94

種豬選拔指數敏感度分析

黃鈺嘉(1) 顏念慈(1) 賴永裕(1) 李世昌(1) 吳明哲(1) 張秀鑾(2)

(1)行政院農業委員會畜產試驗所 (2)國立屏東科技大學畜產系

以日增重(kg/d)、飼料效率及背脂厚度(cm) 1.0: -1.0 : -0.5 及1996年美國豬種改進聯盟推薦的制式豬種改良程序遺傳參數資料為中心，進行敏感度模擬分析。結果發現，遺傳參數變動對指數的係數變動影響很大，相對的有限經濟加權值變動的影響則較可預測，如經濟加權值為 1.0 : -1.0: -0.5 改變至 1.0: -1.2 : -0.8 則指數 b值由 75.2、 -48.0、 -39.6，變化為 60.5 、 -42.2、 -43.7。若性狀間存在高度的表型相關，如日增重與飼料效率間存在高度的表型相關(≦ -0.6，如 –0.7、–0.8、-0.9)，會造成如同迴歸分析中多重共線的迴歸係數估計問題，影響指數的穩定性(指數係數正負號與經濟加權值正負號不一致)。池(1980)首次研訂台灣種豬選拔指數後，已使用於檢定站二十多年，由於豬種在進步，飼料、工資及醫療費用隨國際榖價、國民所得與時間而波動，二十年前首訂的台灣的種豬的選拔指數需要適時加以修訂。依敏感度分析結果，建議小幅度修正經濟加權值，並重新彙整國內外遺傳參數文獻，引用折衷之遺傳參數，作為重新制定中央檢定指數之依據。另一個替代方案則為以混合模式估計各性狀個體育種價，再直接以經濟加權值等比平移轉換( Ī = 100, σI = 25)計算種豬指數。

關鍵語：選拔指數、經濟加權值、遺傳參數。

SENSITITVE ANALYSIS OF SELECTION INDEX FOR PIGS

Y. C. Huang(1), N. T. YEN(1), Y. Y. Lai(1), S. Y. Lee (1), M. C. Wu(1) and H. L. Chang(2)

(1) Livestock Research Institute, Council of Agriculture, Executive Yuan
(2) Department of Animal Science, National Pingtung University of Science and Technology

Based on economic weights of ADG (kg/day), FE (Feed/Gain), and BF thickness (cm) with ratio of 1.0 : -1.0 : -0.5 and genetic parameters of Guidelines for Uniform Swine Improvement Programs (The National Swine Improvement Federation, NSIF, Cynthia and Todd, 1996), trails of sensitive analysis were simulated. Genetic parameters had considerable influence on estimated coefficients of selection indices. However, reasonable changes of economic weight will not sway of b vector of index to unpredictable range. For example, changes of the economic weights from 1.0 : -1.0: -0.5 to 1.0: -1.2 : -0.8 would change the b vector from 75.2, -48.0, -39.6 to 60.5, -42.2, -43.7 only. However, high phenotypic correlation(rp), such as rp of ADG and FE ≦ -0.6(-0.7, -0.8, or -0.9), may produce unstable estimates as multicollinearity problems in the regression analysis. Because, swine selection indices of Taiwan were first proposed by Chyr(1980) in 25 five years ago. Many improvements achieved in the pig industry, reevaluation of the swine selection index becomes crucial for further efficient breeding practice. For changes of the new index, suggestions includes, 1) re-ratio the economic weights by new market information 2) review new published genetic parameters of Taiwan and other countries. Suggestions 1) and 2) can be used to rebuild the selection index. But, alternative option is using the mixed model to estimate breeding value of each trait of individual first, and then weight traits by economic values directly, finally, transform the coefficients of index to achieve Ī = 100, σI = 25.

Key Words: Selection index, Economic weight, Genetic parameters.