A variação do tempo de residência e o fluxo contínuo de água da bacia hidrográfica em direção à barragem fazem dos reservatórios um sistema de transição entre rios e lagos com características físicas, químicas e biológicas particulares que influenciam a sua hidrodinâmica e o seu metabolismo. O estudo do metabolismo em ecossistemas aquáticos subtropicais, particularmente em reservatórios, e sua relação com a bacia hidrográfica tem sido pouco explorado nos trabalhos que utilizam a modelagem matemática como ferramenta de análise. O presente estudo teve como objetivo quantificar e analisar a variabilidade temporal do metabolismo em um reservatório subtropical de altitude localizado na cidade de Caxias do Sul/RS e sua relação com as contribuições da bacia hidrográfica em termos de vazão e nutrientes. Para tanto, foi aplicado o modelo ecológico baseado em processos IPH-ECO, em conjunto com o modelo hidrológico IPH-II. Após o ajuste dos níveis e a calibração do módulo de qualidade de água do IPH-ECO, foi determinada a variabilidade temporal das taxas metabólicas de produção primária (GPP) e respiração (R) no período de nov/2011 a dez/2012, que, comparado a outros períodos, foi marcado por uma estiagem anormal e um consequente aumento do tempo de residência do reservatório. Foi analisado o comportamento temporal das seguintes variáveis: nível d’água, temperatura, oxigênio dissolvido, clorofila-a, nutrientes, GPP e R, juntamente com as condições de contorno do modelo como as características hidrológicas e de uso e ocupação do solo da bacia hidrográfica. O reservatório apresentou um metabolismo autotrófico 97% do período de simulação, tornando-se heterotrófico em alguns dias da primavera, quando ocorreu uma diminuição da concentração de clorofila-a e fósforo total e uma intensificação do processo de nitrificação. As variações temporais das taxas GPP e R acompanharam o crescimento do fitoplâncton, que esteve mais associado ao padrão de estratificação e ao tempo de residência do reservatório. Nos meses com chuva, o aumento do escoamento superficial da bacia hidrográfica provocou um maior aporte de fósforo no reservatório contribuindo para o aumento do fitoplâncton e, consequentemente, das taxas de GPP e R. A biomassa fitoplanctônica foi o principal fator responsável pelo estabelecimento do metabolismo autotrófico no reservatório o qual respondeu mais em relação à dinâmica interna do sistema que à bacia hidrográfica. / Variation of residence time and continuous water inputs make reservoirs a transition aquatic system between lakes and rivers with physical, chemical and biological features that affect its hydrodynamics and metabolism. The study of aquatic metabolism in subtropical ecosystems, particularly in reservoirs, and its relation with the watershed has been poorly explored in researches that use mathematical models as a tool for analyzing this relation. This study aimed to quantify and analyse the temporal variability of metabolism in an altitude subtropical reservoir located in the city of Caxias do Sul/RS together with biotic and abiotic variables components of the reservoir, considering its relation with the physical features of the watershed. For this purpose, the process-based ecological model IPH-ECO was applied in the reservoir in combination with the hydrological model IPH-II. After the fit of simulated water levels to observation data and the calibration of IPH-ECO water quality module, the temporal variability of primary production (GPP) and respiration (R) was obtained in the period from nov/2011 to dez/2012, when, compared to other time periods, an unusual dry period occurred followed by an increase of reservoir residence time. The temporal pattern of the following variables was analysed: water levels, temperature, dissolved oxygen, chlorophyll-A, nutrients, GPP and R, along with the model boundary conditions, such as watershed land use and hydrology. The reservoir was net autotrophic during 97% of the simulation time period, becoming net heterotrophic for a few days in the spring due to the decreasing of chlorophyll-a and total phosphorus concentration and the strengthening of nitrification process in the reservoir. Temporal variations of GPP and R followed the phytoplankton growth which in turn had been driven by the stratification pattern and the residence time of reservoir. A watershed “fast-flush” was observed during storm events with higher rates of leaching of phosphorus, leading to an increase of chlorophyll-a concentration and thus GPP and R in the reservoir. It could be inferred that, during the simulation time period, the phytoplankton biomass in the reservoir was the main factor in maintaining the autotrophic metabolism which in most of the time was more affected by the internal dynamic of the system than by the physical aspects of watershed.

A variação do tempo de residência e o fluxo contínuo de água da bacia hidrográfica em direção à barragem fazem dos reservatórios um sistema de transição entre rios e lagos com características físicas, químicas e biológicas particulares que influenciam a sua hidrodinâmica e o seu metabolismo. O estudo do metabolismo em ecossistemas aquáticos subtropicais, particularmente em reservatórios, e sua relação com a bacia hidrográfica tem sido pouco explorado nos trabalhos que utilizam a modelagem matemática como ferramenta de análise. O presente estudo teve como objetivo quantificar e analisar a variabilidade temporal do metabolismo em um reservatório subtropical de altitude localizado na cidade de Caxias do Sul/RS e sua relação com as contribuições da bacia hidrográfica em termos de vazão e nutrientes. Para tanto, foi aplicado o modelo ecológico baseado em processos IPH-ECO, em conjunto com o modelo hidrológico IPH-II. Após o ajuste dos níveis e a calibração do módulo de qualidade de água do IPH-ECO, foi determinada a variabilidade temporal das taxas metabólicas de produção primária (GPP) e respiração (R) no período de nov/2011 a dez/2012, que, comparado a outros períodos, foi marcado por uma estiagem anormal e um consequente aumento do tempo de residência do reservatório. Foi analisado o comportamento temporal das seguintes variáveis: nível d’água, temperatura, oxigênio dissolvido, clorofila-a, nutrientes, GPP e R, juntamente com as condições de contorno do modelo como as características hidrológicas e de uso e ocupação do solo da bacia hidrográfica. O reservatório apresentou um metabolismo autotrófico 97% do período de simulação, tornando-se heterotrófico em alguns dias da primavera, quando ocorreu uma diminuição da concentração de clorofila-a e fósforo total e uma intensificação do processo de nitrificação. As variações temporais das taxas GPP e R acompanharam o crescimento do fitoplâncton, que esteve mais associado ao padrão de estratificação e ao tempo de residência do reservatório. Nos meses com chuva, o aumento do escoamento superficial da bacia hidrográfica provocou um maior aporte de fósforo no reservatório contribuindo para o aumento do fitoplâncton e, consequentemente, das taxas de GPP e R. A biomassa fitoplanctônica foi o principal fator responsável pelo estabelecimento do metabolismo autotrófico no reservatório o qual respondeu mais em relação à dinâmica interna do sistema que à bacia hidrográfica. / Variation of residence time and continuous water inputs make reservoirs a transition aquatic system between lakes and rivers with physical, chemical and biological features that affect its hydrodynamics and metabolism. The study of aquatic metabolism in subtropical ecosystems, particularly in reservoirs, and its relation with the watershed has been poorly explored in researches that use mathematical models as a tool for analyzing this relation. This study aimed to quantify and analyse the temporal variability of metabolism in an altitude subtropical reservoir located in the city of Caxias do Sul/RS together with biotic and abiotic variables components of the reservoir, considering its relation with the physical features of the watershed. For this purpose, the process-based ecological model IPH-ECO was applied in the reservoir in combination with the hydrological model IPH-II. After the fit of simulated water levels to observation data and the calibration of IPH-ECO water quality module, the temporal variability of primary production (GPP) and respiration (R) was obtained in the period from nov/2011 to dez/2012, when, compared to other time periods, an unusual dry period occurred followed by an increase of reservoir residence time. The temporal pattern of the following variables was analysed: water levels, temperature, dissolved oxygen, chlorophyll-A, nutrients, GPP and R, along with the model boundary conditions, such as watershed land use and hydrology. The reservoir was net autotrophic during 97% of the simulation time period, becoming net heterotrophic for a few days in the spring due to the decreasing of chlorophyll-a and total phosphorus concentration and the strengthening of nitrification process in the reservoir. Temporal variations of GPP and R followed the phytoplankton growth which in turn had been driven by the stratification pattern and the residence time of reservoir. A watershed “fast-flush” was observed during storm events with higher rates of leaching of phosphorus, leading to an increase of chlorophyll-a concentration and thus GPP and R in the reservoir. It could be inferred that, during the simulation time period, the phytoplankton biomass in the reservoir was the main factor in maintaining the autotrophic metabolism which in most of the time was more affected by the internal dynamic of the system than by the physical aspects of watershed.

FundaÃÃo Cearense de Apoio ao Desenvolvimento Cientifico e TecnolÃgico / Estudar padrÃes de distribuiÃÃo dos organismos à relevante na ecologia, biogeografia e biologia da conservaÃÃo fornecendo subsÃdios para compreender a evoluÃÃo da biodiversidade. Dimorphandra Schott à um gÃnero de distribuiÃÃo neotropical com 25 espÃcies no Brasil, distribuÃdas tanto em formaÃÃes savÃnicas (Cerrado) como ombrÃfilas (Floresta AmazÃnica e AtlÃntica). Este trabalho objetivou entender como està distribuÃda e quais os fatores ambientais estÃo associados à riqueza e a diversidade de Dimorphandra. Foram utilizadas ferramentas de Sistemas de InformaÃÃo GeogrÃfica, modelagem ecolÃgica e dados de ocorrÃncia obtidos da rede SpeciesLink, informaÃÃes bibliogrÃficas e visitas à herbÃrios. Esses dados foram classificados dentro de provÃncias biogeogrÃficas de acordo com um esquema de regionalizaÃÃo biogeogrÃfica e com os grupos gerados por uma anÃlise de agrupamento â UPGMA realizada no programa PC-ORD e uma anÃlise parcimoniosa de endemicidade (PAE) feita por meio do programa PAUP. Paralelamente, modelos ecolÃgicos de distribuiÃÃo baseados em algoritmos (BIOCLIM e CCSM) foram formulados para opassado, presente e futuro em ambiente do MAXENT e DIVA-GIS. Ambos os centros de riqueza e diversidade foram a regiÃo amazÃnica. As espÃcies com mais ampla distribuiÃÃo foram D. mollis Benth. e D. gardneriana Tul., ambas ocorrendo no corredor de formaÃÃes vegetacionais abertas e secas no sentido nordeste-sudeste. TrÃs blocos fitogeogrÃficos foram determinados pela anÃlise de agrupamento: A) AmazÃnia; B) EcÃtono Cerrado/Floresta AtlÃntica; e C) Escudo das Guianas. A anÃlise de PAE evidenciou dois grandes clados com suporte estatÃstico (bootstrap) acima de 80%. Foram estabelecidos os seguintes padrÃes de distribuiÃÃo: 1) Amplo na AmazÃnia, 2) Restrito ao Norte da AmazÃnia (acima do Rio Negro), 3) Escudo das Guianas, 4) EcÃtono Cerrado/Floresta AtlÃntica, e 5) MicroendÃmico aos Cerrados de Minas Gerais. Os resultados da modelagem preditiva demonstraram uma tendÃncia de reduÃÃo da maior parte da Ãrea de ocorrÃncia potencial do gÃnero em um cenÃrio futuro, com exceÃÃo de pequenas regiÃes no Escudo das Guianas, as quais teriam condiÃÃes de expandir a distribuiÃÃo das espÃcies. Como conclusÃes, conheceu-se a distribuiÃÃo do gÃnero pela AmÃrica do Sul, seus centros de riqueza e diversidade, no caso a regiÃo amazÃnica, alÃm de ter sido verificada uma tendÃncia geral na reduÃÃo da Ãrea potencial de ocorrÃncia do gÃnero no futuro e expansÃo em Ãreas de altas altitudes, consideradas refÃgios. / Studying patterns of distribution of organisms is relevant in ecology, biogeography and conservation studies under even to understand the evolution of biodiversity. Dimorphandra Schott. is a widespread neotropical genus with 25 taxa in Brazil, occurring in savanna (Cerrado) and ombrophylous vegetation (Amazonia and Atlantic Rain Forest). This work aims to understand the factor that drives the richness and diversity of Dimorphandra. We used SIG, ecological modeling and databank of SpeciesLink and bibliography. These data were classified within biogeographic provinces according to a biogeographically regionalization scheme and according to groups generated by a cluster analysis in the PC-ORD program and a Parsimony Analysis of Endemicity (PAE) by PAUP program. At the same time, ecological distribution models were formulated to the past, present and future based on algorithms (BIOCLIM and CCSM) using the softwares MAXENT and DIVA-GIS, referring to each group according to the UPGMA results. The richness and diversity centers were in the Amazon region. The species more widespread were D. mollis Benth. and D. gardneriana Tul., occurring in the Seasonal Tropical Dry Forests (STDFs) in the northeast-southeast direction. Three phytogeographic groups were determined by the cluster analysis: A) Amazon and its two subgroups: Broad Amazon and North of Negro River Basin; B) Ecotones Brazilian Savanna/Atlantic Forest; and C) Guyana Highlands. PAE analysis pointed two major clades with good support (>80% bootstrap). Some distribution patterns were determined: 1) Broad Amazon Pattern, 2) Restricted to northern Amazon Pattern, 3) Guyana Shield Pattern, 4) Ecotones Brazilian Savanna/Atlantic Forest Pattern and 5) Micro Endemic Minas Geraisâ Brazilian Savanna Pattern. Predictive modeling results demonstrated a trend on reduction of most potential occurrence areas, except in small regions in Guyana Highlands, which expanded in a future scenario. Finally, we can conclude that distribution of the genus through South America was known as well its richness and diversity centers, besides it has been checked a general trend on reduction of potential occurrence area of this genus in a future scenario except in high areas which are considered refugia.

Two ecological models of nitrogen processes in treatment wetlands have been evaluated and compared. These models have been implemented, simulated, and visualized in the Modelica language. The differences and similarities between the Modelica modeling environment used in this thesis and other environments or tools for ecological modeling have been evaluated. The modeling tools evaluated are PowerSim, Simile, Stella, the MathModelica Model Editor, and WEST. The evaluation and the analysis have been performed using McCall’s factors for software quality (McCall et al, 1977), a correlation analysis and the Constant Comparative Method (Glaser&Strauss, 1999). The results show that the modeling tools and the models can both be separated into two categories: Simple Components and Complex Components for the modeling tools, and Simple Models and Complex Models for the models. The major difference between the Simple Components and the Complex Components is the higher possibility of the Complex Components to create and reuse separate components and the higher complexity in these components. The similarities between the categories are that they are consistent, easy to overview and use, if no new components are to be created. The major difference between the Simple Models and the Complex models lies in the number of functions and in the possibility of reuse and expansion. The similarities between all the models are that they are all consequent, logical, valid, specialized, and easy to use if the user has programming skill. To conclude thisthesis, the nitrogen decrease in a constructed treatment wetland can well be simulated using the Nitrification/Denitrification model expressed in Modelica and the MathModelica Model Editor. However, some changes to the Model Editor are recommended to make the creation of the model easier. The most important of these changes are the addition of a tutorial, the ddition of useful error handling and messages, and the removal of unnecessary Visio features.

Climate information is valuable in understanding the ecology of systems affecting wildlife. This information is often unavailable at the landscape scale. This study evaluated the applicability of several climate factor estimates at the landscape-scale, and illustrated the usefulness of estimated climate factors in ecological investigations. Climate variables estimated for each month of the year were solar radiation, temperature, precipitation, and potential evapotranspiration. Map layers for combined temperature and precipitation, and a moisture index were also created. Accuracy of the estimates for temperature and precipitation for each 300 m x 300 m pixel were quantitatively assessed. The methods used estimated mean monthly temperature within 1 degree C. Precipitation estimates were within 9 mm of actual recorded value. The estimates for monthly solar radiation were qualitatively assessed, and provided a reasonable relative index to actual solar radiation. Estimates of potential evapotranspiration were determined to be reasonably accurate. Landscape-scale estimated climate factors were used in 2 case studies. The first used logistic regression to examine the importance of climate factors to the observed distribution of 21 select forest cover-types in Virginia.The second compared the observed climate characteristics for the distributions of 3 species of terrestrial salamanders in Virginia. Winter temperature was the most important climate variable in determining forest cover-type distribution. Several differences in the climate characteristics of the 3 salamander distributions were observed and discussed. The conclusions of this study were that landscape-scale climate factors can be accurately estimated, and the estimates may be helpful in ecological investigations. / Master of Science

京都大学 / 0048 / 新制・課程博士 / 博士(情報学) / 甲第21914号 / 情博第697号 / 新制||情||120(附属図書館) / 京都大学大学院情報学研究科先端数理科学専攻 / (主査)教授 木上 淳, 教授 磯 祐介, 准教授 藤原 宏志 / 学位規則第4条第1項該当 / Doctor of Informatics / Kyoto University / DFAM

Living at the northern limits of its geographic range, the Florida manatee is particularly susceptible to cold stress-related mortality during the winter months, with most deaths occurring in the lower two-thirds of the state. Contributing to this cold stress susceptibility is the manatee's limited physiological and behavioral responses available when thermally stressed. While capable of migrating south in response to falling water temperatures, manatees must still find warm water when ambient river temperature drops below 20°C for more than a few days. This is in part due to the species low metabolic rate, limited capacity for thermogenesis, and limited ability to raise its metabolic rate. Prolonged exposure to cold temperatures may result in cold stress syndrome, which involves a number of potentially life-threatening, if not fatal physiological changes. Survival during the winter months is therefore, dependent upon the manatee's ability to balance basic physiological needs, primarily the need to forage and to obtain fresh water with the need to stay warm. When identifying which animals are most susceptible and where, analyses of statewide manatee mortality records from 1996 through 2011 (n = 823) indicated that, size and location matter. Medium to large-sized calves accounted for the majority of documented death from cold stress (46.6%), while subadults and small calves were the least represented size classes (14.3 % and 9.5%, respectively). Adults slightly outnumbered subadults (15.8%). Males outnumbered females in all size classes but gender differences were not statistically significant. With regards to location, two areas of the state, the southwest and central east coasts showed the highest incidents of cold stress-related mortality. Both are regions with no primary, natural warm-water springs and whose principal warm-water refugia are power plant effluents. Brevard County on the central east coast is the area most at risk during cold weather events accounting for more than 25% of all cold stress deaths statewide. Warm-water sites within this region are few and relatively underrepresented in the literature in an area well-studied in terms of manatee abundance and distribution relative to the operational power plant. Results from cold stress data analyses emphasize the importance of identifying and characterizing the physical attributes of both known and suspected secondary warm-sites used by manatees in this region for both long and short term protection of the species, and its critical habitat. Three locations within Brevard County identified as passive thermal basins (PTBs), and classified as secondary warm-water sites, have been documented supporting in excess of 100 manatees on numerous occasions, and during winters of varying severity. Unique in physical appearance, distance to forage, hydrology including thermal profiles, and when it was used by manatees, each site challenged the accepted definitions and criteria of what constitutes an acceptable and appropriate warm-water site. Through analyses of photo-identification records, site fidelity at two of these warm-water sites, the Berkeley Canal and the Desoto Canal, was established for a minimum of 20 highly identifiable animals, 15 of which used adjacent sites within the same year, and 6 that used both sites but during different years. Observations of daily use patterns within the sites supported optimization of thermoregulation through adjustments in both vertical and horizontal movement, the latter of which seemed to follow the path of the sun. Manatees using the sites also made use of bottom sediment presumably to stay warm at all three locations. Temperature data indicated that water temperatures monitored in the sediment at secondary sites were some of the highest in the county. The predictable movements during all but the coldest weather fronts indicated that manatees utilized these sites during the early morning and afternoon hours when ambient river temperatures were coldest, gradually returning to the river to feed as ambient temperatures began to rise later in the afternoon. The availability of PTBs in proximity to primary warm-water sites within the region may provide an important component needed for manatees to successfully balance the need to forage with the need to stay warm by providing a network that allows for more efficient foraging while reducing exposure to sub-critical ambient river temperatures. The challenge of balancing the need to forage and to maintain homeostasis in the face of thermal stress is complex. This complexity was best approached and better understood through use of a manatee energetics model. The model was designed to facilitate simulation of an unlimited number of different case scenarios involving the exposure of virtually created manatees to a variety of winter conditions as might be experienced by real manatees in a natural system. Sixty-four different simulations were run using six virtual manatees of differing ages, gender, physical parameters, and knowledge of warm-water sites. Simulations were conducted using actual winter water temperature data from Brevard secondary sites and the ambient river from both a mild and a severe winter season. Outcomes, measured as changes in physical parameters indicative of body condition (i.e. mass, percent body fat, blubber depth, girths, etc.), showed that all else being equal, calves in the 2 year-old range fared poorly in all scenarios when compared to individuals of larger size. Subadults fared better than larger adults. This outcome illustrates the complex relationship between size, energy requirements and the synergistic effects of body mass, body fat and blubber thickness on SA:V ratio. Model outcomes agree closely with manatee cold stress mortality analyses predicting that medium to large-sized calves are most susceptible to CS, followed by adults, then subadults. Because all models are simplifications of complex systems, the manatee energetics model is not without its flaws and limitations. The current version of the model could not predict the point at which cold stress mortality would occur. However, a cold stress warning system incorporated into the design alerts the user if potential CSS is likely based on changing physical parameters. Another limitation was the inability of the model to account for the behavioral plasticity of individual subjects since virtual manatees respond to water temperatures based on the user defined rules. A number of additional limitations related to gaps in existing manatee data the gaps were identified and defined. Despite these gaps, the model is designed to allow for incorporation of additional interactions, feedback loops and relevant data as it becomes available and as additional physiological interactions and energy requirements are more clearly defined. Sensitivity analyses, a feature of the model that allowed for modifications in a number of physical as well as environmental parameters, provided an otherwise unlikely opportunity to see how incremental changes in input values, specifically the starting values for mass, percent body fat and blubber depth affected the model's outcome. Ultimately the goal of the model was to facilitate a better understanding of complex relationships by challenging our preconceived understanding of the manatee and its environment.

Land use/land cover (LULC) plays a vital role in the provision of ecosystem services. Changing natural vegetation to agricultural or urban uses affectss the socio-economic dynamics of the surrounding populations. Changing land use may also result in the impairment of ecosystem services. Rural and urban communities have different realities and perceptions on the importance of ecosystem services, which tend to vary by economic, generational, social, and cultural factors. This study assesses the subjective value of people in urban and rural communities on provisioning, regulating, and cultural ecosystem services. A quantitative and qualitative survey using interviews and spatial analysis with GIS is used to study ecosystem services. The survey evaluates the values, perceptions, and benefits of ecosystems in urban and rural regions. Environmental conservation plans that have been implemented on a national level and the traditional ecological activities that local communities play a crucial role in the governance of ecosystem services. There is a need to understand the effectiveness and awareness o actions resulting from governance regimes to be both applicable and useful in protecting ecosystem services in fragile and diverse ecosystems such as Ecuador.

Forested mountain watersheds provide essential resources and services (e.g., water supply) to downstream ecosystems and human communities. Fast-growing mountainside residential development not only modifies the terrestrial system but also aquatic systems by changing the nutrient input from the terrestrial to aquatic. However, the impacts of mountainside residential development on stream ecosystems are complex because interactions between in- stream process and hillslope soils control in-stream nutrient dynamics, and it is difficult to experimentally study these interactions at broad spatial scales. In my dissertation research, I first developed models for leaf decomposition in a forested headwater stream by synthesizing several important ecological concepts, including ecological stoichiometry, microbial nutrient mining, and microbe-substrate interaction. I then extended the single stream model to a stream network model and further linked the stream network model with a terrestrial model that simulates nutrient processes and hydrology in hillslope soils. With this complete modeling framework, I conducted a global sensitivity analysis to evaluate the importance of terrestrial nutrient input versus in-stream processes in modifying nitrogen export. I also conducted a simulation to investigate the impacts of housing density, buffer zone protection, and stream travel distance from the residential development to the catchment outlet on nitrogen export at the local and regional scale. The model for leaf decomposition performed better for predicting detritus decay and nutrient patterns when microbial groups were divided into immobilizers and miners and when leaf quality was included as a variable. The importance of terrestrial nutrient input versus in-stream nutrient processes greatly depended on the level of terrestrial nutrient input. When terrestrial nitrate input was low, nitrogen export was more sensitive to in-stream net microbial nitrogen flux (mineralization - immobilization) than nitrate input. However, when terrestrial nitrate input was high, nitrate input was more important than in-stream net nitrogen flux. Greater impacts, i.e., higher nitrogen export at the local scale or greater change in nitrogen export at the regional scale, were associated with higher residential density, a lack of buffer zone protection, and shorter stream travel distance from the residential development to the catchment outlet. Although subject to model assumptions and further validation through field experiments, this research provides a general modeling framework for in-stream processes and aquatic-terrestrial linkages and expands an understanding of interactions between terrestrial and in-stream nitrogen dynamics and the impacts of mountainside development on stream ecosystems, identifies directions for further research, and provides insights for land and river management in mountainous areas. / Ph. D.

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