The Hispano-Arab Horse breeders’ Association aims

to improve the genetics of both body conformation

targeted to riding and sport activities, as

temperamental qualities (attitudes and behaviors) in

order to facilitate dressage management, and to be

distinguished in the required disciplines.

The objective of the present paper was to study

diverse methodologies for the genetic parameters

estimation of conformational and movements traits

in Spanish-Arab horses with a view to determine

which mathematical philosophy give the best

performance in traits based of objective

assessments. Harmony, walk, trot and gallop

assessments were the traits observed as selection

criteria.

Introduction

M. Gómez1*, S. A. Attalla3, J. Fernández2, J.M. León1, N. Carolino4, L.T. Gama5, J. V. Delgado1

Preliminary comparative analysis of the application of infinitesimal and threshold models in the estimation of genetic parameters of

harmony and movements traits in Spanish-Arabian horses

Materials and Methods

A single trait animal model with repeated

observations, taking into account as fixed effects the

management groups (36 farm clusters according to

their zootechnical management), sex, year, season

and qualifier. Random effects were the individual

genetic additive value and the permanent

environmental effects.

This model was resolved using both an infinitesimal

model using the Restricted Maximum Likelihood

theory and the derivate free algorithm; and a

threshold model using the Gibbs sampling method.

1. For the threshold model , THRGIBBS1F90

Software, was used considering 750000 cycles, being

50000 of them discarded (burn-in), afterwards one of

every 1000 samples were stored to calculate the

posterior distributions, remaining 700 samples taken

into account in the study.

2. For the REML model WOMBAT Software was used,

with a convergence criterion of 10-9. Once the

solutions from both methodologies obtained, genetic

parameters and their errors were compared.

Conclusion

In conclusion we have determined that both

methodologies shown similar estimations. It is

pointing out that the use of threshold models do not

improve the efficiency of the methods presently used

in the breeding program of the breed base on REML

estimations. All traits taken into account are

continuous and distributed according to normal

characteristics.

Results

Heritabilities obtained by mean of the Threshold

models and their correspondent errors of estimation

were;

1Departamento de Genética, Facultad de Veterinaria. Universidad de Córdoba. España. E-mail: 2 Union Española de Ganaderos de Pura Raza Hispano-Árabe. Sevilla, España.

3 Department of Animal Production, Faculty of Agriculture, Cairo University, Egypt. 4 Unidade de Recursos Genéticos, Reprodução e Melhoramento Animal. INRB, IP. Santarém. Portugal.

5 Faculdade de Medicina Veterinária. Universidade Técnica de Lisboa. Portugal.

mayritagom@yahoo.com

Data set

1704 registers

Recorded along 11 years (2000-2011)

Kinship matrix was integrated by 1184

Belonging to 559 both sexes individuals

36 farms

Heritabilities obtained using the REML model were;

Traits h2 (SE) r (SE)

Harmony 0.28 (0.05) 0.33 (0.06)

Walk 0.02 (0.004) 0.17 (0.03)

Trot 0.20 (0.05) 0.35 (0.02)

Gallop 0.12 (0.03) 0.20 (0.01)

Traits h2 (SE) r (SE)

Harmony 0.22 (0.07) 0.32 (0.03)

Walk 0.13 (0.05) 0.20 (0.03)

Trot 0.16 (0.06) 0.30 (0.03)

Gallop 0.11 (0.05) 0.18 (0.03)

0,000

0,050

0,100

0,150

0,200

0,250

0,300

Harmony Walk Trot Gallop

threshold

linear

Figure 1. Difference between heritabilities estimated using linear

and threshold models.