Genetic evaluation of production, reproduction and survival in a flock of Ethiopian Horro sheep

English: Genetic analyses of economically important traits were carried out using data collected for 20 years (1978-1997) from a flock of Horro sheep at Bako Research Center, Ethiopia. (Co)variance components and genetic and environmental parameters were estimated for pre- and post- weaning average daily gain (ADGl, ADG2) and Kleiber ratio (KRl, KR2), birth weight (BWT), weaning weight (WWT) and bi-monthly weights to 12 months of age (WT2 to WTI2), and weight at 18 months of age (WTI8). using ASREML. Twelve models, formed with inclusion or exclusion of the maternal genetic, permanent environmental, and temporary (litter) environmental variance components and the covariance between the direct and maternal additive effect on the basic direct additive genetic model, were used. Bivariate analyses were also done. Maternal genetic and temporary environmental components were found to be important (P<0.05) sources of variation for ADG 1 and KRl while only the temporary environmental component was found to be important for ADG2 and KR2. Total heritability estimates for ADGl, ADG2, KRl and KR2, were 0.13±0.04, 0.04±0.03, 0.13±0.03, and 0.01±0.02 respectively. For weights to about six months of age, the inclusion of the temporary maternal effect in the models was found to have significant (P<0.05) contribution to the total variance. Maternal genetic components were important for weights until about eight months of age, while the direct-maternal covariance was important for BWT, WT2, WWT and ADGl. Direct genetic correlations of ADGl with BWT, WWT and WT6 were 0.04±0.21, 0.96±0.02 and 0.92±0.09 while with KRl they were -0.41±0.21, 0.74±0.10 and 0.66±O.16 respectively. The inclusion of maternal genetic, temporary and permanent environmental effects in analytical models for early weights improved model fit. Heritability estimates of average daily gains and Kleiber ratios are low to moderate. The correlations with weight traits, with the exception of BWT, are moderate to high. Thus, it seems more practical to select on weight traits to improve gain and efficiency. Genetic and maternal environmental parameters of perinatal, pre- and post-weaning survival in Horro lambs were studied using 3864 lamb records. The perinatal period covered the first three days after birth (3DS) while pre-weaning (3MS) and post- weaning (3-6MS) survival were taken as survival in the pre- and post-weaning three-month periods respectively. Cumulative survival to six months of age (6MS) was also considered. The same twelve linear genetic models used for growth traits were applied. Additionally logit and probit analyses were carried out. The most appropriate model was chosen which was consistently superior in all of the analytical methods and used to estimate genetic and environmental parameters. A bivariate analyses between survival on the observed scale and birth weight was carried out. For survival to all ages, the model, which included the direct genetic and maternal temporary (litter) environmental effect was found to be the most consistent of all the models compared. Heritability estimates for 3DS, 3MS, 6MS and 3-6MS were 0.00±0.00,O.02±0.03, 0.05±0.03 and 0.07±0.03 respectively. Genetic correlations with birth weight were 0.20±0.23, 0.68±0.34, 0.26±0.17 and 0.45±0.26 respectively, for 3DS, 3MS, 6MS and 3-6MS. The low heritability estimates of survival in this study suggest acceptable progress with respect to this trait through selection will be difficult to attain. The relatively higher genetic correlation of 3MS with birth weight indicates survival may be improved indirectly through selection for birth weight. Despite the low rate of prolificacy (34% twinning) there exists a sizeable litter effect on survival. This should be looked into, since this may entail a separate management for twin born lambs. Genetic parameters of productivity expressed as ewe fertility, incidence of twinning, number of lambs weaned, and total lamb weight at birth (TBW) and at weaning (TWW) per ewe per lambing were estimated using direct additive and repeatability models. Fertility and twinning were analyzed by a logit analysis, using similar models. In the analysis of fertility, service sire was included either as random or fixed effect. Additionally cumulative total weight of lamb weaned in the first four parities (TWW4) was also analyzed, using the direct additive model. Except for fertility where service sire was considered as random effect, the repeatability model was found to have a better fit than (P<0.05) the direct animal model. Bivariate analyses were done, using the direct animal model. Direct heritability estimates for fertility, twinning and total weight of lambs at birth and at weaning were 0.02±0.02, 0.15±0.03, 0.08±0.02 and 0.04±0.02 respectively. Where service sire was considered as fixed effect, heritability estimate of fertility from a direct model increased to 0.06. Heritability estimates for number of lambs weaned were 0.03 and 0.00 for the direct and repeatability models respectively. Repeatability estimates were 0.10, 0.07, 0.16, 0.06, 0.02 and 0.08 for TBW, TWW, twinning, litter size weaned and fertility under random and fixed sire models respectively. Estimate of the variance ratio in fertility attributed to service sire was 0.16±0.02. Heritability estimate for TWW4 was 0.05. Genetic correlations among TBW, TWW, twinning and number of lambs at weaning were in the range of 0.57 to 0.86. Except for twinning, heritability estimates were very low. Twinning was found to have medium heritability and repeatability, and moderate to high genetic correlation with number of lambs weaned, TBW, and TWW. These suggest that twinning can be used as a selection criterion for improvement in productivity despite increase in lamb mortality with increase in twinning. Weight (kg)-age (days) data of 524 Horro sheep surviving past 3.5 years of age were fitted to Brody's function to estimate growth curve parameters. Genetic and phenotypic parameters of the curve parameters and their relationships among themselves and with birth weight (BW), weaning weight (WW), 6-month weight (WT6) and yearling weight (YW) were estimated using direct univariate and bivariate animal models. Least squares means of growth curve parameters: A (asymptotic mature weight, kg), B (proportion of mature weight attained after birth) and k (rate of maturity, kg gain k-1 body weight day-1) were 37.6, 0.88 and 0.27x10-2 respectively. Female and male Horro sheep have asymptotic mature weights of 33.3 kg and 41.8 kg respectively. Heritability estimates were 0.29±0.10, 0.18±0.09 and 0.14±0.09 for A, Band k respectively. Genetic correlations between A and B, A and k, and B and k were 0.39±0.31, -0.07±0.34 and 0.25±0.39 respectively. Genetic correlations of A with BW, WW, WT6 and YW were 0.27±0.34, 0.34±0.27, 0.44±0.20 and 0.67±0.17 respectively. Genetic correlations of k with BW, WW, WT6, and YW were -0.13±0.69, 0.37±0.35, 0.61±0.25 and 0.66±0.22 respectively. Presence of medium heritability for A and k indicates that progress in improving these traits can be made through selection. WT6 and YW have medium genetic correlations (with reasonable standard error) with the growth curve parameters and these may allow the use of these weights as an indirect selection criterion to shorten generation interval. Weight data from birth to about one year (maximum of 408 days) were analyzed using the random regression model. A total of22149 weight records of 1951 lambs were used. Three different orthogonal (legendre) polynomial orders (three, four and five) of fit for additive genetic and animal permanent environmental effect were considered along with two different error measures. The residual error for weight at the different ages was assumed to be either homogeneous (one error measure) throughout the growing period or heterogeneous (four error measure classes: birth, 10 to 90 days, 91 to 180 days and >180 days). Likelihood ratio test (LRT) and Akaike's Information Criteria (AIC) were used for model comparison. Model fit improved with increased order of fit and with assumption of heterogeneity of error variance. Sizeable differences were found in heritability estimates from the different models, particularly for weight at birth and early age. Components for both additive genetic and animal permanent environmental covariances increased with an increase in age. Heritability estimates from the 'best' model have shown declining and increasing values at different parts of the trajectory. The lowest estimate was 0.14 for weight at birth, while the highest estimate was 0.40 for weight at about 400 days of age. The ratio of permanent environmental variance to the total variance increased from 0.42 at birth to 0.67 at about 270 days and decreased slightly thereafter. With the exception of weight at birth and at about yearling, heritability estimates from this study were lower than estimates from univariate studies where maternal effects were considered. The difference between estimates decreased with age, implying the higher estimates from this study are a result of ignoring maternal genetic and environmental (temporary and permanent) effects. Therefore consideration of maternal genetic and environmental effects needs attention. Choice of appropriate order of polynomial has significant influence and increased order of polynomial may need to be tested to refine estimates with a larger data set (additional weights per animal).