Agronomie 10 (1990) 627-640
DOI: 10.1051/agro:19900803

Etude de la variabilité d'un agent pathogène, Xanthomonas campestris pv manihotis, par l'analyse factorielle multiple

F. Grousson^a, J. Pagès^b and B. Boher<sup>c

a</sup>  DDAF du Nord, cité administrative, BP 505, 59022 Lille Cedex
^b  ENSAR, chaire de mathématiques, 65, route de Saint-Brieuc, 35042 Rennes Cedex, France
^c  ORSTOM, laboratoire de pathologie végétale, BP 375, Lomé, Togo

Résumé - Cette étude de la variabilité de Xanthomonas campestris pv manihotis, agent de la bactériose vasculaire du manioc, porte sur 54 souches originaires d'Amérique du Sud, et de 5 pays d'Afrique. Des inoculations sur feuilles et sur tiges de 2 cultivars de manioc permettent de reproduire les différents symptômes observés au champ. L'analyse des résultats des inoculations montre une variabilité importante de l'agressivité. In vitro, les souches de Xanthomonas campestris pv manihotis sont relativement homogènes. Cependant, l'étude de leur croissance sur différents substrats, de la réponse à une gamme d'antibiotiques et de la lysotypie fait apparaître une certaine variabilité. La synthèse de l'ensemble des résultats, par une analyse factorielle multiple, montre l'existence d'un facteur commun aux différents caractères étudiés. Il existe, en particulier, une liaison entre l'agressivité des souches et leur comportement in vitro.

Abstract - Study of variability of a pathogenic agent Xanthomonas campestris pv manihotis, by multiple factorial analysis. This study of the variability of Xanthomonas campestris pv manihotis, agent of bacterial blight of cassava, deals with 54 strains from South America and from 5 African countries. Leaf and stem inoculations on 2 cultivars of cassava allow the reproduction of the different symptoms observed in the field. Analysis of inoculation results allows one to define, according to their pathogenicity, 4 groups of strains. The influence of geographical origin is variable: for certain countries or certain areas it appears clearly, while for others it does not seem to exist. In vitro, strains of Xanthomonas campestris pv manihotis are relatively homogenous. Nevertheless, the study of their growth on various substrates, of the response to several antibiotics and of phage typing shows variability. This one is not connected to the geographical origin. Each characteristic allows one to define a stable partition from the set of strains. The synthesis of all the results by way of Multiple Factorial Analysis shows the existence of a common factor to the different characteristics studied. Specifically, a link exists between strains pathogenicity and their in vitro behaviour. The complete data are shown in tables II, IV and VI. Table VII shows the structure of the data table submitted to Multiple Factorial Analysis which takes into account 4 groups of active variables (1, affect induced by inoculation on 2 cultivars; 2, phage typing; 3, growth on various substrates; 4, response to several antibiotics) and 3 supplementary groups ( groups 3 and 4 recoded in order to measure the influence of coding - geographical origin). Figure 2 gives a factorial representation of variables; the dual representation of individuals is given in figure 3. The first factor is correlated to variables of all the groups and points out a structure common to the 4 active groups (this result is confirmed by indicators in table VIII).

Key words: cassava / pathogenicity / phage typing / antibiotic / factorial analysis

Mots clés : manioc / agressivité / lysotypie / antibiotique / analyse factorielle