Genetic Evaluation

Back Home Up Next



International Beef cattle Genetic Evaluation

Contribution of ICAR to an International Genetic Evaluation on Beef Herds

Improving the Genetics of Reproductive Traits

Prototype EBVs for Calving ease in Limousin cattle in the USA

Mature Weight, Height and Condition Score in Limousin Cattle

Ultrasound scanning of Carcase Traits in Limousin Cattle

International Beef Cattle Genetic Evaluation

Presented to the International Limousin Conference, France September 2000

Written by J. K. Bertrand, D. K. Lee, and L. L. Benyshek The University of Georgia, Athens


Improvements in computing power and models and the establishment of genetic ties across countries due to the widespread use of common sires may provide the opportunity to combine the data bases from several countries in order to conduct multi-country genetic evaluations. Across country genetic evaluation using animal models could provide genetic values on all animals, including young seed-stock, that could be used to compare individuals on a global basis. This could increase the accuracy of prediction and could enhance the worldwide marketing of germplasm. However, the usefulness of a combined multi-county genetic evaluation would be hampered by heterogeneous genetic and environmental parameters. Genotype by environment interactions that cause animals to re-rank for genetic performance across countries could be a concern. Across country genetic evaluation will also require some standardization of performance and pedigree information across countries and will require cooperation among the organizations responsible for performance programs in participating countries. This paper provides research results that examined the similarity of genetic and environmental parameters and the importance of genotype by country interactions across populations of Hereford cattle in four countries. The paper also provides some discussion on technical and organizational issues involved with global beef cattle genetic evaluation.

Investigation of Heterogeneous Parameters and Genotype by Country Interactions

Multi-country genetic evaluation will be the easiest to implement for those countries where the populations within each country can be consider sub-populations of a large global population with the same genetic and environmental parameters and also where animals rank genetically the same in each country. Meyer (1995) examined Angus populations in Australia and New Zealand and found that genetic and environmental parameters were very similar for birth, weaning (200-day), yearling (400-day) and final (600-day) weight. Johnston (1992) also found that Hereford populations in Canada and the U.S. were similar for weaning weight (205-day) genetic and environmental parameters. Both of these studies involved countries that were very close geographically. A study was conducted at the University of Georgia that compared genetic and environmental parameters for 205-day weaning weight from populations of Herefords in Argentina, Canada, Uruguay and the U.S. A portion of this study is found in de Mattos et al. (2000b). The original data sets consisted of 2,333,722, 44,005, 487,661, and 102,986 edited weaning weight records from the American Hereford Association (AHA), the Asociacion Argentina Criadores de Hereford, the Canadian Hereford Association (CHA) and the Sociedad Criadores de Hereford del Uruguay, respectively. These weaning weight records had been subjected to standard edits that included eliminating records of animals outside the range of three standard deviations from the overall mean and eliminating single record contemporary groups. The basic definition of contemporary groups in all four within country data sets was similar and was based on herd, sex, weaning management code (related to feeding regime of calf), producer assigned pasture code and the date a group of calves were weighed at weaning. Weaning weight records within all four countries were preadjusted for age-of-calf and age-of-dam. The American Hereford Association and CHA used the same age-of-dam and age-of-calf adjustments in the combined genetic evaluation for both associations. The Hereford populations in Uruguay and Argentina used different adjustments. Different age-of-dam adjustments were also used for the Angus populations in Australia and New Zealand in the study reported by Meyer (1995).

In order to reduce the size of the data set within each country, small herds with less than 500 weaning weight records were eliminated, then the data remaining for each country was sampled 10 times. To produce a sample, herds were selected at random with replacement from the data set available for each country. Selection of herds at each sampling was halted when the data set size was between 20,000 to 40,000 records. The data was then checked for direct sire connectedness across contemporary groups and disconnected contemporary groups were eliminated. A single trait animal model that contained fixed contemporary group effects, random additive direct genetic effects, random additive maternal genetic effects, random permanent maternal environment effects and random error was fit to the adjusted weaning weights in each of the 10 samples within each country. Variances and covariances were estimated using REML and the means and empirical standard deviation (SD) of the parameter estimates from the ten samples were calculated.

Table 1 contains weaning weight genetic and environmental parameter estimates for Herefords in Argentina, Canada, Uruguay and the U.S. In general, the parameters across the four were similar. All the genetic and environmental parameter estimates for Canada and Uruguay were within 2 SD of the U.S. estimates, when the mean and SD of the U.S. was the basis for comparison. Furthermore, the genetic parameter estimates for Argentina, Canada and Uruguay were within 2 SD of the U.S., when the mean and SD of the U.S. was the basis for comparison. Estimates of maternal permanent environmental and error variance for Argentina were between 2 and 3 SD of the U.S. estimates, while the phenotypic variance estimate was outside of 3 SD. Even though the weaning weight phenotypic variance in Argentina may be larger than the other three countries, the other genetic and environmental parameters appear to be proportional across all four countries, thus the data could be combined after correcting for differences in phenotypic variances across countries.

Even though genetic and environmental parameters appear to be very similar across three of the countries studied and correctable in the fourth, the presence of genotype by environment interactions that cause animals to re-rank in different countries could still make multi-country genetic evaluation difficult to implement. Many studies have reported important sire by environment interactions in beef cattle weaning weight field data. Bertrand et al. (1985) and Notter et al. (1992) found sizeable sire by herd interactions, and Bertrand et al. (1985), Bertrand et al. (1987), and de Mattos (1996) reported that sire by region within country or maternal grandsire by region within country interactions were large enough to indicate sire rank changes across regions. However, Herring et al. (1995) reported that sire by country interactions were small for weaning weight in Hereford populations in Canada and the U.S. All the above studies used either sire or sire-maternal grandsire models to investigate genotype by environment interactions. Hyde et al. (1998) used animal models to predict genetic values separately for Charolais populations in Australia, Canada, New Zealand, and the U.S. The data from these countries along with data from the United Kingdom was combined and a joint evaluation was conducted to predict genetic values. This study compared actual correlations of the breeding values of animals in two different populations with expected correlations that were computed by taking the product of the approximate accuracies of animals in both populations. Actual correlations were well below the expected correlations for sires with weaning weight accuracies greater than or equal to .70 and for sires registered in more than one country. The authors concluded that genotype by environment interactions may exist in this population and may be the result of different reporting policies across countries. In another study, Meyer (1995) analyzed weaning weight as a different trait for Angus cattle in Australia and New Zealand using a multiple trait animal model and reported direct genetic and maternal genetic correlations of .97 and .82, respectively. Meyer (1995) citing earlier work by Robertson (1959) concluded that genetic correlations for the same trait in different environments > .80 indicated that genotype by environment interactions were of little biological importance.

To further investigate the importance of genotype by country interactions, another study was conducted at the University of Georgia. A portion of this study was also reported by de Mattos et al. (2000a). Sires that had a "large" number of progeny in each country were identified for each pairwise ample of countries. In samples involving the U.S. or Canada, sires were required to have 50 progeny. In samples involving Argentina or Uruguay, sires were required to have 25 progeny. Pairwise country samples were then obtained by eliminating herds within each country that contained less than 500 records, that had an average contemporary size of nine or less, and that did not contained records from progeny or maternal grand-progeny of the across country identified males. Further reduction of the data sets in each country was accomplished by selecting a random sample of the remaining herds so that the data set within each country contained between 20 and 30 thousand records. Again, direct sire connectedness across contemporary groups were checked and disconnected contemporary groups were eliminated. Four regions, Upper Plains (UP), Cornbelt (CB), South (S) and Gulf Coast (GC) were defined by procedures provided by Leighton et al. (1982). Figure 1 provides the location of these regions. These regions were selected because some are similar in climate and production systems (UP vs CB and S vs GC) and some are diverse in climate and production systems(UP-CB vs S-GC). Figure 1 provides the location of these regions along with other regions that were defined in the study by Leighton et al. (1982). The same sampling criteria and common international sires were used to form data sets to investigate interactions across regions within U.S. Pair wise analyses was done with UP-CB vs S-GC; UP vs CB and S vs GC. These pair wise region data sets were similar in size to the pair-wise country data sets. In order to estimate the direct and maternal genetic correlation between the same trait in two different environments, a multiple trait animal model that used all available pedigree information was fit to the data samples between countries and regions within U.S.

Direct genetic correlation estimates across Argentina, Canada, Uruguay and the U.S. ranged from .87 to .82. Maternal genetic correlations ranged from .81 to .82 across Canada, Uruguay and the U.S. The maternal genetic correlations across Argentina-Canada and Argentina-U.S. were both estimated at .80. The maternal genetic correlation estimate for Argentina-Uruguay was .73. The only correlation that was not greater than or equal to .80 was the correlation estimate for maternal genetic effects between Argentina and Uruguay; however, these two data sets had the smallest number of observations, and we expect the correlations to improve as more data is collected. Both the direct and maternal genetic correlation estimates for weaning weight across regions within the U.S ranged from .84 to .87. The correlations across regions within the U.S. were similar to the same genetic correlation estimates across countries, indicating that genotype by country interactions were of minor importance for weaning weight.

Direct genetic correlations across countries were also estimated for postweaning gain. In Canada and the U.S., 160-day postweaning gain was the trait analyzed, and in Argentina and Uruguay, 345-day postweaning gain was the trait analyzed. The genetic correlations for postweaning gain between Argentina-Canada, Argentina-Uruguay, Argentina-U.S., Canada-Uruguay, Canada-U.S. and Uruguay-U.S. were .64, .79, .53, .82, .92 and .82, respectively. Postweaning gain correlations were > .80 across Canada, Uruguay and the U.S. The correlation for postweaning gain between Argentina and Uruguay was very close to .80; however, the postweaning gain genetic correlations of between Argentina-Canada and Argentina-U.S. indicated some re-ranking of sires. It is expected that correlations involving Argentina will increase as more data is collected. The postweaning gain phenotypic variance for Argentina was also three times the magnitude of the U.S. and Canadian postweaning gain phenotypic variance. Therefore for the purposes of Hereford weaning weight and postweaning gain evaluation across the four countries, the best analysis procedure would be to use a multiple trait model that fit weaning weight direct and maternal as the same trait across all countries and treated 160-day postweaning gain in Canada and the U.S. and 345-day postweaning gain in Argentina and Uruguay as separate traits. There is also a high likelihood that birth weight will perform similarly to weaning weight and would also be a good candidate trait for across country genetic evaluation. However, additional research is needed to investigate the importance of heterogeneous parameters and genotype by country interactions for other traits of economic importance.

Important Technical and Organizational Concerns

An important technical issue will be the determination of the minimum amount of connectedness that would be necessary before a country could be included in a global genetic evaluation. In the case of Herefords, between 15 and 20 % of the weaning weight records in Uruguay and Argentina are from progeny and maternal grand-progeny of bulls that originated in either Canada or the U.S., so lack of connectedness should not be problem for a genetic evaluation of Herefords across these four countries. A large amount of semen from several breeds in the U.S. and Canada has been exported to temperate areas in Australia, Argentina, New Zealand, and Uruguay. Therefore, for several breeds it is highly likely that 15% or more of the performance records for birth and growth traits in these countries are from bulls that originated in North America. However, lack of connectedness and genetic ties across countries may be a problem for some breeds. Kennedy and Trus (1993), in a simulation study, concluded that the variance of the estimated difference between management unit effects was highly correlated with the average prediction error variance of pair wise differences between animals in the management groups. It may be possible to compute the variance of the difference between country effect estimates, and use this approach as a measure of connectedness to determine the impact of including the information of a country on the accuracy of across country evaluation. This approach could be particularly useful in identifying which countries have the least amount of connectedness. These countries would then be prime candidates for a program designed to increase the ties between countries through the use of semen of high accuracy internationally used sires. Research needs to be conducted to develop criteria for the minimum amount of connectedness needed for a country to participate in across country genetic evaluation.

Another technical problem is the identification of common animals across countries. Breed associations and other performance organizations need to develop protocols to retain country of origin, country of origin registration or unique identification numbers, tatoos, birth dates, names, and sire and dam identification numbers for any animal imported from another country. It may be that centralization of performance programs for organizations in different countries is needed to facilitate the accurate identification of animals and to standardize the handling of performance information. At the least, some standardization of data formats and field descriptions need to be developed to facilitate ease of data exchange across countries. However, global standardization of beef performance programs may be a difficult task because of across country differences in criteria for assigning market value to beef cattle, differences in the amount of governmental control over performance data collection and genetic evaluation programs, and differences in the type of performance data that has traditionally been collected. Calf weaning weights are generally adjusted to either 200, 205, or 210 days of age. Most countries also collect some type of yearling or post-weaning weight between 300 to 540 days of days of age. Because of the close similarity across countries for these traits, it is likely that across country genetic values for birth weight, growth to weaning and yearling, and weaning maternal performance will be the first values produced and used to compare potential breeding stock on a world-wide basis. Calving ease is another trait that may be a candidate for international evaluation. However, countries vary in number of calving ease categories that are used in their within country evaluation programs. Research needs to be conducted to determine whether the best way to conduct international evaluation for calving ease is to combine categories so that only two categories are used in all countries, or to retain the category differences across countries and employ the fitting of multiple thresholds across countries to scale to the underlying trait. Genetic values for carcass traits will be difficult to produce on an international basis. There are considerable differences across countries in the management of slaughter cattle, in the age at slaughter, in the sex classes (bulls vs steers) of slaughter cattle, in the amount of finish, and in the type of carcass measurements obtained. Across country genetic values for reproductive traits like scrotal circumference and gestation length are may be possible for countries that collect these traits. Much research must be conducted to determine if genetic values for reproductive traits such as cow longevity and stayability can be predicted on an across country basis. Because of the large influence that the environment plays in the expression of reproductive traits, it is likely that these traits will be more subject to genotype by environment interactions and heterogeneous variances than other traits.

An important organizational issue that will arise with across country evaluations involves the extent that data between countries will be shared in order to provide service to breeders. In order for breeders in each country to receive information on why an animal's EPD changed from one analysis to the next, it will be necessary for organizations in each country to have access to the entire across country data set. The sharing of information can be a sticky issue because organizations within each country have invested significant time and money in their performance programs and in the collection of performance data for the sole purpose of helping their own breeders compete in the marketing and selecting of cattle. Breeders within each country will need to be convinced that across country genetic evaluation program and the sharing of information between countries will better help them compete when marketing and selecting animals on a global basis. Some other organizational concerns that will need to be addressed are the timing of and number of yearly analyses, standardization in presentation and use of results, the process of funding the evaluation and the decision of which traits should be included in the evaluation.


Across country genetic evaluation for beef cattle is already a reality between several countries that are in close geographical proximity. Recent research is indicating that genotype by country interactions are not of major importance between countries with temperate climates, even when those countries are in different hemispheres. It appears likely that across country genetic evaluation that involves the combining of data into a single evaluation can be conducted for birth and weaning weight, provided that participating countries and organizations work together to make it happen.

Literature Cited

Bertrand, J. K.; P. J. Berger and R. L. Willham. 1985. Sire by environment interactions in beef cattle weaning weight field data. J. Anim. Sci. 60:1396-1402.

Bertrand, J. K., J. D. Hough and L. L. Benyshek. 1987. Sire x environment interactions and genetic correlations of sire progeny performance across regions in dam-adjusted field data. J. Anim. Sci. 64:77-82.

De Mattos, D., J. K. Bertrand, W.O. Herring and L. L. Benyshek. 1996. Sire and maternal grandsire by environment interactions for weaning weight in a Hereford beef cattle population in Uruguay. J. Anim. Sci. 74 ( Suppl. 1):117 (Abstr.).

De Mattos, J. K. Bertrand, and I. Misztal. 1999a. Investigation of genotype by environment interactions for weaning weight for Herefords in three countries. J. Anim. Sci 78:(In press).

De Mattos, D., I. Misztal and J. K. Bertrand. 1999b. Variance and covariance components for weaning weight for Herefords in three countries. J. Anim. Sci 78:33-37.

Herring, W.O.; J. K. Bertrand and L. L. Benyshek. 1995. Genotype by environment interactions in Hereford beef cattle data in the United States and Canada. J. Anim. Sci. 73 (Suppl. 1):163 (Abstr.).

Hyde, L.R., R. M. Bourdon, B. L. Golden and C. M. Comstock. 1998. Genotype by environment interactions for gowth and milk traits in an international population of Charolais cattle. J. Anim. Sci. 76 (Suppl. 1):60 (Abstr.).

Johnston, D. J. 1992. International genetic evaluation for the North American Hereford and Polled Hereford populations. Ph.D. dissertation. University of Georgia, Athens.

Kennedy, B. W. and D. Trus. 1993. Considerations on genetic connectedness between management units under an animal model. J. Anim. Sci. 71:2341-2352.

Leighton, E. A., R. L. Willham and P. J. Berger. 1982. Factors influencing weaning weight in Hereford cattle and adjustments factors to correct records for these effects. J. Anim. Sci. 54:957-963.

Meyer, K. 1995. Estimates of genetic parameters and breeding values for New Zealand and Australian Angus cattle. Aust. J. Agric. Res. 46:1219-1229.

Misztal, I. 1998 . REMLF90.Manual. Available at:

Notter, D. R.; B. Tier and K. Meyer. 1992. Sire by herd interactions for weaning weight in beef cattle. J. Anim. Sci. 70:2359-2365.

Robertson, A. 1959. The sampling variance of the genetic correlation coefficient. More details in press, Biometrics 15:469-485.

Figure 1. Boundary definitions for nine geographic regions of the United States.

Table 1. Means and SD for the (co)variance estimates for weaning weight from the 10 samples within each country




































































F2d = direct additive genetic variance (kg2), h2d = direct heritability for weaning weight,

F2m = maternal additive genetic variance (kg2), h2m = maternal heritability for weaning weight, Fd,m = direct and maternal covariance (kg2), rd,m = genetic correlation between direct and maternal genetic effects,

F2pe = permanent maternal environmental variance (kg2), c2 = permanent maternal environment variance as a proportion of the phenotypic variance,

F2e = error variance (kg2),

F2P = phenotypic variance (kg2).

Back to Top



Presented to the International Limousin Conference France September 2000

by Dr. Hans J. SCHILD

Landeskuratorium der Erzeugerringe in Bayern Germany.


The International Committee for Animal Recording ICAR is an international non-profit and non-governmental Organisation. The main objective of ICAR is the establishment of standards for animal recording by appropriate guidelines. In this respect there is a close co-operation with other standard setting organisations like ISO.

ICAR members mainly are national umbrella organisations, which deal with animal recording. At present almost 50 countries and three world breed associations (Holstein, Ayrshire, Guernsey) are involved.

Besides its members ICAR consists of the board, the secretariat in Rome, three permanent sub-committees (e.g. INTERBULL) and 13 working groups, which in case of overlapping problems act in close co-operation.


The beef working group was founded in 1990 and consists of 13 members from different countries, which are involved, either in national beef recording schemes or in genetic evaluation of beef traits.

According to ICAR's objectives the ICAR Beef Group is responsible for:

International standardisation of beef recording
Development of recommendations and guidelines for beef performance recording
Safeguarding of consistency and agreement with guidelines of other ICAR working groups
International surveying beef performance recording

Up to now the ICAR Beef Group developed the following guidelines:

Linear scoring of muscle shape
Comparable genetic evaluation for beef traits
Recommendation for testing schemes
                                    - In suckler herds
                                    - Individual test station
                                    - In abattoirs
                                    - In finishing herds
                                    - At official animal sales

With respect to the big diversity of applied recording schemes the latter guideline up to now a rough frame than a detailed recommendation. However, based on a worldwide survey which will be completed by the end of this year, a comprehensive and detailed update is planned for the next future. The aim is to draw a general synthesis of different nationally applied guidelines for beef recording.

The new guidelines probably will be completed by 2001/2002. They will not only refer to beef characteristics including quality items and more advanced recording technology like video imaging but also to female characteristics and male fertility.

Based on the results of the survey, it is planned to establish an international data dictionary according to the ISO standards for agricultural data exchange (ADIS/ADED). The detailed data description then will apply worldwide and can be used for automatic data exchange at any data processing level.

Beef data recording forms the base for farm management tools as well as for genetic evaluation. Although not being directly involved in genetic evaluation the ICAR Beef Group provides the link between data recording and use of this data by appropriate genetic evaluation procedures thus being a part of a complex system. By this reason the guideline "Comparable genetic.evaluation for beef traits" was developed. The guideline refers to the difficulties being faced if genetic proofs from different countries are compared.


3.1 Problems Arising with Comparisons of Different National Results

Genetic evaluation up to now is mostly performed by a national scheme not taking into account that an increasing international genetic exchange takes place. In this respect it might be difficult to compare results from different national evaluation schemes even for the same or at least similar trait. The main reasons are:

Calculation methods

Although the application of multiple trait animal models including maternal and direct genetic effects has become a certain standard, there remain some systematic differences between countries.

Besides additive genetic effects taken into account by within breed estimation, non additive genetic effects by across breed evaluation when across breed evaluation applies. Furthermore procedures have been developed which try to incorporate foreign genetic proofs, which might affect national breeding values.

The manner how non-genetic effects are taken into account differs significantly with regard to fixed vs. random, to the nature and number of considered levels and to pre-adjustments vs. direct consideration in MME.

The number of traits taken into account affects the genetic proof for a certain trait by appropriate genetic correlations. This applies especially for negative correlated traits when using multiple trait models in various forms vs. single trait models.

Despite the fact that heritabilities of the same trait frequently are at least in the same range, the assumed correlations can differ significantly. This affects the genetic proofs, especially in a situation with an unbalanced data design.

Expression of genetic proof

Due to differences in evaluation history and to differences in national units of weights and measures, the expression of genetic proofs differs across countries.

The expression of genetic proofs can either be predicted difference or breeding value. Furthermore the units of measurement show big differences like kg vs. pounds, ratio out of 100, scores with different meaning

The genetic base usually show big differences such as rolling vs. fixed and according to the composition of the bases like birth years vs. genetic groups across birth years or males vs. all sexes.

Composed indexes depend from traits, their mutual correlation and the economic weight per trait being taken into account taken into account. For this reason they can hardly be compared.

Expression of accuracy

Due to different evaluation histories the accuracy of genetic proof is expressed in different ways. In some countries the coefficient of determination: R 2 Al is used. Other countries use the appropriate square root RAI or other expressions like: 1- (1 - R2 )1/2


The publication usually is confined to sires used for natural mating or Al. Usually it is performed once a year by sire summaries. However there are differences in the frequencies and date of publication. Depending on traits taken into account the publication criteria show differences in accuracies and connectedness across herds.

3.2 Use of Conversion Formulas

A first approach to use foreign genetic proofs might be the application of conversion formulas. The aim of conversion formulas is to refer genetic proofs to a scale which is used within country and thus to provide an easy direct comparison between foreign and domestic proofs. With regard to the above-mentioned problems ICAR guideline recommends the following:

Conversion formula should be based on the concepts, which have been implemented for dairy traits in former times:

limp = a + b I exp

limp foreign proof on domestic scale

a fixed constant

b factor (= Simp/Sexp)

I exp foreign proof on domestic scale

The calculation of a- and b-values is due to the responsibility of the domestic country. It should only performed by original genetic proofs of domestic and foreign proofs. A re-calculation to the original scale is not allowed. The use of conversion formula requires the following prerequisites:

Performance recording should fully follow the ICAR guidelines
Similar genetic evaluation according a standard to be defined mainly to take the consequences of a wide use of multi trait model into account and the correlations between direct and maternal effect Genetic model should be documented
Minimum level of genetic connection between populations.

There are numerous attempts to use conversion formula calculation according INTERBULL recommendations. However their use in the beef context might be difficult when practically applied, mainly because of:

Small size of the recorded herds whose available data frequently is not fully used
Low accuracy of sire proofs which even decreases by the use of conversion formula
Low number of sires which are used simultaneously in different countries but the other sources of genetic connection like common ancestors are not fully used
Low percentage of inseminated cows which might lead to a higher degree of genetic connectedness between herds from different countries or even between herds within the same country.


4.1 Problems Arising with Joint International Genetic Evaluation

Compared with dairy breeds the joint genetic evaluation of beef breeds appears to be much easier as the amount of raw data is much more limited. However one must be aware that in practice many problems exist, which in specific situations even can prevent any joint evaluation. The following parts briefly summarise some of the most severe problems, which might arise:

Data base

Common database requires at least a unique identification of animals. In case of identical traits, units of measurement and code sets used for traits as well as for fixed effects also should be identical. The animal records should be performed by similar production conditions. Otherwise severe G x E interactions may occur.

Animal identification

Frequently different animal numbers are used for identical animals in different countries. The identification of identical animals is a laborious but indispensable task to be undertaken by the involved breeding organisations. In this connection ICAR recommends the general use of the ISO number of the animal's origin country (ISO/TC23/SC19/WG3)

Recorded traits animals

Apart from the fact that most of the recorded populations are small, heterogeneous recording systems may lead to different information traits and different traits in the evaluation objective. The definition of recorded traits may differ in different national databases. In this respect differences in standardised weaning age (ICAR recommendation: 200 days) do not play a role, as the weaning age can be considered in the evaluation model directly. This would of course require non-corrected weights.

However especially for scored traits like female fertility, calving ease and to a certain extent muscular and skeletal development different scales with different levels may be used in the national databases. This will result in reduced correlations among similar traits across countries.

Furthermore the extent of data recording is different for different traits used in different countries. When using a multiple trait model this affects the resulting breeding values. This aspect applies not only for traits but also for animals being included in the recording scheme. Recording of selected animals/herds vs. whole herd recording arises the danger of information loss and data pre-selection.

Recording of non genetic effects

Apart from continuous distributed effects like age other non-genetic effects have to be taken into account. It is not clear if any of important effects is recorded in the involved countries. Even in case that yes, there is a big variety in grouping the contemporaries. The grouping of animals has a direct effect on the expression of genetic and phenotypic variances.

Genetic and phenotypic variation of traits

Frequently little genetic diversity within herds but a big diversity of environmental conditions in performance-recorded animals can be observed. In general, the expression of genetic and phenotypic variances depends from production intensity. Usually increasing production intensity causes increasing genetic variances. However, even within comparable production systems the genetic and phenotypic variation among animals within a contemporary group may change due to the definition of contemporary groups. As a consequence genetic correlations even among the same similar traits across countries can be reduced.


Compared to dairy breeds there is only little genetic exchange among regions and herds. However, poor connectedness between populations is a severe problem, which has to be solved prior to any joint evaluation.

As demonstrated at the recent INTERBULL Meeting in Bled 2000, even in dairy population with a high amount of Al and big semen exchange frequently an insufficient connectedness between populations can be observed. By this reason, connectedness probably is the sensitive point for a joint genetic evaluation in beef cattle.

This aspect applies not only for routine evaluation but also for the data analysis, which has to be made prior to any routine procedure. The estimation of genetic correlations between the same or at least similar traits across country only can be performed if there is a sufficient connectedness between the involved populations.

If there is no connectedness between populations, breeding animals from different countries never refer to the common scale, a correct range applies only within country.

If there is only low connectedness by common ancestors, limited comparability applies for the group of common animals. However, due to the genetic trend this does not apply for younger animals

If there is biased connectedness in case that foreign bulls are mated with highly selected dams and their superiority is not taken into account, the genetic proofs of are biased as well.

G x E interactions

Beef production conditions cover a large range from seasonal grazing systems with extremely low intensity to highly intensive feeding systems with voluntary feed intake. As a consequence of this, different abilities for growth and development by the animal must be expected, which results in genotype x environment interactions.

As a consequence the order of genetic superiority is different due to the production system applied. This leads to reduced genetic correlations across countries.

An example might demonstrate this: For 142 German Simmental sires which in dual purposed use produced young finishing bulls in intensive South German production systems and on the other hand produced weaners in South African suckler herds we estimated the following correlations between breeding values:


South African weaners

German finishing bulls


BV-birth weight - direct

BV net gain - direct


BV - weaning weight - direct

BV net gain - direct


BV - growth - 12 mths

BV net gain - direct


BV - growth - 18 mths

BV net gain - direct




Even when weighing the results by reciprocals of the appropriate accuracies (mutual R2 approx. 0.7-0.9) there remain poor and unexpected genetic correlations.

4.2 Practical Approaches With A Common Data Base

In many international breed federations a strong tendency towards the use of one joint international genetic evaluation can be observed at present. However with regard to the arising problems shown in the chapter above, practical approaches are limited up to now.

US American university institutes, especially the Animal Breeding Department of the State University of Georgia developed procedures using common data bases for Hereford, Limousine and Charolais. Besides this the Australian BREEDPLAN made some attempt, with the inclusion of foreign breeding values at a domestic scale.

The ICAR subcommittee INTERBULL up to now confined its activities to dairy breeds only. However INTERBULL is one of the most experienced organisations with regard to the general problems arising from a joint international genetic evaluation.

Dr. Brian Wickham, the former president of INTERBULL who now is working for the Irish Cattle Breeding Federation, initiated an Irish attempt by planning the use of a common beef database of Ireland, UK and France. The terms of reference are already fixed and the establishment of an experienced evaluation team will follow. It is planned to develop most of the next steps by the assistance of INTERBULL which later will become involved in international beef evaluation.


Any joint international genetic evaluation of beef characteristics cannot be successful without a significant contribution of the breeders. In this respect the demand can be summarised as:

Define the traits of common interest
Define the minimum fixed effects to be taken into account
Define the expression of genetic proof and its accuracy
Define the common evaluation center
Use more Al in order to connect herds/countries with different genetic level
Provide one unique number per identical animal used in different countries
Check ancestry and pedigree data for identical animals
Provide one unique number per identical animal used in different countries
Provide uniform recording standards by the adoption of appropriate (ICAR-)


(e.g. use 200-d standardised weaning weight instead of 210-d-weaning weight)

Support the activities of the ICAR Beef Group

Back to Top

Improving the Genetics of Reproductive Traits Defining Reproduction

Presented to the International Limousin Conference France, September 2000

By Kent Anderson of North American Limousin Foundation

Few would argue that reproductive traits are often the most economically important category of traits to producers. Given their economic importance, development of genetic predictions in the form of Expected Progeny Differences (EPDs) or Estimated Breeding Values (EBVs) for reproductive traits should be a priority for beef producers throughout the world. For most cattle breeders, successful reproductive performance means pregnancy and a live calf produced every year, and includes:

replacement heifers with the genetic potential to cycle and become pregnant early in their first breeding season,
young cows which breed back early during their second breeding season,
consistent annual calving after limited fixed breeding seasons until at least "break-even age", and
as much longevity as possible to minimize annual replacement rates.

EPDs for Reproductive Traits

Unfortunately, genetic prediction tools are not widely available for directly improving reproduction. While scrotal EPDs are good indicators of genetic differences in age at puberty in heifers and fertility of yearling bulls, research is beginning to indicate that the genes which control puberty are not necessarily the same genes which influence heifer pregnancy and sustained cow lifetime reproductive performance (longevity). In other words, while puberty is a prerequisite for heifer pregnancy, genetic differences in puberty do not adequately describe economically important genetic differences in heifer pregnancy and longevity.

With the help of researchers at Colorado State University, the North American Limousin Foundation has studied heifer pregnancy, as well as stayability, and has computed EPDs for each of these traits. Using new analytical techniques, heifer pregnancy was found to be more highly heritable than previously thought (.20), as was stayability (.20).

When considered along with EPDs for scrotal circumference, the Limousin breed is unique in the United States, in that the genetics for reproductive traits have been more extensively studied than in any other breed. Heifer pregnancy EPDs are expressed in units of probability, and indicate genetic differences in the likelihood that daughters are pregnant at the end of their first breeding season. Stayability EPDs are also expressed in units of probability, and indicate genetic differences in the likelihood that daughters remain in production to the age of six years or beyond, given that they entered production. For both of these traits, higher EPDs are favored, and represent genetics for greater inherent fertility.

Reproductive EPD Profiles

The concept that the genes which control puberty are different than the genes which influence heifer pregnancy and stayability is demonstrated by the actual EPDs for these traits on the Limousin sires in table 1. Each of these sires has a high accuracy EPD for scrotal circumference, but represent very different genetic potentials for age at puberty, heifer pregnancy and stayability. Consequently, it strongly appears that in order to adequately select for superior overall reproductive performance, additional reproductive EPDs must be considered besides just EPDs for scrotal circumference.

Table 1. Actual EPDs and Accuracies for Scrotal Circumference (SC), Heifer Pregnancy (HP) and Stayability (ST) for Example Limousin Sires.



A 1.3 (.76) 8 (.16) 19 (.49)

B 1.1 (.82) -5 (.24) 20 (.17)

C -.6 (.80) 11 (.33) 12 (.72)

D .1 (.85) -5 (.16) -2 (.49)


Both scrotal circumference and stayability EPDs are included as part of NALF's genetic evaluation program. Genetic predictions for indicator traits, such as scrotal circumference, effectively describe genetic differences in age a puberty, but may not adequately describe genetics for heifer pregnancy and sustained reproductive performance. As it relates to expressed fertility, interactions exist between genetic potentials for inherent fertility and traits such as milk production, mature size, and maintenance requirements. In order to adequately describe genetic differences in reproduction, it appears that genetic predictions must be developed for several different measures of inherent fertility and longevity.

Back to Top

Prototype EPDs for Calving Ease in Limousin Cattle Developed

Presented to the International Limousin Conference France, September 2000

By Kent Anderson of North American Limousin Foundation


The Limousin breed has generally enjoyed a very favorable reputation for calving ease and associated calf vigor and survivability. However, even low levels of calving difficulty results in increased veterinary and labor costs, calf death loss, calf susceptibility to disease, cow mortality, delayed return to estrus and lower conception rates. While birth weight EPDs are useful indicators of potential calving ease, NALF's database of over one million calving ease scores offers the potential for more powerful genetic predictions. Expected progeny differences (EPDs) for calving ease could help users of Limousin genetics more effectively select for high levels of calving ease in replacements.

A Cooperative Effort

Prototype EPDs for Limousin calving ease were made possible through the combined efforts of researchers at Cornell University, the American Simmental Association (ASA) and NALF. Cornell University researchers are leaders in the technology of genetic predictions for threshold traits such as calving ease. Over the years, Cornell has provided genetic evaluation services for the ASA, which helped to prepare NALF calving ease data for this analysis. Researchers to be credited for this study include: Dr. Zhiwu "Joe" Zhang, Ellen Chaffee, Dr. Richard Quaas and Dr. John Pollak from Cornell, and Dr. Bruce Cunningham from the ASA.

Indicator Traits versus the Economically Relevant Traits

As defined by Dr. Bruce Golden and Dr. Richard Bourdon in a recent paper entitled "New EPDs: A Rational Vision for the Future", economically relevant traits directly effect profit by being directly associated with a specific cost of production or an income stream from the sale of a product. Traits which do not directly affect profit, but are used to indicate the merit an animal has for another trait are called indicator traits. In the case here, calving ease is the economically relevant trait, while traits such as birth weight, gestation length, and pelvic measures are indicator traits.

Technical Notes for Prototype Calving Ease EPD Calculations

Researchers at Cornell University utilized a multiple trait, sire and maternal grandsire, threshold model for the computation of Limousin calving ease EPDs. Data contributing to the predictions included approximately a million records each for calving ease scores and birth weights from the NALF herd book. Genetic parameters used in the EPD calculations were from the ASA, and included the heritabilities for calving ease direct and calving ease maternal of .18 and .19 respectively, and genetic correlations of: -.41 between calving ease direct and birth weight, .14 for calving ease maternal and birth weight, and .13 for the genetic relationship between calving ease direct and calving ease maternal.

Since the vast majority of calving difficulty occurs in heifers, calving ease EPDs were expressed in units of percentage of unassisted births from first-calf, two-year-old heifers. Consequently, calving ease scores from Limousin heifers contributed the most meaningful data for the calculation of EPDs. The predictions also incorporated calving ease score information from mature cows and all available calf birth weight information. In birth groups with no calving difficulty, variation in calf birth weights contributed information to the predictions. For the purpose of this evaluation, the base or zero-EPD-point was defined as the average of all animals born in 1990.

Understanding Direct and Maternal Calving Ease (CED and CEM)

Genetically, there are two "types" of calving ease which are of interest. The first and most obvious type involves the question of genetic differences in the ease with which offspring (of a sire or dam) are expected to be born. This is called direct calving ease (CED). The second "type" of calving ease involves genetic differences in the ease with which daughters (of a sire or dam) are expected to give birth. This is called maternal calving ease (CEM).

The two "types" of calving ease are easier to understand if thought of as either traits of the calf or traits of the daughter. Direct calving ease is a trait of the calf, while maternal calving ease is a trait of the daughter. Numerous indicator traits exist for the two types of calving ease. Calf birth weight is the primary indictor trait for direct calving ease, while pelvic size of the daughter (dam) is the highest ranking and commonly measured maternal indicator trait.

EPDs for Calving Ease Direct (CED)

Direct calving ease EPDs predict genetic differences in the percentage of unassisted births when bred to first-calf, two-year-old heifers. Higher EPDs for CED are desirable, and indicate differences in the genetics of the calf which result in a greater proportion of unassisted births and lower incidence of calving difficulty (dystocia). For example, consider the CED EPDs for the following two sires:

Calving Ease Direct (CED) EPD

Sire A +5%

Sire B -5%

Difference 10%

If sire A and sire B were each bred to comparable groups of replacement heifers, based on differences in genetics for direct calving ease, we would expect 10% more of the calves out of sire A to be born unassisted as compared to the calves born from sire B (+5% to -5% = difference of 10%). Calving ease direct EPDs are especially useful when deciding which sires to breed to replacement heifers.

EPDs for Calving Ease Maternal (CEM)

Maternal calving ease EPDs predict genetic differences in the percentage of unassisted births from daughters. Higher CEM values are desired, and indicate genetic differences in daughters ability to calve unassisted (when bred to sires of equal genetics for calving ease direct). As an example, consider the EPDs for CEM for sires A and B below:

Maternal Calving Ease (CEM) EPD

Sire A +3%

Sire B -7%

Difference 10%

Based on genetic differences between these two sires in MCE, 10% (+3% to -7% = difference of 10%) more of Sire A's daughters would be expected to calve unassisted as compared to daughters of sire B, when each are bred to mates of comparable genetics for CED. Maternal calving ease EPDs aid in the selection of sires used to produce easy calving replacement females.

EPD Statistics for Direct and Maternal Calving Ease

The statistics for direct and maternal calving ease EPDs listed in table 2 help provide benchmarks to determine where sires rank in the population. Most sires (95%), have calving ease EPDs which fall within plus or minus two standard deviations from the average calving ease EPDs.

Table 2. Statistics for direct and maternal calving ease EPDs (all sires).

Trait Average Standard EPD

EPD Deviation Range

Calving Ease Direct 1.0 3.7 -33 to +18

Calving Ease Maternal -.1 3.4 -25 to 16

Limousin Genetic Trends for Calving Ease

Genetic trends for Limousin calving ease in the U.S. indicate that since the early seventies, a favorable, steady trend in direct calving ease was observed. For calving ease maternal, there was essentially no genetic trend from the early seventies until about 1990, after which a slightly negative, unfavorable trend developed during the last decade. Incorporation of calving ease EPDs into NALF's genetic evaluation program would provide breeders with direct selection tools to help facilitate simultaneous genetic improvement in both direct and maternal calving ease.


The prototype EPDs for direct and maternal calving ease provide powerful selection information for users of Limousin genetics wanting to more effectively minimize calving problems. Since calving ease is the economically relevant trait of interest, serious consideration should be given to the incorporation of calving ease EPDs into the Limousin breed's performance programs throughout the world.

Back to Top

Mature Weight, Height and Condition Score in Limousin Cattle

Presented to the International Limousin Conference France, September 2000

By Kent Anderson of North American Limousin Foundation

Genetic predictions for birth, weaning and yearling weight help to describe genetic differences in the growth curve for Limousin cattle up until yearling age. However, in order to more fully understand genetic differences in mature size, predictions are needed for mature size and fleshing ability. Predictions for these traits are potentially useful as indicators of maintenance energy requirements, doing ability and maturing patterns which influence the likelihood that progeny are within optimums for carcass weight.

NALF is currently involved in a study with the University of Arizona to develop genetic parameters and experimental EPDs for these traits. Data from 805 herds representing 38,804 cows is currently being analyzed. Preliminary estimates indicate a heritability of .20 for condition score, with permanent environmental effects accounting for 12% of the variation. While parameter estimates for mature weight are still yet to be completed, the average actual mature weight of cows has been calculated to be 1258 lbs. Technically, advanced statistical analysis techniques which utilize random regression are presently being programmed to accommodate multiple measures of weight in subsequent years on the same cows. When this project is completed later this year, genetic differences for the entire growth curve in Limousin cattle will be documented.

Back to Top

Ultrasound Scanning for Carcass Traits in Limousin Cattle

Presented to the International Limousin Conference France, September 2000

By Kent Anderson of North American Limousin Foundation

Back in late 1998, the NALF and the American Simmental Association were the first breeds to join the Centralized Ultrasound Processing (CUP) program established by Iowa State University and the American Angus Association (AAA). Currently, the number of breeds participating in the CUP program has grown to include13 different breeds. While Angus cattle account for nearly 85% of the cattle scanned and processed through CUP (in the year 2000), a growing number of Limousin cattle and cattle of other breeds are being added to the CUP data base.

During the 1999 calendar year, 1,779 Limousin cattle were processed though the CUP lab. Thus far in the year 2000, as of March 21, 792 Limousin cattle were processed through CUP. When added to the cattle scanning in 1998, NALF's database of ultrasound information now includes over 3,000 head of Limousin cattle. While the number of Limousin cattle scanned is encouraging, during 1999 over 37,000 head of Angus cattle were scanned (table 1), and as of March 21, over 20,000 have been scanned in the year 2000. Based on this comparative data, opportunity exists for Limousin breeders to more fully exploit use of ultrasound technology to determine composition and percent intramuscular fat differences in our population of cattle.

Following the Spring 2000 scanning season, NALF plans to develop updated breed specific adjustment factors for scan data and estimate heritabilities and genetic correlations for ultrasound carcass traits. A primary objective of this phase of the research project is to derive age adjustment factors for a wider range of possible scan ages. Currently, animals must be between 330 and 450 days of age for their scans to be adjusted to 365 days. A number of cattle in the scan data base are only about 300 days of age, and our objective is to attempt to derive an accurate formula to adjust these animals, such that their records can be used in the calculation of experimental EPDs.

If the ultrasound traits are reasonably heritable, our plan is to compute experimental EPDs derived from scan records. For NALF members who have had cattle scanned and images processed through the CUP lab, experimental EPDs are anticipated to be available for the scanned cattle as well as their sires and dams. Subsequently, for the higher accuracy sires, current carcass EPDs will be compared to the EPDs generated from the scan data to help validate the usefulness of the scan EPDs. Similar research efforts by the AAA have shown a relatively strong relationship between EPDs computed from ultrasound information versus EPDs computed from actual carcass data collected through progeny testing programs. Depending upon the results of this NALF sponsored research, the NALF board of directors will then decide whether or not to incorporate ultrasound information and EPDs into our performance program.

Genetic evaluation programs are run in the following countries. Information is available for those sites underlined:

Australia/New Zealand
United Kingdom
USA/Canada -

(addresses shown for linkage purposes)

Back to Top


c Copyright International Limousin Council.
For problems or questions regarding this web contact
Last updated: October 09, 2000 .