Phytoplankton forms the base of aquatic food webs and element cycling in diverse aquatic systems. The fate of phytoplankton-derived organic matter, however, often remains unresolved as it is controlled by complex, interlinked remineralization and sedimentation processes. We here investigate a rarely considered control mechanism on sinking organic matter fluxes: fungal parasites infecting phytoplankton. We demonstrate that bacterial colonization was promoted 3.5-fold on fungal-infected phytoplankton cells in comparison to non-infected cells in a cultured model pathosystem (diatom Synedra, fungal microparasite Zygophlyctis, and co-growing bacteria), and even ≥17-fold in field-sampled populations (Planktothrix, Synedra, and Fragilaria). The Synedra–Zygophlyctis model system further revealed that fungal infections reduced the formation of aggregates. Moreover, carbon respiration was 2-fold higher and settling velocities 11–48% lower for similar-sized fungal-infected vs non-infected aggregates. Our data imply that parasites can effectively control the fate of phytoplankton-derived organic matter on a single-cell to single-aggregate scale, potentially enhancing remineralization and reducing sedimentation in freshwater and coastal systems.

Formation of sinking aggregates (model system, cylinder rotation)

Triplicates of non-infected and fungal-infected cultures were grown for six days until the infection prevalence had reached 59±3% (N=3 flasks) in the infected treatment. Thereafter, the cultures were diluted with medium to approx. 5,000 cells mL^-1 (resembling cell abundances during bloom scenarios) and transferred into rotating cylinders (r=8.5 cm, h=10 cm, V=2.3 L, bubble-free), to facilitate the aggregation of cells due to their stickiness and differential settling behavior (setup is shown in Supplementary Fig. S2). The aggregate formation was recorded over time using a Mini Deep Focus Plankton Imager (MDPI, Bellamare, La Jolla, CA, USA). The MDPI is a ‘shadowgraph imager’ using a near-infrared LED light source (A007 Indus star, LED dynamics, Randolph, VT, USA) behind a pinhole and a set of identical plano-convex collimator lenses to create parallel light beams between the LED light pod and the camera pod (Supplementary Fig. S2). Using this parallel light, i.e., telecentric optics, ensured that all objects (aggregates) were in focus, independent of their position between the lenses. We recorded two image sequences per cylinder every 2–9 hours. Each sequence lasted for at least one rotation (50 images). The sampling volume of two image sequences covered 2x 0.35 L, equal to 30% of the entire cylinder volume.

Automated particle analyses in ImageJ provided the number of particles per volume and the cross-sectional area A of each particle (used to calculate the equivalent circular diameter, denotes as diameter, assuming spherical geometry). Aggregates were counted, measured, and binned into size classes, in which the upper diameter was 1.3-times the lower size class. The resulting eleven size classes ranged from 0.4 to 5.6 mm (aggregates with ECD<0.4 mm were excluded from the data set, and aggregates with ECD>5.6 mm were not present).

Data can be accessed using Image J, Notepad and Excel.