**Water Column Sampling:**

Water column sampling was performed on four cruises during the spring and the summer of 2011 and 2012 at the Bermuda Atlantic Time-series Study station (31’40°N 64’10°W, BATS) and in the mesoscale eddies found in the surrounding area of the Sargasso Sea. For each cruise, two stations were sampled, usually in the center of a mesoscale eddy and at BATS. The edge of the eddy was sampled two times, as well. To be able to get a better reproducibility of data, each experiment was replicated.

For each experiment, seawater samples were collected pre-dawn (on deck 2:30-4:00, local time) at four different depths within the euphotic zone (20m, 50m, 80m and the Deep Chlorophyll Maximum, DCM). Twenty-one 10L Niskin bottles were attached to a rosette with conductivity, temperature, depth sensors (CTD), and an *in vivo* fluorometer. This sensor allowed for recording in real time of chlorophyll fluorescence and the DCM for each station. The water that was collected from the 10L Niskin bottles was sampled for abundance and biomass of the plankton community.

**Dilution experiments:**

The growth and grazing rates of the phytoplankton community were determined using the dilution method with a two-point modification (Landry et al. 1984, Landry et al. 2008a, Selph et al. 2011). To check the applicability and resolution of the two point modification, two experiments at two different stations (one replicate of C3 and one replicate of B3) were carried out with a complete set of dilutions (25, 50, 75 and 100%) on-deck; the results of these experiments confirmed the validity of the 2 point dilution experiment (data not shown). The Landry-Hassett dilution method was used because it can separate autotrophic and heterotrophic processes with relatively little manipulation (Landry & Hassett, 1982, Landry et al 1984). The method is based on the concept that the dilution lowers the encounter probability of grazer and prey and enables one to calculate from the net phytoplankton population growth (k) in the diluted and undiluted incubation bottles the instantaneous growth rate (u) and grazing mortality (g), where k= u - g (Landry & Hassett, 1982). This method is based on three assumptions. The first one is that the instantaneous growth rate of any phytoplankton group is not influenced by the dilution of the “seawater/sample” with particle free water. The second assumption is that the probability of a phytoplankton to be grazed by the consumer is directly related to the abundance of the consumer itself, which means that with a higher density of grazers the probability of the prey to be grazed is higher as well, making the grazing a function of the dilution. The third assumption is the sum of the first two assumptions, that the change in the abundance of the phytoplankton community over time can be expressed by the exponential growth equation:

Pt=P0 *e ^* [(u -xg)t]

where P0 and Pt are the initial and the final abundances or biomass of bulk or specific phytoplankton groups, t is time in day (d), u is the instantaneous phytoplankton growth rate (d-1), g is the grazing mortality (d-1), and x represents the dilution factor (Landry and Hassett 1982; Landry et al. 1984; Neuer and Cowles 1994). M and g are calculated as the y-intercept and the slope of the linear regression plotted against the dilution factors where k=1/t ln (Pt/P0) (Landry & Hassett, 1982; Landry et al. 1984). For the two-point dilution, this converts mathematically to solving the equations for u and g (Landry et al., 2008a).

**Bulk measurements**:

Chlorophyll-a was extracted from seawater (250 ml and 400 ml depending on the dilution), with 90% acetone and measured after 24hrs at 4 degrees C in the dark onboard the ship using a TD 700 Laboratory Fluorometer using the non-acidification technique (Welschmeyer 1994). These data were used as a proxy for the phytoplankton biomass in the water column and to calculate the bulk growth and grazing rates of the phytoplankton community.