For many years, the importance of microorganisms attached to surfaces in littoral zones and wetlands has been disregarded when describing aquatic ecosystem dynamics. Supporting evidence is scarce but convincing that these microbial communities are not only very productive but can often serve as major regulators of nutrient and carbon dynamics in many freshwaters. In order to determine the quantitative importance of epiphytic bacteria for the overall carbon turnover, I compared the relative contribution of epiphytic bacteria on the submerged macrophyte Ranunculus circinatus, sediment and free-living bacteria to the total bacterial production. Sediment bacteria generally dominated total bacterial biomass in the littoral zone. Although the epiphytic biomass on R. circinatus was ten times lower than the biomass of sediment bacteria, it often contributed at least equally to the total bacterial production. Thus, the results presented in this thesis confirm that most bacterial biomass and production in shallow lakes is associated with surfaces, and that in littoral zones with dense macrophyte stands, epiphytic bacteria can contribute significantly to the overall carbon turnover. There is increasing evidence that not all cells in natural bacterial communities are metabolically active. In Lake Erken, there were large differences in the fraction of active bacteria between different habitats, while the within-habitat differences were small. The sediments had the largest bacterial fraction, followed by epiphytic bacteria, while in the water column only a few percent of the bacteria were active. In this thesis the fraction of active bacteria is connected to environmental fluctuations. I hypothesize that smaller fluctuations in chemical, biological or physical factors result in large active bacterial fractions. Thus, small environmental fluctuations within a habitat allow large active bacterial fractions, while the active fraction is constrained when the environmental fluctuations are large.

A otimização e padronização de métodos moleculares são fundamentais para evitar distorções nos resultados causadas por processamento de ácidos nucleicos da abundância relativa de micro-organismos de uma amostra. Nos estudos sobre identificação microbiana, a etapa de validação é, na maioria dos casos, negligenciada. As principais etapas em que podem ocorrer vieses capazes de alterar a informação sobre a composição da comunidade microbiana são a extração e a RT-PCR. O rRNA é a molécula ideal para identificar os micro-organismos metabolicamente ativos por estar em maior quantidade dentro da célula. A inexistência de um protocolo de extração de RNA gold standard e de kits comerciais específicos para cada tipo de amostra ambiental pode comprometer as etapas posteriores à extração. O protocolo de extração de RNA adaptado neste trabalho mostrou alta eficácia na lise celular e na remoção de contaminantes co-extraidos, garantindo a integridade e a qualidade do rRNA extraído de amostras contendo altas concentrações de contaminantes. Devido as diferentes características químicas das amostras ambientais, o uso deste protocolo adaptado pode preservar a informação sobre os indivíduos metabolicamente ativos de uma comunidade microbiana. Foram utilizados consórcios artificiais na validação da etapa de PCR. O efeito sinérgico do uso de aditivos, da Taq polimerase de alto rendimento, do programa de sub-ciclagem e da limitação do programa de amplificação em 10 ciclos, mantiveram a proporção dos moldes dos consórcios analisados próximo ao esperado. Após a padronização da técnica de ARDRA, foi definido que há necessidade de entre 500 e 550 UFC para uma cobertura completa da diversidade de lodo ativado. O uso de enzimas combinadas MspI/HaeIII e HhaI/RsaI em uma mesma reação double digest proporcionou a separação dos clones de lodo ativado utilizando menos etapas de ARDRA. E o critério de corte em 3% do total agrupamentos, possibilitou selecionar os perfis mais representativos sem a perda de grupos importantes para a análise das bibliotecas. Foi concluído que, o uso de um protocolo inadequado de extração de RNA de amostras complexas pode gerar extratos contendo contaminantes co-extraidos que interferem na atividade da Taq polimerase e nas enzimas de restrição. Após o sequenciamento de nova geração, foi observado que o uso de diferentes iniciadores de PCR e do pré-processamento manual para os dados obtidos, foram essenciais para ampliar a cobertura observada para a amostra de lodo e melhorar a resolução dos resultados e a profundidade de identificação taxonômica, respectivamente. / The optimization and standardization of molecular methods are fundamental to avoid distortions in the results caused by nucleic acid processing of the relative abundance of microorganisms in a sample. In the studies on microbial identification, the validation step is, in most cases, neglected. The main steps in which biases capable of altering the composition of the microbial community can occur are extraction and RT-PCR. The rRNA is the ideal molecule to identify metabolically active microorganisms because they are in the greatest amount within the cell. The lack of a gold standard RNA extraction protocol and specific commercial kits for each type of environmental sample may compromise the post-extraction steps. The RNA extraction protocol adapted in this work showed high efficacy in cell lysis and in the removal of co-extracted contaminants, guaranteeing the integrity and quality of the rRNA extracted from samples containing high concentrations of contaminants. Due to the different chemical characteristics of the environmental samples, the use of this adapted protocol can preserve the information about the metabolically active individuals of a microbial community. Artificial consortia were used to validate the PCR step. The synergistic effect of the use of additives, the high yield Taq polymerase, the sub-cycling program and the limitation of the amplification program in 10 cycles, kept the proportion of the molds of the consortia analyzed close to the expected. After standardization of the ARDRA technique, it was defined that there is a need for between 500 and 550 CFU for a complete coverage of the diversity of activated sludge. The use of MspI / HaeIII and HhaI / RsaI combined enzymes in the same double digest reaction provided the separation of activated sludge clones using fewer ARDRA steps. And the criterion of cut in 3% of the total groupings, allowed to select the most representative profiles without the loss of important groups for the analysis of the libraries. It was concluded that the use of an inadequate RNA extraction protocol from complex samples can generate extracts containing co-extracted contaminants that interfere with Taq polymerase activity and restriction enzymes. After the sequencing of the new generation, it was observed that the use of different PCR primers and manual preprocessing for the obtained data were essential to increase the coverage observed for the sludge sample and to improve the resolution of the results and the depth of taxonomic identification, respectively.