Os objetivos deste trabalho foram avaliar diferentes modelos de seleção com base nos efeitos aleatórios maternos considerados para características de crescimento e perímetro escrotal, estimar parâmetros genéticos para pesos do nascimento aos 12 meses (pesos ao nascer, a desmama e aos 12 meses de idade); perímetro escrotal aos 12 meses e correlações genéticas entre as características para bovinos compostos Montana Tropical, em análises uni, bicaracterísticas. Estimar parâmetros genéticos para as características categóricas de musculosidade, precocidade e conformação aos 12 meses em modelos uni e bicaracterísticas utilizando diferentes metodologias para análise (modelos mistos, modelos thresholds bayesianos e modelos lineares generalizados mistos) e compará-los. Nas análises feitas para as características de crescimento e perímetro escrotal os efeitos maternos influenciaram os pesos do nascimento aos 12 meses de idade. As estimativas de herdabilidade direta obtidas das análises bicaracterísticas foram superiores àquelas obtidas das análises unicaracterísticas e as estimativas pela análise bicaracterística para as herdabilidades foram 0,27 para peso ao nascer; 0,18 para peso à desmama; 0,20 para peso aos 12 meses; e 0,19 para perímetro escrotal aos 12 meses. As correlações genéticas estimadas entre pesos obtidos em idades jovens com peso ao ano foram moderadas a baixas (< 0,60). A correlação genética obtida entre perímetro escrotal e características de crescimento foram 0,04, e <0,01 com peso a desmama indicando baixa associação entre as características e de 0,38 com peso aos 12 meses o que pode ao longo do tempo gerar animais mais pesados aos 12 meses. A seleção com base em características de crescimento em qualquer idade pode promover ganhos genéticos moderados no peso corporal de animais do composto Montana Tropical. É importante considerar nas análises os pesos prévios à seleção para estimar parâmetros genéticos para pesos após a seleção. Para as características morfológicas as estimativas de herdabilidade foram de baixas a moderadas e houve diferença entre o modelo linear e o de limiar e o modelo linear generalizado misto, na obtenção de estimativas de herdabilidades e correlações genéticas, de características categóricas morfológicas multinomiais. As estimativas dos parâmetros genéticos obtidas por modelo de limiar foram superiores aos demais métodos avaliados, com valores de 0,42; 0,37 e 0,25 para musculosidade, precocidade e conformação aos 12 meses, e as correlações genéticas estimadas em conjunto com peso a desmama para musculosidade, precocidade e conformação aos 12 meses foram 0,89; 0,22 e 0,83 respectivamente. Os escores visuais de conformação, precocidade e musculatura aos 12 meses podem responder rapidamente à seleção individual. / The objectives this study were to evaluate different models of selection based on maternal random effects considered for growth traits and scrotal circumference; to estimate genetic parameters for weights from birth to 12 months (birth weight, weaning and 12 months old), scrotal circumference at 12 months and to estimate genetic correlations between traits for cattle composite Montana Tropical, by univariate and two-trait analysis. To estimate genetic parameters for categorical traits of muscling, precocity and conformation at 12 months in uni-and two-trait models using different methodologies for analysis (mixed models, Bayesian models thresholds and generalized linear mixed models) and compare them. In the analysis made for growth traits and scrotal circumference, maternal effects influenced the weights from birth to 12 months of age. Direct heritability estimates obtained from two-trait analyzes were higher than those obtained using univariate models, and the estimates of heritability for the two traits analysis were 0.27 for birth weight, 0.18 for weaning weight, 0.20 for weight at 12 months , and 0.19 for scrotal circumference at 12 months. The genetic correlation obtained between weights at young ages and yearling weight were moderate to low ( < 0.60). Genetic correlation obtained between scrotal circumference and growth traits were 0.04, and < 0.01 with weaning weight indicating a low correlation between the traits and 0.38 from weight at 12 months which may over time provide heavier animals at 12 months. Selection based on growth traits at any age can provide moderate genetic gains in body weight of animals of the composite Montana Tropical. It is important to consider in the analysis to selecting the prior weights to estimate genetic parameters for weights after selection. For the morphological traits, heritability estimates were low to moderate and there was a difference between the threshold and linear model and generalized linear mixed model, for estimates of heritability and genetic correlations of morphological multinomial categorical traits. Estimates of genetic parameters by the threshold model were higher than the other methods evaluated, with values of 0.42, 0.37 and 0.25 for muscling, precocity, and conformation at 12 months, and the genetic correlations estimated in with weaning weight for muscling, precocity and conformation at 12 months were 0.89, 0.22 and 0.83 respectively. The visual scores of conformation, precocity and muscling at 12 months can respond quickly to individual selection.

The overall accuracy, efficiency and profitability of livestock improvement strategies can be greatly increased by incorporating quantitative genetics into livestock selection and breeding. Since the introduction of quantitative genetics, a range of traits describing the animal e.g. in terms of health, growth, fecundity, production, have been extensively evaluated in terms of genetics and are now commonly manipulated through breeding to achieve specific selection goals. An industry led enquiry as to the possibility of including spine traits in genetic selection to increase back length in sheep was the basis of the present thesis. Collecting information on spine traits (spine length, vertebrae length and vertebrae number) is of particular interest and use to the sheep breeding industry as there may be the potential to increase meat yield from the highly valuable longissimus thoracis et lumborum (LTL or loin), located parallel to the spine, with little associated change in production costs. The thesis focusses on the use of X-ray computed tomography (CT) scanning as a technique which would allow spine traits to be measured in vivo, hence being useful for genetic selection. The topogram scans produced from the CT scanning procedure were analysed to derive spine trait information for the thesis. The scans were from Scottish Blackface (maternal breed stock), Texel (terminal sire breed), Texel cross Mule and Poll Dorset cross Mule (three-way cross slaughter lambs) so as to represent the divergent genotypes found across the different levels of the United Kingdom’s (UK) three-tier crossbreeding structure of sheep. The present study explored as a first step intra- and inter-operator repeatability of assessment of spine traits from CT derived topograms, as a means to investigate the suitability of the approach for widespread uptake within industry where operators will vary. The results showed that there was high repeatability for intra- and inter-operator assessment of spine trait measurements verifying that the CT method could be accepted as a reliable alternative (to slaughter for example) to quantify spine traits. To determine whether spine traits are similar across the range of breeds representing the key genotypes and crosses in the UK sheep industry, numerous CT topograms were analysed. The results showed marked variation in spine traits within and between Scottish Blackface, Texel, Texel cross Mule and Poll Dorset cross Mule breeds and crosses. For example, the Texel breed was found to have the largest within-breed range for thoracolumbar vertebrae number (17 – 21; the majority possessing 19), but the spine length of these animals was, on average, significantly shorter than the other breed/cross groups. The present study concluded that the significant differences between the breeds and breed types for the particular spine traits were possibly indicative of a genetic control for these traits. Furthermore, investigation into the phenotypic correlations between spine and production traits revealed some directional associations which may prove beneficial for meat production. For example, Scottish Blackface lambs which had a longer length of a specific spine region had an associated decrease in the volume of carcass fat. Texel lambs which had a longer length of a specific spine region had a slightly larger loin muscle area, at a given weight. The present study also examined animals from a population of Texel lambs already heavily selected for increased muscling. The Texel muscling quantitative trait locus (TM-QTL), segregating in these animals and generally in the UK’s Texel sheep population, is expressed through a polar overdominance pattern of inheritance and its effect on the loin (localised muscle hypertrophy) is commonly utilised in the selection and breeding of Texel sheep to improve meat production. Examination of topograms from lambs bearing the whole range of TM-QTL genotypes showed little evidence to suggest that the change in loin shape/increased loin muscling, as a result of the TM-QTL and its inheritance, has led to any associated change in the underlying spine characteristics. This suggests that selection for increased muscling associated with the TM-QTL may be achieved independently of changes in the spine traits studied. The potential to breed for certain spine traits to increase vertebrae number and hence chops or loin yield can be enhanced by establishing the genetic parameters for the traits. The present study employed a collection of performance trait records from Texel lambs to provide the basis for genetic analysis. The results showed different levels of heritability for the different spine traits but also high standard errors. For example, heritability of vertebrae number was dependent on vertebra location: for thoracic vertebrae heritability was high (ℎ2 = 0.99; SE = 0.42), for lumbar vertebrae heritability was low (ℎ2 = 0.08; SE = 0.12), whereas in contrast, thoracolumbar vertebrae heritability was moderate (ℎ2 = 0.44; SE = 0.27). Phenotypic and genetic correlations between all combinations of traits were also obtained. Accurate predictions of the size and direction of response to selection can be achieved through such genetic analysis of traits. The more that is known of the genetic characteristics of traits and their genetic correlations with other economically important traits, the more efficiently it can be built into breeding programmes improving the overall performance of stock. The results of this study showed that providing spine measurements can contribute to the diversity of trait information available to breeders. The present study also suggests that there may be opportunities to select for increased spine length/vertebrae number which would benefit the sheep industry in terms of increased chop number/loin yield. Although more data are needed prior to implementation. Practical uptake of selection for spine traits would be enhanced due to the straightforward nature of the measurements and the high operator repeatability.

Os objetivos deste trabalho foram avaliar diferentes modelos de seleção com base nos efeitos aleatórios maternos considerados para características de crescimento e perímetro escrotal, estimar parâmetros genéticos para pesos do nascimento aos 12 meses (pesos ao nascer, a desmama e aos 12 meses de idade); perímetro escrotal aos 12 meses e correlações genéticas entre as características para bovinos compostos Montana Tropical, em análises uni, bicaracterísticas. Estimar parâmetros genéticos para as características categóricas de musculosidade, precocidade e conformação aos 12 meses em modelos uni e bicaracterísticas utilizando diferentes metodologias para análise (modelos mistos, modelos thresholds bayesianos e modelos lineares generalizados mistos) e compará-los. Nas análises feitas para as características de crescimento e perímetro escrotal os efeitos maternos influenciaram os pesos do nascimento aos 12 meses de idade. As estimativas de herdabilidade direta obtidas das análises bicaracterísticas foram superiores àquelas obtidas das análises unicaracterísticas e as estimativas pela análise bicaracterística para as herdabilidades foram 0,27 para peso ao nascer; 0,18 para peso à desmama; 0,20 para peso aos 12 meses; e 0,19 para perímetro escrotal aos 12 meses. As correlações genéticas estimadas entre pesos obtidos em idades jovens com peso ao ano foram moderadas a baixas (< 0,60). A correlação genética obtida entre perímetro escrotal e características de crescimento foram 0,04, e <0,01 com peso a desmama indicando baixa associação entre as características e de 0,38 com peso aos 12 meses o que pode ao longo do tempo gerar animais mais pesados aos 12 meses. A seleção com base em características de crescimento em qualquer idade pode promover ganhos genéticos moderados no peso corporal de animais do composto Montana Tropical. É importante considerar nas análises os pesos prévios à seleção para estimar parâmetros genéticos para pesos após a seleção. Para as características morfológicas as estimativas de herdabilidade foram de baixas a moderadas e houve diferença entre o modelo linear e o de limiar e o modelo linear generalizado misto, na obtenção de estimativas de herdabilidades e correlações genéticas, de características categóricas morfológicas multinomiais. As estimativas dos parâmetros genéticos obtidas por modelo de limiar foram superiores aos demais métodos avaliados, com valores de 0,42; 0,37 e 0,25 para musculosidade, precocidade e conformação aos 12 meses, e as correlações genéticas estimadas em conjunto com peso a desmama para musculosidade, precocidade e conformação aos 12 meses foram 0,89; 0,22 e 0,83 respectivamente. Os escores visuais de conformação, precocidade e musculatura aos 12 meses podem responder rapidamente à seleção individual. / The objectives this study were to evaluate different models of selection based on maternal random effects considered for growth traits and scrotal circumference; to estimate genetic parameters for weights from birth to 12 months (birth weight, weaning and 12 months old), scrotal circumference at 12 months and to estimate genetic correlations between traits for cattle composite Montana Tropical, by univariate and two-trait analysis. To estimate genetic parameters for categorical traits of muscling, precocity and conformation at 12 months in uni-and two-trait models using different methodologies for analysis (mixed models, Bayesian models thresholds and generalized linear mixed models) and compare them. In the analysis made for growth traits and scrotal circumference, maternal effects influenced the weights from birth to 12 months of age. Direct heritability estimates obtained from two-trait analyzes were higher than those obtained using univariate models, and the estimates of heritability for the two traits analysis were 0.27 for birth weight, 0.18 for weaning weight, 0.20 for weight at 12 months , and 0.19 for scrotal circumference at 12 months. The genetic correlation obtained between weights at young ages and yearling weight were moderate to low ( < 0.60). Genetic correlation obtained between scrotal circumference and growth traits were 0.04, and < 0.01 with weaning weight indicating a low correlation between the traits and 0.38 from weight at 12 months which may over time provide heavier animals at 12 months. Selection based on growth traits at any age can provide moderate genetic gains in body weight of animals of the composite Montana Tropical. It is important to consider in the analysis to selecting the prior weights to estimate genetic parameters for weights after selection. For the morphological traits, heritability estimates were low to moderate and there was a difference between the threshold and linear model and generalized linear mixed model, for estimates of heritability and genetic correlations of morphological multinomial categorical traits. Estimates of genetic parameters by the threshold model were higher than the other methods evaluated, with values of 0.42, 0.37 and 0.25 for muscling, precocity, and conformation at 12 months, and the genetic correlations estimated in with weaning weight for muscling, precocity and conformation at 12 months were 0.89, 0.22 and 0.83 respectively. The visual scores of conformation, precocity and muscling at 12 months can respond quickly to individual selection.