General experimental setup

A mesocosm and a microcosm experiment were conducted using the same initial phytoplankton diversity gradient and different nutrient treatments.

For generating the diversity gradient, a natural phytoplankton community was collected from a lake (Grafschaftsee, Germany, 53°330 05″ N; 7°580 49″ E) at the end of the summer (2017). Grazers were excluded by filtering the water through a 53μm mesh and six diversity levels were generated by dilution of the natural community and thus, removal of rare-species. Nutrient treatments were performed by manipulating N and P concentrations while other elements were supplied in excess according to WC phytoplankton growth medium (Guillard & Lorenzen, 1972). Mesocosm experimental setup

The six diversity treatments were inoculated under two nutrient levels (low and high) in 600L indoor mesocosms comprising 12 experimental units. The phytoplankton inoculum (corresponding to the six diversity levels) was added in small volumes (between 80 and 150mL) to the mesocosms assuring the same initial biomass amount in all treatments. For the low nutrient treatments, 0.65μmolL^-1 P and 14.2μmolL^-1 N were added, and to simulate nutrient enrichment, but maintaining the same N:P ratios, 1.9μmolL^-1 of P and 42.6μmolL^-1 of N were added in the high nutrient treatments. The mesocosms were sampled every four days from day 7 to day 27 comprising a total of six samplings. The experiment was continued until the phytoplankton communities reached the stationary growth phase. In every sampling event, 10% (60L) of water was replaced with water containing the design nutrient concentrations.

Microcosm experimental setup

The microcosm experiment was carried out using a factorial design where each of the six diversity levels was incubated under 25 different nutrient conditions leading to a total of 150 experimental units. For that, cell culture flaks (250mL polystyrene, Sarstedt Ltd. Leicester, UK) were filled with 200mL water taken from the six diversity treatments of the mesocosms (from the low nutrient treatment) on day 7. The nutrient conditions were based on combinations of 0.1, 0.7, 1.3, 1.9 and 2.6µmol P L^-1 and 3, 19, 35, 48 and 62µmol N L^-1 and added in each diversity level treatment. Samples were taken when the stationary growth phase was reached.

Laboratory analysis

Water samples for particulate organic carbon (POC), nitrogen (PON) and phosphorus (POP) were filtered onto acid-washed pre-combusted glass-fibre filters (Whatman GF/C). Filters for POC and PON were measured using an elemental analyser (Flash EA 1112, Thermo Scientific). Water samples for dissolved nitrogen fractions (NO_x^- and NH₄⁺) were filtered and determined following the method described by Schnetger & Lehners (2014) for NO_x^- (NO₃^- and NO₂), and a modified version of Benesch & Mangelsdorf (1972) method for ammonium (NH₄⁺). POP and dissolved phosphorus (PO₄^3-) were measured by molybdate reaction of filtered water samples for the dissolved fraction and after digestion with potassium peroxydisulfate (K₂S₂O₈) solution for the particulate fraction (Wetzel & Likens, 2013). Phytoplankton samples corresponding to the initial communities of each experiment and the end of the mesocosm experiment were preserved in lugol solution until counted and identified under inverted microscope based on Utermöhl's method (Utermöhl, 1958). Phytoplankton was identified to the species level and morphospecies were used when clear assignment of a species name was not possible.

Calculations

We used richness and inverse Simpson diversity as diversity indices for estimating realized initial diversity in the experiments and at the end of the mesocosm experiment. The standing biomass was measured as POC, and RUE was calculated as unit of biomass per unit of total nutrient in the system (RUE_P= POC/total P or RUE_N= POC/total N, respectively), where total nutrients represent the sum of dissolved and particulate N or P, respectively.