Detection and characterisation of markers associated with unknown QTL is the initial step in marker assisted selection (MAS) or genomic selection (GS) practiced in livestock breeding. Good response to MAS or GS depends on a set of markers that can explain most of the total genetic variation and on estimating their effects with sufficient accuracy. It is generally not economically feasible to genotype a large population of animals to obtain estimates of the marker associated effects, using currently available techniques. Therefore alternative approaches are tested in order to maximise the utility of genotype data, limiting the number of animals genotyped. Partial genotyping is one approach used to reduce genotyping costs, carried out either by randomly selecting a proportion of animals from the population concerned, or selectively genotyping the top and bottom extreme animals in the phenotypic distribution. Selective genotyping (SG) is reported to have more power in detecting QTL associated with a trait. However, this is associated with biased estimates of the effects of the QTL or markers tested. Further, it is possible to get a higher number of false positive detections of markers with this approach, especially when multiple markers are evaluated, as comparisons of animals are made from the phenotypic extremes. This thesis investigates the use of partial genotyping methods in detecting and characterising single nucleotide polymorphism (SNP) markers with the use of a combination of methods: selective and randomised genotyping methods using genotype probabilities for the "missing" genotypes. In addition, a modified method of SG was developed which includes animal relationships as well as phenotypic dispersal, and this was tested in a population with a high degree of animal relationships. The performance of the markers selected with each method was evaluated using MAS and GS.

The aim of this thesis is to explore the specific structure in livestock populations to unravel hidden information such as recombination events and parental origin of markers in the genomic data. This information then can be used to improve the accuracy of prediction of breeding values which is one of the main aims of animal breeding. In the first experimental chapter an efficient method for detecting opposing homozygotes was proposed. This method makes the detection of opposing homozygote for thousands of individuals and millions of markers feasible. An opposing homozygote matrix can be utilised to identify Mendelian inconsistency and to fix pedigree errors. The second experimental chapter used opposing homozygotes between individuals in a half-sib family to identify recombination events in the sire, to impute sire haplotype and to reconstruct haplotype of offspring. The algorithm was compared with other frequently used methods, using both simulated and real data. The accuracy of detecting recombination events and of haplotype reconstruction was higher with this algorithm than with other algorithms, especially when there were genotyping errors in the dataset. For example, the accuracy of haplotype reconstruction was around 0.97 for a half-sib family size of 4 and the accuracy of sire imputation was 0.75 and 1.00 for a half-sib family size of 4 and 40, respectively. In the third experimental chapter hsphase was developed which implements the algorithms used in the first two chapters into an efficient R package. In addition, an algorithm for grouping half-sib families utilising the opposing homozygote matrix was developed and verified with real datasets. The results show that the algorithm can group the half-sib families accurately, however the accuracy was depended on sample size and genetic diversity in the population. The package includes several diagnostic functions to visualise and check half-sib's pedigree, parentage assignments, and phased haplotypes of offspring in a half-sib family. The fourth experimental chapter utilised the half-sib population structure to fix switch errors. The switch error is a common problem in many haplotype reconstruction algorithms where the haplotype phase is locally correct but paternal and maternal strand are not consistently and correctly assigned across the longer segments (or across the entire genome). The algorithm partitions the genome into segments and creates a group matrix which is used to identify the switch points. Then the switches are fixed with a second algorithm. The results showed that this algorithm can fix the switch problems efficiently and increase the accuracy of genome-wide phasing. In chapter five relationship matrices generated from haplotype segments were used to improve the accuracy of predicting breeding values. The haplotypes were partitioned in three ways and with various size. The new relationship matrices were evaluated with three sets of real data and with simulated data. In all cases the accuracy of prediction and log-likelihood were significantly increased although the amount of increase was trait dependent.

The world's food production will have to double by 2025 to provide for an estimated population of eight billion people. Fish is a vital component of food security in developing countries while demand is increasing in the developed world due to greater awareness of the health benefits of eating fish. As capture fisheries have stagnated, aquaculture is expected to play a crucial role in meeting this increased demand for high quality protein. However, there is a large disparity between the need for increased aquaculture production and the genetic quality of stocks available to meet that need. At present, the majority of global aquaculture production is based on unimproved species and strains. If aquaculture is to fulfil its role then improved breeds that use resources more effectively are needed. ... In this study the principles of quantitative genetics have been applied to an Australian freshwater aquaculture species to improve growth rate. Intra-specific heterosis was evident, but not consistent, highlighting the unpredictable expression of dominance genetic variance. This supports recent work in freshwater fish suggesting that the relative weight of dominance and additive genetic variance vary over development time. It also shows that in a highly variable pond environment, non-genetic or environmentally induced size variation can mask the genetic components of growth. Females are the preferred sex for culture and future breeding programs could exploit mono-sex populations. Finally, the overall aim of fish breeding programs worldwide is improved profitability and this study has demonstrated that improvements in growth rate result in improvements in revenue.

This thesis aimed to investigate genotype by environment (birth or rearing type) interaction (GEI) in Merino sheep for growth traits using univariate and bivariate mixed models and also random regression model analysis with phenotype data and using genome wide association study with genome data. The impact of accounting for sire by birth and rearing type interaction in sheep breeding programs was investigated. In the first research chapter, univariate and bivariate mixed models were used for the analysis of birth weight (BWT), weaning weight (WWT), post weaning weight (PWWT), scan fat (PFAT) and eye muscle depth (PEMD) with and without inclusion of sire by birth type (BT) or rearing type (RT) interactions. Models fitted sire by BT or RT to estimate its contribution to phenotypic variation and to estimate breeding values for expression of traits both in single and twin BT or RT environments. Univariate analysis showed a significant sire x BT interaction accounting for 1.59% and 2.49% of the phenotypic variation for BWT and WWT, respectively, and no significant effect for PWWT, PFAT and PEMD. Sire x RT interaction effects were much smaller and only significant for PEMD. The genetic correlation between breeding values for growth in single and twin BT varied from 0.47 to 0.99. In the next research chapter, univariate and bivariate random regression models (RRM) were fitted to repeated measurements of body weight of Merino lambs from 75 to 300 days of age to explore genetic differences in growth paths between single and twin lambs. A sire by birth type and sire by rearing type interaction contributed around 2-6% to variation in body weight at 200 until 300 days of age, while before 225 days of age the contribution was almost zero. Genotype expression differed in different birth types and in different rearing types particularly in earlier life of lambs for birth type and in late age for rearing type with birth type having a larger effect than rearing type, with the genetic correlation between weight in different BT varying over the trajectory from 0.75-0.97. In the third research chapter, a genome wide association study was applied to investigate the association of genetic markers with birth weight (BWT) and the interaction of significant marker effects with birth type (BT) in Merino lambs. Eleven and fifteen genome-wide significant SNPs for birth weight were detected for birth weight in singles and twins, respectively, with 11 of those overlapping, and 17 genome-wide significant SNPs were found when using all data. OAR6_41936490.1 and OAR6_41877997.1 were the most significant SNPs for single and twin birth weight, respectively. Among 17 significant SNPs detected by GWAS there were 9 that had a significant SNP by BT interaction, indicating that gene by BT interaction contributed to BWT variation. In the final research chapter, the impact of accounting for sire by birth type interaction on percentage of incorrect animals and sires selected was investigated by comparison with using a bivariate "expected" model for birth weight (BWT), weaning weight (WWT) and post weaning weight (PWWT) in Merino lambs. Correlation of EBV of lambs or sires and genetic gain between 4 EBVs from 3 proposed models (for BWT, WWT and PWWT) were investigated. The impact of accounting for sire by birth type interaction on the percentage of animals and sires incorrectly selected was small but this percentage was about 10% compared with the best model. The impact of accounting for sire by birth type interaction was also small when considering loss in genetic gain, but the loss was below 4% compared to the best model. In conclusion, although the existence of genotype by birth type interaction for growth traits in Merino sheep was small, but its impact on incorrect animals and sires can be ignored in breeding program.

A total of 1009 primi- and multiparous sows from two maternal (Large White and Landrace) and two terminal (Duroc and Large White) lines were recorded between early spring (N=494 sows) of 2009 and late summer (N=515 sows) of 2010. Approximately 84% of these sows farrowed within the experimental periods, producing 9133 piglets from 847 litters (610 sows). Both sow and piglet traits were recorded to investigate their associations with piglet survival until weaning, which was estimated from an inventory of individual piglet's deaths. Using logistic regression, in addition to season, line, and parity effects, both sow and piglet attributes were identified that had significant phenotypic associations with individual piglet survival until weaning. It was important to consider the sow and piglet attributes simultaneously to identify factors associated with piglet survival. Important sow factors included health pre-farrowing, interval between transfer and farrowing dates, and teat access. Piglet factors included birth weight, rectal temperature, body condition, ponderal index, skin colour, respiration, muscle tone, shivering, hydration level, fostering status, and gender. The genetic basis of the traits recorded was examined to identify which traits were heritable and potential selection criteria. Estimates of genetic correlations between traits can then be used to identify consequences of selection.

Although for most traits, there is a difference between phenotypic expression in ewes versus rams, they have been traditionally treated equally in breeding programs. This thesis has investigated whether there is any variation between them in how they inherit growth and whether there are any benefits to be gained from divergent selection objectives. A bioeconomic model was used to show that divergent selection for growth in ewes and rams could increase the feed efficiency and profitability of a self-replacing production system. The magnitude of the increase was found to be influenced by the age structure of the flock, fertility, base weight (average of both sexes) and the annual feed availability and price cycles. To test whether divergent selection for growth is possible, sex specific genetic parameters and genetic correlations between ewes and rams were estimated. Accordingly, a series of univariate, multivariate and random regression models using combinations of direct additive genetic, maternal additive genetic, maternal permanent environmental and direct permanent environmental effects were fitted. The results from these analyses revealed that ewes have larger heritability and smaller phenotypic variances than rams and that the intersex genetic correlations were less than unity and decreased with age (range 0.59 to 0.97). Furthermore, the phenotypic and BLUP selection responses predicted using these parameters revealed that rams had higher selection responses than ewes. The results for fat depth (corrected for live weight) were the sole exception to these results. Fat in ewes was deeper and had a lower heritability and higher phenotypic variances than rams. Little variation in sexual dimorphism was observed between the breeds used in this study despite variation in selection history/objectives and founder effects in each breed. In conclusion, these results reveal that it is useful to select for divergent weight, fat and eye muscle depth objectives in ewes and rams and that sexual dimorphism can have a positive effect on the feed efficiency and profitability of a self-replacing flock. Other potential uses of sexual dimorphism in livestock breeding including in species other than sheep, single sex production systems (using sexed semen), reducing product diversity (equal ewes and rams) and the influence of recording/selecting a single sex were also discussed.

This thesis presents the comparison of results of applying tactical and strategic approaches for different breeding programs according to diverse objectives targeting commercial beef production systems in Argentina. The use of the Differential Evolution algorithm for Mate Selection, in which those objectives are considered in a Mate Selection Index function, allows the optimization of profit over herds, and the control of other genetic aspects such as maintenance of genetic diversity and avoidance of inbreeding. The techniques used in this thesis for investigation are based of methods for progressive implementation of breeding programs, and as such bear direct relevance to progressive sectors of the beef industry.

Genetic markers provide the Australian beef industry with the opportunity to increase rates of genetic gains. However, accurate estimates of the gene frequencies and the marker size of effects are first required. Stochastic simulation was used to examine the methods and models required to estimate SNP effects. Results showed for a single SNP explaining 2% of the phenotypic variation, 1,500 animals were required to estimate the SNP effects when the favourable allele frequency (p) was 0.5. However, increasing the additive SNP effects decreased the number of animals required. SNP effect estimates were inflated when the power to detect genotype effects was low. In addition, when the allele frequency was rare (p=0.1), biased dominance effects were estimated. For SNPs that were in linkage disequilibrium with the causative SNP, the SNP effects were accurately estimated when linkage disequilibrium was greater than D'=0.9. This thesis found that when there was linkage disequilibrium between the two direct SNPs (explaining 8% of the phenotypic variation, collectively), and one SNP was ignored in the model, estimates of the SNP effects were biased upwards. Ignoring epistatic effects (1% of the phenotypic variance) also increased the estimates of the SNP effects. To estimate accurate SNP and epistatic effects 3,000 animals were required. If SNPs were excluded in the model, the SNP and epistatic variance was partitioned as polygenic and residual variances, respectively. The inclusion of SNPs was shown to increase the accuracy of the estimated breeding value (EBV); the more phenotypic variation explained by the SNPs the higher the increase in EBV accuracy (or estimated genetic merit when non-additive (i.e. epistasis) effects were included). This thesis shows that the population size, allele frequency, statistical power to detect genotype effects, statistical models and the data structure all affect the ability to accurately estimate the size of SNP effects.

The relationship between rumen microbial ecology and the host is regulated by multiple factors including diet, the microbial inoculum entering the gut, and host genetics. Two hypotheses associated with this statement were developed and tested during this project. The first focused on the relationship between diet, early life microbial inoculum and rumen microbiota and the second concentrated on the association between host genetics and gut microbial ecology. Hypothesis one was that the rumen microbiome of lambs could be altered by post-natal diet and by early-life microbial intervention, to achieve differences in rumen fermentation, development and animal performance that persist beyond weaning. Secondly, it was hypothesized that sheep with different genetic merit for wool growth harbour differences in their rumen microbiome that are associated with differences in gut morphology, physiology, digesta retention time and microbial protein outflow which underpin their wool phenotypes.

This thesis explores various methods to optimise breeding programs that use female reproductive technologies and genomic selection. Simulation studies have shown that female reproductive technologies such as multiple ovulation and embryo transfer (MOET) and juvenile in vitro embryo production and transfer (JIVET) can increase rates of genetic gain through increased female selection intensity and decreased generation interval. Furthermore the use of genomic selection has facilitated better selection decisions to be made on younger selection candidates that may not have phenotypic measurements. When combining genomic selection with reproductive technologies the rate of genetic gain could be further accelerated. However intensive use of the best females in breeding programs can also increase the rate of inbreeding to unsustainable levels. This thesis aimed to stochastically simulate breeding programs where reproductive and genomic technologies are optimally implemented while maintaining a sustainable increase of inbreeding. The impacts of using reproductive technologies and/or genomic selection were evaluated for breeding programs across species. Furthermore, the thesis investigated a cost-benefit analysis of using reproductive technologies which led to a further study that optimized the use of reproductive technologies that considered their costs as well as future co-ancestry during selection.

Sheep are a highly versatile and adaptable domestic species. Dissection of the genetics responsible for the ovine domestic phenotype relies on an understanding of the genetic variability that resides within and between breeds and also knowledge regarding both the maternal and paternal origins of sheep. Mitochondrial DNA was investigated in the search for novel 'Ovis aries' matrilines and the complete mitogenome sequenced from a subset of domestic and wild sheep to resolve the phylogenetic relationships between these groups. A fifth domestic lineage was identified in sheep from the Near East, a proposed centre of domestication. Mitogenome analysis revealed no wild sheep introgression in the five 'O. aries' groups. To contrast this maternal picture, variation in the male-specific region of the ovine Y chromosome was investigated. Seven novel single nucleotide polymorphisms and a previously uncharacterised microsatellite from the ovine sex determining gene region were used to generate 17 paternal haplotypes. Analysis of these markers across wild and domestic sheep again failed to identify signatures of wild Ovis introgression in modern sheep. The emerging picture of male mediated domestication suggested that there are at least two patrilines present within 'O. aries'. One of these has a possible European origin and the other, a less restricted distribution. Genetic diversity and linkage disequilibrium (LD) was characterised using autosomal microsatellites. Five domestic populations were investigated, with the result that in genetically diverse breeds, LD extended for only short distances, whilst more homogeneous populations displayed extensive patterns of LD. This clearly illustrated the impact of population history on the extent LD and will inform subsequence gene mapping studies in sheep. The three classes of genetic variation investigated (autosomal, paternal and maternal), each reveal aspects of the genetic architecture present within domestic sheep and only by assaying each of these, will the true picture of ovine domestic and diversity be revealed.

The objectives of this thesis were to perform a genome scan for important production traits in Australian Merino sheep and to investigate issues related to the prediction of marker assisted selection (MAS) response in single and multiple trait selection schemes in animal breeding programs. In the first experimental chapter, a genome scan was performed for 4 growth and 7 fleece traits in Australian Merino sheep. Quantitative trait loci (QTL) parameters were estimated through single QTL interval mapping within and across four paternal half-sib families. The genome scan yielded 21 significant QTL for all traits at the 1% chromosome wise significance threshold level in within family analysis. Across family analysis supported most of the highly significant results from single family analysis but did not show any common significant QTL at the 1% chromosome wise significance threshold level across all families. Because of the relatively small progeny group sizes the power and the precision of the analysis are probably low and the QTL allele substitution effects are overestimated. In the subsequent three chapters some important issues related to the application of QTL information in MAS, including efficient prediction of MAS response and effect of the accuracy of estimated QTL effect in single and multiple trait selection, were investigated.

The successful breeding of ewes before the age of 12 months presents an opportunity to improve flock reproduction efficiency and increase returns for sheep producers. However, uptake of this practice is currently low in Australia and New Zealand. A contributing factor to this is that reproductive performance at 1 year of age is lower and more variable than in older ewes; between flocks and also from year to year. Whilst previous research has explored what factors contribute to successful reproduction at 1 year of age, there is no universally accepted measure of puberty and sexual maturity in sheep nor is there an accurate phenotypic predictor of yearling reproductive success. Although established genetic correlations between reproduction and production traits exist, reproduction has previously been analysed as a repeated records trait across parities, rather than treating first parity as a separate trait. Model studies show that genomic information offers the opportunity to select animals more accurately at younger ages. However the implications of a lower and more variable fertility rate of ewes mated prior to 1 year of age has been largely ignored. The aim of this thesis was to evaluate the impact of mating ewes prior to 1 year of age on flock genetic progress and to quantify the relationship between potential pubertal traits, yearling reproduction traits, reproduction traits at later parities and other key production traits.

This thesis explores the accuracy of methods to estimate the breed composition of crossbred animals which have unknown pedigree. Herein I present the use of SNP technologies to estimate the breed composition of small-holder crossbred dairy cattle in developing countries for the first time. Before this could be done there was a need to determine: what are the accuracies of different methods for estimating breed composition? The genetic structure of animals, the design of reference populations, the number of SNP markers and the model selected has possible consequences for estimation of breed composition. Once the effect of the above factors on the accuracy of estimation of breed composition is identified, it is possible to estimate with confidence the breed composition of crossbred animals that have no recorded pedigree. The overall aim of this thesis was to investigate the use of high-density SNP data to understand the livestock breed's population structure and estimate the breed composition of crossbred animals.

The aim of the present thesis was to identify genomic regions associated with parasite resistance in sheep and to evaluate the potential improvements in genomic prediction accuracies when incorporating genomic information in estimating breeding values. Data were derived from a large reference population of sheep developed in Australia, based on the CRC Information Nucleus Flock (INF). Worm egg counts (WEC) were collected from animals that were naturally infected in the field with mixed gastrointestinal worm species. Egg counts determined the presence of three predominant strongyle species; Teladorsagia circumcincta, Haemonchus contortus, and Trichostrongylus colubriformis. Heritability estimate for WEC based on pedigree relationships (0.20±0.03) was similar to those obtained from genomic relationships calculated from 50k and 600k genotypes. In a genome partitioning analysis, the genetic variance explained by each chromosome was proportional to the chromosomal length, providing strong evidence that parasite resistance is a polygenic trait with a large number of loci underlying the mechanism of resistance.
Genome wide association studies (GWAS) and regional heritability mapping (RHM) identified a significant region on OAR2 associated with parasite resistance. Haplotype analysis confirmed a haplotype block within this region on OAR2, which overlaps with GALNTL6 (Polypeptide N-Acetylgalactosaminyltransferase Like 6) gene, responsible for mucus production. Fine-mapping RHM analysis with smaller window sizes identified more significant regions on OAR6, OAR18, OAR24 as well as OAR20 within the major histocompatibility complex (MHC). Each region explained only a small proportion of WEC heritability, ranging from 2% to 5%. Pathway analyses revealed key genes involved in innate and acquired immune system pathways as well as cytokine signalling pathways. Mucus production and haemostasis are also relevant in protecting the host from parasite infections.
The accuracy of genomic predictions was evaluated for different groups of animals that had varying degree of relationships to their respective training populations. A closer relationship between the training and validation groups led to a higher accuracy of genomic prediction for WEC. GBLUP predicted breeding values more accurately than pedigree-based BLUP, especially when the relationship between training and validation groups was distant. These results highlight the importance of the relationships between animals in training and validation sets as a key factor in determining prediction accuracies.
The increased availability of whole-genome sequence (WGS) data, combined with a larger number of genotyped animals, made it possible to split datasets into QTL discovery and training/validation subsets and evaluate the prediction accuracy across the three marker densities. The performance of genomic prediction was evaluated using cross-validation design across sire families. Prediction accuracy of WEC improved slightly from 0.16±0.02 using 50k genotypes to 0.18±0.01 and 0.19±0.01 when using HD and WGS data, respectively. Variants selected from WGS data using GWAS and RHM methods improved the prediction accuracy substantially, when fitted alongside 50k genotypes, compared to when the 50k genotypes were fitted alone. However, when variant selection was based only on GWAS, the prediction accuracy increased by 5%, whereas when selection was limited to variants with the lowest GWAS p-values in windows identified by RHM, the prediction accuracy increased by 9%. These findings offer potentially important implications for future genomic prediction studies for parasite resistance.

The importance of extending trait measurement to commercial operations to improve rates of genetic improvement in breeding schemes is of increasing research interest due to the improved ability to identify and record animals using new technologies. In integrated multitiered breeding schemes, selection of commercial ewes and multiplier rams is frequently driven by soundness and overall condition of selected animals and sometimes, based on whether or not the replacement candidate was born in a multiple litter. In breeder flocks, selection is usually supported by knowledge of the genetic merit of candidates. In breeding schemes forming a part of a multi-tier integrated production system, genetic progress is disseminated from the nucleus to the commercial tier through the transfer of males. The opportunity to increase the rate of progress and reduce the genetic lag between tiers in integrated production systems might be exploited by using information from lower tiers to reduce generation interval through candidates’ selection in these tiers and to increase prediction accuracy in individuals from the nucleus.

Through deterministic simulation this study estimated the economic benefits that can be generated in multi-tier breeding schemes by implementing performance recording in conjunction with either DNA pedigree assignment or genomic selection. The overall economic benefits of improved performance in the commercial tier, brought about by higher merit multiplier rams, offset the costs of recording the multiplier tier. The net cumulative benefit per commercial ewe over a period of 40 years ranges from $117 to $249 of additional genetic progress depending on the level of GxE and breeding objective of the production system. Applying genomic selection was the most beneficial strategy if testing costs can be reduced or by genotyping only a proportion of the selection candidates. When the cost of genotyping was reduced, i.e. from $50 to $25 per test, scenarios that combined performance recording with genomic selection were more profitable and reached breakeven point at about 10 years, as opposed to 20 years. This gain however, does not have the permanent and cumulative properties of genetic progress achieved in the nucleus flock, and so needs to be continued each year for the benefits to be maintained. Also, the strategies in which recorded multiplier ewes and rams were selected as replacements for the nucleus flock did modestly increase profitability by 2 to 5% when compared to a closed nucleus structure.

The opportunity to expand the benefits of data recorded on commercial farm animals in livestock industries was then estimated if these data were used to improve selection accuracy of breeding nucleus individuals. In multi-tiered breeding schemes, the top tier is frequently maintained under different conditions to that of the commercial tier, which may lead to a genotype by environment interaction (GxE) that affects performance in target environments. If there is a genotype by environment interaction such that the expression of traits in the nucleus is not fully transmitted to the commercial tier, there will be an additional advantage to using commercial information for genomic prediction. With multi trait selection, both ranking and scale-type GxE may lead to significant re-ranking for overall economic merit, which can be exacerbated when economic values are different between different environments.

The findings of our study indicate minimal evidence of significant sire re-ranking for most traits. Nevertheless, based on relatively limited data, correlations among nucleus and commercial flocks were low for weaning weight (0.79 or less) and dag score (0.85 or less). Also, the spread of EBVs and heritability estimates on other economically important traits differed significantly across environments, which infers distinct responses to selection.

The inclusion of phenotypic and genomic data from lower tiers led to increased accuracy of prediction in the breeding tier. When trait phenotypes and genomic information from related commercial individuals were available for genetic evaluations, these could be used as training animals in genomic prediction and link a wider range of selection candidates. Our results indicate that for a genetic correlation of 0.75 between the nucleus and the commercial environment, the accuracy of prediction increased from 0.28 to 0.77 when the number of commercial individuals recorded per year increased from 0 to 2,000. This has potential to benefit larger numbers of individuals across tiers, as it allows accurate selection of candidates that perform well across environments. The extent of genomic relationships between nucleus and commercial individuals was critical to predict the improved accuracy contributed by the commercial information, as was the number of commercial individuals tested. In the presence of a substantial GxE between the nucleus and commercial tiers, the value of the commercial phenotypic records and genotypes was further enhanced.

The objective of the final chapter was to model the potential impact of using commercial information on the additional genetic gain, relative to a typical scenario without commercial individual’s performance records, pedigree or genotype information, nor accounting of GxE. Greatest impacts were found to arise from the use of trait measurement and genotype data from commercial operations to more accurately select nucleus sires. Benefits from shortening the genetic lag between commercial and nucleus flocks through improved selection of males for use in the multiplier and commercial tiers were more modest, and the cost effectiveness was very dependent on low genotyping costs. In practice, performance records and genotypes from commercial individuals may strengthen the links between breeders and commercial farmers leading to more relevant genetic progress although novel business agreements will be required for implementation in situations where commercial tiers are separate business entities to the nucleus and/or multiplier tiers.