# Estimation of Genetic Parameters of First Lactation and Herd Life Traits by Different Animal Models in Murrah Buffaloes

2 Department of Animal Genetics and Breeding College of Veterinary and Animal Sciences G.B. Pant University of Agriculture and Technology Pantnagar–263 145 Distt, U.S., Nagar (Uttarakhand), India

Author Correspondence author

Animal Molecular Breeding, 2016, Vol. 6, No. 5 doi: 10.5376/amb.2016.06.0005

Received: 11 May, 2016 Accepted: 01 Jun., 2016 Published: 15 Jul., 2016

Yadav T.B. and Singh C.V., 2016, Estimation of Genetic Parameters of First Lactation and Herd Life Traits by Different Animal Models in Murrah Buffaloes,Animal Molecular Breeding, 6(5): 1-12 (doi: 10.5376/amb.2016.06.0005)

In order to compare different animal models, the methodology of mixed models under animal models was used to predict (Co) variance components of 8 traits related to production, reproduction and life time traits of 1 312 Murrah buffaloes descendent from 122 Sires and 341 dams raised at four military dairy farm viz. Ambala, Agra, Lucknow and Bareilly.The least square means (LSD) of body weight at first calving, age at first calving, first dry period, first calving interval, first service period first lactation milk yield, herd life milk yield, and first lactation period were 503.73±2.68 kg, 1 268.39±15.32 days, 173.01±5.18 days, 467.74±7.42 days, 171.37±7.04 days, 1 702.44±31.015 kg, 5 459.97±76.21 kg, 296.30±3.95 days respectively under model 2 and 501.72±2.58 kg, 1281.33±11.89 days, 173.38±4.49 days, 468.39±6.12 days, 171.74±5.89 days, 1 709.04±25.04 kg, 5 457.49±65.01 kg, 296.74±3.17 days respectively under model 8.The least squares means estimated by model 8 were slightly higher than the means estimated by model 2.The sire had accounted more variation under model 8 than model 2 for body weight at first calving. Since all the traits were analyzed simultaneously under model 2, therefore, there may be some confounding among traits which decreased sire contribution in total variation with body weight at first calving. But that was not the case with model 8 as all the traits were analyzed separately.Additive (direct), environmental and phenotypic variances estimated from univariate and multivariate animal models agreed closely for all the traits. Similarly variance estimated from model 2 and model 8 analyses agreed for environmental and phenotypic variance for all eight traits. Model 2 had slightly higher additive genetic variance for FLP, FDP, FCI and FSP but lower for WFC, AFC, FLMY and HLMY, than model 8. Univariate animal model had slightly lower additive (direct) variance for WFC and AFC than the values estimated under multivariate animal model. Higher additive covariance was estimated between WFC and HLMY followed by FLMY, whereas negative covariance was estimated between WFC and AFC. FLMY had higher additive covariance with HLMY. Additive covariance between AFC and HLMY was higher followed by FLMY.Model 2 had higher h^{2} estimates among all the four methods employed for all traits, except WFC, and FLMY. The heritabilities estimates under univariate are generally lower than estimates of model 8, except for WFC and HLMY. Higher h^{2} estimates for AFC, FLP, FDP, FCI, FSP and FLMY, under model 8 were observed than the univaritate animal model. Estimates of heritabilities for FDP and HLMY from multivariate analysis were almost similar to h^{2} estimates of univariate but slightly lower for WFC and FLMY. The genetic and phenotypic correlations among all first lactation and life time performance traits ranged from very low to very high. FLMY had positive genetic and phenotypic correlations with LTMY.Genetic correlations among different traits ranged from very low to very high under model 2. Phenotypic and environmental correlations also showed same trend.

**1 Introduction**

**2 Materials and Methods**

**2.1 Statistical methodology**

**2.2 Univariate**

**2.3 Multivariate model**

**3 Results and Discussion**

Table 1 Least squares means for WFC, AFC, FDP,FCI, FSP, FLMY, HLMY and FLP under model 2 |

Table 2 Least squares means for WFC, AFC, FDP,FCI, FSP, FLMY, HLMY and FLP under model 8 |

Table 3 Observed between and within sire variance (expressed as percentage of phenotypic variation) for WFC, AFC, FDP, FCI, FSP, FLMY, HLMY and FLP under model 2 and 8 Note: σ2 s=between sire, component of variance, σ2w=within sire component of variance PHS=Paternal half-sibs |

Table 4 Estimates of variance components and heritability for WFC, AFC, FDP, FCI, FSP, FLMY, HLMY and FLP traits from model 2 |

Table 5 Estimates of variance components and heritability for WFC, AFC, FDP, FCI, FSP, FLMY, HLMY and FLP traits from model 8 of LSA |

Table 6 Estimates of variance components and heritability for WFC, AFC, FDP, FCI, FSP, FLMY, HLMY and FLP traits from univariate REML analysis |

Table 7 Estimates of variance components and heritability for WFC, AFC, FDP, FCI, FSP, FLMY, HLMY and FLP traits from multivariate REML analysis |

^{2}estimates for weight at first calving were 0.038, 0.39, 0.09 and 0.08 respectively under model 2 and model 8, univariate and multivariate REML analysis. The univariate and multivariate REML animal model estimated higher h

^{2}than the LSA methods. Because of its desirable properties, the univariate and multivariate animal model estimators were considered to be more appropriate than that of LSA method. The above differences observed among h

^{2}estimates may be due to the different methods applied in the present study. The low h

^{2}estimates of weight at first calving under different methods indicated the presence of non additive genetic variance in the herd. Therefore, weight at first calving may be improved through better feeding and management of the herd. The h

^{2}estimated by four methods for age at first calving were 0.39, 0.44, 0.32 and 0.27 respectively under model 2 and model 8 univariate and multivariate REML analysis. The moderate estimates of the heritability for age at first calving suggested that there is chance of selection of animals for age at first calving as sufficient amount of additive genetic variability exists in the herd and age at first calving could be reduced upto optimal level. The h

^{2}estimated by four methods for first lactation period were 0.26, 0.24, 0.18 and 0.20 respectively under model 2, model 8 univariate and multivariate REML analyses. The h

^{2}estimated by four methods for first dry period were 0.14, 0.12, 0.09 and 0.09 respectively under model 2, model 8 univariate and multivariate REML analyses. The low magnitude of h

^{2}clearly indicated that this is influenced by the environmental factors and this trait may be improve by better feeding and management practices. The h

^{2}estimated by four methods for first calving interval were 0.22, 0.21, 0.18 and 0.19 respectively under model 2, model 8 univariate and multivariate REML analyses. The h

^{2}estimated by four methods for first service period were 0.18, 0.14, 0.08 and 0.12 respectively under model , model 8 univariate and multivariate REML analyses. The lower estimates of h

^{2}for first service period reveal that all the variation in the first service period present are due to non-additive genetic causes, viz. management, nutrition and season etc. The h

^{2}estimated by four methods for first lactation milk yield were 0.27, 0.28, 0.23 and 0.22 respectively, under model 2 and model 8, univariate and multivariate REML analyses. The h

^{2}estimated by four methods for herd life milk yield were 0.18, 0.17, 0.19 and 0.19respectively under model 2, model 8 univariate and multivariate REML analyses. These results were in close agreement with those reported by Suresh et al. (2004), and Kumarvelu et al. (2006), Dass and Sadana (2000), Suresh et al. (2004), Yadav et al. (2007), and Singh and Barwal (2012).

^{2}estimates of univariate but slightly lower for WFC and FLMY. AFC had the highest h

^{2}estimates followed by FLMY (0.23), HLMY (0.19), FLP (0.18), FCI (0.18) and WFC (0.09) while FSP had lowest h

^{2}estimate under univaraite REML. Heritability estimated under multivariate REML analysis also displayed the similar tendency as h

^{2}estimates under univariates. The results revealed that the h

^{2}estimated under both models of REML were found almost similar or near to similar value.

Table 8 Heritability (±S.E.) estimates for WFC, AFC, FDP, FCI, FSP, FLMY, HLMY and FLP traits from obtained by different methods |

^{2}estimates among all the four methods employed for all traits, except WFC, and FLMY. Raheja et al. (2001) reported lower h

^{2}estimates by restricted maximum likelihood and maximum likelihood procedures than the paternal half sib based on least squares mixed model methodology for age at first calving (AFC) and first lactation milk yield (FLMY). Univariate and multivariate animal models had similar heritabilities for all trait except AFC but the heritabilities estimated by model 2 differ from model 8 for all traits, except WFC. Raheja (1992) stated that the heritabilities calculated from use of residual and sire variances, obtained from single traits (Henderson Model 3) were over estimated by about 15-20%.

^{2}estimates for AFC, FLP, FDP, FCI, FSP and FLMY, under model 8 were observed than the univaritate animal model. The differences observed under different models could be due to model effect and factors considered under different models. The model 2 and model 8 used did not take in to account all relationship between animals but the animal model fitted does. Since sire was the random effect in model 2 and 8, using paternal half-sibs, therefore, there was no maternal effect included. Such types of paternal half sib’s estimates of h

^{2}might be reduced due to selection of sires.

^{2}estimates when using an animal model with real data. When the maternal variance components and the direct maternal covariance are not fitted in an animal model, this variation is included in the animal additive genetic components of variance. Out of 15 variance ratios obtained from DFREML, 5 were further away from the best solution than 0.03. Since the positive trait correlation has a direct physical origin, it might be expected that their variances, or at least their heritabilities, would show positive correlation (Koots et al., 1994a).

^{2}estimates under REML method were found lower than the estimates obtained under LSA method but the former were more reliable as they had similar S.E. of variance using REML procedure and preferable to those from an animal model accounting for direct effects only (Raheja et al., 2001).

^{2}estimates under model 2 in general had higher standard error for all traits under the study and multivariate animal model had lowest, with same experimental data. The standard errors of h

^{2}estimates are an obvious criterion for comparing the precision, but not the bias of the estimation procedure (Raheja, 1992).

Table 9 Covariance matrices for WFC, AFC, FDP, FCI, FSP, FLMY, HLMY and FLP from model 2 Note: Where,a. Indicates direct additive genetic covariance; b. Indicates environmental covariance; c. Indicates phenotypic covariance |

**4 Conclusion**

^{2}estimates among all the four methods employed for all traits. The heritabilities estimates under univariate are generally lower than estimates of model 8. Genetic correlations among different traits ranged from very low to very high under model 2. Phenotypic and environmental correlations also showed same trend.

. PDF(397KB)

. HTML

Associated material

. Readers' comments

Other articles by authors

. Singh C.V.

Related articles

. Weight at first calving

. Age at first calving

. First lactation milk yield

. Herd life milk yield

. Univariate and multivariate animal models

Tools

. Email to a friend

. Post a comment