Genetic evaluation of grade, appendix and pedigree cow classes in Holstein, Jersey and crossbred dairy breeds in Zimbabwe

S.M. Makuza, V. Muchenje and S. Chiyanike
Department of Animal Science, University of Zimbabwe
P.O. Box MP 167 Mount Pleasant, Harare, Zimbabwe

Abstract

Introduction

Objectives

Materials and methods

Source of data

Data editing

The animal model

Results and discussions

Conclusions

Recommendation

Acknowledgements

References

Abstract

Genetic parameters of milk yield of pedigree, appendix and grade Holstein, Jersey and crossbred cows were estimated using the Average Information Restricted Maximum Likelihood (AIREML) animal model. The objectives were to determine the best cow class, in terms of milk production, and determine whether the cow classes are different in the three major dairy breeds and whether there are any genetic differences among the three cow classes namely pedigree, appendix and grade.

The records were obtained from the Zimbabwe Dairy Services Association (ZDSA) for total lactation from 1979 to 1994. The fixed effects in the animal models were herd-year, month of calving, age at calving and parity. Random effects were residual error, permanent environmental effects of the cow and the individual animal or cow effects. The cow classes were analysed separately within each breed group because of limitations of computer memory.

Appendix cows had the highest milk yields, followed by pedigree cows, with grade cows producing the least. Heritabilities (h²) for milk yield in Holsteins were 0.23, 0.24 and 0.23 for pedigree, appendix and grade cows, respectively. Similar values for Jerseys pedigree, appendix and grade cow classes were 0.36, 0.36 and 0.25, respectively. Heritabilities for milk yield in crossbred cows were 0.28 and 0.33 for appendix and grade cows, respectively. Repeatabilities (r_p) were similar and averaged 0.42 for all breeds and all cow classes. Major differences among the cow classes across the breeds were observed in the additive, residual and phenotypic variances.

The variances observed suggest adequate genetic variation in the Zimbabwean dairy breeds cow classes for an effective direct selection programme. This genetically improves milk yield and its positively correlated traits.

Introduction

Zimbabwean dairy cattle can be divided into three classes, which are pedigree, appendix and grade. Registered classes include pedigree and appendix cows for which the Zimbabwe Herd Book (ZHB) will issue a certificate and confirms the breed stated on the certificate. A grade daughter mated to a registered sire will produce appendix A offspring. By continued use of registered sires, offspring of succeeding generations go from appendices A to B to C and finally to pedigree, thus there is an open herdbook on registered Holsteins, Jerseys, crossbred cows and other dairy breeds. Grades are those not included in the herdbook. The registration by ZHB is done under the appendix scheme, which is an upgrading process where non-registered or grade animals can achieve full pedigree status in three generations. However, it should be emphasised that this registration is not based on performance but on identification of pedigree and parentage. It has been reported that the pedigree has the highest milk yields, followed by the appendix and the grade cow with the least milk yields (Makuza 1988; Trigg 1989). However, Muchenje (1996) found that the appendix cows had the highest milk yields, followed by the pedigree then the grade cow class. Makuza (1995) reported heritabilities and repeatabilities for milk yield of 0.19 and 0.45 in Holsteins, 0.27 and 0.45 in Jerseys, and 0.26 and 0.40 in crossbred cows, respectively. He also reported additive standard deviations for milk yield of 488, 571 and 658 kg in Holsteins, Jerseys and crossbred cows, respectively. The corresponding phenotypic standard deviations were 1120, 1104 and 1288 kg, respectively. Reported repeatabilities for milk yield in the literature range from 0.35 to 0.60 (Mao 1984). Similarly the range of heritabilities for milk production is from 0.15 to 0.40 (Mao 1984). Banga (1992) reported a heritability estimate of 0.54+ 0.16 for milk yield in Zimbabwean Jersey cows. This is a little bit too high possibly due to the method of estimation which was Harvey's least squares mixed models. The estimates vary due to different measurements, samples, populations, models or estimation procedures. Literature genetic parameter estimates of different breeds also differ due to differences in population structures among breeds. Small heritabilities (less than 0.10) mean that regardless of genetic evaluations and selection methods used, the genetic gain would be relatively small.

Due to the contrasting findings by the different Zimbabwean researchers on the performance of breed cow classes, there was a need to evaluate the genetic and phenotypic performance of Holstein, Jersey and crossbred cow classes. The Holstein, Jersey and crossbred cows are the three major dairy breed groups in Zimbabwe (Makuza 1995). The hypothesis tested in this study was that the pedigree class produces the highest milk yield in all the three dairy breeds of Zimbabwe.

Objectives

The objectives of the study were to:

• establish the best class of cow in three Zimbabwean dairy breeds
• determine whether there are genetic differences (heritabilities, repeatabilities and standard deviations) among cow classes of Holsteins, Jerseys and crossbred cows
• compare genetic estimates from pedigree, appendix and grade daughters across breeds.

Materials and methods

Source of data

Total lactation milk yield records for multiple lactations were available from 1979 to 1994 calvings of Holstein, Jersey and crossbred commercial dairy cattle (Table 3). The records were provided by the Zimbabwe Dairy Services Association (ZDSA). The field records consisted of parentage, production and reproduction data.

Data editing

To remove outliers and incorrect data, records with age at calving outside the range of 16 to 200 months were deleted. Days in milk considered ranged between 250 and 300 days with a calving interval between 300 and 600 days. Lactations 1 to 12 were evaluated. Lactations greater than 4 were combined with lactation 4 into one subclass. Milk yield was restricted between 1000 to 15000 kg per lactation. Records with cow classes incorrectly recorded apart from 1 (pedigree), 2 (grade) or 3(appendix) were deleted. The data also had to be consistent in birth dates, calving dates, year of calving and region of the country. The records were also deleted for missing sire and dam identification or duplication or misidentification of parentage. Each sire was supposed to have at least 2 daughters for AIREML to run.

The Animal model

The animal model used in the analyses was as follows:

Y_ijklmno = : + HY_i + M_j + P_k + Age_l

+ Animal_n + PE_n(m) + E_ijklmno

where:

Y_ijklmno = response variable milk yield;

: = overall mean common to all observations;

HY_i = i^th combined effect of herd and year of calving on milk yield;

M_j = j^th effect of month of calving on milk yield;

P_k = k^th effect of parity on milk yield;

Age_l = l^th effect of age at calving on milk yield;

Animal_n = n^th individual random animal effect on milk yield;

PE_n(m) = n^th random permanent environmental effect of the m^th animal on milk yield;

E_ijklmno = random residual error.

The inverse of the matrix of additive genetic relationships was added to improve linkage between sires and improve the accuracy and precision of the estimates. The average information restricted maximum likelihood (AIREML) FORTRAN algorithm of Gilmour (1995) was used for analyses. Sparse matrix techniques were employed to calculate those elements of the inverse of the coefficient matrix required for the first derivatives of the likelihood. Residuals and fitted values for random effects can be used to derive additional right-hand sides for which the mixed model equations can be repeatedly solved in turn to yield an average of the observed and expected second derivatives of the likelihood function.

This Newton method, using average information, generally converges in less than 10 iterations. The Restricted Maximum Likelihood algorithm (REML), using average information, is about five times faster than a derivative-free algorithm, using the simplex method, which in turn is about three times faster than an expectation - maximisation algorithm (Johnson and Thompson 1995). The REML animal model accounts for losses in degrees of freedom of fitting fixed effects and selection bias. The cow classes were analysed separately within each breed. Each dairy breed was therefore also analysed separately. Grade and appendix Jersey cows and grade crossbred cows were analysed as single records. We could not run multiple records due to the small sample sizes resulting in a non full rank matrix meaning that it had dependencies in the factors fitted in the models. This could have resulted in convergence reaching a local rather than a global maximum.

The values reported therefore correspond to a generalised inverse approach and are therefore less reliable as they are not unique solutions to the mixed model equations.

Results and discussions

In the preliminary analyses with SAS's PROC GLM (SAS 1994) for the three cow classes and in all dairy breed groups, fixed effects of herd-year, month of calving, parity and age at calving affected milk yields (P<0.001). These results are in general agreement with those reported in the literature (Makuza 1995; Muchenje 1996; Chiyanike 1997).

Table 1 shows the percentages of cow classes by breed group. The majority of cow classes are grade, which comprise 65% in Holsteins, 45% in Jerseys and 81% in crossbred cows. The proportion of pedigree cows, which are close to 100% pure-bred exotic blood, is small. They comprise 11% and 22% in Holstein and Jerseys respectively. As is expected there are no pedigree Crossbred cows.

Table 1. Percentages of cow classes by breed group.

The percentages of cow classes by breed group and agroecological region are given in Table 2. For Mashonaland, the distribution of cow classes by breed is similar to that of Table 1. However, for Midlands and Manicaland, the percentage of grade cows by breed group is higher than those in Table 1. This indicates that these two regions have the majority of unregistered cows in Zimbabwe. Almost 71% of Jersey cows in Matabeleland are pedigree cows. As this is the driest region of the country, this finding is rather surprising and unexpected. Again, as was expected, there are no pedigree crossbred cows in all the four milk recording scheme regions of Zimbabwe. Herds with some of each type of cow may have represented management systems that were attempting to increase the proportion of registered cows i.e. pedigree and appendix. Grade cows in these herds would have faced lower probabilities of survival than appendix or pedigree cows or even comparable cows in all grade herds of Holstein, Jersey and Crossbred cow classes. This could be true for Mashonaland with the majority of cows for the three breed groups and shows a deviation of proportions from those of other regions (Table 2).

Table 2. Percentages of cow classes by breed and agroecological region.

Table 3 shows the raw means and standard deviations for various traits of the Holstein, Jersey and Crossbred cow classes. Milk yield, lactation number, days dry and calving intervals in all the three breed groups were highest in the appendix cows which were also the oldest with an average age at calving of 59 months compared to 54 months for grade and 50 months for pedigree. Fat percentages tended to be slightly higher in grade cows for all the three breed groups. Days in milk were similar for all cow classes in all the breeds. Grade and pedigree cows left herds at younger ages. Grade and pedigree cows averaged 2.7 lactations across breeds versus 3.0 lactations for appendix cows.

Table 3. Raw means and SD for various traits of Holstein, Jersey and Crossbred cow classes.

Heritabilities and repeatabilities by class of cow and breed are given in Table 4. Generally, heritabilities and repeatabilities were higher in Crossbreds followed by Jerseys. Higher results were previously reported in Zimbabwean Jersey cows (Banga 1992). The genetic parameters (h² and r_p) were slightly higher in appendix cows but those for grade and pedigree cows were similar. As has been explained in the materials and methods section, heritabilities for Jersey pedigree and appendix cows and Crossbred appendix cows had some permanent environmental components in them because they were run as single records. The programme could not separate permanent environmental effects from additive effects. The fact that heritabilities and repeatabilities are higher in Jerseys and Crossbreds than in Holsteins could also be due to sampling variation due to the small sample sizes compared to Holstein cow classes. There could also be differences in the breed structures of these breeds. However, generally these estimates are in agreement to the findings of Makuza and McDaniel (1995), working with the same data. The fact that Crossbreds had higher heritabilities is expected because they are a mixture of different breeds, some indigenous, resulting in more additive, dominance and epistasis (genetic) variation compared to the pure-breds.

Table 4. Heritabilities, repeatabilities for milk yield in cow classes of three major breeds of Zimbabwean dairy cows

¹Run as single records and used a generalised inverse approach due to a non full rank matrix.

Standard deviations by cow class and by breed are in Table 5. In general, standard deviations (Ф_a, Ф_pe, Ф_e and Ф_p) were highest in Holsteins followed by Crossbreds, with Jerseys having the lowest. The same trend was reported by Makuza and McDaniel(1995).

Table 5. Additive _a permanent environmental (_pe ) standard deviations (kg) for milk yield in cow classes of three major Zimbabwean dairy cows.

¹ Run as single records therefore could not separate permanent environmental effects from additive effects.

Appendix cows had higher standard deviations across breed groups followed by pedigree cows. Grade cows had the lowest standard deviations.

These findings suggest that in a large population of any breed, including both registered animals and grades resulting from several generations of use of pure-bred sires, there will inevitably be some animals in the lower strata of the breed structure which are genetically superior to some in elite herds.

Even if selection programmes in elite herds are highly effective and directed to goals of commercial importance, this will occur due to genetic recombination or heterotic effects and the fact that phenotypes are imperfect indicators of genotypes.

The fact that the data analysed covered the whole country of Zimbabwe, might indicate that appendix cows have the right breed composition for general adaptation to the Zimbabwean environment. Appendix cows are referred to as appendix for at least three generations under the grading up appendix scheme. Usually the first crossbred progeny shows maximum heterosis and this will envisage maximum milk production by appendix A cows.

Our study has indicated basic genetic and phenotypic differences in field data for grade, appendix and pedigree milk production.

Conclusions

• Appendix cows outperformed pedigree and grade cows in milk production in all three breed groups.
• Genetic parameters (h², r_p and variances) were highest in appendix cows in Holsteins, Jerseys and Crossbreds. Pedigree and grade cows had similar genetic parameters in the three breed groups.
• Genetic parameters (h², r_p and variances) were highest in Crossbreds followed by Jerseys.
• There is adequate genetic variation in Zimbabwean dairy breed cow classes for an effective direct selection programme for milk yield.
• Milk yield in Zimbabwe is as heritable and repeatable as in the temperate and tropical regions of the world.

Recommendation

There is a need to determine what genotype proportions of Pedigree and Grade cows achieve the maximum milk production in Zimbabwe.

Acknowledgements

We would like to thank the Zimbabwe Dairy Services Association (ZDSA) for providing the records as input to this study. We also send our gratitude to Mrs V. Choruma for typing this manuscript.

References

Banga C. 1992. Genetic parameters for milk production traits in Jersey cattle. Zimbabwe Journal of Agricultural Research 30:45–48.

Chiyanike S. 1997. Genetic evaluation of Grade, Appendix and Pedigree Holstein cows in Zimbabwe. BSc honours thesis, Department of Animal Science, University of Zimbabwe.

Gilmour A. 1995. Average Information REML Manual. NSW Agriculture, Australia

Johnson D.L. and Thompson R .1995. Restricted maximum likelihood estimation of variance components for univariate animal models using sparse matrix techniques and average information. Journal of Dairy Science 78:449–456.

Makuza S.M. 1988. Influence of imported sires on production of dairy herds in Zimbabwe. MSc thesis, Michigan State University, East Lansing, Michigan, USA.

Makuza S.M. 1995. Studies on the genetics of dairy cattle in Zimbabwe and North Carolina. PhD dissertation, North Carolina State University, Raleigh, North Carolina, USA

Makuza S.M. and McDaniel B.T. 1995. Genetic and phenotypic parameters for production traits in Zimbabwean dairy breeds. Journal of the Zimbabwe Society for Animal Production 7:151–156.

Mao I.L. 1984. Variation in dairy cattle population: causes and consequences. Proceedings of the National Invitational Workshop on Genetic Improvement of Dairy Cattle. Milwaukee, Wisconsin, USA pp.25–43.

Muchenje V. 1996. Development of age-month adjustment factors for lactation milk yield of Zimbabwean Holstein dairy cattle. MSc thesis, Department of Animal Science, University of Zimbabwe.

SAS (Statistical Analysis System). User's Guide; Statistics, Version 6.10 Edition. 1994. SAS Inst., Inc., Cary, NC.

Trigg C. 1989. Holstein, sire and cow evaluation for milk production traits in Zimbabwe. MPhil dissertation, Department of Animal Science, University of Zimbabwe. Zimbabwe.