Developing flexible models for genetic evaluations in smallholder crossbred dairy farms

The productivity of smallholder dairy farms is very low in developing countries. Important genetic gains could be realized using genomic selection, but genetic evaluations need to be tailored for lack of pedigree information and very small farm sizes. To accommodate this situation, we propose a flexible Bayesian model for the genetic evaluation of milk yield, which allows us to simultaneously account for nongenetic random effects for farms and varying SNP variance (BayesR model). First, we used simulations based on real genotype data from Indian crossbred dairy cattle to demonstrate that the proposed model can separate the true genetic and nongenetic parameters even for small farm sizes (2 cows on average) although with high standard errors in scenarios with low heritability. The accuracy of genomic genetic evaluation increased until farm size was approximately 5. We then applied the model to real data from 4,655 crossbred cows with 106,109 monthly test day milk records and 689,750 autosomal SNPs. We estimated a heritability of 0.16 (0.04) for milk yield and using cross-validation, a genomic estimated breeding value (GEBV) accuracy of 0.45 and bias (regression of phenotype on GEBV) of 1.04 (0.26). Estimated genetic parameters were very similar using BayesR, BayesC, and genomic BLUP approaches. Candidate genes near the top variants, IMMP2L and ARHGEF2, have been previously associated with milk protein composition, mastitis resistance, and milk cholesterol content. The estimated heritability and GEBV accuracy for milk yield are much lower than those from intensive or pasture-based systems in many countries. Further increases in the number of phenotyped and genotyped animals in farms with at least 2 cows (preferably 3-5, to allow for dropout of cows) are needed to improve the estimation of genetic effects in these smallholder dairy farms.

The patterns of admixture, divergence, and ancestry of African cattle populations determined from genome-wide SNP data

Background

Humpless Bos taurus cattle are one of the earliest domestic cattle in Africa, followed by the arrival of humped Bos indicus cattle. The diverse indigenous cattle breeds of Africa are derived from these migrations, with most appearing to be hybrids between Bos taurus and Bos indicus. The present study examines the patterns of admixture, diversity, and relationships among African cattle breeds.

Methods

Data for ~ 40 k SNPs was obtained from previous projects for 4089 animals representing 35 African indigenous, 6 European Bos taurus, 4 Bos indicus, and 5 African crossbred cattle populations. Genetic diversity and population structure were assessed using principal component analyses (PCA), admixture analyses, and Wright’s F statistic. The linkage disequilibrium and effective population size (Ne) were estimated for the pure cattle populations.

Results

The first two principal components differentiated Bos indicus from European Bos taurus, and African Bos taurus from other breeds. PCA and admixture analyses showed that, except for recently admixed cattle, all indigenous breeds are either pure African Bos taurus or admixtures of African Bos taurus and Bos indicus. The African zebu breeds had highest proportions of Bos indicus ancestry ranging from 70 to 90% or 60 to 75%, depending on the admixture model. Other indigenous breeds that were not 100% African Bos taurus, ranged from 42 to 70% or 23 to 61% Bos indicus ancestry. The African Bos taurus populations showed substantial genetic diversity, and other indigenous breeds show evidence of having more than one African taurine ancestor. Ne estimates based on r² and r²_adj showed a decline in Ne from a large population at 2000 generations ago, which is surprising for the indigenous breeds given the expected increase in cattle populations over that period and the lack of structured breeding programs.

Conclusion

African indigenous cattle breeds have a large genetic diversity and are either pure African Bos taurus or admixtures of African Bos taurus and Bos indicus. This provides a rich resource of potentially valuable genetic variation, particularly for adaptation traits, and to support conservation programs. It also provides challenges for the development of genomic assays and tools for use in African populations.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12864-020-07270-x.

Farmers' Perceptions of Dairy Cattle Breeds, Breeding and Feeding Strategies: A Case of Smallholder Dairy Farmers in Western Kenya

To understand farmers’ preference and perceptions of breed attributes, breeding and feeding practices, 419 households in western Kenya were interviewed in a cross-sectional survey. Respondents scored their preference for cattle breeds, traits and breeding methods on a scale of 1 (most preferred) to 5 (least preferred). Preferences were compared using multinomial logistic regression models on weighted scores. The Ayrshire breed was most preferred followed by the Friesian. Using hardship tolerance as a reference trait, the Friesian was preferred 4.86 times more for high milk production and Ayrshire, Jersey and Guernsey breeds 4.61, 4.60 and 4.18 times (p < 0.01) more, respectively, for milk fat content. The Ayrshire was preferred 4.16 times more for its perceived low feed requirement and 1.22 times more (p < 0.01) for resistance to diseases. Friesian was the only breed preferred (3.18 times more) (p < 0.01) for high growth rate of calves. Artificial insemination (AI) was the breeding method of choice, but majority (>68%) of respondents used natural mating, because it was readily available and cheaper. The current study highlights the importance of taking into account farmers’ objectives and the production environment when designing breed improvement programmes and recommends packaging of breeding together with feeding interventions.

The feasibility of using low-density marker panels for genotype imputation and genomic prediction of crossbred dairy cattle of East Africa

Cost-effective high-density (HD) genotypes of livestock species can be obtained by genotyping a proportion of the population using a HD panel and the remainder using a cheaper low-density panel, and then imputing the missing genotypes that are not directly assayed in the low-density panel. The efficacy of genotype imputation can largely be affected by the structure and history of the specific target population and it should be checked before incorporating imputation in routine genotyping practices. Here, we investigated the efficacy of imputation in crossbred dairy cattle populations of East Africa using 4 different commercial single nucleotide polymorphisms (SNP) panels, 3 reference populations, and 3 imputation algorithms. We found that Minimac and a reference population, which included a mixture of crossbred and ancestral purebred animals, provided the highest imputation accuracy compared with other scenarios of imputation. The accuracies of imputation, measured as the correlation between real and imputed genotypes averaged across SNP, were around 0.76 and 0.94 for 7K and 40K SNP, respectively, when imputed up to a 770K panel. We also presented a method to maximize the imputation accuracy of low-density panels, which relies on the pairwise (co)variances between SNP and the minor allele frequency of SNP. The performance of the developed method was tested in a 5-fold cross-validation process where various densities of SNP were selected using the (co)variance method and also by alternative SNP selection methods and then imputed up to the HD panel. The (co)variance method provided the highest imputation accuracies at almost all marker densities, with accuracies being up to 0.19 higher than the random selection of SNP. The accuracies of imputation from 7K and 40K panels selected using the (co)variance method were around 0.80 and 0.94, respectively. The presented method also achieved higher accuracy of genomic prediction at lower densities of selected SNP. The squared correlation between genomic breeding values estimated using imputed genotypes and those from the real 770K HD panel was 0.95 when the accuracy of imputation was 0.64. The presented method for SNP selection is straightforward in its application and can ensure high accuracies in genotype imputation of crossbred dairy populations in East Africa.

Use of body linear measurements to estimate liveweight of crossbred dairy cattle in smallholder farms in Kenya

Body linear measurements, and specifically heart girth (HG), have been shown to be useful predictors of cattle liveweight. To test the accuracy of body linear measurements for predicting liveweight, crossbred dairy cattle of different genotypes were measured and weighed. A total of 352 mature cows and 100 heifers were weighed using an electronic weighing scale and measurements of HG, body length, height at withers were taken using an ordinary measuring tape and body condition scored (BCS) using a five-point scale. The animals were grouped according to genotype and age. Genotype classification was undertaken from farmer recall and by visual appraisal as 40–60, 61–80 or 81–100 % exotic (non-indigenous). Age classification was simply as mature cows or heifers. Liveweight of the animals ranged from 102 to 433 kg. Liveweight was strongly correlated with HG (r = 0.84) and body condition scores (r = 0.70) and moderately correlated with body length (r = 0.64) and height at withers (0.61). Regressing LW on HG measurements gave statistically significant (P < 0.01) equations with R² ranging from of 0.53 to 0.78 and residual standard deviation ranging from 18.11 to 40.50 kg. The overall model developed (adjusted R² = 0.71) had a prediction error of 26 kg (or 11 % of the mean) and predicted LW of over 95 % of crossbred dairy cattle in the range of 100–450 kg, regardless of age and breed group. Including BCS in the model slightly improved the model fit but not the prediction error. It was concluded that the model can be useful in making general management decisions in smallholder farms.

How many markers are enough? Factors influencing parentage testing in different livestock populations

Reliability of parentage test panels is usually based on its power to exclude wrong parentage assignments based on allele frequencies. We evaluated the rates of false exclusions and inclusions in parentage assignments, and how these results are affected by allele frequencies, panel sizes and the number of allowed mismatches. We also evaluated the reliability of parentage testing by comparing populations with distinct genetic backgrounds using pure and composite families of cattle and sheep. Allowing for 1% genotype mismatches in true parent-offspring relations provided the best compromise between false-positive and false-negative assignments. Pure breeds needed at least 200-210 single-nucleotide polymorphism (SNP) markers to correctly assign relations, but between 700 and 890 markers to avoid assigning incorrect relationships. Composite breeds needed between 220 (sheep) and 500 (cattle) markers for correct assignment; 680 (cattle) to 4400 (sheep) SNPs were needed to eliminate false-positive assignments. Allowing 0% genotype mismatches decreased false-positive but increased false-negative assignments, whilst a higher threshold of 2% showed the opposite effects. Panels with high minor allele frequencies (0.35-0.45) provided the best chance for correct parentage resolutions requiring fewer markers. Further, we propose that a dynamic threshold would allow adapting to population specific error rates. A comparison to the performance of the official International Society for Animal Genetics SNP panel for cattle and a recently published SNP panel for sheep showed that randomly selected markers performed only slightly worse for the applied parentage test based on opposing homozygotes. This suggests that even with carefully selected panels, only marginal assignment improvements are obtainable for a particular number of SNPs. The main point for improvement is the number of markers used. We recommend using at least 200 SNP markers for parentage testing if the aim is to reduce false-negative results. To fully exclude false positives at least 700 markers are required.

Performance of different SNP panels for parentage testing in two East Asian cattle breeds

The International Society for Animal Genetics (ISAG) proposed a panel of single nucleotide polymorphisms (SNPs) for parentage testing in cattle (a core panel of 100 SNPs and an additional list of 100 SNPs). However, markers specific to East Asian taurine cattle breeds were not included, and no information is available as to whether the ISAG panel performs adequately for these breeds. We tested ISAG's core (100 SNP) and full (200 SNP) panels on two East Asian taurine breeds: the Korean Hanwoo and the Japanese Wagyu, the latter from the Australian herd. Even though the power of exclusion was high at 0.99 for both ISAG panels, the core panel performed poorly with 3.01% false-positive assignments in the Hanwoo population and 3.57% in the Wagyu. The full ISAG panel identified all sire-offspring relations correctly in both populations with 0.02% of relations wrongly excluded in the Hanwoo population. Based on these results, we created and tested two population-specific marker panels: one for the Wagyu population, which showed no false-positive assignments with either 100 or 200 SNPs, and a second panel for the Hanwoo, which still had some false-positive assignments with 100 SNPs but no false positives using 200 SNPs. In conclusion, for parentage assignment in East Asian cattle breeds, only the full ISAG panel is adequate for parentage testing. If fewer markers should be used, it is advisable to use population-specific markers rather than the ISAG panel.

Genome-wide detection of signatures of selection in Korean Hanwoo cattle

The Korean Hanwoo cattle have been intensively selected for production traits, especially high intramuscular fat content. It is believed that ancient crossings between different breeds contributed to forming the Hanwoo, but little is known about the genomic differences and similarities between other cattle breeds and the Hanwoo. In this work, cattle breeds were grouped by origin into four types and used for comparisons: the Europeans (represented by six breeds), zebu (Nelore), African taurine (N'Dama) and Hanwoo. All animals had genotypes for around 680 000 SNPs after quality control of genotypes. Average heterozygosity was lower in Nelore and N'Dama (0.22 and 0.21 respectively) than in Europeans (0.26-0.31, with Shorthorn as outlier at 0.24) and Hanwoo (0.29). Pairwise FST analyses demonstrated that Hanwoo are more related to European cattle than to Nelore, with N'Dama in an intermediate position. This finding was corroborated by principal components and unsupervised hierarchical clustering. Using genome-wide smoothed FST , 55 genomic regions potentially under positive selection in Hanwoo were identified. Among these, 29 were regions also detected in previous studies. Twenty-four regions were exclusive to Hanwoo, and a number of other regions were shared with one or two of the other groups. These regions overlap a number of genes that are related to immune, reproduction and fatty acid metabolism pathways. Further analyses are needed to better characterize the ancestry of the Hanwoo cattle and to define the genes responsible to the identified selection peaks.

Hormonal growth implants affect feed efficiency and expression of residual feed intake-associated genes in beef cattle

Hormonal growth promotants (HGP) have been used to improve feed conversion ratio (FCR) and growth rates of cattle by modifying protein turnover rates. Residual feed intake (RFI) is correlated with FCR and has been adopted in Australia as a measure of feed efficiency in cattle for the purpose of genetic improvement. Eight genes (AHSG, GHR, GSTM1, INHBA, PCDH19, S100A10, SERPINI2 and SOD3) have been previously reported to be highly associated with RFI and could potentially be used to predict RFI in bulls and steers. In this study, expression levels of these genes in liver tissue of 46 cattle were measured by quantitative real-time PCR. These cattle were part of a larger tenderness gene marker experiment consisting of two breeds (Angus and Brahman); two sexes (steers and heifers) and HGP treatments (implanted vs control). Cattle were measured for growth, feed efficiency, body composition and carcass traits. Results showed the expression of these eight genes was significantly correlated with RFI. However, HGP treatment did not affect RFI or the expression of the RFI-associated genes. HGP treatment increased average daily gain by 20%, improved FCR by 18%, and increased rib eye-muscle area by 7.5%. HGP treatment was effective in improving growth rate, presumably by its known action in the protein turnover mechanism. This mechanism has been hypothesised as one of the regulators of RFI. Lack of effect of HGP treatment on RFI does not support this hypothesis.

Expression of candidate genes for residual feed intake in Angus cattle

Residual feed intake (RFI) has been adopted in Australia for the purpose of genetic improvement in feed efficiency in beef cattle. RFI is the difference between the observed feed intake of an animal and the predicted feed intake based on its size and growth rate over a test period. Gene expression of eight candidate genes (AHSG, GHR, GSTM1, INHBA, PCDH19, S100A10, SERPINI2 and SOD3), previously identified as differentially expressed between divergent lines of high- and low-RFI animals, was measured in an unselected population of 60 steers from the Angus Society Elite Progeny Test Program using quantitative real-time PCR. Results showed that the levels of gene expression were significantly correlated with RFI. The genes explain around 33.2% of the phenotypic variance in RFI, and prediction equations using the expression data are reasonably accurate estimators of RFI. The association of these genes with economically important traits, such as other feed efficiency-related traits and fat, growth and carcass traits, was investigated as well. The expression of these candidate genes was significantly correlated with feed conversion ratio and daily feed intake, which are highly associated with RFI, suggesting a functional role for these genes in modulating feed utilisation. The expression of these genes did not show any association with average daily gain, eye muscle area and carcass composition.

QTL and gene expression analyses identify genes affecting carcass weight and marbling on BTA14 in Hanwoo (Korean Cattle)

Causal mutations affecting quantitative trait variation can be good targets for marker-assisted selection for carcass traits in beef cattle. In this study, linkage and linkage disequilibrium analysis (LDLA) for four carcass traits was undertaken using 19 markers on bovine chromosome 14. The LDLA analysis detected quantitative trait loci (QTL) for carcass weight (CWT) and eye muscle area (EMA) at the same position at around 50 cM and surrounded by the markers FABP4SNP2774C>G and FABP4_μsat3237. The QTL for marbling (MAR) was identified at the midpoint of markers BMS4513 and RM137 in a 3.5-cM marker interval. The most likely position for a second QTL for CWT was found at the midpoint of tenth marker bracket (FABP4SNP2774C>G and FABP4_μsat3237). For this marker bracket, the total number of haplotypes was 34 with a most common frequency of 0.118. Effects of haplotypes on CWT varied from a -5-kg deviation for haplotype 6 to +8 kg for haplotype 23. To determine which genes contribute to the QTL effect, gene expression analysis was performed in muscle for a wide range of phenotypes. The results demonstrate that two genes, LOC781182 (p = 0.002) and TRPS1 (p = 0.006) were upregulated with increasing CWT and EMA, whereas only LOC614744 (p = 0.04) has a significant effect on intramuscular fat (IMF) content. Two genetic markers detected in FABP4 were the most likely QTL position in this QTL study, but FABP4 did not show a significant effect on both traits (CWT and EMA) in gene expression analysis. We conclude that three genes could be potential causal genes affecting carcass traits CWT, EMA, and IMF in Hanwoo.

Identification of quantitative trait loci affecting resistance to gastrointestinal parasites in a double backcross population of Red Maasai and Dorper sheep

A genome-wide scan for quantitative trait loci (QTL) affecting gastrointestinal nematode resistance in sheep was completed using a double backcross population derived from Red Maasai and Dorper ewes bred to F(1) rams. This design provided an opportunity to map potentially unique genetic variation associated with a parasite-tolerant breed like Red Maasai, a breed developed to survive East African grazing conditions. Parasite indicator phenotypes (blood packed cell volume - PCV and faecal egg count - FEC) were collected on a weekly basis from 1064 lambs during a single 3-month post-weaning grazing challenge on infected pastures. The averages of last measurements for FEC (AVFEC) and PCV (AVPCV), along with decline in PCV from challenge start to end (PCVD), were used to select lambs (N = 371) for genotyping that represented the tails (10% threshold) of the phenotypic distributions. Marker genotypes for 172 microsatellite loci covering 25 of 26 autosomes (1560.7 cm) were scored and corrected by Genoprob prior to qxpak analysis that included Box-Cox transformed AVFEC and arcsine transformed PCV statistics. Significant QTL for AVFEC and AVPCV were detected on four chromosomes, and this included a novel AVFEC QTL on chromosome 6 that would have remained undetected without Box-Cox transformation methods. The most significant P-values for AVFEC, AVPCV and PCVD overlapped the same marker interval on chromosome 22, suggesting the potential for a single causative mutation, which remains unknown. In all cases, the favourable QTL allele was always contributed from Red Maasai, providing support for the idea that future marker-assisted selection for genetic improvement of production in East Africa will rely on markers in linkage disequilibrium with these QTL.

Genetic polymorphisms of the bovine fatty acid binding protein 4 gene are significantly associated with marbling and carcass weight in Hanwoo (Korean Cattle)

The objective of this study was to investigate an association between polymorphisms in the FABP4 gene and phenotypic variation for marbling and carcass weight (CWT) in a population of Hanwoo steers. We re-sequenced 4.3 kb of the FABP4 gene region in 24 Hanwoo bulls and identified 16 SNPs and 1 microsatellite polymorphism. Of these 16 SNPs, three SNPs [g.2774G>C (intron I), g.3473A>T (intron II) and g.3631G>A (exon III, creating a p.Met >Val amino acid substitution)] were genotyped in 583 steers to assess their association with carcass traits. The g.3473A allele showed a significant increasing effect on CWT (P = 0.01) and the g.3631G allele was associated with higher marbling score (P = 0.006). One haplotype of these three SNPs (CAG) was significantly associated with CWT (P = 0.02) and marbling score (P = 0.05) and could potentially be of value for marker assisted selection in Hanwoo cattle. The CAG haplotype effect for CWT was larger (11.14 +/- 5.03 kg) than the largest single locus effect of g.3473A>T (5.01 +/- 2.2 kg).

Genetic variation in resistance to repeated infections with Heligmosomoides polygyrus bakeri, in inbred mouse strains selected for the mouse genome project

Since the publication of the mouse genome, attention has focused on the strains that were selected for sequencing. In this paper we report the results of experiments that characterized the response to infection with the murine gastrointestinal nematode Heligmosomoides polygyrus of eight new strains (A/J, C57BL/6, C3H, DBA/2, BALB/c, NIH, SJL and 129/J), in addition to the well-characterized CBA (poor responder) and SWR (strong responder) as our controls. We employed the repeated infection protocol (consisting of 7 superimposed doses of 125L3 each administered at weekly intervals, faecal egg counts in weeks 2, 4 and 6 and assessment of worm burdens in week 6) that was used successfully to identify quantitative trait loci for genes involved in resistance to H. polygyrus. SWR, SJL and NIH mice performed indistinguishably and are confirmed as strong responder strains to H. polygyrus. CBA, C3H and A/J mice all tolerated heavy infections and are assessed as poor responders. In contrast, DBA/2, 129/J and BALB/c mice performed variably between experiments, some tolerating heavy worm burdens comparable to those in poor responders, and some showing evidence of resistance, although only in one experiment with female 129/J females and one with female BALB/c was the pattern and extent of worm loss much like that in SWR mice. Because the genetic relationships between six of the strains exploited in this study are now well-understood, our results should enable analysis through single nucleotide polymorphisms and thereby provide more insight into the role of the genes that control resistance to H. polygyrus.

Combination of multiple microsatellite data sets to investigate genetic diversity and admixture of domestic cattle

Microsatellite markers are commonly used for population genetic analyses of livestock. However, up to now, combinations of microsatellite data sets or comparison of population genetic parameters from different studies and breeds has proven difficult. Often different genotyping methods have been employed, preventing standardization of microsatellite allele calling. In other cases different sets of markers have been genotyped, providing differing estimates of population genetic parameters. Here, we address these issues and illustrate a general two-step regression approach in cattle using three different sets of microsatellite data, to combine population genetics estimates of diversity and admixture. This regression-based method is independent of the loci genotyped but requires common breeds in the data sets. We show that combining microsatellite data sets can provide new insights on the origin and geographical distribution of genetic diversity and admixture in cattle, which will facilitate global management of this livestock species.

Use of molecular markers to enhance resistance of livestock to disease: a global approach

The improvement and utilisation of host genetic resistance to disease is an attractive option as a component of livestock disease control in a wide range of situations. This paper reviews the situations where genetic resistance of the host is likely to be a useful component of disease control and provides a framework for deciding whether genetic improvement of resistance is likely to be worthwhile. Discussion is focused on low-input production systems in the developing world, where disease resistance is particularly important. The authors propose an integrated strategy for the use of molecular markers in assessing genetic diversity and in utilising and improving host genetic resistance to disease. The integrated approach assures that there is value in the molecular genetic information whether or not it proves useful in genetic selection, a feature that should prove attractive to funding and executing agencies.

No detectable association of the ESR PvuII mutation with sow productivity in a Meishan x Large White F2 population

The polymorphism at the PvuII recognition site in the ESR gene showed no statistically significant association with sow productivity traits in a Meishan x Large White F2 population. Estimates of the effect on litter size were, however, in the opposite direction and statistically different from previously published estimates. Taken together with results from other publications, results here indicate that this PvuII polymorphism displays different degrees of linkage disequilibrium with a gene or genes controlling litter size in different populations.

Bayesian segregation analysis of somatic cell scores of Ontario Holstein cattle

Bayesian segregation analysis using a Gibbs sampling approach was applied to four sets of simulated data and one set of field data to detect evidence of major genes affecting the evaluated trait. The substitution effect of a major gene and its allelic frequency were estimated for each set of data. For two datasets simulated with a model with no major gene effect, the resulting estimates of polygenic variance and heritability agreed with the simulated values and tests for the presence of a major gene were not significant. Analyses of two sets of data simulated with a major gene produced posterior distributions that gave significant evidence of major gene effects but underestimated the substitution values of the major gene. The segregation analysis of field data suggested that a major gene significantly affected somatic cell score (SCS) in the population of Ontario Holstein cattle. The estimated heritability of SCS was approximately 0.16. The major gene variance accounted for about 17% of the total genetic variance and the point estimate of the frequency of the allele having a positive effect on SCS was 0.30. However, the precision of these estimates is questionable based on the simulation results. The effect of the major gene may be underestimated.

Multi-primer target PCR for rapid identification of bovine DRB3 alleles

Multi-primer target polymerase chain reaction (MPT-PCR) is a rapid method for the identification of specific BoLA-DRB3 alleles. In a single PCR reaction, the presence of two alleles associated with increased risk, DRB3.2*23 (DRB3*2701-2703, 2705-2707) and decreased risk, DRB3.2*16 (DRB3*1501, 1502), of mastitis in Canadian Holstein can be detected. Two outer primers amplify exon 2 of DRB3. Simultaneously, two inner, allele-specific primers amplify individual alleles. Initially, 40 cows previously typed by PCR-restriction fragment length polymorphism (PCR-RFLP) were genotyped using the multi-primer approach. An additional 30 cows were first genotyped by multi-primer target PCR, then by PCR-RFLP. All animals were correctly identified and there were no false positives. This technique can readily be modified to identify other BoLA alleles of interest.

Detection of quantitative trait loci affecting milk production traits on 10 chromosomes in Holstein cattle

Sons (n = 71 to 75) of each of six Holstein sires were genotyped at 69 microsatellite loci covering a total of 676 cM on chromosomes 3, 5, 9, 10, 13, 15, 17, 20, 23, and 26. Estimates of quantitative trait loci (QTL) effect and location were made using a least squares interval mapping approach based on daughter yield deviations of sons for 305 d milk, fat, and protein yield and fat and protein percentage. Thresholds for statistical significance of QTL effects were determined from interval mapping of 10,000 random permutations of the data across the bull sire families and within each sire family separately. Analyses combining data across sires indicated the presence of QTL affecting milk, fat, and protein yield on chromosomes 20 and 26 and a QTL affecting fat and protein percentage on chromosome 3. Analyses within each sire family separately indicated the presence of segregating QTL in at least one family on 7 of the 10 chromosomes. Statistically significant estimates of QTL effects on breeding value ranged from 438 to 658 kg of milk, from 17.4 to 24.9 kg of fat, 13.0 to 17.0 kg of protein, 0.04 to 0.17% fat, and 0.07 to 0.10% protein.

Responses to selection for growth and backfat in closed nucleus herds of Hampshire and Duroc pigs

A herd of purebred Hampshire and a herd of purebred Duroc pigs were created between 1987 and 1989 and subsequently selected on an index of growth rate and backfat from 1989 to 1995. Strict rules were put in place to promote rapid turnover of generations and allow only minimal culling for structural soundness. In 1990, 1991, 1993 and 1994, a number of sows coming from the Hampshire and Duroc selection lines were bred using frozen semen collected from a sample of control boars born in 1988 and 1989. Estimates of genetic trend based on single trait animal model analyses of age at 100 kg and backfat estimated rates of response to selection of approximately –1.4 and –0.4 days and –0.5 and –0.3 mm backfat per annum in Durocs and Hampshires, respectively. These were close to the original predictions based on the index employed for Durocs, but somewhat less than expected for Hampshires. Comparison of progeny of selected versus control boars was consistent with significant selection responses having been achieved, though estimates of the magnitude of the response had high standard errors. During the period of the trial, both the Hampshire and Duroc herds went from slightly below average on the Canadian National Genetic Evaluation Program to become the number one herds within their breed for the sire line index based on backfat and growth. The trial demonstrated that selection indexes based on estimated breeding values can be used effectively to achieve genetic progress. Key words: Pig breeding, selection index, estimated breeding values, selection response

Quantitative trait loci for upper thermal tolerance in outbred strains of rainbow trout (Oncorhynchus mykiss)

The expression of three putative QTL for upper thermal tolerance (UTT) was examined in two strains of outbred rainbow trout unselected for this trait using simple-sequence repeat (SSR; microsatellite) markers associated with UTT in backcrosses of lines selected on this trait. Two-way diallel lots in the third generation of an outbred pedigree were exposed to an acute thermal challenge. QTL detection was performed separately by each second-generation parent within each diallel lot, incorporating the effects of full sib families and correlated traits. Inheritance of different alleles at the SSR Ssa20.19NUIG from the sire 93-32-1 was strongly associated with the thermal tolerance of his half sib progeny, explaining 7.5% of their phenotypic variance in this trait. A hierarchical linear model incorporating allelic inheritance from all four grandsires of the experimental diallels (in addition to family specific and covariate trait effects) was also used to detect associations between the SSR and thermal tolerance in their third-generation grandprogeny. Ssa20.19NUIG was strongly associated with thermal tolerance in the grandprogeny of the grandsire G(0)SVM2. The generally stronger marker-trait associations found in male parents may be partially due to reduced chromosomal recombination rates in male salmonids compared to females. These results indicate the effects of a QTL on a fitness-related trait in unselected populations of rainbow trout.

The porcine gonadotropin-releasing hormone receptor gene (GNRHR): genomic organization, polymorphisms, and association with the number of corpora lutea

The interaction of gonadotropin-releasing hormone (GNRH) and its receptor (GNRHR) is critical in the endocrine regulation of reproduction. The gene (GNRHR) encoding the receptor has been mapped to porcine chromosome 8. There is evidence for three quantitative trait loci (QTL) influencing ovulation rate on this chromosome. We obtained an almost complete sequence (3993 bp, excluding intron 1) of the porcine GNRHR gene using PCR-based comparative genomic walking and inverse genomic walking approaches. Twelve polymorphisms were detected by sequencing of pooled DNA of Chinese Taihu and European Large White pigs, including 7 base substitutions and 5 insertions-deletions (indels). A F2 population of Meishan x European Large White pigs was genotyped for a TG indel in the promoter region, and a C/G substitution in the 3' UTR (untranslated region). A significant association of the C/G substitution with number of corpora lutea at first parity was observed.

Detection of QTL for milk production on Chromosomes 1 and 6 of Holstein cattle

Seventy to 75 sons of each of six Holstein sires were assayed for genotypes at a number of microsatellite loci spanning Chromosomes (Chrs) 1 and 6. The number of informative loci varied from three to eight on each chromosome in different sire families. Linkage order and map distance for microsatellite loci were estimated using CRI-MAP. Estimates of QTL effect and location were made by using a least squares interval mapping approach based on daughter yield deviations of sons for 305-d milk, fat, protein yield, and fat and protein percentage. Thresholds for statistical significance of QTL effects were determined from interval mapping of 10,000 random permutations of the data across the bull sire families and within each sire family separately. Across-sire analyses indicated a significant QTL for fat and protein yield, and fat percentage on Chr 1, and QTL effects on milk yield and protein percentage that might represent one or two QTL on Chr 6. Analyses within each sire family indicated significant QTL effects in five sire families, with one sire possibly being heterozygous for two QTLs. Statistically significant estimates of QTL effects on breeding value ranged from 340 to 640 kg of milk, from 15.6 to 28.4 kg of fat, and 14.4 to 17.6 kg of protein.

The effect of using different culling regimens on genetic response with two-trait, two-stage selection in a nucleus broiler stock

Stochastic simulation was used to study the effect on genetic response and inbreeding of various two-stage two-trait culling strategies. Four different parameter sets were considered for the two traits, BW and egg number. Selection of replacement animals was based on animal model best linear unbiased prediction (BLUP) to obtain estimated breeding values (EBV) at the second stage. Culling at Stage 1 was based on either animal model BLUP or phenotypes, and information from culled animals was either available or not available for calculation of second stage EBV. Besides founder individuals, six discrete generations were considered. Culling based on BLUP of two traits at Stage 1 produced higher response than culling on phenotypic evaluations. It was found that culling based on phenotypic evaluation and not carrying information to the second stage reduce rates of response by 9 to 17% and produced inbreeding higher than or close to that of BLUP selection. This study clearly shows that a double penalty of less response and higher inbreeding is generally paid for not using all information. Optimum selection schemes will depend on relative costs and benefits of collecting and processing the extra information required for full BLUP selection schemes.

Using recent versus complete pedigree data in genetic evaluation of a closed nucleus broiler line

Stochastic simulation was used to study the effect of using full data and pedigree structure vs more recent data and pedigree structure to obtain best linear unbiased predictors (BLUP) of breeding values for single trait selection. Simulations used heritabilities of 0.10 and 0.50, with a population structure of 20 sires each mated to two dams, each producing 10 progeny, with 11 hatches from an unselected base population under both discrete and overlapping generations. Selection of parents was based on BLUP of breeding values using an animal model. The use of the last two generations of data and pedigrees gave the same selection response as when using full data and pedigree structure, for both heritabilities. Under discrete generations with use of only the last generation data and pedigree, which is similar to phenotypic evaluation, response to selection decreased by 21 and 3.8% at Generation 10 compared to selection response when using the full data and pedigree for heritabilities of 0.10 and 0.50, respectively. Corresponding decreases in inbreeding were 72 and 37%. The amount of central processing unit time for genetic evaluation when using the last six, four, and two generations of data and pedigree was reduced to 70, 40, and 11% of that when using the full data set, for a heritability of 0.10 and discrete generations. Very similar values were observed for a heritability of 0.50 and also under overlapping generations.

Bi-PASA genotyping of a new polymorphism in the APOB gene shows no evidence for an association with fatness in pigs

Sequence analysis of PCR products of a 343-bp fragment from exon 29 of the porcine APOB gene of four Erhualian and four Landrace pigs revealed a missense G/C substitution at position of 6117 in this gene. Two allele-specific primers were designed to genotype this polymorphism using the Bi-PASA technique. Genotyping of 146 animals from Erhualian, Hampshire, Large White, Landrace and Duroc breeds revealed large breed differences in allele frequency. No association with fatness was observed within each of the four European breeds, where animals had been selected to be those with the highest and those with the lowest backfat depth at 100 kg liveweight from a large carcass dissection project.

Applying breeding objectives to dairy cattle improvement

Otherwise well-researched definitions of breeding objectives and selection criteria may never be used in practice if those definitions do not take into account the perceptions and wishes of the breeders for whom they are designed. Finding selection criteria that are widely accepted and implemented is a daunting task that requires considerable time and full interaction between the scientists and the industry. We review various aspects of this process and, in particular, how scientific principles can be used to ensure that the outcome best meets both the perceptions and needs of the users while remaining as close as possible to the technical economic optimum. Alternative methods of presenting and delivering selection indexes, such as index expression, index formulation, focus on response to selection rather than on index weights, construction of component indexes, and the use of direct accounting for costs of constraints rather than rescaling methods can all help in improving acceptance of an index. Development and implementation of selection criteria also involve consideration of the selective mating decisions that form an integral part of selection decisions in the field. The technical basis of factors that foster emphasis on individual mating decisions in the field are discussed in relation to formulation of the breeding goal and selection index and in relation to nonlinear economic and genetic parameters. Strategies that focus on use of a linear index for the selection of sires and dams followed by selective mating of selected parents have the greatest potential for implementation in the industry. We focus on examples taken from the Canadian dairy industry, but principles apply generally.

Defining multiple-trait objectives for sustainable genetic improvement

Genetic improvement is inherently a long-term process in which progress in the future is built upon improvement in the past. Discounting of future returns is often used in deriving economic values of traits under selection, but this gives a short-term perspective that is in conflict with the long-term nature of genetic improvement. Changes in management, market environment, and genetic potential over time can negatively affect the attainment of breeding goals. Nonlinear optimization techniques can be used to find optimum economic weights each year over any time horizon. Nearly optimal solutions can be found by deriving economic weights for a single, specified future date. Uncertainty about future production and marketing environments creates risk that might be lessened by maintaining or selecting for diverse genetic stocks that could be used in the future. Such programs may need to be coordinated internationally because they may be too expensive for individual companies to undertake. Consideration of risk and careful analyses of future technical and environmental conditions are needed to define multiple trait objectives for long-term genetic change.

The tyrosine-17 residue of Nef in SIVsmmPBj14 is required for acute pathogenesis and contributes to replication in macrophages

The variant simian immunodeficiency virus termed SIVsmmPBj14 induces a rapidly fatal disease in pig-tailed macaques. The acute pathogenic effects of this virus appear to be associated with at least two in vitro characteristics: the ability to induce lymphocyte proliferation; and the ability to replicate in unstimulated PBMC. Two of the amino acids in Nef of PBj14 (the No. 17 residue, tyrosine, and the No. 18 residue, glutamic acid) appear to be linked to the virus' ability to induce lymphocyte activation. To further study the effects of these amino acids on PBj14-induced pathogenesis, we generated two mutant viruses from our molecular clone, PBj6.6, containing either changes in both the No. 17 and No. 18 residues (termed PBj6.6YE-RQ), or a single change in the No. 17 residue (termed PBj6.6Y-R). In vitro analyses of these viruses showed that while their replicative abilities in stimulated peripheral blood mononuclear cells (PBMC) were altered, they still maintained the ability to replicate in unstimulated PBMC. Replication of these viruses in macrophage populations was impaired relative to the wild-type virus. Both mutant viruses were unable to induce proliferation of macaque PBMC in vitro. Virus derived from PBj6.6Y-R was unable to induce acute disease in macaques, but did maintain the ability to induce lymphopenia and intestinal lymphoid hyperplasia. These results show that the tyrosine-17 residue of Nef is linked to lymphocyte proliferation and disease development, but also suggest that the pathogenic characteristics of SIVsmmPBj14 are dependent upon multiple genetic determinants.

Associations between milk-protein production and reproduction, health, and culling

Associations between protein production and individual-cow reproductive performance, health, and culling were investigated in a 2-year observational study involving a convenience sample of 75 Ontario, 5 Alberta, and 3 Nova Scotia dairy farms. Protein production was defined by 305-day lactation protein yields and by estimated breeding values for protein yield. After controlling for the level of milk production, herd, parity, breed, and season of calving, there were no significant associations between either measure of protein production and days open or days to first breeding. The only associations between protein production and disease were small positive associations between the estimated breeding value for protein yield and cystic ovaries and mean lactation somatic cell count. The risk of culling, after controlling for the level of milk production, was negatively associated with previous-lactation 305-day protein yield for parity three animals only. The estimated breeding value for protein yield had a small negative association with the overall risk of culling, although the associations were not significant for individual lactations.

Effects of production and marketing circumstances on economic values for beef production traits

The effect of altering production and marketing circumstances on economic values is quantified for a complete beef production system. Absolute and relative economic values were found to vary substantially with large, but realistic fluctuations in prices and costs. In addition, several examples of different management and different genotypes gave markedly different economic values than in the base situation. Also investigated were the effects of rescaling the enterprise to accommodate three alternative limitations; fixed feed available from pasture, fixed dollars available for feed or fixed amount of beef produced. The effects of rescaling were highly dependent on whether or not fixed costs were accounted for. When fixed costs were ignored (corresponding to a small positive profit) the economic value for mature size decreased while that for fertility increased, but other traits were largely unaffected by rescaling. Overall, production circumstances that reduced survival and fertility yielded the largest changes to economic values. Key words: Economic values, beef cattle, rescaling

Economic values for beef production traits from a herd level bioeconomic model

A bioeconomic model of an integrated Canadian beef production system was developed to derive economic values for genetic improvement of multiple traits. The breeding objective was assumed to be profit maximization of the integrated enterprise. Sixteen input traits were identified as potentially influencing returns and costs in the system. These were mature size, direct and maternal calving ease (in heifers and cows separately), cow fertility, calf survival, cow survival, peak milk yield, residual post-weaning growth rate, residual feed intake in growing animals, residual feed intake in mature animals, residual slaughter weight and dressing percentage at constant backfat thickness, marbling and lean percentage. Most traits were defined to be functionally independent of each other. Thus, traits related to mature size were redefined as residual traits after accounting for the nonlinear relationships among mature size, growth and feed intake traits following mammalian size scaling rules. The base model, which incorporates average returns and costs under production and marketing systems typical of eastern Canada, is described. Economic values in the base model suggest that calf survival, fertility, residual feed intake, and dressing percentage are of primary economic importance in a purebreeding system. These traits also ranked highly in dam lines and (with the exception of fertility) in sire lines in terminal crossbreeding systems. Key words: Beef cattle, economic values, bioeconomic model

Associations between individual cow factors and milk-protein production

Associations between stage of lactation, cow characteristics, and protein production were evaluated using data from a 2-year period on 75 Ontario, 5 Alberta, and 3 Nova Scotia dairy farms. Individual-cow protein production was defined by 305-day protein yield and by the estimated breeding value for protein yield. Lactation curves for average daily protein yield were computed by parity, breed, and season of calving. Mean protein yield was highest in early lactation. However, there was no pronounced peak in daily protein yield. Parity was positively associated with 305-day protein yield and negatively associated with the estimated breeding values for protein yield. First-calf heifers had lower protein yields in early lactation and a slower rate of decline in protein yield in late lactation, as compared to later parity cows. Holstein cows had higher unadjusted protein yields and lower protein yields after adjusting for milk yield than other breeds. Holstein cows had significantly higher protein yields early in lactation compared to other breeds, but the rate of decline in protein production in late lactation was also greater. Season was associated with 305-day protein yield; the highest protein yields occurred in cows calving in the fall and winter months, but these cows had the greatest rate of decline in protein production in late lactation.

Estimation of variance of maternal lineage effects among Canadian Holsteins

Bayesian posterior estimates of variances of maternal lineage effects were obtained with a procedure that used Gibbs sampling. Data were records of yield during first parity from 245,510 Holstein cows that calved in Canada between 1990 and 1994. Maternal lineages were defined by tracing the maternal ancestry of cows to common female ancestors. Traits were standardized yields of milk, fat, and protein and percentages of fat and protein. Estimates of maternal lineage variance were < 0.5% of the total variance for all traits. Effects of this size hardly affected the cow rankings for estimated breeding value. Another analysis defined maternal lineages by establishing groups of dams and their daughters in an attempt to maximize the variance of environmental effects associated with maternal lineages, but only 1.1% of the variance in milk yield was associated with effects of the dam and daughter groups. Analyses of simulated data indicated that positive estimates of maternal lineage effects were not a result of restriction of estimates to positive values and that estimates of maternal lineage variance were biased slightly downward because of incomplete pedigree information. The partitioning of maternal lineage and additive genetic variance was correct when pedigree information was complete, according to results from simulation.

Effect of total allelic relationship on accuracy of evaluation and response to selection

Total allelic relationship (TA) as a possible alternative to the pedigree-derived additive genetic relationship (RA) is defined. The TA measures the actual allelic identity between individuals for loci segregating for the trait concerned. Its value was studied by simulation in populations of different family structure, different numbers of loci, different numbers of alleles per locus, and different heritability levels. The alternative types of relationship matrices were used in mixed model equations to derive best linear unbiased prediction estimates (EBV) of breeding values (BV). Accuracies of evaluations were calculated as correlations of EBV with true breeding values. In populations with random selection and mating, EBVTA derived using TA had higher accuracies than EBVRA derived using RA. In populations with selection, EBVTA was more accurate and resulted in higher responses than selection on EBVRA. We conclude that not accounting for variation in average measures of relationship and identity in state can be important sources of variance of prediction error, and taking account of them increases the accuracy of selection.

Realized sampling variances of estimates of genetic parameters and the difference between genetic and phenotypic correlations

A data set of 1572 heritability estimates and 1015 pairs of genetic and phenotypic correlation estimates, constructed from a survey of published beef cattle genetic parameter estimates, provided a rare opportunity to study realized sampling variances of genetic parameter estimates. The distribution of both heritability estimates and genetic correlation estimates, when plotted against estimated accuracy, was consistent with random error variance being some three times the sampling variance predicted from standard formulae. This result was consistent with the observation that the variance of estimates of heritabilities and genetic correlations between populations were about four times the predicted sampling variance, suggesting few real differences in genetic parameters between populations. Except where there was a strong biological or statistical expectation of a difference, there was little evidence for differences between genetic and phenotypic correlations for most trait combinations or for differences in genetic correlations between populations. These results suggest that, even for controlled populations, estimating genetic parameters specific to a given population is less useful than commonly believed. A serendipitous discovery was that, in the standard formula for theoretical standard error of a genetic correlation estimate, the heritabilities refer to the estimated values and not, as seems generally assumed, the true population values.

Estimation of additive genetic variance in commercial layer poultry and simulated populations under selection

Changes in genetic parameters over generations for a selected commercial population and simulated populations of poultry with different sizes were studied. The traits analyzed from the commercial population were rate of lay, age at first egg, egg weight, deformation, and body weight. In the simulated population, a trait measured on both sexes and a sex-limited trait, measured only on one sex, each with a heritability of 0.1 and 0.5, were analyzed. In the commercial and simulated populations, males and females were selected on the basis of family selection indexes and data was available only after many generations of selection. Parameters for each generation were estimated by fitting an animal model using derivative free maximum likelihood (DFREML) with different data structures. In structure 1, data included the given (base) generation for which the parameters were to be estimated, and all subsequent generations. In structure 2, only data on birds in the given generation and their progeny were included. In both structures, parents of base-generation birds were assumed unrelated and pedigrees traced back to these parents. With commercial data using structure 1, estimates of σ a (2) and h(2) decreased by 14 to 37% across five generations. With structure 2, no trends were observed, though estimates were lower than for structure 1. For simulated data, with a heritability of 0.1, both structures yielded apparently unbiased estimates of the observed additive genetic variances in the (selected) base generation, no matter how many generations of data were utilized, for both sex-limited and normal traits. However, with a heritability of 0.5 the estimated additive genetic variance for both types of trait decreased with a decrease in the number of generations used in the estimation. Estimates based on the first two generations underestimated, while estimates based on five generations of data overestimated, the observed genetic variances in the defined base. The combinations of conditions that lead to varying degrees of bias remain undefined.

Including genetic relationships in selection decisions: alternative methodologies

Investigations are made of variations in an iterative methodology previously introduced for reducing inbreeding by including genetic relationships in selection decisions, using adjusted estimated breeding values (EBV). An alternative computing strategy for maximising the value of the population selection criterion is shown to involve less computation, which results in function values as great or greater than the original method. Alteration of weights for different types of relationships in the adjusted EBV has no detectable effect on genetic gain at a given level of inbreeding. Selection using the adjusted EBV method in one sex and truncation on EBV in the other sex results in less genetic gain at a given level of inbreeding than using adjusted EBV in both sexes, but results in more gain at a given level of inbreeding than three selection strategies that do not include genetic relationships in selection decisions. The advantage of the adjusted EBV method over these three methods is retained when selection is for a sexlimited trait.

Comparison of index selection and best linear unbiased prediction for simulated layer poultry data

The advantage of using best linear unbiased prediction (BLUP) of breeding value over different selection indexes was examined for a sex-limited trait in a simulated layer poultry population. The base breeding population consisted of 30 males and 300 females that were unrelated to each other. Heritability for different analyses was assumed to be either .1, .2, or .5. Each generation was reproduced from two hatches each year, with a hatch variance of 3.165% of the phenotypic variance, except for one simulation, in which it was assumed to be 40% to test the effect of a large fixed effect. Parents were selected on 1) BLUP of breeding value, 2) optimum selection index (individual, full-, and half-sibs), 3) classical selection index (as for optimum, but index weight constant across generations), 4) reduced selection index (individual and full-sibs only), or 5) combination of classical and BLUP. The relative selection response with the selection indices compared to the BLUP estimates (except for the reduced selection index) were from 94.5 to 99.4% of BLUP. Inbreeding was higher in the BLUP selected populations, which could offset any advantage of BLUP if the populations were structured so that inbreeding could rise too rapidly.

Balancing selection response and rate of inbreeding by including genetic relationships in selection decisions

An iterative selection strategy, based on estimated breeding values (EBV) and average relationship among selected individuals, is proposed to optimise the balance between genetic response and inbreeding. Stochastic simulation was used to compare rates of inbreeding and genetic gain with those of other strategies. For a range of heritabilities, population sizes and mating ratios, the iterative strategy, denoted ADJEBV, outperforms other strategies, giving the greatest genetic gain at a given rate of inbreeding and the least breeding at a given genetic gain. Where selection is currently by truncation on the EBV, with a restriction on the number of full-sibs selected, it should be possible to maintain similar levels of genetic gain and inbreeding with a reduction in population size of 10-30%, by changing to the iterative strategy. If performance is measured by the reduction in cumulative inbreeding without losing more than a given amount of genetic gain relative to results obtained under truncation selection on the EBV, then with the EBV based on a family index, the performance of ADJEBV is greater at low heritability, and is generally greater than where EBV are based on individual records. When comparisons of genetic response and inbreeding are made for alternative breeding scheme designs, schemes which give higher genetic gain within acceptable inbreeding levels would usually be favoured. If comparisons are made on this basis, then the selection method used should be ADJEBV, which maximises the genetic gain for a given level of inbreeding. The results indicated that all selection strategies used to reduce inbreeding had very small effects on the variance of gain, and so differences in this respect are unlikely to affect choices among selection strategies. Selection criteria are recommended based on maximising a selection objective which specifies the desired balance between genetic gain and inbreeding.

Genetic improvement of production while maintaining fitness

Selection for production tends to decrease fitness, in particular, major components such as reproductive performance. Under an infinitesimal genetic model restricted index selection can maintain reproductive performance while improving production. However, reproductive traits are thought to be controlled by a finite number of recessive alleles at low frequency. Culling for low reproduction may weed out the negative homozygous genotypes for reproduction in any generation, thus controlling the frequencies of alleles negative for reproduction. Restricted index selection, culling for low reproduction and a new method called empirical restricted index selection were compared for their efficiency in improving production while maintaining reproduction. Empirical restricted index selection selects animals that have on average the highest estimated breeding values for production and on average the same estimated breeding values for reproduction as the base population. An infinitesimal genetic model and models with a finite number of loci for reproduction with rare deleterious recessive alleles, which have additive, dominant or no pleiotropic effects on production, were considered. When reproduction was controlled by a finite number of loci with rare recessive alleles, restricted index selection could not maintain reproduction. The culling of 20% of the animals on reproduction maintained reproduction with all genetic models, except for the model where loci for reproduction had additive effects on production. Empirical restricted selection maintained reproduction with all models and yielded higher production responses than culling on reproduction, except when there were dominant pleiotropic effects on production.