We conducted 10 on-deck incubation experiments (Table 1) on board the R/V Hesperides during the Malaspina 2010 Expedition¹. The first set of experiments (#1–4) was performed in the Pacific Ocean during the expedition’s leg from Auckland, New Zealand to Honolulu, U.S. during April-May 2011. The second set of experiments (#5–10) was performed in the Atlantic Ocean during the expedition’s leg from Cartagena de Indias, Colombia to Cartagena, Spain during June–July 2011 (Figure 1).

Water samples were collected at 3 m depth in 30 L Niskin bottles. The water was filtered through a 50 μm mesh to remove the largest fraction of grazers. Water samples were dispensed into duplicate 2 L bottles for each treatment: quartz flasks, which allowed the full solar light spectrum (FULL treatment) to pass through, polycarbonate Nalgene bottles, which removed UVB and most UVA, and between 35% of transmittance at 399 nm and 86% of transmittance at 419 nm, allowing visible wavelengths to pass through (-UV treatment), and black bottles, which blocked all solar radiation (DARK treatment). The bottles were incubated in an open-air tank (V = 1.9 × 10³ L) and maintained at in situ temperature by using a continuous surface water running system. The experiments started after sunrise, at about 07:00 h local time (initial time T₀), and lasted for approximately 24 h (before dawn, final time T_f). The sampling time at T₀ and T_f were used for Chl a and TEP analysis (except for Exp 4 during which extra sampling at T₃ in the FULL treatment was also performed). Intermediate samplings coupled with the photoperiod were also performed to study picoplankton abundances: a first sampling (T₁) after 4 h of sunlight exposure, a second sampling (T₂) after 6 to 8 h, a third sampling at sunset (T₃) and a fourth sampling (T₄) 1 h after sunset.

In experiments 4, 5, and 6 an additional treatment to control for potential abiotic TEP production, due to spontaneous assemblage from dissolved acidic polysaccharide precursors was included by incubating seawater filtered by gravity through a 0.2 μm cartridge (to eliminate phytoplankton and bacterial cells) into duplicate 2 L quartz bottles.

Incident solar radiation (W/m²) was measured every 2 min by a Radiation Sensor (2770; Aanderaa Instruments) installed in the meteorological station of the vessel (SADO) placed over the deck. The sensor is sensitive in the range of 300–2500 nm, but is covered by a glass (borosilicate) dome, filtering out most of the incident UVR. To measure solar radiation in the ultraviolet bands, we used an incident UVA (315–400 nm) and UVB (280–315 nm) radiometer (UVS-AB-T; Kipp & Zonen, Version T2.07.217) also installed in the vessel’s meteorological station. Doses of solar radiation received during the experiments were then calculated by integrating radiation values for the day’s duration and expressed in kJ/m².

Transparent Exopolymer Particles were stained with alcian blue, a group of basic, polyvalent and water-soluble dyes, derived from phthalocyanine. In a solution of 3% acetic acid (pH = 2.5) alcian blue stained carboxylated and esther-sulfate groups of acidic mucopolysaccharides. TEP were retained on polycarbonate filters (Poretics) with 0.4 μm pore size and of 25 mm in diameter and stained with 1 mL working solution of alcian blue pre-filtered through a 0.2 μm filter to a final concentration of 0.02%. TEP concentrations were determined spectrophotometrically following the colorimetric method proposed by Passow and Alldredge (1995). For each treatment of the experiments, 400 to 1000 mL volumes of the sample (depending on the saturation of the filter) were collected at the initial (T₀) and final (T_f) times (in Exp 4 also at T₃). Duplicates were filtered under low and constant pressure (150 mmHg). Filters were subsequently stained with 1000 μL of a 0.02% working solution of Alcian blue (pre-filtered through a 0.2 μm filter) in 0.06% acetic acid (pH = 2.5), left for few seconds, filtered again and rinsed twice with MilliQ water to eliminate excess dye. The filters were frozen at -80°C and stored until further processing in the laboratory. The filters were then thawed, soaked in 80% sulfuric acid for 2–3 h, and the absorbances were measured by means of a spectrophotometer (PerkinElmer UV-VIS Lambda 35) at a fixed wavelength of 787 nm. Empty stained filters (stained and frozen in parallel with the samples) were used as blanks.

Alcian blue absorption was calibrated using a Xanthan Gum (XG) solution that was homogenized by a tissue grinder and measured by weight. TEP concentrations were, therefore, expressed in μg of XG equivalents per liter and calculated as follows:

where a_sample is the absorbance of the sample, a_blank is the absorbance of the blank, V is the filtered volume of the sample and F is the calibration factor, which is the inverse of the slope between μg of XG and alcian blue absorption at 787 nm. The detection limit of the method was 1 μg of XG equivalents L^-1 and the coefficient of variation was ca. 20%.

The TEP production rate per day (α) was calculated as the natural logarithm of the ratio between the concentrations (μg XG Eq. L^-1) observed at the final and initial experimental times (t = 1 day):

where TEP t_f and t₀ are the TEP concentrations at the final and initial times, respectively.

We also calculated the normalized (Passow, 2012) or net (Egge et al., 2009; MacGilchrist et al., 2014) TEP production (TEP_PR) expressed as μg XG Eq. L^-1 d^-1 and calculated from the difference between the TEP concentration at the end of the experiment (TEP t_f) minus the initial TEP concentration (TEP t₀) divided by the experimental time:

The concentration of chlorophyll a (Chl a) was quantified by using the fluorimetric method described by Parsons et al. (1984). Seawater samples (250 mL) were collected at the initial (T₀) and final (T_f) time of each experiment and filtered through 25 mm Whatman GF/F filters. Chlorophyll a in the filters was extracted in 90% acetone and left 24 h in the dark at 4°C until processing in a Turner Design fluorimeter, calibrated with pure chlorophyll a.

Picoplankton populations were quantified by using flow cytometry.

To quantify the abundance of heterotrophic prokaryotes, we fixed 1.2 mL of each sample treatment from all sampling times (T₀, T₁, T₂, T₃, T₄, T_f) with 48 μL of glutaraldehyde (final concentration 1%). The samples were left for 10 min in the dark and then were frozen in liquid nitrogen and stored at -80°C until analysis in a FACScalibur (Becton Dickinson) flow cytometer equipped with a 488 nm argon laser. The samples were defrosted and analyzed following Marie et al. (1997) by staining 400 μL with 4 μL of 10 × SYBR Green I (Molecular probe S-7563) solution (final dilution 1:1000 [vol/vol]) for 10 min and letting the stained samples run through the FACScalibur at low speed (ca. 25 μL min^-1) and adding 10 μL of a fluorescent 1 μm latex beads solution as an internal standard for cytometric counts. The run terminated when about 10,000 particles had been recorded or after 90 s. The threshold was set in the green channel. Stained bacteria were detected and discriminated from other non-bacterial particles with a light side scatter (SSC) and green (FL1) and red fluorescence (FL3).

To measure the abundance of phototrophic prokaryotes (Prochlorococcus and Synechococcus), we ran 400 μL of each sample treatment from all sampling times (T₀, T₁, T₂, T₃, T₄, T_f) at medium speed (ca. 55 μL min^-1) in the flow cytometer. A sample (10 μL) of a calibrated solution of 1-μm diameter yellow-green fluorescent latex beads (Polysciences) was added as an internal standard for the quantification of cell concentrations. Beads concentrations [9.33 × 10⁵mL^-1 for the Pacific Ocean experiments] [4.63 × 10⁵mL^-1 for the Atlantic Ocean experiments] in the standard solution were calculated by filtering duplicate aliquots onto black Nucleopore filters and counting under an epifluorescence microscope. All samples were fixed with glutaraldehyde (1% final concentration) and stored at -80°C until their analysis in the laboratory. Synechococcus populations were detected by an orange fluorescence dot plot (FL2 = interval between 585 ± 20 nm of wavelength, corresponding to the fluorescence of the phycoerythrin pigment) versus red fluorescence (FL3 ≥ 635 nm, corresponding to the fluorescence of chlorophyll). Prochlorococcus had a lower FL3 signal and no FL2 signal. Moreover, the populations were detected in a SSC versus FL3 dot plot (Marie et al., 1997). Decay rates and growth rates (h^-1) of Synechococcus, Prochlorococcus, and heterotrophic bacterial populations were then calculated as the slope of the linear regression between the natural logarithms of the concentrations (cells mL^-1) and the experimental time (hours) during each sampling interval.

To corroborate that Prochlorococcus could be a source of TEP we set up two non-axenic cultures of a strain of P. marinus (RCC 0156). This strain was grown in the laboratory under one-liter batch aseptic conditions with PCR-11 culture medium at 22°C under low light of 30 μmol Photons m^-2 s^-1. We did not include any UV light incubation due to the difficulty and limitations of keeping this species growing in the laboratory. The cultures were sampled regularly during 2 months to monitor changes in cell abundance and TEP generation during the exponential and stationary phases. Triplicated volumes of 20 mL for TEP quantification were sampled by syringe, gently filtered under low vacuum pressure and immediately analyzed as previously described (Passow and Alldredge, 1995). P. marinus cell abundance (cells mL^-1) was quantified in replicated 1 ml fresh samples in a BD FACScalibur flow cytometer.

Statistical analyses of TEP and picoplankton data were carried out using JMP software, Student’s t-test was used to test for significant differences between oceans, treatments (DARK, FULL, -UV) and other changes in parameters, and linear regression was used for the analysis of the relationship between TEP production and picoplankton cell decay.

Surface temperature and salinity data collected from the stations near the sampling areas indicated that the experiments were conducted in warm waters (Table 1), ranging from 29.4°C in the Caribbean Sea to 21.8°C at the most north-eastern station of the Atlantic Ocean (Figure 1); the most saline waters were located in the North Atlantic Ocean (Table 1).

The chlorophyll a concentration was low in all the biogeographic provinces studied, ranging from 0.05 μg L^-1 in the Northern Atlantic to 0.31 μg L^-1 in the equatorial Pacific Ocean.

Initial abundances of picoplankton were generally higher in the waters sampled in the Pacific Ocean than in the waters sampled in the Atlantic Ocean. Heterotrophic bacterial concentrations were also higher in the Pacific Ocean (mean 9.23 ± 2.00 × 10⁵ cells mL^-1) than in the Atlantic Ocean (mean 3.96 ± 1.39 × 10⁵ cells mL^-1), reaching maximum abundance at the Equatorial station (1.45 ± 0.4.3 × 10⁶ cells mL^-1). Prochlorococcus sp. was more abundant (mean Pacific 6.42 ± 0.38 × 10⁴ cells mL^-1; mean Atlantic 2.51 ± 0.17 × 10⁴ cells mL^-1) than Synechococcus sp. (mean Pacific 3.71 ± 0.50 × 10³ cells mL^-1; mean Atlantic 4.05 ± 0.33 × 10³ cells mL^-1).

UVR conditions measured during the experiments in the Pacific Ocean were slightly higher at the northern latitudes and were the highest during Exp 4 (Table 1). During the Atlantic Ocean experiments, UVR doses were slightly higher at the northern latitudes, in the North Atlantic Gyre, than in the Caribbean experiments but in all the cases lower than during the Pacific Ocean experiments (Table 1). UVB daily doses measured during the Pacific Ocean experiments (mean 43.6 ± 3.1 KJ m^-2) were significantly higher than those measured during the Atlantic Ocean experiments (mean 30.5 ± 2.5 KJ m^-2; t = 13.09, df = 8, p < 0.005).

Initial TEP concentrations in the Pacific Ocean were significantly higher (mean 24.45 ± 2.3 μg XG Eq. L^-1) than in the Atlantic Ocean (mean 8.18 ± 4.56 μg XG Eq. L^-1; t-test: t = 16.26, df = 8, p < 0.05) (Figure 2).

In all the experiments performed in the Pacific Ocean, we detected, after the 24 h of incubation, a significant increase in TEP concentrations under full solar radiation, but not in the DARK and -UV treatments (Figure 2A; t-test: Full-Initial t = 81.05, df = 7, p < 0.05). The highest concentration of TEP was consistently observed in the full solar radiation treatments, reaching values up to 187.3 μg XG Eq. L^-1 in the Exp 3 (Figure 2A). The potential that TEP was produced by self-assembly due to abiotic processes was tested in Exp 4. When microorganisms were excluded from the waters, there was no detectable increase in TEP (Figure 2C). An additional sampling time at the end of the day in Exp 4 revealed that the production of TEP during the day increased to 60.2 ± 8 XG Eq. L^-1, although no significantly higher than the value observed after 24 h (Figure 2A, 47.4 ± 3.83 XG Eq. L^-1).

In the experiments performed in the North Atlantic Ocean (Exps 5–10, Table 1) we also observed an increase in TEP production after the incubation period in most of the experiments, but, in contrast with the Pacific Ocean experiments, this increase in production was not consistently induced by solar radiation exposure (Figure 2B). Maximum concentrations of 26.9 μg XG Eq. L^-1 were produced during Exp 7 (Figure 2B). In the experiments 8, 9, and 10, we observed higher concentrations in the full solar radiation treatments but these increases were not statistically significant (Figure 2B). In the two experiments (#5 and 6) in which an extra treatment excluding the microorganisms was performed, we observed contrasting results (Figure 2C). In Exp 5 significantly more TEP was produced in the 0.2 μm-filtrated treatment (Figure 2C) (ca. 50 μg XG Eq. L^-1) than in the full treatment including microorganisms (Figure 2B) (ca. 10 μg XG Eq. L^-1). This fact suggests the relevance of abiotic self-assembly, although no difference was observed in the case of the Exp 6 (Figure 2C).

Transparent Exopolymer Particles production rates were generally higher in the Pacific Ocean experiments, resulting in a mean net daily TEP production of 44.3 ± 14.2 μg XG Eq. L^-1, than in the Atlantic Ocean experiments, which averaged 6.6 ± 1.1 μg XG Eq. L^-1. Furthermore, the TEP production rates (d^-1) in the Pacific Ocean experiments were significantly higher under the FULL solar radiation treatment (Figure 3A). By contrast, TEP production in the Atlantic Ocean did not differ significantly among treatments (Figure 3B).

Consistently, in all the experiments performed in the Pacific Ocean, the populations of Prochlorococcus sp. were negatively affected by solar radiation (p < 0.05), resulting in the decay of the cell population over time (Figure 4), without significant differences in the decay rates between FULL (mean -0.06 ± 0.01 h^-1) and -UV treatments (mean -0.07 ± 0.009 h^-1), but with significantly higher cell decay rates in comparison with the DARK treatments (Figure 4; mean -0.01 ± 0.002 h^-1; t-test: DARK-FULL t = 0.058, df = 7, p < 0.001; t-test: DARK-UV t = 0.048, df = 7, p < 0.005). The most pronounced decay in cell concentrations during the sampling time was observed during Exp 2. Dark treatments provided a control for other sources of cell mortality than light, including grazing by small ciliates and protists (passing through a net of 50 μm mesh) present in the experimental communities. Prochlorococcus decay rates in the dark treatments in the Pacific experiments were low, significantly lower than those in the light treatments. However, in the Atlantic, the differences between dark and light treatments were not significant (p > 0.05) and may reflect losses due to other processes. In contrast to Prochlorococcus, the Synechococcus population exhibited low cell decay rates, which were not significantly different (p > 0.05) among treatments (Figure 5). Heterotrophic bacterial cell abundance did not decline significantly across the different treatments and experiments in both Pacific and Atlantic experiments. Overall, picoplankton populations from the Atlantic experiments did not exhibit significant decay rates under light exposure treatments except for Exp. 5, which showed the highest decay rates (Figure 5). Cell decay of Prochlorococcus in the Pacific (mean -0.047 ± 0.009 h^-1) was significantly higher than in the Atlantic Ocean (mean -0.01 ± 0.005 h^-1; t-test: t = 0.036, df = 19, p < 0.001) and significantly higher than that of Synechococcus in the Pacific Ocean (mean -0.004 ± 0.002 h^-1; t-test: t = 0.043, df = 23, p < 0.001). However, Prochlorococcus and Synechococcus decay rates were not significantly different in the Atlantic Ocean (Figure 5).

We did not obtain significant correlations between Chl a concentration and bacterioplankton, Synechococcus and TEP concentrations or between Chl a concentration and TEP production rates (α). However, the TEP production rate (d^-1) increased significantly with increasing Prochlorococcus sp. decay rates (R² = 0.41, p < 0.0001). This relationship improved when only the Pacific Ocean experiments were considered (α TEP (d^-1) = 0.047-18.04^∗Prochlorococcus decay (h^-1), R² = 0.60, p < 0.001) in which the Prochlorococcus cell decay accounted for a larger percentage of the variability (Figure 6).

Prochlorococcus marinus in the cultures grew at a moderate rate requiring 40 days to reach the stationary phase (Figure 7). TEP concentrations varied highly, increasing with time as P. marinus cell abundance increased (Figure 7A). The highest concentration of TEP was observed before reaching the stationary phase and remained high during the following days (Figure 7A and Table 2). TEP production rates in the cultures were low (Figure 7B) in comparison with the production rates in the field (Figure 3). There was no net production of TEP during the first 3 weeks of the culture although P. marinus was growing actively (Figure 7B). When growth rates of P. marinus declined, TEP production still remained at the highest values (Figure 7B). When P. marinus experienced negative growth the production of TEP stopped, although the TEP concentration remained high in the culture (Figure 7).