Characterization of plant type and epidemiological development in the pathosystem "upland rice/rice blast" (Pyricularia grisea) by means of multivariate statistical methods

With a holistic experimental approach and using repeated measurements under field conditions, a ranking of six upland rice cultivars (five traditional Philippine and one improved cultivar, 'C22') according to their morphology and leaf blast development was established showing different plant type ex pression. The development of culm length and plant height was parallel duri ng three field experiments and displayed a mean ranking of cv. 'Tayak' > 'M alos' > 'Sinalognan' > 'Lubang' > 'Suacong' > 'C22'. Similar observations w ere made for leaf area. Cvs. 'Tayak' and 'Malos', representing typical trad itional cultivars with long, broad leaves, showed lower disease severities (number of susceptible type lesions/cm(2) leaf area) not at the beginning, but at the end of the growing season. The two intermediate plant types show ed either higher (cv. Suacong) or lower disease (cv. Sinalognan). The tradi tional Indica-cv. Lubang with very narrow, long leaves showed high lesion d ensities throughout the growing period. Indica-cv. 'C22' was a typical impr oved upland rice with short and narrow leaves, but also showed high densiti es of leaf blast lesions, especially if the disease onset was early during the growing period. This data set containing morphological (MP) and disease parameters (DP) was also analyzed with the use of three multivariate metho ds. The cluster analysis of the parameters revealed a stronger positive rel ationship of MP and DP among themselves and a weaker negative relationship between MP and DP. Leaf length and area parameters as well as their growth rates showed a strong association with DP. With the use of the canonical co rrelation analysis, the variance of the DP could be explained between 34-71 % by the canonical variables of the MP, mainly due to the influence of lea f area parameters. Larger plant types showed lesser lesion densities in two field experiments, whereas in one FE (young plants) the relationship was p ositive. The canonical discriminant analysis separated the six cultivars pr actically into three groups usings three MP and their growth rates as well as four DP. This grouping could be displayed graphically with the first two canonical components. For a comparison of epidemiological behavior of cult ivars their morphology, c. g., leaf size, are relevant parameters.