Please use this identifier to cite or link to this item: http://hdl.handle.net/11455/65884
標題: 從具有混雜數據性質的27年生後裔檢定林分估計香杉的育種價及遺傳增益
Estimating Breeding Values and Genetic Gain of Cunninghamia lanceolata (Lamb.) Hook. var. konishii from a 27-Year Progeny Test with Messy Data.
作者: 蘇力禾
Su, Susi Li-Ho
關鍵字: China fir
香杉
efficiency of blocking
border effect
progeny test
breeding value
best linear unbiased prediction (BLUP)
heritability
區集效率
邊界效應
後裔檢定
育種價
最佳線性無偏差預測
遺傳率
出版社: 森林學系
摘要: 本研究係針對臺灣的鄉土樹種香杉(Cunninghamia lanceolata (Lamb.) Hook. var. konishii)後裔檢定林分作育種價預測以及遺傳增益估計,共有18個單親家系。原始區集配置係採「逢機完全區集設計」(randomized complete block design, RCBD),共有四個東西向緊鄰並排的縱向完全區集,但經由群內相關係數分析得知,區集配置應採二個南北向之橫向區集規劃較為妥當。更改區集配置後之數據結構呈不完全區集設計,且邊界效應嚴重影響邊緣木的橫向生長性狀,故剔除邊緣木後重作分析。 本研究發現2H、27WD之生長不受環境影響,且家系間沒有顯著差異,其他生長性狀6H、25H、25DBH、25DBA、25VOL、GR2-6以及GR6-25在區集間(P=0.0005~0.0229)與家系間(P=0.0163~0.1088)皆有顯著差異。高生長2H、25H有家系與區集間交感效應,而橫向生長25DBH、25DBA以及25VOL幾乎消除交感效應(0%)。 25H具高度個體遺傳率(0.61),且樹高之遺傳率通常會隨林齡增加而增大,林木的高生長較橫向生長的遺傳率(0.39~0.48)要高。樹高在不同林齡之間存在著高度的遺傳相關性(0.61),同林齡之樹高與橫向生長也存在著高遺傳相關性(0.59~0.76)。本後裔檢定由於所估計之木材密度缺乏家系變異,因而無法估計木材密度與其他生長性狀之「遺傳相關性」。 以一個單位(i=1)的表型偏差作為選拔差下所估計之遺傳增益有2.29公尺25H、2.34公分之25DBH以及0.08立方米之25VOL。以6H進行早期選拔,可以獲得1.10公尺之25年生樹高,相當於每年可以獲得0.18公尺,較直接選拔25H每年所獲得之0.09公尺之遺傳增益要高。 無論家系視作「固定效應」或是「逢機效應」,各性狀之MEAN與BLUP二育種價之相關係數也相當高(0.9231~0.9572)。在BLUP中,家系7、9、10、12與23,這五個家系在二個不同林齡之樹高皆入選前九名;同時入選25H與25DBH二性狀之前九名家系有7、9、10、11、23以及29家系;若同時考慮入選6H、25H與25DBH三性狀之前九名家系有7、9、10和23號家系;同時落在最後七名之家系有8、21、31及33家系。以雙性狀模式(25H和25DBH)估計25DBH之育種價,前九名之家系可以改良幅度為10.29%∼0.97%,對多性狀選拔為良好之策略。本研究結果對於香杉遺傳選拔的可能發展,提供了足以參考的經驗。
Eighteen open-pollinated (OP) families of a local variety of China fir (Cunninghamia lanceolata (Lamb.) Hook. var. konishii) were grown in a field trial to evaluate the magnitude of genetic parameter estimates and ranking of the parental clones, for growth (height, DBH, and diameter at basal area; DBA) and wood density traits up to 27 years. The field-trial layout was randomized complete block design (RCBD) with four adjacent south-north longitudinal blocks. After checking the intra-class correlations it was found that it would be better to divide the 4 east-west blocks into 2 south-north blocks, so that the long axis of the block is parallel to the contour. That made the field design incomplete blocks. It was also found that edge effects seriously influenced the stem diameter growth for the border trees. We therefore also deleted those seriously biased border trees from the data. Statistical analyses showed second-year height and 27-year wood density were not influenced by environmental impacts, and there were no significant differences among families. Other growth traits (six-year height, 25-year height, 25-year DBH, 25-year DBA, 25-year stem volume, relative growth between second- and six-year heights, and relative growth between six- and 25-year heights) showed significant differences among families and among blocks. Height growth (second-year and 25-year heights) showed significant family by block interactions. Stem diameter growth (25-year DBH, DBA, and Volume) traits had no interactions at all. Tree height between ages demonstrated substantial genetic correlation (0.61). Tree height and stem diameters of the same age also exhibited substantial genetic correlations (0.59). Twenty-seven-year wood density showed no family variation. Twenty-five-year height showed high individual-tree heritability. Heritability estimates increased as the trees grew older. Height growth traits in general have higher heritabilities than stem diameter growth. Using one unit of standard phenotypic deviation as selection differential to estimate genetic gain would result in 2.29m in 25-year height, 2.34cm in 25-year DBH, and 0.08m3 in 25-year stem volume. Using six-year height as an early-selection trait, indirect selection would result in 1.10m gain in 25-year height. This is equivalent to 0.18m per year, which is much higher than a gain of 0.09m direct selection in 25-year height. No matter whether family effect is taken as “fixed effect” or “random effect”, arithmetical means and BLUP values showed high correlations (0.9231~0.9572), which implies that both are reliable estimates of the breeding values. Using BLUP values to rank families, five families were both selected into the top nine families for tree height at different ages. Using bi-character (25-year height and DBH) to adjust the estimate of breeding values of 25-year DBH is effective. The improvement magnitude of the top nine families is from 10.29% to 0.97%. This method was therefore proven to be a superior strategy of multi-trait selection. Current research results provide a useful reference for genetic selection of China fir in the future.
URI: http://hdl.handle.net/11455/65884
Appears in Collections:森林學系

文件中的檔案:

取得全文請前往華藝線上圖書館



Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.