-Declarations of matings (535 746) and 308 549 consecutive declarations of birth from 1989 to 1999 were analysed by logistic regression in order to determine the effects of year, breed and age of parents on numerical productivity (the number of foals declared per mated mare per year). For the years 1994 to 1999, the status of the mare, type of mating and month of first mating, were also available. The effect of inbreeding and, for warm-blooded horses, the effect of the level of performances or the effect of the level of breeding value estimation were also analysed. The main results are the following: numerical productivity progressed in France more for draught breeds than for saddle breeds and trotters. Thoroughbreds progressed less and just reached the level of significance. Cold-blooded horses, however, appeared less productive than warm-blooded horses for which thoroughbreds were at the lower level. It cannot be concluded if this figure reveals biological differences in fertility or if it is only the result of differences in managing the official declarations. For warm-blooded horses, the absence of negative relationships between the trends of selection and numerical productivity results appeared clearly. A high performance level for the mare was positively associated with higher productivity results in sport and trotting horses and showed no significant influence for galloping horses. The relationships with breeding value estimation illustrated the same trends.horse / fertility / logistic model / factors of variation / selection / inbreeding

No abstract

-The aim of this study was to estimate the breeding value of English Thoroughbreds in the Czech Republic using racing results from a 22-year period . The data include the performance of two and three-year-old horses which raced in flat races at hippodromes in the Czech Republic. The racing results (30 203) were available corresponding to 6 333 horses descending from 762 sires and 2 836 dams. Different criteria were applied in order to analyse the performance: Log of earnings per race, a normalized ranking value, distance of the race when placed, earnings and number of starts for 2, 3, 2+3 year-old horses. After preliminary studies, a year effect or a sex by year effect was finally retained. Variance component estimation using VCE software gave the following values for heritability (±standard errors): 0.14 ± 0.01 and 0.16 ± 0.01 for the Log of earnings per race and the ranking value. Repeatability was 0.31 and 0.35, respectively. The maternal environment component was evaluated as 0.02 ± 0.004 for the Log of earnings per race and 0.03 ± 0.004 for the ranking value. We found that the Log of earnings per race and the ranking value were two appropriate criteria when taking into account racing performance in selection for Thoroughbreds in the Czech Republic. The genetic correlation of the two criteria was 0.98 ± 0.003. The heritability for the distance when placed was 0.18 ± 0.01. The genetic correlation of the Log of earnings per race and distance was medium, 0.38 ± 0.05 and of the same order, 0.39 ± 0.05 for ranking value and distance. In the case where we used the Log of annual earnings and the number of starts, the heritabilities were for the Log of earnings: 0.15 ± 0.03 for two-year-olds, 0.34 ± 0.03 for three-year-olds and 0.32 ± 0.03 for two and three-year-old careers together and respectively, 0.12 ± 0.03, 0.21 ± 0.03 and 0.20 ± 0.02 for number of starts. The genetic correlations between the earnings and the number of starts were respectively: 0.26 ± 0.14, 0.33 ± 0.06 and 0.19 ± 0.07. The genetic correlation between the number of starts for two and three year-olds which was restricted to horses earning money for two consecutive years, was average: 0.35 ± 0.05 and between earnings for the same ages was high 0.80 ± 0.04.

-The aim of this study was to estimate the breeding value of Arab horses in Tunisia. Racing results (36203) were available corresponding to 2432 horses issued from 811 dams and 218 sires registered in the Tunisian stud book and in the races organized from 1980 to 2001 by the "Société des courses de Tunis". Performance was described by two criteria: the earning and the rank in the race used after some transformations: (i) attribution of "theoretical" earnings to the not placed horses and normalization by a Log transformation, (ii) normalized rank according to statistical tables on which a "performance rate" calculation was applied (see further). The effects of year, age and sex were significant. The effect of weight carried was also significant but the results appeared paradoxical because the more they carried the better were the performances. Variance component estimation using VCE software gave the following values for heritability: 0.09 and 0.12, respectively, for the Log of earning and the ranking value. Repeatability was the following: 0.25 and 0.35, respectively. The maternal environment component was evaluated as 0.04 for the ranking value and 0.03 for the Log of earning. The genetic correlation of the two criteria was estimated to 0.97 standard error and was around 0.01 in all cases. In conclusion, the Log of earning and ranking value are two appropriate manners to take into account racing performance in selection for Arab horses in Tunisia. It can be remarked, however, that the heritability of the ranking value was higher. The adjustment of earnings with the level of races was probably not precise enough and intra race as a function of the ranks, the respect of a constant relative decrease of the earnings according to the place was too approximate.horse / racing ability / purebred Arabian / BLUP / animal model / Tunisia Résumé -Évaluation génétique par un BLUP en modèle animal des chevaux arabes d'après leurs résultats en course en Tunisie. L'objectif de l'étude est l'estimation de la valeur génétique additive des chevaux de pur-sang arabe en Tunisie, il est mené sur 36203 données relatives à 2432 chevaux issus de 811 mères et 218 pères de pur-sang arabe enregistrés dans le stud book tunisien et (ii) un classement normalisé d'après les tables statistiques sur lequel une procédure de « performance rate » a été appliquée. Les effets année, âge et sexe s'avèrent significatifs. Pour le poids porté, le résul-tat paraît paradoxal puisque les chevaux les plus chargés expriment en moyenne de meilleures performances. L'analyse par un modèle animal avec répétabilité grâce au logiciel VCE a conduit à l'estimation des paramètres génétiques : une héritabilité de 0,09 ± 0,01 et 0,12 ± 0,01 respectivement pour le log du gain et la valeur classement ; la répétabilité est de 0,25 ± 0,01 pour le log du gain et 0,35 ± 0,01 pour le classement. L'effet maternel est de 0,04 ± 0,01 pour le classement et 0,03 ± 0,01 pour le log du gain. Il ressort alors que le log du gain et la valeur classement sont deux bonnes façons...

No abstract