
V
an
de
rm
ei
j 
(1
98
9)
 
F
ra
nc
e 
S
ho
w
 ju
m
pi
ng
 
B
L
U
P
 
G
en
de
r,
 a
ge
, p
er
m
an
en
t e
nv
ir
on
m
en
t 
T
av
er
ni
er
 
(1
99
0a
) 
F
ra
nc
e 
R
ac
e 
ho
rs
es
 
(t
ro
tte
rs
) 
B
L
U
P
 a
ni
m
al
 m
od
el
 
N
o 
fi
xe
d 
ef
fe
ct
s 
w
er
e 
us
ed
 a
s 
pa
rt
ic
ul
ar
 c
on
di
tio
ns
 
of
 e
ac
h 
ra
ce
 w
er
e 
th
e 
sa
m
e 
fo
r 
ea
ch
 h
or
se
 a
nd
 
ef
fe
ct
 o
f 
ri
de
r 
ca
nn
ot
 b
e 
ut
ili
se
d 
in
 a
 d
at
as
et
 w
he
re
 
th
e 
m
aj
or
it
y 
of
 r
id
er
s 
ha
ve
 o
nl
y 
on
e 
ho
rs
e 
T
av
er
ni
er
 
(1
99
1)
 
N
or
w
ay
 
R
ac
e 
ho
rs
es
 
(t
ro
tte
rs
) 
M
at
e 
co
rr
ec
te
d 
si
re
 
m
od
el
 
G
en
de
r,
 b
ir
th
 y
ea
r 
K
le
m
et
sd
al
 
(1
99
4)
 
Ta
bl
e 
2-
6:
 G
en
et
ic
 e
va
lu
at
io
n 
sy
st
em
s 
th
at
 h
av
e 
be
en
 u
ti
li
se
d 
on
 h
or
se
 p
op
ul
at
io
n 
da
ta
. 
 
 
27
  
F
ra
nc
e 
D
re
ss
ag
e,
 s
ho
w
 
ju
m
pi
ng
 a
nd
 
ev
en
tin
g 
G
L
M
 in
 S
A
S
 
G
en
de
r,
 a
ge
, s
ea
so
n,
 m
on
th
 o
f 
bi
rt
h 
L
an
gl
oi
s 
an
d 
B
lo
ui
n 
(1
99
8)
 
Sw
ed
en
 
R
ac
e 
ho
rs
es
 
(t
ro
tte
rs
) 
R
E
M
L
 a
ni
m
al
 m
od
el
 
A
 v
ec
to
r 
of
 u
nk
no
w
n 
fi
xe
d 
sy
st
em
at
ic
 
en
vi
ro
nm
en
ta
l e
ff
ec
ts
 
A
rn
as
on
 (
19
99
) 
G
er
m
an
y 
M
ar
e 
pe
rf
or
m
an
ce
 
te
st
 
M
ul
ti
 tr
ai
t a
ni
m
al
 
m
od
el
 in
 V
C
E
 4
.2
.5
 
Se
as
on
, m
on
th
 a
nd
 p
la
ce
 o
f 
te
st
, a
ge
 g
ro
up
, 
br
ee
d 
D
ie
tl 
et
 a
l. 
(2
00
5)
 
B
el
gi
um
 
S
ho
w
 ju
m
pi
ng
 
B
ay
es
ia
n 
sa
m
pl
in
g 
us
in
g 
th
e 
G
ib
bs
 
sa
m
pl
er
 
A
ge
 o
f 
ho
rs
e,
 g
en
de
r,
 c
om
pe
ti
ti
on
, r
an
do
m
 
pe
rm
an
en
t e
nv
ir
on
m
en
ta
l e
ff
ec
t 
Ja
ns
se
ns
 e
t a
l. 
(2
00
7)
 
U
ni
te
d 
K
in
gd
om
 
E
ve
nt
in
g 
A
nt
ed
ep
en
de
nc
e 
m
od
el
 
of
 o
rd
er
 1
 u
si
ng
 
A
S
R
em
l 
G
en
de
r,
 a
ge
, c
om
pe
ti
ti
on
, s
ir
e,
 r
es
id
ua
l h
or
se
, 
ri
de
r 
K
ea
rs
le
y 
 e
t a
l. 
(2
00
8)
 
H
un
ga
ry
 
S
ho
w
 ju
m
pi
ng
 
R
ep
ea
ta
bi
lit
y 
an
im
al
 
m
od
el
 
G
en
de
r,
 b
re
ed
er
, a
ge
 o
f 
ho
rs
e,
 s
ea
so
n,
 p
la
ce
 o
f 
co
m
pe
ti
ti
on
, r
id
er
, p
er
m
an
en
t e
nv
ir
on
m
en
t  
P
os
ta
 e
t a
l. 
(2
00
9)
 
U
ni
te
d 
K
in
gd
om
 
D
re
ss
ag
e 
R
E
M
L
 u
si
ng
 th
e 
pr
og
ra
m
m
e 
A
SR
em
l 
G
en
de
r,
 c
om
pe
ti
ti
on
 s
ta
nd
ar
d,
 a
ge
, h
ei
gh
t 
St
ew
ar
t e
t a
l. 
(2
01
0)
 
Sw
ed
en
 
D
re
ss
ag
e 
an
d 
sh
ow
 
ju
m
pi
ng
 
D
M
U
 p
ac
ka
ge
 f
or
 
m
ul
tiv
ar
ia
te
 m
ix
ed
 
m
od
el
s  
B
ir
th
 y
ea
r 
an
d 
ge
nd
er
 
V
ik
lu
nd
 e
t a
l. 
(2
01
0)
 
 
 
28 
 
2.14.3 Advantages of the use of breeding values to the New Zealand sport horse 
industry 
If EBVs with adequate reliability can be calculated for New Zealand sport horses then 
they could be used to increase the accuracy of selection towards specified breeding 
objectives, assuming such objectives are in place. Increased genetic gain would ideally 
lead to the production of higher quality sport horses, which would increase the chances 
of success of New Zealand horses and riders in international competition. Increased 
success would result in greater recognition of New Zealand horses on the international 
market; hence, would lead to higher prices for export sales. This could also lead to an 
increase in the funding available for equestrian sport in New Zealand. The use of clearly 
defined EBVs also has the potential to develop a clear and accurate system for both 
breeders and buyers, which would allow them to make more accurate and informed 
decisions on breeding (Rogers and Firth, 2005). 
2.14.4 Transformations of sport horse data 
Sport horse competition data commonly lacks a normal distribution due to the high 
frequency of zero records that result from many horses earning zero points or prize 
money. The most common reasons for non-earning in sport horses are likely to be lack 
of talent, injury or unsuitable temperament (Janssens et al., 2007, Ricard and Fournet-
Hanocq, 1997). The non-normal distributions seen in sport horse performance data 
mean that data must frequently be transformed for analysis (Langlois and Blouin, 1998). 
In France, logarithmic transformed earnings are analysed (Tavernier, 1990a). In 
Germany, logarithmic transformed earnings per start are analysed (Huizinga and 
Vandermeij, 1989). Data transformations which have been effectively used on sport 
horse performance data are logarithmic, power and square root transformations as 
shown in table 2-7. 
 
 
 
 
 
 
 
29 
 
 
Table 2-6: Previous transformations of sport horse and racing horse performance data. 
Type of transformation Country of 
population 
studied 
Discipline(s) Authors who have 
used this method 
Logarithmic 
transformations 
United states Dressage, show 
jumping and 
eventing 
Hintz (1980) 
Germany Dressage, show 
jumping and 
eventing 
Bruns (1981) 
United States Race horses Buttram et al. 
(1988) 
France Sport horses and 
race horses 
Tavernier (1988, 
1989, 1991) 
France Race horses Langlois (1980a, 
1989) 
France Dressage, show 
jumping and 
eventing (1998) 
and race horses 
(2004) 
Langlois and 
Blouin (1998, 
2004) 
France Eventing Ricard and Chanu 
(2001) 
Sweden Dressage and 
show jumping 
Wallin et al. (2003) 
United 
Kingdom 
Dressage, 
eventing and 
show jumping 
Whitaker and Hill 
(2004) 
New Zealand Eventing Rogers and Firth 
(2005) 
France Show jumping Dubois and Ricard 
(2007) 
Power transformations 
Sweden Race horses 
(trotters) 
Arnason et al. 
(1989) 
Norway Race horses 
(trotters) 
Klemetsdal (1994) 
Sweden Race horses 
(trotters) 
Arnason (1999) 
Square root 
transformations 
The Netherlands Dressage and 
show jumping 
Huizinga and 
Vandermeij (1989) 
The Netherlands Race horses 
(trotters) 
Minkema (1975, 
1981) 
The Netherlands Dressage and 
show jumping 
Koenen et al. 
(1995) 
Germany Dressage, 
eventing and 
show jumping 
Luehrs-Behnke et 
al. (2002) 
 
 
30 
 
 
Materials and Methods 
3.1 DESCRIPTION OF THE SPORT HORSE POPULATION 
3.1.1 Dataset 
Official performance records for the 2008/09 and 2009/10 seasons for the disciplines of 
dressage, eventing and show jumping were obtained from ESNZ, the body responsible 
for administering equestrian sport within New Zealand. Data were provided as an 
electronic extract of each horse’s performance, total points and prize money won in the 
2008/09 and 2009/10 seasons and as another extract of rider and horse details which 
contained a unique registration number, horse’s name, sire name and breed code.  
3.1.2 Editing the datasets 
Within Microsoft Office Excel 2007, data were manipulated, cleaned and uploaded to a 
customised Microsoft Office Access 2007 database. Data editing involved removing 
records from ponies7 and horses with an unknown gender or sire. Performance records 
from stallions were also removed due to very small numbers. The spelling of sire names 
was also checked and corrected. Data linking horse performance with sire and rider 
records were extracted for analysis. Performance records from both seasons were 
combined to produce a dataset which included all records from the 2008/09 and 2009/10 
seasons for each discipline using Microsoft Office Excel 2007. Records on stallions, 
ponies, horses of unknown gender or age and horses with no sire listed were removed 
from the datasets from all three disciplines. A ‘new sire’ list was created for all horses 
registered with ESNZ in which spelling errors of sire names were corrected. New 
variables of sire ID, Year ID, official birth year and amateur (1) or professional (2) rider 
were created for each dataset. Variables which came from separate ESNZ data sheets 
were linked to the three datasets using the query design function of Microsoft Office 
Access 2007. These variables were gender of horse, rider ID and new sire name. 
                                                          
7 147.5cm or shorter when measured at the whither. 
 
 
31 
 
3.1.3 Investigation of opportunities for starts 
The official ESNZ Bulletin schedules (October 2010) and the Eventing Omnibus for the 
2010/11 spring season were used to investigate the average number of starts available 
and the opportunities to win points and/or prize money in dressage, eventing and show 
jumping during the 2010/11 spring competition season. Investigations were conducted 
using Microsoft Office Excel 2007 . The three disciplines were considered separately 
and each event available within each discipline was listed, along with all grades, points 
and/or prize money available at the event. When multiple events were held on the same 
day in different locations they were considered as one opportunity to start. 
3.1.4 Performance measures 
The criteria to quantify performance differed between the disciplines. For show jumping 
records were usually only available for horses which placed (1st-8th) in a competition, 
whereas for eventing and dressage an event was recorded every time they entered a 
competition. The performance data suitable for analysis was number of starts and total 
points awarded (dressage and eventing) and total annual prize money (show jumping). 
Within dressage, points are awarded according to the horses final percentage score for 
each event, (7, 5, 3 and 1 point irrespective of grade of competition). Points awarded in 
eventing use an ascending scale, with increasing points awarded with higher grades, 
such that a win earns 6, 9 or 12 points for Novice, Intermediate or Advanced grade. 
Prize money won in show jumping is positively associated with height of jumps and 
difficulty of competition, and is usually allocated with each placing generally receiving 
60% of the prize money given to the placing above it.   
Attempts were made to examine the effect of rider level on performance records. Riders 
were categorised into professional (riders of 3 or more horses in the season) and 
amateur (up to 2 horses registered for sport in a given season) (Rogers and Firth, 2005). 
However, these classifications were very subjective; hence, were not included in the 
model for genetic analysis. 
3.1.5 Statistical analysis 
Descriptive analysis was performed within Microsoft Office Excel, SAS v9.1.3 (SAS 
Institute, 2009) and PASW Statistics 18 (IBM Corporation, 2009). The raw data and the 
transformation of the raw data or the transformation of the raw data plus 1 were 
 
 
32 
 
examined for normality using the Shapiro-Wilk test. Data used for transformation were 
for all animals with performance records for the 2008/09 or the 2009/10 season.  Hence, 
all horses registered for either season that entered at least one competition (dressage and 
eventing) and all horses which placed during the season (show jumping) were included 
in the analyses. Show jumping horses with no records were not included as they were 
not available from the dataset. For all analysis significance was set at p <0.05. 
In this analysis a mixture of both the French and German systems were used to calculate 
EBVs in sport horses due to the type of data available for each discipline. For the 
disciplines of dressage and eventing information on points won and number of starts 
during the season was available. Therefore, EBVs can be calculated based on points per 
start, which permits separation of horses which won a large number of points in a few 
competitions from horses which have won the same number of points over many 
competitions. In show jumping, only the number of starts in which the horse was placed 
were recorded and available, therefore, number of starts cannot be utilised. Hence, 
EBVs were calculated for prize money won. 
3.2 ESTIMATED BREEDING VALUES 
3.2.1 Definition of the estimated breeding value traits 
Three traits were analysed for the calculation of estimated breeding values for New 
Zealand sport horse sires. Points per start (dpps and epps) and number of starts (dnst 
and enst) were analysed for dressage and eventing and total prize money (tpm) was 
analysed for show jumping. These traits were defined as: 
1. Points per start (dpps for dressage and epps for eventing): The average number 
of points per start the horse has earnt based on records from two competition 
seasons (2008/09 and 2009/10). Using average points per start allows the difference 
to be seen between a horse which earns 10 points over two competitions and a 
horse which earns 10 points over ten competitions. EBVs for dressage and eventing 
horses were based on this. A horse which earns a greater number of points per start 
is performing better in its chosen discipline. Data was transformed in such a way 
that EBVs are for log10(pps+1). 
 
 
 
33 
 
2. Number of starts (dnst for dressage and enst for eventing): One start is any time 
the horse begins a competition, whether they completed it or not, in which points 
are on offer. One start in eventing is one weekend of competition. One start in 
dressage is one dressage test. Therefore, in dressage a horse may have several starts 
on the same day. If new points are offered, this counts as a new start. Having more 
starts during the season suggests a horse is performing well in competition and is 
also indicative of soundness. The variable for analysis was log10(number of starts). 
 
3. Total prize money (tpm): An estimate of a horse’s ability to perform in 
competition based on the total amount of prize money the horse has earnt per 
season. Records were available from two seasons (2008/09 and 2009/10). EBVs for 
show jumping are based on this, as true start data is absent. A horse which earns 
more prize money is performing better in its chosen discipline. Data was 
transformed in such a way that EBVs are for log10(tpm+1). 
3.2.2 Model and method 
The model for genetic analysis was as follows: 
yij = season + gender + age + animal + eij 
where yij is log10(points/start+1), log10(number of starts) or log10(total prize money+1) 
for the ith animal in the jth season. The fixed effects were season, gender and age. Season 
was either 2008/09 (1) or 2009/10 (2). Gender was either mare (1) or gelding (2). In 
dressage and show jumping, for horses between 5 years and 18 years, age was defined 
in steps of one year. For horse 3-4 years old this effect was combined as a single age 
class because there were few records in these categories. Horses 19-24 years old were 
combined as a single age class for the same reason. The same procedure was followed 
for the eventing dataset but with horses 6-17 years defined in steps of one year and 4-5 
years old and 18-21 years old combined as two single age classes. The assumptions of 
the model were no inbreeding, no genetic groups, no covariates, unrelated sires and no 
full sibs (half sibs only). The assumption that there are no full sibs must be assumed in 
the absence of dam details. 
Data and pedigree files were created in Notepad. The order of variables in the data file 
was animal, sire, dam, season, gender, age, dpps or epps, dnst or enst (or tpm for show 
 
 
34 
 
jumping). The pedigree files contained animal, sire and dam in that order. As 
information on dam was not available, all dams were denoted by a zero. AIREML was 
used to calculate the EBVs (Johnson, 2010). Heritability was assumed at 0.15 for all 
traits (Bruns, 1981). These values were estimated from previous research, as they could 
not be calculated from the ESNZ data due to lack of convergence. EBV outputs from 
AIREML were then linked with the ESNZ datasets (Microsoft Office Excel 2007) to 
retrieve sire names via Microsoft Office Access 2007. The distribution of EBVs for 
each variable were tested for normality using the Shapiro-Wilk test with significance set 
at p<0.05.  
 
 
35 
 
Chapter 3  
Results 
4.1 DESCRIPTION OF THE SPORT HORSE DISCIPLINES BASED ON 
OPPORTUNITIES TO START 
4.1.1 Eventing 
In eventing it was found that the number of starts available decreased as level of 
competition increased. In the 2010/11 spring season there were 33 Novice, 17 
Intermediate and 5 Advanced level competition starts available nationally. The average 
points on offer for each grade increased as level of competition increased. The average 
points per start available for horses which place (1st-6th) were 3.5, 5 and 7 for Novice, 
Intermediate and Advanced grades respectively (Figure 4-1).  
 
Figure 3-1: Average number of points per start and number of starts available in eventing 
during the spring 2010/11 season, by level of competition. 
  
0
5
10
15
20
25
30
35
Novice Intermediate Advanced
Grade
Average points/start
Number of starts
 
 
36 
 
4.1.2 Dressage 
The number of tests that could be entered for in the 2010/11 competition season 
decreased as competition level increased. 26 tests were available at level 1 (Preliminary) 
and 10 tests at level 9 (Grand Prix) (Figure 4-2). Levels 7, 8 and 9 were often combined 
as one competition class where riders could choose which test to ride. The number of 
points won at each level was the same as 1, 3, 5 and 7 points are awarded to horses 
scoring over 57%, 60%, 63% and 67% respectively, regardless of grade. As the system 
for earning points is so different between the disciplines of dressage (points earnt by 
percentage and eventing (points earnt by placing), similar opportunities to earn points 
do not exist between the two disciplines and horses in dressage tend to win more points 
than those competing in eventing. Hence, EBVs are unlikely to be comparable across 
these two disciplines for New Zealand sport horse sires. 
 
 
 
 
 
 
 
 
 
 
 
Figure 3-2: Number of tests available at each level in dressage during the spring 2010/11 
season. 
 
 
  
0
5
10
15
20
25
30
1 2 3 4 5 6 7 8 9
Nu
m
be
r o
f t
es
ts
Level
 
 
37 
 
4.1.3 Show jumping 
In show jumping the number of starts available for the 2010/11 spring season appeared 
normally distributed with the greatest number of starts available at the 1.1-1.19m class. 
The average prize money on offer across all placings at a competition (1st-8th) increased 
from $6.03 for classes jumping the height of 0.8-0.89m to $164.44 for the 1.4-1.49m 
classes. The average prize money offered increased dramatically to $490.83 and $375 
for the 1.5-1.59m and 1.6-1.69m classes respectively (Figure 4-3). The use of total prize 
money as a measure of performance makes it unlikely that EBVs for show jumping 
sires will be comparable with those calculated for sires in dressage or eventing. 
 
Figure 3-3: Number of starts available and average money per start on offer for show jumping 
during the spring 2010/11 season. 
  
0
10
20
30
40
50
60
70
$-
$100.00 
$200.00 
$300.00 
$400.00 
$500.00 
$600.00 
.8-.89 .9-.99 1-1.09 1.1-1.191.2-1.291.3-1.391.4-1.491.5-1.591.6-1.69
nu
m
be
r o
f s
ta
rt
s a
va
ila
bl
e
Av
er
ag
e 
m
on
ey
 p
er
 st
ar
t
Height (m)
Average 
money 
per start
Number 
of starts 
oct-jan 
 
 
38 
 
4.2 ANALYSIS OF THE 2008/09 AND 2009/10 COMPETITON SEASONS 
4.2.1 Description of the population 
During the 2008/09 season there were 1123, 902 and 1326 horses registered for the 
equestrian disciplines of dressage, eventing and show jumping respectively. Sire was 
not recorded for 13.2%, 15.0% and 16.3% of registered horses within each discipline. 
There were 51, 115 and 43 riders with 3 or more horses registered for dressage, 
eventing and show jumping (Table 4-1). 26.8% of horses registered for show jumping in 
the 2008/09 season did not have a rider identification number available and were 
included as amateur. 
Table 3-1: Description of the data set for the equestrian disciplines of dressage, eventing and 
show jumping during the 2008/09 and 2009/10 competition seasons. 
 Discipline 
 Dressage Eventing Show jumping 
Parameter 2008/09 2009/10 2008/09 2009/10 2008/09 2009/10 
Number of horses 
registered 
1123 1472 902 1255 1326 1331 
Number of riders 860 1200 583 835 950 836 
Number of professional 
riders (3 or more horses 
registered) 
51 52 115 72 43 106 
Total number of starts 
within discipline during 
the season 
10400 13677 4133 5169 n/a n/a 
Horses with no sire 
listed (%) 
13.2% 14.2% 15.0% 10.5% 16.3% 17.2% 
Number of sires 
represented 
511 621 434 236 533 566 
Sires contributing only 1 
progeny record (%) 
70.8% 69.4% 63.6% 75.0% 63.8% 67.7% 
Horses earning zero 
points on record (%) 
1.8% 1.1% 64.6% 13% n/a n/a 
Horses earning no prize 
money on record (%) 
43.9% 56.6% 70.8% 16.3% 2.7% 5.6% 
 
 
39 
 
During the 2009/10 season there were 1472, 1255 and 1331 horses registered for the 
equestrian disciplines of dressage, eventing and show jumping respectively. Sire was 
not recorded for 14.2%, 10.5% and 17.2% of registered horses within each discipline. 
There were 52, 72 and 106 riders with 3 or more horses registered for dressage, 
eventing and show jumping (Table 4-1). 
4.2.2 Progeny numbers 
There were very few stallions registered for competition in all three disciplines. In 
dressage, there were 21 and 30 stallions with competition records for the 2008/09 and 
2009/10 seasons respectively. In eventing, there were 5 and 2 stallions with competition 
records for the 2008/09 and 2009/10 seasons respectively. In show jumping, there were 
22 and 26 stallions with competition records for the 2008/09 and 2009/10 seasons 
respectively. The number of progeny records available for each sire was on average 
very small indicating a highly heterogeneous sport horse population (Tables 4-2, 4-3 
and 4-4). The allocation of sire names was relatively inaccurate as there were 2487 
spelling errors from 11926 records (20.9%) for sire names in the original data set 
(records of all horses registered with ESNZ) which had to be corrected before analysis 
to enable linking of progeny groups.  
 
  
 
 
40 
 
Table 3-2: The ten stallions with the greatest number of progeny records (nprog) in dressage 
during the 2008/09 and 2009/10 seasons. 
Sire nprog 
Anamour 193 
Dream Boy 46 
Voltaire II 45 
Gymnastik Star 37 
HP Corlando 31 
Falkensee 22 
Worldwide P.B 21 
Littorio 18 
Jaguar 18 
Zabalou 17 
 
Table 3-3: The ten stallions with the greatest number of progeny records (nprog) in eventing 
during the 2008/09 and 2009/10 seasons. 
Sire nprog 
Voltaire II 17 
Drums of Time 17 
Grosvenor 15 
Anamour 14 
Hula Town 14 
Internet 12 
Lone Arrow 10 
Senor Pete 8 
Laughton's 
Legend 
8 
Ebony Groove 8 
 
Table 3-4: The ten stallions with the greatest number of progeny records (nprog) in show 
jumping during the 2008/09 and 2009/10 seasons. 
Sire nprog 
Voltaire II 69 
Ramirez 26 
Kingsway Diamond 25 
HP Corlando 24 
Littorio 24 
Anamour 23 
Laughton's Legend 21 
Kahurangi Valiant 18 
Hula Town 18 
Distelfink 18 
 
 
41 
 
(a)  
 
 
(b)  
 
 
 
 
 
 
(c) 
 
 
 
 
 
 
Figure 3-4: Histograms of raw (left) and transformed (log10 data plus 1) (right) data for the 
equestrian disciplines of (a) dressage (points/start); (b) eventing (points/start) and (c) show 
jumping (prize money). Data is combined for the 2008/09 and 2009/10 seasons. Normality 
curves show the ideally desired distribution for the given data.  
 
 
42 
 
(a) 
 
 
 
 
 
 
 
(b) 
 
 
 
 
 
(c) 
 
 
 
 
Figure 3-5: Histograms of log(points per start+1) with records valued at zero removed to show 
improved normality of excluding these records for horses in dressage (a) and eventing (b) and 
log(total money+1) for show jumping (c). Data is combined for the 2008/09 and 2009/10 
seasons. Normality curves show the ideally desired distribution for the given data. 
 
 
43 
 
 
(a) 
 
 
 
 
 
 
 
(b) 
 
 
 
 
 
 
 
(c) 
 
 
 
 
 
 
 
Figure 3-6: Histograms of the age distribution of horses competing in dressage (a), eventing (b) 
and show jumping (c). Data is combined for the 2008/09 and 2009/10 seasons. Normality 
curves show the ideally desired distribution for the given data. 
 
 
44 
 
(a)  
 
 
(b)  
 
 
Figure 3-7: Histograms for number of starts and log(number of starts) in dressage (a) and 
eventing (b). Data is combined for the 2008/09 and 2009/10 seasons. Normality curves show the 
ideally desired distribution for the given data. 
  
  
 
 
45 
 
4.2.3 Normality tests 
Tests for normality identified that none of the raw performance data or data on number 
of starts was normally distributed. Transformation of the data plus 1(log10) significantly 
improved the distribution of the performance data (Figures 4-4 and 4-7), though the 
large number of horses with zero performance records continued to skew the 
distributions (p<0.05). Removal of horses with zero records resulted in the transformed 
data approaching a normal distribution (p<0.05) (Figure 4-5). However, these records 
were retained for the calculation of EBVs as removing them would result in a biased 
progeny sample for each sire. 
4.2.4 Population age structure 
The population age structure for all three disciplines shows that there is a large loss of 
horses from competition at a young age. This is a common trend in sport horses as many 
horses are lost from competition early on as a result of lameness/musculoskeletal injury 
or lack of talent (Rogers and Firth, 2005). The official ages of competing horses ranged 
from 3-24 years for dressage, 4-21 years for eventing and 3-24 years for show jumping. 
The average age of a horse in competition was 9.6 years, 9.8 years and 9.3 years for 
dressage, eventing and show jumping respectively. The median age was 7 years for all 
three disciplines (Figure 4-6). 
4.2.5 Investigation of the variables for genetic analysis 
An investigation of the raw means of points per start, number of starts and total prize 
money was performed. The average number of points per start for stallions in dressage 
during the 2008/09 season was zero. The number of stallions in dressage competition 
was low in both seasons with 21 and 30 competing in the 2008/09 and 2009/10 seasons 
respectively (Figure 4-8). In eventing the average points earnt per start and number of 
starts appeared much lower than seen in dressage due to the different systems for 
earning points between the two disciplines and the different competition formats. The 
average number of points per start for stallions during the 2008/09 and 2009/10 seasons 
was zero. The number of stallions in competition was low in both seasons with 5 and 6 
competing in the 2008/09 and 2009/10 seasons respectively (Figure 4-9). In show 
jumping, the number of stallions in competition was low in both seasons with 22 and 26 
competing in the 2008/09 and 2009/10 seasons respectively (Figure 4-10). These zero 
 
 
46 
 
values suggest poor performance of the stallions. However, stallions are usually kept 
entire as they are potentially superior performers, hence, averages of zero points per 
start could be the result of the very small numbers of stallions. Stallions are also 
physiologically different from geldings; hence, should be considered as a separate 
gender group. Stallions as progeny records were removed from genetic analysis as 
grouping them with geldings could have overvalued them and stallion numbers were too 
low for them to be considered as a separate gender (Stewart et al., 2010). 
 
 
 
 
 
 
 
 
 
Figure 3-8: Breakdown of the variables included in the model for genetic analysis showing the 
average points per start and number of starts for mares, geldings and stallions in dressage during 
the 2008/09 and 2009/10 seasons. 
 
 
 
 
 
 
 
47 
 
 
 
 
 
 
 
 
 
Figure 3-9: Breakdown of the variables included in the model for genetic analysis showing the 
average points per start and number of starts for mares, geldings and stallions in eventing during 
the 2008/09 and 2009/10 seasons. 
 
 
 
 
 
 
 
 
Figure 3-10: Breakdown of the variables included in the model for genetic analysis showing the 
average prize money won by mares, geldings and stallions in show jumping during the 2008/09 
and 2009/10 seasons. 
 
 
48 
 
4.3 ESTIMATED BREEDING VALUES 
4.3.1 Description of the estimated breeding values calculated for New Zealand 
sport horse sires 
The limitations in the New Zealand dataset suggest that the EBVs calculated here 
should only be considered as initial indicators of the genetic merit of New Zealand sport 
horse sires. The variables dpps and dnst measured in dressage and epps and enst 
measured in eventing showed a smaller range of EBVs than the variable tpm in show 
jumping. The population means were zero to four decimal places in all three disciplines 
(Table 4-5). The standard errors of all EBVs were reasonably large, due in part to the 
fact that only 11.8%, 11.3% and 14.2% of sires in dressage, eventing and show jumping 
respectively contributed 5 or more progeny records for genetic analysis. 
Table 3-5: Percentile bands for the distribution of EBVs calculated for the New Zealand data 
set using data from the 2008/09 and 2009/10 seasons in dressage, eventing and show jumping. 
The variables are dpps and epps (log(points per start+1)), dnst and enst (log(number of starts)) 
and tpm (log(total prizemoney+1)). 
Dressage Eventing Show jumping 
dpps dnst epps enst tpm 
Top 10% 0.021 0.025 0.048 0.043 0.191 
Top 20% 0.011 0.015 0.020 0.026 0.105 
Top 30% 0.006 0.009 0.000 0.017 0.058 
Top 40% 0.003 0.004 -0.006 0.007 0.020 
Top 50% 0.000 -0.001 -0.009 -0.002 -0.010 
Top 60% -0.005 -0.005 -0.013 -0.008 -0.038 
Top 70% -0.009 -0.010 -0.016 -0.020 -0.075 
Top 80% -0.015 -0.018 -0.022 -0.031 -0.110 
Top 90% -0.021 -0.027 -0.030 -0.043 -0.175 
µpopulation -0.000008 -0.00001 0.00002 -0.000009 0.00002 
Range -0.066-0.158 -0.117-0.232 -0.076-0.182 -0.101-0.168 -0.523-0.993 
 
 
 
 
49 
 
4.3.2 Sire results 
The six top ranked sires based on EBVs for dpps in dressage were also included in the 
list of ten highest ranked sires based on number of progeny records available for 
analysis. The four top ranked sires based on EBVs for dnst were included in the ten 
highest ranked sires based on number of progeny records. This could in part be a 
reflection of the increased number of records available to calculate EBVs for these sires. 
Anamour had the greatest number of progeny records (193). Dream Boy had the highest 
EBV for dpps (0.158±0.062) and Anamour had the highest EBV for dnst (0.232±0.052). 
There were 5 records in dressage (for both dpps and dnst) rated in the top ten stallions 
based on progeny records for which negative EBVs were calculated (Tables 4-6 and 4-
7). 
Two of the ten highest ranked sires based on EBVs for epps in eventing were also 
included in the list of ten highest ranked sires based on number of progeny records 
available for analysis. These sires were Grosvenor and Voltaire II. Four of the ten top 
ranked sires based on EBVs for enst were included in the ten highest ranked sires based 
on number of progeny records. Voltaire II and Drums of Time had the greatest number 
of progeny records (17). A Touch of Hillbilly had the highest EBV for epps 
(0.182±0.099) and Grosvenor had the highest EBV for enst (0.168±0.109). There were 
4 records in eventing (for both epps and enst) rated in the top ten stallions based on 
progeny records for which negative EBVs were calculated (Tables 4-8 and 4-9). 
One of the ten highest ranked sires based on EBVs for tpm in show jumping was also 
included in the list of ten highest ranked sires based on number of progeny records 
available for analysis. This sire was Kahurangi Valiant. Voltaire II had the greatest 
number of progeny records (69) but did not feature on the list of ten best stallions by 
EBV. Wishing Well had the highest EBV for tpm (0.993±0.307). There were 3 records 
in show jumping (for tpm) rated in the top ten stallions based on progeny records for 
which negative EBVs were calculated. These were Kingsway Diamond (-0.013±0.264), 
Laughton’s Legend (-0.025±0.282) and Distelfink (-0.061±0.294) (Table 4-10). 
Histograms of the distribution of EBVs8 for each discipline show that the values 
obtained for dressage (dpps and dnst), eventing (epps and enst) and show jumping (tpm) 
approach normality around a mean approaching zero (p<0.05) (Figures 4-11 to 4-15).  
                                                          
8 The full EBV lists can be found on a CD at the back of this thesis. 
 
 
50
  
T
ab
le
 3
-6
: 
C
om
pa
ri
so
n 
of
 s
ir
es
 w
ith
 th
e 
gr
ea
te
st
 n
um
be
r 
of
 p
ro
ge
ny
 r
ec
or
ds
 (
np
ro
g)
 a
nd
 s
ir
es
 w
it
h 
th
e 
hi
gh
es
t E
B
V
s 
(f
or
 th
e 
va
ri
ab
le
 d
pp
s)
 f
or
 d
re
ss
ag
e.
 V
al
ue
s 
ar
e 
fr
om
 d
at
a 
co
m
bi
ne
d 
fo
r 
th
e 
20
08
/0
9 
an
d 
20
09
/1
0 
se
as
on
s.
 
        
 
T
op
 1
0 
st
al
lio
ns
 b
y 
po
pu
la
ri
ty
 
T
op
 1
0 
st
al
lio
ns
 b
y 
E
B
V
 
Si
re
 
E
B
V
dp
ps
 
SE
 
np
ro
g 
Si
re
 
E
B
V
dp
ps
 
SE
 
np
ro
g 
A
na
m
ou
r 
0.
15
2 
0.
04
0 
19
3 
D
re
am
 B
oy
 
0.
15
8 
0.
06
2 
46
 
D
re
am
 B
oy
 
0.
15
8 
0.
06
2 
46
 
G
ym
na
st
ik
 S
ta
r 
0.
15
6 
0.
06
4 
37
 
V
ol
ta
ir
e 
II
 
0.
07
9 
0.
06
1 
45
 
A
na
m
ou
r 
0.
15
2 
0.
04
0 
19
3 
G
ym
na
st
ik
 S
ta
r 
0.
15
6 
0.
06
4 
37
 
V
ol
ta
ir
e 
II
 
0.
07
9 
0.
06
1 
45
 
H
P
 C
or
la
nd
o 
0.
07
2 
0.
06
6 
31
 
W
or
ld
w
id
e 
P
.B
 
0.
07
8 
0.
07
1 
21
 
Fa
lk
en
se
e 
-0
.0
29
 
0.
07
0 
22
 
H
P
 C
or
la
nd
o 
0.
07
2 
0.
06
6 
31
 
W
or
ld
w
id
e 
P
.B
 
0.
07
8 
0.
07
1 
21
 
In
ju
st
ic
e 
0.
05
9 
0.
08
1 
4 
L
it
to
ri
o 
0.
01
1 
0.
07
3 
18
 
W
oo
ds
to
ck
 
R
ec
ar
o 
R
ee
l 
0.
05
4 
0.
08
1 
5 
Ja
gu
ar
 
-0
.0
27
 
0.
07
2 
18
 
R
iv
er
m
an
 
0.
05
3 
0.
08
3 
2 
Z
ab
al
ou
 
-0
.0
20
 
0.
07
3 
17
 
H
or
ac
e 
0.
05
2 
0.
08
1 
4 
 
 
51
  
T
ab
le
 3
-7
: 
C
om
pa
ri
so
n 
of
 s
ir
es
 w
ith
 th
e 
gr
ea
te
st
 n
um
be
r 
of
 p
ro
ge
ny
 r
ec
or
ds
 (
np
ro
g)
 a
nd
 s
ir
es
 w
it
h 
th
e 
hi
gh
es
t E
B
V
s 
(f
or
 th
e 
va
ri
ab
le
 d
ns
t)
 f
or
 d
re
ss
ag
e.
 V
al
ue
s 
ar
e 
fr
om
 d
at
a 
co
m
bi
ne
d 
fo
r 
th
e 
20
08
/0
9 
an
d 
20
09
/1
0 
se
as
on
s  
      .   
 
T
op
 1
0 
st
al
lio
ns
 b
y 
po
pu
la
ri
ty
 
T
op
 1
0 
st
al
lio
ns
 b
y 
E
B
V
 
Si
re
 
E
B
V
dn
st
 
S
E
 
np
ro
g 
Si
re
 
E
B
V
dn
st
 
SE
 
np
ro
g 
A
na
m
ou
r 
0.
23
2 
0.
05
2 
19
3 
A
na
m
ou
r 
0.
23
2 
0.
05
2 
19
3 
D
re
am
 B
oy
 
0.
11
7 
0.
08
0 
46
 
G
ym
na
st
ik
 S
ta
r 
0.
18
9 
0.
08
3 
37
 
V
ol
ta
ir
e 
II
 
-0
.0
10
 
0.
07
9 
45
 
W
or
ld
w
id
e 
P
.B
 
0.
15
2 
0.
09
0 
21
 
G
ym
na
st
ik
 S
ta
r 
0.
18
9 
0.
08
3 
37
 
D
re
am
 B
oy
 
0.
11
7 
0.
08
0 
46
 
H
P
 C
or
la
nd
o 
0.
02
3 
0.
08
5 
31
 
L
on
e 
A
rr
ow
 
0.
10
4 
0.
09
7 
12
 
Fa
lk
en
se
e 
0.
00
7 
0.
09
0 
22
 
W
itz
bo
la
 
0.
09
2 
0.
09
9 
8 
W
or
ld
w
id
e 
P
.B
 
0.
15
2 
0.
09
0 
21
 
S
al
ut
at
io
n 
0.
08
4 
0.
10
3 
7 
L
it
to
ri
o 
0.
02
6 
0.
09
3 
18
 
W
el
tm
ey
er
 
0.
08
3 
0.
10
2 
5 
Ja
gu
ar
 
-0
.0
41
 
0.
09
3 
18
 
G
en
iu
s 
0.
07
3 
0.
09
8 
10
 
Z
ab
al
ou
 
0.
01
8 
0.
09
3 
17
 
M
cH
am
is
h 
0.
06
1 
0.
10
3 
4 
 
 
52
  
T
ab
le
 3
-8
: 
C
om
pa
ri
so
n 
of
 s
ir
es
 w
ith
 th
e 
gr
ea
te
st
 n
um
be
r 
of
 p
ro
ge
ny
 r
ec
or
ds
 (
np
ro
g)
 a
nd
 s
ir
es
 w
it
h 
th
e 
hi
gh
es
t E
B
V
s 
(f
or
 th
e 
va
ri
ab
le
 e
pp
s)
 f
or
 e
ve
nt
in
g.
 V
al
ue
s 
ar
e 
fr
om
 d
at
a 
co
m
bi
ne
d 
fo
r 
th
e 
20
08
/0
9 
an
d 
20
09
/1
0 
se
as
on
s.
 
            
T
op
 1
0 
st
al
lio
ns
 b
y 
po
pu
la
ri
ty
 
T
op
 1
0 
st
al
lio
ns
 b
y 
E
B
V
 
Si
re
 
E
B
V
ep
ps
 
SE
 
np
ro
g 
Si
re
 
E
B
V
ep
ps
 
SE
 
np
ro
g 
V
ol
ta
ir
e 
II
 
0.
13
0 
0.
07
5 
17
 
A
 T
ou
ch
 o
f 
H
il
lb
il
ly
 
0.
18
2 
0.
09
9 
2 
D
ru
m
s 
of
 T
im
e 
0.
00
6 
0.
07
4 
17
 
G
ro
sv
en
or
 
0.
13
6 
0.
07
7 
15
 
G
ro
sv
en
or
 
0.
13
6 
0.
07
7 
15
 
T
ur
bu
le
nt
 
D
an
ce
r 
0.
13
5 
0.
09
0 
6 
A
na
m
ou
r 
0.
01
6 
0.
07
8 
14
 
V
ol
ta
ir
e 
II
 
0.
13
0 
0.
07
5 
17
 
H
ul
a 
T
ow
n 
-0
.0
76
 
0.
07
8 
14
 
W
ai
ki
ki
 S
ta
r 
0.
12
9 
0.
08
9 
6 
In
te
rn
et
 
0.
07
9 
0.
08
2 
12
 
L
an
th
am
 
0.
12
0 
0.
09
9 
2 
L
on
e 
A
rr
ow
 
-0
.0
43
 
0.
08
3 
10
 
T
el
es
un
 
0.
11
1 
0.
09
3 
4 
S
en
or
 P
et
e 
0.
06
8 
0.
08
7 
8 
B
ah
ha
re
 
0.
10
8 
0.
09
6 
3 
L
au
gh
to
n’
s 
L
eg
en
d 
0.
04
2 
0.
08
6 
8 
H
aa
ji
is
an
 
0.
10
1 
0.
09
6 
3 
E
bo
ny
 G
ro
ov
e 
0.
03
8 
0.
08
6 
8 
E
ng
ag
em
en
t 
0.
09
4 
0.
09
9 
2 
 
 
53
  
T
ab
le
 3
-9
: 
C
om
pa
ri
so
n 
of
 s
ir
es
 w
ith
 th
e 
gr
ea
te
st
 n
um
be
r 
of
 p
ro
ge
ny
 r
ec
or
ds
 (
np
ro
g)
 a
nd
 s
ir
es
 w
it
h 
th
e 
hi
gh
es
t E
B
V
s 
(f
or
 th
e 
va
ri
ab
le
 e
ns
t)
 f
or
 e
ve
nt
in
g.
 V
al
ue
s 
ar
e 
fr
om
 d
at
a 
co
m
bi
ne
d 
fo
r 
th
e 
20
08
/0
9 
an
d 
20
09
/1
0 
se
as
on
s.
 
         
 
T
op
 1
0 
st
al
lio
ns
 b
y 
po
pu
la
ri
ty
 
T
op
 1
0 
st
al
lio
ns
 b
y 
E
B
V
 
Si
re
 
E
B
V
en
st
 
SE
 
np
ro
g 
Si
re
 
E
B
V
en
st
 
SE
 
np
ro
g 
V
ol
ta
ir
e 
II
 
0.
11
8 
0.
10
8 
17
 
G
ro
sv
en
or
 
0.
16
8 
0.
10
9 
15
 
D
ru
m
s 
of
 T
im
e 
0.
10
0 
0.
10
6 
17
 
A
na
m
ou
r 
0.
13
1 
0.
11
0 
14
 
G
ro
sv
en
or
 
0.
16
8 
0.
10
9 
15
 
V
ol
ta
ir
e 
II
 
0.
11
8 
0.
10
8 
17
 
A
na
m
ou
r 
0.
13
1 
0.
11
0 
14
 
C
ha
rl
to
n 
Ja
ve
li
n 
0.
11
1 
0.
12
2 
6 
H
ul
a 
T
ow
n 
-0
.0
05
 
0.
11
0 
14
 
L
it
tl
e 
B
ro
w
n 
Ju
g 
0.
11
1 
0.
12
4 
5 
In
te
rn
et
 
0.
07
1 
0.
11
4 
12
 
K
in
gd
om
 B
ay
 
0.
10
3 
0.
12
4 
5 
L
on
e 
A
rr
ow
 
-0
.0
33
 
0.
11
5 
10
 
N
K
 P
ri
nc
e 
Fe
rd
in
an
d 
0.
10
1 
0.
12
3 
6 
S
en
or
 P
et
e 
0.
06
4 
0.
11
8 
8 
D
ru
m
s 
of
 T
im
e 
0.
10
0 
0.
10
6 
17
 
L
au
gh
to
n’
s 
L
eg
en
d 
0.
04
7 
0.
11
9 
8 
Sp
ec
ta
cu
la
r 
L
ov
e 
0.
08
7 
0.
12
7 
4 
E
bo
ny
 G
ro
ov
e 
0.
03
3 
0.
11
8 
8 
A
st
on
 M
ag
nu
m
 
0.
08
3 
0.
12
2 
5 
 
 
54
  
T
ab
le
 3
-1
0:
 C
om
pa
ri
so
n 
of
 s
ir
es
 w
ith
 th
e 
gr
ea
te
st
 n
um
be
r 
of
 p
ro
ge
ny
 r
ec
or
ds
 (
np
ro
g)
 a
nd
 s
ir
es
 w
it
h 
th
e 
hi
gh
es
t E
B
V
s 
(f
or
 th
e 
va
ri
ab
le
 tp
m
) 
fo
r 
sh
ow
 ju
m
pi
ng
. 
V
al
ue
s 
ar
e 
fr
om
 d
at
a 
co
m
bi
ne
d 
fo
r 
th
e 
20
08
/0
9 
an
d 
20
09
/1
0 
se
as
on
s.
 
       
T
op
 1
0 
st
al
lio
ns
 b
y 
po
pu
la
ri
ty
 
T
op
 1
0 
st
al
lio
ns
 b
y 
E
B
V
 
Si
re
 
E
B
V
tp
m
 
SE
 
np
ro
g 
Si
re
 
E
B
V
tp
m
 
SE
 
np
ro
g 
V
ol
ta
ir
e 
II
 
0.
20
5 
0.
19
2 
69
 
W
is
hi
ng
 W
el
l 
0.
99
3 
0.
30
7 
16
 
R
am
ir
ez
 
0.
32
1 
0.
26
1 
26
 
S
al
ut
e 
th
e 
S
ta
rs
 
0.
71
2 
0.
31
8 
13
 
K
in
gs
w
ay
 
D
ia
m
on
d 
-0
.0
13
 
0.
26
4 
25
 
S
an
 M
at
eo
 
Se
co
nd
 C
ha
nc
e 
0.
61
9 
0.
38
3 
4 
H
P
 C
or
la
nd
o 
0.
25
3 
0.
26
6 
24
 
C
ab
du
la
 d
u 
T
ill
ar
d 
0.
6 
0.
33
9 
9 
L
it
to
ri
o 
0.
17
3 
0.
26
4 
24
 
O
m
ni
co
rp
 
0.
57
7 
0.
33
1 
11
 
A
na
m
ou
r 
0.
11
7 
0.
27
4 
23
 
K
ah
ur
an
gi
 
V
al
ia
nt
 
0.
54
6 
0.
29
6 
18
 
L
au
gh
to
n’
s 
L
eg
en
d 
-0
.0
25
 
0.
28
2 
21
 
In
do
ct
ro
 
0.
48
6 
0.
37
2 
5 
K
ah
ur
an
gi
 
V
al
ia
nt
 
0.
54
6 
0.
29
6 
18
 
Z
ig
 Z
ag
 
0.
47
8 
0.
38
3 
4 
H
ul
a 
T
ow
n 
0.
17
8 
0.
28
9 
18
 
Fe
tic
he
 d
u 
pa
s 
0.
47
2 
0.
34
8 
8 
D
is
te
lf
in
k 
-0
.0
61
 
0.
29
4 
18
 
L
eo
 C
ay
lo
n 
0.
44
9 
0.
33
6 
10
 
 
 
55 
 
4.3.3 The distribution of estimated breeding values for dressage sires 
 
 
 
 
 
 
 
 
 
 
Figure 3-11: Histogram of the distribution of EBVs calculated from New Zealand data over the 
2008/09 and 2009/10 competition seasons for the variable log(points per start+1) in dressage 
(dpps). 
 
 
 
 
 
 
 
 
 
 
 
Figure 3-12: Histogram of the distribution of EBVs calculated from New Zealand data over the 
2008/09 and 2009/10 competition seasons for the variable log(number of starts) in dressage 
(dnst).  
 
 
56 
 
4.3.4 The distribution of estimated breeding values for eventing sires 
 
 
 
 
 
 
 
 
 
 
Figure 3-13: Histogram of the distribution of EBVs calculated from New Zealand data over the 
2008/09 and 2009/10 competition seasons for the variable log(points per start+1) in eventing 
(epps). 
 
 
 
 
 
 
 
 
 
 
Figure 3-14: Histogram of the distribution of EBVs calculated from New Zealand data over the 
2008/09 and 2009/10 competition seasons for the variable log(number of starts) in 
eventing(enst) . 
 
 
57 
 
4.3.5 The distribution of estimated breeding values for show jumping sires 
 
 
 
 
 
 
 
 
 
Figure 3-15: Histogram of the distribution of EBVs calculated from New Zealand data over the 
2008/09 and 2009/10 competition seasons for the variable log(total prize money+1) in show 
jumping (tpm). 
 
 
 
 
 
 
58 
 
Chapter 4  
Discussion 
5.1 SYSTEM OF EARNING POINTS AND PRIZE MONEY AND STARTS 
AVAILABILE IN NEW ZEALAND SPORT HORSE COMPETITIONS 
5.1.1 Starts, points and prize money availability 
The number of starts available and the number of points and amount of prize money 
available during the 2010/11 spring season were found to differ between competition 
grades for all three disciplines. In eventing, it was found that the average number of 
points per start available increased as competition level increased (from novice to 
advanced). However, the number of starts available decreased as competition level 
increased (Figure 4-1). Based on this, talented eventing horses competing at advanced 
level may not appear sufficiently superior to horses in lower grades based solely on 
points data.  
In dressage, the same trend was observed, the number of tests available decreased as 
competition level increased (from level 1 to level 9). The points available per start were 
the same for each level (Figure 4-2). In show jumping it was also seen that the number 
of starts available decreased as competition level increased (apart from in very low 
grades where few starts were offered) and average prize money on offer increased as 
level increased (Figure 4-3). This is useful for genetic analysis as horses earning more 
prize money are likely to be more talented than horses earning less. Including 
competition level in the genetic model would standardise the points earnt per start 
across each level. Hence, competition level was excluded from the model as 
competition levels test different aptitudes and points earnt in a higher grade should hold 
greater weight (Aldridge et al., 2000). 
There were a similar number of starts available in novice eventing (33) as in level 1 
dressage (26) and advanced eventing (5) and level 9 dressage (10). Hence, the 
opportunities to earn points were similar in both disciplines. In show jumping, the 
highest number of starts was much higher than in eventing and dressage (65 in the 
1.10m-1.19m division) but there was only 1 start in the highest grade (1.60-1.69m). 
However, how the number of starts compares to the other disciplines is of little 
 
 
59 
 
relevance as the system for show jumping is so different from that of eventing and 
dressage and analysis must be based on prize money instead of points per start.  
5.1.2 Comparisons across disciplines 
International comparisons of EBVs are difficult due to different data recording systems 
in different countries (Koenen, 2002a). Similarly, EBVs will not be comparable across 
disciplines for New Zealand sport horse sires due to different data recording systems 
between disciplines. 
5.2 THE UNIQUE SITUATION IN EQUESTRIAN SPORT 
5.2.1 Normality of the data 
Points per start data for dressage and eventing and total prize money data for show 
jumping had a highly skewed distribution, with the majority of horses earning no points 
or prize money and a select few which were high earners (Figure 4-4). The same highly 
skewed distribution was seen for number of starts data analysed for dressage and 
eventing, with a large number of horses only starting once during the season (Figure 4-
7). Hence, transformation of the points, prize money and number of starts data was 
required for genetic analysis (Langlois and Blouin, 1998). Of the three transformations 
attempted (logarithmic (base 10), square root and power (x0.2)), the logarithmic 
transformation was found to be most effective for improving normality of the data. This 
is consistent with previous research, as logarithmic transformations have been widely 
used as a scale on which to measure horse performance (Tavernier, 1991). 
Very non-normal distributions are common in equestrian sport data due to the high 
frequency of zero records. Assigning horses which earnt zero points or prize money 
with a zero record categorises these horses as bad performers. However, care should be 
taken when doing this, as some of these horses will simply not yet have been tested in 
competition (Ricard and Chanu, 2001). There were no missing records on points or 
prize money in the ESNZ dataset and all horses had at least one start during the 
competition season, hence, a zero record was assumed to be the result of poor 
performance. Therefore, as in the study of Klemetsdal (1994) and Minkema (1989) zero 
records were assigned to all nonearning or nonscoring horses. The main reasons for 
 
 
60 
 
nonearning in sport horses are likely to be lack of talent, injury or unsuitable 
temperament (Janssens et al., 2007, Ricard and Fournet-Hanocq, 1997).  
In the case of points and prize money, one was added to each record before 
transformation to enable logarithmic transformation of zero records. These zero records 
meant that the data still showed a non-normal distribution after transformation. 
Removal of these zero records improved the normality of the data (Figure 4-5). 
However, these records were retained for genetic analysis, as removing them would lead 
to a biased sample of progeny from each sire (Aldridge et al., 2000). The effect of zero 
records on normality was greatest for the eventing data, as the proportion of zero 
records was much higher than in dressage or show jumping. Hence, the distribution of 
eventing data changed significantly (appeared far more normal) when zero records were 
removed. Logarithmic transformation of number of starts data resulted in less than ideal 
normality of the data for analysis, particularly for eventing (P<0.05) (Figure 4-7). 
The age distribution of the sport horse population was similar for all three disciplines 
and showed that a small number of horses started in competition at a young age (3 years 
in dressage and show jumping or 3-4 years in eventing) and a large number of horses 
competed in the general range of 4 to 19 years of age (Figure 4-6). Very few horses 
were retained in competition past the age of 19 years. Mares in particular are often 
removed from sport early to be used for breeding (O’Brien et al., 2005). There were 
large losses of horses at a young age which is common in sport horse populations. The 
main reasons for loss of horses from competition at a young age are 
lameness/musculoskeletal injury and lack of talent (Rogers and Firth, 2005). 
5.2.2 Heterogeneity of the sport horse population 
Sport horse populations tend to be relatively heterogeneous (Wallin et al., 2003). The 
New Zealand data set was found to be particularly heterogeneous with 63.6%-75% of 
sires contributing only one progeny record (Table 4-1) and only 11.3% to 14.2% of sires 
had 5 or more progeny records. A minimum of approximately 5-10 progeny records is 
required to achieve sufficient reliability in EBV calculation (Sun et al., 2009, Phocas et 
al., 2008). Hence, the reliability of the EBVs for the majority of sport horse sires in 
New Zealand, based on current data, would have been extremely low (Woolliams, 
2006). This heterogeneity will partly be the result of missing information on sire 
affecting the data records. 
 
 
61 
 
The number of progeny competing in sport horse disciplines demonstrates how 
heterogeneous the New Zealand sport horse population is with even popular stallions in 
each discipline having relatively few progeny. Anamour’s progeny in dressage were the 
only exception with 92 and 101 progeny competing during the 2008/09 and 2009/10 
seasons respectively. 
5.2.3 Missing records 
There were many missing or inaccurate sire records in the ESNZ database. Information 
on sire was not available for 3.82% to 17.2% of horses competing in dressage, eventing 
and show jumping (Table 4-1). Horses with an unknown sire had to be excluded from 
genetic analysis; hence, progeny information from these horses is missing, which 
influences the variance of sire EBVs, leading to a reduced response to selection (Harder 
et al., 2005, Sanders et al., 2006). Information on sire was not available as either it was 
not provided when the horse was registered or the sire was simply not known. 
Inaccurate spelling of sire names was a large problem within the New Zealand dataset. 
The impact of wrong sire information on reliability and genetic gain has been found to 
be twice that of missing sire information (Woolliams, 2006). Across all three disciplines 
20.9% of horses registered with ESNZ had incorrectly spelt sire names. If the spelling 
of sire names is not corrected prior to analysis, different versions of the same sire will 
be counted as different sires, as no sire identification number is provided in the dataset. 
Hence, the sport horse population will appear more heterogeneous and EBVs will be 
based on incomplete progeny samples. 
5.2.4 The effect of rider 
Equestrian sport is unique in the animal breeding sector in that there is a confounding 
effect of the rider on sport horse performance (Kearsley et al., 2008). This rider effect 
makes it difficult to separate a horse’s own talent from that of its rider, as more 
experienced riders tend to achieve better results on less talented horses than less 
experienced riders. Horses which are thought to be better performers also often end up 
with better riders (Bruns, 1981). Hence, horses are often categorised as ridden by either 
amateur or professional riders for genetic analysis. Categorisation of riders as amateur 
or professional is simple in European systems, as professional riders are registered 
under a different licence to amateur riders and they compete in separate classes 
 
 
62 
 
(Federation Francais d'Equitation, 2009). However, in New Zealand all riders are 
registered under the same licence and compete in the same classes (Equestrian Sports 
New Zealand, 2009b). Hence, riders could only be assumed to be amateurs or 
professionals based arbitrarily on the number of horses they registered to ride in a given 
discipline during the season.  
Riders were assumed to be amateurs if they registered one or two horses during the 
season and professional if they registered three or more horses during the season (Table 
4-1). This cut off was decided based on the sudden decrease in the number of riders 
registering three or more horses, as the number of professional riders was assumed to be 
much smaller than that of amateurs in all disciplines. Rider identification numbers were 
only available for 26.8% of riders in show jumping during the 2008/09 season, and 
therefore all horses with unknown riders were categorised as amateurs for analysis. As 
horses ridden by professionals are assumed to be on average greater earners of points 
and prize money, classifying a horse ridden by an unknown rider as having an amateur 
rider is less presumptuous. The classification of amateur and professional riders was so 
ambiguous for the New Zealand dataset that rider effect was not included in the model 
to calculate EBVs, as it would have introduced a large source of bias (Rogers and Firth, 
2005).  
5.2.5 Stallions 
There were very few stallions in competition (between 2 and 30) in all three disciplines 
in New Zealand. This is very different from European countries where competing 
stallions is common and up to 89% of sires had a performance record of their own by 4 
years of age (Dubois and Ricard, 2007). Due to the small numbers of stallions in the 
New Zealand dataset, the average points and prize money earnt by stallions were zero in 
many cases, particularly in eventing. Hence, the gender stallion could not be included in 
the genetic model. Categorising stallions and geldings together as ‘male’ could 
potentially overvalue the stallions, as there could be selection for stallions remaining 
entire because they are believed to be superior individuals. Stallions are also 
physiologically different from geldings. For these reasons, stallions as progeny records 
were excluded from genetic analysis (Stewart et al., 2010). The effect of gelding is 
important in the New Zealand sport horse population as it introduces a high selection 
 
 
63 
 
rate prior to entry into competition; hence, it reduces the effectiveness of utilising 
competition results for stallion selection (Rogers and Firth, 2005). 
5.2.6 Mares 
There were few horses registered with ESNZ which included information on the dam. 
Hence, dams were all assumed to be unknown and all sires were assumed unrelated for 
genetic analysis. Mares often are removed from competition early to become breeding 
animals if they are not successful competitors or suffer an injury. Hence, there could be 
a bias towards mares in competition performing better, on average, than geldings as 
only successful mares are kept in competition (Hennessy and Quinn, 2007). This also 
leads to the problem of a long generation interval for successful competition mares 
(Langlois, 1983). 
5.3 INTERPRETATION OF ESTIMATED BREEDING VALUES 
5.3.1 Usefulness of estimated breeding values 
The use of EBVs by the New Zealand sport horse population would lead to 
improvements in the selection of competition sires, as sires with reasonable EBVs can 
be used more effectively to achieve genetic gain, particularly in show jumping. 
However, more accurate and complete recording of sport horse pedigree information is 
required in order for the New Zealand dataset to be used to calculate breeding values 
with reasonable reliability. The issues of normality and heterogeneity are common to all 
sport horse breeding industries; hence, must be dealt with in the genetic model (Wallin 
et al., 2003, Langlois and Blouin, 1998). The range of breeding values was very small 
(Table 4-5) for all variables and all disciplines; therefore, there appear to be limited 
opportunities to achieve genetic gain in the current New Zealand sport horse population. 
Hence, there may be insufficient differences for effective selection (Van Vleck et al., 
1992).  
Many traits analysed in sport horse competition data are subjective and difficult to 
measure (Koenen et al., 2003). The variables analysed were dpps and epps (log(points 
per start+1)), dnst and enst (log(number of starts)) and tpm (log(total prize money+1)). 
Positive EBVs were desired, as positive increases in the number of points and amount 
of prize money won would be indicative of improved performance. Horses which are 
 
 
64 
 
capable of a greater number of starts during the season could also be desirable, as this 
may suggest that the horse is successful in competition and could also be indicative of 
soundness (Dubois and Ricard, 2007).  
5.3.2 Heritability value assumptions 
As the dataset was not sufficiently complete to calculate values for heritability, a 
heritability value of 0.15 was assumed for all variables (dpps, epps, dnst, enst and tpm). 
This lack of convergence was also encountered by Koerhuis and Van der Werf (1994) in 
the bivariate analysis of a simulated horse population. Assumptions of heritability 
values of 0.16-0.36 for conformation and movement traits have previously been made 
for genetic analysis of New Zealand data based on European values (Rogers, 1993a). A 
heritability value of 0.15 was chosen based on heritability values which have previously 
been calculated for sport horse competition variables from overseas data which ranged 
from 0.10-0.27 (Bruns, 1981, Huizinga and Vandermeij, 1989, Ducro et al., 2007, 
Olsson et al., 2008). As European sport horses are commonly from uniform 
environments, heritability values may be higher than those applicable to New Zealand’s 
more heterogeneous environment (Rogers, 1993a). To account for this, a lower 
heritability value (0.15) from within the range generally seen in Europe was used. The 
EBVs calculated appeared reasonably normally distributed around a mean near zero 
(Figures 4-11 to 4-15). 
5.3.3 Individual sire indicators 
Based on the EBVs calculated in this analysis, there appear to be sires in dressage, 
eventing and show jumping which could be utilised more effectively in order to 
improve genetic gain in the sport horse population. Sires which were included in the ten 
highest EBVs in dressage based on dpps which could be utilised more effectively were 
Injustice, Woodstock Recaro Reel, Riverman and Horace. Based on dnst, stallions 
which could be utilised more effectively were Lone Arrow, Witzbola, Salutation, 
Weltmeyer, Genius and McHamish (Tables 4-6 and 4-7). 
Sires which were included in the ten highest EBVs in eventing based on epps which 
could be utilised more effectively were A Touch of Hillbilly, Turbulent Dancer, Waikiki 
Star, Lantham, Telesun, Bahhare, Haajiisan and Engagement. Based on enst, sires 
which could be utilised more effectively were Charlton Javelin, Little Brown Jug, 
 
 
65 
 
Kingdom Bay, NK Prince Ferdinand, Spectacular Love and Aston Magnum (Tables 4-8 
and 4-9). 
Sires which were included in the ten highest EBVs in show jumping based on tpm 
which could be utilised more effectively were Wishing Well, Salute the Stars, San 
Mateo Second Chance, Cabdula du Tillard, Omnicorp, Indoctro, Zig Zag, Fetiche du 
pas and Leo Caylon (Table 4-10). 
5.3.4 Multi-trait animal model 
A multi-trait animal model could be useful for future genetic evaluation of New Zealand 
sport horses, assuming sufficient data can be made available to effectively utilise this 
model. Effective use of a multi-trait animal model would make it possible to predict 
EBVs with better reliability; hence, greater increases in genetic gain would be seen. In 
this way, the New Zealand sport horse industry could benefit from increased quality of 
their sport horses, leading to better results in international competition and increased 
profits from selling New Zealand horses overseas. 
5.4 RECOMMENDATIONS 
5.4.1 A template for estimated breeding values 
Due to the limitations of the current sport horse performance dataset in New Zealand, 
the EBVs calculated here are primarily aimed to be an initial template for future 
calculation rather than a recommendation of which current sires to utilise.  
Based on the above analyses of the current sport horse population in New Zealand, data 
available for analysis and the EBVs calculated, the following recommendations have 
been made with the objective of improving the data available for use in future genetic 
analyses: 
1. Recording sires on registration: The number of horses registered with a sire needs 
to be increased. An incentive to include sire name, such as a discount on 
registration fees or an additional charge for registering a horse without a sire could 
improve completeness of these records. A similar system is implemented by the 
FFE (France), as a horse cannot be registered under list A (to compete in official 
competitions) with an unknown sire, except in special cases (Federation Equestre 
 
 
66 
 
Internationale, 2011). Longer term, implementing a similar scheme to encourage 
the inclusion of dam information in horse registration would further strengthen the 
dataset. 
 
2. Accurate sire names: The implementation of a system to improve the accuracy of 
spelling of sire names would help reduce inaccurate heterogeneity in the sport horse 
population. One way to do this would be for ESNZ to utilise a programme which 
stores sire names that have been entered before and comes up with past options 
which can be selected when a new record is added. Alternatively, a unique sire 
identification number could be assigned to each sire so that there is no longer any 
reliance on the spelling of sire names. Implementing the UELN system in New 
Zealand could be a way to achieve this. 
 
3. Differentiating amateur and professional riders: It is impractical to implement 
separate amateur and professional grades in New Zealand due to relatively small 
competitor numbers and limited competition funding (Rogers and Firth, 2005). 
However, it would be advantageous to record whether a rider is an amateur or 
professional in the ESNZ database. This could be achieved by asking riders 
whether equestrian sport is their main source of income and recording those who 
respond with ‘yes’ as professional riders. 
 
4. More stallions competing: Encouraging riders to compete stallions would help 
reduce the high selection prior to competition. This could be achieved by improving 
the facilities to keep stallions at competitions, such as having a greater number of 
stallion yards on offer at competition venues. 
5.4.2 Conclusion 
Implementation of these recommendations could help to improve the accuracy and 
completeness of the data available on sport horses in New Zealand; hence, improve the 
accuracy of calculating EBVs for sport horse sires.  If reasonably accurate EBVs could 
be obtained for New Zealand sport horse sires, their use could lead to increases in 
genetic gain and improved performance of New Zealand’s sport horses on the 
international stage. 
 
 
67 
 
References  
 
ALDRIDGE, L. I., KELLEHER, D. L., REILLY, M. & BROPHY, P. O. 2000. Estimation of 
the genetic correlation between performances at different levels of show jumping 
competitions in Ireland. Journal of Animal Breeding Genetics, 117, 65-72. 
 
ARNASON, T. 1999. Genetic evaluation of Swedish standard-bred trotters for racing 
performance traits and racing status. Journal of Animal Breeding and Genetics-
Zeitschrift Fur Tierzuchtung Und Zuchtungsbiologie, 116, 387-398. 
 
ÁRNASON, T. 1987. Contribution of various factors to genetic evaluations of stallions. 
Livestock Production Science, 16, 407-419. 
 
ARNASON, T., BENDROTH, M., PHILIPSSON, J., HENRIKSSON, K. & DARENIUS, A. 
1989. Genetic evaluations of Swedish trotters. In: LANGLOIS, B. (ed.) State of 
Breeding Evaluation in Trotters. 
 
AURICH, J. & AURICH, C. 2006. Developments in European horse breeding and 
consequences for veterinarians in equine reproduction. Reprod Dom Anim, 41, 275-279. 
 
BARNEVELD, A. 1996. The role of breeding in sports performance and health with special 
emphasis on lameness prevention. Pferdeheilkunde, 12, 689-692. 
 
BRUNS, E. 1981. Estimation of the breeding value of stallions from the tournament 
performance of their offspring. Livestock Production Science, 8, 465-473. 
 
BRUNS, E., RICARD, A. & KOENEN, E. 2004. Interstallion - on the way to an international 
genetic evaluation of sport horses. Bled, Slovenia: 55th Annual meeting of the 
European Association of Animal Production. 
 
BUTTRAM, S. T., WILLHAM, R. L., WILSON, D. E. & HEIRD, J. C. 1988. Genetics of 
racing performance in the American quater horse: I. Description of the data. Journal of 
Animal Science, 66, 2791-2799. 
 
DIETL, G., HOFFMANN, S. & REINSCH, N. 2005. Impact of trainer and judges in the mare 
performance test of Warmblood Horses. Archiv Fur Tierzucht-Archives of Animal 
Breeding, 48, 113-120. 
 
DUBOIS, C., MANFREDI, E. & RICARD, A. 2008. Optimisation of breeding schemes for 
sport horses. Livestock Science, 118, 99-112. 
 
DUBOIS, C. & RICARD, A. 2007. Efficiency of past selection of the French sport horse: Selle 
Francais breed and suggestions for the future. Livestock Science, 112, 161-171. 
 
DUCRO, B. J., KOENEN, E. P. C., TARTWIJK, J. M. F. M. V. & BOVENHUIS, H. 2007. 
Genetic relations of movement and free-jumping traits with dressage and show-jumping 
performance in competition of Dutch Warmblood horses. Livestock Science, 107. 
 
EQUESTRIAN SPORTS New Zealand. 2009a. 2009 Annual Report [Online]. Available: 
www.nzequestrian.org [Accessed 15 August 2009]. 
 
EQUESTRIAN SPORTS New Zealand. 2009b. Equestrian Sports New Zealand [Online]. 
Available: www.nzequestrian.org [Accessed 24 April 2009]. 
 
 
68 
 
FEDERATION EQUESTE INTERNATIONALE. 2010. Federation Equeste Internationale 
[Online]. Available: www.fei.org [Accessed 16 August 2009]. 
 
FEDERATION EQUESTRE INTERNATIONALE. 2011. List A [Online]. Available: 
www.fei.org [Accessed 22 March 2011]. 
 
FEDERATION FRANCAIS D'EQUITATION. 2009. FFE competition rules 2009, Edition T1 
[Online]. Available: www.ffe.com [Accessed 15 August 2009]. 
 
FRIEDRICH, C. 2010. Examination of longevity in dressage horses - A comparison between 
sport horses in New Zealand and Hanoverians in Germany. Massey University. 
 
GILMOUR, A. R., THOMPSON, R. & CULLIS, B. R. 1995. Average information REML: an 
efficient algorithm for variance parameter estimation in linear mixed models. 
Biometrics, 51, 1440-1450. 
 
GROENEVELD, E., KOVAC, M. & MIELENZ, N. 2008. VCE User's Guide and Reference 
Manual Version 6.0. 
 
HAMANN, H. & DISTL, O. 2008. Genetic variability in Hanoverian warmblood horses using 
pedigree analysis. Journal of Animal Science, 86, 1503-1513. 
 
HARDER, B., BENNEWITZ, J., REINSCH, N., MAYER, M. & KALM, E. 2005. Effect of 
missing sire information on genetic evaluation. Archiv Tierzucht, 48, 219-232. 
 
HARRIS, D. L. & NEWMAN, S. 1994. Breeding for profit: Synergism between genetic 
improvement and livestock production. Journal of Animal Science, 72, 2178-2200. 
 
HASSENSTEIN, C. 1998. [Genetic-statistical analysis of newly developed traits derived from 
show competitions for riding horses]; Genetisch statistische Analyse von 
neuentwickelten Merkmalen aus Turniersportpruefungen fuer Reitpferde. Schriftenreihe 
des Instituts fuer Tierzucht und Tierhaltung der Christian-Albrechts-Universitaet zu 
Kiel (Germany). 
 
HAZEL, L. N. 1943. The genetic basis for constructing selection indexes. Genetics, 28, 476-
490. 
 
HAZEL, L. N. & LUSH, J. L. 1942. The efficiency of three methods of selection. Journal of 
Heredity, 33, 393-399. 
 
HELLSTEN, E. T., JORJANI, H. & PHILIPSSON, J. 2008. Connectedness among five 
European sport horse populations. Livestock Science, 118, 147-156. 
 
HELLSTEN, E. T., VICKLUND, A., KOENEN, E. P. C., RICARD, A., BRUNS, E. & 
PHILIPSSON, J. 2006. Review of genetic parameters estimated at stallion and young 
horse performance tests and their correlations with later results in dressage and show 
jumping competition. Livestock Science, 103, 1-12. 
 
HENDERSON, C. R. 1953. Estimation of variance and covariance components. Biometrics, 9, 
226-252. 
 
HENDERSON, C. R. 1974. Rapid method for computing the inverse of a relationship matrix. 
Journal of Dairy Science, 58, 1727-1730. 
 
 
 
69 
 
HENDERSON, C. R. & QUAAS, R. L. 1976. Multitrait evaluation using relatives' records. 
Journal of Animal Science, 43, 1188-1197. 
 
HENNESSY, K. & QUINN, K. 2007. The future of the Irish sport horse industry. University 
College Dublin, commissioned by the Irish Horse Board. 
 
HINTZ, R. L. 1980. Genetics of performance in the horse. Journal of Animal Science, 51, 582-
594. 
 
HOLSTEINER-VERBAND. 2003. Verband der zuchter des Holsteiner pferdes E.V. (Holst) 
[Online]. Available: www.holsteiner-verband.de [Accessed 23 March 2009]. 
 
HOLSTEINER VERBAND. 2010. Stallion Performance Test [Online]. Available: 
http://www.holsteiner-verband.de [Accessed 17 April 2010]. 
 
HUIZINGA, H. A. & VANDERMEIJ, G. J. W. 1989. Estimated parameters of performance in 
jumping and dressage competition of the Dutch Warmblood horse. Livestock 
Production Science, 21, 333-345. 
 
IBM CORPORATION 2009. PASW statistics 18. Somers, New York, USA. 
 
INTERSTALLION 2005. Recommendations on choice of scale and reference population for 
publication of breeding values in sport horse breeding. Interstallion. 
 
INTERSTALLION. 2010. Interstallion [Online]. Available: 
http://www.biw.kuleuven.be/genlog/livgen/chgs_interstallion.html [Accessed 16th June 
2010]. 
 
JANSSENS, S., BUYS, N. & VANDEPITTE, W. Year. Sport status and the genetic evaluation 
for show jumping in Belgian sport horses. In:  EAAP, 2007. 
 
JOHNSON, D. D. 2010. aiREML. Palmerston North. 
 
KEARSLEY, C. G. S., WOOLLIAMS, J. A., COFFEY, M. P. & BROTHERSTONE, S. 2008. 
Use of competition data for genetic evaluations of eventing horses in Britain: Analysis 
of the dressage, showjumping and cross country phases of eventing competition. 
Livestock Science, 118, 72-81. 
 
KLEMETSDAL, G. 1994. Application of standardised, accumulated transformed earnings in 
breeding of Norwegian trotters. Livestock Production Science, 38, 245-253. 
 
KOENEN, E. 2002a. Genetic evaluations for competition traits of warmblood sport horses. 
Arnhem: World Breeding Federation for Sport Horses. 
 
KOENEN, E. 2002b. The interstallion questionnaire: a valuable source for sport horse breeders. 
Budapest: World Breeding Federation for Sport Horses. 
 
KOENEN, E. 2010. Sport horse breeding: what are we aiming at? : Interstallion. 
 
KOENEN, E. & PHILIPSSON, J. 2007. Report on Interstallion activities 2006-2007. 
Interstallion. 
 
KOENEN, E. P. C. & ALDRIDGE, L. I. 2002. Testing and genetic evaluation of sport horses in 
an international perspective. 7th World Congress Applied to Livestock Production. 
Montpellier. 
 
 
70 
 
KOENEN, E. P. C., ALDRIDGE, L. I. & PHILIPSSON, J. 2003. An overview of breeding 
objectives for warmblood sport horses. Livestock Production Science, 88, 77-84. 
 
KOENEN, E. P. C., VELDHUIZEN, A. E. V. & BRASCAMP, E. W. 1995. Genetic parameters 
of linear scored conformation traits and their relation to dressage and show-jumping 
performance in the Dutch Warmblood riding horse population. Livestock Production 
Science, 43, 85-94. 
 
KOERHUIS, A. & VAN DER WERF, J. 1994. Uni-and bivariate breeding value estimation in a 
simulated horse population under sequential selection. Livestock Production Science, 
40, 207-213. 
 
KWPN. 2003. Koninklijk warmbloed paardenstamboek Nederland (KWPN) [Online]. 
Available: www.kwpn.nl [Accessed 27 March 2009]. 
 
KWPN. 2010. Stallion statistics: Breeding values stallions sport and conformation 2009-2010 
+ Breeding statistics stallions 2009 [Online]. Available: www.kwpn.nl [Accessed 26 
November 2011]. 
 
LANGLOIS, B. 1980a. Estimation de la valeur génétique des chevaux de sport d’après les 
sommes gagnées dans les compétitions équestres françaises. Annales de génétique et de 
sélection animale, 12, 15-31. 
 
LANGLOIS, B. 1980b. Heritability of racing ability in thoroughbreds -- A review. Livestock 
Production Science, 7, 591-605. 
 
LANGLOIS, B. 1983. Genetic problems in horse breeding. Livestock Production Science, 10, 
69-81. 
 
LANGLOIS, B. 1989. Breeding evaluation of French trotters according to their race earnings .1. 
Present situation. In: LANGLOIS, B. (ed.) State of Breeding Evaluation in Trotters. 
 
LANGLOIS, B. & BLOUIN, C. 1998. Effect of a horse's month of birth on its future sport 
performance. Ii. Effect on annual earnings and annual earnings per start. Annales De 
Zootechnie, 47, 67-74. 
 
LANGLOIS, B. & BLOUIN, C. 2004. Practical efficiency of breeding value estimations based 
on annual earnings of horses for jumping, trotting, and galloping races in France. 
Livestock Production Science, 87, 99-107. 
 
LUEHRS-BEHNKE, H., ROEHE, R. & KALM, E. 2002. Genetic associations among traits of 
the new integrated breeding evaluation method used for selection of German 
warmblood horses. Veterinarija Ir Zootechnika, 18, 90-93. 
 
LUSH, J. L. 1949. Heritability of quantitative characters in farm animals. Hereditas, 35, 356-
375. 
 
LYNCH, M. & WALSH, B. 1998. Genetics and analysis of quantitative traits, Massachusetts, 
Sinauer Associates, Inc. 
 
MILLER, R. H. 1977. Economics of selection programs for artificial insemination. Journal of 
Dairy Science, 60, 683-695. 
 
MINKEMA, D. 1975. Studies on the genetics of trotting performance in Dutch trotters. Annales 
de genetique et de selection animale, 7, 99-121. 
 
 
71 
 
MINKEMA, D. 1981. Studies on the genetics of trotting performance in Dutch trotters. III.–
Estimation of genetic change in speed. Genetics, Selection, Evolution: GSE, 13, 245. 
 
MRODE, R. A. & THOMPSON, R. 2005. Linear models for the prediction of animal breeding 
values, Midlothian, CABI Publishing. 
 
NEUMAIER, A. & GROENEVELD, E. 1998. Restricted maximum likelihood estimation of 
covariances in sparse linear models. Genetics Selection Evolution, 30, 1-24. 
 
New Zealand HANOVERIAN HORSE ASSOCIATION. 2009. NZHHA [Online]. Available: 
http://www.nzhanoverian.com [Accessed 26 September 2009]. 
 
New Zealand WARMBLOOD BREEDERS ASSOCIATION. 2010. The Warmblood in NZ 
[Online]. Available: http://www.horsetalk.co.nz/warmbloodhorses/history.shtml  
[Accessed 13 April 2010]. 
 
O'FERRALL, G. J. M. & CUNNINGHAM, E. 1974. Heritability of racing performance in 
thoroughbred horses. Livestock Production Science, 1, 87-97. 
 
O’BRIEN, E., STEVENS, K., PFEIFFER, D., HALL, J. & MARR, C. 2005. Factors associated 
with the wastage and achievements in competition of event horses registered in the 
United Kingdom. Veterinary record, 157, 9. 
 
OLDENBURGER-PFERDE. 2004. Verband der zuchter des Oldenburger pferdes E.V. (Oldbg) 
[Online]. Available: www.oldenburger-pferde.com [Accessed 17 March 2009]. 
 
OLSSON, E. 2006. Multi-trait evaluation of Swedish Warmblood stallions at station 
performance tests including field and competition records. Master of Science, Swedish 
University of Agricultural Sciences. 
 
OLSSON, E., NASHOLM, A., STRANDBERG, E. & PHILIPSSON, J. 2008. Use of field 
records and competition results in genetic evaluation of stallion performance tested 
Swedish Warmblood stallions. Livestock Science, 117, 287-297. 
 
PEARSON, R. E. & MILLER, R. H. 1981. Our industry today: Economic definition of total 
performance, breeding goals and breeding values for dairy cattle. Journal of Dairy 
Science, 64, 857-869. 
 
PHOCAS, F., DONOGHUE, K. & GRASER, H. 2008. Comparison of alternative strategies for 
an international genetic evaluation of beef cattle breeds. Oceania, 38, 40.8. 
 
POSTA, J., KOMLÓSI, I. & MIHÓK, S. 2009. Breeding value estimation in the Hungarian 
Sport Horse population. The Veterinary Journal, 181, 19-23. 
 
REINHARDT, F. & SCHMUTZ, M. 1997. Estimation of breeding values for sport horses in 
Germany. Sto arstvo, 51, 461-466. 
 
RICARD, A. & CHANU, I. 2001. Genetic parameters of eventing horse competition in France. 
Genetics Selection Evolution, 33, 175-190. 
 
RICARD, A. & FOURNET-HANOCQ, F. 1997. Analysis of factors affecting length of 
competitive life of jumping horses. Genetics Selection Evolution, 29, 1-17. 
 
ROGERS, C. 1993a. Examination of alternative selection policies for sport horse breeding in 
New Zealand. 423-426. 
 
 
72 
 
ROGERS, C. 2008. Study guide three: The German Equestrian Federation (FN). Palmerston 
North: Massey University. 
 
ROGERS, C. 2010. Breed Improvement Program for the NZ Sport Horse Promotion Board. 
Palmerston North: Massey University. 
 
ROGERS, C. & FIRTH, E. C. 2005. Preliminary examination of the New Zealand event horse 
production system. Proceedings of the New Zealand society of animal production, 65, 
372-377. 
 
ROGERS, C. & WICKHAM, G. 1993a. N.Z. breeders motivations and methods revealed. 
 
ROGERS, C. W. & WICKHAM, G. A. 1993b. Studies of alternative selection policies for the 
New Zealand sport horse. New Zealand Society of Animal Production, 53, 423-426. 
 
ROGERS, C. W. W. 1993b. Examination of alternative selection policies for sport horse 
breeding in New Zealand. Master of Agricultural Science, Massey University. 
 
RUHLMANN, C., JANSSENS, S., PHILIPSSON, J., HELLSTEN, E. T., CROLLY, H. & 
QUINN, K. 2009. Genetic correlations between horse show jumping competition traits 
in five European countries. Livestock Science, 122, 234-240. 
 
SANDERS, K., BENNEWITZ, J. & KALM, E. 2006. Wrong and Missing Sire Information 
Affects Genetic Gain in the Angeln Dairy Cattle Population. Journal of Dairy Science, 
89, 315-321. 
 
SAS INSTITUTE 2009. SAS. 9.3.1 ed. Maryland, USA  
 
STEWART, I. D., WOOLLIAMS, J. A. & BROTHERSTONE, S. 2010. Genetic evaluation of 
horses for performance in dressage competitions in Great Britain. Livestock Science, 
128, 36-45. 
 
SUN, C., MADSEN, P., NIELSEN, U., ZHANG, Y., LUND, M. & SU, G. 2009. Comparison 
between a sire model and an animal model for genetic evaluation of fertility traits in 
Danish Holstein population. Journal of Dairy Science, 92, 4063-4071. 
 
TAVERNIER, A. 1988. Advantages of BLUP animal-model for breeding value estimation in 
horses. Livestock Production Science, 20, 149-160. 
 
TAVERNIER, A. 1989. Breeding characteristics of the French trotter using a BLUP animal-
model. Annales De Zootechnie, 38, 145-155. 
 
TAVERNIER, A. 1990a. Estimation of breeding value of jumping horses from their ranks. 
Livestock Production Science, 26, 277-290. 
 
TAVERNIER, A. 1990b. Performances and BLUP procedure: new ways for breeding race and 
sport horses. CEREOPA. 
 
TAVERNIER, A. 1991. Genetic evaluation of  horses based on ranks in competitions. Genetics 
Selection Evolution, 23, 159-173. 
 
TEEGEN, R., EDEL, C. & THALLER, G. 2009. Population structure of the Trakehner horse 
breed. Animal, 3, 6-15. 
 
 
 
73 
 
THE New Zealand HANOVERIAN SOCIETY. 2010. The Hannoverian in New Zealand 
[Online]. Available: www.nzhanoverian.com [Accessed 15 April 2010]. 
 
THOREN, E. & PHILIPSSON, J. 2010. Interstallion study shows: Efficient young horse testing 
procedures in Europe - but further improvements possible! Uppsala: SLU. 
 
TRAKEHNER-VERBAND. 2005. Verband der zuchter und freunde des ostpreubischen 
warmblutpferdes Trakehner abstammung (Trak) [Online]. Available: www.trakehner-
verband.de [Accessed 17 March 2009]. 
 
UELN. 2010. Unique Equine Life Number [Online]. Available: http://www.ueln.net/ [Accessed 
17th June 2010]. 
 
VAN VLECK, L. D., HAKIM, A., CUNDIFF, L. V., KOCH, R., CROUSE, J. & BOLDMAN, 
K. 1992. Estimated breeding values for meat characteristics of crossbred cattle with an 
animal model. Journal of Animal Science, 70, 363. 
 
VANDENPLAS, J. 2010. Integration of foreign breeding values for stallions into the Belgian 
genetic evaluation for jumping horses. Master of Science, University de Leige-
Gembloux Agro Bio-Tech. 
 
VIKLUND, Å., NÄSHOLM, A., STRANDBERG, E. & PHILIPSSON, J. 2010. Effects of long-
time series of data on genetic evaluations for performance of Swedish Warmblood 
riding horses. Animal, 1, 1-9. 
 
VISSCHER, P. M., BOWMAN, P. J. & GODDARD, M. E. 1994. Breeding objectives for 
pasture based dairy production systems. Livestock Production Science, 40, 123-137. 
 
WALLIN, L., STRANDBERG, E. & PHILIPSSON, J. 2003. Genetic correlations between field 
test results of Swedish Warmblood riding horses as 4 year olds and lifetime 
performance results in dressage and show jumping. Livestock Production Science, 82, 
61-71. 
 
WHITAKER, T. C. & HILL, J. 2004. Analysis of the population of competing British sport 
horses as measured by lifetime performance: structural distributions at different levels 
of competition. Equine and comparative excercise physiology, 2, 43-51. 
 
WOOLLIAMS, J. A. 2006. Technical Note: A Note on the Differential Impact of Wrong and 
Missing Sire Information on Reliability and Gain. Journal of Dairy Science, 89, 4901-
4902. 
 
WORLD BREEDING FEDERATION FOR SPORT HORSES. 2010. World Breeding 
Federation for Sport Horses [Online]. Available: http://www.wbfsh.org/?GB.aspx 
[Accessed 10th June 2010]. 
 
WORLD BREEDING FEDERATION FOR SPORT HORSES. 2011. The members and 
associate members [Online]. Available: http://www.wbfsh.org/?GB/Members.aspx 
[Accessed 12 January 2011].