Abstract: In this study, a total of 733 milk yield records of Brown Swiss cows raised at Konuklar State Farm in Konya Province in Turkey were used for estimation of phenotypic and genetic parameters for milk yield, lactation period and fat percentage. The Phenotypic and genetic parameters were estimated by the MTDFREML program using Multiple Trait Animal Model. The model included individual, permanent environment and errors as random effects, year and season of calving, parity, year and age as fixed effects and days in milk as a covariate for milk yield; milk yield as a covariate for lactation period and milk yield and lactation period as covariate for fat percentage. Genetic parameters and breeding value of cow, sire and dam for milk yield in kg, for lactation period in days and for fat percentage in percent were estimated. Cow breeding values ranged from -3006-1724 kg for milk yield, from 10.81-14.22 days for lactation period and from -1.48-0.97% for fat percentage. Likewise, dam breeding values ranged from -1628-862 kg, from -5.69-7.74 days and from -0.76-0.48% for the same traits, respectively. Sire breeding value ranged between -1129 and 862 kg, -8.63 and 5.73 days and -0.68 and 0.83% for the above mentioned traits, respectively. Estimates of heritability were 0.33, 0.11 and 0.39 for milk yield, lactation period and fat percentage, respectively. The genetic correlation between milk yield and fat percentage was positive and high (0.95), whereas the genetic correlation between lactation period and milk yield and between lactation period and fat percentage was negative, -0.49 and -0.73, respectively. Repeatability estimates were 0.34, 0.47 and 0.54 for the same traits, respectively.

INTRODUCTION

Accurate knowledge of genetic parameters is required in a selection effort, especially one using selection index with multiple traits (Falconer and Mackay, 1996). Heritabilities are used to estimate genetic change in between generations and genetic correlations are used to estimate how traits change in the next generation in relation to each other. Selection schemes as an alternative to progeny testing depends on the heritability of the trait considered (Santus et al., 1993). Estimates of repeatability help culling decisions; animals with an inferior performance in high-repeatability traits may be culled early.

Brown Swiss cattle are quite common in Turkey and they are in need of genetic improvement, which requires selection for various traits. Though Holsteins have been replacing these cattle, they do not have the high capacity for fat percentage and do not have the adaptability of Brown Swiss to environmental conditions in Turkey. Brown Swiss cattle are relatively low maintenance and are thus, preferred by the common farmer in Turkey.

Estimating genetic parameters in these cattle can help calculate more accurate estimates of genetic improvement and increase accuracy of breeding value calculations (Falconer and Mackay, 1996). The estimations give an idea to breeders and farmers what to expect in a selection program and what kind of a genetic improvement scheme should be used.

Milk traits are influenced by maternal effects and permanent environmental effects in addition to direct genetic effects. Including maternal effects in the model decreases the variance of direct genetic effects (Meyer, 1992; Hoque et al., 2007). These effects should be taken into consideration in a selection program, increasing the accuracy of the estimates (Meyer et al., 1994).

Major aim of this study, was to estimate heritabilities, repeatabilities and genetic correlations for milk yield, fat percentage and lactation period in Brown Swiss cattle raised in Konya, Turkey.

MATERIALS AND METHODS

The data used in this study were collected from Brown Swiss cattle reared at Konuklar state farm in Turkey. The 203 cows, 182 dams and 41 sires constituted pedigree data. Cows were artificially inseminated by using frozen semen. Parity of cows varied from 1-9, year-season of calving from January-December, year 1984-1996 and age 2 from 14. Records arranged with integer fixed fields to left (parity, year-season of calving, year and age) and real fields to right (Lactation Period (LP) as a covariate for Milk Yield (MY), MY as a covariate for LP and MY and LP as covariate for Fat Percentage (FP)). Data were analyzed by Multiple Trait Derivative Free Restricted Maximum Likelihood (MTDFREML) according to Boldman et al. (1995), using repeatability animal model multiple trait analysis. Table 1 shows the data structure considered in the analysis, mean of Milk Yield (MY) in kg, Days in Milk (DIM) and Fat Percentage (FP) in days, number of mixed model equations and number of iterations.

To ensure global convergence, the algorithm by (Boldman et al., 1995) was restarted with estimates until the log likelihood did not change at the fourth decimal (Robison et al., 2002). The solutions given are from the final round of iteration. Fixed effects for the model included year and season of calving, parity, year and age and lactation period was included as a covariate for MY. Milk Yield was included as a covariate for lactation period and MY and LP were included as covariate for FP.

Permanent environmental effects for each cow were used to calculate the permanent covariance between each two traits, while the genetic and residual covariances were obtained using the Mixed Model Least Squares and Maximum Likelihood (LSMLMW) in the computer program of (Harvey, 1990) for all traits. Duncan multiple comparison test (Duzgunes, 1993) was used to test the differences between factors. Experiment was carried out according to Selcuk University, Faculty of Agriculture guidelines.

Table 1 shows the data structure considered in the analysis, means of Milk Yield (MY) in kg, Lactation Period (LP) in day, Fat Percentage (FP) in percent, number of mixed model equations and number of iterations.

Table 1:	Data structure, unadjusted mean, Standard Deviation (SD) and Coefficient of Variation (CV%) for Milk Yield (MY), Lactation Period (LP) and Fat Percentage (FP)

Variance components were estimated using the following animal model:

Variance-covariance structure of the model described by El-Arian et al. (2003) was used:

To estimate heritability (h²) and repeatability (r) the following equation was used:

The Mixed Model Equations (MME) for the Best Linear Unbiased Estimator (BLUE) of estimable functions of b and for the Best Linear Unbiased Prediction (BLUP) of a and p in matrix notation were as follows:

where, α₁ = σ²_e/σ²_α and α₂ = σ²_e/σ²_p.

RESULTS AND DISCUSSION

Unadjusted Mean and Standard Deviation (SD) for MY in kg, LP in days and FP in percent were 4713.12±1412.13 kg, 308.16±29.92 day and 3.67±0.06%, respectively (Table 1). Estimates of Coefficient of Variations (CV%) are given in Table 1. The highest CV% value for MY (29.96) reflects a medium variation between individuals. The lowest CV% value for FP (1.64) reflects a small variation between individuals. The Least Squares Means (LSM) and Standard Deviations (SD) of milk yield, lactation period and fat percentage according to year and season of calving, parity, year and age is given Table 2.

The effect of year and year-season of calving on milk yield was statistically significant (p<0.01). The highest milk yield was obtained from 1996 year and the lowest milk yield was obtained from 1984 year.

Table 2:	The Least Squares Means (LSM) and Standard Deviations (SD) of milk yield, lactation period and fat percentage according to year and season of calving, parity, year and age
a, b: Means in a column with different superscripts differ (p<0.01), A, B: Means in a column with different superscripts differ (p<0.05)

Table 3:	Estimates of variance and covariance components, heritability (h2), repeatability (t) and genetic correlation (rG), for Milk Yield (MY), Lactation Period (LP) and Fat Percentage (FP)
σ2a = Additive genetic variance; σa = Additive genetic variance, σ2p = Permanent environmental variance, σp = Permanent environmental covariance, σ2e = Temporary environmental variance, σe = Temporary environmental covariance, -2 log L = log likelihood, h2 = heritability, t = repeatability, rG = genetic correlation, R2 = determination coeefficient

Table 4:	Range of predicted Cows’ Breeding Values (CBV’s), Sires (SBV’s) and Dams (DBV’s) their accuracy for MY, LP and FP

On account of milk yield was occurred considerable differences until from 1984-1996. This might have resulted from effect of improvement level performed along the years in herd. The highest milk yield was obtained from winter season and the differences between winter and spring season was not statistically significant. Likewise, the lowest milk yield was determined to summer season and the differences between summer and spring season was not statistically significant. The differences between summer, spring and autumn, winter seasons were statistically significant (p<0.01). Increase in the milk yield of animals giving birth to the winter and spring seasons can be due to excessive green feed crops in the spring season and an applying of feed diet by concentrated feed in the winter season. The effect of age and parity on milk yield was statistically significant (p<0.05, p<0.01). The effect of investigated traits on lactation period was not statistically significant. The effect of year on fat percentage was statistically significant (p<0.01) and other traits not significant.

The heritability estimates in this study for MY, LP and FP were 0.33, 0.11 and 0.39, respectively (Table 3). The present estimate was higher than Espinoza et al. (2007) findings for MY as 0.14-0.17, Wiggans et al. (2002) findings for MY for Brown Swiss as 0.29, for FP as 0.26, Rosati and Van Vleck (2002) findings for Buffalo as 0.23 for MY and as 0.14 for FP, Ilatsia et al. (2007) findings for MY as 0.16 and for LP as 0.07, Ojango and Pollott (2001) findings for MY as 0.29 and for LP as 0.087, similar to Atil et al. (2001) finding for LP as 0.13, lower than Costa et al. (2008) finding as 0.32 for 305 day LP and Atil et al. (2001) finding for MY as 0.38.

Low h² estimates for LP (0.11) indicate that this trait is affected by environmental factors. Improvement of herd management, feeding, service period, arrangement of heat, proper milking, inseminated at proper time of animals by good quality semen would help in improving of LP. Medium h² estimates for MY and FP indicates that these traits can be improved by mass selection in addition to increased level of environmental conditions.

The repeatability estimates in this study for MY, LP and FP were 0.34, 0.47 and 0.54, respectively (Table 3). Repeatability estimates were higher than Ojango and Pollott (2001) for FP as 0.11, (Meyer et al., 1994) for MY as 0.228 for Hereford, Sawalha et al. (2005) as 0.36 for fat, as 0.52 for MY, lower than Paura et al. (2002) finding for MY and higher than for FP; Wiggans et al. (2002) for Brown Swiss as 0.47 for MY, as 0.42 for fat, Ilatsia et al. (2007) finding for MY as 0.49 and lower than for LP as 0.40, similar to Ojango and Pollott (2001) for MY as 0.34. The highest repeatability was obtained by FP. According to this, It is possible that to say sufficiently reliable of using to fat percentage at first lactation for early selection of animal.

The results in Table 3 show that the genetic correlation between MY and FP was positive and high (0.95), MY and LP was negative and medium (-0.49) and LP and FP was negative and high (-0.73). This result similar to Ozcelik and Dogan (1999) findings for MY and LP Genetic correlation as -0.16, Atil et al. (2001) findings for MY and FP as 0.43, different from Farhangfar et al. (2003) findings for MY and FP as -0.69, Rosati and Van Vleck (2002) findings for MY and FP as -0.08.

This result indicates that high yielding cows may have the capacity for high FP. The genetic correlation between MY and LP was negative, indicating cows with increased genetic capacity for LP may have lower capacity for MY. Similarly, cows with longer lactation periods may have lower capacity for FP. Based on these results, selecting animals based on longer lactation periods and higher milk yield can be difficult. However, selection using both milk yield and fat percentage should be simple and effective. Although, these are contrary to the expectations based on phenotypic observations, many times genetic correlations can be different than phenotypic correlations, even in the sign.

Estimates of minimum and maximum Predicted Breeding Values (PBV) and their accuracies for MY, LP and FP estimated from Cow Breeding Values (CBV’S), Sire Breeding Values (SBV’S) and dam breeding values (DBV’S) are given in Table 4.

Breeding values were calculated from 203 cows, fathered by 40 sires and mothered by 182 dams. Estimates of minimum and maximum predicted Breeding value and their accuracies for milk yield ranged from -3006 and 1724, 0.76-0.80 for cows; -1129 and 862, 0.38-0.90 for sires; -1628 and 862, 0.38-0.43 for dams, for lactation period ranged from -10.80 and 14.20, 0.45-0.61 for cows; - 8.60 and 5.70, 0.47-0.59 for sires; - 5.69 and 7.70, 0.22-0.30 for cows; for fat percentage ranged from -1.48 and 0.96, 0.59-0.79 for cows; - 0.68 and 0.82, 0.66-0.88 for sires; - 0.76 and 0.47, 0.42-0.60, respectively (Table 4). Obtained BV in this study for MY was higher than Espinoza et al. (2007) and Peixoto et al. (2006) findings.

Results in Table 4 show the importance of cow, since it gave the higher range of breeding value for MY, LP and FP. Thus, selection of cows for the next generation would lead to higher genetic improvement in the herd for these traits. Moderate improvement can be obtained with mass selection for milk yield and fat percentage because of the heritability value as 0.33 and 0.39, respectively. Also, the accuracy of the estimates of cow breeding value was higher than the accuracy of dam and sire breeding values, which may be due to the higher number of cows than dam and sire number.

CONCLUSION

This model caused the greatest differences between genetic and residual correlations, the highest heritability values for the milk yield and fat percentage, the highest values of EBV’s difference for the best and worst cows, as well as the greatest correlation among estimated genetic values. The present estimates showed large genetic differences between cows for different traits for milk yield and fat percentage, which indicate the high potential for rapid genetic improvement in milk traits of Brown Swiss cattle in Turkey through selection. So, the results presented here show that the multiple trait animal models could be used appropriately for genetic evaluation of milk yield, lactation period and fat percentage for Brown Swiss cattle in Turkey.

ACKNOWLEDGEMENT

We thank to Konuklar State Farm in Konya Province for providing data sets.