Dans cette recherche, nous abordons le problème de le recherche de services qui répondent à des besoins des utilisateurs exprimés sous forme de requête en texte libre. Notre objectif est de résoudre les problèmes qui affectent l'efficacité des modèles de recherche d'information existant lorsqu'ils sont appliqués à la recherche de services dans un corpus rassemblant des descriptions standard de ces services. Ces problèmes sont issus du fait que les descriptions des services sont brèves. En effet, les fournisseurs de services n'utilisent que quelques termes pour décrire les services souhaités. Ainsi, lorsque ces descriptions sont différentes des phrases dans les requêtes ce qui diminue l'efficacité des modèles classiques qui dépendent de traits observables au lieu de traits sémantiques latents du texte. Nous avons adapté une famille de modèles de recherche d'information (IR) dans le but de contribuer à accroître l'efficacité acquise avec les modèles existant concernant la découverte de services. En outre, nous avons mené des expériences systématiques afin de comparer notre famille de modèles IR avec ceux de l'état de l'art portant sur la découverte de service. Des résultats des expériences, nous concluons que notre modèle basé sur l'extension des requêtes via un thésaurus co-occurrence est plus efficace en terme des mesures classiques utilisées en IR que tous les modèles étudiés dans cette recherche. Par conséquent, nous avons mis en place ce modèle dans S3niffer, qui est un moteur de recherche de service basé sur leur description standard. / In this research, we address the problem of retrieving services which fulfil users' need expressed in query in free text. Our goal is to cope the term mismatch problems which affect the effectiveness of service retrieval models applied in prior re- search on text descriptions-based service retrieval models. These problems are caused due to service descriptions are brief. Service providers use few terms to describe desired services, thereby, when these descriptions are different to the sentences in queries, term mismatch problems decrease the effectiveness in classical models which depend on the observable text features instead of the latent semantic features of the text. We have applied a family of Information Retrieval (IR) models for the purpose of contributing to increase the effectiveness acquired with the models applied in prior research on service retrieval. Besides, we have conducted systematic experiments to compare our family of IR models with those used in the state-of-the-art in service discovery. From the outcomes of the experiments, we conclude that our model based on query expansion via a co-occurrence thesaurus outperforms the effectiveness of all the models studied in this research. Therefore, we have implemented this model in S3niffer, which is a text description-based service search engine.

Dans cette recherche, nous abordons le problème de le recherche de services qui répondent à des besoins des utilisateurs exprimés sous forme de requête en texte libre. Notre objectif est de résoudre les problèmes qui affectent l'efficacité des modèles de recherche d'information existant lorsqu'ils sont appliqués à la recherche de services dans un corpus rassemblant des descriptions standard de ces services. Ces problèmes sont issus du fait que les descriptions des services sont brèves. En effet, les fournisseurs de services n'utilisent que quelques termes pour décrire les services souhaités. Ainsi, lorsque ces descriptions sont différentes des phrases dans les requêtes ce qui diminue l'efficacité des modèles classiques qui dépendent de traits observables au lieu de traits sémantiques latents du texte. Nous avons adapté une famille de modèles de recherche d'information (IR) dans le but de contribuer à accroître l'efficacité acquise avec les modèles existant concernant la découverte de services. En outre, nous avons mené des expériences systématiques afin de comparer notre famille de modèles IR avec ceux de l'état de l'art portant sur la découverte de service. Des résultats des expériences, nous concluons que notre modèle basé sur l'extension des requêtes via un thésaurus co-occurrence est plus efficace en terme des mesures classiques utilisées en IR que tous les modèles étudiés dans cette recherche. Par conséquent, nous avons mis en place ce modèle dans S3niffer, qui est un moteur de recherche de service basé sur leur description standard. / In this research, we address the problem of retrieving services which fulfil users' need expressed in query in free text. Our goal is to cope the term mismatch problems which affect the effectiveness of service retrieval models applied in prior re- search on text descriptions-based service retrieval models. These problems are caused due to service descriptions are brief. Service providers use few terms to describe desired services, thereby, when these descriptions are different to the sentences in queries, term mismatch problems decrease the effectiveness in classical models which depend on the observable text features instead of the latent semantic features of the text. We have applied a family of Information Retrieval (IR) models for the purpose of contributing to increase the effectiveness acquired with the models applied in prior research on service retrieval. Besides, we have conducted systematic experiments to compare our family of IR models with those used in the state-of-the-art in service discovery. From the outcomes of the experiments, we conclude that our model based on query expansion via a co-occurrence thesaurus outperforms the effectiveness of all the models studied in this research. Therefore, we have implemented this model in S3niffer, which is a text description-based service search engine.

This paper deals with corpus linguistics. There are two applications under its scrutiny. Both of these applications are processing data from the corpus DeReKo via corpus-driven approach. It is a co-occurrence analysis and a Co-occurrence database. The aim of the work is to evaluate whether the results obtained by the co-occurrence analysis of the current scope of DeReKo are different from the results of the Co-occurrence database, which was created on a basis of a smaller scale corpus. In addition, this thesis offers illustrative examples of the use of both applications and the evaluation of their effectiveness, depending on the purpose of the research. The theoretical part of the thesis deals with the terminology of corpus linguistics and with the mentioned corpuses, which serve as a basis for the practical part of the thesis. The empirical part of the thesis consists of analyses of the randomly picked words (one from each word class) in both applications. The results confirm that the data obtained with Co-occurrence database and co-occurrence analysis are in many respects different and thus confirm the hypothesis that the corpus size plays a crucial role in the results. Both applications have their advantages and disadvantages. The paper offers a comprehensive overview and by doing so it...

Parmi les problèmes les plus fréquemment soulevés en biogéographie, figure celui de l’intégration des interactions écologiques dans les modèles de distribution d’espèces. Bien que la littérature scientifique apporte un ensemble de preuves soulignant le rôle prépondérant des interactions dans la structuration des communautés locales, on trouve relativement peu d’études révélant les empreintes laissées par les interactions dans les données de distribution d’espèces. Proposer une explication simple et claire à ce problème demeure un défi important que la biogéographie doit mener. Le problème majeur que pose l’absence de réponse claire sur le rôle des interactions aux larges échelles spatiales est que la plupart des scénarios de changements de biodiversité partent de l’hypothèse que les interactions sont négligeables. Si cette hypothèse est régulièrement rejetée, alors il faut réviser ces scénarios et soutenir le développement de méthodologies incluant les relations entre les espèces. Je commence cette thèse par un travail théorique sur le sujet car les théories classiques en biogéographie relèguent souvent au second plan les interactions écologiques. Au premier chapitre, je traite de l'intégration des interactions écologiques dans un modèle théorique de distribution d'espèces issue d'une des théories les plus importantes en biogéographie: la théorie de la biogéographie des îles. Ce travail montre comment les effets conjoints des facteurs biotiques et abiotiques changent les attendus de la théorie classique. En m'appuyant sur ce premier chapitre, je montre au second chapitre comment les interactions peuvent se répercuter dans les données de co-occurrence d’espèces. Ces données indiquent la présence ou l’absence de plusieurs espèces sur un même ensemble de sites dispersés sur de larges étendues spatiales. À l’aide d’un modèle probabiliste, j'obtiens des résultats théoriques liant les données de co-occurrence et l’information contenue dans les réseaux écologiques.Je démontre clairement que les interactions affectent les données de co-occurrence. Je montre également que plus le nombre d’interactions séparant deux espèces est grand, moins leur interactions indirect est détectable. De même si une espèce entretient de nombreuses interactions, il sera difficile de trouver une quelconque trace des interactions dans les données de co-occurrence pour cette espèce. Au troisième chapitre, je présente l’analyse de cinq jeux de données de co-occurrence pour lesquels la description des interactions était disponible. Avec ces donnés, j'ai été capable de confirmer les hypothèses du second chapitre en montrant que les espèces qui interagissent co-occurrent différemment de celles n’interagissant pas. Mes résultats indiquent aussi que l’abondance d'interactions est un frein à leur détection dans les données de co-occurrence. Cependant, en intégrant la similarité des facteurs abiotiques pour les différents sites, je montre que les signaux de co-occurrence s’affaiblissent pour parfois disparaitre. Mes résultats suggèrent donc qu’en utilisant des facteurs abiotiques pour inférer les probabilités de co-occurrence,une partie du lien entre les espèces est capturée, mais cette part est entachée d’une grande incertitude. Ceci vient questionner la qualité des prédictions données par les modèles classiques de distribution d'espèces actuellement utilisés. Les résultats de ma recherche apportent des éléments théoriques nouveaux sur le rôle des interactions écologiques dans le tracé des aires de répartition des espèces en plus de proposer une méthode originale pour étudier les données de co-occurrence d’espèces : les regarder à la lumière des réseaux écologiques. Avant de conclure ma thèse, je propose au chapitre 4 une démarche prometteuse pour aller encore améliorer l’intégration des interactions en biogéographie : les introduire par le biais des contraintes énergétiques, ce qui offre une base solide pour une théorie métabolique de la biogéographie. / One of the most pressing challenges currently in the field of biogeography is the successful integration of ecological interactions in species distribution models. Although the scientific literature points out the evidence of the controlling role interactions play on local community structure, relatively few studies have demonstrated its importance over large geographical gradients. Developing a concise, clear explanation for this issue remains a significant challenge that biogeographers need to answer. The main issue associated to the lack of a clear answer concerning the role of interactions at broad spatial scales is that most of scenarios of biodiversity changes assume that interactions can be ignored. When tested, if this hypothesis is proven false, then a re-consideration of species distribution models and their development must be undertaken to include relationships among species. I begin this thesis with a theoretical investigation on this topic, where classical theories have typically ignored ecological interactions. In the first chapter of the thesis I present the integration of interaction networks into a theoretical model of species distribution coming from one of the most important theory in biogeography: the theory of island biogeography. This work shows how together the biotic and abiotic factors can affect the expectations derived from the classical theory. Building upon the findings in the first chapter, in the second chapter, I show how interactions can affect co-occurrence (between species) data. Such data contains the presence or absence of several species for a similar set of sites dispersed along large latitudinal gradients. Using a probabilistic model, I obtain theoretical results linking co-occurrence data and the information included in ecological networks. I clearly demonstrate that interactions shape co-occurrence data. Furthermore, I show that the higher the number of links between two species, the more difficult it is to detect their indirect interaction. Similarly, if a species experiences many interactions, it is then challenging to detect any sign of interactions in co-occurrence data for this species.In the third chapter of the thesis, I assess five sets of co-occurrence data, which had descriptions of their interactions available. Using this data, I was able to confirm my hypotheses put forth in my second chapter, by showing that species co-occur differently from non-interacting one. These results also point out that the abundance of interaction must preclude their detection in co-occurrence data. However, when accounting for abiotic similarities among sites, signals of interactions are weakened. Therefore, my results suggest that using abiotic factors to infer co-occurrence probabilities capture a part of the link between species and further pinpoint the uncertainty associated to this part. As a result of these findings, the predictive power of classical species distribution models used to date is brought into question. My research findings bring new theoretical elements to the forefront when considering the influence of ecological interactions and how they shape species geographical distributions, while also introducing an original methodology for studying species co-occurrence: examining them in the light of ecological networks. Before concluding, my fourth and final chapter, I propose a promising new avenue to further investigate integrating species interactions in biogeography. Here, I introduce interactions in terms of energetic constraints, which will provide a sound basis for a metabolic theory of biogeography.

This thesis is focused on texture analysis methods. The project contains an overview of widely used methods. The main aim of the thesis is to develop a method for texture analysis of retinal images, which will be used for distinction of two patient groups, one with glaucoma eyes and one healthy. It is observed that glaucoma patients don´t have a texture on the eye ground. Preprocessing of the images is found by transfer of the image to different color spaces to achieve the best emphasis of the eye ground texture. Co-occurrence matrix is chosen for texture analysis of this data. The thesis contains detail description of the chosen solutions and feature discussion and the result is a list of features, which can be used for distinction between glaucoma and healthy eyes. The method is implemented in Matlab environment.

A Mata Atlântica é uma floresta tropical úmida considerada um \"hotspot\" de biodiversidade e endemismo. É uma das florestas mais antigas do mundo e uma das maiores florestas da América, abrangendo aproximadamente 150 milhões de hectares em condições ambientais altamente heterogêneas. Estudos em diferentes ambientes da Mata Atlântica, nos núcleos de Picinguaba e Santa Virginia no Parque Estadual da Serra do Mar (PESM), têm sido realizados para determinar a diversidade de espécies e alterações da estrutura das comunidades de bactérias, tanto na filosfera, quanto na dermosfera e solo rizosférico. No entanto, pouco se sabe sobre as funções ecológicas dessas bactérias, e sobre as interações ecológicas entre as comunidades microbianas e os ambientes onde se desenvolvem. Assim o objetivo desse trabalho foi explorar as interações entre as comunidades microbianas da filosfera, dermosfera e solo coletado sobre a projeção da copa de duas espécies arbóreas da Mata Atlântica ao longo de um gradiente altitudinal, usando análises de co-ocorrência, a partir dos dados obtidos por pirosequenciamento da região V4 do gene rRNA 16S de bactérias, para determinar padrões de associações de bactérias em diferentes níveis taxonômicos em cada microambiente. Para esse estudo, foi proposta a hipótese de que mesmo que as condições ambientais sejam diferentes em cada tipo de floresta (gradiente altitudinal), pode existir grupos de bactérias específicos que co-ocorrem na filosfera, dermosfera ou solo das plantas, funcionando como taxons chaves na estruturação das comunidades bacterianas. Com base do sequenciamento dos genes rRNA 16S, as comunidades bacterianas associadas à filosfera e dermosfera de E. edulis e G. opposita nas diferentes florestas foram mais similares entre si do que as do solo. Actinobacteria, Firmicutes, Bacteroidetes e Proteobacteria foram mais abundantes em todos os microambientes estudados. Diferenças nas estruturas das comunidades bacterianas na filosfera, dermosfera e solo foram observadas ao longo do gradiente altitudinal, independente da espécie de planta. Na floresta de terras baixas, a comunidade bacteriana associada à filosfera foi mais similar entre E. edulis e G. opposita. No solo, a comunidade bacteriana foi mais similar dentro de cada tipo de floresta do que entre florestas, sugerindo um efeito da fisionomia da floresta nas comunidades de bactérias dos solos. Explorando as redes de co-ocorrência das comunidades bacterianas em cada microambiente observou-se que no nível de UTOs, cada microambiente têm diferentes táxons chaves que podem regular as interações ecológicas da comunidade. Embora táxons chaves não representam as UTOs mais abundantes em cada microambiente, eles pertencem, predominantemente às classes Alphaproteobacteria e Gammaproteobacteria, sugerindo que na filosfera, dermosfera e solo o core microbioma não pode ser definido ao nível de UTO, mas possivelmente a níveis taxonômicos mais elevados representando grandes grupos microbianos que apresentam funções redundantes. / The Atlantic Forest is a rainforest considered a hotspot of biodiversity and endemism. It is one of the oldest forests in the world and one of the largest forests of America, covering approximately 150 million hectares in highly heterogeneous environmental conditions. Studies in different environments of the Atlantic forest, in the Picinguaba and Santa Virginia areas in the Serra do Mar State Park (PESM) have been conducted to determine the species diversity and changes in the structure of the bacterial communities in the phyllosphere, dermosphere and rhizosphere. However, little is known on the ecological functions of these bacteria, and on the ecological interactions between microbial communities and the environment in which they develop. The aim of this study was to explore the interactions between the microbial communities of the phyllosphere, dermosphere and rhizosphere of two tree species of the Atlantic Forest along an altitudinal gradient. Co-occurrence analysis based on data obtained by pyrosequencing of the 16S rRNA gene V4 region of bacteria to determine patterns of bacterial associations in different taxonomic levels in each microenvironment. For this study, the hypothesis that even if the environmental conditions are different in each type of forest (altitudinal gradient), there may be specific groups of bacteria that co-occur in the phyllosphere, dermosphere or rhizosphere, functioning as keystone taxa in the bacterial communities. Based on the sequencing of 16S rRNA genes, bacterial communities associated with the E. edulis and G. opposita phyllosphere and dermosphere in different forests were more similar to each other than the rhizosphere. Actinobacteria, Firmicutes, Proteobacteria and Bacteroidetes were the more abundant taxa in all studied microenvironments. Differences in the bacterial community structures in the phyllosphere, dermosphere and rhizosphere were observed along the altitudinal gradient, regardless of the plant species. In the lowland forest, the bacterial community associated with the phyllosphere was more similar between E. edulis and G. opposita. The rhizosphere bacterial community was more similar within each forest type than between forests, suggesting an effect of the forest physiognomy on the bacterial communities of the rhizosphere. Exploring the co-occurrence networks in the bacterial communities of each microenvironment it was observed that at the OTU level each microenvironment has different keystoine taxa that may regulate the ecological interactions in the community. Although the keystone taxa do not represent the most abundant OTUs in each microenvironment, they belong predominantly to Alphaproteobacteria and Gammaproteobacteria classes, suggesting that in the phyllosphere, dermosphere and rhizosphere the core microbiome cannot be determined at the OTU level, but possibly at higher taxonomic levels representing microbial groups having redundant functions.

Madurese, a Western Malayo-Polynesian language spoken on the Indonesian island of Madura, exhibits a three-way laryngeal contrast distinguishing between voiced, voiceless unaspirated and voiceless aspirated stops and an unusual consonant-vowel (CV) co-occurrence restriction. The CV co-occurrence restriction is of phonological interest given the patterning of voiceless aspirated stops with voiced stops rather than with voiceless unaspirated stops, raising the question of what phonological feature they may share. Two features have been linked with the CV co-occurrence restriction: Advanced Tongue Root [ATR] and Lowered Larynx [LL]. However, as no evidence of voicing during closure for aspirated stops is observed and no other acoustic measures except voice onset time (VOT), fundamental frequency (F0), frequencies of the first (F1) and the second (F2) formants and closure duration relating to the proposed features have been conducted, it remains an open question which acoustic properties are shared by voiced and aspirated stops. Three main questions are addressed in the thesis. The first question is what acoustic properties voiced and voiceless aspirated stops share to the exclusion of voiceless unaspirated stops. The second question is whether [ATR] or [LL] accounts for the patterning together of voiceless aspirated stops with voiced stops. The third question is what the implications of the results are for a transparent phonetics-phonology mapping that expects phonological features to have phonetic correlates associated with them. In order to answer the questions, we looked into VOT, closure duration, F0, F1, F2 and a number of spectral measures, i.e. H1*-A1*, H1*-A2*, H1*-A3*, H1*-H2*, H2*-H4* and CPP. We recorded fifteen speakers of Madurese (8 females, 7 males) reading 188 disyllabic Madurese words embedded in a sentence frame. The results show that the three-way voicing categories in Madurese have different VOT values. The difference in VOT is robust between voiced stops on the one hand and voiceless unaspirated and voiceless aspirated stops on the other. Albeit statistically significant, the difference in VOT values between voiceless unaspirated and voiceless aspirated stops is relatively small. With regard to closure duration, we found that there is a difference between voiced stops on the one hand and voiceless unaspirated and aspirated stops on the other. We also found that female speakers distinguish F0 for the three categories while male speakers distinguish between F0 for voiced stops on the one hand and voiceless unaspirated and voiceless aspirated stops on the other. The results for spectral measures show that there are no significant differences in H1*-A1*, H1*-A3*, H1*-H2*, H2*-H4* and CPP between vowels adjacent to voiced and voiceless aspirated stops. In contrast, there are significant differences in these measures between vowels adjacent to voiced and voiceless unaspirated stops and between vowels adjacent to voiceless aspirated and voiceless unaspirated stops. Regarding the question whether voiced and voiceless aspirated stops share certain acoustic properties, our findings show that they do. The acoustic properties they share are H1*-A1* for both genders, H1*-H2* for females, H1*-A3* and H2*-H4* for males, and CPP for females at vowel onset and for males at vowel midpoint. However, they do not share such acoustic properties as VOT, closure duration and F0. Voiceless unaspirated and voiceless aspirated stops can be distinguished by VOT, F0 and spectral measures, i.e. H1*-A1*, H1*-A3*, H1*-H2*, H2*-H4* and CPP. However, these two voiceless stop categories have similar closure durations. As regards the question if [+ATR] or [+LL] might be responsible for the patterning together of voiceless aspirated stops with voiced stops, our findings suggest that either feature appears to be plausible. Acoustic evidence that lends support to the feature [+ATR] includes lower F1 and greater spectral tilt measures, i.e. H1*-A1*, H1*-A3*, H1*-H2* and H2*-H4*, and lower CPP values. Acoustic evidence that supports the feature [+LL] includes lower F1 and greater spectral tilt measures, i.e. H1*-A1*, H1*-A3*, H1*-H2* and H2*-H4*, and lower CPP values. However, the fact that voiceless aspirated stops are voiceless during closure raises a problem for the feature [+ATR] and the fact that F0 for voiceless aspirated stops is higher than for voiced stops also presents a problem for the feature [+LL]. The fact that not all acoustic measures fit in well with either feature is problematic to the idea that the relationship between phonetics and phonology is transparent in the sense that phonological features can be directly transformed into their phonetic correlates. Following the view that not all phonological features may not be expected to be phonetically grounded, for example, when they are related to historical sound change, we hold the idea of a phonetics-phonology mapping which allows for other non-phonetic factors to account for a phonological phenomenon. We also provide historical and loanword evidence which could support that voiceless aspirated stops in Madurese may have derived from earlier voiced stops, which probably retain their historical laryngeal contrast through phonologisation.

A Mata Atlântica é uma floresta tropical úmida considerada um \"hotspot\" de biodiversidade e endemismo. É uma das florestas mais antigas do mundo e uma das maiores florestas da América, abrangendo aproximadamente 150 milhões de hectares em condições ambientais altamente heterogêneas. Estudos em diferentes ambientes da Mata Atlântica, nos núcleos de Picinguaba e Santa Virginia no Parque Estadual da Serra do Mar (PESM), têm sido realizados para determinar a diversidade de espécies e alterações da estrutura das comunidades de bactérias, tanto na filosfera, quanto na dermosfera e solo rizosférico. No entanto, pouco se sabe sobre as funções ecológicas dessas bactérias, e sobre as interações ecológicas entre as comunidades microbianas e os ambientes onde se desenvolvem. Assim o objetivo desse trabalho foi explorar as interações entre as comunidades microbianas da filosfera, dermosfera e solo coletado sobre a projeção da copa de duas espécies arbóreas da Mata Atlântica ao longo de um gradiente altitudinal, usando análises de co-ocorrência, a partir dos dados obtidos por pirosequenciamento da região V4 do gene rRNA 16S de bactérias, para determinar padrões de associações de bactérias em diferentes níveis taxonômicos em cada microambiente. Para esse estudo, foi proposta a hipótese de que mesmo que as condições ambientais sejam diferentes em cada tipo de floresta (gradiente altitudinal), pode existir grupos de bactérias específicos que co-ocorrem na filosfera, dermosfera ou solo das plantas, funcionando como taxons chaves na estruturação das comunidades bacterianas. Com base do sequenciamento dos genes rRNA 16S, as comunidades bacterianas associadas à filosfera e dermosfera de E. edulis e G. opposita nas diferentes florestas foram mais similares entre si do que as do solo. Actinobacteria, Firmicutes, Bacteroidetes e Proteobacteria foram mais abundantes em todos os microambientes estudados. Diferenças nas estruturas das comunidades bacterianas na filosfera, dermosfera e solo foram observadas ao longo do gradiente altitudinal, independente da espécie de planta. Na floresta de terras baixas, a comunidade bacteriana associada à filosfera foi mais similar entre E. edulis e G. opposita. No solo, a comunidade bacteriana foi mais similar dentro de cada tipo de floresta do que entre florestas, sugerindo um efeito da fisionomia da floresta nas comunidades de bactérias dos solos. Explorando as redes de co-ocorrência das comunidades bacterianas em cada microambiente observou-se que no nível de UTOs, cada microambiente têm diferentes táxons chaves que podem regular as interações ecológicas da comunidade. Embora táxons chaves não representam as UTOs mais abundantes em cada microambiente, eles pertencem, predominantemente às classes Alphaproteobacteria e Gammaproteobacteria, sugerindo que na filosfera, dermosfera e solo o core microbioma não pode ser definido ao nível de UTO, mas possivelmente a níveis taxonômicos mais elevados representando grandes grupos microbianos que apresentam funções redundantes. / The Atlantic Forest is a rainforest considered a hotspot of biodiversity and endemism. It is one of the oldest forests in the world and one of the largest forests of America, covering approximately 150 million hectares in highly heterogeneous environmental conditions. Studies in different environments of the Atlantic forest, in the Picinguaba and Santa Virginia areas in the Serra do Mar State Park (PESM) have been conducted to determine the species diversity and changes in the structure of the bacterial communities in the phyllosphere, dermosphere and rhizosphere. However, little is known on the ecological functions of these bacteria, and on the ecological interactions between microbial communities and the environment in which they develop. The aim of this study was to explore the interactions between the microbial communities of the phyllosphere, dermosphere and rhizosphere of two tree species of the Atlantic Forest along an altitudinal gradient. Co-occurrence analysis based on data obtained by pyrosequencing of the 16S rRNA gene V4 region of bacteria to determine patterns of bacterial associations in different taxonomic levels in each microenvironment. For this study, the hypothesis that even if the environmental conditions are different in each type of forest (altitudinal gradient), there may be specific groups of bacteria that co-occur in the phyllosphere, dermosphere or rhizosphere, functioning as keystone taxa in the bacterial communities. Based on the sequencing of 16S rRNA genes, bacterial communities associated with the E. edulis and G. opposita phyllosphere and dermosphere in different forests were more similar to each other than the rhizosphere. Actinobacteria, Firmicutes, Proteobacteria and Bacteroidetes were the more abundant taxa in all studied microenvironments. Differences in the bacterial community structures in the phyllosphere, dermosphere and rhizosphere were observed along the altitudinal gradient, regardless of the plant species. In the lowland forest, the bacterial community associated with the phyllosphere was more similar between E. edulis and G. opposita. The rhizosphere bacterial community was more similar within each forest type than between forests, suggesting an effect of the forest physiognomy on the bacterial communities of the rhizosphere. Exploring the co-occurrence networks in the bacterial communities of each microenvironment it was observed that at the OTU level each microenvironment has different keystoine taxa that may regulate the ecological interactions in the community. Although the keystone taxa do not represent the most abundant OTUs in each microenvironment, they belong predominantly to Alphaproteobacteria and Gammaproteobacteria classes, suggesting that in the phyllosphere, dermosphere and rhizosphere the core microbiome cannot be determined at the OTU level, but possibly at higher taxonomic levels representing microbial groups having redundant functions.

O consumo de qualquer parte do organismo de um fungo denomina-se fungivoria e são raros os estudos que abordam a temática da interação trófica entre suas estruturas reprodutivas (basidiomas) e os insetos. Mesmo que seja um hábito alimentar freqüente entre os coleópteros, com muitas famílias obrigatoriamente fungívoras, pouco se conhece sobre a estruturação dessa assembléia. Entretanto, sabese que existem preferências por determinadas espécies de fungos, tanto para a postura dos ovos quanto para o próprio consumo. Algumas questões que este estudo aborda referem-se à riqueza e à especificidade dos besouros fungívoros. Adicionalmente investigarei a influência da taxonomia dos fungos, bem como da consistência e do sistema hifal dos basidiomas para o processo de seleção que as espécies de besouros realizam. Também foi verificado se a posição taxonômica dos besouros está relacionada com o grupo de fungos que estes utilizam ou com a consistência dos seus basidiomas. Foram realizadas oito coletas em um ambiente de Floresta Ombrófila Mista no Estado do Rio Grande do Sul, obtendo-se 376 indivíduos das Ordens Polyporales e Hymenochaetales. Estes espécimes foram mantidos individualizados em laboratório por três meses para a constante retirada dos besouros adultos e então dissecados. Todos os fungos que apresentaram besouros fungívoros associados (N = 207) foram identificados e totalizaram 40 espécies. Os coleópteros totalizaram 447 ocorrências e foram classificados em 90 morfoespécies e 20 famílias, sendo Ciidae e Staphylinidae as mais representadas. Os basidiomas foram classificados quanto às suas classes de consistência seguindo um gradiente de durabilidade e rigidez. A riqueza de besouros variou conforme a espécie do hospedeiro e não foram encontrados fatores que influenciam a variação de riqueza de besouros em fungos. A especificidade dos besouros não apresentou relação com a longevidade dos basidiomas. Também se verificou que a posição taxonômica dos fungos, bem como a consistência dos basidiomas, interferem na escolha do hospedeiro pelos besouros. Além disso, se observou que besouros mais relacionados filogeneticamente não utilizam uma assembléia de fungos mais semelhante nem basidiomas com consistências semelhantes. Isso demonstra que a habilidade em consumir as mesmas espécies não está tão relacionada com a filogenia dos besouros como ocorre com os herbívoros. / Fungivory is the consumption of any part of the fungi organism and just a few studies were done about the trophic interactions of its reproductive structures, so called basidiomes, and insects. Even though this is a disseminated habit among Coleoptera, which has many obligated fungivores families, little is known about its community organization. However, there have been works about feeding preferences and oviposing sites of some fungi species. Some of the questions of this study refer to richness and specificity of fungivores beetles on their host. It was also investigated the influence of fungi taxonomy, basidiomes consistency and hyphal system for the selection process of the beetle species, as well as the importance of the taxonomic position of the beetles regarding the host group they use and the basidiome´s consistency. The basidiomes of Polyporales and Hymenochaetales Orders were collected in an Araucaria angustifolia forest of Rio Grande do Sul State. It was obtained 376 fungal individuals, which were kept in the laboratory, individually separated in containers for three months, while the newly emerged beetles were captured, and the fungi were dissected afterwards. All individuals of fungi that had associations (207) were identified (40 species). It was found 447 occurcences of Coleoptera, which were classified in 90 morphospecies belonging to 20 families, with Ciidae and Staphylinidae being the most abundant. The basidiomes where classified by its consistency, following a longevity and hardness gradient. Beetle richness varied among fungal species and no factors that might influence the richness variation in fungi was found. Its specificity did not seem to be related to the longevity of the basidiomes. It was also verified that the fungus taxonomy as well as the consistency of its basidiomes demonstrated some importance for the process of host selection. Regarding the Coleoptera taxonomy, it was not verified that the more phylogenetic related ones use a more similar group of hosts or basidiomes that have similar consistency. This demonstrates that the ability to feed on the same species is not so related to the phylogeny of fungivore beetles as it occurs with herbivores.

The predictability of an upcoming word has been found to be a useful predictor in eye movement research, but is expensive to collect and subjective in nature. It would be desirable to have other predictors that are easier to collect and objective in nature if these predictors were capable of capturing the information stored in predictability. This paper contributes to this discussion by testing a possible predictor: conditional co-occurrence probability. This measure is a simple statistical representation of the relatedness of the current word to its context, based only on word co-occurrence patterns in data taken from the Internet. In the regression analyses, conditional co-occurrence probability acts like lexical frequency in predicting fixation durations, and its addition does not greatly improve the model fits. We conclude that readers do not seem to use the information contained within conditional co-occurrence probability during reading for meaning, and that similar simple measures of semantic relatedness are unlikely to be able to replace predictability as a predictor for fixation durations. Keywords: Co-occurrence probability, Cloze predictability, frequency, eye movement, fixation duration.