Seasonal variability and vertical distribution of autotrophic and heterotrophic picoplankton in the Central Red Sea

Sample collection and environmental properties

Periodic samplings were conducted from January 2015 to May 2017 on board of RV Thuwal at a mesopelagic station (ca. 700 m depth) located in the central Red Sea, 6 km off the coast of King Abdullah Economic City (KAEC) in Saudi Arabia. Sailing permission were approved by Saudi Coast Guard. We performed 15 vertical profiles evenly distributed along the four seasons (only winter had 3 samples rather than 4, Table 1). Samples were taken at regular depths from the surface to the bottom: 5, 20, 40–80 targeting the deep chlorophyll maximum (DCM), 100, 200, 300, 400, 550, 600 and 700 m. Temperature, salinity, fluorescence and photosynthetically active radiation (PAR) data were acquired with SeaBird SB9 Plus or IDRONAUT 305 CTDs. PAR was available for only 7 sampling times. The depth of the photic layer was determined by the vertical light attenuation coefficient (Kd) as the depth receiving 1% of surface irradiance (Calvo-Díaz & Morán, 2006). Stratification index (SI) was calculated as the density at 100 m minus that at the surface (Calvo-Díaz & Morán, 2006). The upper mixed layer (UML) depth was determined as the first depth in which the difference in density with the shallower 5 m was ≥ 0.05 kg m⁻³ (Calvo-Díaz & Morán, 2006).

Water samples were taken from Niskin bottles in a rosette sampler with an attached CTD probe (Fig. S1). In 2015 total chlorophyll a concentration (Chl a) was obtained after filtering 500 to 2,000 ml of the sample through Whatman GF/F filters (25 mm diameter). After checking for the minimum volume yielding reliable results, in 2016 and 2017 we performed sequential filtration of 200 ml samples through filters of 20, 2 and 0.2 µm of pore-size (IsoporeTM Membrane Filters, RTTP, 47 mm diameter), so that Chl a was the sum of the corresponding size-fractions: micro- (above 20 µm), nano- (between 2 and 20) and picophytoplankton (between 0.2 and 2 µm). Filters were frozen at −80 °C until analysis in the laboratory. Pigments were extracted in 90% acetone for 24 h in the dark at 4 °C and chlorophyll a fluorescence was measured with a Trilogy fluorometer (Turner) calibrated with pure extracts.

Samples for dissolved inorganic nitrogen (DIN = NO³⁻ + NO²⁻), dissolved inorganic phosphorus (DIP = PO₄³⁻), dissolved organic carbon (DOC) and total dissolved nitrogen (TDN) were filtered through pre-combusted GF/F filters and analyzed as previously reported by Calleja et al. (2018). The nutricline depth was defined as the depth where nitrate concentration first reached 1 µmol L⁻¹ (Calvo-Díaz & Morán, 2006).

Analysis of picoplankton by flow cytometry

Picoplankton samples (1.8 mL) were preserved with 1% paraformaldehyde + 0.05% glutaraldehyde final concentration and placed in the dark for approximately 10 min, then frozen in liquid nitrogen and stored at −80 °C once in the laboratory. After thawing, samples were analyzed with a FACSCanto II flow cytometer (BD-Biosciences). Molecular Probes fluorescent latex beads of 1 µm were used as an internal standard for size and fluorescence measurements. We analyzed aliquots of 0.6 mL for autotrophs and 0.4 mL for heterotrophs, at high (mean 117.9 µL min⁻¹) and low (17.9 µL min⁻¹) flow rates, respectively, until acquiring 10,000 events. Before analysis, heterotrophic bacteria were stained with 2.5 µmol L⁻¹ of the DNA fluorochrome SYBR Green II (Gasol & Morán, 2015). All cytograms were analyzed with FCSExpress 5 software. Autotrophic prokaryotic cells were classified as cyanobacteria (Synechococcus and Prochlorococcus) and picoeukaryotes according to their orange (PE, 433 nm) and red (PerCP-Cy5-5, 498 nm) fluorescence and light scatter at 90° or side scatter (SSC) signals. Two groups of heterotrophic prokaryotes were distinguished based on their relative green fluorescence (FITC, 360 nm) signal: low and high nucleic acid content (LNA and HNA, respectively). Cell size was determined by an empirical calibration between relative SSC and cell diameter according to Calvo-Díaz & Morán (2006). Spherical shape was assumed for all groups for estimating biovolume, which was transformed into individual biomass by using the biovolume-to-carbon conversion factor of 237 fg C µm⁻³ for autotrophs (Worden, Nolan & Palenik, 2004) and the equation biomass = 108. 8 × (biovolume)^0.898 for heterotrophs (Gundersen et al., 2002). The biomass of each picoplanktonic group was finally obtained by multiplying the individual biomass estimate by the corresponding abundance.

Statistical analyses

Picoplankton abundance, biovolume and biomass data were log10-transformed to attain normality and assess their relationship with environmental variables by Spearman’s rank correlation coefficient. One-way ANOVAs and post hoc Tukey’s pairwise comparisons were used to determine significant variations between seasons (P < 0.05) with OriginPro software. A non-metric multidimensional scaling (NMDS), a distance-based ordination technique, was performed on the Bray-Curtis dissimilarity distances together with pairwise PERMANOVAs in order to summarize the seasonal and vertical changes in the abundance of the different picoplankton groups and their relation with environmental variables in the upper epipelagic zone. Four groups of samples were considered according to depth: surface, above DCM, DCM depth, below DCM and 100 m. NMDS stress values, a measure of goodness-of-fit, can be used to evaluate the proper choice of dimensions. Low values (0.05–0.1) provide a good fit in reduced dimensions while values >0.3 indicate that the ordination is arbitrary and potentially uninterpretable (Ramette, 2007; Zhu & Yu, 2009). The NMDS analysis was done in R (http://www.r-project.org) and we used the “envfit” function in order to estimate the correlations between the environmental variables and the NMDS axis scores.

Vertical and seasonal variability in hydrographic conditions

The mean vertical profiles of selected environmental variables for the four seasons are shown in Fig. 1. As expected, significant differences (ANOVA: F = 14.4, p = 0.0004, n = 15) in mean surface temperature were found, with summer values 6.3 °C higher than in winter (Table 2). The temperature remained constant year-round from 200 m down to the bottom at 21.7 ± 0.02 °C SE (Fig. 1A). Surface salinity displayed slight seasonal variations from 38.8 ± 0.2 in spring to 39.6 ± 0.1 in winter (Table 2), but there was no seasonal difference below 200 m (40.6 ± 0.0). Differences in SI were not significant despite some seasonality (Table 2), but the UML was significantly shallower in summer than in the other seasons (Table 2, ANOVA: F = 18.3, p = 0.0002, n = 14). The euphotic layer depth varied from 63 to 89 m, with similar mean values across seasons (Table 2).

Dissolved inorganic nitrogen (DIN = nitrate + nitrite) presented uniformly low concentrations at the surface (0.17 ± 0.11 µmol L⁻¹) but reached 20.5 ± 1.6 µmol L⁻¹ at depths higher than 200 m (Fig. 1C), with an average nutricline depth of 67 ± 6 m (Table 2). Dissolved inorganic phosphate (DIP) followed the same pattern as DIN, with low seasonal mean values at the surface (0.10 ± 0.04 µmol L⁻¹) and increasing with depth to a seasonal mean maximum of 1.17 ± 0.1 µmol L⁻¹at around 600 m depth (Fig. 1D). The concentration of dissolved organic carbon (DOC, Fig. 1E) declined with depth from a mean 76.1 ± 7.5 µmol L⁻¹ at the surface to 52.2 ± 5.8 µmol L⁻¹ below 200 m.

Total and size-fractionated chlorophyll a concentration

The vertical distribution of total chlorophyll a (Chl a) concentration (Fig. 1F) showed a consistent and clear deep chlorophyll maximum (DCM) located at an average depth of 56 ± 4 m (Table 2). Surface seasonal mean values ranged from 0.09 to 0.15 µg L⁻¹ (Fig. 1F). Mean integrated Chl a values for the upper 100 m increased from 16.1 mg m⁻²in spring and summer (±2.06 and 1.09, respectively) to 19.0 ± 2.9 mg m⁻² in winter and 24.0 ± 2.7 mg m⁻² in fall (Fig. 2A). The picoplankton size fraction contributed, on average, 70.8 ± 1.0% to total integrated values, with nanoplankton and microplankton making up 21.9 ± 1.5% and 7.3 ± 1.9%, respectively, with no significant differences in the relative contributions of the three size classes (Fig. 2B).

Vertical distribution of picoplankton abundance and cellular characteristics

Prochlorococcus, Synechococcus and picoeukaryotes were mostly restricted to the upper 100 m, with none of the groups detected in significant numbers at or below 150 m depth. Figure 3 shows the average vertical distribution of picophytoplankton abundance, cell size and relative red fluorescence (as a proxy of Chl a content) for each season. Prochlorococcus abundance was generally low at the surface (1.11–5.81 × 10⁴ cells mL⁻¹) and peaked at the DCM (1.32 ± 0.16 × 10⁵ cells mL⁻¹ in summer) (Fig. 3A). The two groups of Synechococcus discriminated by low (LF-Syn) and high (HF-Syn) phycoerythrin fluorescence were consistently less abundant than Prochlorococcus. LF-Syn and HF-Syn tended to show higher numbers in the surface layers, with averages of 2.29 ± 0.53 × 10⁴ and 3.47 ± 0.54 × 10⁴ cells mL⁻¹, respectively (Figs. 3B and 3C). HF-Syn reached deeper than LF-Syn with the latter virtually absent at 80 m (Figs. 3B and 3C). Two groups of picoeukaryotes according to size were consistently distinguished, hereafter referred to as Small (Speuk) and large (Lpeuk). Speuk vertical distribution was similar to that of Prochlorococcus (Fig. 3D), while Lpeuk usually disappeared deeper than 40–60 m except in fall, where the highest values were found in the DCM (Figs. 3D and 3E). Coincident with declining abundances, the biovolume of all groups increased steadily with depth from 40 m downwards except for Lpeuk (Figs. 3F–3J). Similar to biovolume, the relative red fluorescence increased consistently with depth, with less marked patterns for LF-Syn and Lpeuk due to their shallower distribution (Figs. 3K–3O).

The mean seasonal distribution of heterotrophic prokaryotes abundance and cell size with depth is shown in Fig. 4. The abundances of both LNA and HNA cells were highest in the upper 100 m (maxima of 2. 92 × 10⁵ and 2. 51 × 10⁵ cells mL ⁻¹a t 20 and 40 m, respectively) but remained relatively stable for the entire mesopelagic layer (Figs. 4A and 4B). LNA were more abundant than HNA cells in the upper epipelagic, resulting in a contribution of HNA cells to total abundance (%HNA) that ranged from 38.3 to 47.1% at the surface. Values increased to 52.3 –57.4% at 200 m and remained pretty homogeneous down to the sea floor (Fig. 4C). The biovolume of HNA cells was consistently larger than that of LNA cells throughout the water column. Differences were observed between seasons, with maxima in winter for both groups and minima in summer for HNA and in fall for LNA cells (Figs. 4D and 4E).

Seasonal variation of picoplankton abundance and cellular characteristics

Although some differences between seasons were already apparent in the vertical distributions (Figs. 3 and 4), depth-weighted averages from the surface to 100 m were calculated to better capture the seasonal changes. Prochlorococcus abundance displayed a clear seasonal pattern, with minimum values in winter (1.4 × 10⁴ cells mL⁻¹, January 2015) and maximum values in summer (7.8 × 10⁴ cells mL⁻¹, July 2015) (Fig. 5A). The two groups of Synechococcus shared similar dynamics, with maximum values in spring for LF-Syn (2.7 × 10⁴ cells mL⁻¹, March 2015) and winter for HF-Syn (3.2 × 10⁴ cells mL⁻¹, January 2015) while the lowest values were observed between late spring and early summer for both groups (Fig. 5B). Consequently, the ratio between Prochlorococcus and Synechococcus abundance, which was higher than 1 for most of the year, was occasionally lower in winter and at the end of fall (Fig. 5C). Speuk abundance presented low values in summer (9.9 × 10² cells mL⁻¹, July 2015) and higher in winter (3.60 × 10³ cells mL⁻¹, January 2015) while Lpeuk, generally less abundant, peaked in fall (7.1 × 10² cells mL⁻¹, October 2016) (Fig. 5D). On an annual basis, Prochlorococcus contributed 57.6 ± 4.2% to total picophytoplankton cell numbers, followed by Synechococcus (38.9 ± 3.9%) with picoeukaryotes 21- to 79-fold lower abundances than cyanobacteria. Depth-weighted biovolumes (0.14–0.25 µm³ Prochlorococcus, 0.08–0.19 µm³ LF-Syn, 0.27–0.42 µm³ HF-Syn, 1.09–1.66 µm³ Speuk, 2.50–3.62 µm³ Lpeuk) did not show any clear seasonal pattern with slightly increased values of Speuk in early summer and Lpeuk in fall (Figs. S2A–S2C). The seasonality of 0–100 m mean relative red fluorescence as a proxy for Chl a content followed the expected summer minimum only for Prochlorococcus and LF-Syn (Figs. S2D–S2F). Differences in biovolume affected little the changes mentioned above in abundance when calculating the biomass of the different picophytoplankton groups. Integrated autotrophic picoplankton biomass for the upper 100 m showed higher values in summer (387.4 mg C m⁻²) with a significant contribution of Prochlorococcus (46.6 ± 6%) (ANOVA: F = 4.2, p = 0.03, n = 15) except in winter when Synechococcus contributed 49.5% with a high contribution of HF-Syn (37.02%).

The mean total abundance of heterotrophic prokaryotes (HNA + LNA prokaryotes) in the upper 100 m ranged from 2.29 to 4.21 × 10⁵ cells mL⁻¹, with higher values in spring and fall and lower in summer (Dataset S1). Figure 6A shows the corresponding values for the LNA and HNA groups with respective annual means of 1.87 ± 0.01 × 10⁵ and 1.38 ± 0.07 × 10⁵ cells mL⁻¹. Although their abundances failed to show marked seasonal patterns, a clear seasonality in the contribution of HNA bacteria emerged. Upper epipelagic-averaged %HNA values ranged from 35.9% in late spring and early summer to ca. 50% in winter and fall (Fig. 6B). In contrast, the seasonality in biovolume and relative nucleic acid content was not clear for any of the two groups (Figs. S3A and S3B). The integrated biomass of heterotrophic prokaryotes in the upper 100 m ranged from 86.9 to 257.6 mg C m⁻², with seasonal means shown in Fig. 7. Regarding the contribution of LNA and HNA cells to total heterotrophic prokaryotes biomass in the upper epipelagic, differences were minor with the HNA group prevailing in winter (53.8%) and fall (51.3%) and the LNA groups in summer (53.4%) and spring (52.3%) (Fig. 7). Overall, the biomass of autotrophic picoplankton groups was consistently higher than that of heterotrophic bacteria in the upper epipelagic, with annual averages of 348.1 ± 20.5 and 140.8 ± 10.1 mg C m⁻², respectively. However, when values were integrated over the entire water column (0–700 m), the mean biomass of heterotrophic bacteria increased to 410.7 ± 27.0 mg C m⁻², thus exceeding the total biomass of autotrophic picoplankton (Fig. 7). Moreover, when the entire water column was considered HNA cells clearly dominated total biomass regardless of the season.