Optimization of dairy cattle breeding programs for different environments with genotype by environment interaction

Dairy cattle breeding organizations tend to sell semen to breeders operating in different environments and genotype × environment interaction may play a role. The objective of this study was to investigate optimization of dairy cattle breeding programs for 2 environments with genotype × environment interaction. Breeding strategies differed in 1) including 1 or 2 environments in the breeding goal, 2) running either 1 or 2 breeding programs, and 3) progeny testing bulls in 1 or 2 environments. Breeding strategies were evaluated on average genetic gain of both environments, which was predicted by using a pseudo-BLUP selection index model.When both environments were equally important and the genetic correlation was higher than 0.61, the highest average genetic gain was achieved with a single breeding program with progeny-testing all bulls in both environments. When the genetic correlation was lower than 0.61, it was optimal to have 2 environment-specific breeding programs progeny-testing an equal number of bulls in their own environment only. Breeding strategies differed by 2 to 12% in average genetic gain, when the genetic correlation ranged between 0.50 and 1.00. Ranking of breeding strategies, based on the highest average genetic gain, was relatively insensitive to heritability, number of progeny per bull, and the relative importance of both environments, but was very sensitive to selection intensity. With more intense selection, running 2 environment-specific breeding programs was optimal for genetic correlations up to 0.70-0.80, but this strategy was less appropriate for situations where 1 of the 2 environments had a relative importance less than 10 to 20%. Results of this study can be used as guidelines to optimize breeding programs when breeding dairy cattle for different parts of the world.     

Short communication: genotype by environment interaction due to heat stress

Heat stress was evaluated as a factor in differences between regional evaluations for milk yield in the United States. The national data set (NA) consisted of 56 million first-parity, test-day milk yields on 6 million Holsteins. The Northeastern subset (NE) included 12.5 million records on 1.3 million first-calved heifers from 8 states, and the Southeastern subset (SE) included 3.5 million records on 0.4 million heifers from 11 states. Climatic data were available from 202 public weather stations. Each herd was assigned to the nearest weather station. Average daily temperature-humidity index (meanTHI) 3 d before test date was used as an indicator of heat stress. Two test-day repeatability models were implemented. Effects included in both models were herd-test date, age at calving class, frequency of milking, days in milk × season class, additive genetic (regular breeding value) and permanent environmental effects. Additionally, the second model included random regressions on degrees of heat stress (t = max[0, meanTHI - 72]) for additive genetic (breeding value for heat tolerance) and permanent environmental effects. Both models were fitted with the national and regional data sets. Correlations involved estimated breeding values (EBV) from SE and NE for sires with ≥100 and ≥300 daughters in each region. When heat stress was ignored (first model) the correlations of regular EBV between SE and NE for sires with ≥100 (≥300) daughters were 0.85 (0.87). When heat stress was considered (second model), the correlation increased by up to 0.01. The correlations of heat stress EBV between NE and SE for sires with ≥100 (≥300, ≥700) daughters were 0.58 (0.72, 0.81). Evaluations for heat tolerance were similar in cooler and hotter regions for high-reliability sires. Heat stress as modeled explains only a small amount of regional differences, partly because test-day records depict only snapshots of heat stress.     

Holstein-Friesian dairy cows under a predominantly grazing system: interaction between genotype and environment

A 5 yr whole-system study, beginning in June 1994, compared the productivity of high [HGM; Australian Breeding Value (ABV) of 49.1 kg of fat plus protein] and low [LGM; ABV of 2.3 kg of fat plus protein] genetic merit cows. Cows from both groups were fed at 3 levels of concentrate (C): 0.34 (low C), 0.84 (medium C), and 1.71 (high C) t of DM/cow per lactation. Thus, there were 6 treatments (farmlets) composed of 18 cows each. The 30 blocks of pasture on each farmlet were matched between farmlets for pasture growth before the study (and soil characteristics and aspect). Cows were culled, and pasture and feed use were managed so as not to bias any one treatment. Genetic merit, level of feeding, and their interaction were significant effects for protein content, protein/cow, and milk and protein/ha. For fat and milk yield/cow, genetic merit and level of feeding were significant, whereas there was no significant effect of genetic merit on fat content. The difference of 46.8 kg of fat plus protein yield between the ABV of HGM and LGM cows and the actual difference in production between the 2 groups was not significantly different except for low C (27 kg) cows. This was due to a 3-fold lower protein yield difference (6 kg/cow) compared with an ABV difference for protein yield of 17.9 kg/cow. The dramatic effect of treatment on protein is in line with differences in the mean protein content (2.89% for the HGM - low C cows compared with a mean of 3.02% for the remaining groups) and mean body condition score [4.3 for HGM - low C cows compared with 4.8 for the mean of the remaining groups (scale 1 to 8)], both indicators reflecting a higher negative energy balance in the HGM - low C cows. When individual cow production was plotted against ABV for production of milk or protein yield all relationships were quadratic, but the slope was relatively flat (low response to ABV) for the low C cows, steeper for the medium C cows and steepest (but not linear) for the high C cows. The relationship between ABV for fat yield and actual fat yield was linear for all levels of concentrate. The mean milk yield/ha from pasture for the 6 farmlets over the 5 yr was 11,868 L, 11,417 L, or 7,761 L for the HGM cows fed at low C, medium C, or high C, respectively, and 10,579 L, 9,800 L, or 5,812 L for LGM cows, fed at low C, medium C, or high C, respectively. The response to concentrates fed was very high for the HGM - medium C cows at 0.115 kg fat plus protein or 1.75 L milk/kg of concentrate fed, with comparable figures of 0.083 kg and 1.0 L, 0.86 kg and 1.47 L and 0.066 and 0.92 L/kg of concentrate fed for the HGM - high C, LGM - medium C, and LGM - high C, respectively. The results show a significant genetic merit by environment (level of feeding) interaction for reproduction and most production parameters when considered in terms of the individual cow and the whole farm system.     

Short communication: Changes in the association between milk yield and age at first calving in holstein cows with herd environment level for milk yield

The aim of this study was to evaluate the effect of herd environment class on the genetic and phenotypic relationships of mature equivalent milk yield (MY) on age at first calving (AFC). Data analyzed were 248,230 first parity records of Holstein cows, daughters of 588 sires in 3,042 herds in the United States. Heritability for AFC was 0.33 ± 0.01 and 0.20 ± 0.01 in high and low environment herds, respectively, and 0.47 ± 0.01 in the complete data set. The correlation between AFC sires’ predicted breeding values of low and high classes was 0.69. Genetic correlations between MY and AFC were −0.52 ± 0.02 and −0.31 ± 0.03 in high and low environment herds, respectively, and −0.44 ± 0.02 in the complete data set representing intermediate environments. If selection is based on the whole data set, expected correlated responses for AFC estimated as a result of 1,000 kg of genetic gain in MY, for high and low herd environment herds were −26.1 and −15.3 d, respectively, and −32.6 for the complete data set; hence the highest reduction in AFC occurs in intermediate environment herds. Different estimates of the heritability of AFC, the correlation between AFC breeding values of low and high classes as well as changes in the genetic correlation between MY and AFC across environments indicate genotype × environment interaction. Caution in interpretation is warranted because genetic relationships are dynamic, especially in populations undergoing selection. Current relationships may differ from those during the time period of the present study (1987–1994). Notwithstanding this possibility, methods and findings from the present study provide insight about the complexity of genetic association and genotype × environment interactions between AFC and MY.     

Genomic selection improves the possibility of applying multiple breeding programs in different environments

One joint breeding program (BP) for different dairy cattle environments can be advantageous for genetic gain depending on the genetic correlation between environments (r_g). The break-even correlation (r_b) refers to the specific r_g where genetic gain with 1 joint BP is equal to the genetic gain of 2 environment-specific BP. One joint BP has the highest genetic gain if r_g is higher than r_b, whereas 2 environment-specific BP have higher genetic gain if r_g is lower than r_b. Genetic gain in this context is evaluated from a breeding company's perspective that aims to improve genetic gain in both environments. With the implementation of genomic selection, 2 types of collaboration can be identified: exchanging breeding animals and exchanging genomic information. The aim of this study was to study genetic gain in multiple environments with different breeding strategies with genomic selection. The specific aims were (1) to find r_b when applying genomic selection; (2) to assess how much genetic gain is lost when applying a suboptimal breeding strategy; (3) to study the effect of the reliability of direct genomic values, number of genotyped animals, and environments of different size on r_b and genetic gain; and (4) to find r_b from each environment's point of view. Three breeding strategies were simulated: 1 joint BP for both environments, 2 environment-specific BP with selection of bulls across environments, and 2 environment-specific BP with selection of bulls within environments. The r_b was 0.65 and not different from r_b with progeny-testing breeding programs when compared at the same selection intensity. The maximum loss in genetic gain in a suboptimal breeding strategy was 24%. A higher direct genomic value reliability and an increased number of genotyped selection candidates increased genetic gain, and the effect on r_b was not large. A different size in 2 environments decreased r_b by, at most, 0.10 points. From a large environment's point of view, 1 joint BP was the optimal breeding strategy in most scenarios. From a small environment's point of view, 1 joint BP was only the optimal breeding strategy at high r_g. When the exchange of breeding animals between environments was restricted, genetic gain could still increase in each environment. This was due to the exchange of genomic information between environments, even when r_g between environments were as low as 0.4. Thus, genomic selection improves the possibility of applying environment-specific BP.     

Optimizing the design of small-sized nucleus breeding programs for dairy cattle with minimal performance recording

Dairy cattle breeding programs in developing countries are constrained by minimal and erratic pedigree and performance recording on cows on commercial farms. Small-sized nucleus breeding programs offer a viable alternative. Deterministic simulations using selection index theory were performed to determine the optimum design for small-sized nucleus schemes for dairy cattle. The nucleus was made up of 197 bulls and 243 cows distributed in 8 non-overlapping age classes. Each year 10 sires and 100 dams were selected to produce the next generation of male and female selection candidates. Conception rates and sex ratio were fixed at 0.90 and 0.50, respectively, translating to 45 male and 45 female candidates joining the nucleus per year. Commercial recorded dams provided information for genetic evaluation of selection candidates (bulls) in the nucleus. Five strategies were defined: nucleus records only [within-nucleus dam performance (DP)], progeny records in addition to nucleus records [progeny testing (PT)], genomic information only [genomic selection (GS)], dam performance records in addition to genomic information (GS+DP), and progeny records in addition to genomic information (GS+PT). Alternative PT, GS, GS+DP, and GS+PT schemes differed in the number of progeny per sire and size of reference population. The maximum number of progeny records per sire was 30, and the maximum size of the reference population was 5,000. Results show that GS schemes had higher responses and lower accuracies compared with other strategies, with the higher response being due to shorter generation intervals. Compared with similar sized progeny-testing schemes, genomic-selection schemes would have lower accuracies but these are offset by higher responses per year, which might provide additional incentive for farmers to participate in recording.     

Genotype x environment interaction for milk yield in Holsteins using Luxembourg and Tunisian populations

Test-day (TD) milk yield records of first-lactation Holstein cows in Luxembourg and Tunisia were analyzed using within-and between-country random regression TD models. Edited data used for within-country analysis included 661,453 and 281,913 TD records in Luxembourg and Tunisia, respectively. The joint data included 730,810 TD records of 87,734 cows and 231 common sires. Both data sets covered calving years 1995 to 2006. Fourth-order Legendre polynomials for random effects and a Gibbs sampling method were used to estimate variance components of lactation curve parameters in separate and joint analyses. Genetic variances of the first 3 coefficients from Luxembourg data were 46 to 69% larger than corresponding estimates from the Tunisian data. Inversely, the Tunisian permanent environment variances for the same coefficients were 52 to 65% larger than the Luxembourg ones. Posterior mean heritabilities of 305-d milk yield and persistency, defined as estimated breeding values (EBV) at 280 days in milk-EBV at 80 days in milk, from between-country analysis were 0.42 and 0.12 and 0.19 and 0.08 in Luxembourg and Tunisia, respectively. Heritability estimates for the same traits from within-country analyses, mainly from the Tunisian data, were lower than those from the joint analysis. Genetic correlations for 305-d milk yield and persistency between countries were 0.60 and 0.36. Product moment and rank correlations between EBV of common sires for 305-d milk yield and persistency from within-country analyses were 0.38 and 0.41 and 0.27 and 0.26, respectively. Differences between genetic variances found in both countries reflect different milk production levels. Moreover, low genetic and rank correlations suggest different ranking of sires in the 2 environments, which implies the existence of a genotype × environment interaction for milk yield in Holsteins.     

Genotype x environment interaction for protein yield in Dutch dairy cattle as quantified by different models

Variance components and breeding values for protein yield were estimated with REML without and with correction for heterogeneity of variances. Three different sire models were applied, which all accounted for genotype x environment (G x E) interaction. The first model included a sire x herd-year-season subclass (HYS) interaction. The second model divided all records in four different types of management groups, based on estimated HYS subclass effect. The third model, the reaction norm model, performed a random linear regression on the estimated HYS effect. For comparison, a standard model that did not take G x E interaction into account was also applied. Data consisted of 102,899 305-d first-lactation protein records of Holstein Friesians of 1,000 ofthe largest Dutch dairy herds. All animals calved in 1997, 1998, or 1999. Estimated breeding values (EBV) for 2,150 bulls with at least five daughters were calculated. The interaction model detected an interaction variance of 2.5% of the phenotypic variance. The EBV showed a correlation of 1.00 with those of the standard model without interaction. The model with the division in groups showed correlations between groups ranging from 0.73 to 0.86. The EBV showed correlations from 0.84 to 0.91 with the EBV of the standard model. The reaction norm model calculated EBV that had a correlation of 1.00 with the EBV of the standard model. The reaction norm model was not able to detect significant variance of the slope for the protein data corrected for heterogeneity of variances.     

Identification of factors that cause genotype by environment interaction between herds of Holstein cattle in seventeen countries

Currently, the International Bull Evaluation Service calculates international dairy sire evaluations using the multiple-trait across country evaluation procedure. This method depends implicitly on political boundaries between countries, because the input data are national evaluations from each participating country. Therefore, different countries are treated as different production environments. The goal of this study was to identify factors that describe the production system on each farm. Such factors could be used to group herds across countries for borderless genetic evaluations. First lactation milk records of Holstein cows calving between January 1, 1990 and December 31, 1997 in Australia, Austria, Belgium, Canada, Czech Republic, Estonia, Finland, Germany, Hungary, Ireland, Israel, Italy, The Netherlands, New Zealand, South Africa, Switzerland, and the USA were used in this study. Thirteen genetic, management, and climatic variables were considered as potential indicators of production environments: peak milk yield, persistency, herd size, age at first calving, seasonality of calving, standard deviation of milk yield, culling rate, days to peak yield, fat to protein ratio, sire PTA milk, percentage of North American Holstein genes, maximum monthly temperature, and annual rainfall. Herds were grouped into quintiles based on herd averages for each of these variables. Genetic correlations for lactation milk yield between quintiles were significantly less than one for maximum monthly temperature, sire PTA milk, percent North American Holstein genes, herd size, and peak milk yield. The variables can be used to group herds into similar production environments, regardless of country borders, for the purpose of accounting for genotype by environment interaction in international dairy sire evaluation.     

Genotype by environment interaction for milk production traits between organic and conventional dairy cattle production in The Netherlands

Estimates of genetic parameters for organic dairy farming have not been published previously, and neither is information available on the magnitude of genotype by environment interaction (G×E) between organic and conventional farming. However, organic farming is growing worldwide and basic information about genetic parameters is needed for future breeding strategies for organic dairy farming. The goal of this study was to estimate heritabilities of milk production traits under organic farming conditions and to estimate the magnitude of G×E between organic and conventional dairy farming. For this purpose, production records of first-parity Holstein heifers were used. Heritabilities of milk, fat and protein yield, and somatic cell score (SCS) were higher under organic farming conditions. For percentages of fat and protein, heritabilities of organic and conventional production were very similar. Genetic correlations between preorganic and organic, and organic and conventional milk production were 0.79 and 0.80, respectively. For fat yield, these correlations were 0.86 and 0.88, and for protein yield, these were 0.78 and 0.71, respectively. Our findings indicate that moderate G×E was present for yield traits. For percentage of fat and protein and SCS, genetic correlations between organic and conventional and preorganic production were close to unity, indicating that there was no G×E for these traits.     

Characterization of the dairy farm environment in Great Britain and the effect of the farm environment on cow life span

Dairy farms vary a great deal in the feeding and management systems that are used. These differences affect the performance of the cows, and some genotypes may be affected more than others. If effects of such genotype-by-environment interactions (G×E) are large, then farmers must be made aware of them to make informed breeding decisions. To investigate G×E, a classification system for farm environments was developed based on national- and fine-level data from dairy herds across the United Kingdom. The national data included herd and yield characteristics and local weather information. The fine-level data included information on feeding and management systems on farms, and was obtained from survey results from 778 farms. A principal components analysis of the surveys identified 2 major dimensions characterizing the data. The first dimension explained 14.6% of the variation and was related to the level of production intensity. The second dimension explained 11.5% of the variation and was related to climate. Information on milk yield, herd characteristics, and climate was then extracted from national databases for the survey farms. A canonical correlation analysis was used to relate the survey data to the variables extracted from the national data set to determine the most relevant variables. The canonical correlation between the chosen sets of national data and survey variables was 0.62. This environmental classification was then used to determine how the farm environment affects the life span of dairy cows. The life span of the daughters of 1,000 sires was related to the type of farm environment. The daughters of a majority of sires showed a “plastic” response, with increased life span in less intensive farms. The daughters of a smaller number of sires showed a more generalized response, with life span being less affected by the environment. This G×E suggested that sires vary in the sensitivity of their daughters to different farm environments. This variation in response could allow breeding companies and farmers to match sires to particular farm environments.     

Optimal strategies for the use of genomic selection in dairy cattle breeding programs

The objective of the present study was to conduct a stochastic simulation study on the possible benefits of an application of genomic selection in dairy cattle breeding programs according to a variety of selection schemes. In addition, the heritability of the trait in question, the accuracy of genomic breeding values, and the number of animals to be genotyped were varied. Specifically, the question of genotyping males, females, or both, was addressed. Selection schemes were compared with a young bull breeding program. The main criterion for comparison was the average of true breeding values of selected young males to be used as replacements for artificial insemination bulls. Stochastic simulations were run with 50 repetitions each to generate individuals with phenotypes, breeding values estimated by BLUP, and true breeding values. Genomic breeding values were generated from true breeding values with defined accuracy. Examined scenarios included a group of selection schemes that featured genotyping of parents of future bulls only. Such schemes can be viewed as improvements of young bull programs, and they were found to be competitive with or superior to a classical young bull program. However, a genomic breeding program usually involves at least genotyping young male candidates. A second group of selection schemes reflected this requirement. Scenarios in this group were found to be superior over the young bull program by 1.0 to 1.2 standard deviations of the average true breeding value of young male candidates. Within this group of scenarios, one scheme referred to an ideal situation under which genotypes for male calves were available without limitation. Using the average of true breeding values as the criterion for comparison, this idealistic scenario was competitive with other scenarios only if the reliability of genomic breeding values was larger than 0.50. Conventionally, not all males available will have genotypes, and the 2 most promising scenarios included a preselection step for dams of future bulls. This preselection step can be based on conventional BLUP estimated breeding values for bull dams, because differences with a scheme under which both parents and the resulting male offspring are genotyped were marginal. Genotyping of young male candidates should be the focus of activities of today's breeding organizations.     

Genotype by environment interaction for somatic cell score across bulk milk somatic cell count and days in milk

The objective of this paper was to investigate the importance of a genotype × environment interaction (G × E) for somatic cell score (SCS) across levels of bulk milk somatic cell count (BMSCC), number of days in milk (DIM), and their interaction. Variance components were estimated with a model including random regressions for each sire on herd test-day BMSCC, DIM, and the interaction of BMSCC and DIM. The analyzed data set contained 344,029 test-day records of 24,125 cows, sired by 182 bulls, in 461 herds comprising 13,563 herd test-days. In early lactation, considerable G × E effects were detected for SCS, indicated by 3-fold higher genetic variance for SCS at high BMSCC compared with SCS at low BMSCC, and a genetic correlation of 0.72 between SCS at low and at high BMSCC. Estimated G × E effects were smaller during late lactation. Genetic correlations between SCS at the same level of BMSCC, across DIM, were between 0.43 and 0.89. The lowest genetic correlation between SCS measures on any 2 possible combinations of BMSCC and DIM was 0.42. Correlated responses in SCS across BMSCC and DIM were, on some occasions, less than half the direct response to selection in the response environment. Responses to selection were reasonably high among environments in the second half of the lactation, whereas responses to selection between environments early and late in lactation tended to be low. Selection for reduced SCS yielded the highest direct response early in lactation at high BMSCC.     

Culling reasons in organic and conventional dairy herds and genotype by environment interaction for longevity

Dairy cow longevity combines all functional traits and is thought to be especially important in organic production, which is an established, increasing part of Swedish dairy production, representing approximately 6% of the market. The aim of this study was to compare dynamics in culling reasons between organic and conventional production and to analyze genotype by environment interactions for longevity. The data contained information from all organic herds with information available from official recording (n = 402) and from approximately half of the conventional herds (n = 5,335). Records from Swedish Holsteins (n = 155,379) and Swedish Red cows (n = 160,794) that had their first calf between January 1998 and September 2003 were included. The opportunity period for longevity was at least 6 yr. Six longevity traits were defined: length of productive life; survival through first, second, and third lactations; fertility-determined survival; and udder health-determined survival. Twenty codes were used to describe the cause of culling, and these were divided into 8 groups: udder health, low fertility, low production, leg problems, metabolic diseases, other diseases, other specified causes, and unspecified cause. The main reason for culling cows in organic herds was poor udder health, whereas for cows in conventional herds it was low fertility. Furthermore, the shift in main culling reason from fertility, which was most common in first lactation regardless of production system, to udder health occurred at a lower age in organic production. Heritabilities and genetic correlations for the longevity traits expressed in organic and conventional herds were estimated from a bivariate animal model. The genetic correlations were close to unity (>0.88), except for fertility-determined survival in the Swedish Red breed (0.80). Heritabilities were low to moderate, and no clear pattern was identified for production system or breed. In general, the results indicate that farmers’ culling criteria differ between organic and conventional production. Different preferences may influence the need for alternative selection indexes for organic production, with different weightings of traits, or a separate breeding program. However, no genotype by environment interaction of importance was found between the production systems.     

Controlling inbreeding by constraining the average relationship between parents of young bulls entering AI progeny test programs

In breeding is known to impair the health, fertility, and productivity of dairy cattle and other livestock species. Mating programs can address inbreeding concerns on the farm, at least in the short term, but long-term control of inbreeding in a dairy population requires consideration of relationships between young bulls entering AI progeny test programs. The present study discusses an application of optimal contribution methodology to selection of young AI bulls in the five major US dairy breeds. Elite cows and active AI sires from the Ayrshire, Brown Swiss, Guernsey, Holstein, and Jersey breeds were considered as potential bull parents. Genetic merit of selected sires and dams was maximized subject to various constraints on the mean additive genetic relationship within the selected group. Relationships between selected parents can be reduced substantially relative to current levels, but the corresponding reduction in genetic merit may be large. This loss in genetic merit occurs due to lower selection intensity, although it is mainly a reflection of a larger number of bull parents (with progeny more evenly distributed among these parents), rather than selection of genetically inferior "outcross" parents that wouldn't otherwise have been considered. Selected parents were generally older and slightly less inbred than those that would have been chosen had inbreeding been ignored. Although severe restrictions on relationships can be costly, in terms of lost genetic progress, it appears that moderate constraints can keep relationships at a manageable level without a significant loss in genetic merit. Cooperation between breed associations and several competing AI companies may be required to facilitate implementation of this methodology in dispersed populations, but if this can be accomplished, prospects for achieving a balance between inbreeding and selection seem positive.