頁籤選單縮合
題 名 | 應用主成分分析法研究秈稻品種雜交親本之選擇 |
---|---|
作 者 | 林中茂; | 書刊名 | 中華農學會報 |
卷 期 | 82 1973.06[民62.06] |
頁 次 | 頁1-19 |
關鍵詞 | 主成分分析法; 秈稻; 品種; 雜交; |
語 文 | 中文(Chinese) |
中文摘要 | 本研究攻勢品種(系)計有67品種,包括本地秈稻34品種、大陸秈稻5品種以及外來秈稻28品種,均於60年第2期作栽培於嘉義農專試驗田。田間設計係採用系統排列。單本植。性狀調查係每品種(系)逢機取樣10株,分別測定一株穀粒產量、穗數、穗重、穗長、一穗粒數、百粒重、粒長、稔實率、葉長、葉寬、葉角、節間數、節間長、葉面積指數、株高以及抽穗期等16性狀。根據調查測定結果,估算各性狀平均值、變異係數、相關係數並應用M.G. Kendall以及奧野忠一氏的主成分分析法(Principal component analysis),將具有主成分相類似的品種(系)歸類為若干品種?(Varietal group),然後在合乎育種目標的品種?中,選拔雜交親本,俾利增進育種效果。茲將資料分析結果摘要如下: (一)供試秈稻品種(系)間,性狀之變異係數以葉角、LAI、穗數以及節間長較大(表2)。該等性狀均為關係株型(Plant type)的主要性狀。因變異性大者,育種材料也豐富,故其在株型改良上,極為有利。 (二)根據16個調查性狀之相關係數(表3),分為8個性狀?(Character group)(表4)。並以各?代表性狀(Representative character)間的相關係數作成相關行列(Correlation matrix)(表5)。經主成分分析,估算固有值 λj (Eigen value)和固有向量 ljk (Eigen vector),結果分析至第4主成分時,其% Variance的和(=Σλj/p)已達79.03%(表6)。 1. 第一主成分關連的性狀有穗長、粒長、節間長、葉角、穗數以及LAI等。 2. 第二主成分關連的性狀有節間數、LAI、葉角、生育日數等。 3. 第三主成分關連的性狀有生育日數、粒長、節間數等。 4. 第四主成分關連的性狀有節間數、LAI、葉角等。 (三)以上列四個主成分為軸,組合成六種(4C2)座標(Coordinates),作成六種品種散佈圖(Scatter diagram)(圖1~2)。凡是品種在座標上的位置相近者,又歸類為八個品種?,再根據各主成分在育種學上的意義,闡明各品種?的育種特性(表8)。 (四)在本試驗材料中,品種?V5與V6均具有中早熟、多穗、直立葉、短稈等的優良特性,但為短粒是其缺點,而品種?V1與V2卻均具有長粒之特性,適可補救V5與V6兩品種?之缺點,故可採用回交育種法以V5和V6為輪迴親本(Recurrent parent)而以V1和V2為非輪迴親本(Non-recurrent parent)?可於其回交後裔中,選拔兼具兩親本優良性狀之植株。 |
英文摘要 | Materials used in this study were 67 indica rice varieties (strains) including 34 Taiwan native, 5 mainland China, and 28 foreign varieties which were grown in the second crop of 1971 at the Rice Experimental Farm of Taiwan Provincial Chia-yi Institute of Agriculture. One seedling was planted to each hill and field design was systematic arragement. Ten plants were selected at random from each variety, and the following characters were measured: grain yield per plant, number of panicles per plant, panicle weight, panicle length, number of grains per panicle, 100 grains weight, grain length, percentage of fertility, leaf length, leaf width, leaf angle, internode number, internode length, LAI, plant height and heading date. From these data, the means and correlation coefficient were calculated for each character, and principal components were also analyzed. In this study, the classification of varieties based on four principal component axes were discussed in relation to plant breeding. The results from the analyses of data are summarized as follows: 1. The coefficients of variation for the important characters of plant type such as leaf angle, LAI, panicle number and internode length were apparently larger than those of other characters among indica rice varieties (table 2). Thus, these characters appear usefully for the improvement of the plant type. 2. Classification of sixteen characters into eight character groups were made (Table 4). From them, eight representative characters which were of significance in breeding were selected, and the correlation matrix of these eight characters was calculated (Table 5). According to the principal component analysis, 35, 21, 14 and 9 percentages of the total variation of eight characters were accounted for by the first four Principal Components respectively (Table 6). The first principal component corresponded to panicle length, grain length, internode length, leaf angle, panicle number and LAI; the second principal component corresponded to internode number, LAI, leaf angle and heading date; the third principal component corresponded to heading date, grain length and internode number; and the fourth corresponded to internode number, LAI and leaf angle. 3. So as to classify 67 varieties into 8 varietal groups (Table 8), six scatter diagrams of 67 varieties were made from the first four principal components' scores of varieties (Fig. 1. ). The smaller the distance between varieties, the more similar the charactaristics of varieties were expected. So the varieties with small distances among them were grouped into a varietal group. 4. From the analysis of data mentioned above (Table 8), it was noted that varietal groups V5 nad V6 were satisfactory recurrent parents with early maturity, large number of panicle, erect leaves and short culm while V1 and V2 were good donor parent because of their long grains which the recurrent parents were deficient. Thus the backcross method of breeding can be used for improving varieties in this case. 5. It appears that the classification of indica rice varieties and selection of breeding materials can be successfully through the application of the Principal component analysis. |
本系統中英文摘要資訊取自各篇刊載內容。