Project report: Sustainable Breeding in the Nordic Red Dairy Breeds Bærekraftig utnyttelse av den røde nordiske mjølkekupopulasjonen som genetisk ressurs Project report: Sustainable Breeding in the Nordic Red Dairy Breeds Bærekraftig utnyttelse av den røde nordiske mjølkekupopulasjonen som genetisk ressurs The project group has had the following memebers: Torstein Steine (Geno) Gunnar Klemetsdal (University of Life Sciences, Norway) Lars-Olof Bårström (Svensk Avel) Hans Eström (Nordic Gene-bank FarmAnimals) Erling Strandberg (Sveriges Lantbruksuniversitet) Jaana Kiljunen (Finnish Animal Breeding Association) Jarmo Juga (Finnish Animal Breeding Association) Morten Kargo (Danish Institute of Agricultural Sciences) Lisbeth Holm (Dansire), and AstridKarlsen (Geno) 2 1.0 Introduction .................................................................................................................................. 5 2.0 Assesing total profit of alternative levels of co-operation between Nordic cattle populations .............................................................................................................................................................. 6 2.1 Introduction ................................................................................................................................ 6 2.2 Aim ............................................................................................................................................ 6 2.3 Literature review of critical points around comparisons of single and multiple objective breeding schemes ............................................................................................................................. 6 2.4 Simulation study of three strategies of co-operation. ................................................................ 7 2.4.1 Material and methods .......................................................................................................... 7 2.4.2 Results ............................................................................................................................... 15 2.4.3 What is the exchange of red cattle bulls between Finland, Sweden and Norway today? . 22 2.4.4 Correlations between the breeding goals in the three countries ....................................... 22 2.4.5 Final conclusions .............................................................................................................. 22 2.5 References ................................................................................................................................ 23 3.0 Genotype by environment interaction ..................................................................................... 24 3.1 Introduction .............................................................................................................................. 24 3.2 Methods to study genotype by environment interaction .......................................................... 24 3.3 Does GxE exist? ....................................................................................................................... 25 3.4 Does GxE exist in Nordic countries? ....................................................................................... 26 3.5 Consequences for selection ...................................................................................................... 27 3.6 References ................................................................................................................................ 28 4.0 Breeding objectives and optimum use of sires in the Nordic red populations ..................... 30 4.1 Introduction .............................................................................................................................. 30 4.2 Methods.................................................................................................................................... 30 4.3 Results and discussion ............................................................................................................. 30 5.0 Use of EVA in Dairy Cattle Breeding Schemes ....................................................................... 35 5.1 Summary .................................................................................................................................. 35 5.2 Inbreeding – Cause and Consequences .................................................................................... 35 5.2.1 Balance between Genetic Gain and Relationship ............................................................. 36 5.2.2 Programmes to Handle Genetic Gain vs. Inbreeding........................................................ 37 5.2.3 ”Economic” Weighting of Genetic Gain and Relationship .............................................. 38 5.2.4 Implementation ................................................................................................................. 38 5.2.5 The Use of EVA for Red Dane ......................................................................................... 39 5.2.6 The use of EVA for Danish Holstein ................................................................................ 40 5.3 References ................................................................................................................................ 42 6.0 New techniques in dairy cattle breeding schemes ................................................................... 43 6.1 Use of Marker Assisted Selection in Nordic Breeding Schemes ............................................. 43 6.2 Use of sexed semen in the Nordic Red Breeds ........................................................................ 45 6.3 References ................................................................................................................................ 46 7.0 Nucleus breeding in Practice – a Nordic approach................................................................. 48 7.1 Introduction .............................................................................................................................. 48 7.2 The accuracy in the selection of bull dams .............................................................................. 48 7.3 Simulation of nucleus breeding schemes ................................................................................. 51 7.4 Nucleus breeding in practice .................................................................................................... 52 7.5 Recommendations and conclusion ........................................................................................... 54 7.6 References ................................................................................................................................ 54 8.0 Imports from other populations to the Scandinavian Red Dairy Breeds. ............................ 56 8.1 Introduction .............................................................................................................................. 56 8.2 What are we particular looking for for the Scandinavian Red breeds? ................................... 56 3 8.3 Where do we find possible sources for importations to the Scandinavian red breeds? ........... 57 8.4 Conclusions .............................................................................................................................. 58 9.0 Competitive advantages in the Nordic red breeds. ................................................................. 59 9.1 The Nordic Red Breeds ............................................................................................................ 59 9.2 Conclusions .............................................................................................................................. 61 10.0 Prospects for cooperation between Baltic and Nordic countries in dairy cattle breeding 62 10.1 Background ............................................................................................................................ 62 10.2 Milk recording in Baltic countries ......................................................................................... 62 10.3 Genetic evaluation ................................................................................................................. 63 10.4 Breeding programs in Baltic countries .................................................................................. 63 10.5 The possibilities for cooperation ............................................................................................ 63 10.6 Possibilities in marketing ....................................................................................................... 65 10.7 Contact people in Baltics ....................................................................................................... 65 10.8 References .............................................................................................................................. 65 11.0 Discussion/summing up ........................................................................................................... 67 Appendix 1. The use of EVA-programme in Finnish Ayrshire and Finncattle breeding schemes.............................................................................................................................................. 69 4 1.0 Introduction The breeding associations for the Nordic red cattle breeds Swedish Red (SRB), Norwegian Red (NRF) and Finnish Ayrshire (FA) have a long history of exchanging semen of the best elite sires. In later years also Danish Red (RDM) has been a part of this exchange system. Due to the exchange of semen, genetic ties exist between the four populations, and it is obvious that the main genetic pool for any of the red Nordic cattle populations is the red populations in the neighbouring Nordic countries. One of the reasons for the exchange of semen between the Nordic red populations is the similarity in breeding schemes in these countries compared to the breeding schemes for other dairy populations. The breeding programs for the Nordic red dairy breeds have focused on functional traits such as health traits, fertility, stillbirths, and calving ease as well as production. This is in contrary to the situation in the main dairy breed in the western world today, Holstein, where the breeding program has focused entirely on production and conformation until recently. It is well known that functional traits and production have unfavorable genetic correlations. Therefore selection for production will result in deterioration of functional traits. This has resulted in a competitive advantage for the Nordic red breeds on the international market, since these breeds are able to offer genetic material that is strongly selected for functional traits as well as production. Although semen has been exchanged between the Nordic red breeds during the last decades, the degree of exchange of semen has been decided upon by each individual breeding association. There has not been done any work to estimate the maximum, minimum or recommended use of sires across countries to obtain the optimum genetic gain. Another very important topic in this regard is inbreeding. Extensive use of a limited amount of elite sires in dairy populations have increased the rate of inbreeding rapidly. One way to reduce the amount of inbreeding is through import of semen from sires that are relatively non-related to the population they are imported to. A major effect of the exchange of semen between the Nordic red breeds, is that it has kept the inbreeding at a relatively low level in these populations. In the last decade we have seen a reduction in the number of cows in the Nordic countries, while the cost of running breeding programs are constant. It has become increasingly important to manage the breeding schemes within the Nordic countries as efficiently and innovative as possible. Fundings were provided by The Norwegian Ministry of Agriculture and Nordic Genbank Farm Animals in 2003 to start a project focusing on the future challenges in managing the Nordic Red dairy cattle populations to maximize genetic gain while focusing on problems regarding inbreeding and competitive advantages for the Nordic Red Breds. Some of the problems raised by the group are the following: How much exchange of semen should there be between the Nordic red breeds to obtain desired genetic gain with a low rate of inbreeding? What would happen if the breeding goals (total merit index) becomes more similar or more dissimilar between the Nordic red breeds? Are the Nordic red breeds able to serve as a genetic pool for each other to avoid increased inbreeding? May other breeds serve as a genetic pool to avoid increased inbreeding? 5 2.0 Assesing total profit of alternative levels of co-operation between Nordic cattle populations Anna K. Sonesson, AKVAFORSK 2.1 Introduction Although the Nordic countries breeding for red dairy cattle have three different breeding goals and different population and production structures, co-operation between the countries could be one alternative to increase the competitiveness of the Nordic red cattle populations. In this study, we will compare the effects of alternative levels of co-operation between the Nordic red cattle populations on profit. This report contains the following parts: 1. Literature review of critical points around comparisons of single and multiple objective breeding schemes. 2. Simulation study of three strategies of co-operation. 3. What is the exchange of red cattle bulls between Finland, Sweden and Norway today? 4. What is the correlation between breeding goals in Finland, Sweden and Norway today? 5. Final conclusions 2.2 Aim The main aim of this study is to compare three different selection strategies on the total profit of Nordic red cattle after 25 years of selection using computer simulation. Strategy 1. Selection within each of the three countries for each of its breeding goals. Strategy 2. Selection over the three countries for each of its breeding goals. Strategy 3. Selection over the three countries for one overall breeding goal. A small literature review of previous studies on comparison of across-country breeding schemes is included in this report. An estimate of the current exchange of bulls between Norway, Sweden and Finland will also be given. The three correlation coefficients of the breeding goals in Norway, Sweden and Finland are included as well. 2.3 Literature review of critical points around comparisons of single and multiple objective breeding schemes Critical points when comparing schemes There are very few studies on comparison of single versus multiple objective/line selection. Banos and Smith (1991) and Smith and Banos (1991) have looked at schemes for multiple objective selection in a single line. This should not be mixed up with multiple line selection for multiple objectives that have been compared to single line selection for single objective (e.g. Ollivier et al., 1990; Smith, 1986). One problem is to compare strategies at the same base, i.e. genetic response, costs of selection and inbreeding. 1. With single line selection, line mean response is the same as population mean response. For multiple line selection, response is the mean of several lines. When selecting for multiple objectives, animals are selected over lines. The need to allocate animals to the lines where they perform best was emphasised by Banos and Smith (1991). Howarth (2001) suggested to use the mean of the animals allocated to each of the lines. Thereby, it is possible to compare selection response for single and multiple objective schemes. Multiple line Closed (MLC), multiple lines open (MLO) and single line with an average objective (SLA) were 6 compared by Howarth (2001) with and without allocation of candidates and with different correlations between selection objectives. Allocation had effect only at low correlations between objectives, such that the two lines were divergent. Net allocated response of MLO was highest for all schemes. 2. In these comparative studies, the total number of individuals over multiple lines (ML) equals the number of individuals in the corresponding single line (SL). However, in the ML approach, selection intensities can be higher, which results in higher rates of inbreeding. In general, genetic gain for different schemes should be compared at the same rate of inbreeding, such that a comparison at different rate of inbreeding is not appropriate. Which are the critical variables for comparison of multiple objectives/lines? In general, the correlation between the objectives has been the variable to compare multiple line selection for multiple objectives at. Correlation of 0.8 between a trait measured in different environments before it was ‘biologically or agriculturally’ important (Robertson, 1959). Correlation of >0.75 between a nucleus and a particular breeder to make participation in a group breeding program useful (Del Bosque Gonzalez and Kinghorn, 1990). Correlation of 0.9, when the number of progeny tested bulls equals the total number of bulls over lines. When the number of progeny tested bulls was 5 % of the total number, the correlation was 0.7. (Goddard, 1992). Correlation of 0.7-0.8 (Meuwissen, 1998). These correlations require that objectives be composed of the same traits. Therefore, the correlation of selection indices that are composed of different traits may be more important (Howarth, 2001). However, also the means and variances of two populations are important variables when comparing single and multiple breeding objectives (Banos and Smith, 1991 and Smith and Banos, 1991). Howarth (2001) optimised size and proportional weight given to one or two breeding objectives. In general, the line with the highest mean or variance increases in size, whereas the other lines decrease in size. This convergence was faster when the correlation between breeding goals was high. Summary Not only the correlation between breeding objectives is an important variable to compare single and multiple objective schemes for, but also genetic means and variances are important variables. Because the selection ‘drifts’ towards a population with high genetic variance, it is important to reduce the loss of genetic variation in the population, i.e. to control the rate of inbreeding. 2.4 Simulation study of three strategies of co-operation. 2.4.1 Material and methods Genetic values Multitrait genetic values were simulated according to the infinitesimal model (Bulmer, 1985). Genotypes, gi, of the unrelated base animals were sampled from the distribution N(0, G), where the genetic covariance matrix G was standardised such that all genetic variances were equal to 1.0. Record yi was calculated as yi=gi+ei, where ei is the environmental effect, which was sampled from N(0, E), where 7 E denotes the matrix of environmental covariances. Later generations were obtained by simulating offspring genotypes from gi= +0.5gs+0.5gd+mi, where, s and d denote sire and dam of offspring i and mi=Mendelian sampling component, which was sampled from N(0, 0.5(1-Fsd)G), where Fsd is the average inbreeding coefficients of the sire and the dam. Selection strategies Three different selection strategies were compared for the total profit of Nordic red cattle after 25 years of selection. Strategy 1. Selection within each of the three countries for each of its breeding goals. Strategy 2. Selection over the three countries for each of its breeding goals. Strategy 3. Selection over the three countries for one overall breeding goal. Differences in costs between the strategies were ignored. The effect of these three strategies on the profit within each country and on the total profit across countries will be reported. Population structure and size Each of the three countries Norway (N), Sweden (S) and Finland (F) had two tiers: 1. Nucleus 2. Commercial The number of candidates of males (mal) and females (fem) and the number of selected sires and dams in the tiers are given below for each country. In the figures, arrows indicate where the male (left part of the figure) and females (right part of the figure) candidates come from for each of path of selection (sires in the nucleus, dams in the nucleus, sires in the commercial tier, dams in the commercial tier). Sires are selected among males from the nucleus, because these are assumed to be much superior to the males in the commercial tier. These same sires are also used in the commercial tier, such that the males in the commercial tier are ignored as candidates of selection. However, dams of the nucleus are selected among females from both the nucleus and the commercial tier, because the best females of the commercial tier are assumed to be equally good as the females of the nucleus since a more intense selected is made among the females of the commercial tier. Dams of the commercial tier are selected only among the females of the commercial tier. 8 Norway: In the nucleus, 125 bulls are progeny tested on 250 daughters each year. From these, the best sires will be selected to produce 125 male and 125 female offspring. We assumed that females lived three years, such that there are 375 (=3*125) female candidates from the nucleus each year. Due to computational limitations, the number of female selection candidates in the real commercial tier was corrected for the selection intensity for the bulldam selection step, which is the limiting selection step, due to the potential large use of each sire. In the real Norwegian population, 12000 potential bulldams are selected out of 270 000 dams in the commercial tier, corresponding to 4.0% selected dams. We know that 125 progeny tested bulls are needed and assume that they originate from the 4.0% selected dams. Therefore, 3125 (=125/4%) dams will be necessary each generation, i.e. approximately 1000/yr for three years. In order to get 1000 female offspring/yr, we need to select 2000 commercial dams, which are mated to the 25 best bulls from the nucleus. The scheme is summarised below. NUCLEUS 125 mal /yr No. sires by OC 125 mal/yr 375 fem/ year 125 dams/yr x offspring 125 fem/yr COMMERCIAL 25 sires/yr 1000 mal/yr x 3125 fem/yr 2000 dams/yr offspring 1000 fem/yr 9 Sweden: The correction for the selection intensity for the bulldam selection step was done as follows for Sweden. In the real Swedish population, 3700 potential bulldams are selected out of 160 000 dams in the commercial tier, corresponding to 2.5% selected dams. We know that 105 progeny tested bulls (for 140 daughters) are needed and assume that they originate from the 2.5% selected dams. Therefore, 4500 (=105/2.5%) dams will be necessary each generation, i.e. approximately 1500/yr for three years. In order to get 1500 female offspring/yr, we need to select 3000 commercial dams, which are mated to the 25 best bulls from the nucleus. NUCLEUS 105 mal /yr No. sires by OC 105 mal/yr 315 fem/ year 105 dams/yr x offspring 105 fem/yr COMMERCIAL 25 sires/yr 1500 mal/yr x 4500 fem/yr 3000 dams/yr offspring 1500 fem/yr 10 Finland: The correction for the selection intensity for the bulldam selection step was done as follows for Finland. In the real Finnish population, 3300 potential bulldams are selected out of 210 000 dams in the commercial tier, corresponding to 1.6% selected dams. We know that 120 progeny tested bulls (with 200 daughters) are needed and assume that they originate from the 1.6% selected dams. Therefore, 7635 (=120/1.6%) dams will be necessary each generation, i.e. approximately 2550/yr for three years. In order to get 2550 female offspring/yr, we need to select 5100 commercial dams, which are mated to the 25 best bulls from the nucleus. NUCLEUS 120 mal /yr No. sires by OC 120 mal/yr 360 fem/ year 120 dams/yr x offspring 120 fem/yr COMMERCIAL 25 sires/yr 2550 mal/yr x 7635 fem/yr 5100 dams/yr offspring 2550 fem/yr 11 Traits to select for We assumed that each country selected for two traits: kg protein (kgP) and mastitis (mast). These traits were seen as different traits in each country, such that 2 traits x 3 countries = 6 traits in total were considered. We assumed that both the nucleus and the commercial tier had the same (co)variances. Genetic (G) and error (E) (co)variances are given in Figure 1. The within-country variances were given by the breeding organisation of the countries. The correlation between countries was considered to be 0.9 for kgP and 0.7 for mast. The correlation between kgP in one country and mast in another country was 0.7 * (average correlation over the two countries). In the simulation, genetic variances were standardised to 1, such that all genetic gains are expressed in genetic standard deviation units. Figure 1. Genetic (a) and error (b) (co)variances between traits, kg Protein (kgP) and mastitis (mast), in Norway (N), Sweden (S) and Finland (F). (a) N kgP N mast S kgP S mast F kgP F mast 264 .359 .005 294 .363 .171 .0025 187 .298 .129 .001 405 .291 232 .156 .0025 .177 165 .001 .180 .0008 (b) N kgP N mast S kgP S mast F kgP F mast 674 .687 .152 902 .960 .0845 350 .824 .056 The two traits, kgP and mast, were recorded on females at age two in both the nucleus and commercial tiers. 125 males from Norway, 105 males from Sweden and 120 males from Finland were progeny tested for kgP and mast each year. This data became available at age 5, such that progeny from these sires were born when they were 6 years old. considered as the maximum acceptable rate of inbreeding in ongoing breeding programs (e.g. Goddard, 1992). Results are based on averages over 50 replicated schemes and the average of the last 10 years of selection is given. Economic weights The estimated breeding values of the candidates for selection were calculated using a multitrait across country animal model. EBVs of the two traits were weighted according to the relative economic weight (v) of the trait (kgP and mast) in each country as EBV= vkgP * EBVkgP + vmast * EBVmast for each individual for Selection strategies 1 and 2. 12 For strategies 1 and 2, the economic weights were (in genetic standard deviations): Norway kgP: 0.5= vkgP(N) mast: 0.5 = vmast(N) (1) (2) Sweden kgP: 0.7 = vkgP(S) mast: 0.3 = vmast(S) (3) (4) Finland kgP: 0.7 = vkgP(F) mast: 0.3 = vmast(F) (5) (6) For strategy 3, where selection was over the three countries for one breeding goal, the economical weights were also weighted by the population sizes of each country. The across-country economical weight for kgP and mast are Norway kgP: 0.5*270000/640000 = 0.210 = v’kgP(N) mast: 0.5*270000/640000 = 0.210 = v’mast(N) (7) (8) Sweden kgP: 0.7*160000/640000 = 0.175 = v’kgP(S) mast: 0.3*160000/640000 = 0.075 = v’mast(S) (9) (10) Finland kgP: 0.7*210000/640000 = 0.230 = v’kgP(F) mast: 0.3*210000/640000 = 0.098 = v’mast(F) (11) (12) For strategy 3, EBV= v’kgP(N) * EBVkgP(N) + v’mast(N) * EBVmast(N) + v’kgP(S) * EBVkgP(S) + v’mast(S) * EBVmast(S) + v’kgP(F) * EBVkgP(F) + v’mast(F) * EBVmast(F). Hence, this total breeding goal will for a large part be selecting for the Norwegian breeding goal. This is because Norway has the largest population size and economic values are expressed per genetic standard deviation. The within-country profit was calculated as the total sum of the resulting genetic gain times the economic weights (1-6) for the two traits of each country. The across-country profit was calculated as the sum of the resulting genetic gain times the scaled economic weights (7-12) for the two traits of each country. Selection and mating strategies Dams of both the nucleus and commercial tiers were selected using truncation selection. Within the nucleus tier, sires were selected using optimum contribution selection, where the contributions were adapted to the already selected dams. We assumed that one selection round corresponds to one year. Truncation selection For the paths that used truncation selection, i.e. selection of nucleus dams, selection of commercial sires and dams, the number of individuals with the highest EBVs were selected each year. Optimum contribution selection Optimum contribution selection for overlapping generations was applied for the selection of nucleus bulls as proposed by Meuwissen & Sonesson (1998). This method maximises the genetic level of next generation of animals under three restrictions. The first restriction is on the contribution per sex, i.e. each sex must contribute 50% of the genes to the next generation. The second restriction is on the increase of 13 the average relationship of the selected parents, i.e. on the rate of inbreeding of the progeny. The third restriction is that the females are selected as defined by the truncation selection. For populations with overlapping generations, there are animals of different age-classes at year t from which parents are selected and the different age-classes get different weights, which equal their long-term contributions and which are derived from the gene-flow theory of Hill (1974). The weights indicate how much ageclasses have to contribute in the future. The algorithm iterates on these weights such that the contribution of each individual selection candidate and the contributions of each age-class, are optimised. An addition to the method of Meuwissen and Sonesson (1998) in this study is that these weights were also optimised over tiers by calculating the longterm contributions of tier*age-class. For strategies 2 and 3, since candidates came from all countries and each country had two tiers, there were six tiers in total to be considered. The output from the selection method is a vector of genetic contributions for each selection candidate. These genetic contributions are translated to the number of progeny a certain individual should get. Hence, the number of selected sires in the nucleus is not fixed, but optimised with this algorithm. Mating Random mating was applied. A progeny got a sire or dam assigned by random sampling with sampling probabilities following the optimal contributions of the selection candidates. This vector of genetic contributions was translated to the number of matings each candidate should get by multiplying it by the desired number of male and female progeny of the scheme. In the nucleus tier, a mating pair always got two progeny, one female and one male. In the commercial tier, the male progeny was neglected, because we assumed that all sires came from the nucleus tier. Only female progeny were therefore accounted for in the commercial tier. 14 2.4.2 Results Profit within country Norway: In the nucleus tier, 3, where all countries select for the same overall breeding goal than for strategy 1, where each country selects for its own breeding goal (Table 1, Figure 2). This resulted in a total profit of 0.163 for strategy 3 and 0.153 for strategy 1, i.e. an increase in total profit of 6.5%. This is probably because the overall breeding goal of strategy 3 consists for a large part (42%) of the original Norwegian breeding goal, due to the large size of the Norwegian population, and then it becomes attractive to select from the Swedish (18%) and Finnish (22%) nuclea and Finnish commercial tier (18%) (Table 2). Overall, Norway had the lowest profit (Tables 1, 3, and 5), probably because there is a high relative economic weight on mastitis, which is a lowly heritable trait and more difficult to improve than kg P, even though the Norwegian cow population is the largest of the three countries. Strategies 2 and 3 mainly had a large advantage for strategy 1 (Figure 2). This is because initial selection response is larger in the large unrelated population under strategies 2 and 3, whereas it becomes similar to response under strategy 1 as relationships between populations build up. This effect is not important when calculating profit, 15 Table 1. Number of male (Malcand) and female (Femcand) candidates and selected sires (Malsel) and dams (Femsel using strategies 1-3 * Strategy Malcand Femcand Malsel Femsel MalL FemL 1 2 3 379 1058 1060 3525 16972 16957 17.8 4.5 9.3 125 125 125 6.62 7.18 7.26 2.78 2.96 3.17 F 0.001 0.001 0.001 G kgP G mast Profit (sd) (sd) 0.172 0.193 0.207 0.133 0.120 0.119 0.153 0.156 0.163 * Strategy 1. Selection within each of the three countries for each of its breeding goals Strategy 2. Selection over the three countries for each of its breeding goals Strategy 3. Selection over the three countries for one overall breeding goal Table 2. Use of genes from the nucleus (Nucl) or commercial (Com) tiers of Norway (N), Sweden (S) and Finland (F) in the nucleus of Norway using strategies 1-3* Strategy %NuclN %ComN %NuclS %ComS %NuclF %ComF 1 2 3 0.72 0.52 0.30 0.28 0.14 0.05 x 0.08 0.18 x 0.05 0.08 x 0.10 0.22 x 0.11 0.18 * Strategy 1. Selection within each of the three countries for each of its breeding goals Strategy 2. Selection over the three countries for each of its breeding goals Strategy 3. Selection over the three countries for one overall breeding goal 16 Genetic response (sd) 6 5 1 kgP 4 1 mast 2 kgP 3 2 mast 3 kgP 2 3 mast 1 25 23 21 19 17 15 13 11 9 7 5 3 1 0 Year Figure 2. Genetic response in the nucleus of Norway for kg protein (kg P) and mastitis (mast) using strategies 1-3 Sweden In the Swedish nucleus, the effect of the three strategies was very small (Table 3). For strategy 1, profit was 0.170, for strategy 2, profit was 0.168 and for strategy 3, profit was 0.172. Sweden has the smallest scheme, and will therefore gain from collaborating with especially Finland (Table 4) that has a similar breeding goal. The contribution from the nuclea of the different countries is approximately the same for all three countries for strategy 3 (Tables 2, 4 and 6). Hence, the same animals are probably, to a large extent, selected in all countries, because they select for the same breeding goal. 17 Table 3. Number of male (Malcand) and female (Femcand) candidates and selected sires (Malsel) and dams (Femsel), male (MalL) and female (FemL) generation interval, rate of inbreeding ( using strategies 1-3 * Strategy Malcand Femcand Malsel Femsel MalL FemL 1 2 3 317 1058 1060 5046 16972 16957 14.5 4.7 7.8 105 105 105 6.4 7.2 7.3 3.0 3.1 3.2 F 0.001 0.001 0.001 G kgP G mast Profit (sd) (sd) 0.200 0.200 0.204 0.100 0.092 0.096 0.170 0.168 0.172 * Strategy 1. Selection within each of the three countries for each of its breeding goals Strategy 2. Selection over the three countries for each of its breeding goals Strategy 3. Selection over the three countries for one overall breeding goal Table 4. Use of genes from the nucleus (Nucl) and commercial (Com) tiers of Norway (N), Sweden (S) and Finland (F) in the nucleus of Sweden using strategies 1-3* Strategy %NuclN %ComN %NuclS %ComS %NuclF %ComF 1 2 3 x 0.08 0.28 x 0.02 0.05 0.69 0.44 0.18 0.31 0.18 0.08 x 0.17 0.24 x 0.11 0.18 * Strategy 1. Selection within each of the three countries for each of its breeding goals Strategy 2. Selection over the three countries for each of its breeding goals Strategy 3. Selection over the three countries for one overall breeding goal 18 Genetic response (sd) 6 5 1 kgP 4 1 mast 2 kgP 3 2 mast 3 kgP 2 3 mast 1 25 23 21 19 17 15 13 11 9 7 5 3 1 0 Year Figure 3. Genetic response in the nucleus of Sweden for kg protein (kg P) and mastitis (mast) using strategies 1-3. Finland: The Finnish nucleus had a large increase in profit for strategy 1 (0.180) compared to strategy 2 (0.167) and strategy 3 (0.164) (Table 5). Hence, Finland will loose most in the co-operation with Norway and Sweden. This is probably because they have a high relative economic weight on kgP (0.7), which is highly heritable and because the number of cows is larger (210 000) than Sweden (160 000) that has the same relative economic weights on the two traits. Another reason for the low profit of strategies 2 and 3 for Finland is that bulls are selected more intensely than in strategy 1 (number of selected sires is reduced and the number of candidates is increased), which might have reduced the genetic variance in the next generation more than when Finland selected on its own in strategy 1. Table 5. Number of male (Malcand) and female (Femcand) candidates and selected sires (Malsel) and dams (Femsel), male (MalL) and female (FemL) generation interval, rate of inbreeding ( F), genetic gain ( G) for kg protein (kg P) and mastitis (mast) and total profit in the nucleus of Finland using strategies 1-3 * Strategy Malcand Femcand Malsel Femsel MalL FemL 1 2 3 365 1058 1060 8377 16972 16957 15.8 6.1 9.8 120 120 120 6.4 7.2 7.7 3.1 3.2 3.2 F 0.001 0.001 0.001 G kgP G mast Profit (sd) (sd) 0.205 0.194 0.188 0.121 0.106 0.108 0.180 0.167 0.164 * 19 Strategy 1. Selection within each of the three countries for each of its breeding goals Strategy 2. Selection over the three countries for each of its breeding goals Strategy 3. Selection over the three countries for one overall breeding goal Table 6. Use of genes from the nucleus (Nucl) or commercial (Com) tiers of Norway (N), Sweden (S) and Finland (F) in the nucleus of Finland using strategies 1-3* Strategy %NuclN %ComN %NuclS %ComS %NuclF %ComF 1 2 3 x 0.13 0.29 x 0.02 0.05 x 0.11 0.17 x 0.04 0.08 0.64 0.40 0.24 0.36 0.31 0.18 * Strategy 1. Selection within each of the three countries for each of its breeding goals Strategy 2. Selection over the three countries for each of its breeding goals Strategy 3. Selection over the three countries for one overall breeding goal Genetic response (sd) 6 5 1 kgP 4 1 mast 2 kgP 3 2 mast 3 kgP 2 3 mast 1 25 23 21 19 17 15 13 11 9 7 5 3 1 0 Year Figure 4. Genetic response in the nucleus of Finland for kg protein (kg P) and mastitis (mast) using strategies 1-3 20 Profit over countries The total profit over countries was very similar for the three strategies (Table 7). The genetic gain was about the same for all traits within a country, except for Norway that had higher genetic gain for kg P for strategies 2 and 3 than for strategy 1. Table 7. Profit for kg protein (kg P) and mastitis (mast) for Norway (N), Sweden (S) and Finland (F) using strategies 1-3 * Strategy 1 2 3 kgP N mast N kgP S Profit mast S kgP F mast F Total profit 0.036 0.041 0.043 0.028 0.025 0.025 0.035 0.035 0.036 0.007 0.007 0.007 0.047 0.045 0.043 0.012 0.010 0.011 0.166 0.163 0.165 * Strategy 1. Selection within each of the three countries for each of its breeding goals Strategy 2. Selection over the three countries for each of its breeding goals Strategy 3. Selection over the three countries for one overall breeding goal 21 2.4.3 What is the exchange of red cattle bulls between Finland, Sweden and Norway today? In order to get an overview of the exchange of bulls of today, the three breeding organisations gave a data-file on the use of foreign bulls. We required that sires should have their first daughter group of at least 20 daughters born 1990 or later and with EBV for kgP in both countries. Finland had used 16 bulls from Sweden and 1 from Norway. Sweden had used 36 bulls from Finland and 9 from Norway. Norway had used 15 bulls from Sweden and 4 from Finland. 2.4.4 Correlations between the breeding goals in the three countries Correlation between breeding goals is one critical parameter to compare of single/multiple objectives. When using the real values of the genetic (co)variances matrix (G) (Figure 1) and the relative economic weights (1-6) the following correlations between the breeding goals were calculated as rT1T2 = v1´Gv2/ (v1´Gv1v2´Gv2)1/2 , where v1 contains the relative economic weights of country 1 and v2 contains the relative economic weights of country 2. The resulting correlations were: Norway-Sweden: 0.798 Norway-Finland: 0.800 Sweden-Finland: 0.832 Hence, all correlations are around 0.8, which is the ‘breakeven point’ for when there is any advantage of co-operating for two populations (see literature review). The highest correlation is between Sweden and Finland, probably because they use the same relative economic weights for kgP and mast. 2.4.5 Final conclusions The exchange of bulls between Norway, Sweden and Finland is small today. Especially the exchange with Norway is small.When using the real values of the variances with the artificial relative economic weights, the correlation between the breeding goals of Sweden and Finland is highest (0.832), whereas the correlation between Norway and Sweden or Finland was around 0.800. These correlations are at the ‘breakeven point’ for when there is any advantage of co-operating for two populations, higher correlations indicating that the breeding goals are the same and thus that the populations would gain in co-operating. There was only a small effect of the three different strategies on the profit of the schemes. However, these results will be sensitive to some assumptions made in the study, e.g. the number of selected sires, the relative economic weights, and the covariances used. At the start of a co-operation, a genetic lift can be expected, because the populations are still unrelated and selection intensities can be high. When co-operating, a too high selection intensity might lead to reduced genetic variances in the next generation, which in turn will reduce genetic gains. 22 2.5 References Banos, G. and Smith, C. (1991) Selecting bulls across countries to maximise genetic improovemnt in cattle. J. Anim. Breed. Genet. 108: 174-181 Bulmer M.G., The mathematical theory of quantitative genetics. Clarendon Press, Oxford, 1985. Del Bosque Gonzalez, A.S. and Kinghorn, B.P. (1990) Implications of differing selection objectives within open nucleus breeding schemes. In. Proc. Australian Association of Animal Breeding and Genetics, Hamilton and Palmerston North, vol. 8, pp. 95-102 Goddard, M.E. (1992) Optimal effective size for the global population of black and white dairy cattle. J. Dairy Sci. 75: 2902-2911 Hill, W.G. (1974) Prediction and evaluation of response to selection with overlapping generations. Anim. Prod. 18: 117-139. Howarth, J.M. (2001) Selection strategies to exploit diversity between economic breeding objectives. PhD thesis, The University of New England, Australia Meuwissen, T.H.E. and Sonesson, A.K. (1998) Maximizing the response of selection with a predefined rate of inbreeding- overlapping generations. J. Anim. Sci. 76: 2575-2583 Meuwissen, T.H.E. (1998) Risk management and the definition of breeidng objectives. In. Proc. 6 th WCGALP, vol XXV, pp. 347-354 Ollivier, L., Guéblez, R., Webb, A.J. and van der Steen, H.A.M. (1990) breeidng goals for nationallya nd internationally operating pig breeder’s organisations. In. Proc 4th WCGALP, vol. XV, pp. 383-394. Robertson, A. (1959) The sampling variance of the genetic correlation coefficient. Biometrics 15: 469-485 Smith, C. (1986) Variety of breeding stocks for the production-marketing range and for flexibility and uncertainty. In. Proc. 3rd WCGALP, Lincoln, vol X, pp. 14-20 Smith, C. and Banos, G. (1991) Selection within and across populations in livestock improvement. J. Anim. Sci. 69: 2387-2394 23 3.0 Genotype by environment interaction Erling Strandberg1 and Rebecka Kolmodin2, 1Dept. of Animal Breeding and Genetics, SLU, 75007 Uppsala, 2Swedish Red Cattle Breed Organisation, Örnsro, 532 94 Skara 3.1 Introduction The ability to respond to changes in the environment is a vital characteristic of all organisms. This ability is sometimes called phenotypic plasticity or environmental sensitivity. Differences in environmental sensitivity between individuals result in genotype by environment interaction (GxE). Strictly speaking GxE means that the difference between the phenotypic values of two genotypes is not the same in two environments. If the difference changes sign between environments, there is reranking of genotypes. If the difference changes in size only, there is a scaling effect (Falconer and Mackay, 1996; Kolmodin, 2003). In Figure 1 these two situations are exemplified. 30 25 20 15 A B C 10 5 Phenotypic value Phenotypic value 30 25 20 15 A B C 10 5 1 2 1 Environment 2 Environment Figure 1. Example of GxE resulting in scaling (left) and re-ranking (right) for three genotypes. 3.2 Methods to study genotype by environment interaction There are basically three different methods to describe the extent of GxE. For all methods, observations on the same or related individuals in two or more different environments are needed to study GxE. The common use of artificial insemination in dairy cattle makes it possible to compare the performance of daughters of the same sires in different environments (de Jong, 1995). 1) Interaction term model. In the first method the phenotypic value of an individual is simply described as the sum of the genotypic value, the environmental value and the residual. The environmental value E could e.g. be classification into herds, or herd production classes. P = G + E + e [1] When interaction between genotype and environment exists an interaction component, GxE, is added to the equation: P = G + E + GxE + e [2] The phenotypic variance ( P2 ) of the observed phenotypes (P) can be derived from [2] as: 2 P2 G2 E2 GE e2 [3] 24 assuming all covariances being zero. 2) Multiple trait model. The second method used to describe GxE is based on phenotypic values in different environments and genetic correlations (rg) between these. The phenotypic expression in the two environments is seen as two separate traits and rg can be studied to see whether GxE exists. When rg between the phenotypic values of the same genotype expressed in different environments is high, the phenotypic expression is considered as the same trait in the different environments (Falconer and Mackay, 1996). In other words, if rg between the phenotypic expressions of the trait in two different environments is equal or close to 1 there is no GxE (Robertson, 1959). When rg is low, the phenotypic expressions in the different environments are not the same trait and this is an indication of GxE (Falconer & Mackay, 1996). The genetic correlation (rg) can be estimated using a multiple trait analysis based on grouping herds with similar production environments to clusters and treating the observation from the different clusters as separate traits. GxE is indicated by low rg between clusters (Falconer and Mackay, 1996). 3) Reaction norm model. When the production environment can be described as a continuous variable, a third method called the reaction norm model, is possible to use (de Jong, 1995). The phenotypic expression of a genotype as a function of the environment is described by the reaction norm (Kolmodin, 2003). A given difference of an environment can have a greater effect on one genotype than on another (Falconer and Mackay, 1996). The reaction norm model has an advantage in describing the environment on a continuous scale that the multiple trait model does not possess (Fikse et al., 2003). 3.3 Does GxE exist? For dairy cattle, GxE has been studied for various environmental factors. Between herd production levels, feeding regimes or management systems within a country, or a group of neighbouring countries, there is seldom reranking of genotypes (Cromie, 1999; Kolmodin et al., 2002; Boettcher et al., 2003; Calus and Veerkamp, 2003). Between countries or regions that differ considerably, e.g. in climate or management system, reranking of genotypes is more common. For example, the genetic correlation between the same milk production trait evaluated in any country in Western Europe, the USA, or Canada (the northern hemisphere group) is high (0.85-0.9), while the correlation between the northern hemisphere group and New Zealand and Australia is lower (0.750.84)(Emanuelson et al., 1999), indicating that reranking occurs. Low genetic correlations have also been estimated between milk yield evaluated in Mexico and the US (0.63) (Cienfuegos-Rivas et al., 1999), milk yield in the UK and Kenya (0.49) (Ojango and Pollott, 2002), and longevity in Canada and New Zealand (-0.07—0.21) (Mwansa and Peterson, 1998). It was also found that there was GxE between environments with different heat-humidity indexes, with a lowest correlation of around 0.6 between low and high stress environments (Ravagnolo and Misztal, 2000). The phenotypic and genetic variances between animals are often smaller in low than in high yield environments, i.e. there is a scaling effect of GxE. As genetic progress is a function of the genetic variance (Falconer and Mackay, 1996), the expected response to selection is smaller in low than in high yield environments. For example, the expected response in Kenya, a low yield environment, to selection based on UK breeding values is only 44 % of the expected response in the UK, a high yield environment (Ojango and Pollott, 2002). Thus, investments in high merit semen from bulls evaluated in high yield environments may not pay off for farmers in low yield environments. 25 3.4 Does GxE exist in Nordic countries? In the following we will summarize results on GxE in Nordic dairy cattle. Some results are based on estimated genetic correlations between countries as currently used by Interbull (www.interbull.org, November, 2004), and some results are based on research done on Nordic or Swedish dairy cattle. Results for Nordic Red cattle are given, unless otherwise stated. Milk production traits The genetic correlations between countries for protein production are 0.86-0.92 (Interbull) with similar values for amount of milk and fat as well. Kolmodin et al. (2002), using a reaction norm approach on Nordic Red cattle, showed no GxE for protein yield between average and high protein yielding herds but a rank correlation of 0.93 between low and average yielding herds.They also found a rank correlation of down to 0.9 between average and very large herds, but not between herds in the normal range of size. Udder health Genetic correlations used by Interbull for somatic cell count (SCC) were around 0.9-0.95 between the three countries (for Norway, no information was available as SCC is not used for genetic evaluation). For clinical mastitis, the values were somewhat lower (0.84-0.91). Correlations with Norway were even lower, 0.68-0.75. Jansson (2004) found GxE for SCC using both a multiple-trait approach and a reaction norm approach, between herds with low and high average SCC levels. In the multiple-trait approach, the genetic correlation was 0.8-0.84 between low and high quartile SCC herds. There was also an indication of GxE for mastitis between low and high SCC herds (0.89, n.s.). Fertility Svendsen et al. (2001) estimated genetic correlations between countries for non-return rate (number of inseminations for Sweden) of 0.76-0.95, the highest between Denmark and Finland and the lowest between Finland and Sweden. Kolmodin et al. (2002) found no GxE for calving to last insemination (CLI) between herds with average and long CLI but some indication for herds with short average CLI. Calving performance For the direct (sire of calf) effect of calving ease, genetic correlations were above 0.88 among all countries except Norway, for which correlations with other countries ranged from 0.38-0.76. A similar situation existed for stillbirths, where correlations were 0.36-0.57 when Norway was involved but higher (0.75-0.93) among other countries (Svendsen et al., 2001). For the maternal effect (maternal grandsire effect) of calving ease correlations were above 0.75 among all countries, including Norway. For stillbirths correlations were generally between 0.65 and 0.8, but with a high exception of 0.95 between Denmark and Finland and a low value of 0.44 between Sweden and Norway (Svendsen et al., 2001). Longevity Petersson et al. (2005) found a genetic correlation for longevity between herds with short or long average longevityof around 0.74. They also found an indication of GxE for longevity between herds with average or low 305-day protein yield (correlation about 0.77) but less so between herds with 26 average and high yielding herds (0.9). For Holstein the genetic correlation between Denmark and Finland for direct longevity (i.e. not using indicator traits of longevity but only length of productive life by itself) 0.9. Summary For production traits there seems to be little or no GxE both between average levels in the Nordic countries as indicated by the genetic correlation between countries, and between various herd environments with respect to production level. Therefore, using the same breeding value for all environments would not be expected to give too wrong results. For SCC the correlations also indicate little GxE between the three countries where this information was available. For clinical mastitis the correlations were slightly lower, perhaps owing to the lower accuracy in estimating breeding values for this trait. For these traits there may, however, be withincountry GxE that is not visible on the across-country scale. For fertility traits, there is currently work on harmonization of trait definitions which may help in increasing the correlation among countries somewhat. For calving performance traits, correlations were not very strong, except possibly for direct calving ease among Denmark, Finland, and Sweden. For the maternal effects it is not quite clear whether the breeding values are the maternal grandsire solution unadjusted or adjusted for direct genetic effects. For longevity rather little is known across countries but there seems to be some GxE within the Swedish population with respect to herd average longevity and production. The genetic correlations may be underestimated owing to weak genetic connectedness between countries. This may be especially so for correlations involving Norway. Other reasons for underestimation can be that different definitions, edits, and genetic evaluation methods are used in the different countries. The on-going work in the joint Nordic genetic evaluation (NAV) regarding harmonization can be expected to decrease these effects in the future. However, the current assumption in NAV of unity genetic correlation between countries will cover up any true GxE than might exist. 3.5 Consequences for selection The simplest consequence of GxE is that we select the wrong animals and that we miss out in performance in those environments not similar to that which the genetic evaluation assumes. How much one would lose is a complex question, considering that one would have to compare two (or more) breeding programs, with corresponding decrease in population size, incresed inbreeding and risk in outcome, with one breeding program applied to a larger population. Some of these issues are addressed in the study by Sonesson in this publication. One question that often arises is how much reranking occurs with a given genetic correlation. In Figure 2 some examples are given where a certain correlation between true and predicted breeding values is assumed (or between two breeding values predicted from two methods or environments). We rank on “true” breeding value and select the top fraction. Then we rank on predicted breeding value and select the top fraction. The graph shows how large a proportion of those in the top fraction for predicted breeding values are also in the top fraction for the true breeding values. If this 27 proportion is 100% then we would select exactly the same individuals in the top fraction on both criteria. Figure 2 shows that if the genetic correlation is very high, we more or less correctly select the top fraction, however, even with a correlation of 0.95 only about 80% of the top 10% are correctly selected. With a correlation of 0.9 this proportion decreases to below 70%. With larger top fraction the correct proportion increases (and obviously goes to unity for all correlation as the top fraction goes to 100%). Prop. correctly among top fraction . 100% 90% 80% 70% Correlation 60% 0.99 0.95 0.9 0.8 0.7 50% 40% 10 20 30 40 50 Top fraction, % Figure 2. Proportion correctly ranked on predicted breeding value for various correlations between true and predicted breeding values for various top fractions. In total 100 bulls available, 1000 replicates. 3.6 References Boettcher, P., J. Fatehi, and M. Schutz. 2003. Genotype x environment interactions in conventional versus pasture-based dairies in Canada. J. Dairy Sci. 86:383-389. Calus, M. and R. F. Veerkamp. 2003. Estimation of environmental sensitivity of genetic merit for milk production traits using a random regression model. J. Dairy Sci. 86:3756-3764. Cienfuegos-Rivas, E. G., P. A. Oltenacu, R. W. Blake, S. Schwager, H. Castillo-Juarez, and F. Ruiz. 1999. Interaction between milk yield of Holstein cows in Mexico and the United States. J. Dairy Sci. 82:2218-2223. Cromie, A. 1999. Genotype by environment interaction for milk production traits in Holstein Friesian dairy cattle in Ireland. PhD-thesis, Queens University of Belfast, Ireland, de Jong, G. 1995. Phenotypic plasticity as a product of selection in a variable environment. Am. Nat. 145:493-512. Emanuelson, U., G. Banos, and J. Philipsson. 1999. Interbull Centre Report. Interbull Bulletin 22:16. 28 Falconer, D. S. and T. F. C. Mackay. 1996. Introduction to quantitative genetics, 4th ed. Longman Group, Essex. Fikse, W. F., R. Rekaya, and K. Weigel. 2003. Assessment of environmental descriptors for studying genotype by environment interaction. Livestock Prod. Sci. 82:223-231. Jansson, K. 2004. Genotype by environment interaction for udder health in Swedish Holstein cows. MSc-thesis, Dept of Animal Breeding and Genetics, SLU, Kolmodin, R. 2003. Reaction norms for the study of genotype by environment interaction in animal breeding. PhD-thesis, Agraria 437, Swedish University of Agricultural Sciences, Uppsala. Kolmodin, R., E. Strandberg, P. Madsen, J. Jensen, and H. Jorjani. 2002. Genotype by environment interaction in Nordic dairy cattle studied by use of reaction norms. Acta Agric. Scand. Section A (Animals) 52:11-24. Mwansa, P. and R. Peterson. 1998. Estimates of GxE effects for longevity among daughters of Canadian and New Zealand sires in Canadian and New Zealand dairy herds. Interbull Bulletin 17:110-114. Ojango, J. and G. Pollott. 2002. The relationship between Holstein bull breeding values for milk yield derived in both the UK and Kenya. Livestock Prod. Sci. 74:1-12. Petersson, K.-J., R. Kolmodin, and E. Strandberg. 2005. Genotype by Environment Interaction for Productive Life in Swedish Red and White Dairy Cattle. Acta Agric. Scand. Section A (Animals) submitted. Ravagnolo, O. and I. Misztal. 2000. Genetic component of heat stress in dairy cattle, parameter estimation. J. Dairy Sci. 83:2126-2130. Robertson, A. 1959. The sampling variance of the genetic correlation coefficient. Biometrics 15:469-485. Svendsen, M., T. Mark, U. Nielsen, J. Pösö, and M. Gundel. 2001. Genetic relationships among functional traits in the Nordic Ayrshire populations. Proceeding of the 2001 Interbull meeting, August 30-31, Budapest, Hungary. Interbull Bulletin 27:60-63. 29 4.0 Breeding objectives and optimum use of sires in the Nordic red populations Erling Sehested, GENO 4.1 Introduction Genetic gain in traits of interest is a function of breeding objectives, genetic parameters and design of breeding program. In general, genetic parameters can be regarded as constants, and thereby is not a variable component in the process of optimising the breeding plan. Definition of the breeding objective in terms of aggregate genotype (economic values of the traits) has traditionally been given much attention. Several approaches have been proposed. Examples are using partial derivatives of profit functions, desired gains and multiple regression analysis of observed profit from herd data. When the breeding objective is defined, the design of the breeding plan should be optimised with respect to this objective. The term to optimise is maximum gain in aggregate genotype with minimum costs. The most important aspects of a breeding plan are number of test bulls per year, fraction of cows inseminated with test bulls, number of elite bulls, number of bull sires, selection intensity of bull dams and fraction of bull dams being heifers. Given the number of cows in the population, progeny group size is a function of number of test bulls per year and fraction of cows inseminated with test bulls. The purpose of this study was to investigate effects of these two variables on genetic gain in different sets of aggregate genotype. 4.2 Methods Breeding goals and characteristics of the breeding populations of the 4 Nordic red populations were approximately as shown in tables 1 and 2 around 2000. These assumptions together with assumed genetic parameters as shown in table 3 were used in a deterministic simulation to compute expected gain in the aggregate genotype. Selection intensities were corrected for deviations from normal distribution due to selection in finite (small) sample sizes. Eventually expected genetic changes with varying levels of number of test bulls per year and fraction of cows inseminated with test bulls were computed according to: T iSS SS iSD SD i DS DS i DD DD LSS LSD LDS LDD where: i SS SD DS DD L = selection intensity = Standard deviation on breeding values = Bull sire = Cow sire = Bull dam = Cow dam = Generation interval Selection and culling effects of cow-dams were ignored (iDD=0) 4.3 Results and discussion Results for the four populations are shown in figures 1-4. For all populations the distance from current to optimal combination of the two variables is rather large. In general all populations should test more bulls per year and inseminate a larger proportion of the cows with test bulls. Using more test bulls can be expensive. Substantial increase of the proportion of cows inseminated with test bulls will probably not be accepted by the farmers. Therefore it is important to investigate the net effect of changing. For NRF, SRB and RDM this effect is only marginal, from .8 to 1.7% increase in genetic gain. Hence for these breeds a change does not seem to be worthwhile. For FAY, 30 however, moving from current to optimal will give 6.2% increase in genetic gain. On the other hand, increasing number of test bulls per year from 150 to 415 is far from realistic. The optimum combination differs between the populations, NRF and FAY being somewhat extreme while SRB and RDM are intermediate. The breeding goal in NRF is characterized by heavy weights on low heritable traits such as disease resistance and fertility. The optimum design for such breeding goals is to have large progeny groups in order to achieve high accuracies for these traits. In FAY highly heritable traits such as milk production and conformation dominates. Optimum for FAY is therefore a situation with many test bulls and small progeny groups. This gives high selection intensity and acceptable accuracies. Table 1. Alternative breeding goals Trait Kg protein Protein % Mastitis Other diseases Udder Body/leg conf Beef Milking ease Temperament Non-return Calving ease Stillbirth NRF 21 0 21 3 11 6 12 0 4 14 4 4 SRB 31 0 16 5 16 6 7 0 0 12 3,5 3,5 RDM 32 0 13 0 15 6 6 9 2 10 7 0 FAY 41,7 12,5 12,5 0 12,5 0 0 0 0 20,8 0 0 Table 2. Characteristics of breeding population NRF No cows 290 000 % Test bulls 40 No test bulls 125 No elite bulls 10 No bull sires 5 Bull dam selection, % 5 % Heifers as bull dams 34 Progeny group sizes 150-300 SRB 193 000 30 110 5 5 5 34 50-130 RDM 59 000 40 70 5 5 5 34 50-100 FAY 208 000 45 150 15 5 5 34 50-200 Table 3. Assumed genetic parameters Heritability Protein Mastitis Prot. Mast Udder Beef Milk. speed ,20 ,03 ,10 ,15 ,20 -,31 ,17 ,16 ,05 ,09 -,09 -,01 Legs Temp Fert Calv. ease ,10 ,10 ,05 ,05 -,12 ,14 -,10 -,03 ,11 ,05 ,01 -,01 Stillb Other dis. ,05 ,02 ,11 -,25 ,02 ,27 P% ,30 ,24 ,19 31 Udder Beef Milk. Speed Legs Temperament Fertility Calving Ease Stillbirth Other diseases -,05 ,25 ,01 ,15 ,26 ,37 ,03 -,19 ,25 ,06 ,12 -,21 -,02 -,13 ,08 -,07 ,02 -,27 ,14 -,06 -,03 -,07 -,12 -,06 -,13 ,00 ,01 ,61 ,11 ,17 -,20 ,36 -,18 -,04 -,03 ,00 ,01 ,01 -,08 ,00 ,01 ,01 -,11 -,07 ,19 NRF 80 75 70 65 55 50 45 % test bulls 60 40 35 30 25 90 20 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 100 No test bulls 100,8 Figure 1.Optimal design, NRF 32 SRB 80 75 70 65 55 50 45 % test bulls 60 40 35 30 25 70 80 20 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 100 No test bulls 101,4 Figure 2. Optimal design, SRB RDM 80 75 70 65 55 50 45 % test bulls 60 40 35 30 25 50 60 70 100 80 20 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 101,7 No test bulls Figur 3. Optimal design, RDM 33 FAY 80 75 70 65 55 50 45 % test bulls 60 40 35 30 25 100 No test bulls 420 400 380 360 340 320 300 280 260 240 220 200 180 160 140 120 100 20 106,2 Figure 4. Optimal design, FAY 34 5.0 Use of EVA in Dairy Cattle Breeding Schemes Morten Kargo Sørensen, Anders Christian Sørensen, Peer Berg, Danish Institute of Agricultural Sciences, and Søren Borchersen, A. I. Centre Dansire 5.1 Summary EVA is a programme, which among the potential parents of next group of offspring, optimises the number of offspring per breeding animal, after which the programme will suggest a mating scheme for the selected animals. The average breeding value is calculated for the selected matings as well as the average relationship between the matings. The information used is pedigrees going back as far as the pedigree can be traced and estimated breeding values of the parents. The selection of animals depends on how much weight is being attached to avoid future inbreeding. As future inbreeding is depending on the relationship among animals in the next generation EVA analyses the relationship between the selected animals as well as the relationship between the selected animals and the animals used in previous years. Therefore, it is important that the animals previously used in the breeding scheme are included in the analysis i.e. the present young bulls, waiting bulls as well as the planned young bulls (contracts) must be included in the analysis. EVA differs from the computer inseminating schemes in that they only consider the inbreeding of the calf in relation to the insemination of the actual cow or heifer, while EVA considers the total population. 5.2 Inbreeding – Cause and Consequences Because of inbreeding depression, inbreeding weakens genetic improvement in general and genetic improvement of health, fertility and the productivity of the dairy cows in particular. However, in the long term the largest problem of inbreeding is the expected reduction of future genetic gain because of reduced genetic variation. The breeding program including insemination, breeding value estimation and selection of the best animals for breeding has been so successful that we are now facing a problem of large rates of inbreeding if nothing is being done – and that is so even though the breeds are numerous in animals, but they are small in effective number of breeding animals. The increase of inbreeding of the production animals reduces the productivity, health and fertility of the animals. Research results from United States shows that a 1% increase of inbreeding within Holstein Friesian resulted in a loss of 37 kg milk, 1.2 kg fat, 1.2 kg protein per lactation, an increasing age at first lactation of 0.4 days and a prolonged calving interval of 0.3 days as well as 6 days shorter productive life (Smith et al., 1998). Danish results show an increased incidence of mastitis with inbreeding (Sørensen et al. 2005a). The conclusion of all research indicates that inbreeding reduces the economic results in a herd. At breed level inbreeding is expected to reduce the genetic variation and thus the genetic gain in the long term. Calculations indicate that genetic gain per time unit in an efficient breeding scheme is reduced by about 20% over a period of 20 years only because of the increased inbreeding, if inbreeding has not been taken into account (Sørensen,1999). Therefore, inbreeding has now gained more attention in the breeding program of the dairy breeds than previously. 35 For the time being the degree of inbreeding is not alarming. With the existing pedigree information in the Danish cattle databases the degree of inbreeding in the potential Danish Holstein and Jersey breeding animals is calculated (Sørensen et al., 2005b) to be in the interval of 3-4%, which is a little lower than for the corresponding breeds in the United States, while it is about 1% for RDM. The rate of the inbreeding is, however, alarming - >1% generation. That is why Dansire is giving this problem full attention. The increase of inbreeding can be kept under control by assuring that the future animals are not too closely related. The control of relationship between members of the breeds in the long term demands a consideration of the relationship between the young bulls produced for testing and those used in the previous years. The EVA Programme is capable of weighing up the genetic gain and future increase of inbreeding. It has now been taken into use in the dairy cattle association, Dansire. In the long term it will likely mean that more sires of sons and bull dams are used. 5.2.1 Balance between Genetic Gain and Relationship B A Inbre e ding / Re lationship Figure 1 Schematic connection between average relationship and genetic gain A: Only focus on genetic gain B: Focus on both genetic gain and the average relationship in next generation. B A Total genetic gain Genetic gain If the focus is only on genetic gain when selecting sires of sons and bull dams it will lead to a major genetic gain in the next generation and also an increased relationship between the selected young bulls. A closer relationship will result in a higher degree of inbreeding in future generations. If only a smaller increase of degree of relationship can be accepted among the selected bulls the genetic gain will be reduced in the next generation. Figure 1 shows a schematic relation between the relationship and the maximum genetic gain in the next generation. Numbe r of ge ne ration Figure 2 The genetic gain of the two different selection strategies in Figure 1 36 In the short term a little of the genetic gain is lost when considering the relationship in the next generation. e.g. by choosing strategy B in stead of strategy A. By choosing strategy B more sires of sons will be used leading to a reduced increase in average relationships, whereby a higher degree of genetic variation is maintained. In the long term it will pay off to choose strategy B rather than strategy A, as can be seen from Figure 2. Add to this that a population following strategy B will suffer from a smaller degree of inbreeding depression than the population following strategy A. The time where the two lines are intersecting is dependent on where on the curve in Figure 1 B is placed – the longer to the left B has been placed the later the curves will intersect. That means that the more weight is being attached to the relationship, the longer time will pass to obtain the genetic gain, which is lost in the short term. 5.2.2 Programmes to Handle Genetic Gain vs. Inbreeding The above problems are not unique for cattle breeding – but can be found for all species. For species with shorter generation intervals the inbreeding increase will even find place at a more rapid pace. Therefore, there has been a major focus on developing programmes adressing this issue. That is also the case in Denmark where Peer Berg has developed the computer programme EVA (EVolutionary Algorithms). Also a Dutch developed programme (GenCont) for handling these problems developed by Theo Meuwissen, now Norway, is available. EVA is based on the method described by Grundy et al. (2000), where GenCont is based on the method described by Meuwissen & Sonesson (1998). EVA is different because of its ability to limit the inbreeding increase both per year and per generation. In a comparison of the methods (Sonesson et al. 2000) small deviations were found among the methods, and it was concluded that the method developed by Meuwissen and Sonesson cannot be further developed to limit the inbreeding increase per generation. There are differences in principles of the calculation methods used in the optimisation. In GenCont an iterative method has been used, which can give negative genetic contributions, and therefore it is necessary that restrictions are made iteratively on animals getting negative contributions. In return the method is deterministic, which means that it is quick and presents the same result every time. EVA uses a calculation method, which ensures the genetic contributions to be positive, but as the method is stochastic the result might vary a little depending on convergence of the algorithm. However, this has not turned out to be a problem in practice. In addition EVA suggests a mating list given the optimum genetic contributions based on minimising the degree of inbreeding in the next generation. Both methods are very computer intensive, both in terms of memory and CPU time. There has not been made a direct practical comparison, but it is expected that EVA will be the best to handle large data sets as it is not based on inverting matrices as is the case with GenCont. During the summer EVA has been adapted to a user-friendly Windows based version of the programme. 37 5.2.3 ”Economic” Weighting of Genetic Gain and Relationship EVA is capable of maximising the genetic gain with a control of the inbreeding increase simultaneously. As mentioned above the programme is a further development of methods developed by John Woolliams, Robin Thompson and Brian Grundy, all Edinburgh, where it was shown that both genetic gain and degree of inbreeding can be described as functions of genetic contributions. Optimising genetic contribution for each sire and dam thus allows for a simultaneous consideration of genetic gain and inbreeding. The programme put weight on the average relationship among the selected combinations and the previously used animals in the breeding scheme and to the average breeding value of the selected animals. In principle a variable S is generated S = b1 * average breeding value - b2 *average relationship where b1 is the economic value of a breeding value unit and b2 is the cost of inbreeding. Depending of the weights chosen, EVA finds the particular use of the potential parents, which optimise the variable S and thus the breeding scheme. As the average relationship of a particular candidate to the group of selected candidates changes depending on the set of selected candidates, the S value also change. Therefore complex calculations are needed. The selected matings are printed on a mating list. It is difficult to specify the weights b1 and b2, and it depends on various subjective estimates. The literature gives, however, methods describing functions to specify these weights. The value of the genetic gain, the chosen time horizon as well as the cost of the inbreeding depression all form part of these functions. Again it is a subjective estimate, which time horizon is to be considered. The longer time horizon, the lower weight on the genetic gain and the higher weight on relationship. The reason for that is that the genetic variation will be reduced with inbreeding, and the longer time horizon considered the more important it is to keep the inbreeding increase under control. Another possibility is to choose the weights in such a way that a maximum rate of inbreeding per generation will not be exceeded. The literature gives arguments that the rate of inbreeding should not exceed 0.5 to 1% per generation, e.g. Bijma, 2000. Another and more practical access to the problem is that the dairy cattle association cannot accept to reduce the genetic gain for the next generation of young bulls too much. This is so because there is a limit to how much genetic gain they are willing to loose in the short term to get more back in the long term. Considering it over a longer period of time it might be profitable to give up more genetic gain in the short term against having more genetic gain in the long term. To do this it is essential to have the acceptance of the users. Without this, there is a risk that they might choose another supplier, who would provide a larger genetic gain just now – a reaction that is easy to understand if farmers are economically strained or are supposed to be out of business the day the genetic gain would pay off. 5.2.4 Implementation EVA is used as a facility in Dansire’s breeding programme. The results from the programme are not being used uncritically but more as a pointer in the work in optimising the selection. As can be seen from the two following examples EVA now gives guidelines for which bulls 38 should be used and in which scale to keep a rational and acceptable genetic gain and rate of inbreeding. We would, however, like to keep improving the applications, while we continue to develop the programme and the procedures used in connection with the calculations by making them more efficient. 5.2.5 The Use of EVA for Red Dane When selecting young bulls more or less weight might be attached to the relationship between the selected bulls and the bulls previously used in the mating scheme for Red Danes (RDM). The more weight being attached to the relationship the lesser degree of inbreeding will be the result in the future and the larger genetic gain in the long term. Unfortunately, it will cost genetic gain in the short term. The table below shows the distribution of sires of the young bulls selected depending on no weight, a moderate weight or a large weight being attached to relationship. The 30 bulls have been selected among 90 bull calves after cows on bull dam level in September 2004. 39 Table 1 Distribution of sire of sons for 30 RDM young bulls depending on how much weight has been attached to the relationship Sire of Sons FYN Aks Fyn Cent Jobladin Syd Hallas Vest Andy Micmac Fyn Cima SYD Garant T Hejnsvig T Moberg VEST Bæk ØDA Best Number of young bulls No weight attached Moderate weight to relationship attached to relationship 4 4 2 1 3 1 4 7 7 4 5 1 4 1 5 5 2 Great weight attached to relationship 4 2 1 1 1 6 1 4 1 2 4 3 The table shows that the sires of the selected bulls are distributed more evenly on more sires the more weight has been attached to the relationship. 5.2.6 The use of EVA for Danish Holstein By using EVA for Danish Holstein (SDM-DH), 150 (it is of course possible to vary this number) mating combinations are selected. The mating combinations would vary depending on the ratio of the weights on genetic gain and the average relationship in next generation. Out of these mating combinations the distribution of sires of sons can be calculated, as well as the distribution of sires of the selected cows. Table 2 shows the distribution of sire of sons and sire of dams at the varying weight on relationship 40 Table 2 The distribution (%) of sires and maternal grandsires to the selected combinations at different weight attached to relationship 0 indicates that all weight has been attached to genetic gain, at increased number of +’es the weight is being increased on relationship. ++++ indicates that weight has only been attached to relationship. Furthermore, the table shows the average total merit index and the average relationship. The calculations have been made using predicted breeding values from 15-4 2004. 0 + ++ +++ ++++ Av. Ebv Progeny 115,95 115,55 114,12 113,65 113,20 Average 0,118 0,110 0,109 0,109 relationship Distribution of sires* Bob 117 67 45 Laudan 116 33 Bjørn 111 1 6 Dynasty 112 1 3 O. Justi 116 55 35 23 11 Chuck 114 16 13 7 Okendo 111 2 Puck 110 7 15 Chassee 113 2 3 3 Boliver 109 47 52 53 Other distribution of sires of dams** Farmer (16) 4 5 8 9 9 Huxley (34) 2 4 4 5 7 Lord Lily (10) 3 1 Marshall (21) 5 5 3 1 Funkis (52) 35 35 20 17 8 Lambada (11) 2 2 1 1 TB Steven (13) 3 1 1 Hesne (34) 3 5 10 12 15 Bojer (14) 3 4 1 1 Others (150) 39 39 53 55 58 * After the name the bull’s total merit index has been indicated in brackets. ** The number of potential bull mothers after which each sire of dam has been indicated in brackets. The table shows that the more weight is being attached to relationship the more the sires of sons and the maternal grand sires are used and in more even proportions. If great weight has been attached to the genetic gain almost only sires of sons with a high total merit-index have been selected, whereas the total merit-index of the sire of sons is of no importance, if all weight has been attached to the average relationship. When all weight has been attached to relationship, the bulls are selected with the smallest degree of relationship with the population in general. The population in this case is the animals in the pedigree file. In Table 3 the average relationship between some sires of sons and the animals in the population distributed on year of birth is shown. 41 Table 3 The average relationship between some sire of sons and animals in the pedigree file distributed on year of birth Bull sire Sire Maternal 2004 2003 2002 2001 grand sire RGK Bob Lukas Marconi 12.3 10.4 10.5 8.8 Laudan Lukas Raider 12.5 10.5 10.2 9.2 Chuck Manfred E. Elton 7.4 6.5 7.1 6.7 O Justi Novalis E Lutz 8.8 6.2 6.9 6.7 V Exces Luxemburg Leadman 9.5 8.6 8.8 9.1 V Elo Lukas Luke 13.6 10.5 10.6 9.1 Chassee Novalis Jabot 8,5 6.5 7.5 6.6 Dynasty Lucky Leo Luke 7.6 6.8 7.5 7.3 VAR Elvis Luxemburg Dannix 9.0 7.8 8.3 8.6 Okendo Winchester Mountain 7.7 8.1 6.8 6.8 RGK Bjørn Mattie G Luke 8.2 6.9 7.7 6.2 Puck Patron Merrill 7.4 7.0 7.1 6.2 Tresor C M Val Fatal 8.7 7.7 8.0 8.1 P Bandores C M Val Hunter 9.4 8.6 9.3 8.8 E. Boliver Amel P Mathie 3.7 3.7 4 3.7 When estimating these relationships it is important to draw the attention to the fact that some animals have lacking pedigree information with only two complete generations of pedigree. generation. It may reduce the relationship to the population considerably. In general it can be seen, however, that sires of sons with a high degree of average relationship with the population as e.g. RGK Bob, V Elo or Laudan are not used at all as sires of sons (Table 2) when the weight is only attached to relationship. 5.3 References Bijma, P. 2000. Long-term genetic contributions, prediction of rates of inbreeding and genetic gain in selected populations, Ph.D. Thesis, Wageningen, Holland. Grundy, B., Villanueva, B. & Woolliams, J.A., 2000. Dynamic selection for maximizing response with constrained inbreeding in schemes with overlapping generations. Anim. Sci. 70:373-3832. Meuwissen, T.H.E. & Sonneson, A.K., 1998. Maximizing the response of selection with a predefined rate of inbreeding: overlapping generations. J. Anim. Sci. 76:2575-2583. Smith, L.A., Cassell, B.G. & Pearson, R.E. 1998. The effects of inbreeding on the lifetime performance of dairy cattle. J. Dairy Sci. 81: 2729-2737. Sonesson, A.K., Grundy, B., Woolliams, J.A. & Meuwissen, T.H.E., 2000. Selection with control of inbreeding in populations with overlapping generations - a comparison of methods. Anim. Sci. 70:1-8. Sørensen, A.C., Madsen, P., Sørensen, M.K., Berg, P. 2005a. Danish Holstein Show Inbreeding Depression for Udder Health. In preparation. Sørensen, A.C., Sørensen, M.K., Berg, P. 2005b. Inbreeding in Danish Dairy Cattle Breeds. J. Dairy Sci. Accepted. Sørensen, M.K. (1999). Stokastisk simulering af avlsplaner for malkekvæg. DJF, Rapport nr. 13, 183 pp. 42 6.0 New techniques in dairy cattle breeding schemes 6.1 Use of Marker Assisted Selection in Nordic Breeding Schemes Most traits in animal breeding are influenced by both genotype and environment. Traditional quantitative genetics is based on the infinitesimal model, assuming that a trait is controlled by an infinite number of genes, each with a small effect. The additive effect of these genes determines the total genetic effect. The infinitesimal model has proven to work and the success of modern dairy cattle breeding is one proof. The knowlegde of genetic regulation has, however, increased, and today it is well known that most traits are regulated by a combination of genes with large effect and genes with small effect. In 1990, designs to identify Quantitative Trait Loci (QTL), or single loci with a large influence on a trait, was suggested for cattle (Weller et al., 1990). These designs, the Daughter Design and the Grand Daughter Design, have been used to identify QTL affecting various traits, such as milk yield, milk composition, various conformation traits, and funcitonal traits (e.g. Freyer et al., 2003; Ron et al., 2004; Schrooten et al., 2000;Van Tassel et al., 2004). Today, genetic evaluation in dairy cattle breeding is based mainly on progeny testing, and it is generally accepted that it is possible to obtain genetic progress for any heritable trait in a progeny testing scheme. The limitations are that the traits are reliably recorded, bulls are progeny tested with a sufficient number of daughters to gain high accuracy on the index, and the trait is given sufficient weighting in the net merit index. Although it is possible to select successfully for most traits in a progeny testing scheme, there are some challenges: most of the economic important traits (e.g. milk production, daughter fertility, mastitis resistance) can only be measured on females after first calving. With the long generation interval in cattle, sires will be 5-6 years old before genetic proofs are available, and the elite sires can be selected. Progeny testing programs are also expensive to run: bulls are kept alive until they get their proofs at 5-6 years of age, and then only a few percent of the bulls are selected to be used in the breeding program as elite sires. To obtain maximum genetic improvement in a population, the best sires must be used to a maximum (with constrains to avoid inbreeding). However, in a progeny testing scheme, a certain amount of the cows have to be bred by young bulls to produce daughters for progeny testing. If it was possible to select elite sires as young calves based on genetic markers linked to favourable alleles of a gene, or on the gene itself, the generation interval for elite sires would be dramatically reduced, and progeny testing would be redundant. The costs of running a breeding scheme would then be represented by the cost of genotyping a sufficient number of bulls to find the best sires for the next generation, and to produce and distribute semen from these sires. It would also be possible to use the best elite sires to breed all cows, resulting in maximum genetic progress in the population. This approach, however, requires that markers/genes for all traits in the breeding goal are known, and that breeding values can be estimated with high accuracy based on this information. Based on the challenges in the progeny testing scheme, several studies have suggested implementation of markers and/or genes in breeding schemes, where sires are selected based on genetic markers (Marker Assisted Selection; MAS) in addition to, or as a substitute for 43 progeny testing schemes. Various schemes have been suggested, the two major being withinfamily selection and MAS in BLUP evaluations. The BLUP based evaluations require genotypes on all animals (Fernando and Grossman, 1989), which would represent a major cost. In within family MAS conventional breeding value estimation (progeny testing) is combined with MAS within families to select sons carrying the favourable QTL-allele (e.g. Spelman and Garric, 1998). To our knowledge, MAS has not so far been implemented for sufficient many generations to study long-term effect in dairy cattle breeding. However, various simulation studies have investigated the effect of MAS on genetic gain (e.g. Abdel-Azim and Freeman 2002; Fernando and Grossman 1989; Meuwissen and Van Arendonk 1992; Spelman and Garrick 1998; Stella et al. 2002). In general, results show that the superiority in genetic gain is large for the first generations, then the superiority decline as the QTLs reach fixation in the population (e.g. Abdel-Azim and Freeman 2002). Simulation results show highest genetic gain when MAS is included in a scheme where all animals are genotyped, for example in a closed nucleus system (e.g. Fernando and Grossman 1989); when the QTL explaines a large proportion of the genetic variance; when the position of the QTL is determined by high accuracy; and when the trait is rare or have a low heritability. Although MAS might be a useful supplement in dairy breeding schemes, this technique has several challenges. QTL/genes influencing important traits need to be identified, and the position of the QTL has to be accurate to avoid the wrong selection decision due to cross-over between the marker and the QTL. At the moment, QTL has been found for a limited number of traits included in the Nordic Red breeding schemes. Another limiting factor is the price of genotyping and the number of available markers/SNPs. The bovine genome is being sequenced at the moment (http://www.genome.gov). The sequencing will provide important knowledge on the genes on the bovine genome, and also provide important genetic markers (SNPs), however, these genes and SNPs have to be validated in the Nordic Red populations, and linked to important traits before this information could be implemented in the breeding scheme. The breeding schemes in the Nordic countries have a special position worldwide because they also include functional traits with low heritability in the net merit index. As mentioned, traits with low heritabilities are challenging to include in progeny testing schemes. Good recording systems and large daughter groups, as well as sufficient weight in the net merit index, are necessary to obtain genetic progress for these traits. In the Nordic countries, functional traits with low heritability have been included in the breeding schemes for decades, and genetic improvement for these traits have been documented in some of the breeds (e.g. AndersenRanberg et al., 2005). Several competing breeding organizations do not include these traits in their breeding schemes at the moment due to lack of good recording systems. The need for developing good recording systems in these countries could decrease if MAS could be implemented to obtain genetic progress for these traits. If QTL/genes affecting traits like fertility or health are identified with high accuracy, it would be possible for competing organizations to select for functional traits without including these in the progeny testing program (i.e. only select sires in to the progeny testing program that have been shown to carry the favourable QTL/allele for these traits). This would result in a new competitive situation for the Nordic Countries. It should also be noted that the genetic regulation of traits is complicated. Favourable and unfavourable genetic correlation between traits are well known in quantitative genetics. 44 Selection for a gene that would be beneficial for one trait, could have a unfavourable effect on another important trait. To fully implement MAS in animal breeding schemes, it would therefore be extremely important to know how various genes interact, and how selection for one trait would influence other traits. Although it is not likely that MAS will be used as a substitute for progeny testing for the next decade in the Nordic countries, it is likely that genetic information could be (are) used as a supplement to progeny testing schemes. Bulls entering the progeny testing program could be pre-selected based on which allele they carry for a known gene. This could be genes controlling a certain disease, like BLAD (Bovine Leukocyte Adhension Deficiency, genetic test presently used by breeding organisations) or other traits controlled by a single gene (like polled/horned). 6.2 Use of sexed semen in the Nordic Red Breeds Sexed semen are now public available for farmers (http://www.cogentuk.com/). Semen in cattle is separated using fluorescent activated cell sorting (Johnson et al., 1987a, 1987b). The X chromosome contains more DNA than the Y chromosome, and it is therefore possible to treat sperm with a DNA specific fluorescent dye, and subsequently sort the sperm through high throughput flow cytometry. It has to be noted that the method is not perfect: a proportion of the sperm will be damaged or unsorted. It is well known (discussed by Weigel 2004) that sexed semen are considerably more expensive than non-sexed semen, and is known to give lower conception rates. Results for fertility from several field trails in Holstein, concludes that conception rates in virgin heifers are expected to drop from 55-60% with unsexed semen to 35-40% with sexed semen (Weigel 2004). Despite the lowered conception rates and the higher price, the commercial availability of sexed semen has been impatiently expected by the industry. The main reason for this is probably the reduced fertility (lower conception rates and more days open) in the Holstein breed in recent years, resulting in to few replacement heifers. Commercial availability of sexed semen could change this, resulting in more female calves being born, and thereby more replacement heifers. In the Nordic Red breeds, there has not been a decreasing genetic trend for fertility, and there is no reason to expect that sexed semen will be used to increase the number of replacement heifers. The Nordic Breeding Associations will, however, face other challenges when sexed semen is commercially available at affordable prices. Semen from young sires for progeny testing is currently used in regular herds all accross the Nordic countries (within country random use). The use of semen from young sires vary between the Nordic countries. Today semen from young sires are used to 30-40% of the cows. It must be expected that in a situation where sexed semen is commercially available at a reasonable price, farmers will buy sexed semen from the best elite sires to produce replacement heifers. Cows that are not inseminated with sexed semen from the best elite sires can be expected to be inseminated with the purpose of beef production. In this situation it will be difficult to get the necessary number of daughters after young sires in production to perform progeny testing, and in the worse case, it would not be possible to run a progeny testing scheme. Facing a situation like this, it might be necessary for the breeding associations to change their progeny testing system into contract use of semen from 45 young sires, where farmers have a contract to inseminate with semen from young sires and use these heifers for replacement for the purpose of progeny testing. It must be assumed that farmers on contract must get an economic compensation for the use of semen from young sires, which would result in a additional cost for the breeding associations. It should however be noted that semen from young sires in a contract situation could be sexed to reduce the amount of semen to produce daughters for progeny testing. For sexed semen to be commercially used in the Nordic countries, the increased cost of sexed semen has to be considerably lower compared to today, where sexed semen represents a major increase in price per straw. With the current technology, 150-200 straws can be sorted per machine per day, resulting in a high increase in price (Weigel, 2004). It would also be reasonable to expect that a major increase in conception rate for sexed semen would be a requirement for the successful commercial implementation of sexed semen in the dairy industry in the Nordic countries. A reduction in virgin heifer conception rate to 35-40% with the use of sexed semen would be dramatically for the dairy farmers in the Nordic countries. Although there are limitations with the commercial use of sexed semen at the moment, it must be expected that some of these problems will be solved. When semen can be sexed at higher speed resulting in a lower price, and if the reduction in conception rate is decreased, it should be expected that dairy farmers in the Nordic countries wants the opportunity to purchase sexed semen from their breeding associations. It is therefore necessary that the Nordic breeding associations have a strategy to meet this situation. 6.3 References Abdel-Azim., G., and A. E. Freeman. 2002. Superiority of QTL-assisted selection in dairy cattle breeding schemes. J. Dairy Sci., 85:1869-1880 Andersen-Ranber, I. M., G. Klemetsdal, B. Heringstad. and T. Steine. 2005. Heritabilities, genetic correlations, and genetic change for female fertility and protein yield in norwegian dairy cattle. J. Dariy Sci., 88:348-355 Fernando, R. L., and M. Grossman. 1989. Marker assisted selection using best linear unbiased prediction. Genet. Sel. Evol., 21: 467-477 Freyer, G., P. Sørensen, C. Kühn, R. Weikard, and I. Hoeschele. 2003. Search for Pleiotropic QTL on chromosome BTA6 affecting yield of milk production. J. Dairy Sci., 86:9991008 Johnson, L. A., J. P. Flook, and M. V. Look. 1987a. Flow cytometry of X and Y chromosome bearing sperm for DNA suing and improved preparation method and staining with Hoechst 33342. Gamete Res. 17:203-212 Johnson, L. A., J. P. Flook, M. V. Look, and D. Pinkel. 1987b. Flow sorting of X and Y chromosome-bearing spermatozoa into two populations. Gamete Res. 16:1-9 Meuwissen, T. H. E., and J. A. M. Van Arendonk. 1992. Potential improvements in rate of genetic gain from marker-assisted selection in dairy cattle breeding schemes. J. Dairy Sci., 75:1651-1659 Ron, M., M. Golik, I. Tager-Cohen, D. Klinger, V. Reiss. R. Domochovsky, O. Alus, E. Seroussi, E. Ezra, and J. I. Weller. 2004. A complete genome scan of the Israeli Holstein population for quantitative trait loci by a daughter design. J. Dairy Sci., 87:476490 Schrooten, C, H. Bovenhuis, W. Coppiters, and J. A. M. Van Arendonk. 2000. Whole genome scan to detect quantitative trait loci for conformation and functional traits in dairy cattle. J. Dairy Sci., 83:795-806 46 Spelman, R. J., and D. J. Garrick. 1998. Genetic and economic responses for within-family marker-assisted selection in dairy cattle breeding schemes. J. Dairy Sci., 81-2942-2950 Stella., A., M. M. Lohuis, G. Pagnacco, and G. B. Jansen. 2002. Strategies for continual application of marker-assisted selection in an open nucleus population. J. Dairy Sci., 85:2358-2367 Van Tassel, C. P., T. S. Sonstegard, and M. S. Ashwell. 2004. Mapping Quantitative Trait Loci affecting dairy conformation to chromosome 27 in two holstein grandsire families. J. Dairy Sci., 87:450-457 Weigel, K. A. 2004. Exploring the role of sexed semen in dairy production systems. J. Dairy Sci., 87:(E.Suppl):E120-E130 Weller, J. I., Y. Kashi, and M. Soller. 1990. Power of daughter and granddaughter designs for determining linkage between marker loci and Quantitative Trait Loci in dairy cattle. J. Dairy Sci., 73: 2525-2537 47 7.0 Nucleus breeding in Practice – a Nordic approach Morten Kargo Sørensen, Dep. of Genetics and Biotechnique,and Søren Borchersen, Dansire 7.1 Introduction For more than 25 years the breeding plans for dairy cattle in most Western countries have been based on the waiting bull principle. According to this highly selected young bulls are used on a limited scale and subsequently wait until breeding values based on daughter phenotypes are estimated. Thereafter final selection of proven bulls and sire of sons can take place. The breeding schemes will constantly be adapted, and modern reproduction technology will be used in many breeding plans. The theoretic possible genetic gain in a conventional breeding scheme with the use of artificial insemination is 0.26 genetic standard deviation units in total economic merit per year. However, many international examinations show that the genetic gain in practice has been far below the theoretically obtainable. Through the latest 10 years the implemented breeding growth has for most breeds in Denmark been about 0.17 standard deviation units in total economic merit per year. This is due to both lack of consequence in the management of the breeding scheme and to the fact that the accuracy of bull dam selection in private breeding herds is considerably lower than expected. Because of that there has been an increasing interest in nucleus breeding schemes with use of embryo-transfer in many countries. Numerous simulating studies have proven that open nucleus breeding with test herds and heifer flushing has a considerable potential, and several nucleus breeding programmes have been initiated throughout Europe. If Nordic cattle breeders plan far ahead to maintain their competitiveness and position as a serious international partner it is important to continue the existing Nordic nucleus breeding programmes at a highly professional and commercial level. A description will be given below on the following: Accuracy in the selection of bull dams The theoretic background of nucleus breeding programmes Practical experience of those subjects 7.2 The accuracy in the selection of bull dams The selection of bull dams is very important for the total breeding result. In the so-called conventional breeding schemes the theoretically expected contribution of the bull dams to the genetic gain varies from 25 to 40%. The importance of the bull dams is highest in the breeding plans with a wide use of embryo-transfer and a large share of young bull insemination (Pedersen, 1992). In the latest decade many results have been published showing that the accuracy of the selection of bull dams is lower than expected. A Danish study from the early 1990ies has shown that the estimated breeding values of bull dams for milk was overestimated by 225 to 350 kg (Pedersen, 1992), and this tendency has just been confirmed (Hostrup, 2003). In both examinations the deviation was largest for SDM-DH. 48 The expectation of the estimated breeding value of a progeny equals the average of the estimated breeding value of the parents. The similar regression coefficients are shown in table 1. The regression coefficients show how much the breeding value of the dam can tell about a son’s breeding value. The theoretic value of this regression coefficient is 0.5. The Regression Coefficient The Regression Coefficient of y on x is an expression of how much y is expected to change when x is changed by one unit. Table 1. The regression coefficients of the breeding value of bulls and cows for the yield of butter fat on the similar breeding value figures of the parents. Bull dams . Cow dams RDM SDM-DH DJ Ped1) 0,23 0,25 0,30 Sire of sons. Hos2) 0,27 0,30 0,30 Ped1) Hos2) RDM 0,48 0,42 SDM-DM 0,37 0,47 DJ 0,40 0,42 1) DOA breeding value figures - Pedersen, (1992) 2) Animal Model breeding value figures - Hostrup, (2003) Hos2) 0,48 0,48 0,48 Sire of cows Hos2) 0,50 0,51 0,51 As shown the regression coefficient for bull dams is only about half the theoretically expected of 0.5. The Danish results are in agreement with Finnish results where the average empirical bias in pedigree indices was estimated to 13.6 kg protein. The correlation between the final proof of the bull and the EBVs of the bull sire or dam were 0.45 and 0.17, respectively. The low correlation with bull dam EBV indicates the unreliability of the bull dam EBVs (Uimara & Mäntysaari, 1995). The reason is primarily preferential treatment of some cows. This has an unpleasant ring to many people, but fundamentally it just covers up the fact that there is more awareness to good animals – which can be called good farmer practice. The implementation of an increased registration in the herds or automatic milking does not solve any problem, as it is a question of a real additional yield to be obtained on the basis of a slightly better environment for each animal. In relation to breeding value estimation this is however a problem as the higher yield is not a result of the animal’s genes. Preferential treatment is the most essential source of error in the breeding value estimation of bull dams, and until now it has not been possible to describe the effect in a way permitting a correction. Preferential treatment hinders the ability of the estimated breeding value to foresee the breeding value of untested bulls, which limits the genetic gain. It has been expected that the increasing size of herds as well as more cows in free stalls would partly solve the problem. However, until now there has been no sign that the preferential treatment will be reduced, why the breeding plan must be adjusted accordingly. However, there is reason to believe that an improved method to correct the herd variation will reduce the problem. This method will be applied together with the NAV (Nordisk Avlsværdivurdering) testday model for yield. 49 The positive preferential treatment of bull dams is very much dependent on herds. Fig. 1 shows that the average difference between the estimated breeding values of the sons for protein yield and the breeding value of the parents for protein yield vary from –1.5 kg to 17 kg in the 9 Danish SDM-DH herds, in which from 1994 to 1997 more than 15 progeny tested bulls were born. On average this difference should be zero. The reason for not being zero is preferential treatment. The higher negative differences between the estimated breeding values of bulls and the average breeding values of parents, the higher degree of preferential treatment in the herd. Herd Number. 1 2 3 4 5 6 7 8 9 0 -2 kg protein -4 -6 -8 -10 -12 -14 -16 -18 Fig. 1. The difference between the bulls’ and parents ’estimated breeding values for protein in 9 different SDM herds (Hostrup, 2003). The anticipation of preferential treatment within bull dams is confirmed by the fact that daughter/mother regression for production cows is 0.48 – which is very close to the theoretically expected. Also the production cow/father regression is high, while the son/father regression shows some variation, which may be caused by inaccurate estimated breeding values of some of the imported sire of sons. The Danish and Finish results are confirmed by international examinations, e.g. the results from the Delta nucleus breeding programme, where bull dams are tested and selected in test herds. Here the overestimation of bull dams is only 1/5 of the overestimation found for bull dams in normal breeding herds (Fig. 2). A dispersed nucleus where the nucleus animals are placed in private herds does not create a solution. The Canadian nucleus breeding programme TEAM where bull dams were selected and flushed in private herds has found an extremely low regression value of about 0.1 (Lohuis, 1998). 50 INET 200 180 160 140 Av. parent EBV 120 Av. bull EBV 100 80 60 Normal breeding herds Delta herd Fig. 2. Pedigree EBV and own breeding value for Dutch AI bulls born in normal breeding herds or in the DELTA-project. The Dutch EBV for yields is called INET. By the implementation of special nucleus herds it is possible to avoid preferential treatment. Because of the equal environment it is also possible to obtain a generally higher heritability. In the Danish FY-Bi project it was documented for yield, where the heritability for both milk, fat and protein was about 0.42 against about 0.28 in field surveys (Jakobsen et al. 1997, Sørensen 1999). The heritability is also expected to be higher for functional traits. e.g. somatic cell count, so that it might be possible to calculate estimated breeding values for these traits including own performance. If a central nucleus is a part of the breeding programme it is important to make sure that the environment in the nucleus herd reflects the future (10 – 15 years) production circumstances, as a genotype environment interaction may arise. It may have the effect that the theoretically possible breeding growth in the population cannot be obtained as indicated in another chapter. 7.3 Simulation of nucleus breeding schemes The nucleus breeding programmes are characterized by exploiting the possibilities of advanced reproduction techniques to the optimum so that the generation interval for the female animals is reduced, and the number of offspring from the best female animals will be increased. By collecting the bull dam candidates in special nucleus herds (test stations) it may as mentioned be possible to obtain a more accurate breeding value estimates of bull dams. There is a differentiation between open and closed nucleus breeding programmes. The open programmes are characterized by involving competitive breeding material from the total population according to requirements, as also future insemination bulls are selected between all progeny tested bulls. In the following only simulation results from open nucleus breeding schemes will be mentioned. 51 New Finish simulation studies found a very large effect of nucleus breeding schemes - from 40% to 77% larger genetic gain for yield (Stranden et al. 2001). An essential reason for the large effect was that the comparison standard was normal conventional breeding plans and not effective screening schemes. The latest Danish simulations have shown that nucleus breeding plans will increase the genetic gain by about 5% (Nielsen et al. 2001). For a population of 50.000 cows a nucleus of 100 lactating cows was assumed, and for a population of 190.000 cows the nucleus was 200 cows. However, it must be mentioned that the effect of preferential treatment and any other advantages of additional registrations in a nucleus have not been taken into account in these calculations. The effect of preferential treatment in breeding schemes has been calculated several times. The results showed that total genetic gain for production was reduced by about 10% if the estimated breeding value of bull dams has been overestimated in practice due to preferential treatment (Glacius 1999,Meuwissen and Ruane 1989, Schrooten et al. 1993). Screening Plan: A screening plan is a breeding scheme, where breeding animals are selected exclusively on estimated breeding values. Thus the conclusion to be drawn is that the genetic gain will increase by at least 5% by changing an effective screening plan to an open nucleus programme. But as shown above considerable preferential treatment still exists in many actual breeding herds, why the real effect of nucleus breeding will be at least 10 to 15%, depending on the weigth put on production in the breeding goal. To this must be added that to days conventional breeding schemes only exploits 60 – 70% of the potential of a screening plan. The expected larger consequence in the selection procedure will contribute additionally to the genetic gain. Danish calculations show that having a population size of 500.000 cows the value of the extra genetic gain will be 50 to 75 mill. kr. per year after payment and interest of the costs involved. 7.4 Nucleus breeding in practice There are many other nucleus breeding programmes in Europe, many of them established within the latest years. Nucleus breeding programmes in Holland, Finland, Germany and Sweden have been examined closely; plans and experience from these projects are mentioned below. Delta Project in Holland The Delta project is a nucleus breeding program, which has been running over a longer period, and from this much practical experience has been gained. In Holland about 270 Holstein Friesian bulls, 60 red Holstein Friesian and 10 Dutch red and white bulls are tested. About half of the black and white bulls originate from the Delta project. Holland Genetics now own all animals in the nucleus herd. The heifer calves are placed there at the age of no more than 6 weeks, raised under standardized conditions, flushed 2 –3 times at the age of 12-16 months and then inseminated at the age of about 17 months. Subsequently OPU are carried out on the best third of the heifers. The oocytes are fertilized by the test tube method (IVP) 52 During first lactation the heifers are performance tested on milk production. About 300 heifers are tested every year, and the best 60 are selected as bull dams. Delta has made contracts with recipient farms, and many embryos are sold with an option on the calves born. In 2002 there were about 500 bulls born in the Delta project, which have been progeny tested, and as mentioned above their pedigree value has been determined with a higher degree of accuracy than the contemporary young bulls from other breeding herds. This results in a larger part of the Delta young bulls being selected as proven bulls, which strengthens both the genetic gain as well as the economy of the AI organisation. ASMO Project in Finland The breeding nucleus in the ASMO project consists of 70 first lactation cows placed in the test station at the age of 8-12 months. Asmo herd is owned by Alkiokeskus Oy - Emryocenter Ab and MTT together, Alkiokeskus Oy is owned by AI-FABA, Breeding-FABA, dairy industry and Svensk Avel. The heifers are flushed twice prior to the insemination at the age of about 17 months. Furthermore the OPU/IVP is made on the selected first lactation cows. When the cows have ended their first lactation then the best cows are selected as donors. According to the plan this should be 15-20 cows per year, in reality this is only about 7, partly because of phenotypic selection on conformation. A network of recipients consisting of 50 herds has been established. These recipients buy embryos relatively cheap on the condition that ASMO has the option. Finland initiates every year 120 young bulls, and 25% of them should come from ASMO. Today approximately 27% of the test bulls come from ASMO, so the goal has been reached. The project has during the latest years experienced a growing support and importance. NOG Project in Germany NOG is a breeding cooperation between four North German AI organisations, and they started in 2001 a nucleus breeding project including bull dam testing. NOG plans each year to place 280 Holstein Friesian young bulls, and out of this number 5060% is to be found in the nucleus breeding project. Each year about 200 heifer calves are selected for the nucleus breeding programme. The heifers are flushed twice and subsequently inseminated. If less than 10 useful embryos is obtained then subsequent OPU/IV is made. Two months prior calving the heifers are placed in a test herd with space for 150 cows, owned by the university in Kiel. The test period is 180 days, and each year 30 cows / bull dams are selected. They complete the lactation in the test herd, and produce more embryos. Viken Project in Sweden In 2002 Svensk Avel started a nucleus breeding project called Viken. The first heifers were bought for the project in spring 2002, and the first elite doners were ready in December 2004. The project is shared equally between Läntmannen, Svalöf Wiebull and SEAB (Svensk Embryoservice AB). SEAB is owned by Svensk Avel ( 51 %), Dansire (36.5 %) and FABA (12.5 %). The project has a capacity to test 100 Swedish Red Cattle and 100 Holstein Friesian first lactation cows per year. Heifer calves are placed there at the age of 10 months and prior to insemination they are flushed twice. An annual production of 1800 embryos is expected. 53 Those embryos are sold to the network herds where Viken has the option on the calves born, both bulls and heifers. In the first part of second lactation the “elite donors” are selected. The “elite doners” are subsequently flushed. According to the plan 60 SRB and 60 HF young bulls from the projects are initiated every year. 7.5 Recommendations and conclusion The implementation of nucleus breeding programmes results in a major increase in genetic gain. The reason is higher accuracy in the bull dam selection, shorter generation intervals and a stronger management of the practical part of the breeding work. To this must be added that the nucleus breeding programmes and test stations have a considerable marketing effect. Therefore it is important that the Nordic nucleus herds continue in an integrated cooperation. This is an essential condition for the Nordic cattle breeding to keep its professional and commercial competitiveness both nationally and internationally. Well established nucleus programs (ASMO and Viken) with participation from both Denmark, Sweden, and Finland already exist. Norway is not involved and will probably not be in the near future, since preferential treatment does not seem to be a problem in Norway. In the existing programs it is important only to focus on indices in the selection procedure. Practice have shown that phenotype in some cases play a role in the selection, this rely reduce the possible effect of a nucleus program. Due to the Nordic breeding profile it’s obligate to include more registrations on functional traits in the existing and coming nucleus programmes in the Nordic countries. In relation to that feed intake will be an essential measurement, since intake can not be registered accurate in production herds. Inbreeding has as mentioned in chapter 5 growing interest in dairy cattle breeding. The more offspring from the highly selected animals the more important is inbreeding, therefore this has to be taken into account when heifers are selected for the nucleus herds. The tools mentioned in chapter 5 just have to be implemented in the selection procedure. 7.6 References Glacius, A., 1999. Effekt af særbehandling og fysiologiske funktionsprøver i avlsplaner for malkekvæg. Speciale afhandling, Institut for Husdyrbrug og Husdyrsundhed, KVL. 93 sider. Hostrup, H. 2003. The relation between bull mothers and the breeding value figures of their sons – using Animal Model for yield for RDM, SDM-DH and DJ. Tesis, ”Institut for Husdyrbrug og Husdyrsundhed”, KVL. 54 sider. Jakobsen, J.H., Liboriussen, T. & Jensen, J., 1997. Genetiske parametre for produktionsegenskaber. I : Liboriussen, T. & Andersen, B.B. (ed.) Ægtransplantation, fysiologiske funktionsprøver og kerneavl med malkekvæg. Beretning nr. 737 fra Danmarks JordbrugsForskning, 47-72. Jørgensen, J.N. & Sørensen, M.K. , 2003. Breeding schemes for improvement of genetic response in dairy breeds of minor population size. Godkent til publikation i Acta Agric. Scand. 54 Lohuis, M. & Bagnato, (1998). What have we learned from the TEAM projekt ? http://www.aps.uoguelph.ca/cgil/pub/mmlpapers/hjjuly98.html. Meuwissen, T.H.E., Ruane, J., 1989. The value of daughters of bull dams in dairy cattle progeny testing schemes. Livest. Prod. Sci. 23: 267-274. Nielsen, L.P., Sørensen, M.K., Berg, P. & Noesgaard, J., 2001. Alternative avlsstrategier for rød dansk malkerace. DJF-rapport nr. 32 - Husdyrbrug, 99 sider. Pedersen, G.A., 1992. Assessment of breeding value and selection of bull mothers. Examination of the expected and realized efficiency in the Danish dairy races. Ph.D. thesis,” Institut for Husdyrbrug og Husdyrsundhed,” KVL. 83 sider. Schrooten, C., Steverink, M.H.A. & van Arendonk, J.A.M., 1993. Stochastic simulation of dairy cattle breeding schemes: Influence of breeding strategy and baised breeding values in the population. J. Anim. Breed. Genet. 110:268-280. Stranden, I., Korpioha, P., Pakula, M. & Mäntysaari, E.A., 2001. Bull selection in MOET nucleus breeding schemes with limited testing capacity. Acta Agric. Scand. 51: 235-245. Sørensen, M.K., 1999. Stokastisk simulering af avlsplaner for malkekvæg. Ph.D. afhandling, Institut for Husdyrbrug og Husdyrsundhed, KVL. 226 sider. Uimari, P., Mäntysaari, E. A. 1995. Relationship between bull dam herd characteristics and bias in estimated breeding value of bull. Agricultural science in Finland 4, 5-6: 463-472. 55 8.0 Imports from other populations to the Scandinavian Red Dairy Breeds. Lars-Olof Bårström, Svensk Avel 8.1 Introduction The Scandinavian Red Dairy Breeds, Finnish Ayrshire (FAY), Norwegian Red (NRF), Red Dane (RDM) and Swedish Red (SRB), are the most productive red breeds in the world besides Red Holstein. The breeding programmes in the red populations in the Scandinavian countries have since many years been based on Total Merit Index. All traits of economical importance for the milk production have been included in the TMI, which is a PROFITABILITY INDEX. Important traits except production and functional conformation are female fertility, calving ease and calf-mortality, udder health, beef production traits and longevity. Since many years there has been an exchange of semen from top bulls between the Scandinavian red populations. In Finnish Ayrshire and Swedish Red a number of the best bulls from the North American Ayrshire population have also been used as bull sires. Until end of the 1990th, a number of Swedish Holstein Frisian bulls have been used as bull sires in the NRF population. In the RDM breeding programme bulls from three more breeds, Brown Swiss, Montbeliarde and Red Holstein Frisian have been used. Exchange/use of bulls from different breeds in the Nordic red population SRB RDM FAy NRF RHF Brown Swiss Montbeliarde Ayrshire, US & Can HF, Swe, Dk In the ongoing work to improve the Scandinavian red breeds it is important to optimise the national breeding programs. It is also important to look for interesting genetics from other breeds or populations, which could offer the Scandinavian red population desirable traits. 8.2 What are we particular looking for for the Scandinavian Red breeds? Positive traits according to our breeding goal Traits with a positive correlation to longevity and udder health 56 Udder conformation (important to improve) o For-udder attachment o Udder depth Outcross pedigrees 8.3 Where do we find possible sources for importations to the Scandinavian red breeds? Today all countries with organized breeding programs are members of the Interbull. There are six breed groups in the Interbull system and the number of bulls in are shown below. (November 2004) Ayrshire - The red group 10780 Brown Swiss 6430 Guernsey 816 Holstein 83000 Jersey 6870 Simmental 21350 . Characteristics for breeds / populations from which it can be possible to import bulls from to the Scandinavian red breeds. Breeding values for Nordic profile traits Breeding goals close to the Scandinavian. Efficient breeding program Large population Ayrshire The Scandinavian red breeds are all in the Ayrshire group. Bulls from these breeds are totally dominating both in number and also the ranking for BV for production. In November 2004 breeding values were calculated for about 10780 bulls in the Interbull Ayrshire group and 77,3 percent of the bulls were from the Scandinavian breeds. Percentage of bulls from different populations in the Ayrshire / Red group in Interbull November 2004. Breed Percentage SRB 21,7 Finnish Ayrshire 21,6 NRF 21,2 RDM 12,8 New Zeeland Ayrshire 6,0 Canadian Ayrshire 4,1 Australian red breeds 3,3 US Ayrshire 2,5 Other ay-populations 6,8 A number of Canadian Ayrshire bulls have been used in Sweden and Finland and the conclusions from the results in Sweden are; Production obvious lower production 57 Female fertility lower fertility in average Calving traits lower calving index sire Conformation obvious higher results for udder Holstein Red Holstein bulls have been used to ”improve” many red breeds. Is this also a alternative for Scandinavian red? What will be positive and negative by using Holstein bull in the Scandinavian red breeds? Positive Milk yield Udder conformation Size Nice-looking cows?? Negative Lower components Calving traits Female fertility SCC / Udder health Beef production traits Feet & legs Scandinavian Red will not be a real alternative to HF. Holstein bulls have during a period been used as bull sires in both NRF and RDM but since some years this is ended. Guernsey, Jersey , Brown Swiss and Simmental. These breeds are quite different from the Scandinavian red breeds and there today no reason to consider importations of bulls from these breeds to the Scandinavian population. Brown Swiss bulls were during 1980th used in the RDM breed but they couldn’t compete production wise. A other problem was that many of the US Brown Swiss bull used were carrier of genetic defects. Montebeliarde is a French breed in the Simmental family with a breeding goal rather close to the Scandinavian. Bulls from the breed have been used in the RDM program with positive results for production, beef traits, female fertility and udder health. For traits like udder conformation, feet and legs and calving ease the results were negative. During the last year two for these traits very positive Montebeliarede bulls have been used in the RDM programme. 8.4 Conclusions The breeding work to improve the Scandinavian red breeds have to be done within the population. There is today no real source for importation of bulls to the Scandinavian red population. Hopefully some of the best bulls from red populations with great influence of Scandinavian red can be brought back. (Aussi Red in Australia, New Zeeland Ayrshire and also Canadian Ayrshire) North American Ayrshire bulls will improve udder conformation but aren’t competitive for other traits. A number of the highest LPI-cows in Canada have been flushed to some of the best bulls from Sweden and also Finland and resulting young sires have been tested in Swedish and Finnish breeding programs. Holstein bulls are not real alternative in the Nordic Red population. For the future it is important that the Nordic Red population is ”holstein free” 58 9.0 Competitive advantages in the Nordic red breeds. Torstein Steine, Geno, and Lars-Olof Bårstrøm, Svensk Avel 9.1 The Nordic Red Breeds The Nordic red breeds have been selected according to a broad breeding goal over many years. Unlike breeding programs in other countries, the nordic breeding organisations have all included health and fertility in the breeding goal. Another important difference is the fact that these traits are based on direct observations of health and fertility, not only some traits with some correlations to the real traits. The nordic breeding objectives have without doubt had a significant positive effect on health and fertility in the red nordic breeds. On the other hand such a breeding goal also leads to less weight on milk production, which is quite necessary to be able to obtain anything on health and fertility. In this report we might look into all possible details with regard to the development of the four red nordic breeds. We do not find that to be a very fruitful operation. It is of more value to see how the red nordic breeds perform compared with the major dairy breeds in the world. Now we have recent results from both USA and Ireland. In USA both NRF and SRB are used in crossbreeding with Holstein, especially in California. In Ireland and Nortern Ireland NRF are compared with Holstein as a pure breed. It may be argued against these comparisons that it should be done just within the group of red breeds. But today semen is sold globally, and the commercial dairy farmers will hardly look at colour if there are clear economical differences between breeds. In order to have a future on the international scene the red nordic breeds have to compete with Holstein. Table 1. Results from California for production during the first 150 days of lactation. Published by Les Hansen and coworkers, University of Minnesota, USA NormandeMontbeliarde- ScandinavianTrait Holstein Holstein Holstein Holstein * Cows Milk (kg) 294 31.8 171 28.6 194 31.0 120 33.3 Fat (kg) 1.10 1.01 1.10 1.17 Protein (kg) 0.94 0.89 0.95 1.01 SCS 2.13 2.40 2.33 1.88 Fat + Protein (kg) 2.03 1.90 2.05 2.18 ̶ 6% 1% % of Holstein +7% * Scandinavian = NRF or SRB Table 1 show the results from California for crosses between Holstein and NRF or SRB compared with pure Holstein. There will be more results available in March 2005. At that time also fertility and full lactation records will be included. This analysis has used records 59 from very large Californian herds, more than 1000 cows. The French breeds, Montbelliard and Normande have also been used in these herds. Therfore we get a comparison with these breeds too. Table 2. Results from Ireland, milk yield 2. lactation, kg. Published by Frank Buckley and coworkers, Teagasc, Ireland. Breed Low level of High level of concentrates, concentrates, 548 kg 1251 kg Holstein 5433 6360 NRF 5208 5871 Montbelliarde 4826 5700 Normande 4307 4661 Table 3. Results from North-Ireland, milk production. Published by Conrad Ferris and coworkers, ARINI, North-Ireland. NRF Holstein Trait 305 day milk yield, kg 5704 5957 Fat % 3,88 3,82 Protein % 3,29 3,22 Days in production 319 316 Full lactation yield, kg 6084 6317 Table 4. Results from North-Ireland. Fertility, culling and still births. Published by Conrad Ferris and coworkers, ARINI, North-Ireland. NRF Holstein Trait Heifers: Non return(%) 62 54 Culled due to infertility 2 3 1. lactation cows: Non return(%) 50 39 Culled due to infertility 6 17 All: Culled of the 230 animals per 26 54 breed starting in the project 11% 24% Still births 5,5% 15,5% These results demonstrate that the Nordic red breeds are very competitive with Holstein. This is especially true for functional traits like fertility, somatic cell count and therefore probably also mastitis resistance, but they are also at almost the same level as Holstein for milk yield. Economically the Nordic red breeds seem to be quite superior to Holstein. Experiences from the Nordic countries indicate less clear differences between the breeds than shown here. This may be a result of a systematical selection of cow sires based on nordic breeding objectives in the nordic Holstein populations. This has probably had some positive effect even though most of the bull sires have been imported. 60 But there are more challenges in making an international market for the red Nordic breeds than just to demonstrate the quality of the animals. In USA the results from the cross breeding trial in California was reported in a farmers magazine where NRF and SRB were called “whohave-ever-heard-of-them-breeds”. This shows that we have to start from nowhere to sell the Nordic dairy cattle genetics. The international results and all the publicity around these results have however helped very much in the marketing, but we still have some way to go. 9.2 Conclusions The quality of the red Nordic dairy cattle genetics has proven to be at a high international level. The Nordic red breeds are outperforming Holstein in functional traits as health and fertility in trials in California and Ireland. The Nordic red breeds are close to Holstein in milk yield. There is still some work to do in marketing the Nordic dairy cattle genetics. If we are able to continue the efficient breeding work and also succeed in marketing, the Nordic countries may become a dairy cattle breeding nucleus for a large part of the world. 61 10.0 Prospects for cooperation between Baltic and Nordic countries in dairy cattle breeding Jarmo Juga, FABA 10.1 Background Four Nordic countries, Denmark, Finland, Norway and Sweden, have signed an agreement of joint testing and use of AI-bulls. The motivation for this agreement is to improve the efficiency of current breeding programs in four countries by mowing towards a joint breeding program and to better utilise the existing resources. Joint use of best bulls also guarantees a better assortment of proven bulls for commercial breeding in a single country. It also increases the competitivity in export of Nordic semen to other countries. Accurate comparison and selection of bulls and cows across Nordic countries is of great importance in such cooperation. To improve the possibilities for across country comparison the breeding organisations have established a new company Nordic Cattle Genetic Evaluation (NAV), which has a responsibility to develop and run joint genetic evaluations for Denmark, Finland and Sweden. There is a mutual interest to expand this Nordic cooperation also to Baltic countries: Estonia, Latvia and Lithuania. The aim of this report is to evaluate the possibilities for the future cooperation in prediction of breeding values and sustainable selection of animals across Nordic and Baltic countries and also to look at the future market for export in Baltic countries. 10.2 Milk recording in Baltic countries In Estonia the recording scheme is quite developed and about 90 % of the cows belonged to milk recording in 2003. One reason for high participation is that belonging to milk recording is required for state subsidies. The number of cows in Estonian milk recording was 102788 of which about 26 %, 74 % were Estonian Red and Estonian Holstein, respectively. Estonia has also some native cattle left in recording. The average production in 2003 was 5119 kg milk with 4,44% butter fat and 3,39 % protein for Estonian Red and 5906 kg milk with 4,27 % butter fat and 3,27 % protein for Estonian Holstein. In Latvia quite small number of herds belongs to milk recording at the moment due to a big number of small herds. The goal is not to have all the herds in recording, but to have the biggest and most progressive herds, which will most likely continue the production in the future. There was an increase in the number of herds in recording in 2003, which is a challenge for the milk recording system, since many new herds enter to the system with incomplete information. Due to the high integration of databases all new herds have pedigree and calving information, however. In the end of 2003 Latvia had 186300 dairy cows, of which 62143 belonged to milk recording. Of these 42219 were Latvian brown breed, 17377 Holstein and few hundred were Angler, Danish red or Swedish red and white. The average yield for Latvian brown was 4550 kg with 4,45 % of fat and 3,22 % of protein. The average yield for Holstein in 2003 was 5296 kg of milk with 4,21 % of fat and 3,09 % of protein. The total number of dairy cows in Lithuania in 2002 was 441800 of which 84237 Holstein cows and 31634 red cows belonged to milk recording. The average milk yield for Holstein 62 was 5136 kg with 4,2 % of fat and 3,29 % of protein. The milk yield for red breed in 2002 was 4695 kg with 4,38 % of fat and 3,43 % of protein. To my knowledge the recording scheme is the least developed in Lithuania. Latvia has applied to ICAR Special Stamp in the end of 2003 with a reasonable good auditing report from ICAR auditors. There are a lot of small herds in all Baltic countries, but also quite a high number of very large herds, which are formed from old state farms. Some of these big farms are very well developed, especially in Estonia, and looking very much forward to future production under EU system. I have also experienced the same optimism in Latvia among the larger dairy herds. These herds are investing a lot to management, but also to foreign genetics. 10.3 Genetic evaluation Sampling data from milk recording database to the evaluation system is problematic in Latvia, since a lot of data is lost due to different reasons. This might be a reflection of new herds in recording system with incomplete information, but also a quality problem of recording on farm. A review of sampling criteria and classification of fixed (environmental) effects might be advisable. I assume the situation is very similar in Lithuania. Estonia is the most developed country when genetic evaluation is concerned. Estonia also participates to INTERBULL evaluation. Latvia and Lithuania are members in ICAR and INTERBULL, but do not send data to international evaluation. 10.4 Breeding programs in Baltic countries Both in Estonia and Lithuania the most common breed in milk recording is Holstein, which has increased very fast. The most common breed in Latvia is still the Latvian brown breed, which makes 74% from all livestock Estonia runs a breeding program for Holstein and Estonian red. The breed organisations have recently merged. Latvia has also a national breeding program on both Latvian brown and Holstein breed. The number of annually tested bulls is small in all Baltic countries in all breeds. Increasing the number of tested bulls would speed up the genetic progress, but obviously the small number of recorded cows makes it difficult to carry out more inseminations with young bull semen. Increasing the number of herds in recording scheme will produce more capacity for the testing program, too. Breeding programs rely very much on imported semen. In Latvian brown many of the red European populations including red Holstein are accepted as sires. Recently the Swedish SRB has been found very successful (Strautmanis, 2003). Also Estonia and Lithuania have used Nordic red semen, especially Swedish and Danish red. 10.5 The possibilities for cooperation All Baltic countries are willing to develop their recording and evaluation systems, breeding programs and use of new technology in animal breeding. The biggest obstacle currently is the lack of money and skilled people. The cooperation with Nordic countries could grow from joint R&D projects and education. In research and development the possible areas of cooperation are to map the possibilities for a joint genetic evaluation and sustainable selection program. 63 The problem on Nordic side is that there is much less money available to projects in Baltic countries compared to the situation some years ago. All Nordic countries and Nordic Ministerrådet were funding Baltic projects through development funds, but after Baltic countries joint the EU these funds are much less. Now the best possibility is to apply money to joint EU projects, but the competition is much harder and the amount of work needed to write the application is much bigger. Genetic evaluation Before planning any joint evaluation between Nordic and Baltic populations the genetic links across populations should be studied. Between population links are of two kinds, either genetic links, through the additive genetic relationship matrix, or data links, where animals have offspring in the same contemporary group. The amount of such links makes up connectedness in the data. The studies have shown that Ayrshire populations in Finland and Sweden have good genetic links. Norwegian red is quite well linked to Swedish red and moderately well linked to Finland. Danish red is more isolated due to its past breeding policy, but recent exchange of semen has increased the amount of links to Swedish red population. In Holstein the populations in all three Nordic countries Denmark, Finland and Sweden have good genetic links. Nordic countries have realised the need for increment in exchange of semen between populations to enable more accurate comparison. Since only Estonia has delivered data to Interbull evaluations it is of interest to quantify the genetic links between the Baltic and Nordic red populations. There was a report of genetic links and model validation of Estonian Red population for Interbull Ayrhshire evaluation (Uba and Kruus, 2004) in the conference for Animal Breeding in Baltics in 2004. Selection program Differences in breeding programs between countries may be due to differences in economic conditions. Currently the economic values used in the breeding goals are calculated according to different methods and principles. Different weights will lead to different selection decisions in different countries, even if the same bulls are available for selection everywhere. It would be of interest to compare the breeding goals in Baltic and Nordic countries and to study the possibilities for harmonisation of the methods to calculate economic values. It is also important to describe the benefit of animals with better health, fertility and other functional traits, especially when the consumers are putting more emphasis in the future to non-economic values, such as ethical value of better health and welfare. Global breeding and more effective selection will also introduce a higher risk of decreasing the effective population size and hence increasing the change in inbreeding (Meuwissen and Woolliams, 1994). This risk can be accounted for by utilising the results of modern selection theory, for example by maximising the response of selection with a predefined rate of inbreeding or constraining the variance of response (Meuwissen, 1997). Common use of sires across countries (Mark et. al., 2002) would increase connectedness, but probably also the rate of inbreeding. It is hence important to look for the optimum exchange 64 of genetic material between populations. One aim in R&D could be to calculate the change in average inbreeding in Baltic populations and to estimate the contribution of imported sires to the current breeding populations. This information could be used in guiding the use of imported material in a sustainable way in Baltic breeding programs. 10.6 Possibilities in marketing The Baltic countries make a small market with a hard competition from the central Europe. The Nordic countries are strong in red breeds, but unfortunately the red breeds are playing a smaller and smaller role in Baltic countries. The Holstein breed is taking very much over and will do so very fast if the Nordic countries don’t help the Baltic red breeding programs. There is an interest to import Nordic red genetics to Baltic countries, but the money available is limited and the numbers of doses will be quite small. The total number of red cows in milk recording in Baltics is about 100000. If foreign bulls would be used for 10 % of the population the total marked would be about 20000 doses per year. Some years ago the Baltic countries used old AI bulls from Sweden and Denmark. Now they import mainly semen. I think that the best strategy would be to help the Baltic countries to improve their own breeding programs and sell semen to be used for best cows. In the long run the Baltic programs could cooperate on a more equal level, but it will take at least ten years. There might be some room for selling the red genetics to crossing in the very large herds. 10.7 Contact people in Baltics Estonia: Tanel-Taavi Bulitko, Animal Breeders´ Association of Estonia, must@estpak.ee Olav Saveli, Estonian Animal Breeding Association email: saveli@eau.ee Mart Uba, Estonian Animal Recording Centre (EARC), email: mart.uba@reg.agri.ee Kaivo Ilves, Estonian Animal Recording Centre (EARC), e-mail: kaivo.ilves@reg.agri.ee Latvia: Ilona Miceikiene, Lithuanian Veterinary Academy, e-mail: genetikalab@lva.lt Rita Zutere, Latvian State information Data Processing Centre of Domestic Animals, e-mail: Rita.Zutere@vcidac.lv Erna Galvanovska, Latvian State Information Data Processing Centre of Domestic Animal Pedigree (LSIDPC), email: erna.galvanovska@vcidac.lv Lithuania: Donata Uchockiene, Holstein Breeders Association, Lithuania Arunas Svitojus, Rural Business Development and Information Centre in Lithuania (RBDIC), e-mail: arunas@vic.lt 10.8 References Mark, T., W.F. Fikse, H. Jorjani & J. Philipsson. 2002. Monitoring changes in the structure of global dairy cattle populations. Proc. 7th World Congress on Genetics Applied to Livestock Production, 19-23 August 2002, Montpellier, France. 65 Meuwissen, T.H.E. 1997. Maximizing the response of selection with a predefined rate of inbreeding. J. Anim. Sci. 75, 934-940. Meuwissen, T.H.E. & J. Woolliams. 1994. Effective sizes of livestock populations to prevent a decline in fittnes. Theor. Appl. Genet. 89, 1019-1026. Strautmanis, D. 2003. Improving Latvian Brown dairy breed by using different breeds. In Proc. Baltic Animal Breeding Conference, Sigulda 2003, pp. 37-40. Uba, M. and M. Kruus. 2004. Data of Estonian Red breed for Interbull Ayrshire evaluation; Genetic links and model validation. In Proc. Animal Breeding in the Baltics, Tartu 2004, pp. 100-1004. 66 11.0 Discussion/summing up One of the major concerns in modern animal breeding is to avoid a rapid increase in rate of inbreeding. As it is discussed in chapter 5, breeding associations have to decide how much increase in inbreeding they are willing to accept in the short and long term scale. This is the major challenge in modern animal breeding where one sire has the potential to be used for all cows in the Nordic red population to obtain maximum genetic gain. There are several aspects to consider when an optimum breeding strategy should be chosen in the Nordic red breeds. A major concern is the number of sires that should be used across breeds, and how much they should be used within the country of origin to avoid inbreeding, and also how much they should be used in other Nordic countries before the other red Nordic populations does not serve as a genetic pool in case of inbreeding problems. Simulation results in chapter 2 shows that only a limited exchange of sires across countries will result in a rapid decrease in the possibility of using the other red population as genetic pools. At the same time it should be noted that a certain exchange of genetic material is necessary to obtain genetic links between countries. In addition to genetic links, traits have to be defined similarly and recorded in similar ways countries to be able to rank sires across countries. It is also obvious from the discussion in chapter 2 that the correlation between breeding goals needs to be at a certain level before exchange of sires are important. It should be discussed whether this is the case today. In the results presented in chapter 2, the correlation between breeding goals are ≈0.8, however this number are very sensitive for the parameters used as input. In the simulation scheme in chapter 2, the breeding populations in Norway, Sweden and Finland are used as a base, and only two traits are considered; milk and mastitis resistance. We would, however, expect that the realistic correlation between countries are much lower, because there are large variation amongst countries for breeding goals when the other traits included in the breeding goals are considered. Further work is to be done to predict the results of exchange of sires between countries with larger differences in breeding goals and with more extreme variation in genetic parameters. Based on the results in chapter 2 there are now reason to doubt that exchange of sires between breeding associations at a higher level than what we see today, will result in extra genetic gain at the short term, but in a longer term the opportunity to use the other Nordic red breeds as a genetic pool if one of the populations should get heavily inbred will disappear. It should also be noted that the conclusion from chapter 8 is that there are no other alternative populations to import from if the Nordic red breeds become inbred. Another aspect that should be considered when it is discussed at what level sires should be exchanged, is genotype-environment (GXE) interaction. As it is discussed in chapter 3, there have been found indications of GxE interaction, indication that different genotypes have different ranking (reranking or scaling) between the Nordic countries. This indicates that a sire that are ranked as number 1 in one Nordic country doesn’t necessarily rank as number 1 in another Nordic country or within environments in a country. This indicates that the wrong animals could be selected for a given environment, or that we would observe lower performance in the environments different to those where the genetic evaluation was performed. Although certain environmental factors are quite similar in the Nordic countries, there are undoubtedly differences in, for instance, feeding regimes, production level, amount of daylight, and temperature. Breeding associations select those sires with the best results in 67 the given country, and when considering exchange of sires it has to be kept in mind that there may be some differences in ranking between countries. It is also an important question if and how breeding nucleuses may be integrated in the Nordic breeding schemes. Today there are such nucleuses in Sweden and Finland. Denmark is an active part in the Swedish nucleus, and Sweden and Finland have been working together on the nucleus breeding for some years. Norway is the only country without participation in any nucleus. This reflects some differences in the view on what is the best way of utilising the limited resources available in dairy cow breeding. There are some disagreements on the benefit from nucleuses in a scheme with many low heritable traits in the breeding goal, and about the benefit from a nucleus when the rate of increase in inbreeding has to be kept low. These questions can be better answered after some years of experience with the existing nucleus herds in Sweden and Finland. The Nordic red breeds (especially NRF and SRB) have during the last 2 years become important competitors to the Holstein breed. These breeds can offer genetic material bred with special emphasis on health and fertility, which have become major problems in the Holstein breed. To maintain this competitive advantage, it is extremely important that the breeding associations continue to run breeding schemes with large emphasis on functional traits, that the breeding programs are run as optimal as possible (as discussed in chapter 4), and that the breeding organisations work to maintain the high participation in the recording systems in the Nordic countries, so that data of high quality are recorded on a routine basis and can be used for progeny testing in the future. Summing up Modern animal breeding consists of possibilities, risks and challenges. It is possible to increase the selection intensity by utilising sires across countries, or using the best sires more intensely within country. But this may also lead to more inbreeding and reduced selection intensity in the long term. Theoretical studies on nucleus herds demonstrate that such herds may give an extra lift to the genetic gain. But is it possible to use this and still keep inbreeding under control? Or will the risk of increase in inbreeding reduce the extra benefit from the nucleus resulting in no extra benefit? This report does not give any final answer to these questions, but it highlights some of the topics of importance when discussing more close breeding cooperation between the Nordic countries. We hope this report will start many discussions on the topics included, and that it will help in continuing and improving a Nordic cooperation in dairy cattle breeding, balancing the risks and the benefits. The situation today is quite unique: We have three internationally competitive cow populations with some relations between them, but also with some differences. With all the existing knowledge and experiences in dairy cattle breeding it should be possible to develop this into a way of running dairy cattle breeding which makes the Nordic countries even more competitive, but without loosing the genetic variation. 68 Appendix 1. The use of EVA-programme in Finnish Ayrshire and Finncattle breeding schemes Terhi Vahlsten, Finnish Animal Breeding Association Ayrshire The main idea with Ayrshire was to find an answer to the question how to breed bullsires and bulldams in order to keep genetic progress high and rate of inbreeding low? There were several different strategies to explore this question. The first strategy was to mate the current bullsires to current bulldams i.e. cows that were expected to inseminate during the next few months. This strategy is rather easy to carry out and the animals are already selected by the breeding committee (bullsires) and breeding advisor (bulldams). Basically, this strategy produces a mating plan for bulldams and this strategy can be done four times a year after every breeding value evaluation. The second strategy was to approximate the number of sons AI should by out of each bullsire. The idea was to include all cows that have been inseminated during the last 12 months and which had the bulldam status at the insemination time. The bulls were those bulls which were selected as bullsires during the last 12 months. This strategy is not very useful because it is difficult to set limits to the time period because all bullsires are not changed after every breeding value evaluation. And this strategy answers the question what should have been done instead of what should be done. The results from this strategy were not promising although the EBV’s were high but the rate of relationship per generation was too high. The third strategy included cows calved in a certain time with high TMI and udder conformation index. Heifers were also included if they filled certain requirements. The bulls were the best available Finnish and foreign bulls. The results weren’t very good; especially the rate of relationship per generation was very high. The results changed dramatically when only half of the cows and heifers were mated. EVA-programme selected the mated females and the rate of relationship per generation was in acceptable level and the EBV’s maintained high. This strategy showed that the programme loses some of its powers when the females are selected strictly and every female has to be mated. Finncattle The Finncattle breed is old traditional Finnish breed and there are three different types: Western, Northern and Eastern types. The Western type is most common and over 2400 cows are included in milk recording scheme. The Eastern type is endangered as well as the Northern type. Although the Western type is not endangered the situation is not good for this type. Especially the average relationship is high and the young bulls are rather close related to their potential mates. The effect of few old bulls has become too large and the gene pool is narrow. About 70 % of inseminations in Western population are made with young sire semen in order to try to keep the genetic base as wide as possible. The young sires are also used as bullsires and bulldams need to fill certain requirements. The EVA-programme was used for making mating plan for bullsires and bulldams but the results were not very promising. The main problem with the Western type is that the best cows come from the very popular families. 69 The Eastern and Northern populations are so small that the selection is based almost entirely on avoiding inbreeding. However, the inbreeding is not a problem with these two types. For both types there are some bulls that are used rather much and their sons are bought for AI. It is very important that the effect of one bull or family doesn’t become too significant in a very small population. The object with these types is to make a mating plan for the population with the EVA-programme. However, there are some requirements for the females, thus the plan is not made for the entire population. The females include all live cows, heifers and calves whose sire and maternal grandsire are same type. Thus the females are at least 75 % Eastern or Northern type. The selected bulls need to have enough sperm doses and they shouldn’t have many live daughters in the population and their average relationship to the selected females should be rather low. The object is to do this plan once a year and the farmer receives a letter where the bull candidate is informed. This is only a proposition and the farmer makes the final decision of the bull. 70 71