The objective of this study was to characterize the wild bee fauna of a managed mixed-grass prairie in southwestern Manitoba. Weekly sampling using two methods, sweep-netting and bee bowls, was conducted over two years (2005-2006) at three sites within the Yellow Quill Mixed-grass Prairie Preserve. Spatial and seasonal patterns in diversity indices, taxonomic composition and ecological composition of the bee fauna were identified and investigated in relation to sampling method, environmental conditions, and floral resource availability. A total of 7014 individual bees representing five families and 100 species were collected. Numerically, social nesters from the genera Lasioglossum and Bombus dominated, while mining species of Andrena represented the greatest species richness. Observed spatial and seasonal patterns in the abundance, diversity and composition of the bee community were strongly modified by sampling method, resource limitation in the second year and the presence of the exotic invasive plant leafy spurge (Euphorbia esula).

The objective of this study was to characterize the wild bee fauna of a managed mixed-grass prairie in southwestern Manitoba. Weekly sampling using two methods, sweep-netting and bee bowls, was conducted over two years (2005-2006) at three sites within the Yellow Quill Mixed-grass Prairie Preserve. Spatial and seasonal patterns in diversity indices, taxonomic composition and ecological composition of the bee fauna were identified and investigated in relation to sampling method, environmental conditions, and floral resource availability. A total of 7014 individual bees representing five families and 100 species were collected. Numerically, social nesters from the genera Lasioglossum and Bombus dominated, while mining species of Andrena represented the greatest species richness. Observed spatial and seasonal patterns in the abundance, diversity and composition of the bee community were strongly modified by sampling method, resource limitation in the second year and the presence of the exotic invasive plant leafy spurge (Euphorbia esula).

This thesis considers the problem of performing inference on undirected graphical models with continuous state spaces. These models represent conditional independence structures that can appear in the context of Bayesian Machine Learning. In the thesis, we focus on computational methods and applications. The aim of the thesis is to demonstrate that the factorisation structure corresponding to the conditional independence structure present in high-dimensional models can be exploited to decrease the computational complexity of inference algorithms. First, we consider the smoothing problem on Hidden Markov Models (HMMs) and discuss novel algorithms that have sub-quadratic computational complexity in the number of particles used. We show they perform on par with existing state-of-the-art algorithms with a quadratic complexity. Further, a novel class of rejection free samplers for graphical models known as the Local Bouncy Particle Sampler (LBPS) is explored and applied on a very large instance of the Probabilistic Matrix Factorisation (PMF) problem. We show the method performs slightly better than Hamiltonian Monte Carlo methods (HMC). It is also the first such practical application of the method to a statistical model with hundreds of thousands of dimensions. In a second part of the thesis, we consider approximate Bayesian inference methods and in particular the Expectation Propagation (EP) algorithm. We show it can be applied as the backbone of a novel distributed Bayesian inference mechanism. Further, we discuss novel variants of the EP algorithms and show that a specific type of update mechanism, analogous to the mirror descent algorithm outperforms all existing variants and is robust to Monte Carlo noise. Lastly, we show that EP can be used to help the Particle Belief Propagation (PBP) algorithm in order to form cheap and adaptive proposals and significantly outperform classical PBP.

The aim of this methodically oriented thesis was to find spatial distribution of water bugs in fishpond by parallelly used two different sampling methods - classic method (netting) and light traps as well as to find, if sampling method can influence a result of ecological study by substantial way. Study was realized in the one locality - fishpond Polepšil (region: Vysočina: Moravskobudějovicko) in years 2007 and 2008. Spaces {--} transects for sampling were difined. Efficiency of water bug sampling by nett is conspicuously less than sampling by light traps. More species and individuals (in total sum of samples and in samples from individual transects too) were sampled by light traps under comparable conditions. Chemical yelow-green light of light traps was more attractive for water bugs than the pink one. Spatial preferences of some corixid species and their developmental stages were noted by application of light trap method of sampling. Light trap method of samplings brings more complete and valuable data for ecological studies of water bug communities.

Nessa dissertação, avaliei a eficiência e congruência de métodos para amostragem de mamíferos de maior porte e investiguei a influência de fatores múltiplos correlacionados, comuns em paisagens fragmentadas, sobre a distribuição destes animais em remanescentes de Mata Atlântica. Através de amostragens padronizadas em 24 remanescentes florestais de uma paisagem rural com 49% de remanescentes florestais no Planalto Paulista e de um delineamento pareado, no capítulo 2 comparei o desempenho e a congruência de dois métodos (pegadas em parcelas de areia e armadilhas fotográficas) e de dois tipos de isca (banana e iscas de cheiro) para a estimativa da riqueza e taxa de ocorrência de mamíferos de maior porte. Ambos os métodos se mostraram adequados para o estudo destes animais em florestas tropicais e dos fatores que afetam sua distribuição em paisagens alteradas, pois (1) registram as espécies de menor porte e noturnas, (2) podem ser padronizados entre áreas heterogêneas, (3) apresentam desempenho semelhante no registro da maioria das espécies e da riqueza de espécies, e (4) refletem de maneira similar o padrão de ocorrência das espécies entre diferentes áreas. Frente às iscas de cheiro utilizadas, a banana foi a isca mais eficiente na amostragem tanto das espécies herbívoras/frugívoras quanto das espécies onívoras, destacando a necessidade da padronização das iscas utilizadas e do uso de iscas complementares ou de iscas que atraiam uma ampla gama de animais. Usando os dados obtidos com armadilhas fotográficas, avaliei quais fatores condicionam a presença e a freqüência de ocorrência de cães domésticos (capítulo 3) e de mamíferos de maior porte (capítulo 4) nos remanescentes. Esta avaliação foi realizada através da seleção de modelos de regressão construídos com base na relação causal entre os fatores estudados - que pode ser estabelecida a partir de como usualmente se dá a expansão das atividades humanas em florestas neotropicais - e comparados através do critério de informação AIC (The Akaike Information Criterion). Para o cão doméstico, foram considerados quatro fatores: extensão de estradas, quantidade de mata e número de cães domésticos no entorno, e qualidade da vegetação. Foi observada uma forte relação positiva entre o total de residentes e o total de cães domésticos no entorno dos remanescentes. Dentro dos remanescentes, o cão doméstico foi mais registrado do que sete das 11 espécies nativas, sua ocorrência foi mais bem explicada pela quantidade de cães no entorno, e sua freqüência de ocorrência pela diminuição da quantidade de mata no entorno, o que mostra a necessidade de controle populacional desta espécie no entorno de áreas naturais e da manutenção de áreas florestadas mais extensas, menos suscetíveis a entrada deste invasor. Para os mamíferos de maior porte, os fatores considerados foram: quantidade de mata e número de residentes no entorno dos remanescentes, distância à estrada de asfalto, freqüência de ocorrência de cães domésticos nos remanescentes, e qualidade da vegetação dos remanescentes. Os resultados demonstraram que, apesar da ampla variação na resposta das espécies a estes fatores, a distância a estradas de asfalto e a freqüência de ocorrência de cães domésticos, fatores que são correlacionados a disponibilidade de habitat e não são freqüentemente considerados, determinaram a distribuição de um número maior de espécies. A comunidade de mamíferos de maior porte encontrada é simplificada, dominada por espécies generalistas e com poucos dispersores e predadores de sementes e plântulas, o que pode levar a conseqüências negativas para outras espécies e para o funcionamento e regeneração das florestas remanescentes. Estes resultados sugerem que para a conservação de uma comunidade íntegra de mamíferos de maior porte na Mata Atlântica é necessário, não apenas a manutenção de grandes contínuos de mata nativa ou de paisagens com percentual alto de matas remanescentes, mas a redução drástica das perturbações associadas à ocupação e atividades humanas. O manejo e restauração de paisagens devem considerar a proximidade entre os remanescentes florestais e estradas, principalmente as asfaltadas e de alto tráfego, e a importância de campanhas de controle populacional e vacinação de cães domésticos. / In this master thesis, I evaluated the efficiency and congruence of methods for sampling large mammals, and investigated the influence of multiple and correlated factors, common in fragmented landscapes, on the distribution of these animals in Atlantic Forest remnants. Through standardized surveys at 24 forest remnants in a rural landscape with 49% of remaining forest in the Planalto Paulista and using a paired design, in chapter 2 I compared the performance and the congruence of two methods (tracks in sand plots and camera-trapping) and of two types of baits (banana and scent lures) for estimating richness and rate of occurrence of large mammals. Both methods are suitable for studying these animals in tropical forests as well as the factors that affect their distribution in disturbed landscapes, because (1) they record medium-sized and nocturnal species, (2) can be standardized among heterogeneous sites, (3) present similar performance for recording most species and richness, (4) reflect in similar ways the pattern of species occurrence among different sites. Compared with scent lures, banana was the most efficient bait for sampling herbivores/frugivores as well as omnivores, highlighting the need to standardize baits and use complementary baits or baits that attract a wide range of animals. Using camera-trapping data, I evaluated which factors determine the presence and the frequency of occurrence of domestic dogs (chapter 3) and of large mammals (chapter 4) in remnants. This evaluation was done through the selection of regression models buit based on the causal relationship among studied factors which can be established from the way expansion of human activities usually happens in neotropical forests and compared through the AIC, The Akaike Information Criterion. For the domestic dog, four factors were considerate: extension of roads, amount of forest and number of domestic dogs in the surroundings, and vegetation quality. It was observed a strong positive relationship between the total number of residents and the total number of domestic dogs in the surroundings. Inside remnants, the domestic dog was more frequently recorded than seven of the 11 native species, its occurrence was better explained by the number of dogs in the surroundings, and its frequency of occurrence by the reduction of the amount of forest in the surroundings, which indicates the need of population control of this invasive species in the surroundings of natural areas and of the maintenance of extensive forests areas that are less susceptible to the entrance of this invasive species. For native large mammals, the considered factors were: amount of forest and number of residents in the surroundings, distance from paved roads, frequency of occurrence of domestic dogs and vegetation quality. Results suggested that, despite the large variation in the responses of species to the studied factors, distance from paved roads and frequency of occurrence of domestic dog, factors that are correlated with habitat availability and are not frequently considered, determined the distribution of a larger number of species. The large mammal community found is simplified, dominated by generalist species and with few seed and seedling dispersers and predators, which can lead to negative consequences to other species and to forest functioning and regeneration. These results suggest that for conserving a complete large mammal community at the Atlantic Forest, not only the maintenance of large tracts of continuous forest or landscapes with high percentage of remaining forest is needed, but also a drastic reduction of disturbances associated with human activities and occupation. Landscape management and restoration should consider the proximity between forest remnants and roads, especially those that are paved and present high traffic, and the importance of campaigns of domestic dog population control and vaccination.

Made available in DSpace on 2016-12-23T14:37:34Z (GMT). No. of bitstreams: 1 JANAINA SILVA HOFFMAN REDLING.pdf: 895970 bytes, checksum: aaadd8d1e660ee210c4aceb4ea8f3c98 (MD5) Previous issue date: 2007-02-15 / The objective of this work was to compare the sampling method plots of equal size with the sampling methods of distance (point-center quarter and Prodan) for floristic and phitosociology characterization in tree species community of a forest fragment with 94,38 ha approximately in the Fazenda Recanto Verde, located in Alto Jequitibá, surroundings of the Parque Nacional do Caparaó-MG, Zona da Mata Mineira. For the plots of equal size method were allocated 17 parcels of 20 x 30 m (600 m2). For localization and marking in the field a GPS of navigation was used having as reference the central point of the parcel. For marking of referring points to the distance methods a grid in a map was created from generated image of the satellite IKONOS, gotten in August/2004, being allocated 104 points with 100 m of distance between lines and 50 m between points inside the lines of sampling, In each parcel and point, the trees with equal or superior DAP the 5 cm was measured. The families Araliaceae, Burseraceae, Celastraceae, Chrysobalanaceae, Lecythidaceae, Rhamnaceae, Simaroubaceae, Thymeliaceae and Vochysiaceae are observed only in the plots of equal size method, whereas the families Elaeocarpaceae and Lamiaceae are shown only in the point-center quarter and Prodan methods. The species with greeter VI(%) are: Planthymenia foliolosa plots of equal size method, Dalbergia nigra point-center quarter and Mabea fistulifera Prodan method. The diametric structure of the forest parcel studied follows the general standard reversed-J-shape of the uneven-aged forests for the three methods of sampling. Considering all the estimates for wealth, diversity and horizontal and diametric structures gotten by the three methods of sampling, beyond the sampling intensity and time in the application of the each method, the plots of equal size method demonstrates to be the more trustworthy and with better performance in terms of time in the proposal to portray the arboreal structure studied. / Este trabalho teve como objetivo comparar o método de amostragem de parcelas de área fixa com métodos de amostragem de distância (quadrantes e Prodan) para caracterização florística e fitossociológica da comunidade arbórea do fragmento florestal com aproximadamente 94, 38 ha na fazenda Recanto Verde, localizado no município de Alto Jequitibá entorno do Parque Nacional do Caparaó- MG, Zona da Mata Mineira. Para o método de parcelas de área fixa foram alocadas 17 parcelas 20 x 30m (600 m2). Para localização e marcação no campo, utilizou-se um GPS de navegação tomando-se como referência o ponto central da parcela. Para marcação dos pontos referentes aos métodos de distâncias, foi criada uma malha em um mapa gerado a partir de uma imagem do satélite IKONOS, obtida em agosto de 2004. Foram alocados 104 pontos com distância entre linhas amostrais de 100 metros e distância entre pontos dentro da linha de amostragem igual a 50 metros. Em cada parcela e ponto, foram medidas todas as árvores com DAP (diâmetro à altura do peito a 1,30 m de altura) igual ou superior a 5 cm. As famílias Araliaceae, Burseraceae, Celastraceae, Chrysobalanaceae, Lecythidaceae, Rhamnaceae, Simaroubaceae, Thymeliaceae e Vochysiaceae são observadas apenas no método de parcelas de área fixa, enquanto que as famílias Elaeocarpaceae e Lamiaceae são amostradas apenas nos métodos de quadrantes e Prodan. As espécies com maior VI(%) são: Plathymenia foliolosa método de parcelas de área fixa, Dalbergia nigra método de quadrantes e Mabea fistulifera método de Prodan. A estrutura diamétrica do fragmento florestal estudado segue o padrão geral das florestas ineqüiâneas, em J invertido para os três métodos de amostragem. Considerando-se todas as estimativas para riqueza, diversidade e estruturas horizontal e diamétrica obtidas pelos três métodos de amostragem, além da intensidade amostral e tempo gasto na aplicação de cada método, o método de parcelas de área fixa demonstra ser o mais fidedigno e com melhor desempenho em termos de tempo na proposta de retratar a comunidade arbórea estudada.

Este trabalho faz parte do projeto Biota/FAPESP, Métodos de Inventário da Biodiversidade de Espécies Arbóreas, e analisou a diferença na estrutura florística entre as áreas estudadas. Na Estação Ecológica de Assis, SP, área de Cerradão, foram amostradas 102 espécies, 72 gêneros e 43 famílias; 67% das espécies foram exclusivas desta fitofisionomia. Na Floresta Estacional Semidecidual da Estação Ecológica dos Caetetus, município de Gália-SP, foram amostradas 208 espécies, 138 gêneros e 49 famílias; sendo 65% das espécies exclusivas. Na Floresta Ombrófila Densa do Parque Estadual de Carlos Botelho, município de São Miguel Arcanjo-SP, foram encontradas 410 espécies, pertencentes a 152 gêneros e 64 famílias; 84% das espécies exclusivas desta fitofisionomia. A Floresta Estacional apresentou maior similaridade com a Floresta Ombrófila do que com o Cerradão. Apesar de apresentar maior número de espécies, a Floresta Ombrófila Densa apresenta concentração de espécies nas famílias Myrtaceae e Lauraceae. Em outra análise foram calculados os índices de distinção taxonômica nas 5 grades amostrais para os métodos de amostragem testados. O método de Bitterlich apresentou tendência nos índices de diversidade e distinção taxonômica, conforme a diversidade taxonômica do sub-bosque. Esses índices apresentaram média independente e variância decrescente com o aumento da amostra. Os índices de distinção taxonômica média e variância da distinção taxonômica média também apresentaram variância decrescente. Porém, a estabilização da média ocorreu com amostras de maior tamanho, principalmente para as parcelas de área fixa na Estação Ecológica dos Caetetus. Na Floresta Ombrófila Densa, a distinção taxonômica média foi menor e a variância da distinção taxonômica média foi maior, devido a alta concentração de espécies de Myrtales e Laurales nesta fitofisionomia. / This research is part of the project Biota / FAPESP, \"Tree Species Biodiversity Inventory Methods, and analised floristic structural differences among three areas. In Assis Ecological Station, Forest Savanna area were sampled 102 species, 72 genus and 43 families, 67% of this species were exclusive of this phytophisiognomy. Semideciduous seasonal forest in the Caetetus Ecological Station, 208 species were sampled, 138 genus and 49 families, with 65% of exclusive species. Rain Forest in the Carlos Botelho State Park, were found 410 species, belonging to 152 genus and 64 families, 84% of exclusive species. The seasonal forest showed greater similarity to the rain forest than the Savanna. In spite of the greater number of species, the Rain Forest presents concentration of species in the Myrtaceae and Lauraceae families. In another analysis were calculated the taxonomic indices in tree species data provided by sampling methods tested in five sampling grids. The Bitterlich sampling method was tendecious in estimates the taxonomic diversity and taxonomic distinctness indices because the taxonomic diversity in the understory. These indices showed average independent and variance decreasing with increasing sample. The average taxonomic distinctness and variation in taxonomic distinctness also showed decreased, but the stabilization of the average occurred with samples of greater size, mainly for fixed area plots at Caetetus Ecological Station. In the Rain Forest the average taxonomic distinctness was lower and the variation of taxonomic distinctness was higher due to high concentration of species of Myrtales and Laurales in this phytophisiognomy.

Submitted by Geandra Rodrigues (geandrar@gmail.com) on 2018-06-19T13:13:53Z No. of bitstreams: 1 rodrigobarbosadesantis.pdf: 2597054 bytes, checksum: b19542ca0e9312572d8ffa5896d735db (MD5) / Approved for entry into archive by Adriana Oliveira (adriana.oliveira@ufjf.edu.br) on 2018-06-27T11:12:01Z (GMT) No. of bitstreams: 1 rodrigobarbosadesantis.pdf: 2597054 bytes, checksum: b19542ca0e9312572d8ffa5896d735db (MD5) / Made available in DSpace on 2018-06-27T11:12:01Z (GMT). No. of bitstreams: 1 rodrigobarbosadesantis.pdf: 2597054 bytes, checksum: b19542ca0e9312572d8ffa5896d735db (MD5) Previous issue date: 2018-03-26 / CAPES - Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / Falta de materiais é um problema comum na cadeia de suprimentos, impactando o nível de serviço e eficiência de um sistema de inventário. A identificação de materiais com grande riscos de falta antes da ocorrência do evento pode apresentar uma enorme oportunidade de melhoria no desempenho geral de uma empresa. No entanto, a complexidade deste tipo de problema é alta, devido ao desbalanceamento das classes de itens faltantes e não faltantes no inventário, que podem chegar a razões de 1 para 100. No presente trabalho, algoritmos de classificação são investigados para proposição de um modelo preditivo para preencher esta lacuna na literatura. Algumas métricas específicas como a área abaixo das curvas de Característica Operacionais do Receptor e de Precisão-Abrangência, bem como técnicas de amostragem e comitês de aprendizado são aplicados nesta tarefa. O modelo proposto foi testado em dois estudos de caso reais, nos quais verificou-se que adoção da ferramenta pode contribuir com o aumento do nível de serviço em uma cadeia de suprimentos. / Material backorder (or stockout) is a common supply chain problem, impacting the inventory system service level and effectiveness. Identifying materials with the highest chances of shortage prior its occurrence can present a high opportunity to improve the overall company’s performance. However, the complexity of this sort of problem is high, due to class imbalance between missing items and not missing ones in inventory, which can achieve proportions of 1 to 100. In this work, machine learning classifiers are investigated in order to fulfill this gap in literature. Specific metrics such as area under the Receiver Operator Characteristic and precision-recall curves, sampling techniques and ensemble learning are employed to this particular task. The proposed model was tested in two real case-studies, in which it was verified that the use of the tool may contribute with the improvemnet of the service level in the supply chain.

La disparition et la fragmentation d’habitat sont considérées comme des causes majeures de l’érosion de la biodiversité.Les forêts sont parmi les plus riches écosystèmes terrestres de la planète. La simplification structurelle et la disparition d’éléments d’habitats clés qui résultent de leur exploitation mettent en péril la biodiversité qu’elles abritent.Grâce à l’analyse de différents jeux de données mutualisés entre plusieurs laboratoires et de deux jeux de donnéesoriginaux générés par mes travaux, l’objectif de cette thèse est d’évaluer le rôle joué par plusieurs éléments de la Trame de Très Vieux Bois pour la diversité des coléoptères saproxyliques aux échelles locale et paysagère. L’effet des caractéristiques du milieu et la biodiversité associée aux ilots de vieillissement, réserves forestières et arbres isolés extra-forestiers ont été évalués.En parallèle, nous avons conduit des explorations méthodologiques concernant la principale technique d’échantillonnagede notre modèle biologique. Nous avons mis en évidence les effets (i) forts de la réplication spatiale ou temporelle des dispositifs d’échantillonnage sur les données obtenues, mais (ii) négligeables de l’exclusion d’une famille couteuse en temps d’identification sur les résultats.Nos résultats écologiques indiquent l’effet de l’arrêt d’exploitation sur la reconstitution des compartiments bois mort etdendromicrohabitats et des assemblages de coléoptères saproxyliques associés, au contraire de l’extension de rotationreprésentée par les ilots de vieillissement. D’importantes densités de réserves forestières dans le paysage (plus de 20%) semblent nécessaires pour une conservation efficace de la faune saproxylique. De plus, nous avons mis en évidence qu’une part importante des espèces saproxyliques est préférentiellement retrouvée sur des arbres solitaires, non-forestiers. La totalité de la faune saproxylique n’est donc pas abritée par la forêt. Ces structures non forestières représentent des éléments primordiaux à prendre en compte pour la sauvegarde de la biodiversité saproxylique. / Habitat loss and fragmentation are considered as major threats to biodiversity in forests, one of the species-richest terrestrial ecosystems worldwide. The structural simplification and the loss of key habitat elements resulting from forest harvesting threaten forest biodiversity. Through the analysis of datasets shared between laboratories and two original datasets created during this thesis, I intended to evaluate the role for associated beetle diversity of several components of the saproxylic habitat network at local and landscape scales. The effect of local environmental variables has been evaluated and the biodiversity associated with ageing stands, forest reserves and isolated non-forest trees were explored. In parallel, we have analyzed the main technique used to sample our study group. We highlighted (i) strong effects of spatial or temporal replication of sampling on data, but (ii) negligible effects of data simplification by excluding a time-expensive family on results. We demonstrated significant effects of forest setting aside on the restoration, mainly of dead wood and tree microhabitats, but also on associated saproxylic beetle assemblages. Extended rotations in ageing stands did not provide such positive effects. Besides, we evidenced that a high density of forest reserves in the landscape (over 20%) seems necessary to efficiently favor the saproxylic fauna. In addition, we observed that a significant proportion of saproxylic species prefer non-forest solitary tree habitats. The whole saproxylic fauna is therefore not hosted by forests. These non-forest saproxylic structures have to be included in conservation strategies dedicated saproxylic biodiversity.

Audit is an interdisciplinary activity which requires, nowadays more than ever in the past, knowledge and skills in the field of accounting and related financial knowledge, as well as from other non-financial sectors (such as the statistics or the information technology). The use of statistics in the audit may not be entirely obvious at first sight and thus the main aim of this thesis is to point out possible ways to apply statistical tools in substantive testing (which represents an important stage of the audit work). The most important use of statistics in the audit can be found in sampling methods. The range of a tested population may be too wide for testing the whole population that leads to the fact that it is necessary to select an audit sample appropriately. The thesis outlines what kind of method and technique of selection to choose and how to generalize the results obtained during the sample testing to the entire population. Besides the main aim in the form of the application of statistical tools in the audit, the thesis also focuses on a brief description of the audit history, polemic of current understanding of the term audit, professional organizations and regulations of the audit in the Czech Republic and description of individual stages of audit work (especially on a detailed analysis of a substantive testing).