No description available.

Ecólogos de comunidades têm como um dos principais objetivos identificar padrões de abundância e distribuição de espécies ao longo das regiões e ambientes, e inferir sobre os processos norteadores desses padrões. Para tanto, duas linhas teóricas complementares são principalmente utilizadas: limitação de similaridade e filtros ambientais. Enquanto a primeira prediz que as espécies dentro de uma comunidade devem ser distintas nas suas características como resultado da competição (divergência), a segunda espera uma alta similaridade entre espécies coexistentes (convergência), uma vez que elas são filtradas sob certas condições ambientais. Estudos sobre diversidade funcional podem nos dar uma ideia da proporção de divergência e convergência na estruturação das comunidades, enquanto a composição funcional pode indicar a direção das respostas funcionais aos filtros ambientais. Ambientes campestres predominantemente cobrem o bioma Pampa, no sul do Brasil, e nesta região as florestas são restritas a zonas ribeirinhas e a pequenas encostas. As inundações são uma pressão comum em matas ribeirinhas, mas dadas as condições da região elas também são vulneráveis à dessecação por evapotranspiração e ventos. Em contraste, as florestas de encosta são mais sombreadas e úmidas. No presente trabalho, temos como objetivo principal caracterizar a diversidade e composição funcional destes ambientes florestais. Esperamos encontrar padrões de convergência para os atributos de florestas ciliares, devido às condições ambientais mais severas. Estabelecemos 18 unidades amostrais de 250 x 10 m em três regiões do Pampa, nove para cada tipo de floresta. Todas as árvores com pelo menos 5 cm de DAP (diâmetro à altura do peito) foram medidas. As regiões representam um gradiente de precipitação e temperatura. Seis atributos funcionais das espécies arbóreas foram avaliados: área foliar, área foliar específica (SLA), conteúdo de matéria seca foliar (LDMC), altura máxima, síndrome de dispersão e tamanho do fruto. Índices de diversidade funcional e seus tamanhos de efeito padronizados (SES - standardized effect size) dada a riqueza local de espécies foram estimados para cada região e tipo florestal. A composição funcional foi analisada através de média dos atributos funcionais ponderada pelas abundâncias nas comunidades (CWM), com posterior análise de variância para verificar diferenças entre as florestas e regiões. Os valores de CWM de área foliar e proporção de zoocoria foram menores nas florestas ribeirinhas de todas as regiões, enquanto o tamanho do fruto, SLA e LDMC só foram influenciados pelo fator região, sendo a SLA menor na região de menor precipitação e temperatura médias anuais. Altura máxima média das comunidades diferiu tanto conforme o tipo de floresta (maior na encosta) quanto à região (maior na região com condições climáticas mais favoráveis). Os valores do SES para riqueza funcional e dispersão funcional não demostraram padrões de divergência ou convergência significativos. Apenas a riqueza funcional para altura e tamanho de fruto foi distinta de acordo com o ambiente e a região das florestas avaliadas. Os resultados deste trabalho, ao contrário do esperado, demonstram não haver padrões de divergência e/ou convergência funcional para as florestas do Pampa, em termos do ambiente em que ocorrem (ciliar ou encosta) ou mesmo região. Contudo, a composição funcional variou tanto em relação ao ambiente quanto à região. Isso pode indicar uma seleção de espécies com distintas estratégias adaptativas de acordo com as diferentes influências ambientais em escala local e regional. / Community ecologists aim at identifying patterns of species abundance and distribution along regions and environments, inferring on their main driver processes. Two complementary approaches rule community assembly: limiting similarity and environmental filters. While the first predicts that species within a community should be distinct in traits to avoid competition (divergence), the second expects a high similarity between co-occurring species (convergence), since they are filtered out to occur under certain environmental conditions. Functional diversity can give us an idea of the proportion of divergence and convergence in the structure of communities, while the functional composition might indicate the direction of trait responses to the environmental filters. Grasslands predominantly cover the southern Brazilian Pampa, and the forests are restricted to riparian zones and hillslopes. Flooding is a common pressure on riparian forests, but given the environmental conditions of the region they are also vulnerable to desiccation by evapotranspiration and winds. In contrast, slope forests are more shaded and humid. We aimed at characterizing functional diversity and composition of these forest environments. We expected trait convergence for riparian forests because of the harsher environmental conditions. We established 18 sampling units of 250 x 10 m in three regions of the Pampa, nine for each forest type. All trees with at least 5cm DBH were measured. The regions represent a gradient in precipitation and temperature. Six functional traits of the tree species were assessed: leaf area, specific leaf area (SLA), leaf dry matter content (LDMC), plant height, dispersal mode, and fruit size. Functional diversity indices and their standardized effect size (SES), considering the local species richness, were estimated for each region and forest. Functional composition was analyzed through community-weighted mean of traits (CWM), with further analysis of variance to verify differences between forests and regions. CWM of leaf area and zoochory were lower in riparian forests of all regions, while fruit size, SLA and LDMC were only influenced by the region. SLA was lower in the region with lower annual mean precipitation and temperature. Maximum height shows response for the type of forest (higher CWM in hillslope forests) and region (higher in the region under more favorable climatic conditions). SES of functional richness and dispersion did not show a recurrent pattern of significant divergence or convergence related to the factors. The results of SES, contrary to our expectation, are not indicating any patterns of trait divergence and/or convergence for the forests. Only the functional richness of height and fruit size differed according to the habitat and the region of the studied forests. However, both forest environments and regions differed in terms of functional composition, indicating the assembling of species with distinct adaptive strategies in relation to different environment influences for local and regional scale.

O caranguejo de água doce Trichodactylus fluviatilis Latreille, 1828 apresenta uma considerável variabilidade morfológica, aliada a uma ampla distribuição geográfica e ocupação de ambientes costeiros e continentais. Tal variabilidade tem gerado, em alguns casos, dúvidas quanto à delimitação da espécie. O presente trabalho tem como objetivo averiguar se as populações de T. fluviatilis apresentam divergências morfológicas e genéticas compatíveis com o nível intraespecífico avaliando-se a hipótese de validade deste táxon. Para isto, foi realizada a análise da variabilidade genética entre as populações e uma revisão taxonômica. O material analisado foi obtido por meio de coletas, visitas e empréstimos de coleções carcinológicas do Brasil. Foram analisados caracteres descritos na literatura e obtidas sequências parciais dos genes mitocondriais 16S rRNA e citocromo c oxidase subunidade I (COI). Dentre os caracteres morfológicos analisados alguns tiveram muita variação, enquanto outros se mostraram bem informativos para alguns grupos. O resultado da análise molecular mostrou a formação de clados internos com altas divergências genéticas entre eles, tanto para o gene 16S quanto para o COI. Além disso, a espécie T. petropolitanus foi alocada entre os clados reconhecidos morfologicamente como T. fluviatilis. Tais estruturações genéticas, aliada ao polimorfismo morfológico, mostraram claramente que a espécie reconhecida morfologicamente como T. fluviatilis não forma um grupo monofilético, podendo ser considerada um complexo de espécies, que precisa de ajustes taxonômicos consideráveis. / The freshwater crab Trichodactylus fluviatilis Latreille, 1828 presents a considerable morphological variability, as well as a wide geographical distribution and occupancy of coastal and continental environments. Such variability has generated, in some cases, doubts concerning the species delimitation. The present work aims to investigate whether the populations of T. fluviatilis exhibit morphological and genetic divergence compatible with the intraspecific level, assessing the hypothesis of validity of this taxon. Therefore, we have performed the analysis of genetic variability among populations and a taxonomic revision. The material analyzed was obtained from field expeditions, visits and loans from carcinological collections. We analyzed morphological characters described in the literature as well as obtained partial sequences of the 16S rRNA and cytochrome c oxidase subunit I (COI) mitochondrial genes. Some morphological characters analyzed had a wide variation, while others were well informative for some groups. The results of molecular analysis showed the formation of internal clades with high genetic divergence among them, both for 16S and COI. Moreover, the species T. petropolitanus was allocated among clades recognized morphologically as T. fluviatilis. Such genetic structuration and morphological polymorphism clearly indicate that the species morphologically recognized as T. fluviatilis does not form a monophyletic group. Therefore, it must be considered as a species complex, which claims for huge taxonomic adjustments.

Mémoire numérisé par la Direction des bibliothèques de l'Université de Montréal.

Grafické modely jsou vhodným nástrojem statistické analýzy. V poslední době byly aplikovány rovněž ve financích. Práce pojednává o grafických modelech určených pro zpracování dat pocházejících z normálního rozdělení. Jsou zde popsány možnosti testování shody modelů s daty a čtyři selekční algoritmy (včetně odvození vztahů mezi různými STOP pravidly) pro výběr vhodného grafu, který dobře reprezentuje konkrétní data. Tyto dosud nepříliš známé postupy jsou pak použity k zodpovězení otázek z oblasti českého i mezinárodního kapitálového trhu (zkoumání provázanosti odvětvových indexů BCPP, globalizace světových akciových trhů, provázanosti měnových kurzů) a rovněž k testování hypotéz pro vysvětlení časové struktury úrokových sazeb v České republice. Vytvořené programy ve formátu Mathematica 4 lze nalézt v příloze.

The social mole-rats have often been typecast as extreme examples of mammalian sociality. With their pronounced reproductive skew, status-related contrasts in lifespan and morphology, and the suggestion of a division of labour amongst helpers, mole-rat societies have repeatedly been likened to the structurally complex societies of some eusocial insects. However, because few studies of mole-rats have quantified individual variation in growth and behaviour across long periods of development, it has remained unclear the extent to which mole-rat societies, and the features of individuals within them, should be considered unique amongst social vertebrates. In this thesis, I examine life history variation in Damaraland mole-rats Fukomys damarensis from three perspectives- growth, behaviour, and ageing- to explore how individual developmental trajectories contribute to, and are influenced by, the structure of mole-rat societies. First, I use a large longitudinal dataset to test for the presence of behavioural specialisation in non-breeding mole-rat helpers. I find no indication of individual specialisation in cooperative activities. Instead, individual differences in helping behaviour are largely the result of age-related changes in the extent to which individuals commit to all forms of helping (Chapter 3); refuting the notion of helper castes. I then focus on the variation in growth across non-breeders, developing a novel biphasic model to accurately quantify sex differences in growth and explore the influence of social effects on growth trajectories (Chapter 4). Despite the proposition of intense intrasexual competition in mole-rat societies, there was no clear signature of sex-specific competition on helper growth trajectories. A more conspicuous form of socially-mediated growth in mole-rats is the secondary growth spurt displayed by females that have acquired the dominant breeding position, causing them to become larger and more elongated. By experimentally controlling reproduction in age-matched siblings, I show that rather than being stimulated by the removal from reproductive suppression, this adaptive morphological divergence is achieved through a lengthening of the lumbar vertebrae when breeding is commenced (Chapter 5). With contrasts in size and shape following the acquisition of the breeding role, this status-related growth pattern shares similarities with growth in naked mole-rats and other social vertebrates. Breeders also show a twofold greater lifespan than non-breeders in Fukomys mole-rats, prompting the suggestion that the transition to dominance also sets individuals onto a slower ageing trajectory. To date, there is little evidence to support a physiological basis to lifespan extension in breeders. This assertion is bolstered by the absence of longer telomeres or slower rates of telomere attrition in breeding females compared to non-breeding females residing in groups (Chapter 6), each of which might be expected if breeders age more slowly. I argue that previous studies exploring status-related ageing in captive Fukomys mole-rats have overlooked the importance of demographic processes (and associated behavioural influences) on mortality schedules. Irrespective of the proximate basis of the longer lifespan of breeders, at an interspecific level the social mole-rats are unusually long-lived for their size. A recent large-scale comparative analysis concluded that prolonged lifespan is a general characteristic of all mammalian cooperative breeders, but this conclusion is premature, as in most of the major clades containing both cooperative and non-cooperative species there is no consistent trend towards lifespan extension in cooperative species (Chapter 7). In the case of mole-rats, it seems more likely that their exceptional longevity arises principally from their subterranean habits and related reductions in extrinsic mortality. Overall, these findings demonstrate that cooperative breeding has important consequences for individual life histories, but there is no strong basis for the claim that Damaraland mole-rat societies are markedly different in form than other cooperative breeding societies.

The rise of earnings inequality in the United States has garnered attention in both the political and academic spheres. Recently progressive politicians have pointed towards the divergence of wages and labor productivity as a source of this inequality. known as the productivity-pay gap as a source of the rise in inequality. This paper analyzes that divergence with a regression model that evaluates the change in compensation that is attributable to increases in productivity. Results were somewhat surprising with productivity accounting for a larger portion of the growth in wages for the period after 1972 when the divergence in the two growth rates began than in the time between 1948 and 1972 when they were said to grow together. Additionally, results showed more wage growth was attributable to increases in productivity in goods producing sectors like manufacturing, utilities, and construction than financial intermediation in the services sector. However standard errors across our model were relatively large making it difficult to say with certainty the size of effects observed. Future research should seek to better define productivity in the service sector to determine whether other factors like education, occupation or area of residence affect the level of wage growth attributable to compensation.

This thesis identifies and extends techniques that can be linked to the principle of maximum entropy (maxent) and applied to parameter estimation in machine learning and statistics. Entropy functions based on deformed logarithms are used to construct Bregman divergences, and together these represent a generalization of relative entropy. The framework is analyzed using convex analysis to charac- terize generalized forms of exponential family distributions. Various connections to the existing machine learning literature are discussed and the techniques are applied to the problem of non-negative matrix factorization (NMF).

Techniques that can effectively identify authors of texts are of great importance in scenarios such as detecting plagiarism, and identifying a source of information. A range of attribution approaches has been proposed in recent years, but none of these are particularly satisfactory; some of them are ad hoc and most have defects in terms of scalability, effectiveness, and computational cost. Good test collections are critical for evaluation of authorship attribution (AA) techniques. However, there are no standard benchmarks available in this area; it is almost always the case that researchers have their own test collections. Furthermore, collections that have been explored in AA are usually small, and thus whether the existing approaches are reliable or scalable is unclear. We develop several AA collections that are substantially larger than those in literature; machine learning methods are used to establish the value of using such corpora in AA. The results, also used as baseline results in this thesis, show that the developed text collections can be used as standard benchmarks, and are able to clearly distinguish between different approaches. One of the major contributions is that we propose use of the Kullback-Leibler divergence, a measure of how different two distributions are, to identify authors based on elements of writing style. The results show that our approach is at least as effective as, if not always better than, the best existing attribution methods-that is, support vector machines-for two-class AA, and is superior for multi-class AA. Moreover our proposed method has much lower computational cost and is cheaper to train. Style markers are the key elements of style analysis. We explore several approaches to tokenising documents to extract style markers, examining which marker type works the best. We also propose three systems that boost the AA performance by combining evidence from various marker types, motivated from the observation that there is no one type of marker that can satisfy all AA scenarios. To address the scalability of AA, we propose the novel task of authorship search (AS), inspired by document search and intended for large document collections. Our results show that AS is reasonably effective to find documents by a particular author, even within a collection consisting of half a million documents. Beyond search, we also propose the AS-based method to identify authorship. Our method is substantially more scalable than any method published in prior AA research, in terms of the collection size and the number of candidate authors; the discrimination is scaled up to several hundred authors.

Les systèmes de partage collaboratif de connaissance sont des collecticiels qui permettent à des communautés en ligne de partager des connaissances. Ces systèmes offrent des espaces partagés où les membres d'une communauté peuvent construire ensemble une mémoire partagée. Malheureusement, ces systèmes restent peu utilisés car ils sont peu motivant, paraissent artificiels et le partage de connaissance est vécu comme forcé. Un des problèmes avec les systèmes actuels de partage de connaissance est qu'ils voient l'activité de partage comme une accumulation centralisée d'information plutôt que comme un processus évolutif dans lequel un groupe construit de la connaissance et dans lequel des divergences et des inconsistances dans la mémoire peuvent apparaître.<br />Dans ce travail, nous étudions les communautés de partage de connaissance, en portant une attention particulière sur la divergence. Cette divergence dans la connaissance mémorisée correspond à un conflit cognitif et se traduit par l'apparition d'alternatives, d'arguments et de points de vue différents autour d'un concept ou d'un sujet d'intérêt. Dans ce contexte, l'activité de partage de connaissance est comprise comme un processus évolutif englobant notamment des étapes de discussion où les participants expriment leurs points de vue, et dont l'objectif est la construction commune d'un concept partagé. Dans ce cadre, les divergences de point de vue sont des pratiques courantes. <br />En assistant les membres dans le partage et l'expression de leurs divergences, nous pensons améliorer l'utilisabilité des systèmes de partage de connaissance. Dans cette thèse, nous faisons l'hypothèse qu'une attention spéciale dans le support de la divergence et des discussions autour de la divergence est une condition nécessaire pour concevoir et améliorer l'utilisabilité d'une application collaborative pour le partage de connaissance. Cette approche propose des besoins nouveaux aux applications de type TCAO/collecticiel. Ainsi, le thème central de ce travail est d'identifier et concevoir des mécanismes de gestion logicielle de la collaboration pour améliorer le support au partage de connaissance en supportant les divergences et inconsistances.<br />Nous développons un cadre théorique qui conceptualise l'activité de partage de connaissance et considère la divergence comme un objet de premier ordre. Ce cadre propose une vision d'un système de partage de connaissance comme étant basé sur une application collaborative conduite par un procédé de partage de connaissance. Il prend en compte le développement distribué et collaboratif d'une mémoire commune, l'apparition et la conservation de divergences et l'autonomie des participants. En particulier, nos objectifs sont de montrer qu'une approche de construction par augmentation monotone de la mémoire, des espaces de travail différenciés (privé/public) et des services de conscience de groupe adaptés afin de permettre aux utilisateurs de suivre l'évolution de la mémoire partagée, sont des éléments clés dans la conception d'un collecticiel de partage de connaissance.