Simulated phytoplankton bloom input in top-down manipulated microcosms: comparative effect of zooflagellates, ciliates and copepods

International audienceSuccession of microbial populations and carbon flow were studied experimentally in a 2-stage linked system where phytoplankton growth was separated from decay or consumption. Three phytoplankton loss processes were considered: bactenal lysis, protozoan grazing and mesozooplankton grazing. An axenic culture of the &atom Phaeodactylum tricornutum was transferred to 3 second-stage microcosms (150 1) kept in the dark. The first vessel (L: lysis batch) contained no herbivores and was top-limited by bacterivorous flagellates, the second (G: grazing batch) by ciliates and the third (Z: zoo-plankton batch) by copepods. In the L container there was no evidence of grazing on phytoplankton; the particle-associated protozoa were abundant (3.7 X lo3 cell 1-l). In the G vessel, cihates (up to 96 X 103 cell I-') controlled both the phytoplankton and nanoflagellate populations and ciliates ingested phytoplankton at rates from 1.7 to 16 phytoplankton cell ciliate-' h-' (average, 7.5 cell ciliate-' h-'). The average growth ylelds for bacterivorous flagellates and ciliates were 35 % and 45 %, respectively. In the Z treatment, the concentration of > l 0 pm protozoa (able to ingest P. tricornutum) was apparently kept low due to copepod grazing pressure. Consequently, a considerable fraction of the phytoplankton was channelled through the detrital pathway instead of the classical herbivorous pathway. The general trend of the parameters descnbing microbial activity was Z > G > L. An influence of the complexity of the system on the turnover rate of organic matter was evldent with the ratio 'cumulative biomass/POC' in the different vessels showing a lower level of detntus when predators of phytoplankton were present