Abstract
This experiment was carried out in order to classify 12 varieties of orchardgrass grown under individual plant basis by the principal component analysis. Out of 42 characters measured, 11 characters were selected. They were number of tillers per plant, weight of a tiller, tiller angle, stem thickness, plant height, panicle length, 2nd internode length, 3rd leaf breadth, 3rd leaf length×3rd leaf breadth and 3rd leaf length/3rd leaf breadth ordered downward from the flag leaf. The results are summarized as follows: 1) 84% of the total variation of 11 characters were accounted for by 1st, 2nd and 3rd principal components. The 1st principal component is mainly concerned with plant type and plant size. It is assumed that the 1st principal component is almost the same as the plant type index defined in the previous paper (J. Japan. Grassl. Sci., 24, 102-107). In fact, the 1st principal component indicated highly significant correlation with the plant type index (r=0.925). The variety with the large 1st principal component should have the big and elect tillers. Similarly, the 2nd principal component appeared to indicate the size of photosynthetic and non-photosynthetic organs, and the 3rd principal component appeared to correspond to the size of panicle. 2) The classification of 12 varieties was made using these 3 principal components. The result of classification seemed to be short of clearity on this purpose. It was concluded that the classification of grass variety with the plant type index reported in the preceding paper should have a practical meaning. The principal component analysis might be useful in the case of classification of varieties including large numbers of unknown materials.
