I used multivariate, hierarchical analyses to examine the relative influence of watershed-, riparian- and channel-scale environmental characteristics on catch per unit effort (CPUE: fish/angler-hour) and species composition of sport fisheries in Mississippi wadeable streams. Partial canonical correspondence analyses indicated that riparian-scale variables (31.1%) explained more variation in sport fish relative abundances compared to watershed-scale (24.4%) and channel-scale variables (18.9%). Largemouth bass M. salmoides and longear sunfish Lepomis megalotis were more abundant in smaller-watershed streams with dense forest cover and greater woody debris, alkalinity and diverse substrates. Spotted bass M. punctulatus and bluegill L. macrochirus were more abundant in larger-watershed streams with moderate to dense forest cover yet more open riparian canopies. Regional-scale characteristics also influenced relative abundances of these fisheries. Total CPUE, total bass CPUE and largemouth bass CPUE were greatest in watersheds draining the Blackland Prairie-Flatwoods compared to other level III ecoregions. This ecoregion contains fertile soils that influence stream productivity, because alkalinity tends to be large in forested streams draining this ecoregion. I developed and validated watershed-scale models and found that percentage forest cover, stream density, total road density and primary highway density predicted mean total CPUE, mean total sunfish CPUE and mean total bass CPUE accurately (Sign tests comparing observed versus predicted mean CPUE, P > 0.05). The models were precise (R2 > 0.71), explaining 83%, 71% and 80%, respectively, of variation in mean total CPUE, mean total sunfish CPUE and mean total bass CPUE from independent data. Species-specific models performed poorly, suggesting biotic relationships may hinder development of meaningful habitat models for species. My study supports forest conservation to sustain sport fisheries in Mississippi’s wadeable streams. Forests mediate sediment and nutrient loading to stream channels, influence hydrology and channel morphology and provide woody habitat for sport fish and their forage base (benthic macroinvertebrates). My small sample size was small; (N = 13 reaches), thus caution is advised before engaging in comprehensive management of wadeable streams based on my results. Nevertheless, my watershed models can be applied at very low cost using a GIS or topographic maps to identify reaches state-wide that support wadeable stream sport fisheries.

MSc. Thesis, Ryan L. Rolick / Stream temperature, which influences many biogeochemical processes, is controlled by the exchange of water and energy across the stream surface, banks, and bed. A stream’s thermal sensitivity is its sensitivity to changes in air temperature and is increasingly important with projected warming in northern regions. As the thermal dynamics in permafrost underlain headwater streams are poorly documented, this study examines the thermal signals of two neighboring alpine streams (Granger Creek, GC and Buckbrush Creek, BB) within the Wolf Creek Research Basin (WCRB), Yukon Territory, and quantifies the dominant energy fluxes through an energy budget approach. A 1950 m study reach was established in each stream where water temperature, vertical streambed temperature, and stream discharge were measured. Distinct thermal heterogeneity is observed in each stream, with areas of persistent temperature decrease along their lengths, minimal downstream increase in stream temperature in GC, and an overall decrease in downstream temperature in BB. These observations are indicative of focused groundwater upwelling, low thermal sensitivity, and an increasing downstream influence of groundwater. This is inferred from increases in specific conductivity (2 μS/cm – 20 μS/cm) at locations of temperature decrease, and patterns in low (high) mean air-water linear regression slope (intercept) values. With distance downstream, regression slope (intercept) in GC went from 0.37 (1.2 ̊C) upstream to 0.33 (2.6 ̊C) at the outlet. Similarly, in BB regression slope (intercept) went from 0.37 (2.1 ̊C) upstream to 0.24 (3.3 ̊C) downstream. These patterns also indicate an overall lower thermal sensitivity and greater groundwater influence across the study reach of BB. Downwelling is observed at locations of vertical temperature profiles in each stream through efficient downward propagation of the diel temperature signal into the streambed, indicating hyporheic exchange. The energy balance indicates a large unaccounted for sink in BB, further denoting a larger groundwater influence in its study reach. These findings highlight the importance for continued study of thermal regimes and the complicated interconnections between heat exchange processes in alpine catchments in permafrost regions. / Thesis / Master of Science (MSc)

Yes / Data streams have arisen as a relevant research topic during the past decade. They are real‐time, incremental in nature, temporally ordered, massive, contain outliers, and the objects in a data stream may evolve over time (concept drift). Clustering is often one of the earliest and most important steps in the streaming data analysis workflow. A comprehensive literature is available about stream data clustering; however, less attention is devoted to the fuzzy clustering approach, even though the nonstationary nature of many data streams makes it especially appealing. This survey discusses relevant data stream clustering algorithms focusing mainly on fuzzy methods, including their treatment of outliers and concept drift and shift. / Ministero dell‘Istruzione, dell‘Universitá e della Ricerca.

I examined streams in the Joyce Kilmer-Slickrock wilderness to determine if streams responded to logging following 75 y of recovery. Joyce Kilmer was never logged and the Slickrock wilderness was logged from 1917 until 1922. Wood was common in unlogged streams and averaged 417 m³ of wood/ha of streambed. Logged streams had significantly less wood (1.1 m³ of wood/ha of stream), probably because of the construction of railroads in streambeds used to remove timber. Fine substrates (<5.6 mm diameter) were less abundant in logged streams and the retention of fines decreased as wood volume decreased. Species diversity was similar between streams in unlogged and logged catchments; however, the abundance of several taxa and functional feeding groups did differ. Streams in unlogged catchments had significantly greater proportions of shredders while streams in logged catchments had significantly greater proportions of scrapers. Ecosystem parameters showed that the linkages between streams and the forests they drain were weaker in logged than unlogged streams and that unlogged streams derived a greater proportion of fixed carbon from riparian vegetation. Stream-forest linkage strength increased as debris dam abundance increased, while the use of riparian vegetation inputs increased as moss increased. Contrary to predictions, solute storage was significantly greatest in logged catchments and negatively related to debris dam abundance that decreased streambed permeability. Additionally, phosphorus retention, instead of being enhanced by solute storage, was negatively related to transient storage. Uptake velocity was significantly greater in unlogged than logged streams and significantly related to debris dam abundance. Mean breakdown rate of experimental leaf packs and wood veneers was not significantly different between unlogged and logged streams. Leaf breakdown was strongly related to shredder colonization, while wood breakdown was unrelated to variables measured. Much of the persisting disturbance to streams by past logging was directly or indirectly related to differences in wood volume, debris dam frequency, and streambed substrate composition. Results support the hypothesis that logging results in a downstream shift from the headwaters in ecosystem function and that logging disturbance to streams likely persists for centuries. / Ph. D.

O volume de informação manipulada em sistemas apoiados por computador tem crescido tanto no número de objetos que compõem os conjuntos de dados quanto na quantidade e na complexidade dos atributos. Em conjuntos de dados do mundo real, a uniformidade na distribuição de valores e a independência entre atributos são propriedades bastante incomuns. De fato, dados reais são em geral caracterizados pela ampla presença de correlações entre seus atributos. Além disso, num mesmo conjunto podem existir correlações de naturezas diversas, como correlações lineares, não-lineares e não-polinomiais. Todo esse cenário pode degradar a performance dos algoritmos que manipulam e, principalmente, dos que realizam análises dos dados. Além da grande quantidade de objetos a serem tratados e do número elevado de atributos, as correlações nem sempre são conhecidas, o que pode comprometer a eficácia de tais algoritmos. Nesse contexto, as técnicas de redução de dimensionalidade permitem diminuir o número de atributos de um conjunto de dados, minimizando assim os problemas decorrentes da alta dimensionalidade. Algumas delas são baseadas na análise de correlações e, com o objetivo de reduzir a perda de informação relevante causada pela remoção de atributos, procuram eliminar apenas aqueles que sejam correlacionados aos restantes. No entanto, essas técnicas geralmente analisam como cada atributo está correlacionado a todos os demais, tratando o conjunto de atributos como um todo e usando ferramentas de análise estatística. Esta tese propõe uma abordagem diferente, baseada na Teoria dos Fractais, para detectar a existência de correlações e identificar subconjuntos de atributos correlacionados. Para cada correlação encontrada é possível ainda identificar quais são os atributos que melhor a descrevem. Conseqüentemente, um subconjunto de atributos relevantes para representar as características fundamentais dos dados é determinado, não apenas com base em correlações globais entre todos os atributos, mas também levando em consideração especificidades de correlações que envolvem subconjuntos reduzidos. A técnica apresentada é uma ferramenta a ser utilizada em etapas de pré-processamento de atividades de descoberta de conhecimento, principalmente em operações de seleção de atributos para redução de dimensionalidade. A proposta para a identificação de correlações e os conceitos que a fundamentam são validados por meio de estudos experimentais usando tanto dados sintéticos quanto reais. Finalmente, os conceitos básicos da Teoria dos Fractais são aplicados na análise de comportamento de data streams, também constituindo uma contribuição relevante desta tese de doutorado. / The volume of information processed by computer-based systems has grown not only in the amount of data but also in number and complexity of attributes. In real world datasets, uniform value distribution and independence between attributes are rather uncommon properties. In fact, real data is usually characterized by vast existence of correlated attributes. Moreover, a dataset can present different types of correlations, such as linear, non-linear and non-polynomial. This entire scenario may degrade performance of data management and, particularly, data analysis algorithms, as they need to deal with large amount of data and high number of attributes. Furthermore, correlations are usually unknown, which may jeopardize the efficacy of these algorithms. In this context, dimensionality reduction techniques can reduce the number of attributes in datasets, thus minimizing the problems caused by high dimensionality. Some of these techniques are based on correlation analysis and try to eliminate only attributes that are correlated to those remaining, aiming at diminishing the loss of relevant information imposed by attribute removal. However, techniques proposed so far usually analyze how each attribute is correlated to all the others, considering the attribute set as a whole and applying statistical analysis tools. This thesis presents a different approach, based on the Theory of Fractals, to detect the existence of correlations and to identify subsets of correlated attributes. In addition, the proposed technique makes it possible to identify which attributes can better describe each correlation. Consequently, a subset of attributes relevant to represent the fundamental characteristics of the dataset is determined, not only based on global correlations but also considering particularities of correlations concerning smaller attribute subsets. The proposed technique works as a tool to be used in preprocessing steps of knowledge discovery activities, mainly in feature selection operations for dimensionality reduction. The technique of correlation detection and its main concepts are validated through experimental studies with synthetic and real data. Finally, as an additional relevant contribution of this thesis, the basic concepts of the Theory of Fractals are also applied to analyze data streams behavior.

Cette thèse étudie les conditions et l’ampleur des évolutions de la politique nucléaire de l’Union européenne, de 2000 à 2016. Depuis le milieu des années 1990, plusieurs événements – ou « fenêtres d’opportunité » selon la terminologie du Multiple Streams Framework – ont remis les politiques nucléaires au centre de l’attention des décideurs européens : l’adhésion de nouveaux Etats membres en 2004 ; différentes Présidences du Conseil et nominations de Commissaires ayant pris clairement position sur les questions nucléaires ; l’accident de Fukushima de mars 2011 ; et enfin, les deux sorties du nucléaire allemandes, en 2000 et 2011.La première partie de ce travail est consacrée à l’analyse de l’ampleur du changement introduit par les directives adoptées durant la période étudiée: les directives sûreté réacteur de 2009 et 2014, ainsi que la directive déchets radioactifs de 2011. La deuxième partie étudie les facteurs de stabilité qui ont empêché la mise à l’agenda ou l’adoption de certaines propositions, comme les politiques d’harmonisation des régimes de responsabilité civile ou comme les tentatives d’abolition du traité Euratom.Parmi les différentes variables testées, des coûts de mise en œuvre limités, ainsi que l’existence d’indicateurs pour un problème donné, sont apparus comme des conditions nécessaires (mais non suffisantes) pour un changement de politique.La variable la plus pertinente pour expliquer les changements contenus dans les directives adoptées est l’évolution de la place d’un problème dans les priorités des acteurs. L’accident de Fukushima n’a modifié ni les conditions objectives d’existence d’un problème, ni, sur le long terme, les priorités des Etats Membres ou du secteur nucléaire. Ainsi, les propositions post Fukushima tendant à augmenter le rôle de la Commission européenne ou une calibration plus stricte des mesures de sûreté ont donc été partiellement rejetées, pour finalement aboutir à une révision de la directive sûreté, en 2014, avec peu de changements. De même, la priorité accordée à la résolution des problèmes nationaux liés à gestion des déchets radioactifs a abouti à l’adoption d’une directive incluant des mesures plus précises.Les principaux changements ayant eu lieu dans les directives étudiées concernent leur logique instrumentale et leur calibration ; aucun changement n’a été observé dans les buts ou les orientations générales de la politique nucléaire de l’UE. La politique nucléaire de l’UE semble ainsi particulièrement stable.Cette thèse met en évidence plusieurs facteurs d’inertie des politiques et institutions d’Euratom, malgré l’action d’entrepreneurs politiques : le haut niveau de méfiance entre acteurs, la délégation de la sûreté à des agences indépendantes, la dépendance au sentier liée à la technologie nucléaire, et la flexibilité « incorporée » dans Euratom. Ces facteurs réduisent le nombre d’entrepreneurs politiques potentiellement efficaces ainsi que l’éventail de leurs stratégies pour promouvoir un changement. En plus des facteurs institutionnels, la stabilité des politiques et institutions est également promue par des veto players.Néanmoins, les problèmes récents auxquels est confronté le secteur nucléaire européen (intégration croissante de sources d’énergie renouvelable, difficultés économiques) érodent les conditions de reproduction et de stabilité des institutions d’Euratom. Cet aspect se traduit notamment dans le progressif retrait du nucléaire des communications officielles mentionnant les sources d’énergie à développer pour remplir les objectifs de l’Union.En se basant sur le MSF et sur l’institutionnalisme historique, cette thèse démontre ainsi que les évolutions de la politique nucléaire de l’UE requièrent trois conditions : l’action d’un entrepreneur politique, l’émergence d’un problème dans les priorités des principaux acteurs et, pour des changements d’orientation ou d’institution, l’érosion des mécanismes de reproduction. / The goal of this dissertation is to understand the conditions under which the nuclear policy of the European Union has evolved – between 2000 and 2016 — as well as the scope of these evolutions. From the mid-1990s onwards, various events – “windows of opportunity” in the terminology of the Multiple Streams Framework – brought nuclear energy back to the attention of European policy-makers: the accession of new member states in 2004; different Council presidencies and nomination of Commissioners with a strong position on nuclear energy; the Fukushima accident in March 2011; Germany’s first and second nuclear phase-outs (in 2000 and 2011).This PhD proceeds in two parts: First, it analyses the scope of change in policies that have been set on the agenda and finally adopted (the reactor safety and the radioactive waste management directives); second, it analyses the stability factors that have hindered the agenda-setting or the decision-making on proposals for policy change or institutional reforms (harmonisations of civil liability regimes, licensing processes, or decommissioning funds; abolition of the Euratom treaty).Among the several variables tested, low implementation costs, as well as the existence of problem indicators, may be a necessary, but not a sufficient condition for policy change.The most relevant variable accounting for the policy changes included in the adopted directives, was the evolution of the problem in place of actors’ priorities. The Fukushima accident neither modified the objective conditions of existence of a problem, nor the long-term priority of Member States or the nuclear sector. Hence the post-Fukushima proposals suggesting an increase in the role of the European Commission or a stricter calibration of safety measures have been partially rejected. Hence there was no substantial policy change in the safety directive adopted in 2014. In a similar way, the priority granted to overcoming national problems of radioactive waste management led to the 2011 radioactive waste directive, which contained more settings than the 2009 reactor safety directive .The main changes observed in the directives impacted their settings, instrumental logic and calibration; no change has been observed in the goals or orientations of the EU nuclear policy since 2000. Therefore, the EU nuclear policy seems especially stable: in many cases, policy proposals did not even make in onto the agenda of the Council.In order to account for the stability of the policy in spite of the action of policy entrepreneurs, this PhD underlines the particular “stickiness” of the Euratom institutions and policies: The high distrust among actors; the delegation of safety to independent agencies; the technological path-dependence linked to the use of nuclear energy; and the “built-in” institutional flexibility of Euratom. These factors reduce the number of potential successful policy entrepreneurs and the range of strategies of change. Not only the institutional context but also the defiance among actors has accounted for the particular stability of the European nuclear policy and for the absence of change in goals or objectives: Stability was promoted by veto players, whose actions can be taken into account by the Multiple Streams Framework.Still, the recent problems facing the nuclear sector (growing renewable energy share, economic difficulties) are eroding the conditions of stability and mechanisms of reproduction of the European nuclear institutions. This is reflected in the progressive modification of the Commission’s official positions on nuclear energy, which is no longer included in the energy sources to be developed to meet the EU’s energy objectives.Relying both on the MSF and the historical institutionalism, this PhD shows that change in the European nuclear policy requires the action of a policy entrepreneur, the erosion of reproduction mechanisms and the occurrence of a problem among the priorities of main actors.

Thesis (M.S. - Hydrology and Water Resources)--University of Arizona, 1983. / Includes bibliographical references (leaves 117-120).

O volume de informação manipulada em sistemas apoiados por computador tem crescido tanto no número de objetos que compõem os conjuntos de dados quanto na quantidade e na complexidade dos atributos. Em conjuntos de dados do mundo real, a uniformidade na distribuição de valores e a independência entre atributos são propriedades bastante incomuns. De fato, dados reais são em geral caracterizados pela ampla presença de correlações entre seus atributos. Além disso, num mesmo conjunto podem existir correlações de naturezas diversas, como correlações lineares, não-lineares e não-polinomiais. Todo esse cenário pode degradar a performance dos algoritmos que manipulam e, principalmente, dos que realizam análises dos dados. Além da grande quantidade de objetos a serem tratados e do número elevado de atributos, as correlações nem sempre são conhecidas, o que pode comprometer a eficácia de tais algoritmos. Nesse contexto, as técnicas de redução de dimensionalidade permitem diminuir o número de atributos de um conjunto de dados, minimizando assim os problemas decorrentes da alta dimensionalidade. Algumas delas são baseadas na análise de correlações e, com o objetivo de reduzir a perda de informação relevante causada pela remoção de atributos, procuram eliminar apenas aqueles que sejam correlacionados aos restantes. No entanto, essas técnicas geralmente analisam como cada atributo está correlacionado a todos os demais, tratando o conjunto de atributos como um todo e usando ferramentas de análise estatística. Esta tese propõe uma abordagem diferente, baseada na Teoria dos Fractais, para detectar a existência de correlações e identificar subconjuntos de atributos correlacionados. Para cada correlação encontrada é possível ainda identificar quais são os atributos que melhor a descrevem. Conseqüentemente, um subconjunto de atributos relevantes para representar as características fundamentais dos dados é determinado, não apenas com base em correlações globais entre todos os atributos, mas também levando em consideração especificidades de correlações que envolvem subconjuntos reduzidos. A técnica apresentada é uma ferramenta a ser utilizada em etapas de pré-processamento de atividades de descoberta de conhecimento, principalmente em operações de seleção de atributos para redução de dimensionalidade. A proposta para a identificação de correlações e os conceitos que a fundamentam são validados por meio de estudos experimentais usando tanto dados sintéticos quanto reais. Finalmente, os conceitos básicos da Teoria dos Fractais são aplicados na análise de comportamento de data streams, também constituindo uma contribuição relevante desta tese de doutorado. / The volume of information processed by computer-based systems has grown not only in the amount of data but also in number and complexity of attributes. In real world datasets, uniform value distribution and independence between attributes are rather uncommon properties. In fact, real data is usually characterized by vast existence of correlated attributes. Moreover, a dataset can present different types of correlations, such as linear, non-linear and non-polynomial. This entire scenario may degrade performance of data management and, particularly, data analysis algorithms, as they need to deal with large amount of data and high number of attributes. Furthermore, correlations are usually unknown, which may jeopardize the efficacy of these algorithms. In this context, dimensionality reduction techniques can reduce the number of attributes in datasets, thus minimizing the problems caused by high dimensionality. Some of these techniques are based on correlation analysis and try to eliminate only attributes that are correlated to those remaining, aiming at diminishing the loss of relevant information imposed by attribute removal. However, techniques proposed so far usually analyze how each attribute is correlated to all the others, considering the attribute set as a whole and applying statistical analysis tools. This thesis presents a different approach, based on the Theory of Fractals, to detect the existence of correlations and to identify subsets of correlated attributes. In addition, the proposed technique makes it possible to identify which attributes can better describe each correlation. Consequently, a subset of attributes relevant to represent the fundamental characteristics of the dataset is determined, not only based on global correlations but also considering particularities of correlations concerning smaller attribute subsets. The proposed technique works as a tool to be used in preprocessing steps of knowledge discovery activities, mainly in feature selection operations for dimensionality reduction. The technique of correlation detection and its main concepts are validated through experimental studies with synthetic and real data. Finally, as an additional relevant contribution of this thesis, the basic concepts of the Theory of Fractals are also applied to analyze data streams behavior.

abstract: Nitrogen is an essential, often limiting, element for biological growth that can act as a pollutant if present in excess. Nitrogen is primarily transported by water from uplands to streams and eventually to recipient lakes, estuaries, and wetlands, but can be modulated by biological uptake and transformation along these flowpaths. As a result, nitrogen can accumulate in aquatic ecosystems if supply is high or if biological retention is low. Dryland and urban ecosystems offer interesting contrasts in water supply, which limits transport and biological activity in drylands, and nitrogen supply that increases with human activity. In my dissertation, I ask: What is the relative balance among nitrogen retention, removal, and transport processes in dryland watersheds, and what is the fate of exported nitrogen? My dissertation research demonstrates that water is a major control on where and when nitrogen is retained and removed versus exported to downstream ecosystems. I used a mass-balance model based on synoptic surveys to study seasonal and spatial patterns in nitrate loading to a dryland stream network. I found that irrigation diversions transport nitrate from agricultural areas to the stream network year-round, even during dry seasons, and are an important driver of nitrate loading. I further explored how seasonal precipitation influences flood nutrient export in an intermittent desert stream by coupling long-term data of flood-water chemistry with stream discharge and precipitation data. I found that higher precipitation prior to a flood fills water storage sites in the catchment, leading to larger floods. In addition, higher antecedent precipitation stimulates biological nitrogen retention in the uplands, leading to lower nitrogen concentration in floods. Finally, I evaluated the consequences of nitrogen export from watersheds on how urban wetlands attenuate nitrate through denitrification that permanently removes nitrogen, and dissimilatory nitrate reduction to ammonium (DNRA) that retains nitrogen in another biologically reactive form. I found that DNRA becomes proportionally more important with low nitrate concentration, thereby retaining nitrogen as ammonium. Collectively, my dissertation research addresses how dryland and urban ecosystems can be integrated into models of watershed nitrogen cycling. / Dissertation/Thesis / Doctoral Dissertation Biology 2019

Stream food webs are supported by carbon produced within the stream (autochthonous) and from terrestrial environments (allochthonous). Allochthonous carbon (C) inputs are assumed to be the dominant C source supporting food webs within small streams, but few direct estimates of resource use in small streams have been made, especially in boreal streams. The objective of this study was to determine the relative dependence on allochthonous and autochthonous C by consumers in relation to C pools within streams with high terrestrial inputs. Furthermore, this study aimed to investigate if the relative resource use of allochthonous and autochthonous C by consumers differed among seasons (summer and fall), between streams of different sizes, and locations within the catchment. To estimate consumer resource use, δ2H signatures for organic C sources were compared to those of six key consumers in five streams of varying catchment sizes in northern Sweden. Macroinvertebrate biomass was quantified to calculate a taxa-specific biomass-weighted allochthony, and compared with the mass of different C pools potentially available for consumers. The biomass-weighted mean allochthony for all samplings ranged between 43.5-61.5%; there was thus high autochthonous support despite low algal density and high terrestrial C pools within the streams. No significant trend in allochthony was observed over season (linear regression, p-value &gt;0.05). Allochthony differed by invertebrate taxa and was not related to stream size or location in catchment. These results suggest that autochthonous C is far more important for consumers in boreal streams than previously recognized.