Effect of milk yield-based selection on some reproductive traits of Holstein Friesian cows on large-scale dairy farms in Malawi

M G G Chagunda*,***, E W Bruns, C B A Wollny* and H M King**

Institute of Animal Breeding and Genetics, University of Goettingen, Goettingen, Germany
*Department of Animal Science, Bunda College of Agriculture, University of Malawi, Lilongwe, Malawi
^**Animal Breeding and Husbandry in the Tropics and Subtropics, Georg-August University, Goettingen, Germany
***Current address: Department of Animal Health and Welfare, Danish Institute of Agricultural Sciences, Foulum, Denmark
Mizeck.Chagunda@agrsci.dk

Abstract

This paper describes a study aimed at evaluating the effect of selecting sires based on their breeding values for milk yield estimated in their countries of origin on reproductive performance of their daughters on large-scale dairy farms in Malawi. To do this an analysis was done using 2362 records of purebred Holstein Friesian cattle kept in Central and Southern Malawi from 1986 to 1996. Fixed effects of sire group, herd, year and season of calving were tested and random effects of dam and cow were included in the statistical model applied. (Co)variance components were estimated through the restricted maximum likelihood (REML) procedure.

Mean number of services per conception (NSC) was 1.50, gestation interval (GI) was 277days, calving interval between first and second parity (CI1) was 416 days, calving interval between second and third parity two and three (CI2) was 408 days, and age at first calving (AFC) was 32.5 months. In general the majority of the reproductive traits were affected by the non-genetic factors of herd, year and season. Sire group, a proxy of the breeding strategy, had a significant effect on AFC (p<0.01) and CI2 (p<0.05). Factors that had significant effects (p<0.01) on most of the reproductive traits in the study were herd, year of insemination, year of calving, and season of insemination. Phenotypic correlation between NSC and AFC was 0.19, between NSC and GI was -0.05, while that between NSC and CI2 was 0.14. Heritability estimates for NSC, GI, CI1, and AFC were, 0.04, 0.10, 0.001, and 0.20, respectively.

The high variation due to non-genetic factors and the low heritability estimates for the reproductive traits indicate that much improvement could be made through improved management, husbandry practices, and strategically utilizing environmental factors.

Key words: Holstein Friesian, Malawi, reproductive performance

Introduction

In Malawi, dairy production is performed on smallholder and large-scale dairy farms. The major differentiating features of these two dairy sub-sectors are the holding size, the genotype of cattle raised, and the level of management applied. Recent information (Malawi Government 1997) indicates that there are about 3600 smallholder farmers who use over 6000 Holstein Friesian x Malawi Zebu cows and about 1700 smallholder farmers who use an unknown number of Malawi Zebu cattle for commercial milk production in the peri-urban setting. Smallholder farmers supply about 60 % of the milk processed in Malawi each year (Banda 1996). There are about 15 private large-scale dairy farms accounting for about 2200 milking cows. The predominant genotype on the large-scale dairy farms is the Holstein Friesian although some of these farms also have few Ayrshire and Jersey cattle. Data for the current analysis were collected from three farms belonging to the then parastatal organization, Malawi Dairy Industries (MDI). Katete and Capital Hill dairy farms are situated in the central region of Malawi while Ndata Farm is situated in the Southern region of Malawi. These three dairy farms were established under the Malawi-Canada Dairy Development Project to provide Malawi with a foundation herd of Holstein Friesians. Apart from milk production, these farms were meant to supply heifers to smallholder farmers (Malawi Government 1997). Over time, MDI and the other large-scale dairy farms have also supplied bull calves that have been utilized in the national artificial insemination programme for smallholder dairy production (Chagunda et al 1998). This means that breeding strategies on large-scale dairy farms have carry-over effects on the smallholder dairy farms and impact on the national dairy development in general. Currently it is difficult if not impossible to carry out longitudinal evaluations on smallholder farms because of the unavailability of systematically kept records.

Reproductive traits in dairy cattle are not only a measure of fertility but also of productivity and lifetime production potential of an animal. Fertility may be defined as the ability to conceive and produce a viable calf following an appropriately timed insemination (Royal et al 2000). Fertility efficiency can be enhanced by means of better management but quite some variation in it has genetic origin (Biffani et al 2003). Low fertility is of economic importance for dairy enterprises, because it results in higher levels of involuntary replacement, slippage in calving pattern, veterinary intervention, hormonal treatment and reduced annual milk production (Esslemont and Peeler 1993). In Malawi, in an effort to improve milk production, for sometime dairy cattle production has been oriented towards increasing milk yield per animal. One element in this process has been importation of Holstein Friesian bull semen from the temperate regions whose main selection criteria has been milk yield. A concern about such a breeding strategy is the antagonistic relationships, which may exist between milk yield and reproductive traits, which when ignored may result in low overall productivity and below-expected response to selection in the aggregate genotype.

The objective of the current analysis was to investigate the effect of selecting sires based on their breeding values for milk yield estimated in their countries of origin on reproductive performance of their daughters on large-scale dairy farms in Malawi.

Materials and Methods

Climate

Malawi lies between latitude 9°S and 17°S and longitude 32° 42'E and 36° 36'E and has altitude that ranges from 52 to about 1632 m above sea level (in the arable region). During the period for which data was collected (1986 to 1996), the mean annual temperature in the study locations was 21°C. The warmest month was November, with an average maximum temperature of 29.5°C with the standard deviation (SD) of 1.9 and the coolest was July with the average maximum temperature of 22.8°C (SD 1.1). The highest average minimum temperature of 18.4°C (SD 0.9) was in December while the lowest average minimum temperature of 10.1°C (SD 1.8) was in July. The rainfall pattern is uni-modal, confined to the period from early November to April and peaks in January (Figure 1). Three distinct seasons occur: hot-wet (November - April), cool-dry (May -August), and hot-dry (August - November).

Figure 1. Average monthly temperatures (maximum, minimum), rainfall,
relative humidity (RH%), and wind speed for the study area from 1986 to 1997

Data

Performance records on production and reproduction from the three large-scale farms of Katete, Capital Hill and Ndata in Malawi were utilized. Breeding at these farms was done through artificial insemination using Holstein Friesian frozen semen imported mainly from Canada. The bull semen was selected based on breeding values for milk yield estimated in their countries of origin as presented in the sire catalogues. Initially the data set had 2362 records of lactation one to three from cows calving between 1986 and 1996. Cows with missing birth dates, breeding dates, calving dates, and cows with both parents missing were dropped from the analysis. After editing, 1968 records were available for analysis. The reproductive measures analysed were number of services per conception (NSC), gestation interval (GI), age at first calving (AFC), calving interval between lactation one and two (CI1), and calving interval between laction two and three (CI2). For the purposes of this study, breeding values were obtained from the Canadian Dairy Network website. The breeding values ranged from -2225 to +1788 kg and were from 130 sires.

Data analysis

Data were subjected to analysis of variance to determine the effect of sires grouped according to their Canadian breeding values for milk yield on the reproductive traits of their daughters in Malawi. The sires were grouped into three, that is, those with breeding values below -500 (group 1), those with breeding values between -500 and +500 (group 2), and those with breeding values above 500 (group 3). The following mixed model was applied using the GLM procedure of SAS (1989):

Y_ijklm = m + b₁(D)_ijklm + b₂(C)_ijklm + SG_i + P_j + H_k + (YR*SS)_l +e_ijklm

where:

            Y_ijklm = the response reproductive traits
            m = overall population mean
            b₁(D)_ijklm = random effect of dam as a covariate with a linear coefficient b₁
                b₂(C)_ijklm = random effect of cow as a covariate with a linear coefficient b₂
            SG_i = fixed effect of sire group (i =1,2,3)
            P_j = fixed effect of parity (j =1,2,3)
            H_k = fixed effect of herd (k = 1,2,3)
            (YR*SS)_l = fixed effect of season within year
            e_ijklm = random residual effects, e_ijkm ~N (0,s_e²)

The seasons of calving were defined as Hot-Wet (December to April), Cold-Dry (May to August) and Hot-Dry season (September to November). In the analysis the year and season were defined as year or season of birth, insemination, and calving depending on the reproductive trait being analysed.

(Co)variance components were estimated through the Restricted Maximum Likelihood (REML) procedure in an animal model using VCE 4.2 (Groeneveld 1998). Genetic correlation between some traits and also between milk yield for different lactations were obtained. The model equation used for the analysis was:

y = Xb + Zu + e

where:

            y = the vector of observations;
            b = the vector of fixed effects;
            u = the vector of random effects;
            e = the vector of random residual errors

X and Z are the incidence or design matrices for fixed and random effects with the assumption that:

E(y) = Xb; E(u) = 0

            Var (u) = G,
            Var (e) = R and
            Cov (u, e') = 0

The solutions for the fixed effects (b) and simultaneously for the random effects (u) were obtained by solving the mixed models equations (MME) (Henderson 1973). Heritability (h²) was determined as the ratio of additive genetic variance (s²_u) to the total variance (s²_p) i.e. h² = (s²_u / s²_p).

Results and Discussion

Descriptive statistics

In the records available for analysis (Table 1), only up to four inseminations were recorded. Malawi Dairy Industries (who were the owners of the farms) had a breeding policy that stipulated that if a cow failed to conceive after four artificial inseminations, she was put to a bull in the subsequent oestrus for natural mating. The average number of inseminations per conception (NSC) was 1.5 (SD = 0.82). Although the mean NSC was low with a range of 3, the within trait variation was quite high with a coefficient of variation of 54.5%. On average, the cows in the current study took 278 (SD = 23.8) days from conception to parturition. Gestation interval (GI) had the lowest variation of all the traits studied. The coefficient of variation for GI was 8.54%. Holstein Friesian cows performing on large-scale dairy farms calved for the first time at an average age of 32.0 (SD = 4.61) months. Age at first calving (AFC) was found to have moderate variation with the coefficient of variation of 14.4% (Table 1). Results from the study indicated that calving interval (CI) from first to second calving was 416 (SD = 96.9) days and that from second to third lactation was 408 (SD = 80.8) days. Calving interval had moderate variation with coefficient of variation of 23.3% for CI1 and 19.8% for CI2. There was more variation in CI1 than in CI2.

Factors affecting the reproductive traits

Results from the analysis of variance (Table 2) indicate that the environmental effects of herd, year, and season influenced the majority of the reproductive traits more than the animal effects of sire group, dam and cow. This reflects the important influence of management and nutrition on the reproductive traits.

a) Number of services per conception (NSC)

Of the factors tested, cow, parity, herd, year of insemination, year of calving, and season of insemination significantly influenced (p<0.001) NSC. Least square means for NSC by parity are presented in Table 3 while least square means by year are presented in Table 5.

First calf heifers had a significantly (p<0.05) lower NSC than cows in second and third lactation. Least square analysis indicated that cows inseminated for the first time during the hot-wet season (November - April) had a significantly higher NSC (p<0.05) (lsmean = 1.56, SE = 0.031) than those cows that were inseminated for the first time during the other seasons (lsmean = 1.46, SE 0.029 for cool-dry season, and 1.47, SE = 0.037 for hot-dry season). The hot-wet season, which is characterised by high humidity, high temperatures and low wind speed, obviously exerted some stress on the animals.

b) Gestation Interval

Factors that significantly affected gestation interval (GI) were year of calving (p<0.001), season of calving (p<0.001), dam (p<0.01), and year of insemination (p<0.01). Cows calving in the hot-wet season had a relatively shorter GI than those cows that calved down in the hot-dry season (275, SE. = 0.944 vs 279, SE 1.39 days).

c) Age at first calving

Age at first calving (AFC), which is a proxy for age at first service and age at sexual maturity, was significantly affected by cow effect (p<0.001), herd (p<0.001), year of birth (p<0.001), and sire group (p<0.01). One herd in the central region and the other in the southern region of Malawi had least square means for AFC which were not significantly different from each other (33.1 SE 0.45 vs 32.6 SE 0.45 months) but had a significantly higher (p<0.05) AFC than the third herd situated in central Malawi (lsmean = 31.5, SE 0.24 months). Least square means for the sire group effect are presented in Table 4 while those for year of birth are presented in Table 5.

Age at first calving increased as we moved from sire group 1 to sire group 2 and decreased again with the use of sire group 3. There was a curve-linear increase in AFC with increase in sire milk yield breeding values. AFC increased when sire milk yield breeding values increased from low to medium and decreased when there was further increase to high breeding values.

d) Calving interval

CI1 was significantly influenced by the herd (p<0.001) and year of calving (p<0.001) while CI2 was significantly affected (p<0.05) by sire group and calving season (p<0.05). CI2 increased from 407 (SE = 4.53) days for the sire group 1 to 414 (SE = 5.55) days and 446 (SE = 14.2) days for sire group 2 and 3, respectively. This apparent biological antagonistic relationship between production and reproduction traits indicate that selection based on milk yield breeding values may have negative consequences on reproductive traits.

The trends of some of the reproductive traits over the period studied are presented in Table 5.

Over the period of between 1985 and 1997, there have been significant differences (p<0.05) between different years. There do appear to be some trends in the variables considered. Number of services per conception appears to increase over time. There also appear to be slight increases in both calving intervals, which would be related to number of services per conception. Age at first calving does not appear to have a trend although it was much higher in the middle period than either of the extremes.

Relationship among the reproductive traits

In this study (Table 6), we found that NSC was strongly (p<0.01) and positively correlated to age at first calving (AFC) and calving interval between second and third lactation (CI2). Number of services per conception was negatively correlated to gestation length. Gestation interval was significantly and positively correlated to calving interval between first and second lactation (CI1) (p<0.05) and CI2 (p<0.01). The significant correlation coefficients of 0.09 and 0.12 between GI and CI1 and CI2 indicate the resultant increase in calving interval from the length of the gestation interval. Phenotypic correlation indicates a significant (p<0.05) and positive relationship between CI1 and GI. CI2 was found to have a strong (p<0.01) and positive relationship with NSC and GI.

Heritability estimates

Estimated additive genetic variances, residual variances, heritabilities and standard errors for some reproductive traits are presented in Table 7.

In this study, we found an additive genetic variance for AFC to be 3.03 and the residual variance of 12.4 giving a heritability value of 20%. The heritability for NSC was 4% while that for GI was 10%. Calving interval had the lowest heritability of 0.01%. Heritability estimates of the reproductive parameters were low ranging from 0.001 to 0.20. In general reproductive traits tend to have lower heritability values than production traits like milk yield although among the reproductive traits, some of the traits like age at first calving tend to have a relatively high variation with different workers reporting values as low as 0.04 to as high as 0.47 (Rege 1991; Makuza and McDaniel 1996; Baco et al 1998). The low heritability estimates for reproductive traits indicates that little genetic improvement would be expected from selection for such traits; however, improvement could be made through improved management, husbandry practices, and other environmental factors. Apart from the genetic effects discussed above, environmental effects of herd, year and season exerted some considerable influence on reproductive traits in Holstein Friesian cows on large-scale dairy farms in Malawi. These factors reflect the effect of management practices and other environmental conditions that are likely to vary from one herd to another (Van Bebber et al 1997). This has implications on the importance of improving management on smallholder farms if the benefit of utilizing germ-plasm from large-scale dairy farms is to be seen.

Conclusions

• The results indicate that the physical, management and nutrition environment influence reproductive traits of a Holstein Friesian herd performing in a tropical climate of Malawi more than additive genetic factors.
  
  
• Management in smallholder farms, which acquire heifers and semen from bulls culminating from the large-scale farms, needs to be improved in order to benefit from the national dairy programme.
  
  
• Performance recording should become an integral part in both the large-scale and the smallholder dairy sector to enable continued monitoring and evaluation of the dairy sector.
  
  
• In ancillary studies, other factors such as heat stress, physiological balance and health should be investigated and quantified.

Acknowledgements

The financial support provided by GTZ/DAAD and the cooperation by Malawi Dairy Industries Corporation (MDI) Ltd. in allowing us access to data for this study are gratefully appreciated

References

Baco S, Harada H and Fukuhara R 1998 Genetic relationship of body measurements, at registration to a couple of reproductive traits in Japanese black cows. Animal Science and Technology 69 (1): 1 -7

 

Banda J W 1996 A General Survey of the Dairy and Meat sector in Malawi. Consultancy Report. Bunda College of Agriculture, Lilongwe, Malawi

 

Biffani S, Samore A B, Canavesi F and Marusi M 2003 Data quality assessment and preliminary investigations on fertility in the Italian Holstein Friesian. Interbull Bulletin No. 30. Proceedings of the Interbull Technical Workshop. Beltsville, MD, USA. March 2-3, 2003. pp 89-95
 

Esslemont R J and Peeler E J 1993 The scope for raising margins in dairy herds by improving fertility and health. British Veterinary Journal 149: 537-547

 

Groeneveld E 1998 VCE user manual. ftp:/ ftp.tzv.fal.de/pub/vce4.2

 

Henderson C R 1973 Sire evaluation and genetic trends. In: Proceedings of Animal Breeding and Genetics Symposium in Honor of Dr. J. L. Lush, ASAS / ADSA, Champaign, Illinois, 10-41

 

Makuza S M and McDaniel B T 1996 Effect of days dry, previous days open, and current days open on milk yields of cows in Zimbabwe and North Carolina. Journal of Dairy Science 79:702 - 709

 

Malawi Government 1997 Statement of Development policies 1987-96. Malawi Government Printer, Zomba

 

Rege J E O 1991 Genetic analysis of reproductive performance of Friesian cattle in Kenya. 1 Genetic and phenotypic parameters. Journal of Animal Breeding and Genetics 108:6, 412-423

 

Royal M D, Darwash A O, Flint A P F, Webb R, Woolliams J A and Lamming G E 2000 Declining fertility in dairy cattle: changes in traditional and endocrine parameters of fertility. Animal Science, 70: 487-501

 

SAS Institute Inc. 1989 SAS/STAT User's Guide, version 6, Fourth Edition, Volume 2, Cary, NC, USA

 

Van Bebber J, Reinsch N, Junge W and Kalm E 1997 Accounting for herd, year, and season effects in genetic evaluations of dairy cattle: a review. Livestock Production Science 51: 191 - 203