Multivariate analysis of stripe rust assessment and reactions of barley in multi-location nurseries

Abstract

Xi, K., Chen, X. M., Capettini, F., Falconi, E., Yang, R. C., Helm, J. H., Holtz, M. D., Juskiw, P., Kumar, K., Nyachiro, J. and Turkington, T. K. 2013. Multivariate analysis of stripe rust assessment and reactions of barley in multi-location nurseries. Can. J. Plant Sci. 93: 209–219. A total of 1357 entries, mainly consisting of hulled two-row, hulled six-row and hulless barley, were evaluated in stripe rust nurseries at Toluca, Mexico during 2007, Quito, Ecuador during 2007 and 2008, and Pullman and Mt. Vernon, USA [Pacific Northwest (PNW)] during 2007–2009. Disease screening data for barley stripe rust resistance at multiple locations and seasons were analyzed using factor analysis (FA), principal component analysis (PCA) and analysis of variance (ANOVA). Factor analysis led to the removal of a number of disease assessment variables that had inadequate disease levels or an unsuitable rating scale. The PCA scores revealed that the two-row types of barley were generally more resistant than the six-row and hulless types. ANOVA indicated that the effect of seasonal influence on screening was small, while stripe rust susceptible and resistant barley types were differentiated significantly on mean values averaged on infection type (IT) and percentage diseased leaf area (disease severity, DS) during the 3-yr tests in multiple screening nurseries. The significant correlations in disease assessments between IT and DS suggest that either assessment can be used to replace the other without a significant loss of information regarding barley stripe rust reactions. The first principal component is a useful criterion for assessing stripe rust reactions in barley lines.