More than 270,000 km of rivers and streams are impaired due to fecal pathogens, creating an economic and public health burden. Fecal indicator organisms such as Escherichia coli are used to determine if surface waters are pathogen impaired, but they fail to identify human health risks, provide source information, or have unique fate and transport processes. Statistical and machine learning models can be used to overcome some of these weaknesses, including identifying ecological mechanisms influencing fecal pollution. In this study, canonical correlation analysis (CCorA) was performed to select parameters for the machine learning model, Maxent, to identify how chemical and microbial parameters can predict E. coli impairment and F+-somatic bacteriophage detections. Models were validated using a bootstrapping cross-validation. Three suites of models were developed; initial models using all parameters, models using parameters identified in CCorA, and optimized models after further sensitivity analysis. Canonical correlation analysis reduced the number of parameters needed to achieve the same degree of accuracy in the initial E. coli model (84.7%), and sensitivity analysis improved accuracy to 86.1%. Bacteriophage model accuracies were 79.2, 70.8, and 69.4% for the initial, CCorA, and optimized models, respectively; this suggests complex ecological interactions of bacteriophages are not captured by CCorA. Results indicate distinct ecological drivers of impairment depending on the fecal indicator organism used. Escherichia coli impairment is driven by increased hardness and microbial activity, whereas bacteriophage detection is inhibited by high levels of coliforms in sediment. Both indicators were influenced by organic pollution and phosphorus limitation.

Made available in DSpace on 2014-07-29T14:52:08Z (GMT). No. of bitstreams: 1 Tese_UFG_Keyla_ORibeiro.pdf: 2021366 bytes, checksum: 719945fe498cc8ef7b6d15b1343d35dd (MD5) Previous issue date: 2011-04-26 / Soybean is currently the most prominent crop in the Brazilian production of grains. Its main products include oil and meal, which have been used for the production of food and animal feed. The high protein contents represent the main attraction of soybean use in diet. Furthermore, soybean oil has also been used in the production of biodiesel. These industrial segments demands cultivars with similar profiles of fatty acids, in which case the oil guarantees a better oxidative stability. This profile corresponds to oils with low polyunsaturated acids contents, associated to high contents of monounsaturated fatty acids. For that, it becomes essential the chemical characterization of grains in cultivars, which could be extended to the genetic resources preserved in germplasm collections, with the purpose of breeding soybean in these traits. Nowadays the characterization of these resources is still deficient, mainly regarding special traits such as those previously mentioned. In this context, the objective of this study was to characterize a sub-collection of 527 accessions of soybean from the germplasm bank of Secretaria de Agricultura, Pecuária e Abastecimento do Estado Goiás (SEAGRO-GO), based on the physicochemical traits of the grain, including protein and oil contents in the meal, and the main fatty acids contents of the oil. The grains were obtained through a field trial carried out at the experimental unit of SEAGRO-GO (16°43 S; 49°7 W; 765 m), county of Senador Canedo - GO, in the 2007-2008 growing season. The experimental design was augmented blocks (blocks of Federer), with two check cultivars (Conquista and Monsoy 6101). The plots were constituted by one row of plants (5.0 m x 0.5 m). The protein contents were quantified by the Kjeldahl method, the oil by the Soxhlet method, and the fatty acids (palmitic, stearic, oleic, linoleic and linolenic) by FAME method (Fatty Acid Methyl Esters). To assess the genetic divergence among accessions the multivariate analyses used were: canonical variables, principal components and cluster analysis. The last one was performed by the hierarchical clustering method using UPGMA criteria. Canonical discriminant analysis was also performed on established similarity groups and on maturity groups of the accessions (early, intermediate and late). In the principal components analysis it was needed three components to retain 78% of the total variance, whereas in the canonical variables analysis only two of these variables captured 81.2% of this variation. About the phenotypic variation, it was observed that the studied sub-collection presents a relatively lowl variability among its accessions, in protein content (from 35.7% to 44.0%) and oil content (from 14.8% to 18.5%). For fatty acids, the germplasm showed a relatively wide variation, although characteristics of industrial interest such as high contents of oleic acid (greater than 40%) and low contents of polyunsaturated acids (less than 40%) are restricted to few accessions. Regarding the polyunsaturated acids, the germplasm presents satisfactory variability to the linoleic oil (omega-6), however, low variability to the linolenic acid (omega-3), especially in its upper limit. By cluster analysis they were established five similarity groups (A, B, C, D, and E) with the formation of subgroups in some them. The accessions of groups A and B present the largest oil and protein contents, respectively. The accessions of group C outstood with the best profile of oil quality (oxidative stability), and they were followed by the accessions of E and B groups. Among the germplasm with the best profiles of oil oxidative stability, it was still possible to identify in this collection five divergent groups of accessions with significant differences in fatty acids contents. The polyunsaturated and saturated fatty acids are the most relevant variables in the structuring of genetic diversity of this germplasm. / A soja é atualmente a cultura de maior destaque na safra brasileira de grãos. Entre seus principais produtos estão o óleo e o farelo, os quais têm sido matéria-prima para a produção de alimentos e ração animal. O elevado teor de proteína representa o principal atrativo do uso da soja na alimentação. Além disso, o óleo de soja tem sido também utilizado na produção de biodiesel. Esses segmentos industriais buscam cultivares com perfis semelhantes de ácidos graxos, cujo óleo garanta melhor estabilidade oxidativa. Isto corresponde a óleos com baixos teores de ácidos poli-insaturados, associados a elevados teores de ácidos graxos monossaturados. Para isso, torna-se indispensável a caracterização de atributos relacionados à qualidade do grão dos cultivares comerciais; o que pode ser estendido, para os propósitos de melhoramento da soja nesses atributos, aos recursos genéticos preservados em bancos de germoplasma. Atualmente a caracterização desses recursos ainda é deficiente, sobretudo quando se tratam de caracteres especiais como aqueles anteriormente mencionados. Nesse contexto, o objetivo deste trabalho foi caracterizar uma subcoleção de 527 acessos de soja do banco de germoplasma da Secretaria de Agricultura, Pecuária e Abastecimento do Estado de Goiás (Seagro-GO), com base nos teores de proteína, óleo e dos principais ácidos graxos no óleo. Os grãos foram obtidos de ensaio de campo conduzido na Unidade Experimental da Seagro-GO (16°43 S; 49°7 W; 765 m), município de Senador Canedo-GO, na safra 2007/2008. O delineamento experimental foi em blocos aumentados (blocos de Federer), com duas cultivares testemunhas (Conquista e Monsoy 6101) e parcela constituída de uma fileira de plantas (5,0 m x 0,5 m). Os teores de proteína foram quantificados pelo método Kjeldahl; de óleo, por Soxhlet; e os de ácidos graxos (palmítico, esteárico, oleico, linoleico e linolênico), pelo método FAME (Fatty Acid Methyl Esters). Para a avaliação da divergência genética entre os acessos, utilizaram-se as análises multivariadas de variáveis canônicas, de componentes principais e de agrupamento. Esta última foi realizada pelo método hierárquico aglomerativo, associado ao critério de ligação UPGMA. Também se realizaram análises discriminantes canônicas entre os grupos de similaridade identificados e entre os ciclos de maturação dos acessos (precoce, médio e tardio). Na análise de componentes principais foram necessárias três componentes para explicar 78% da variância total, ao passo que na análise de variáveis canônicas, apenas duas destas variáveis explicaram 81,2% dessa variação. Em relação à variação fenotípica, observou-se que a subcoleção apresenta variabilidade relativamente baixa entre os acessos, em seus teores de proteína (de 35,7% a 44,0%) e óleo (de 14,8% a 18,5%). Para os ácidos graxos, o germoplasma apresenta variação relativamente ampla, embora características de interesse industrial como elevados teores de ácido oleico (maiores que 40%) e baixos teores de ácidos poli-insaturados (menores que 40%) sejam restritas a poucos acessos. Quanto aos ácidos graxos poli-insaturados, o germoplasma apresenta variabilidade satisfatória para o ácido linoleico (ômega-6), porém, baixa variabilidade para o ácido linolênico (ômega-3), sobretudo em seu limite superior. Pela análise de agrupamento foram estabelecidos cinco grupos de similaridade multivariada (A, B, C, D e E), com formação de subgrupos em alguns deles. Os acessos dos grupos A e B apresentam os maiores teores de óleo e de proteína, respectivamente. Os acessos do grupo C destacam-se com os melhores perfis em qualidade do óleo (estabilidade oxidativa). Mesmo selecionando apenas o germoplasma com melhor perfil em estabilidade oxidativa do óleo, ainda foi possível identificar nessa coleção cinco grupos de acessos significativamente divergentes nos teores de ácidos graxos. As variáveis que se mostraram mais relevantes na estruturação da divergência genética desse germoplasma foram os teores de ácidos graxos poli-insaturados e saturados.