This study presents an original 18S rRNA PCR survey of the freshwater picoeukaryote community, and was designed to detect unidentified heterotrophic picoflagellates (size range 0.6-5 microm) which are prevalent throughout the year within the heterotrophic flagellate assemblage in Lake Pavin. Four clone libraries were constructed from samples collected in two contrasting zones in the lake. Computerized statistic tools have suggested that sequence retrieval was representative of the in situ picoplankton diversity. The two sampling zones exhibited similar diversity patterns but shared only about 5% of the operational taxonomic units (OTUs). Phylogenetic analysis clustered our sequences into three taxonomic groups: Alveolates (30% of OTUs), Fungi (23%) and Cercozoa (19%). Fungi thus substantially contributed to the detected diversity, as was additionally supported by direct microscopic observations of fungal zoospores and sporangia. A large fraction of the sequences belonged to parasites, including Alveolate sequences affiliated to the genus Perkinsus known as zooparasites, and chytrids that include host-specific parasitic fungi of various freshwater phytoplankton species, primarily diatoms. Phylogenetic analysis revealed five novel clades that probably include typical freshwater environmental sequences. Overall, from the unsuspected fungal diversity unveiled, we think that fungal zooflagellates have been misidentified as phagotrophic nanoflagellates in previous studies. This is in agreement with a recent experimental demonstration that zoospore-producing fungi and parasitic activity may play an important role in aquatic food webs.

Seasonal and depth variations of the abundance, biomass, and bacterivory of protozoa (heterotrophic and mixotrophic flagellates and ciliates) were determined during thermal stratification in an oligomesotrophic lake (Lake Pavin, France). Maximal densities of heterotrophic flagellates (1.9 × 103 cells ml -l) and ciliates (6.1 cells ml -r) were found in the metalimnion. Pigmented flagellates dominated the flagellate biomass in the euphotic zone. Community composition of ciliated protists varied greatly with depth, and both the abundance and biomass of ciliates was dominated by oligotrichs. Heterotrophic flagellates dominated grazing, accounting for 84% of total protistan bacterivory. Maximal grazing impact of heterotrophic flagellates was 18.9 x 106 bacteria 1-Ih-l. On average, 62% of nonpigmented flagellates were found to ingest particles. Ciliates and mixotrophic flagellates averaged 13% and 3% of protistan bacterivory, respectively. Attached protozoa (ciliates and flagellates) were found to colonize the diatom Asterionella formosa. Attached bacterivores had higher ingestion rates than free bacterivorous protozoa and may account for 66% of total protozoa bacterivory. Our results indicated that even in low numbers, epibiotic protozoa may have a major grazing impact on free bacteria.

Summary 1. A substantial fraction of the freshwater available in neotropical forests is impounded within the rosettes of bromeliads that form aquatic islands in a terrestrial matrix. The ecosystem functioning of bromeliads is known to be influenced by the composition of the contained community but it is not clear whether bromeliad food webs remain functionally similar against a background of variation in the understorey environment. 2. We considered a broad range of environmental conditions, including incident light and incoming litter, and quantified the distribution of a very wide range of freshwater organisms (from viruses to macroinvertebrates) to determine the factors that influence the functional structure of bromeliad food webs in samples taken from 171 tank‐bromeliads. 3. We observed a gradient of detritus‐based to algal‐based food webs from the understorey to the overstorey. Algae, rotifers and collector and predatory invertebrates dominated bromeliad food webs in exposed areas, whereas filter‐feeding insects had their highest densities in shaded forest areas. Viruses, bacteria and fungi showed no clear density patterns. Detritus decomposition is mainly due to microbial activity in understorey bromeliads where filter feeders are the main consumers of microbial and particulate organic matter (POM). Algal biomass may exceed bacterial biomass in sun‐exposed bromeliads where amounts of detritus were lower but functional diversity was highest. 4. Our results provide evidence that tank‐bromeliads, which grow in a broad range of ecological conditions, promote aquatic food web diversity in neotropical forests. Moreover, although bromeliad ecosystems have been categorised as detritus‐based systems in the literature, we show that algal production can support a non‐detrital food web in these systems.