應用遺傳與環境交感作用模型於藍瑞斯母豬窩仔數性狀遺傳評估分析

Abstract

藍瑞斯為台灣主要的母系品種之一，然而，台灣的高溫高濕環境可能對其繁殖性能，例如總仔數（TNB）和活仔數（NBA）產生負面影響，因此，透過動物育種模型的改良，對於選拔耐熱母豬提供潛在的解決方法。在動物育種模型中，氣候效應作為類別因子，並可轉換為溫濕度指數（THI）的形式，THI 同時考慮藍瑞斯母豬哺乳到分娩期間的環境溫度與相對濕度。首先，遺傳參數估計可了解台灣藍瑞斯母豬窩仔數性狀的遺傳率和遺傳相關；隨後，反應規格模型（RNM）使得THI 和累加性遺傳效應進行交感作用分析，用以觀察遺傳率、遺傳相關和育種價在不同 THI 下的變化。在本研究中，首先使用線性混合模型（Linear Mixed Model, LMM）評估台灣藍瑞斯族群中 TNB 和 NBA 的遺傳參數，並使用同質性殘差模型（homogeneous residual model）進行分析，結果顯示，TNB 和 NBA 的遺傳率分別為 0.164 和 0.111，兩性狀間的遺傳相關為 0.911。除了使用同質性殘差模型的分析外，並進一步使用異質性殘差模型（heterogeneous residual model）進行分析，該模型將氣候因子與殘差進行交感，結果顯示，TNB 的殘差方範圍為 8.934 至 11.113 且 NBA 的殘差變方範圍為 8.196 至 9.811。因此，不同氣候分類下 TNB 的遺傳率範圍在 0.186 至 0.156 之間，NBA 的遺傳率範圍在 0.125 至 0.106 之間。以上結果顯示，應根據不同氣候分類對母豬的管理與營養進行調整。在 RNM 模型中，將母豬從配種前到懷孕期切分為四個階段，包括配種前 28 天（before）、懷孕前期（early）、懷孕中期（middle）和懷孕後期（late），並估算 THI 對 TNB、NBA 和死胎率（STBR）的影響。在模型中，配種前 28 天的階段與 TNB 和 NBA 的累加性遺傳效應進行交感作用，而懷孕前期則與 STBR 的累加性遺傳效應進行交感作用。結果顯示，TNB、NBA 和 STBR 的遺傳率隨著 THI 從 54 到 83 呈現下降趨勢，分別為 0.23 - 0.11、0.18 - 0.08 和 0.10 - 0.04。三個性狀的最低遺傳相關分別為 0.85、0.64 和 0.80。另一方面，與傳統動物育種模型為每個個體估算育種價（bvc）相比，RNM 為每個個體提供兩個不同的育種價，其中一個是不隨 THI 變化的育種價（bv1），另外則是隨著 THI 變化的育種價（bv2），因此，RNM 所估計的育種價在 THI 的變化下，個體的育種價排序將有所不同。總結來說，在傳統動物育種模型中，異質性殘差顯示台灣藍瑞斯的窩仔數性狀與殘差間有交互作用產生。此外母豬不同懷孕階段的 THI 與累加性遺傳效應的交感也顯示遺傳率和遺傳相關將會隨著 THI 的變化而有不同。因此，應根據不同氣候條件對母豬的管理和營養進行調整，同時，在 RNM 的分析中所得出的 bv1 和 bv2 應用於藍瑞斯母豬窩仔數性狀對 THI 變化的選拔計畫中。

Landrace is one of the major dam lines in Taiwan. However, the environment in Taiwan, which have high temperature and high humidity, could cause negative effects of their reproductive performance, such as total number born (TNB) and number born alive (NBA). Therefore, the modification of a statistical animal model might provide the potential solution to select the sows with heat tolerance. The climate effect, which is a categorical factor in a statistical animal model, could be redefined by temperature-humidity index (THI). The THI considers ambient temperature and relative humidity together from lactating to farrowing in Landrace sows. Firstly, the estimated genetic parameters might provide the fundamental knowledge for each trait in the population. After that, the THI factor was interacting with the additive genetic applied in the model, called Reaction Norm Model (RNM), to observe the changes of genetic parameters and breeding values across THI. In this study, the genetic parameters of litter traits, TNB and NBA, were evaluated firstly for long-term data in Taiwan Landrace population using Linear Mixed Model (LMM) with homogeneous residual variance. The results showed that the heritability of TNB and NBA were 0.164 and 0.111, respectively. The genetic correlation between the two traits was 0.911. In addition to the analysis of LMM with homogeneous residual variance, a LMM with heterogeneous residual variance which used the climate factor interacting with residual variance was conducted with the data. The residual variances ranged from 8.934 to 11.113 for TNB and 8.196 to 9.811 for NBA. Thus, the ranges of heritability fluctuated in TNB (0.186 - 0.156) and NBA (0.125 - 0.106) under different climates. These results indicated that the adjustments of management and nutrients for the sows should be considered in different levels of the climate. In the RNM model, the lactation and pregnancy of sow were divided into four stages, including before, early, middle and late to estimate the impact of average THI in TNB, NBA and stillborn rate (STBR). Then the before stage interacted with additive genetic effect for TNB and NBA, while the early stage interacted with additive genetic effect for STBR. The results showed that the decreasing trend of heritability showed 0.23 – 0.11, 0.18 – 0.08, and 0.10 – 0.04 from 54 to 83 of THI for TNB, NBA and STBR, respectively. The lowest genetic correlations among THI were 0.85, 0.64, and 0.80 for TNB, NBA and STBR, respectively. On the other hand, compared to the conventional LMM with a single breeding value estimated for each individual (bvc), the RNM provides two breeding values for each individual: one breeding value does not change along with THI (bv1), and the other breeding value (bv2) shows a changing trend across THI. Thus, the breeding values predicted by the RNM would genetically re-rank the animals across THI values. In conclusion, the estimated genetic parameters were revealed in several typical Taiwan Landrace herds. The fluctuation of residual variances in the Landrace sow population was shown to have an interaction with the climate in Taiwan. Otherwise, the interaction between THI in pregnant stages and additive genetic effect showed the changing trend of heritability and genetic correlation across THI for litter traits. Therefore, management and nutrients should change in different levels of the climate. At the same time, the estimated breeding values across THI in the RNM should be considered in the genetic improvement programs for litter trait performance in Taiwan Landrace population.