Accounting for Breed Differences
PART 2 OF A SERIES
In the first installment of this series describing the Multiple-Breed Evaluation System developed by the ASA and Cornell University, the new contemporary group definition and age of dam effects were presented. For review, the new contemporary group places all calves in the same contemporary group as long as they are the same sex and have been managed alike in the same herd. The new age of dam effects treat the age of cow as a continuous variable instead placing cows into twelve distinct groups based on their age. Also, the age of dam adjustment will depend on the sex of the calf and the breed makeup of the cow.
Two important effects need to be adjusted for when using records from crossbred calves, breed differences and heterosis. The effects of heterosis are split further by looking at heterosis of the calf (direct) and heterosis of the dam (maternal). If the breed differences and heterosis effects are not accounted for properly, these differences could end up as part of the animal's EPDs.
Breed of Founder Effects
The breed differences in the Multiple-Breed Evaluation are accounted for determining the different breeds that exist in an animal's pedigree. All the pedigrees in the ASA data base are traced back to the most remote ancestor in each pedigree. These distant ancestors are called founders. The breed composition of a calf is determined by the breeds of all founders in the calf's pedigree and the number of generations between the calf and the founder animals. The expected genetic value of an animal is the weighted average of the breed of founder (BOF) effects as shown in table 1. If the calf was 75% Simmental, 25% Angus, its expected genetic merit would be the weighted average of the Simmental and Angus BOF effects. The BOF effect accounts for the genes from various breeds that contributed to the Simmental population through founder animals.
75% Simmental, 25% Angus
3/4 BOF(SM) + 1/4 BOF(AN)
If breeders have been practicing selection, genetic trend will be present in the breed. A yearly BOF effect was created to account for the genetic trend that may be present in a breed. The yearly BOF effect adjusts the evaluation for the animal's breed composition along with the year(s) the genes were sampled from their respective breeds. There are sixty-three breeds represented in the ASA data base. Since only a handful of these breeds are represented in the ASA data to a significant degree, twelve breed groups were created for each year: Simmental, Angus, Hereford, Brahman, Gelbvieh, Charolais, Limousin, American, British, Continental, Dairy, and Mixed. Given the nature of the ASA data base, it is very doubtful that these different BOF effects could be estimated from the data alone. A statistical procedure was used that combines information reported from the scientific literature (priors) and the information contained in the data. The prior information is very useful for those breeds that received very little sampling into the Simmental data base. For breeds like Angus, Hereford or Brahman, the prior values have very little influence since so many animals in the data have genes from those three breeds. The information from the literature was summarized to provide estimates of breed differences (Table 2). Based on roughly fifty published reports, these differences are our best estimates of the differences between the breeds represented in the ASA data. Also, the BOF solutions for 1991 are listed for each trait in table 2. The influence of the data on the BOF solutions can be seen by comparing the BOF solutions with the prior values for each breed group. For a few breed groups, the difference between the solutions and priors is very small indicating one of two things: 1) The prior values were very close to the differences seen in the Simmental data; or 2) There were very few animals with any genes from those breed groups in the Simmental data.
Table 2. Prior Values and 1991 Solutions for Breed of Founder Effects
(Scale: lb., 2xEPD, Simmental Base Breed)
| ||Prior|| 1991|| Prior|| 1991|| Prior|| 1991|| Prior|| 1991|
|Simmental ||0.0||-0.3||0.0 ||4.3||0.0 ||4.5 ||0.0 ||7.4 |
|Angus ||-18.2||-11.9||-58.6||-43.1||-40.7 ||-30.3 ||-21.7||-8.4 |
|Hereford ||-11.4||-9.6||-56.4 ||-49.7||-56.0||-50.7 ||-42.49||-29.8 |
|Brahman ||1.8||0.2||-46.8||-49.8||-51.1||-50.9||-6.5||-10.2 |
|Charolais ||3.7||-1.5||11.8||4.3||-14.6||-19.1|| -9.1||-6.4 |
|Limousin ||-4.3||-6.5||-19.1 ||-21.2||-46.7||-44.7 ||-8.1||-12.8 |
|American ||-10.6||-9.8||-58.6 ||-57.5||-95.9||-90.9||7.2 ||8.2 |
|British||-17.2||-12.1||-76.3 ||-66.8||-26.2||-23.9||-12.5||-2.5 |
|Dairy||-5.1||-3.2||-24.2 ||-27.6||-32.1||-33.7 ||65.9||47.9 |
These breed differences are expressed on a breeding value scale relative to Simmental. The important thing to consider is not the actual values for each breed group but the differences between breed groups. The difference between Angus and Simmental founders in 1991 for birth weight was -11.6 lb. Expressed on an EPD scale, the difference is -5.8. The contrasts between Brahman and Simmental for weaning weight and postweaning gain in 1991 were -54.1 and -55.4 lb., respectively.
The prior values are assumed to the same for each year. The data will provide the information to determine if a genetic trend exists in the genes sampled from the different breed groups represented in the Simmental data. If only a few animals from founder groups enter the population each year, the yearly BOF trends will fluctuate due to the limited sample of genes each year. Another statistical procedure was used to smooth the estimates of the BOF trends so the yearly estimates do not bounce wildly from year to year.
The BOF trend describes the average merit of the genes from a breed group entering the Simmental population in a given year. The founder animals may not be a random sample from their breed so the BOF effects are not true breed differences. The BOF trends for birth weight are shown in figure 1 for Simmental, Angus, Hereford, and Brahman. These trends are represented on a breeding value scale. From 1970 to 1994, the difference between founders from the Brahman and Simmental breeds increased from -1.6 lb. to +0.7 lb. The Angus trend indicated that breeders selected Angus cattle with higher birth weight genetics to enter the Simmental population. Compared to the prior difference between Angus and Simmental, -18.2 lb., the difference between Angus and Simmental founder genes was -12.0 lb. in 1994.
The BOF trends for maternal milk are shown in figure 2 for Simmental, Angus, Hereford, and Brahman. Breeders sampled genes for increased maternal ability from the Angus and Brahman breeds as shown by the increased trends from 1970 to 1994. The Hereford trend increased from 1970 to 1980 then stayed relative flat to 1994. From 1980 to 1994, compared to Angus and Brahman, there has been fewer Herefords entering the Simmental population as founder animals. Also, the Hereford genes present in the Simmental population have received very little selection for improved maternal performance.
The phrase, breed of founder, is used to indicate that a random sample of genes from a contributing breed is not needed to evaluate breed differences. The breed of founder effects are developed by tracing pedigrees and determining the breed(s) represented in the most distant animals (founder) in each pedigree. A time trend is included in the breed of founder effects to account the genetic trend that may be present in the genes sampled from other breeds. The MBE system uses two sources of information to estimate breed differences in the Simmental population: 1) Prior values estimated from the published crossbreeding studies in the scientific literature; and 2) the Simmental data base. The prior values provide a starting point which may be changed by the data in the MBE analysis. If the data contains a lot of animals with some Angus breeding, the prior difference between Angus and Simmental will have little impact on the final estimate. For other breeds that are not represented as well as Angus, Hereford, or Brahman, the prior values have a significant influence on the final estimates.