DisplayTitle

Grouping and Identification of Argentine Maize Races by Principal Component Analysis of Zein Reversed-Phase HPLC Data

¹EEA Pergamino-INTA. CC 31, 2700 Pergamino, BA, Argentina. This research was partially supported by INTA project 80-017. ²Corresponding author. E-mail: perlabtec@pergamino.inta.gov.ar ³National Center for Agricultural Utilization Research, Biomaterials Processing Research, USDA-ARS, 1815 N. University Street, Peoria, IL 61604. Use of trade names in this publication is necessary to report factually on available data. However, INTA and USDA neither guarantee nor warrant the standard of the product, and use of names implies no approval or endorsement to the exclusion of other products not mentioned that may also be suitable.

Racial classification of maize has important taxonomic and phylogenetic implications. It may also serve to organize germ plasm inventory, thus helping breeders choose their stocks. Maize racial classification is usually based on phenotypic descriptors, which may not always accurately express genetic characteristics. In contrast, synthesis and expression of zeins is directly associated with genotype. This study was conducted to determine whether racial grouping and identification of maize can be done by applying principal component analysis to zein reversed-phase high-performance liquid chromatography (RP-HPLC) data. Zeins from samples of 97 landraces (primitive varieties) of the Argentine races Cristalino Colorado, Dentado Blanco, Avatí Morotí, Capia, and Pisingallo, stored at the Pergamino Active Maize Germplasm Bank, were analyzed by RP-HPLC. Data from the ≈21- to 53-min chromatogram region (total zeins [ZT]), the ≈21- to-30 min region (zeins 2 [Z2]), or the ≈38- to 52- min region (zeins 1 [Z1]) were subjected to multivariate analysis based on principal component to group samples by race and to assign unknown samples to predetermined racial groups. Clearly differentiated racial groups were revealed, closely matching groups based on phenotype. Unknown samples could be assigned, with a low percentage of misidentification, to predetermined groups based on Mahalanobis distances. The shortest distances of unknown samples were almost always the distances to their respective groups. Approaches other than multivariate analysis were used to group and assign samples to defined races but they were not as effective. Results indicate the potential of this method as a complementary tool to perform racial grouping and identify maize materials with high genetic variability.