The potential use of sugarcane varieties for the identification of genetic markers.

Abstract

The use of genetic markers that are linked to specific traits in sugarcane has the potential to increase the efficiency of the selection of improved varieties. Conventionally, markers are identified by analysing the segregation of potential markers and traits in the progeny of single crosses. However, this approach is not practical for sugarcane breeding programmes where replicated, well characterized progenies do not exist. The objective of this project was to investigate the potential of using commercial varieties for identifying markers associated with some of the important traits in sugarcane. This approach would be far more effective than dealing with single progenies since the traits of commercial varieties have already been characterized. The DNA of fifty commercial varieties of sugarcane was amplified by RAPD PCR using forty-one arbitrary decamer primers. Analysis of the resulting banding profiles, obtained by agarose gel electrophoresis, yielded fifty-four reliable polymorphic fragments. Two approaches were used to identify putative markers linked to the traits of resistance to eldana, sugarcane mosaic virus, and smut: (1) a correlation approach which attempted to identify whether the presence of any polymorphisms could be used to imply the existence of a particular phenotypic state, and (2) multiple regression analysis, in order to determine whether polymorphisms could be used to predict the performance of the varieties for each of the traits. Both approaches appeared to identify associations between polymorphisms and the traits, although multiple regression analysis yielded the most informative results and was able to assign statistical values to the associations. Using multiple regression, the best predictive model was obtained for sugarcane mosaic virus resistance. This model consisted of four polymorphisms and had an r² of 0.40l. By dividing the resistance ratings into three groups (resistant, intermediate and susceptible), 52% of the varieties were correctly classified and only 2% of the varieties were predicted in opposite groups (i .e. predicted susceptible when actually resistant, and vice versa). The predictive model for eldana resistance consisted offour polymorphisms and had an r² of 0.347. This model classified 30% of the varieties in the correct group of three while none of the varieties were predicted in opposite groups. The predictive model for smut resistance consisted of three polymorphisms and had an r² of 0.316. This model classified 30% of the varieties in the correct group of three while 2% of the varieties were predicted in opposite groups. Further analysis of sugarcane varieties using additional polyrnorphisrns has the potential to identify markers linked to important traits. These markers could be used for marker-assisted selection to increase the efficiency of selecting for improved sugarcane genotypes for commercial release.