The aims of this thesis were to investigate whether characteristics associated with animal welfare are genetically and genomically determined by using quantitative and molecular genetic approaches and to develop strategies indicating how these traits could be used in breeding programmes. Two traits that are closely related to animal welfare and associated with high socio-economic values are piglet survival at birth and aggressive behaviour in pigs. Piglet survival traits were analysed based on quantitative Bayesian approaches using phenotypic and pedigree information only, while aggressive behaviour was analysed based on molecular genetic approaches such as genome-wide association studies and genomic selection using additionally a dense panel of single nucleotide polymorphisms (SNP). The latter approach was validated using behavioural traits related to welfare characteristics in a welldocumented mouse data set. Selection for piglet survival at birth is expected to be effective, because all lines and breeds in this thesis showed considerable variation for this trait and relatively high heritabilities, particularly in lines with low average birth weight. Maternal heritabilities of individual birth weight were mostly at moderate magnitude and thus of great interest for selection. The genetic correlations between piglet survival and birth weight indicated that selection for either individual or average birth weight or variation of birth weight within litter would indirectly increase survival. The genetic associations of piglet survival with economically important (re)production traits are of great importance for breeding organisations. Undesirable genetic correlations between piglet survival and (re)production traits were generally of low magnitude, so that simultaneous improvement of all traits could be achieved. A comparison of five breeds and lines showed that differences in genetic parameters between breeds and lines can be substantial and no single selection strategy would be optimal for all. A unique study of a sire and a dam line originating from one breed but selected for more than 25 years with different breeding goals demonstrated how selection pressure can alter the genetic parameters over years. Breeding organisations should therefore consider selection strategies per breed or line individually to achieve maximum overall improvement. This study gives new insight into the use of genomic selection for traits associated with animal welfare. It is one of the first to present estimates for linkage disequilibrium in the pig using a new 60K SNP panel and the first to evaluate the efficiency of genomic selection against aggressive behaviour in pigs. Genomic selection showed a high predictive ability in comparison to traditional polygenic selection. It was especially advantageous for traits with lower heritabilities. In particular in situations where little family information was available, the performance of polygenic selection was low and genomic selection increased the performance considerably. Reducing the number of SNPs did not significantly change the performance of genomic selection. The consistently high performance across models indicates that low-density SNP panels may be sufficient to ensure a high efficiency of genomic selection, thus reducing the high costs associated with genotyping in pig breeding with its short generation interval. To summarize, this thesis has shown how to optimise quantitative and genomic approaches to improve animal welfare related characteristics efficiently in pig breeding programmes.

Orientadores: Daniella Jorge de Moura, Yamilia Barrios Tolon / Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Engenharia Agricola / Made available in DSpace on 2018-08-11T08:30:36Z (GMT). No. of bitstreams: 1 Silva_WagnerTome_M.pdf: 2437625 bytes, checksum: f516c3f988e62a79c594ca308b78ec78 (MD5) Previous issue date: 2008 / Resumo: As atuais tendências de mercado e exigências dos consumidores, juntamente com regulamentações de normas e padronizações relacionadas ao bem-estar animal têm provocado um forte impacto sobre a produção suinícola brasileira e podem afetar as exportações. Na mesma medida, a preocupação com a produção de alimentos está cada vez mais voltada para a quantidade e a qualidade do produto final e também às suas implicações no meio ambiente. Isso está acontecendo justamente no momento em que novas ferramentas de tecnologia de informação aplicam-se aos sistemas de produção animal, principalmente em sistemas de suporte a decisão, usando dados da produção animal, fisiologia e comportamento. Também, o uso de sistemas computacionais e protocolos específicos foram desenvolvidos para apoiar os produtores, provendo soluções apropriadas e direcionadas aos eventuais problemas. A vocalização dos animais é a expressão do seu estado específico, que pode ocorrer espontaneamente, ou pode ser o resultado de um evento externo, por exemplo, o desmame, a fome, e a dor, que mostrarnndo vocalizações de alta freqüência; por este motivo a vocalização dos animais transformou-se em urna ferramenta muito importante para a avaliação do bemestar animal. O objetivo do presente trabalho foi desenvolver um sistema que permita o registro, a análise e a interpretação da vocalização dos animais com a finalidade de monitorar as indicações de estresse na produção comercial de suínos. O sistema foi desenvolvido para registraro e análisanalisare dea vocalização de suínos. A validação foi realizada em urna granja de suínos localizada no. estado de São Paulo, onde. foram captadas as expressões vocais dos suínos em situações de estresse e não estresse. As seguintes situações de estresse foram consideradas na maternidade: impedimento do acesso ao aleitamento, temperaturas baixas de 15 e 20oe, retirada do leitão da proximidade da matriz, disputa pelo teto, corte da cauda, corte dos dentes, castração e aplicação de ferro; e na creche foram: retirada de alimento e retirada do leitão da baia. Foi usado um microfone acoplado a um computador portátil posicionado a 1,0 m dos animais, para registro dos sons. O sinal sonoro foi digitalizado pela placa de som e gravado através do software Audacity. Para o reconhecimento, tratamento e classificação dos sinais gerados pelas expressões vocais dos leitões optou-se pela utilização da Transformada de Fourier. As seguintes características foram usadas para a classificação dos sinais: duração da vocalização, formato de onda, espectro total e evolução do espectro de vocalização ao longo do tempo. O sistema usou o software "Vocalização", desenvolvido em linguagem Delphi 5 e possui os seguintes recursos: transformação do sinal digital em valores de amplitude em função do tempo; regressão polinomial de ordem até 50 para obtenção de envoltórias em tomo dos pontos coletados; transformada de Fourier do espectro total da vocalização; transformada de Fourier para intervalos de tempo de 40 milissegundos, com os valores de freqüência divididos por cinco faixas que vão de 100 a 5000 Hz. Para a análise estatística dos valores de freqüência e nível de intensidade sonorà utilIzou-se o software Minitab e para a análise estatística dos polinômios foi desenvolvido um software denominado "Poli" que compara estatisticamente os coeficientes dos polinômios gerados pelo software "Vocalização". Os resultados mostraram que, quando comparado o tempo de vocalização dos leitões em creche e em maternidade, concluiu-se que análises de intervalos de vocalização de 1 segundo seriam suficientes para determinar o nível de estresse do animal tanto em creche quanto em maternidade. O sistema de monitoramento do bem-estar usando tecnologia de informação permitiu o registro, a análise e a interpretação da vocalização dos suínos, indicando o nível de estresse na produção comercial de suínos / Abstrat: Today's market and consumers demands trends beside the welfare norms and regulations have imposed a strong impact on Brazilian swine production and may affect export. At the same time food production worry has been turn for final product quantity and quality and to environmental implications as wel1. This is happening at the same time when new control technologies emerge. During the last decade new information technology tools were introduced in the animal production system mainly for decision support systems, using as basics data on animal production, physiology and behavior. Also the use of computational systems and specific protocols were developed providing appropriate solutions directed to eventual problems. Animal vocalization is an expression of its specific state that may occur spontaneously or yet as a result of an external event, for instance weaning, hunger or pain in the form of high frequency noise. Because of that animal vocalization has become an important tool for assessing animal welfare. The objective of this research was to develop a system that allows the recording, analysis and the interpretation of animal vocalization with the goal of monitoring stress indication during swine commercial production. The system was developed for application at vocalization recording and analyzing data. The validation was done in both a farrowing and a growing housing in a commercial swine farm in Sao Paulo state where the swine vocal expressions in both situations of stressful and non-stressful were captured. The following stressful events considered at farrowing were: to restrain access to farrowing, to expose to low temperatures (below 20°C), to remove the piglet from the sow, the dispute of the teats from the piglets, the tail cut, the teeth cut, the castration, and the iron marking; and at growing were: feed removal and removal of the piglet from the room. A microphone placed at 1 m distance from the animaIs was connected to a laptop computer for recording the noise. The noise signal was digitalized and recorded in a sound hardware using the software "Audacity@". For the recognition and classification of the signals generated by the swine vocal expressions Fourier Transform were used. The following characteristics were used for the noise classification: vocalization duration, wave format, total spectrum, and the spectrum evolution along time. The system used the software called V ocalização that was developed in Delphi 5 language and with the output of the following resources: t-ransform the digital signal into amplitude values as function of time; polynomial regression of up to order 50 for obtaining involve functions around the collected points, Fourier Transform for the total spectrum for each 40 s time intervals with frequency values divided into five that go blocks from 100 to 5000 Hz. For performing the statistical analysis of the frequency and leveI of sound intensity the software Minitab was used. For statistically analyzing the polynomials software called Poli was developed, and it compares statistically the polynomials coefficients generated by the software V ocalização.'Resutts showed that when compared the piglets vocalization length time it was concluded that the interval analysis of 1 s was enough to determine their stress leveI both in farrowing and growing rooms. The welfare monitoring system using information technology allowed the recording, analysis and data interpretation of swine vocalization indicating the stress level in commercial production / Mestrado / Construções Rurais e Ambiencia / Mestre em Engenharia Agrícola