Reducing disease prevalence through selection for host resistance offers a desirable alternative to chemical treatment which is a potential environmental concern due to run-off, and sometimes only offers limited protection due to pathogen resistance for example (Chen et al., 2010). Genetic analyses require large sample sizes and hence disease phenotypes often need to be obtained from field data. Disease data from field studies is often binary, indicating whether an individual became infected or not following exposure to infectious pathogens. In genetic analyses of binary disease data, however, exposure is often considered as an environmental constant and thus potential variation in host infectivity is ignored. Host infectivity is the propensity of an infected individual to infect others. The lack of attention to genetic variation in infectivity stands in contrast to its important role in epidemiology. The theory of indirect genetic effects (IGE), also known as associative or social genetic effects, provides a promising framework to account for genetic variation in infectivity as it investigates heritable effects of an individual on the trait value of another individual. Chapter 2 examines to what extent genetic variance in infectivity/susceptibility is captured by a conventional model versus an IGE model. The results show that, unlike a conventional model, which does not capture the variation in infectivity when it is present in the data, a model which takes IGEs into account captures some, though not all, of the inherent genetic variation in infectivity. The results also show that genetic evaluations that incorporate variation in infectivity can increase response to selection and reduce future disease risk. However, the results of this study also reveal severe shortcomings in using the standard IGE model to estimate genetic variance in infectivity caused by ignoring dynamic aspects of disease transmission. Chapter 3 explores to what extent the standard IGE model could be adapted for use with binary infectious disease data taking account of dynamic properties within the remit of a conventional quantitative genetics mixed model framework and software. The effect of including disease dynamics in this way was assessed by comparing the accuracy, bias and impact for estimates obtained for simulated binary disease data with two such adjusted IGE models, with the Standard IGE model. In the first adjusted model, the Case model, it was assumed that only infected individuals have an indirect effect on their group mates. In the second adjusted IGE model, the Case-ordered model, it was assumed that only infected individuals exert an indirect effect on susceptible group mates only. The results show that taking the disease status of individuals into account, by using the Case model, considerably improves the bias, accuracy and impact of genetic infectivity estimates from binary disease data compared to the Standard IGE model. However, although heuristically one would assume that the Case-ordered model would provide the best estimates, as it takes the disease dynamics into account, in fact it provides the worst. Moreover, the results suggest that further improvements would be necessary in order to achieve sufficiently reliable infectivity estimates, and point to inadequacy of the statistical model. In order to derive an appropriate relationship between the observed binary disease trait and underlying susceptibility and infectivity, epidemiological theory was combined with quantitative genetics theory to expand the existing framework in Chapter 4. This involved the derivation of a genetic-epidemiological function which takes dynamic expression of susceptibility and infectivity into account. When used to predict the outcome of simulated data it proved to be a good fit for the probability of an individual to become infected given its own susceptibility and the infectivity of its group mates. Using the derived function it was demonstrated that the use of a linear IGE model would result in biased estimates of susceptibility and infectivity as observed in Chapters 2 & 3. Following the results of Chapter 4, the derived expression was used to develop a Markov Chain Monte Carlo (MCMC) algorithm in order to estimate breeding values in susceptibility and infectivity in Chapter 5. The MCMC algorithm was evaluated with simulated disease data. Prior to implementing this algorithm with real disease data an adequate experimental design must be determined. The results suggest that there is a trade-off for the ability to estimate susceptibility and infectivity with regards to group size; this is in line with findings for IGE models. A possible compromise would be to place relatives in both larger and smaller groups. The general discussion addresses such questions regarding experimental design and possible areas for improvement of the algorithm. In conclusion, the thesis advances and develops a novel approach to the analysis of binary infectious disease data, which makes it possible to capture genetic variation in both host susceptibility and infectivity. This approach has been refined to make those estimates increasingly accurate. These breeding values will provide novel opportunities for genome wide association studies and may lead to novel genetic disease control strategies tackling not only host resistance but also the ability to transmit infectious agents.

When an individual interacts socially with a conspecific, their behavioural phenotype is affected directly by their genotype (‘direct genetic effect’, DGE), but may also be affected indirectly by the genotype of the opposing individual (‘indirect genetic effect’, IGE). While there is no doubt that IGEs occur in various organisms and contexts, it is unknown how properties of the environment may influence the relative magnitude of DGEs, IGEs, and their covariance. To gain insight into this, I examined territorial interactions in Drosophila melanogaster. Due to their short generation time and relatively simple care requirements, D. melanogaster has been used extensively in quantitative genetic research. Using offspring from a half-sib breeding design, I constructed an arena for documenting multiple dyadic territoriality assays with two sizes of a food resource. With this apparatus, 618 territoriality contests between 1,236 individuals were recorded and scored for four key aggressive behaviours. The results revealed significant genetic variation in how opponent effects on focal individuals changed between environments (i.e., genetic variation in the plasticity of IGEs). In addition, changes in DGEs and IGEs between environments were strongly and positively correlated (i.e., there was a DGE × IGE × environment interaction), although confirmation of this result in further studies is warranted because it was non-significant (P = 0.10), likely due to large uncertainties arising in part from some small variance component estimates. As a high throughput system for quantify IGEs in territoriality in Drosophila, my approach holds promise but there are issues to resolve, including automating phenotyping behaviors in place of manual scoring to enable many more trials. Additionally, modifications to increase humidity during trials might result in increased expression of certain territorial behaviours.

This study was carried out in order to investigate the human genetic effects of exposure to benzene in 66 male workers of a refinery population and the results compared with 33 control workers in the same refinery, not known to have had exposure to benzene. Questionnaires were used to determine various life style factors such as smoking, drinking and exposure to ionising radiation as examples of known confounding variables. In addition, experiments were designed to investigate the mechanism of benzene carcinogenicity using cell transformation techniques, together with a molecular dosimetry approach in an attempt to identify and quantify any interaction with benzene metabolites and DNA. The results from the human studies showed no difference between the groups when effects such as mitogen-induced blastogenesis, proliferative rate index, sister chromatid exchange and urine mutagenicity were measured. There was a suggestion of a decrease in mitogenic response with age in both exposed and control individuals in the mitogen induced blastogenesis experiments, which was consistent with other studies. Although no difference in the number of revertant colonies in strain TA 98 and 100 was demonstrated between the high and low urinary phenol groups there was a correlation between the number of revertants and the ages of the individuals as a whole. One statistical test used in the examination of the chromosome aberration data suggested a statistically significant increase in aberrations in the exposed group to the control groups, and this increase could be the result of benzene exposure. Cell transformation studies using C3H10T1/2 cell lines did not indicate that benzene had any initiating carcinogenic properties in vitro using the two stage model of carcinogenesis. Furthermore, molecular dosimetry studies using C[14]-labelled benzene in vivo demonstrated only a very weak interation between benzene metabolites and rat liver DNA. All the methods used in this study generated negative data except for that to detect chromosome damage. This method showed a slight increase in damage in exposed workers by comparison with the controls suggesting that benzene may be a weak clastogen at low doses.

Objective: To investigate the association of previously reported single nucleotide polymorphisms (SNPs) in relation to orofacial clefts and assess their interaction with environmental factors. Methods: Genome-wide SNP genotypes were obtained for case-parent triads from the EUROCRAN and ITALCLEFT studies. Candidate SNPs were selected from a previous genome-wide association study (Shi et al., 2012) along with surrounding SNPS for a total of 2142 genotyped and imputed SNPs. A total of 411 case-parent triads and 25 case-parent dyads were analyzed using log-linear models to test for maternal and parent-of-origin effects along with their interaction with maternal smoking and maternal folic acid consumption. Results: A significant association (q = 0.025) was detected for a region in the ATXN3 gene. This significance refers to the interaction between maternal periconceptional smoking and maternal genetic effects. Nominally significant associations in genes relating to the brain were also detected. Conclusion: SNPs in the ATXN3 region warrant further investigation.

Porcine Reproductive and Respiratory Syndrome (PRRS) is, globally, one of the costliest of diseases to the pig industry. Despite enormous efforts, methods such as vaccination strategies and herd management have failed to fully control the disease. Exploiting the genetic variation in host response could be included as part of a multifaceted approach to mitigate the devastating impact of this disease. Establishing the presence of genetic variation and its underlying genetic architecture are key to implementing genomic selection, which is considered a viable and safe long-term disease control strategy. This thesis explores the effect of natural PRRSV outbreaks on the reproductive performance of sows, and the underlying genetic influences on it. Litter records were available from two farms, where Porcine Reproductive and Respiratory Syndrome Virus (PRRSV) outbreaks had been confirmed using ELISA. One farm had full pedigree information, but for both farms 60K SNP genotypes were available. In both farms, performance records could be partitioned into an epidemic and non-epidemic phase using a previously established threshold method. The partitioning also identified a period of high reproductive failure not coinciding with a diagnosed PRRSV outbreak on one farm. This period was isolated and analysed separately. Linear mixed models were used to explore both genetic and non-genetic factors contributing to differences in reproductive performance associated with the two phases. This analysis identified five disease indicator traits identified showing significant differences (>95% CI) in least squares means between the epidemic and non-epidemic phase. These were the number of mummified, stillborn, dead and alive piglets per litter and the fraction of the total born dead. Alternative statistical models that accounted for differences in the severity of the individual PRRSV outbreaks were also considered throughout. Despite differences in the estimates associated with different models and farms, in general very low heritability estimates were obtained for these disease indicator traits during the non-epidemic phase, whereas the traits were found moderately heritable during the epidemic phase. Two genome wide association analyses methods were used to explore the distribution of the genetic effects throughout the genome: Family-based Score Test for Association (FASTA) and Genome-wide Rapid Analysis using Mixed Model and Regression (GRAMMAR). In addition, regional associations were studied using Regional Heritability Mapping (RHM). Associations were then further characterised using Measured Genotype (MG) analyses. Genome-wide significant associations were identified for five SNPs and one region. The regional association spans the region previously identified in an experimental challenge experiment of growing pigs, in association with viral load and weight gain. Different patterns of linkage disequilibrium (LD) are observed which may explain why this study and others failed to find single SNP effects at this location. One genome wide significant SNP on SSC15 was found between two previously identified SNPs associated with PRRSV mortality. Five further putative SNP associations are indicated by RHM and subsequent measured genotype analysis, two of which flank previously reported associations and indicate an epistatic effect, observed in several traits. In summary, this study showed that reproductive performance of sow is considerably reduced during PRRSV outbreaks and the genetics of the sow significantly affects variance in survival and mortality. Several novel genomic regions associated with the reproductive performance of sows in the absence and during PRRSV outbreaks have been identified in this study. In addition to these, the results suggest the region on SSC4 previously associated with PRRSV viral load and weight gain may also affect foetal mortality. These results demonstrate the potential for genomic selection to be used to mitigate PRRSV related reproductive losses, the greatest financial exposure faced by the pig industry. In addition, RHM is directly shown to capture genetic variance, where single SNP methods fail to identify an effect, highlighting the usefulness of this tool as a method to identify genomic regions with significant effect on production traits.

Complex phenotypes are traits which are influenced by many factors, and not just a single gene, as for classical Mendelian traits. The brain, and its resultant behaviour, gives us a large subset of complex phenotypes to examine. Variation in these traits is affected by a range of different influences, both genetic and environmental, including social interactions and the effects of parents. Systems-genetics provides us with a framework in which to examine these complex traits, seeking to connect genetic variants to the phenotypes they cause, through intermediate phenotypes, such as gene expression and protein levels. This approach has been developed to exploit and analyse massive data sets generated for example in genomics and transcriptomics. In the first half of this thesis, I combine genetic linkage data from the BXD recombinant inbred mouse panel with genome-wide association data from humans to identify novel candidate genes, and use online gene annotations and functional descriptions to support these candidates. Firstly, I discovered MGST3 as a novel regulator of hippocampus size, which may be linked to neurodegenerative disorders. Secondly, I identified that CMYA5, MCTP1, TNR and RXRG are associated with mouse anxiety-like phenotypes and human bipolar disorder, and provide evidence that MCTP1, TNR and RXRG may be acting via inter-cellular signalling in the striatum. The second half of this thesis uses different cross-fostering designs between genetically variable BXD lines and the genetically uniform C57BL/6J strain to identify indirect genetic effects and the loci underlying them. With this, I have found novel loci expressed in mothers that alter offspring behaviour, novel loci expressed in offspring affecting the level of maternal care, and novel loci expressed in offspring, which alter the behaviour of their nestmates, as well as the level of maternal care they receive. Further I provide evidence of co-adaptation between maternal and offspring genotypes, and a positive indirect genetic effect of offspring on their nestmates, supportive of a role for kin selection. Finally, I demonstrate that the BXD lines can be used to investigate genes with parent-of-origin dependent expression, which have an indirect genetic effect on maternal care. In conclusion, this thesis identifies a number of novel loci, and in some cases genes, associated with complex traits. Not only are these techniques applicable to other phenotypes and other questions, but the candidates I identify can now be examined further in vitro or in vivo.

The evolution of social behaviour has been studied using different frameworks based on game theory and quantitative genetics. While both approaches provide a conceptually clear explanation of evolution of social behaviour, both have been limited in their applicability to empirical systems, mainly due to difficulties in measuring model parameters. Here, I develop a new quantitative genetics approach to the study of the evolution of social behaviours based on indirect genetic effects (IGEs), which parameters can be readily determined by empirical studies. IGEs describe effects of an individual's genotype on phenotypes of social partners, which may indirectly affect their fitness. Unlike traditional quantitative genetics assuming a non-genetical, non-heritable environment, IGE models assume that part of the environment is social, provided by parents and other interacting partners, thus has a genetic basic and can be heritable. In this study I explore the effects of IGEs on the magnitude and range of phenotypic values in a focal individual. I show that social interactions may not only cause indirect genetic effects but can also modify direct genetic effects. I demonstrate that interactions can substantially alter group mean phenotype and variance. This may lead to scenarios in which between group phenotypic variation is much higher than within group variation despite similar underlying genetic properties of different groups. Further, I analyse how IGEs influence levels of selection and predictions about evolutionary trajectories. I show that IGEs can create selection pressure at the group level, leading to evolution of behaviours that would not evolve otherwise. Moreover, I demonstrate that IGEs may lead to differences in the direction of evolutionary response between genotypes and phenotypes. Building on these results, I show that IGE models can be translated to and are fully compatible with traditional kin and multilevel selection models. I express costs and benefits in IGE parameters and determine the conditions under which social interactions lead to the evolution of cooperative or harmful behaviours. Therefore, the model I propose combines the conceptual clarity of kin and multilevel selection models with the applicability of IGE models, which parameters can be empirically determined, facilitating the testing of model predictions. Finally, I show that the use of IGE models is strongly limited by the underlying assumption of linearity. I prove that the modelling of interaction dynamics leads to steady state solutions found by IGE models only under limited conditions. In this light, I discuss the relevance of results published previously and propose a solution of how this problem can be addressed.

A trait must genetically correlate with fitness in order to evolve, however, theory suggests that strong directional selection should erode additive genetic variance (Va) in fitness and limit future evolutionary potential. Sexual antagonism and temporal fluctuations in selection are mechanisms that could maintain Va in fitness. Maternal genetic effects could be an additional source of adaptive genetic variation. I used ‘animal models’ to examine a long-term population of red squirrels to determine 1) if either sexual antagonism or temporal fluctuations in selection were maintaining direct Va in fitness or 2) if maternal genetic effects were a source of indirect Va in fitness. While there were environmental trade-offs on juvenile survival, neither sexual antagonism nor temporal fluctuations in selection maintained Va in fitness. Maternal genetic effects on fitness were significant and provide the Va in fitness needed for rapid microevolution. This is the first instance of maternal genetic effects demonstrated as the only genetic variance available for microevolution. / Northern Scientific Training Program, the Arctic Institute of North America, American Society of Mammologists, Queen Elizabeth II Graduate Scholarship in Science and Technology, NSERC Discovery (to Andrew McAdam), NSF (to Andrew McAdam)

Os objetivos deste trabalho foram estimar os efeitos genéticos fixos atuando sobre uma população sintética e testar diferentes modelos e metodologias neste processo de estimação. Os efeitos genéticos fixos testados foram os efeitos aditivos direto e materno de raça e não aditivos diretos e maternos de heterose, perdas epistáticas e complementariedade. Os modelos testados incluem alternada e conjuntamente todos estes efeitos. As metodologias de regressão de cumeeira e regressão por quadrados mínimos foram comparadas assim como dois métodos distintos para determinação do ridge parameter. Uma população sintética, envolvendo as raças Angus e Nelore foi utilizada. Foram utilizados 294.045 registros de desmame e 148.443 registros de sobreano de uma população sintética envolvendo as raças Angus e Nelore. Foram estudadas as seguintes características: ganho de peso do nascimento ao desmame (WG), escores de conformação (WC), precocidade (WP) e musculatura (WM) coletados ao desmame, ganho de peso do desmame ao sobreano (PG), escores fenotípicos de conformação (PC), precocidade (PP) e musculatura (PM) e perímetro escrotal (SC) coletados ao sobreano. Na maioria das análises, os efeitos genéticos fixos estimados foram estatisticamente significativos. O modelo completo, incluindo todos os efeitos genéticos fixos foi o mais indicado nas duas metodologias testadas. Na estimação por regressão de quadrados mínimos, o modelo mais parcimonioso foi o que incluiu apenas os efeitos aditivos de raça e não aditivos de heterose (dominância) e na estimação por regressão de cumeeira o mais parcimonioso foi o aquele que incluiu, além dos dois já referidos, os efeitos não aditivos de perdas epistáticas. As metodologias mostraram-se equivalentes, para os modelos que incluíram apenas efeito aditivo de raça e não aditivo de heterose. Todavia com a inclusão dos efeitos não aditivos de perdas epistáticas e/ou complementariedade, a regressão de cumeeira mostrou-se mais indicada até o momento em que os dados atingiram um determinado volume e estrutura, com grande parte das classes de composições raciais representadas na amostra e, a partir daí os modelos se mostraram equivalentes. Na comparação entre os métodos de determinação do ridge parameter, o mais indicado foi o método que identifica o menor valor possível que produz fatores de inflação de variância abaixo de 10 para todos os regressores estimados. / The objectives of this study were to estimate the fixed genetic effects acting on a synthetic population, as well as test different models and methodologies in this estimation process. The tested fixed genetic effects were the direct and maternal breed additive and direct and maternal heterosis, epistatic loss and complementarity non-additive effects The tested models include alternate and together all these effects. The ridge regression and least square regression methodologies were compared and were also compared two different methods for determining the ridge parameter to use in the ridge regression. A synthetic beef cattle population, involving Angus and Nellore in several breed combinations was used. 294,045 records at weaning and 148,443 records at yearling were used. The traits of weight gain from birth to weaning (WG), phenotypic scores of conformation (WC), precocity (WP) and muscling (WM) collected at weaning, weight gain from weaning to yearling (PG), phenotypic scores of conformation (PC), precocity (PP) and muscles (PM) collected at yearling and scrotal circumference (SC) were used in the analyzes. In most of analyzes, the estimated fixed genetic effects were statistically significant. The complete model, including all fixed genetic effects was the most suitable in the two tested methodologies. In the estimation by least squares regression, the most parsimonious model was the model that included only breed additive and non-additive heterosis (dominance) effects and in the estimation by ridge regression the most parsimonious model was that included, besides the breed additive and non-additive heterosis (dominance) effects, the non-additive epistatic loss effects. Comparing the two methodologies, for models that include only breed additive and non-additive heterosis effects, methodologies proved to be equivalent; with the inclusion of non-additive epistatic loss and / or complementarity effects, ridge regression was more indicated originally. After reached a certain volume and structure, with much of classes of breeds represented in the sample. Both least squares and ridge regression were equivalent. Comparing the methods for determining the ridge parameter, the best method was that which identifies the smallest possible value that produces the variance inflation factors below 10 for all estimated regressors.

Os objetivos deste trabalho foram estimar os efeitos genéticos fixos atuando sobre uma população sintética e testar diferentes modelos e metodologias neste processo de estimação. Os efeitos genéticos fixos testados foram os efeitos aditivos direto e materno de raça e não aditivos diretos e maternos de heterose, perdas epistáticas e complementariedade. Os modelos testados incluem alternada e conjuntamente todos estes efeitos. As metodologias de regressão de cumeeira e regressão por quadrados mínimos foram comparadas assim como dois métodos distintos para determinação do ridge parameter. Uma população sintética, envolvendo as raças Angus e Nelore foi utilizada. Foram utilizados 294.045 registros de desmame e 148.443 registros de sobreano de uma população sintética envolvendo as raças Angus e Nelore. Foram estudadas as seguintes características: ganho de peso do nascimento ao desmame (WG), escores de conformação (WC), precocidade (WP) e musculatura (WM) coletados ao desmame, ganho de peso do desmame ao sobreano (PG), escores fenotípicos de conformação (PC), precocidade (PP) e musculatura (PM) e perímetro escrotal (SC) coletados ao sobreano. Na maioria das análises, os efeitos genéticos fixos estimados foram estatisticamente significativos. O modelo completo, incluindo todos os efeitos genéticos fixos foi o mais indicado nas duas metodologias testadas. Na estimação por regressão de quadrados mínimos, o modelo mais parcimonioso foi o que incluiu apenas os efeitos aditivos de raça e não aditivos de heterose (dominância) e na estimação por regressão de cumeeira o mais parcimonioso foi o aquele que incluiu, além dos dois já referidos, os efeitos não aditivos de perdas epistáticas. As metodologias mostraram-se equivalentes, para os modelos que incluíram apenas efeito aditivo de raça e não aditivo de heterose. Todavia com a inclusão dos efeitos não aditivos de perdas epistáticas e/ou complementariedade, a regressão de cumeeira mostrou-se mais indicada até o momento em que os dados atingiram um determinado volume e estrutura, com grande parte das classes de composições raciais representadas na amostra e, a partir daí os modelos se mostraram equivalentes. Na comparação entre os métodos de determinação do ridge parameter, o mais indicado foi o método que identifica o menor valor possível que produz fatores de inflação de variância abaixo de 10 para todos os regressores estimados. / The objectives of this study were to estimate the fixed genetic effects acting on a synthetic population, as well as test different models and methodologies in this estimation process. The tested fixed genetic effects were the direct and maternal breed additive and direct and maternal heterosis, epistatic loss and complementarity non-additive effects The tested models include alternate and together all these effects. The ridge regression and least square regression methodologies were compared and were also compared two different methods for determining the ridge parameter to use in the ridge regression. A synthetic beef cattle population, involving Angus and Nellore in several breed combinations was used. 294,045 records at weaning and 148,443 records at yearling were used. The traits of weight gain from birth to weaning (WG), phenotypic scores of conformation (WC), precocity (WP) and muscling (WM) collected at weaning, weight gain from weaning to yearling (PG), phenotypic scores of conformation (PC), precocity (PP) and muscles (PM) collected at yearling and scrotal circumference (SC) were used in the analyzes. In most of analyzes, the estimated fixed genetic effects were statistically significant. The complete model, including all fixed genetic effects was the most suitable in the two tested methodologies. In the estimation by least squares regression, the most parsimonious model was the model that included only breed additive and non-additive heterosis (dominance) effects and in the estimation by ridge regression the most parsimonious model was that included, besides the breed additive and non-additive heterosis (dominance) effects, the non-additive epistatic loss effects. Comparing the two methodologies, for models that include only breed additive and non-additive heterosis effects, methodologies proved to be equivalent; with the inclusion of non-additive epistatic loss and / or complementarity effects, ridge regression was more indicated originally. After reached a certain volume and structure, with much of classes of breeds represented in the sample. Both least squares and ridge regression were equivalent. Comparing the methods for determining the ridge parameter, the best method was that which identifies the smallest possible value that produces the variance inflation factors below 10 for all estimated regressors.