臺灣落花生種原之親緣關係(I).應用農藝性狀進行花生種原親緣關係之研究

Abstract

本試驗使用本省過去落花生育種上常用之親本及重要育成品種(系)共30種之農藝性狀來探討落花生種原的緣關係，以期做為育種上親本選用之參考，並進而發展岀一套落花生原判別鑑定之方法。 首先，將19項田間調查所得到的農藝性狀資料進行變方分析(ANOVA)，分析後未達5%顯著水準的成熟莢數和飽滿度等2項性狀資料捨棄，採用其餘17項農藝性狀進行集群分析，並兩群間最小距離(MD)、標準偏差(RMSSTD)、半淨複相關係數(SPRSQ)和複相關係數(RSQ)等四種統計值作為分群模式的判斷慓準。當歐氏距大於79時分成三群,第一群以3個Virginia型的品種(系)為主；第二群以Spanish型的台南12號單獨形成一群,可能因其親本中有Virginia runner型和Valencia型的血緣；第3群則以個Spanish型的品種(系)為主。相對地,當歐氏距離大於0.45時亦可分成六群,第一群Virginia runner型為主:第二群和第三群分別Virginia runner型的Kinorales和Spanish型的台南12號均單獨形成一群;第四群有Virginia runner型的大冇和Spanish型的CJ-444，第五群以4個Spanish型和2個Virginia bunch型的品種(系)為主;第六群除了PI 314817為VaIencia型外，其它16個品種(系)均為Spanish型。因此，當集群數目區分為3群或6群時，可發現分群後的結果大致與落花生依植物型所分類的Spanish型、Valencia型和Virginia型相同，而且各集群的品種(系)間之親緣關係亦可利用五組育種系統，以及本地種和外來引進種兩群不同選育來源的族群之種類和譜系圖來做輔助性的解釋，做為上述三型分群後再組合之參考。 另外，根據因素分析的結果，17項農藝性狀在二個主要共同因素中的因素荷量可以明顯發現，Factor 1的組成因子中大部分為產量構成因子;Factor2的組成因子則以植株外表形態為主。所以，以農藝性狀資料而言，推論影響落花生遺傳距離的遠近主要因子為產量構成因子，其次為植株性狀因子。 The 30 varieties (lines), often used as materials of peanut breeding in Taiwan, were applied to evaluate the genetic relationship of peanut germplasms by agronomic traits. According to the forenamed method, expect to develop a suitable technique for identifying the genetic relationship of peanut germplasms. First, 19 agronomic traits investigated in field were analyzed by ANOVA. The number of maturity pod and the fullness degree were abandoned, because the probabilities of F values were above 5% significant level. Consequently, only use else 17 agronomic traits to proceed with the cluster analysis. According to MD, RMS STD, SPRSQ and RSQ obtained from cluster analysis, the suitable numbers of divided clusters could be distinguished by 17 agronomic traits. When the Euclidean distance was larger than 0.79, the 30 varieties (lines) could be divided into 3 mam clusters. The first cluster mainly contained 3 varieties of Virginia type. The second cluster only contained one variety of Spanish type (Tainan No. 12), because it was led into the genetic compositions of Virginia runner type and Valencia type. The third cluster mainly contained 21 varieties (lines) of Spanish type. Oppositely, when the Euclidean distance was larger than 0.45, they might be also divided into 6 main clusters. The first cluster mainly contained some varieties of Virginia type. The second cluster and the third cluster respectively contained Kinorales of Valencia type and Tainan No.12 of Spanish type alone. The forth cluster contained Ta-yu of Virginia runner type and CJ-444 of Spanish type. The fifth cluster mainly contained 4 varieties (lines) of Spanish type and 2 varieties (lines) of Virginia bunch type. The sixth cluster, except for PI 314817 of Valencia type, also contained 16 varieties (lines) of Spanish type. Consequently, when the cluster numbers were divided into 3 main clusters or 6 main clusters, the result was closely related to Spanish type, Virginia type and Valencia type classified by plant botanic types and growth characteristics of peanut. The genetic relationship between varieties (lines) of main clusters might be reasonably explained by the pedigree diagrams of 5 breeding systems and the sources of 2 selecting populations. In addition, according to the factor loadings of 17 agronomic traits of 2 main common factors, the mainly composed factors of Factor I and Factor 2 respectively belonged to the yield component and the plant botanic type. Therefore, we suggested that the genetic distances among divided clusters by the agronomic traits were mainly influenced by the yield component, then by the plant botanic type.