區域試驗多性狀產量指標之穩定性分析

Stability Analysis of Yield Indices for Multiple Traits in Regional Trials

史凱萱(Kai-Hsuan Shih) ； 陳凱儀(Kai-Yi Chen) ； 呂秀英(Hsiu-Ying Lu) ； 呂椿棠(Chun-Tang Lu) ； 周國隆(Kuo-Lung Chou) ； 劉力瑜(Li-Yu D Liu)

《作物、環境與生物資訊》 11卷2期 (2014/06) Pp. 80-87

https://doi.org/10.30061/CEB.201406_11(2).0003

摘要

作物區域試驗資料之穩定性分析，係探討作物品系在不同環境下是否具有穩定表現。傳統的穩定性分析方法，僅針對單一性狀，尤其是收穫產量；然而作物品種改良，有可能同時考量諸多性狀作為品種選拔基準。因此，本研究以毛豆區域試驗17個性狀資料為分析實例，提出利用多變數統計方法之集群分析與因素分析，先針對多個性狀進行篩選並建立綜合指標後，再進行穩定性分析。結果顯示，運用集群分析篩選出毛豆10個穩健性狀，該等性狀經由因素分析擷取出可分別解釋株型、經濟產量及能量分配之三個因素，以其中經濟產量因素之得點為例進行直線回歸穩定性分析，得以選拔出符合育種目標下具有優良表現且穩定之品系。

關鍵字

區域試驗 ； 穩定性分析 ； 多變數分析 ； 集群分析 ； 因素分析

並列摘要

The stability analysis traditionally investigates the stability of yield for a crop grown in multiple environments. However, in some crop breeding programs, multiple traits are simultaneously considered and it is desired to analyze the data using multivariate statistics. In this study, we proposed to select robust traits by cluster analysis and to generate a composite index for all selected traits by factor analysis; the composite index was then subject to regular stability analysis. Our methodology had been practiced on vegetable soybean regional trial data with records of 17 traits. The results showed that cluster analysis identified ten robust traits and factor analysis extracted three factors representing plant type, economic value, and energy partition of the crop, respectively. We used the scores of economic-value factor to perform linear regression stability analysis and had demonstrated that the stable varieties that mostly achieve the specific breeding objective had been selected.

並列關鍵字

Regional trials ； Stability analysis ； Multivariate data analysis ； Cluster analysis ； Factor analysis