| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
* Department of Animal Science, Agricultural University of Norway,
GENO Breeding and A. I. Association, P. O. Box 5025, N-1432 Ås, Norway
Corresponding author: B. Heringstad; e-mail: bjorg.heringstad{at}ihf.nlh.no.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
Key Words: clinical mastitis • protein yield • selection experiment • selection response
Abbreviation key: CM = clinical mastitis, HMP = high milk production group, HPY = high protein yield group, LCM = low clinical mastitis group, LMP = low milk production group, NRF = Norwegian Dairy Cattle
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
In both experiments, sires were selected based on their progeny testing results in the Norwegian Dairy Cattle (NRF) population. For low milk production, a group of 21 sires was initially selected based on their progeny test results in 1978 and 1979; and these bulls were used exclusively during the entire experiment. Otherwise, for each selection group and year, the 2 to 7 highest ranking sires for the respective trait were selected from the most recent group of progeny tested bulls. No consideration was made with respect to EBV for other traits. This approach implies single-trait selection of sires, which were preselected for the NRF breeding objective. In NRF, 120 to 130 bulls are progeny tested each year with an average of 250 to 300 daughters per sire. Norwegian dairy cattle have been selected for a broad breeding objective, with steadily increasing emphasis on functional traits like health and fertility over time (Heringstad et al., 2001). The current breeding objective results in genetic improvement of both protein yield and mastitis resistance (Heringstad et al., 2001, 2003), despite the unfavorable genetic correlation between these traits. Heringstad et al. (2000) summarized the estimates of genetic correlation between clinical mastitis and milk production based on Nordic field data and found estimates ranging from 0.24 to 0.55, with an average of 0.43. Recent estimates of the genetic correlation between clinical mastitis and milk production range from 0.25 to 0.45 (Heringstad et al., 1999; Lund et al., 1999; Rupp and Boichard, 1999; Hansen et al., 2002).
Due to the unfavorable genetic correlation between mastitis and milk production, selection for increased milk production is expected to result in a genetic deterioration of mastitis if mastitis is ignored in the breeding program. This response has been confirmed in simulation studies (Strandberg and Shook, 1989; Colleau and le Bihan-Duval, 1995) but never demonstrated with real data, because populations that ignore mastitis in the breeding program generally do not record the trait.
The main objective of this paper was to estimate selection responses for clinical mastitis and protein yield in the two previously described selection experiments. Results from these selection experiments may provide estimation of both the correlated response in clinical mastitis resulting from single trait selection for increased milk production, as well as responses to direct selection against clinical mastitis.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
Selection Experiments
Selection experiment I.
The first selection experiment was carried out in the period from 1978 to 1989. Eight herds were included in the experiment (Table 1
). Two selection groups were established based on NRF cows present in the herds when the experiment began. Cows were randomly assigned to either a HMP or a LMP group. Each year, the three to seven highest ranking proven sires for milk production were selected from the most recent group of progeny tested NRF sires and mated to the HMP cows (total 49 sires). Sires were used only 1 yr, so very close inbreeding (e.g., father-daughter mating) was therefore avoided. For the LMP group, 21 sires with milk production indices approximately equal to 90 from progeny testing in 1978 and 1979 were selected and used as sires during the entire experimental period. Sires in LMP were used in a rotational system (six to nine sires were used per year) to avoid inbreeding. The milk production index was a relative index with mean 100 and SD 6. Sires were selected based on EBV for 305-d lactation milk production in first lactation, which had been calculated somewhat differently over time (Ødegård, 2000). Until 1978 the selected trait was kg FCM (kg milk [0.4 + 0.15 x fat%]). In the period 1979 to 1985, the selection criteria was kg milk, fat percentage, and protein percentage, with twice as much weight on protein percentage as on fat percentage (kg milk [0.22 + 0.075 x fat% + 0.15 x protein%]). From 1986 to 1989, the criteria was kg milk with five times as much weight on protein percentage as on fat percentage (kg milk [0.20 + 0.04 x fat% + 0.20 x protein%]). In 1989, the selection criterion was changed to 305-d protein yield (kg protein). Directional selection of cows was not practiced in either group. At the end of this experiment, in 1989, most calves and heifers in the LMP group were culled, while animals from HMP were kept in the herds and produced first-lactation records. Therefore, five generations of HMP cows and four generations of LMP cows were included in the analyses.
|
Selection experiment II.
In 1989 another selection experiment was initiated, based mainly on the HMP cows from the first experiment. Eight herds were included: six of the herds from the first experiment and two new school herds (Table 2). Cows were randomly assigned to either a high protein yield (HPY) or a low clinical mastitis (LCM) group. Each year the 3 to 4 highest ranking proven sires for 305-d protein yield (one selection criterion) were selected from the most recent group of NRF test-bulls and mated to HPY cows. Correspondingly, each year the two to four best proven sires for mastitis resistance (the other selection criterion) were selected and mated to LCM cows. This mating design, corresponding to that of HMP in experiment I, efficiently avoids close inbreeding. As in experiment I, there was no directional selection of cows in either group. Approximately 70% of the cows in generation 0 were HMP cows from experiment I, 8% were LMP cows, and 22% were other NRF cows (mainly cows in the two new herds). In cow generation 1, 75% of LCM cows and 64% of the HPY cows had mothers from HMP. Three generations of cows had completed first lactation at the time of these analyses. This experiment is ongoing.
|
). Data on 305-d protein yield in first lactation were extracted using the same criteria for calving period, age of first calving, and start of lactation as for CM. Only cows with complete 305-d lactation records were considered. Mean and SD of 305-d protein yield was 170.1 and 34.9, respectively (Table 3). The resulting file contained information on 305-d protein yield for 1,576,579 cows (Table 3). The structure of the data sets is shown in Table 3.
|
). The average mastitis frequencies were 18.3% for HMP cows and 12.8% for LMP cows (Table 4). A total of 1815 cows, 1098 from the HMP group, and 717 from the LMP group, completed 305-d of first lactation (Table 4). The average 305-d protein yields were 185.1 kg and 165.1 kg for HMP and LMP cows, respectively (Table 4). A total of 2722 cows had EBV for CM, and the numbers per herd, selection group and cow-generation are given in Table 1
.
|
). A total of 1117 cows, 610 from the HPY group and 507 from the LCM group, completed 305-d of first lactation, with a mean 305-d protein yield of 185.9 kg for HPY cows and 178.5 kg for LCM cows (Table 4). A total of 1590 cows had EBV for CM, distributed over selection groups, cow-generations, and herds as shown in Table 2.
Pedigree File
The initial pedigree file for the NRF population included 3.7 million records. From this file, separate pedigree files for the cows in the CM and protein yield datasets were derived by tracing the pedigrees for cows with data back as far as possible. The resulting pedigree files had 2,316,164 (CM) and 1,897,747 (protein yield) observations, respectively (Table 3). A total of 3.7% of the animals in the pedigree-files were base animals.
Model
Breeding values for CM were calculated using the following linear model:
 |
where
| Yijkl | = | observation of CM (0 = healthy, 1 = diseased) for cow l, calving at age i, in month j, and herd x year class k;
| Ai | = | fixed effect of age at calving i in 15 classes, where <20 mo is the first class, >32 mo the last class, and the other classes are in single months;
| Mj | = | fixed effect of month of calving j in 12 classes;
| HYk | = | fixed effect of herd-year class k;
| al | = | random effect of animal l; and
| eijkl | = | random error term.
|
Protein yield was analyzed using the model:
 |
where
| Yijklm | = | observation of 305-d protein yield (kg protein) for cow m, calving at age i, in month j, days open class k and herd x year class l;
| DOk | = | fixed effect of days open k in 14 classes, where <30 d is the first class, >149 d is the last class, and the other classes are in 10 d intervals; and
|
other effects were as defined previously for CM.
The models included the same fixed effects as the linear sire models used by Heringstad et al. (1999; 2001). Univariate analyses were chosen because a previous study (Heringstad et al., 2001) showed that estimates of genetic trends in this population were the same from univariate analyses and bivariate analysis of CM and protein yield.
Variance components for CM and 305-d protein yield were estimated with VCE4 (Neumaier and Groeneveld, 1998) using linear sire models with the same fixed effects as described above and data restricted to daughters of sires progeny tested in the years from 1978 to 1998. The resulting sire (
) and residual (
) variances were used to calculate variance components relevant for the animal model as follows:
 |
Additive relationship matrices were included in the analyses, and breeding values were estimated using the program PEST (Groeneveld and Kovac, 1990).
Genetic trends were estimated by linear regression of EBV on cow generation and cows birth-year, and t-tests were used to test whether the regression coefficients were different from zero and to assess whether mean EBV were significantly different between selection groups for each cow generation.
|
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
shows mean mastitis frequency per cow generation in the two selection experiments. Cows in the LMP group show a relatively flat phenotypic trend, whereas mastitis frequency in the HMP group increased with cow generation. In selection experiment II, mastitis frequency decreased in the LCM group and increased somewhat for the HPY group. Phenotypic mastitis frequency in cow generation 0 was higher in the LCM group than in the HPY group.
|
) and residual (
) effects were 0.0038 and 0.1198 for CM, and 100.8 and 427.2 for 305-d protein yield, corresponding to heritabilities of 0.03 and 0.19, respectively. The overall mean (SD) EBV for CM and 305-d protein yield in the total dataset were 0.002 (0.029) and 3.07 (6.28), respectively. Annual mean EBV for CM and 305-d protein yield for sires used in the two experiments are given in Tables 5 and 6.
|
|
. The genetic trend for LMP cows was flat, while mean EBV for CM in the HMP group increased with cow generation. Mean EBV was significantly different (P < 0.0001) between the two selection groups from cow generation 1 onwards. The genetic difference in CM between HMP and LMP cows in experiment I, equivalent to 3.1% CM after four cow-generations, is most likely a correlated response in CM after selection for increased milk production in the HMP group. In selection experiment II, the genetic level of CM in cow generation 0 was the same for the two groups (Figure 2). The genetic trend of CM was approximately flat for HPY cows and strongly decreasing in the LCM group. Mean EBV was significantly different (P < 0.0001) in HPY and LCM from cow generation 1 onwards. Here, the genetic difference of 8.6% CM after three cow generations is mainly a result of direct selection against CM in the LCM group.
|
). A genetic improvement equivalent to a reduction of 3.03% CM per cow generation was estimated for the LCM group, and an approximately flat genetic trend was found for the LMP group (Table 7). The regression coefficients for HMP and LCM were significantly different from zero (P < 0.0001), whereas the levels of significance were 0.058 and 0.173 for HPY and LMP, respectively (Table 7).
|
. Genetic trend for CM in the NRF population was approximately flat for cows born before 1990 and slightly decreasing thereafter. The corresponding regression coefficients indicate a genetic change of +0.04% CM per year from 1976 to 1990, and -0.19% CM per year from 1990 onwards (Table 7). This agreed closely with Heringstad et al. (2003) who found a similar result using linear and threshold sire models. However, the estimate of annual genetic change after 1990 was slightly lower from the animal model (-0.19% CM) than the estimate of Heringstad et al. (2003) from a linear sire model (-0.23% CM). The genetic trend may have been somewhat over estimated with the sire model. Genetic improvement implies that the mean genetic level of the mothers of the daughters used to predict sire EBV increases over time, while the sire model assumes that sires are mated to cows at random, i.e., to cows with a constant genetic level over time. On the other hand, Heringstad et al. (2003) compared genetic change estimates from linear and a threshold sire model, and concluded that given that the threshold model holds true, the magnitude of genetic change for CM was slightly understated by the linear sire model in these data. The annual genetic response may therefore also have been underestimated with the linear animal model used in the present study.
|
and 3). In these experiments, sires were from the NRF population, which has changed genetically over time (Figure 3). The difference in genetic trends of CM between the HMP and HPY groups, both selected for increased milk production, can probably be explained by this preselection. The genetic trend for CM in the HPY group is thus not representative for what would happen in a population selected only for increased production. Rather, the difference between genetic trends for CM in the NRF population and in the HPY group describe the indirect selection response in CM resulting from selection for increased milk production.
In the time period when the genetic trend of CM in the NRF population was approximately flat, a genetic increase equivalent to 0.23% CM per year was found for HMP cows, while in the period where the NRF population showed a genetic improvement of -0.19% CM per year an approximately flat genetic trend (+0.04% CM per year) was found for HPY cows (Figure 3 and Table 7). The difference between genetic trend for CM in the NRF population and in the groups selected for increased milk yield, HMP and HPY, respectively, indicates that an increase between 0.19 and 0.23% CM per year may be expected as a correlated response to selection for increased milk production in NRF, given that similar selection intensities are used.
Our study is the first to quantify the correlated genetic change in CM resulting from selection on increased milk production. The results, however, agree with results from several other designed experiments reporting increased health costs as a correlated response to selection for increased milk yield (Hansen et al., 1979; Bertrand et al., 1985; Short et al., 1990; Dunklee et al., 1994; Jones et al., 1994). From a selection herd in Minnesota, Hansen et al. (1979) reported increased labor and health care expenses for cows in a group selected for high milk yield compared with cows in the control group. In the same herd Jones et al. (1994) compared health care costs for cows born from 1975 to 1990 and found that cows in the selection group had higher health expenses as a correlated response to increased milk yield than did unselected controls, and that expenses due to mastitis accounted for most of the difference between groups. Dunklee et al. (1994) evaluated a selection experiment with Holstein cows that included two groups with high and average milk yield, respectively. They reported that the high yielding cows had higher total health costs and higher mammary health costs than did the cows in the average yield group. The mean number of cases of mastitis per lactation was 0.46 in the high yield group and 0.29 in the average yield group. Bertrand et al. (1985) reported that daughters of high Predicted Difference Milk sires produced 16% more milk, had 26% more mammary costs, and 42% more discarded milk costs, than daughters of breed average sires. Also significant differences in health costs found between a line of Jersey cows selected for milk yield and a control line were mainly due to higher mammary health costs (Short et al., 1990).
Mean EBV for 305-d protein yield per cow generation in the two selection experiments are given in Figure 4. The difference between selection groups was significantly different (P < 0.0001) after cow generation 0 in both experiments. The HMP and HPY groups showed as expected a positive genetic trend for protein yield, with a genetic gain of 3.23 and 4.85 kg protein per cow generation (Table 7). In the LMP and LCM groups the genetic trends were negative, corresponding to a genetic change of -2.83 and -1.97 kg protein per cow generation (Table 7). All these regressions of EBV for 305-d protein yield on cow generation were significantly different from zero (P < 0.0001) (Table 7). In Figure 5, mean EBV for 305-d protein yield are given by birth-year for cows in the two selection experiments and also for the NRF population. The figure shows a positive genetic trend for protein yield in the NRF population, as well as for the HMP and HPY groups. The trend for the LCM group was negative, but varied as expected between years, since mean EBV for 305-d protein yield of the sires used in LCM varied strongly between years (Table 6). The negative genetic trend for protein yield in the LCM group is most likely a correlated response after selection against CM, due to the unfavorable genetic correlation between the two traits.
|
|
), expected to be approximately one-half the genetic trend for AI sires, agree with Heringstad et al. (2001). They reported an annual genetic improvement of 0.96 kg protein per year for NRF sires born between 1974 and 1990, based on a linear regression analysis of sire EBV for 305-d protein yield on sires birth-year. The annual genetic trend in the groups selected for increased milk production, HMP and HPY, was lower than the genetic trends for milk yield of US Holsteins, which according to a review by Hansen (2000) were 79, 102, and 116 kg/yr during the 1970s, 1980s, and in the period from 1990 to 1996, respectively (corresponding to between 2.5 and 3.7 kg protein per year). Selection in the HMP and HPY groups were not strict single trait selection. Although sires were selected solely based on their breeding values for milk production, they were selected among progeny-tested bulls and therefore had been preselected for NRF breeding objective with increasing weight on health and fertility over the last 20 yr. |
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
|
ACKNOWLEDGEMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
Received for publication October 4, 2002. Accepted for publication April 22, 2003.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  B. Heringstad, E. Sehested, and T. Steine Short Communication: Correlated Selection Responses in Somatic Cell Count from Selection Against Clinical Mastitis J Dairy Sci, November 1, 2008; 91(11): 4437 - 4439. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. Heringstad, G. Klemetsdal, and T. Steine Selection Responses for Disease Resistance in Two Selection Experiments with Norwegian Red Cows J Dairy Sci, May 1, 2007; 90(5): 2419 - 2426. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Wolfova, J. Wolf, J. Kvapilik, and J. Kica Selection for Profit in Cattle: I. Economic Weights for Purebred Dairy Cattle in the Czech Republic J Dairy Sci, May 1, 2007; 90(5): 2442 - 2455. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. Moroni, C. S. Rossi, G. Pisoni, V. Bronzo, B. Castiglioni, and P. J. Boettcher Relationships between somatic cell count and intramammary infection in buffaloes. J Dairy Sci, March 1, 2006; 89(3): 998 - 1003. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. Moroni, G. Pisoni, M. Antonini, G. Ruffo, S. Carli, G. Varisco, and P. Boettcher Subclinical Mastitis and Antimicrobial Susceptibility of Staphylococcus caprae and Staphylococcus epidermidis Isolated from Two Italian Goat Herds J Dairy Sci, May 1, 2005; 88(5): 1694 - 1704. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. Heringstad, Y. M. Chang, D. Gianola, and G. Klemetsdal Genetic Association Between Susceptibility to Clinical Mastitis and Protein Yield in Norwegian Dairy Cattle J Dairy Sci, April 1, 2005; 88(4): 1509 - 1514. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
I. M. Andersen-Ranberg, G. Klemetsdal, B. Heringstad, and T. Steine Heritabilities, Genetic Correlations, and Genetic Change for Female Fertility and Protein Yield in Norwegian Dairy Cattle J Dairy Sci, January 1, 2005; 88(1): 348 - 355. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
