7th World Congress on Genetics Applied to Livestock Production, August 19-23, 2002, Montpellier, France Session 26. Management of genetic diversity Communication N° 26-06 CONTRIBUTION OF INDIVIDUAL MARKERS TO THE ANALYSIS OF THE RELATIONSHIPS AMONG BREEDS BY CORRESPONDENCE ANALYSIS D. Laloë 1 , K. Moazami-Goudarzi 2 and D. Chessel 3 1 INRA, Station de Génétique Quantitative et Appliquée, 78352 Jouy-en-Josas, France 2 INRA, Laboratoire de Génétique biochimique et de Cytogénétique, 78352 Jouy-en-Josas, France 3 Biométrie et Biologie Évolutive, Université Lyon 1, 69622 Villeurbanne Cedex, France INTRODUCTION Recently the value of genetic ressources has been recognized which has encouraged studies to determine genetic variations within and among breeds. For this purpose, genetic markers have been successfully used. Generally, the approach to determine the relationships among breeds consists in calculating genetic distances and in constructing trees. Alternatively, when admixtures are known to have occurred among the populations under study, multivariate techniques can be used. This approach has been pioneered by Cavalli-Sforza and his colleagues to reconstruct the history of human populations (Cavalli-Sforza et al., 1994) and is commonly used in studies of domestic breeds (MacHugh et al., 1997). Such studies typically involve several loci. However, how each marker contributes to the global typology has not been examined. In this article, we propose, within the framework of a multivariate analysis, a method that deals with this problem. This method is illustrated with a data set of 16 cattle breeds characterized with the 30 microsatellite markers of the reference FAO list. MATERIALS AND METHODS Material. Animals were typed for the 30 microsatellite markers of the reference FAO list (see http://www.ri.bbsrc.ac.uk/cdiv for a description). Protocols are described in Moazami- Goudarzi et al. (1997). Eleven French and 5 West-African cattle breeds were included in the analysis. French breeds were: Aubrac, Bazadaise, Blonde d'Aquitaine, Bretonne Pie-Noire, Charolaise, Gasconne, Limousine, Maine-Anjou, Montbéliarde, Normande, Salers. West African breeds were : Somba, Lagunaire, N'Dama, Sudanese Zebu Peul and Borgou (Shorthorn x Zebu crossbred). More details concerning the animal sampling are in Moazami-Goudarzi et al. (2002). Methods. Correspondence Analysis (CA) (Lebart et al., 1984) is a weighted Principal Component Analysis (PCA) method which is suited for categorical variables. In our context, it leads to a simultaneous representation of breeds and alleles as a cloud of points in an euclidean space. As with the PCA, axes, which are ranked according to their fraction of information, span this space with each axis independent of the others. Inertia measures this information, i.e., the direction of maximum inertia is the direction in which the cloud of points is most scattered. Inertia can be split up according to axes and/or loci or alleles. Contribution of an allele to the construction of an axis is measured by the percentage of the inertia of this axis that is supplied by the allele. CA is suited for simple contingency tables that cross populations by alleles at one locus. Phylogenetic studies typically involve several markers, say k markers, and lead to a stock of k contingency tables, i.e. a partitioned table where populations are crossed with the