ASSESSMENT OF GENETIC-VARIATION IN AVENA-STERILIS USING MORPHOLOGICAL TRAITS, ISOZYMES, AND RFLPS

Genetic patterns revealed by proximity coefficients may be affected by choice of traits, method of scoring, any subsequent transformations of scores, and choice of proximity coefficient. Our objective was to compare restriction fragment length polymorphism (RFLP), isozyme polymorphism, and variation in qualitative and quantitative morphological traits in a geographically stratified set of 177 accessions of a hexaploid wild oat, Avena sterilis L. Jaccard similarity coefficients (SJ) and Russell and Rao similarity coefficients (SRR) were calculated for all pairs of genotypes from restriction fragments and, separately, from isozymes. Standard taxonomic (DSZ), Mahalanobis, and Good man distances were calculated from 26 morphological trait scores. Clustering of mean DSZ values or mean SJ(RFLP) values between pairs of countries produced similar dendrograms while SRR(isozymes) values resulted in different subgroups. Rankings of within-country diversity were similar for different types of traits but were unrelated to geographic proximity of provenances of the accessions. Proximity coefficients based on the same type of trait (either RFLP, isozyme, or morphology) were highly correlated (Mantel statistic) (0.6-0.9),while correlations for proximities based on different types of traits were less than or equal to 0.35. Isozyme and RFLP-based proximities were poorly correlated, and both were poor predictors of morphological relationships with the highest correlation being -0.35 between SJ(RFLP) and DSZ. While broad patterns of variation revealed by different types of traits were similar in this sample of A. sterilis, differences in pairwise estimates of relationship were sufficiently great to question the exclusive use of one type of trait for sampling and management of plant germplasm collections.