An operational bayesian approach for the identification of sexually reproduced cross-fertilized populations using molecular markers

Identification of an individual, clone or pure line with molecular markers is a simple matter, at least in theory. It involves having a sufficient number of reproducible markers to ensure that each of the genotypes corresponds to a unique genetic profile. When the operational taxonomic units (OTU) to be identified are allogamous populations, each of them corresponds to a series of genotypes for which it is impossible to make a list, but which can be characterized, fur example, by allelic frequencies. Identification therefore has to be carried out in terms of probabilities. The method proposed consists in comparing the sample to be identified with a reference database consisting of a certain number of OTUs covering the species' diversity spectrum. The a priori information available on the sample can be exploited by taking a Bayesian approach. The limited size of the sample, and of the reference populations, is also taken into account. The result is the list of possible OTUs matched with probabilities of belonging. This approach was tested on 44 coconut cultivars represented by a total of 288 individuals (4 to 15 per cultivar), with 23 RFLP markers. The advantages offered by the method are optimum utilization of available information, good tolerance of missing data, easy interpretation, and the possibility of optimizing the size of the samples and the number of markers. Lastly, with a simple update of the database, the results obtained during the study can be used to their full advantage.