2.2.36
ASSESSMENT BY RFLP ANALYSIS OF THE POPULATION STRUCTURE OF XANTHOMONAS PATHOVAR MANGIFERAEINDICAE WITHIN A SINGLE FIELD CONTAINING TWO HOST SPECIES

L GAGNEVIN and O PRUVOST

Laboratoire de Phytopathologie, CIRAD-FLHOR, 97455 Saint-Pierre, Ile de la Réunion, France

Background and objectives
A previous study using RFLP typing established that the population of Xanthomonas pv. mangiferaeindicae, a pathogen of mango and present in Réunion, is heterogeneous and contains two types of strains [1]. One type is found on mango plants and the other on pepper tree, a weed plant present throughout the island. Although there is symptom specialization, all strains are pathogenic when artificially inoculated on mango and pepper tree. These findings raised the possibility that pepper tree might serve as an alternative host for X. pv. mangiferaeindicae. The strains that infect the two hosts could not be differentiated by physiological or biochemical tests or by RFLP analysis with repetitive DNA probes, but could be differentiated by means of RFLP using the probe pBSavrXa10, which contains the avirulence gene avrXa10 from X. oryzae pv. oryzae [2].

The purpose of this study was to evaluate the role of the pepper tree strains in the epidemiology of the disease on mango by systematically collecting and typing strains from an infected mango orchard and from a nearby infected pepper tree windbreak hedgerow.

Materials and methods
A total of 365 strains were collected from leaf lesions on mango trees and pepper trees (a windbreak located less than 6 m from the closest mango trees) in a 3-year-old commercial orchard. BamHI-digested total DNA was hybridized against the probe pBSavrXa10. Hybridization profiles were compared by constructing a neighbour-joining tree based on the Jaccard similarity coefficients. The distribution of the haplotypes (groups of strains distinguished by RFLP patterns) was compared with the distribution of the disease in the mango orchard assessed by a spatial autocorrelation analysis.

Results and conclusions
In this mango orchard, the prevailing winds blow through the hedgerow towards the mango trees. The disease distribution is aggregative, with the shape of the core cluster indicating a prevailing spread larger than one row. Therefore, it is likely that the conditions for short-distance bacterial inoculum transport (associated with wind-driven rains) from pepper trees to mango trees occurred prior to strain collection.

Avr haplotypes of strains found on mango and pepper trees are always distinct and constitute separate branches of the neighbour-joining tree. This implies that even if bacterial strains were transported from the pepper tree hedgerow to the mango trees, they did not cause disease on this plant. Several hypotheses can be invoked to explain this: either the lesions on the pepper tree do not produce sufficient amounts of inoculum to infect significantly the mangos, the survival of the pepper tree strains is low during the transport and the epiphytic phases of their life cycle, or the pepper strains can not cause detectable levels of disease or replicate on mango. A combination of these different factors (and others) could also explain the observed absence of pepper tree-type strains on mango.

These results show that, even for strains of a pathovar that cannot be differentiated by means other than genomic studies, there are pathogenicity and epidemiological differences that are probably responsible for the distribution of different members of this pathovar on different hosts. This also raises questions about the relatedness of the haplotypes found on mango and pepper tree. Considering that strains found on the same host are genetically related (as suggested by the neighbour-joining tree) and that some groups of pepper tree-originating strains are close to some mango strains, it is likely that this host differentiation originates from subtle genetic changes revealed by RFLP. This is even more striking considering that the probe used to assess these differences is homologous to avr genes, which are highly involved in host range and probably also in pathogenicity.

References
1. Gagnevin L, Leach JE, Pruvost O, 1997. Applied Environmental Microbiology 63(1), 246-253.
2. Hopkins CM, White FF, Choi SH, Guo A, Leach JE, 1992. Mol. Plant-Microbe Interactactions 5, 451-459.