Arnason, J. T.,Kéita, Sékou Moussa.2009-03-232009-03-2319921992Source: Masters Abstracts International, Volume: 32-01, page: 0168.9780315800205http://hdl.handle.net/10393/7624http://dx.doi.org/10.20381/ruor-6879By means of ANOVA, Duncan's test Student's t test and linear regressions, it was demonstrated that there was no significant correlation between the quantity of gramine and the morphological characters considered (geographic origin, spike row number, spike form, cultivation season, glume length, yield, water content of fresh leaves, seed mean mass, and seed colour). By "Duncan's multiple range test" it was shown that there was a significant (F = 26,81; P 0.0001) variation in gramine content between the different genotypes (minimum = 0 & maximum = 2674). However, canonical discriminant analyses and "cluster"ing by the MODECLUS procedure failed to provide justification for grouping of varieties susceptible to aphid attack, leading to the conclusion that gramine content cannot be sufficiently closely related to morphological characters to serve as a basis for taxonomy. Finally, additional experimentation revealed that the presence of gramine alone is often not sufficient to confer protection against aphid attack, and other factors must be considered, such as anatomy or the existence of other protective constituents. Such research could lead to the development of new resistant biotypes. (Abstract shortened by UMI.)210 p.Agriculture, Agronomy.Thesis