- 學位論文
非介量穩定性分析方法於毛豆區域試驗資料之應用
Application of Nonparametric Stability Methods in Vegetable Soybean Regional Trial Data
摘要
作物產量的穩定性分析 (analysis of stability) 是一針對區域試驗中產量因素進行穩定性探討的研究，透過將品種與地區間交感效應的再分配，判斷該作物品系是否能夠穩定地維持合理產量。傳統常使用介量穩定性統計法 (parametric stability method) 進行作物穩定性分析時，但當試區面積過小或所選作物受地域限制，使收集之數據無法符合常態分布、機差變方同質與效應具有累加性的假設前提時，以介量方法所做出之穩定性分析結果即有待商榷。 針對環境數少或參試品系少而不適合使用傳統介量穩定性分析方法之區域試驗資料，可採用將原始數值轉換成排序值 (rank)，並以排序值進行穩定性分析的非介量穩定性統計法 (nonparametric stability method)。本研究以高雄場所提供毛豆區域試驗資料為例，以Nassar-Huhn與Kang所提出之兩類非介量穩定性分析方法進行分析，探討較不受前提限制之非介量穩定性分析方法在毛豆區域試驗資料的應用性，並使用圖像化的方式對參試品系的穩定程度進行評估。 此外，本研究參照前人研究所提出之穩定性意義之分類方式，將非介量穩定性統計量進行歸類，並針對各種非介量與介量穩定性分析所得的品種 (系) 穩定性之排序結果，運用相關分析、主成分分析及集群分析進行關聯性探討，以確認非介量穩定性統計量在穩定性分類上的妥適性，並建議未來育種者針對不同穩定性需求或受限於特定資料型態下，應使用之較合適的穩定性統計量。
並列摘要
Stability analysis is to test whether the desired agronomic traits (yield) of the variety would not dramatically change over time and locations. Traditional parametric stability methods usually require statistically assumptions, such as normality and the homogeneity of variance, and can be greatly affected by the outliers in the data. When the regional trial data has one or more above assumptions invalid, the nonparametric stability methods are suggested. Nonparametric stability methods measure the stability using the data rank and require less assumption. In this study, we apply two types of nonparametric stability methods, proposed by Nassar-Huhn and Kang, respectively, in vegetable soybean regional trial data in Taiwan. We also proposed a visual presentation to help to determine stable varieties in real practice. The nonparametric stability statistics were further classified into 3 classes by definition along with commonly used parametric ons. We also applied the principle component analysis, cluster analysis, and correlation analysis to verify our classification. The classification may assist to select proper method for stability analysis in the future.