This thesis aimed to define a suitable definition of ewe longevity for the Australian sheep industry and determine the merit of incorporating the trait into the Australian sheep breeding objectives. Therefore, the fundamental requirements for incorporating ewe longevity in the Australian sheep breeding objectives, the genetic parameters and economic values were estimated for ewe longevity and stayability traits followed by an estimation of response to selection.

The first part of this thesis deals with the data exploration of the MERINOSELECT and LAMBPLAN maternal databases within the Sheep Genetics and the estimation of genetic parameters for the ewe longevity and stay ability traits in Merino and maternal breeds. The majority of the flocks submitting data to the MERINOSELECT and LAMBPLAN maternal databases do not have sufficient recording patterns to derive longevity. However, the contemporary groups with regular recording patterns were selected. The contemporary groups were defined as the site × flock × year of birth. These contemporary groups of ewes with regular recording patterns within the MERINOSELECT database were; 1) born since the year 2000, 2) had spent a minimum of 3 years in the flock, 3) had their own annual weight (weaning, postweaning or yearling) or wool record and reproduction record (from 2 yrs) up to 6 years of age, 4) contained at least 30 ewes, and 5) at least 70% of the animals were assigned a sire (chapter 3).The ewe longevity or time in flock (TIF) was defined as the period between birth and the last available production record. The stay ability traits were defined as the presence of a ewe in flock up to certain periods of time. The heritability estimates of the ewe's longevity and stayability traits were moderate if not corrected for the ewes’ production and reproductive performance. However, after correcting for these traits, the ewe's longevity and stayability traits were lowly heritable. The correlation between the ewes’ longevity and stayability traits was strong.

The correlation between ewe longevity and production and reproduction traits was estimated via a series of bivariate analyses. The analysed production traits were weaning weight (wwt), post-weaning weight (pwt), post-weaning C-site fat (pcf), post-weaning eye muscle depth (pemd), post-weaning faecal egg count (pfec), yearling weight (ywt), yearling Csite fat (ycf), yearling eye muscle depth (yemd), yearling faecal egg count (yfec), yearling greasy fleece weight (ygfw), yearling fibre diameter (yfd), adult greasy fleece weight (agfw) and adult fibre diameter (afd). The reproductive traits analysed were fertility (fert), litter size (ls), number of lambs born (nlb), ewe rearing ability (era) and number of lambs weaned (nlw). The ewes’ TIF was lowly heritable and correlated to the production and reproduction traits. Therefore, a breeding objective was to be formulated that considers longevity as an objective trait, which requires calculating the economic value of the ewes’ TIF trait.

The second part of this thesis deals with the estimation of economic value and response to selection. The economic value of the ewes’ TIF was large across the fine wool Merino, dual purpose Merino and maternal production systems. Ewe longevity has a positive correlation with the current breeding objectives suggesting that selection on the current breeding objectives will improve ewe longevity across the three production systems within the Australian sheep industry. However, including longevity trait in the breeding objectives will further increase the overall genetic gain, and particularly improve genetic gain for longevity. In the maternal production system, the genetic gain of the growth and carcase traits slows down a little after including the longevity trait in the breeding scenarios. The results showed a 2 to 3% increase in the total dollar response across the three production systems after incorporating the ewes’ TIF trait into the breeding objectives. However, incorporating genomic information of TIF into breeding objectives increased the overall response by 13 to 16% across three production systems. The results suggest that selection based on the current breeding objectives will improve ewe longevity within the Merino and maternal production systems but noticeably higher genetic gain can be attained if the genomic information of ewes’ TIF is incorporated into the breeding objectives.

The final chapter discusses the research findings and concludes with recommendations for future research areas. These recommendations include important encouragements like improving the data quality, the importance of recording culling reasons, estimating accurate genetic parameters of ewe longevity and the potential of incorporating the ewe longevity as a trait in the Australian sheep breeding objectives to achieve higher genetic gain. This thesis contributes significantly to define ewe longevity and using the genetic parameters in the Australian sheep breeding objectives.

In the last decades, multiple methods using molecular genetic data have been developed and their application in breeding programs have been evaluated. Particularly, due to the development and cost of dense genotyping arrays (SNP chips) and sequencing technologies (whole genome and RNA sequencing), these techniques have become standard methods to study the association between genetic variants and phenotypic variation in important traits in livestock (‘genomics’) as well as levels of expression for contrasting groups in age of development or phenotype (‘transcriptomics’). This thesis explores the use of transcriptomics and genomics applied to better understand economically important complex traits in beef cattle (marbling and residual feed intake) as well as the utility of using information generated by these technologies in the estimation of breeding values.

In this thesis, in chapter 3, we describe the characterization of muscle development and the progress of fat deposition at the transcriptomic level in Hanwoo cattle. This Korean breed has the genetic potential to accumulate intramuscular fat to reach very high levels of marbling and high prices in the market. To examine the key genes and pathways that regulate the differentiation process in satellite cells, we extracted these cells from Longissimus dorsi (LD) and semimembranosus (SM) of three newborn calves, promote cell differentiation in culture cells and evaluated the differences between muscles in a time-series RNA-seq experiment. The histological (differentiation index) results indicated that LD muscle differentiated faster from myoblast into multinucleated myotubes than SM. These results agreed with the gene expression of the myogenic regulatory factors (MRF) which tend to be significantly up-regulated at the end of the differentiation in LD, specifically the genes MYOD, MYF6 and MYOG. The number of genes differentially expressed was larger across time than across muscles. In total, thirteen genes (HOXB2, HOXB4, HOXB9, HOXC8, FOXD1, IGFN1, ZIC2, ZIC4, HOXA11, HOXC11, PITX1, SIM2 and TBX4) were differentially expressed (DE) between muscles, which seem to be involved in modulating the muscle lineage development during myogenesis. In addition, our results indicated, in agreement with previous studies in other species, that some of the DE genes modulated the expression of myogenic regulatory factors (MYOD and MYF5) during the differentiation process.

The use of RNA-seq on the marbling development of Hanwoo helped to better understand variation in gene expression related to high or low marbling phenotypes. In chapter 4, we describe an experiment where muscle samples from Longissimus dorsi were studied at the age of 18 (by biopsy) and 30 months. Twelve animals were grouped according to their marbling score in Low (average 2.4, range from 1-5), and High (average 6.28, range from 6-9). In total, 1,883 differentially expressed genes were identified from multiple contrasts, among them 782 genes were up-regulated and 1,101 were down-regulated. Differences in transcriptome were higher between ages rather than between marbling groups. The genes SLC38A4, ABCA10, APOL6, and two novel genes (ENSBTAG00000015330, ENSBTAG00000046041) were up-regulated in the High marbling group at 18 months of age. These genes are likely to have important roles in energy transport and utilization during growth of steers. Potential markers for marbling development (LEP, MEDAG, FOXO1, ADIG, ADIPOQ, CMKLR1, and FABP4) were identified from the functional analysis as involved in regulation of fat cell differentiation or brown fat cell differentiation. These results imply a potential use of gene expression technologies to identify younger steers that will develop high marbling. Further functional studies would need to be conducted to better understand the role of these genes on marbling.

The combination of multiple omic technologies opens up the possibility to improve the interpretation of a trait from many approaches. The combination of expression studies with mapping quantitative trait loci was applied in chapter 5 for studying the genetic architecture of residual feed intake (RFI) in 2190 Angus steers. First, the imputation from low density to medium density and later to high-density genotyping arrays was performed for 2,190 animals using a larger population of Angus with genotypes as reference. Additionally, the RNA sequences from 126 Angus cattle divergently selected for RFI were analyzed in a multi-tissue experiment (from liver, blood and muscle). The estimated heritability for RFI was 0.3 and we identified 78 SNPs associated with RFI on six QTL located on BTA1, BTA6, BTA14, BTA17, BTA20 and BTA26. The most significant SNP was on chromosome BTA20 (rs42662073) for which STC2 was the closest gene. The genes OAS2, SHOX, XKR4, and SGMS1 were the closest to the significant QTL on BTA17, BTA1, BTA14, and BTA26, respectively. In the 2 Mb windows around the six significant QTL, we identified fifteen genes whose expression was significantly associated with RFI selection line: NEURL1B, CPEB4, RITA1, CCDC42B, OAS2, RPL6, ERP29, ATP6V1H, MRPL15, MFSD1, RARRES1, A1CF, SGMS1, PAPSS2, and PTEN. The results imply that the integration of GWAS and gene expression analysis may help to contribute with knowledge and help to understand better genetic variation in complex traits like RFI.

The ultimate goal in livestock breeding programs is always to find the most accurate way to select breeding animals according to a defined breeding objective and one way to achieve this could be to apply genomic selection. The final study in this thesis (chapter 6) attempts to investigate the utility of integrating information from GWAS or gene expression studies into increasing the accuracy of genomic prediction of RFI. We used the results from chapter 5 and evaluated the accuracy of genomic prediction using data on 2,190 Angus steers with two cross-validation designs; one where the GWAS was performed on the same data used for training the genomic prediction named four cross-validation (4CV) and a design where GWAS and training sets were separate data denoted as four by four cross validation (4x4CV). The accuracy of prediction of RFI did not improve when using a 770k SNP panel compared with 50k SNP panel. There were 1.2% and 2.7% point increase in accuracy when top SNPs from GWAS were added to the 50k or 770k panels in the 4x4CV design. The 4CV design showed lower accuracy when using top SNPs and the predictions were much more biased. Genomic prediction accuracy can be slightly improved when using selected SNPs from GWAS and GSA. Further analysis would need to be conducted using a larger population to confirm these results.

Finally, the thesis ends with some thoughts on the application of “omics” in livestock breeding programs, the impact on accuracy of prediction and the improvement of experimental design. This chapter features a final summary and concluding remarks about the research involved in producing this thesis as well as notes on the interpretation of complex trait research and reflections future research directions regarding the use of gene expression and gene association to improve livestock production systems.

This thesis aimed to lay a foundation for a dairy cattle breeding programme in tropical Sri Lanka using temperate dairy breeds. The fundamental requirements for establishing a dairy breeding programme were studied, i.e. genetic parameters and economic values were estimated, and response to selection was predicted for milk production, udder health, and fertility traits in imported Jersey and Jersey-Friesian crossbred cows in Sri Lanka. Heritability estimates of predicted and realized 305-day milk yield traits ranged from 0.02 ± 0.01 to 0.09 ± 0.02, and heritability estimates for daily milk yield records ranged from 0.002 ± 0.05 to 0.19 ± 0.02 in first lactation. These heritability estimates were low mainly due to the low additive genetic variance. The correlations between the estimated breeding values for 305-day milk yield for bulls in Australia and bull estimated breeding values based on their progeny in Sri Lanka were 0.39 in Jersey cows and -0.35 in Jersey-Friesian cows, suggesting that importing genetic material from Australia to the population in Sri Lanka is not an effective route for genetic improvement. All mastitis traits had a zero heritability. Milk electrical conductivity and milk flow rate are often used as potential indicator traits in selection for mastitis resistance. In this study, milk electrical conductivity and milk flowrate were lowly to moderately heritable with heritability estimates ranging from 0.02 ± 0.01 to 0.11± 0.03 and from 0.02 ± 0.01 to 0.14 ± 0.04. Mastitis had no phenotypic association with milk electrical conductivity and only lowly negative phenotypic correlations with milk flowrate (range -0.17 to 0) demonstrating that milk electrical conductivity or milk flow rate were not good indicators of clinical mastitis in this study. The reproductive traits analysed were the interval from first calving to first service, the interval from the first service to conception, the interval from first calving to conception (days open), the interval between first calving to second calving, gestation length, number of services per conception for second calving and stillbirths. The heritability estimates for fertility traits ranged from 0.01 ± 0.01 to 0.05 ± 0.02, and they were sufficiently heritable to be used in a breeding programme. Genetic improvement of milk yield, age at first calving, number of services per conception, calving interval, and number of mastitis episodes had a positive impact on the farm profitability while genetic improvement of fat and protein yields was not profitable due to high feed cost relative to the additional returns. Percentage change of the traits relative to the current mean after one year of selection for annual milk yield, annual fat yield, annual protein yield, age at first calving, number of services per conception, calving interval, and mastitis episodes were 0.40, 0.34, 0.67, 0.34, 0.31, -0.18 and 0.21, respectively.

Favourable responses to selection were observed for milk, fat, and protein yields, and calving interval. Genetic improvement of milk, fertility, and mastitis traits should be included in the breeding objective for temperate dairy breeds in large-scale intensive dairy farms in Sri Lanka. Proposed structure and function for implementing a single across-herd genetic evaluation for temperate dairy cattle in Sri Lanka, and recommendations for routine data collection were provided.

This thesis examines the methods to improve the productivity of a dual-purpose native chicken breed in Thailand. Five generations of body weights and egg production data on Lueng Hang Kao Kabinburi (LHKK) chickens were used for estimation of genetic parameters in chapters 3 and 4. Traits considered were body weights, measured at four-weekly intervals from hatch to 24 weeks of age (BW1D, BW4, BW8, BW12, BW16, BW20, and BW24), body weight at first egg (BWFE), age at first egg (AFE), egg weight at first egg (EWFE), and egg number (EN).There were 11,588 chickens from 486 cocks and 1,461 hens that had records for growth and egg production traits. Relationships between growth rate and egg production traits were also explored. The level of inbreeding and its effect on growth and egg production were explored. In chapter 5, hatchability (HAT), rate of lay (RL), average daily gain (ADG), and survival rate (SUR) were identified as economically important traits and were used in a selection index to optimize the genetic response to selection. Finally, in chapter 6, population sizes and mating ratios were explored to optimize the genetic response in HAT, RL, ADG and SUR while minimizing the rate of inbreeding in the nucleus flock of LHKK chickens by simulating 20 generations of chickens for recurrent selection.

Univariate and bivariate analysis were used to estimate genetic parameters in body weight and egg production traits. Fixed effects of year and hatch within year were significant for all traits and sex was significant for all body weight traits, except for BWFE. The direct additive genetic effect was significant for all traits and the maternal genetic effect was significant only for growth traits, except BWFE. The maternal permanent environmental effect was significant for all growth traits, except for BW24 and BWFE. The estimates of heritability for direct additive genetic effect ranged from 0.10 to 0.47 for body weight traits and ranged from 0.15 to 0.16 for egg production traits. High positive genetic correlations were estimated between all traits, except for negative genetic correlations between EN and other traits. Inbreeding effect on growth and egg production traits was not significant, except for BW1D, where BW1D reduced by 0.09 g when the rate of inbreeding increased by 1% per generation.

Univariate random regression model was used to estimate genetic parameters for body weight traits along the growth trajectory from BW1D to BW24. A quadratic Legendre polynomial was identified as the best model to estimate variance structure for all random effects fitting heterogeneous residual variances based on six growth periods. Heritability estimates ranged from 0.34 to 0.54 and 0.04 to 0.06 for direct additive and maternal genetic effects, respectively. Estimated variance ratios ranged from 0.19 to 0.48 and 0.10 to 0.12 for direct and maternal permanent environmental effects, respectively. All genetic correlations between body weight traits were high and positive.

Genetic relationships between growth rates, measured at four-weekly intervals, and AFE and EWFE were estimated using bivariate analysis. Estimated heritabilities for growth rates ranged from 0.06 to 0.28. Estimated heritabilities for AFE and EWFW were 0.24 and 0.16, respectively. Genetic correlations between growth rates and AFE ranged from -0.22 to 0.02, and between growth rate and EWFE ranged from -0.05 to 0.40. The result suggested that selecting chicken with a high growth rate at an early growth period (at 28 days of age) would improve the body weight and egg weight at sexual maturity while reducing the age at sexual maturity.

Breeding strategies to improve the meat and egg production of the LHKK chickens under intensive (IPS) and extensive production system (EPS) were explored. A bioeconomic model was developed to calculate economic weights for HAT, RL, ADG and SUR. Estimated economic weights and the response to selection showed that LHKK chicken production was economically viable under the IPS and the EPS in Thailand. Annual economic return per hen in the EPS (621THB) was higher than the annual economic return per hen in the IPS (132THB) due to the lower cost of production under the EPS than the IPS. Calculated combined economic weights from the IPS and the EPS were 21.01THB for HAT, 56.52THB for RL, 106.52THB for ADG, and 15.76THB for SUR. The estimated relative economic weights showed that RL was the most important economic trait in LHKK chicken production. Decreased feed price and increased fattening chicken price is expected to increase the monetary return from the native chicken farms. Using the multi-trait selection index resulted in predicted responses of 0.97% in HAT, 2.41% in RL, 1.38g in ADG and 0.73% in SUR. Thus, implementing a single breeding objective strategy in the nucleus flock of the LHKK chickens will improve the productivity under both IPS and EPS.

A stochastic simulation using an optimal contribution selection (OCS) approach with a target of 25º for mate selection was used to minimize the level of inbreeding and maximize genetic gain in LHKK nucleus flock. The level of inbreeding and genetic gain for a range of population sizes and mating ratios were compared after 20 generations of recurrent selection. The predicted level of inbreeding indicated that increasing the population size by 30% from the current population size would reduce the level of inbreeding in LHKK flock by around 1% per generation. Reducing the nucleus size by 30%, in comparison to the current nucleus size, will increase the level of inbreeding by 1.55% per generation. The highest level of inbreeding was found in the higher mating ratios of one cock to floating hens (one to 10 hens) and the lowest level of inbreeding was found in the lowest mating ratio of one cock to three hens. The predicted genetic responses obtained across the four mating ratios were not significantly different to each other.

In summary, this study found that the growth and egg production traits in LHKK chicken are heritable and also have high genetic correlations between them. Therefore, both meat and egg production of LHKK chickens can be improved by implementing a multiple traits selection strategy. Furthermore, implementing a common selection strategy in the nucleus flock of LHKK chickens is expected to increase the monetary return under an intensive and extensive production system in Thailand. However, the effects of using a system specific selection strategy for IPS and EPS on the profitability under each production system need to be explored. Increasing the current nucleus flock size and reducing the mating ratio will fulfil the main objective of the current LHKK breeding program by further improving the meat and egg production, while maintaining the breed characteristics of LHKK chickens.

The conversion of feed into usable products, also known as feed efficiency (FE), is important given the necessity to increase quality food production. This concept is also important given the environmental sustainability and profitability of the beef production systems. The present thesis analysed different aspects involving FE and their inclusion in beef cattle breeding programs. Thus, this work aimed to increase the understanding of genetic variability of the FE traits and the inclusion of residual feed intake (RFI) into genomic assisted beef cattle breeding programs.

The first research chapter was oriented to better understand the genetic variability of the main FE traits and their association with growth and meat quality traits in an Angus population. The analysed FE component traits were daily weight gain, metabolic midweight, average of daily, feed intake, feed conversion ratio, RFI, and residual gain (RG), evaluated during a 70-day feedlot test period. In this chapter, it was concluded that selection on RFI would have a negative impact in growth and carcass traits. Therefore, it was suggested that selection only for RFI would have negative impacts on growth and meat quality in the studied population.

The second research chapter compared repeatability (REPM) and random regression models (RRM) on feed efficiency component traits during the feedlot test period. Unlike the REPM, the RRM can accommodate changes in parameters of those traits over duration of the test period. First-order RRM applied to body weight (BW) and average daily feed intake (ADFI), shown that genetic parameters tend to change during the feedlot period. By comparing these models, it was concluded that ignoring the change in parameters, regardless of feed costs, resulted in a loss of selection response of approximately 3%

The third research chapter analysed REPM and RRM when using genomic information. Genomic variants associated with BW and ADFI variation were identified and a change of their effect during the feedlot test period was evaluated. For both traits, RRM presented the best fit and only one genomic region was detected with a constant effect throughout the 70-d feedlot test period. For BW and ADFI, the strongest associated variants were rs43350564 and rs109326204, located on chromosomes 20 and 5, respectively. These identified SNP may help to unravel the biology of FE and can be used for more accurate genomic prediction of breeding values.

The final research chapter evaluated various realistic multi-trait selection strategies incorporating RFI, as well as including the use of genomic selection (GS). Here it was concluded that selecting on RFI via a genomic test yielded the largest increase in selection accuracy and overall responses of 0.48 and 64.9%, respectively. Finally, it was concluded that including feed efficiency increased the $ net return without compromising meat quality and cow condition score in the beef cattle breeding programs.

Genetic evaluation of Australian sheep is conducted by Sheep Genetics Australia (SGA) for millions of animals for more than 100 traits. Sheep Genetics uses the Australian sheep genetic analysis software (OVIS), which applies a pre-adjustment of phenotypes for systematic environmental effects rather than fitting all environmental effects (fixed effects) and genetic effects (random effects) jointly in a linear mixed model to estimate breeding values. This thesis aims to compare different methods of genetic evaluation for weight and carcase traits for the Australian sheep industry with a specific emphasis on different methods of accounting for systematic environmental effects. In the first experimental chapter (Chapter 3), various methods of correcting systematic environmental effects were investigated using early body weight phenotypes in White Suffolk and Poll Dorset sheep breed. This was comprised of comparing currently available OVIS pre-adjustment factors for fixed effects with updated pre-adjustment factors, and testing a range of additional interactions between fixed effects. Correlations between EBVs obtained from different models and the regression slopes from forward prediction were used as best model selection criteria. Results showed that the updated pre-adjustment factors produced a slightly better regression slope of progeny performance on sire estimated breeding values (EBVs) than current OVIS pre-adjustment factors: 0.40 and 0.38 versus 0.37 and 0.35 for weaning weight and post-weaning weight, respectively, when averaged over two breeds. Analysis with a linear mixed model produced a significant improvement (P < 0.05) in the regression slopes (0.47 and 0.44) compared to analysis based on pre-adjusted phenotypes. A linear mixed model with flock by sex by age interaction produced better regression slopes (0.48 and 0.46) than the current linear mixed model without this interaction term. Results indicated that flock by sex by age interaction should be included in the linear mixed model for sheep genetic evaluation of early body weight traits in Australia.

This thesis explores the genetic variation of carcass and meat quality traits in beef cattle. Estimation of genetic parameters including heritability, genetic and phenotypic correlations is the first step in the genetic evaluation process to understand the nature of quantitative traits. Subsequently, the estimated parameters are required for the establishing of a selection program in livestock. Alongside with performance data, pedigree information is essential to estimate the relationship between animals more accurately in the conventional breeding program. In beef cattle, carcass traits cannot be recorded on selection candidates and therefore time-consuming progeny tests are often used to gain selection accuracy. However, the discovery of genomic information has enabled to select high merit individuals at an early age without scarifying the selection candidate. Furthermore, the availability of high-density SNP panels and whole genome sequence data has improved the selection accuracy of high merit individuals. Therefore, the general aim of this thesis was to understand the genetic variability of carcass and meat quality traits using pedigree and genomic information in beef cattle.

The first experiment of this thesis explored the genetic variation of carcass and meat quality traits in Hanwoo beef cattle using pedigree information. The phenotypic data were collected from 469,002 Hanwoo beef cattle raised at 3646 farms in the Republic of South Korea. The studied carcass traits were carcass weight, eye muscle area, back fat thickness, body weight, and meat index. In addition, the studied meat quality traits included marbling score, meat colour, fat colours and meat texture. Carcass traits, including carcass weight, eye muscle area, back fat thickness and marbling score showed high genetic variation and moderate to high heritability in the Hanwoo beef cattle population. However, the study also revealed that carcass weight and eye muscle area traits showed low and negative (unfavourable) genetic associations with meat texture, meat and fat color traits. Low heritabilities were observed for meat and fat colour traits, however, the observed moderate and positive genetic correlations among meat texture, meat and fat colour traits suggests that these traits can be jointly improved in a breeding program of beef cattle. In this study, the genetic and phenotypic correlations between carcass and meat quality traits were low in general, indicating that these traits are independent and require careful application in selection schemes. In conclusion, the estimates of genetic parameters in this study could be useful for designing breeding programs to improve various carcass and meat quality traits in Hanwoo cattle.

The second experiment was designed to examine the drawback of the data obtained from slaughterhouses that were used in the first experiment. Therefore, the second experiment assesses bias due to sorting of animals based on body weight for the genetic evaluation of carcass traits using simulation data. Various sources of bias in genetic evaluation including parental selection, sequential selection, culling of animals before records, and misclassification or manipulation of contemporary groups have been discussed widely in the literature. We hypothesized that the sorting of animals into different contemporary groups based on their yearling weight had an impact on the genetic evaluation of this trait and other correlated traits. The experiment aimed to observe the impact of sorting animals into different contemporary groups based on an early measured trait and then examine the effect on the genetic evaluation of subsequent measured traits. Our result showed that when animals are sorted based on yearling weight leads to biased estimated breeding values in genetic evaluation of carcass traits. The magnitude of the bias in the estimated breeding values that was observed in the current study varied with heritability, and genetic and residual correlations between the simulated traits. The current result demonstrated that the detected sorting biases were stronger when higher genetic and residual correlations were allocated to the simulated traits. However, the observed sorting bias in the univariate model was accounted for by multi-trait evaluation methods. In addition, a slight decrease of bias in estimated breeding values was observed when carcass weight was fitted as a linear covariate in the model for the genetic evaluation of subcutaneous fat depths at the 12th/13th rib (CRIB) trait. Overall, the current simulation study provides insights into how the genetic architecture of studied traits affects the genetic evaluation of animals.

The third experiment explored the genetic architecture underlying genetic variability of meat quality traits through the analysis of a genome-wide association study (GWAS) in Hanwoo beef cattle. Genome-wide association studies using common SNP chips have revealed many quantitative trait loci for various production traits such as growth, efficiency, carcass and meat quality traits; however, GWAS using whole genome sequence data are scarce in beef cattle. The current GWAS was conducted using whole-genome sequence data from 2110 Hanwoo beef cattle recorded for marbling score, meat texture, meat and fat color traits. The study identified several chromosomal regions on various chromosomes that contained 107 significant SNPs associated with meat quality traits. Four QTL regions were identified on BTA2, 12, 16, and 24 for marbling score and two QTL regions were found for meat texture trait on BTA12 and 29. Similarly, three QTL regions were identified for meat color on BTA2, 14 and 24 and five QTL regions for fat color on BTA7, 10, 12, 16, and 21. Candidate genes were identified for all studied traits, and their potential influence on the given trait was discussed.

The fourth and fifth experiments examined factors that influence the genomic prediction accuracy in beef cattle. Prediction of the breeding values based on information on DNA of the animals is changing breeding strategies and saving time and costs in a breeding program of beef production. The fourth experiment mainly focused on the impact of the relationship between reference and test population was examined on the accuracy of genomic prediction using distantly versus closely related animals in the reference and test populations. The result showed that when the animals in the reference and test populations were closely related, the prediction accuracy was higher than when the animals were related distantly.

The fifth experiment assessed the effect of SNP densities including 50K, 777K (HD), whole genome sequence (WGS)) and preselected SNP on the accuracy of genomic prediction. The results showed that similar prediction accuracies were observed across all SNP densities. Small sample size in genomic prediction is a limiting factor to capitalize the benefit of using WGS data since the effect of causal mutations on quantitative traits cannot be accurately estimated. Additionally, high-density markers and WGS data may not help to improve the prediction accuracy in a breed with small effective population size such as Hanwoo beef cattle used in the current study. Depending on the SNP selection methods, zero to 5% improvement of genomic prediction accuracy was gained due to the inclusion of SNPs that were significantly associated with the studied traits, as detected in the GWAS previously. Similarly, different magnitudes of bias were observed in the genomic breeding values depending on the SNP selection methods used in the study. Potential reasons for the observed bias due to the inclusion of preselected SNPs have been discussed and found in other relevant literatures. Overall, the study shows that marbling score and meat texture traits had higher genomic prediction accuracy in all scenarios, suggesting that genomic selection for these traits may contribute well to the genetic improvement of meat quality in Hanwoo beef cattle.

Carcase value is predominantly based on hot carcase weight and fat depth without considering price variation in different primal cuts. This research aimed to develop selection strategies to improve carcase value by including valuable primal cuts into existing breeding objectives of pigs and beef. Therefore, this thesis focused on the fundamental requirements, i.e. estimation of genetic parameters for primal cuts and linear or area measurements of live pigs as selection criteria for primal cuts. Further economic values for different primal cuts were derived and predicted genetic responses to evaluate different selection strategies in breeding objectives. Primal cuts were expressed as both weight and percentage traits. Significant exploitable genetic variability in individual primal cuts or groups of primal cuts at a fixed carcass weight was evident for pigs and beef. Heritabilities for primal cuts of pigs were low to moderate. The strongest negative genetic correlation was found between leg and belly primal cuts. Linear and area measurements of pigs were lowly to moderately heritable. Genetic correlations between linear or area measurements and primal cuts indicated that the area measurements were significant selection criteria for all primal cuts in pigs. Beef primal cuts were moderate to highly heritable. Two different primal groups were also derived in beef including high-valued cuts (HVC) and low-valued cuts (LVC), where HVC was highly heritable. Primal cut traits were included in the breeding objectives as a percentage trait rather than weight trait to keep the traits independent of carcase weight. An approach was derived to estimate economic values directly for primal cut traits based on an independent model relevant for primal cuts. In pig breeding objectives, primal cuts were included based on two approaches, either as loin and belly primals separately or as a middle primal. Inclusion of middle primal only was considered better because of the higher response than the inclusion of loin and belly separately in breeding objective. In beef breeding objectives, HVC was included as a breeding objective trait based on two different production systems representing the domestic and Japanese markets. There was a higher response to selection for the Japanese index than the domestic market as the production system for the Japanese market was based on higher carcase values and feed prices. Additional responses were generated for both pig and beef breeding objectives by including valued primal cuts as breeding objective traits. Therefore, expanding knowledge of primal cut traits and including them in breeding programs for both pigs and beef offers new opportunities to improve carcase value in the livestock industry.

The success to implement and conduct a breeding program necessarily requires a good design and a definition of a breeding objective. In addition, accurate pedigree, data quality and performance recording are essential to strengthen the accuracy of the estimated breeding values to establish an effective selection program. The overall objective of this dissertation was to develop strategies to design a breeding program to improve productivity and profitability in the Mexican sheep industry. Data sets from Hampshire, Katahdin and Pelibuey sheep breeds were used. Genetic parameters for early growth traits were estimated from a standard animal model, applying twelve statistical models (base models and sire by flock interaction) using Restricted Maximum Likelihood. The heritability for growth traits was higher in the base models than with sire by flock but reasonably consistent with estimates presented in a range of previous studies. Both direct genetic and phenotypic correlations for base models and sire by flock were positive and moderate to very high. The validation study showed reasonable predictability of sire EBVs, where it is better to fit sire by flock. Validation analyses showed that when using models accounting for sire by flock interaction, sire EBVs predicted progeny performance in line with expectation. Estimated genetic trends showed Mexican sheep breeders have made genetic progress even without a formal evaluation system. The impact of genotype by environment interaction (G by E) on breeding values across the diverse Mexican regions for the Katahdin sheep resulted in an average genetic correlation across regions and traits of 0.55, which could also be managed in a genetic evaluation model by fitting sire by flock interactions. In reproduction analyses for Katahdin sheep, litter size was shown to be heritable (0.05), with small to moderate correlations with body weight traits (0.15 – 0.33). An economic model to derive breeding objectives was developed and applied in selection indexes combining growth, reproduction and carcass traits to predict genetic responses in 10 years relevant to the commercial production system in the Hampshire and Katahdin breeds. The highest influence among the traits included in the selection index was for litter size, followed by growth traits, carcass and adult weight (which had a negative economic value due to the importance of feed costs in the breeding ewe flock). In conclusion, the Mexican sheep industry has the potential and the basic structure to implement these results to increase productivity and profitability. The industry could move in a favourable direction by implementing the results of this study in a modern genetic evaluation system.

Variation in the genetic environmental sensitivity (GES) of livestock can cause genotype by environment interactions (G×E). The impacts of G×E differs depending on whether genotypes express different sensitivities across or within environments. Across environments G×E is caused by macro-GES, while within environments G×E is caused by micro-GES.

Estimation of GES is challenging especially in unbalanced datasets. The number of animals in each macro-environment and the degree of genetic connection across macro-environments both influence the estimation accuracy of genetic variance due to macro-GES. Meanwhile, it has been suggested that balanced datasets with relatively large sire family sizes are required to accurately estimate micro-GES of single recorded traits.

The aim of this thesis was to assess the data structure requirements for estimation of macro and micro-GES in unbalanced data, evaluate the accuracy of modelling micro-GES on one trait in multi-trait models, estimate the relationship between health-related traits and micro-GES of production traits, examine the interaction between macro- and micro-GES, and estimate the magnitude of macro- and micro-GES in livestock.

The data structure requirements for estimation of macro- and micro-GES in unbalanced data, was evaluated using a simulation study in Chapter 3. It was shown that the accuracies and bias of estimated variance components for simultaneous estimation of macro- and micro-GES using double hierarchical generalised linear models (DHGLMs) including a linear reaction norm depended primarily on average sire family size. Accurate and unbiased estimates variance components and EBVs of macro- and micro-GES could be obtained with a dataset with 500 sires with 20 offspring per sire on average.

The impact of differences in the number of records on the accuracy of variance component estimation when analysing multiple traits of which one exhibit micro-GES was assessed in Chapter 5. The genetic correlations were found to be slightly overestimated when the true genetic correlations were 0.5. However, the models were accurately able to identify the presence of non-zero genetic correlations, showing that these models could provide useful information.

The relationship between health-related traits and production traits were examined in Chapter 6 by estimating the genetic correlation between immune competence traits and mean performance and micro-GES of weaning weight, eye muscle area and rib and rump fat depth. It was shown that animals with high immune competence tended to also have high mean performance and micro-GES of rib and rump fat and low mean performance and micro-GES of weaning weight and eye muscle area.

The interaction between macro- and micro-GES of body weight in two subpopulations of the same cross reared in Burkina Faso and France was assessed in Chapter 7. Micro-GES of body weight showed considerable macro-GES with both heterogeneity of heritabilities and reranking between the two subpopulations.

The existence of macro-GES and micro-GES were found for yearling weight of Australian Angus beef cattle and body weight of purebred and crossbred broiler chicken. Furthermore, micro-GES was found in weaning weight, eye muscle area and rib and rump fat in Australian Angus beef cattle.

Survival of growing pigs through to slaughter age is not only a key driver of profitability but also has implications for animal welfare. The primary objectives of this study were to 1) investigate the non-genetic and genetic factors that influence individual piglet pre- and post-weaning mortality, 2) gain a better understanding of the genetic correlation between purebred and crossbred mortality traits, and 3) explore other avenues for decreasing mortality through selective breeding, such as the relationship between immunity and survival outcomes. In this study a piglet was recorded as a pre-weaning death (PREw) if it was born alive and died up until, and including, the day of weaning (0= alive, 1= dead). A post-weaning death (POSTw) was recorded if the piglet had been weaned and was less than 70 days of age at death and is often referred to as nursery mortality. Piglets born still-born were excluded from examination in this study.

To address objective one, it was hypothesised that both non-genetic and genetic factors were significant contributors to piglet mortality outcomes and in the presence of fostering activities, a nurse sow model would be more informative than a biological dam model. The factors that influence individual piglet pre- and post-weaning mortality traits (non-genetic animal factors in Chapter 3 and genetic factors in Chapter 4) were investigated using a dataset that included purebred and crossbred piglets (total N = 614,573), recorded between March 2009 and December 2019. The availability of a large dataset over a long period of time has allowed the investigations of factors into survival that have been poorly quantified in previous studies, often due to a low number of animals recorded.

Chapter 3 analysed mortality records which included datasets over two time-periods, March 2009 to March 2011 and January 2017 to December 2018. The results presented in this chapter firstly identified that total born (TB) had a strong linear association with piglet birthweight (PBWT). The average piglet birthweight decreased by 37.3 ± 0.0003 grams per piglet for every piglet increase in TB. However, non-linear relationships were evident between PBWT or TB and piglet mortality. For example, a curvilinear relationship was evident between PBWT and PREw up until the fourth decile, after which reductions in mortality were more linear. The effects of TB for PREw was greatly reduced when the effect of PBWT was simultaneously fitted in the model, suggesting that much of the effect of litter size on mortality is a consequence of the effects of litter size on piglet birthweight. From this study it was also evident that successful genetic strategies to increase total born does not universally increase mortality. High total born and low mortality are possible as demonstrated between the time-periods outlined above. Individual piglet birthweight (PBWT) was the main factor associated with the ability of individual piglets to survive. Other factors identified to have significant impacts on survival were piglet genotype, gestation length, biological dam and nurse sow parities, fostering status of the piglet, piglet gender and weaning age. Although these effects on mortality were relatively small, they are generally not known under normal commercial production, and could provide valuable information for management decisions. For example, supporting litters who have been born before 114 days of gestation or to very young or old sows, fostering lightweight piglets who have not established a teat within the first day of life, or avoiding the weaning of individuals before 21 days of age, even when weight based criterion might suggest they are ready to wean.

Following the identification of systematic effects in Chapter 3, alternative models for genetic evaluation of the pre- and post-weaning mortality traits were investigated in Chapter 4. For pre-weaning mortality, the best linear model accounted for direct piglet effects, common litter effects of both the nurse sow and biological dam, repeated records of the nurse sow and the maternal nurse sow genetic effects, in preference to a comparable model based on these effects represented by the biological dam only. For post-weaning mortality, the most parsimonious linear model included only direct piglet effects along with the common litter effects of both the nurse sow and biological dam. After accounting for systematic effects, genes of the piglet contributed to both pre- and post-weaning mortality (direct h²: 0.02 ± 0.002 for PREw and POSTw), whereas the nurse sow genes only contributed to pre-weaning mortality (m²: 0.01 ± 0.002). The heritability and variation obtained from sire threshold models (PREw h²: 0.08 ± 0.007 and POSTw h²: 0.07 ± 0.007) were higher, suggesting the linear animal model estimates were biased downwards and that genetic improvements can be made at both the direct and maternal levels

To address objective two, it was hypothesised that genetic correlations between purebred and crossbred mortality would be high when they are recorded in the same production environment and are represented through some common parents. Chapter 5 explored the relationships between purebred and crossbred PREw and POSTw, where bivariate analyses were conducted to estimate genetic correlations. The dataset used for this study had records for purebred and crossbred animals, who were recorded together on the same farm and experiencing similar production conditions, which is very rare to find in the literature for any trait. As the incidence of mortality was similar between purebred and crossbred PREw, the estimated variance components were also similar. Likewise, variance component estimates also reflected the incidence of mortality for POSTw, as crossbred mortality was approximately half that of purebreds. The estimates of the additive genetic correlations between purebred and crossbred performances were high (PREw r_a: 0.78 ± 0.097 and POSTw r_a: 0.94 ± 0.112), indicating that purebred and crossbred mortality traits are controlled by the same genes. The high correlations, within this population and environment, demonstrate that survival traits of crossbred pigs can be improved using genetic evaluation and selection based on purebred records.

To address objective three, it was hypothesised that the immune phenotype measures of mature boars would be associated with survival of their progeny reared in commercial environments, based on the assumptions that immune phenotypes are both variable and heritable. The ability to link immune phenotypes and progeny survival is novel as there is very little literature published, due to the number of progeny required to conduct this type of research. A pilot study conducted on nine mature boars, presented in Chapter 6, found that simple testing procedures could be developed which used commercially available vaccines, containing tetanus toxoid as a model antigen. Vaccination was used to induce measurable immune responses that can be used to assess the immune phenotype measures of mature boars. Using these procedures, a further 87 boars were assessed for their ability to mount antibody (Ab-IR) and cell (Cell-IR) mediated immune responses. The associations between boar Ab-IR and Cell-IR immune phenotypes with their own estimated breeding values (EBVs) for direct (sPREd) and maternal (sPREm) components of preweaning mortality and direct post-weaning (sPOSTd) mortality were tested. These EBVs were estimated from accurate evaluation of the progeny these boars sired and are presented in this thesis as survival percentages. Also, the ability of these boars daughters to rear progeny in their first litter and the ability of their progeny to mount their own antibody mediated immune response were verified.

The results presented in Chapter 7 showed that as sire Cell-IR responses increased, there was an increase in independent estimates of breeding values for sPREd and sPOSTd survival. It was unexpected that there was no association between Ab-IR and sPOSTd based on sire EBVs. Results demonstrated that variation in immune phenotype measures of boars were associated with the survival of their progeny, reared in commercial environments. Results also suggested that CellIR phenotype had a greater influence on sPREd and sPOSTd than Ab-IR in the animals studied. Further to this, an enhanced boar immune phenotype was reflected in a higher maternal performance of daughters in their first parity by improving their progeny survival and therefore the number of piglets weaned. And finally, immune phenotypes of sires was associated with the ability of their progeny to mount an antibody mediated immune response to the same model antigen. These results provide encouraging evidence that the model developed for immune phenotype testing was informative and more extensive testing of selection candidates should be carried out to obtain estimates of genetic parameters and understand the impact on other economically important traits. This would require assessment of suitable testing protocols for selection candidates.

In conclusion the research conducted in this thesis has shown that progeny survival is the outcome of complex non-genetic and genetic interactions between the piglet, the biological dam, the nurse sow and the environment, including management decisions such as fostering. Although heritabilities for direct and maternal effects on survival are low, there is potential for genetic improvement to be delivered to the commercial tier of the breeding pyramid due to high correlations between purebred and crossbred animals, independent of other important traits such as birthweight, gestation length and litter size. Finally, selection for immune phenotype measures has the potential to further improve survival of growing pigs.

Supplementing ruminants with dietary nitrate (NO₃) is an effective methane mitigation strategy if it can be managed so as to not expose ruminants to any risk of clinical nitrite (NO₂) toxicity. The objective of this thesis was firstly to deepen the understanding for NO₃ metabolism in sheep and secondly to develop practical strategies to reducing risk of NO₂ toxicity in sheep supplemented with dietary NO₃.

It has been previously established, that in the rumen NO₃ is reduced to NO₂ and then to NH₃, and that supplementing with excessive amounts of NO₃ can expose ruminants to NO₂ toxicity due to the absorption of NO₂. This thesis reports a series of five investigations of NO₃ metabolism by sheep and identifies:

Nitrate, like urea, is ‘recycled’ within the ruminant. Transfer of ruminal ¹⁵NO₃^--N into the blood and transfer of blood NO₂-N into the rumen being quantified. Only 20% of rumen NO₃^-and 30% of blood NO₂^- were recovered in urine.

That in hourly fed sheep approximately 90% of dietary NO₃^- was rapidly converted to NH₃ in the rumen, with the remainder leaving the rumen by absorption into the bloodstream or passage to the lower gastro-intestinal tract.

Within the rumen, the conversion of NO₃^-to NH₃ is neither simple nor complete. In vitro and in-vivo studies showed NO₃^-is reduced to gaseous nitrous oxide (N₂O) and N₂O may be further metabolised to N₂ gas by the rumen microbiota. Approximately 0.04% and 3.0% of dosed NO₃^--N was recovered over 10 h from sheep as N₂O and N₂ respectively, and this was not affected by whether sheep had prior adaption to NO₃^- or not, identifying denitrification as a reaction not previously reported from the rumen.

From this understanding and a review of the literature on ruminant NO₃ metabolism, eight critical control points for reducing the risk of nitrite toxicity (methaemoglobinaemia), were identified and the potential for manipulating five of these evaluated.

Reducing the rate at which NO₃ became available to the rumen biota by coating calcium nitrate with paraffin wax significantly reduced blood methaemoglobin level (MetHb; an indicator of NO₂ toxicity) in sheep supplemented with NO₃.

The extent of methaemoglobinaemia could also be reduced by the daily ration being consumed at shorter intervals rather than in a single bout, and this established that feed management is pivotal to safe feeding of NO₃^-containing diets.

Enhancing the rumen’s capacity to reduce potentially toxic NO₂ ^-by supplying Propionibactericum acidicpropionici as a direct fed microbial was ineffective in reducing blood MetHb or NO₂^-concentration of sheep fed NO₃^- supplemented diets.

Attempts to increase the rate of removal of NO₂^-from the rumen by providing a substrate (glycerol) to stimulate NADH availability in the rumen, and accelerate the nitrite reductase enzyme system did not reduce the concentration of NO₂ in incubations of rumen contents supplemented with NO₃^-.

We found no evidence that adapting sheep to dietary NO₃^- protected them against NO₂^- toxicity. Indeed, in vitro more NO₂^- accumulated in incubation when donors where adapted to dietary NO₃^-. Also, no signs of reduced MetHb were noticed after several weeks of NO₃^-supplementation in vivo.

Other critical control points such as regulating microbial uptake of NO₃ and ruminal absorption of NO₃ and NO₂ were unable to be assessed in this thesis.

The studies reported here also confirmed the practical impacts of NO₃ as an effective supplement for reducing enteric methane emissions and increasing wool growth of sheep. As well as providing a better understanding of NO₃^-metabolism, studies also showed that the greenhouse gas (GHG) abatement impact of methane mitigation may be partly offset by an associated production of the potent GHG, N₂O. Discovery of the production of N₂O and N₂ from NO₃^-in the rumen and identification of recycling of blood NO₂^- to the rumen has expanded our understanding of NO₃^-metabolism. Coating NO₃^-to decrease the rapidity of NO₃^- release in the rumen as a strategy to reduce NO₂ toxicity was effective but needs further investigation. The applicability of feed grade NO₃^-as a commercially available feed additive will also depend on the cost of NO₃ and the additional cost of the technology to ensure its safe feeding, compared to the cheaper alternative non-protein nitrogen source, urea.

Optimisation of pig breeding programs aims to increase the genetic gain in pig populations and to decrease the rate of inbreeding in the pig nucleus population. Genomic selection is a potential breeding strategy that can increase genetic gain and is also expected to decrease the rate of inbreeding in livestock breeding programs. Pigs are selected based on multiple correlated traits in the nucleus population. It might be difficult to improve response to selection in favourable direction for individual breeding objective traits because of an interplay between complex correlation structure and the economic value of each trait. On top of that, genomic selection might also shift genetic gain towards hard-to-measure traits. More work is needed on how genomic selection benefits the overall merit of breeding objectives and individual breeding objective traits.

Post-weaning survival (PWS) is an important breeding objective trait in the sire line of pigs. The benefits of genomic selection for PWS depend on the structure of the reference population, which should have both genotypes and phenotypes. Animal breeders might not be interested in genotyping dead pigs because dead pigs cannot be selection candidates. However, genotyping dead and live pigs might increase the genetic gain for PWS in comparison to genotyping live animals alone. While improving genetic gain, it is also important to reduce the rate of inbreeding because pigs are selected in a closed elite herd. Genomic markers might also increase genetic diversity because genomic relationships are more accurate than pedigree relationships. With the availability of genomic marker information, it is also easier to account for the dominance effect than pedigree information in the genetic evaluation model in the presence of dominance. Therefore, the broad objective of this thesis was to investigate the benefits of using genomic selection in pig breeding programs. This thesis explored multiple new avenues of using genomic information to increase the rate of genetic gain and decrease the rate of inbreeding in pig breeding programs.

In chapter 3, a premise was tested that the overall pig breeding objective achieves additional genetic gain in genomic selection compared to pedigree selection, but some traits achieve larger additional genetic gain than other traits. Results in a deterministic simulation study showed that genomic selection scenarios based on different sizes of reference populations increased overall response in the breeding objectives by 9% to 56% and 3.5% to 27% in the dam and sire lines, respectively, compared to pedigree selection. In the dam line of pigs, reproductive traits such as sow mature weight, number born alive, and sow longevity achieved 123% to 403%, 73 % to 351%, and 58% to 278% larger genetic gain in genomic selection compared to the pedigree selection respectively. In comparison, backfat thickness, average daily gain, and feed conversion ratio achieved 6% to 14%, 4% to 11%, and 7% to 9% smaller genetic gain in genomic selection than pedigree selection, respectively. In the sire line of pigs, post-weaning survival, drip loss, and middle portion percentage achieved larger genetic gain in genomic selection than the pedigree selection. Achieving larger genetic gain for reproduction traits in the dam line and post-weaning survival and meat and carcass quality traits in the sire line increased the overall merit of pig breeding objectives in genomic selection compared to the pedigree selection.

In chapter 4, a premise was tested that genotyping both live and dead animals realises more genetic gain for PWS (assuming a PWS of 90%) in pigs compared to the scenario where only live animals are genotyped. Stochastic simulation was conducted to compare genetic gain in the scenarios of either genotyping live and dead animals or genotyping live animals only. Genetic gain for PWS in these genotyping strategies was compared at 1% pedigree inbreeding in optimum contribution selection. Results showed that genetic gain with genotyping all live animals was 52% higher than pedigree selection. On top of that, genetic gain with genotyping live and 20 to 100% of dead animals was 14 to 33% higher than genotyping only live animals. Genotyping live and dead animals increased the accuracy of the genomic breeding values of live animals compared to genotyping only live animals, which resulted in a larger genetic gain for PWS.

In chapter 5, a premise was tested that optimum-contribution selection with genomic relationships using only low MAF (minor allele frequency) markers below a predefined threshold to control inbreeding realises less rate of true inbreeding than optimum-contribution selection (OCS) with pedigree relationships. Genetic gain in genomic and pedigree-based OCS was fixed at a predefined value while comparing the rate of inbreeding. Results showed that pedigree-based OCS realised a lower rate of inbreeding than genomic-based OCS at the same rate of genetic gain. Genomic-based OCS fixed more favourable quantitative trait loci than pedigree-based OCS. In addition, genomic-based OCS selected more closely related animals than pedigree-based OCS. Therefore, pedigree-based OCS realised a lower rate of inbreeding than genomic-based OCS at the same rate of genetic gain.

Finally, in chapter 6, a premise was tested that genetic gain in pig breeding programs using dominance models that accounted for both random additive genetic and dominance effects was higher than additive models that included only random additive genetic effects under the presence of dominance. The stochastic simulation was conducted to compare models in thedam and sire line of pigs. In the sire line, similar additive genetic variances were estimated by the two models but with the additive model, the litter and residual variances were biased upward by 42% and 23%, respectively. When the model did not include a common litter effect in the dam line, the additive genetic variance was 10% smaller in comparison to the additive genetic variance estimated using the dominance model. Despite overestimating variance components using additive models, both models realised a similar rate of total true genetic gain. Since animals were selected based on additive genetic merit, the dominance model did not impact the rate of total true genetic gain. Therefore, the additive genetic model can be used for estimating breeding values if animals are selected based on additive genetic merit, even under the presence of dominance.

The results mentioned above showed the potential of genomic selection to increase genetic gain in pig breeding programs. This study investigated multiple new avenues of using genomic information for the genetic improvement in pigs. However, there are still many unanswered question. Use of genomics is beneficial for improving the accuracy of selection and genetic gain, it is not clear how to use genomics to control inbreeding. To take the advantage of genomics, more work is needed to investigate how to use genomics to control inbreeding. In addition, genomics can be useful for accounting non-additive genetic effects such as dominance and epistasis in the model. As more research emerges, use of genomics will be more useful for optimising the pig breeding programs because genomics opens up further opportunities to reveal the biology of traits.

A total of 1103 primi- and multiparous sows from two nucleus farms were recorded in late spring (N=558 sows) of 2017 and late summer (N=545 sows) of 2018. These sows were recorded for a range of health and welfare traits from the transfer to farrowing shed until weaning. Data obtained from electronic sow feeders during the gestation period were used to generate feed intake and feeding behavior traits and establish possible associations with health and welfare traits. These data were recorded in 2015, from 2847 predominantly (90.5%) F1 sows and from 540 pure bred sows from two locations.

The aim of this project was to determine the most useful measures for identifying sows with the risk of undesirable outcome during the gestation period and around farrowing through to weaning, for the purpose of establishing a potential monitoring system to improve management of at risk sows. The second aim was to identify which of these traits were heritable and potentially useful in a breeding program context. The third aim was to evaluate the implications of variation in genetic merit in reproductive or performance traits for health and welfare outcomes.

Using multivariate logistic regression, several predictors were related to undesirable farrowing and lactation outcomes, along with higher rate of premature removals. Predictors relating to body condition and ability to fit into the crate, appetite of sows, udder health (mastitis and injured or regressed teats) and physiological state of sows (respiration rate and haemoglobin levels) appeared to be the most informative to identify sows with higher risk of undesirable outcomes. Early detection of sows with urinary tract infection and the following treatment may reduce forced removals. The results obtained from the electronic sow feeder data (ESF) showed that cumulative missed or low intake feeding events were informative to predict the risk of undesirable outcomes. In addition, while sows in this study were considered to be healthy, correlations between ESF data and subsequent health and welfare outcomes suggested that ESF data can be useful for identifying sows which require further attention.

In the genetic analysis of health and welfare traits it was demonstrated that traits related to appetite of sows (feed refusals before farrowing), udder health (mastitis, un-suckled and regressed teats), physiological state (respiration rate and rectal temperature) and size of the sow (caliper score, crate fit and teat access) were moderately heritable. Along with adjusting management practices, selection for traits related to health and welfare could potentially improve production performances and decrease forced removals. Variation in ESF phenotypes amongst individual sows reflected significant genetic variation in some of these traits. Traits related to feeding behaviour of gestating sows were moderately heritable, whereas heritability for feed intake itself was low to negligible. Thus, individual phenotypes constructed from ESF data could be useful for genetic evaluation purposes, but equivalent capabilities were not available for both ESF systems used in this study.

This thesis investigates and develops breeding strategies to maximise genetic improvement of dairy cattle in different production systems in Kenya. The research questions central to the study were: (1) is there heterogeneity of variance and genotype by environment interaction between the dairy production systems in Kenya?; (2) is the relative economic importance of the breeding objectives traits the same in different production systems?; (3) which selection strategies maximise genetic gain in the overall breeding goal of dairy production systems in Kenya.

Dairy herds in Kenya vary considerably based on the level of input and output, and in the use of breeds and crosses in different systems. Data from multi-breed cows from Dairy Recoding Services of Kenya (54,775 records) were used in this study to classify environments based on production level and evaluate the performance of different genotypes within and across these environments. Herds were grouped into low, medium and high production system environments using mean 305 days milk yield and the K-means clustering method. An animal model was used to estimate variance components and genetic parameters for milk production and fertility traits within and between production systems. Genetic groups were fitted to account for the multi-breed cows and their crosses. To account for the small herd sizes contemporary group effects were fitted as random effects. Genetic correlations between traits under different production systems were used to determine the presence of genotype by environment interaction. This study found that variance components were heterogeneous across the production systems. Genetic correlations between traits in the low, medium and high production systems also suggested that sires should be selected based on genetic evaluation accounting for genotype by environment between production systems.

Further analyses were performed to estimate genetic parameters for milk yield along the lactation trajectory and for lactation persistency (how flat the lactation curve is after the peak yield) in the first four lactations under the three production systems. The genetic association between age at first calving and test-day milk yield in different production systems was also estimated to provide knowledge that can be used to optimize the genetic improvement of milk yield and reproductive efficiency. This was done using multi-variate random regression models. This evaluation produced several key findings: 1) variance components were heterogeneous across the production systems, 2) test-day milk yield and lactation persistency were heritable and therefore genetic improvement can be realised through selection, 3) genetic correlations between test-day milk yield within lactations decreased with increase in time interval, 4) genetic improvement of later lactations could be achieved by selection in the early lactations, 5) sires may be re-ranked between production systems and 6) test-day milk yield and age at first calving are positively correlated (low to medium) at the beginning of the lactation but the correlation decreases along the lactation. This confirms the earlier recommendation that sires should be selected based on a genetic evaluation which accommodates genotype by environment between production systems.

A deterministic bio-economic model was developed to estimate economic values for lactation milk yield, fat yield, age at first calving, calving interval, mature weight and survival under low, medium and high production systems. Economic weights were derived for each trait in the three production systems by discounting the economic values using diffusion coefficients. To allow comparison of the potential economic response in traits the economic weights were standardised using the genetic standard deviations. Traits were reranked across the production systems in order of their economic importance. After performing some sensitivity tests economic values were found to be robust to changes in input and output prices, changes in feeding strategies and milk and surplus heifer marketing strategies. Inclusion of lactation milk yield, fat yield, age-at-first calving, calving interval, mature weight and survival rate is recommended to develop breeding objectives for dairy cattle in Kenya. Selection should also be based on selection indices using estimated breeding values which account for genotype by environment interaction between production systems.

A deterministic simulation using multi-trait selection index theory was used to predict response to selection for alternative selection strategies to maximise genetic gains in the overall breeding objective for dairy cattle under low, medium and high production systems in Kenya. Five different breeding strategies were evaluated including: 1) a breeding program with genetic evaluation and selection of candidate bulls in the high production system only (OPT-S); 2) one joint breeding program with genetic evaluation and selection of bulls in three environments (OPT-J); 3) three environment-specific breeding programs each with an independent genetic evaluation and selection of bulls within each environment (OPT-3); 4) a modified version of OPT-3 simulated to evaluate the effect of using phenotypic and genomic information (OGS-3); and 5) a modified version of OGS-3 using genomic information only (GS-3). The genetic gain in the overall objective was used to evaluate the breeding strategies with varying amounts of information on own performance and relatives.

The OGS-3 strategies generated the highest overall economic gain with a large nucleus (5,000) while the OPT-J strategy generated the highest overall economic gain with smallsized nucleus (500).. OPT-J strategy produced the highest economic gain in the overall breeding objective among the strategies without genomic information, while the OPT-S strategy produced lower overall economic gain compared to other strategies. Genomic selection could generate higher responses compared to the conventional breeding strategies due to a reduced generation interval and higher accuracy of selection with a bigger reference population. This study, therefore, concludes that the dairy cattle industry in Kenya would benefit from a breeding strategy accounting for the differences between the production systems and genotype by environment between them.

The advent of genomic technologies, and its use in the livestock industry has accelerated the increase of genetic merit in multiple breeds. The Hanwoo cattle breed, known for its high-marbled meat, stand to benefits from the integration of genomic information into the breeding system. This thesis aims to explore the potential of utilizing genomic information to enhance the genetic merit of four carcase traits in the Hanwoo breed by exploring the benefits of using pre-selected SNPs and the implementation of female selection in breeding programs.

In Chapter 1 of this thesis, we provide a comprehensive introduction to the overarching research topic. In Chapter 2 a thorough review of relevant literature is presented, setting the stage for a deeper understanding of the current research landscape. Chapter 3 of this research focused on conducting a genome-wide association study (GWAS) for four carcase traits using genotyped animals with relatives’ phenotypic information in Hanwoo. GWAS is a method used to identify genomic regions associated with the phenotypic information, and it has previously been applied to Hanwoo cattle. However, a limitation of these previous studies was the small size of the data set. Although there is a large amount of phenotypic information from carcase recording system in Hanwoo, genotyping of those animals has been restricted due to cost constraints. To overcome this limitation and make use of the available phenotypic information, we utilized a dataset comprising 13,715 animals, which have both genotype and phenotypes, and 454 genotyped sires with progeny phenotype information from 440,284 progeny for this study. There were 33 QTLs and 313 candidate genes identified for carcase weight, as well as 17 QTLs and 122 candidate genes for back fat thickness. Additionally, we detected 19 QTLs and 137 genes for eye muscle area and 12 QTLs and 77 candidate genes for marbling score. This approach has identified additional genomic regions and candidate genes associated with the carcase traits in Hanwoo cattle. Furthermore, these findings demonstrate the potential to increase the power of GWAS by including a large number of animals with phenotypes that are related to genotyped animals.

The identification of selection signatures is a technique employed to reveal the influence of natural or artificial selection in a population. This approach is widely used to identify genomic regions potentially linked to economically important traits. In Chapter 4, we conducted a comprehensive assessment of selection signatures employing three methods. Our analysis encompassed both within and between population evaluations, focusing on the Hanwoo and Angus populations using whole-genome sequence data. Within the Hanwoo population, we identified a total of 374 significant genomic regions, while the Angus population, exhibited selection signatures in 65 genomic regions. When comparing between the two populations, we found 70 genomic regions indicating divergent selection. The candidate genes related with the meat quality traits like HSPA9 and LPL were identified within Hanwoo. Genes associated with the meat quantity and growth traits (ACTC1 and TMEM68) were detected within Angus, and between Hanwoo and Angus. This study showed the selection history of each breed and provided insights into the genomic regions under selection.

In Chapter 5, we assessed the use of pre-selected SNPs from both GWAS and signature of selection analysis to improve the prediction accuracy of carcase traits in Hanwoo cattle. Since the pre-selected SNPs could have the potential to increase the prediction accuracy by explaining a large proportion of the phenotypic variance, the pre-selected SNPs were detected using common and widely used methods, as previously described in Chapters 3 and 4. We conducted a comparison involving BayesR and Genomic Best Linear Unbiased Prediction (GBLUP), the latter using the genomic relationship matrix (GRM), which was fitted with a standard 50K SNP array, augmented with pre-selected SNPs from GWAS, SoS and combined (GWAS + SoS) SNP sets. The result indicated that combining the pre-selected SNPs from both GWAS and signature of selection analysis yield the highest increase in prediction accuracy by 15.1% for carcase weight and by 12.3% back fat thickness using the GBLUP model with two GRMs fitted. For eye muscle area and marbling score, the highest prediction accuracy increase by up to 13.5% was achieved when using only the pre-selected SNPs from GWAS, without incorporating additional SNPs from the signature of selection analysis. By identifying informative genomic regions, we can potentially achieve higher prediction accuracy for these traits. The utilization of pre-selected SNPs from GWAS offers advantages for improving the prediction accuracy of carcase traits in Hanwoo. Expanding the dataset for GWAS emerges as a necessity, enabling the identification of a greater number of genomic regions for traits like MS, which currently exhibits comparatively lower accuracy than its counterparts. Furthermore, it might be useful to expand studies into the signatures of selection analysis to more inter-breed comparison to identify additional SNPs associated with carcase traits. These approaches promises to unveil a broader spectrum of genetic markers for carcase traits.

The Hanwoo breeding program traditionally focused on male selection for carcase traits based on the progeny records, as these traits are difficult to measure directly on selection candidates. With the development genomic information, selection of progeny tested males can be replaced by selection on younger males based on the genomic test. Selecting females with genomic information could be of benefit as well, to increase prediction accuracy and reduced generation interval in the breeding nucleus, however this practice has not been incorporated into the Hanwoo breeding system. Chapter 6 of this thesis comprises a simulation study to assess potential genetic gain in selecting genotyped females and how this is influenced by factors like prediction accuracy, selection intensity, and generation interval. Both deterministic and stochastic simulation models were employed. The results demonstrated that implementing female selection using genomic information provided a predicted improvement in genetic gain by 5.9% and 10.8% in the deterministic and stochastic simulations, respectively. Prediction accuracy of female candidates emerged as the primary factor driving this genetic gain, and the use of genomic selection helped to increase this accuracy. These findings imply that there is a considerable benefit of implementing female selection with genomic information in the Hanwoo breeding system. Moreover, this approach can be extended to include not only carcase traits but also other traits, such as reproduction and growth traits.

The final chapter offers a discussion of the research and concludes with recommendations for future research directions. These recommendations encompass areas like prediction accuracy enhancement and novel breeding programs, aiming to further enhance the genetic improvement of Hanwoo. The thesis serves as a significant contribution to the understanding of Hanwoo breeding and paves the way for more efficient and effective breeding strategies in the future.