The Coorong is a shallow lagoon in South Australia, parallel to the coast and separated from the open ocean by a network of sand dunes (Fig. 1). This coastal lagoon forms the Murray Mouth with the lower lakes (Lake Alexandrina and Lake Albert), which is the terminal lake system of the River Murray 25 . More than 140 km in length, the Coorong is characterised by a strong salinity gradient with salinity values ranging from ca. 2.0% close to the Murray Mouth to more than 15.0% near Salt Creek (Fig. 1). As the saline waters of the Coorong receive inputs from the ocean through the Murray Mouth and from underground, the concentration of salts along the lagoon is similar to seawater. The saline waters of the lagoon are separated from the lower lakes by a series of barrages. The freshwater inputs through the barrages lead to lower salinities in the North-West part of the Coorong, whereas the excess in evaporation over precipitation increases salinity along its North-South axis, especially during the summer period characterised by (i) low water levels (ranging from 0.5 m near the Murray Mouth to 0.9 m in the southern part of the lagoon) and (ii) weak tidal impact 25 .

Samples were collected at 20 locations along the lagoon, from the brackish waters near Goolwa (salinity = 1.8%) to the hypersaline waters near Salt Creek (salinity = 15.5%; Fig. 1) on February 3-4, 2007. At each sampling site, referred to as S_i with i = 1 to 20 (Fig. 1), measurements and water sample collection were performed in 50 cm of water (a depth representative of most parts of the lagoon) from (i) the sub-surface waters and (ii) at the water-sediment interface (WSI). Temperature (°C), conductivity (mS cm^-1) and dissolved oxygen concentrations (DO; mg l^-1) were recorded using a YSI 85 (Fondriest) multiparameter probe. Salinity (%) was calculated from temperature and conductivity following Fofonoff and Millard 26 . 1 liter water samples were collected at each depth using acid-washed 1-liter borosilicate bottles with special care to avoid sediment resuspension.

Dissolved inorganic nutrient concentrations were determined from 12 ml filtered (Whatman GF/C) water samples. Analyses were performed in the field using a portable LF 2400 photometer (Aquaspex^®) according to standard colorimetric methods for (Indophenol blue), (Naphtylethylene diamine), (Naphtylethylene diamine after zinc reduction) and (Ascorbic acid reduction). Ammonium, nitrite, nitrate and phosphate measures ranged from 0.6 to 110 μM, 0.2 to 160 μM, 1.6 to 160 μM and 1.1 to 50 μM, respectively. Samples (50 to 100 ml) for suspended particulate material concentration (SPM; mg l^-1) were filtered through pre-combusted (400°C; 4 hours) and pre-weighted glass-fibre filters (Whatman GF/C; pore size = 1.2 μm), and immediately deep frozen in liquid nitrogen until analysis. In the laboratory, filters were rinsed with MilliQ water, dried at 60°C for 24 h, and reweighed to determine the mass of suspended solid retained on the filter 27 .

Two distinct set of samples (3 × 1 ml) were collected for the identification and enumeration of virus-like particles and picophytoplankton populations by flow cytometry.

Virus-like particles (VLP) were identified through flow cytometry to assess the presence of phytoplankton phages that could potentially infect phytoplankton. These phytoplankton phages can be discriminated from the other groups of virus by their higher side scatter (related to their size) and/or green fluorescence (related to their DNA content) as previously shown in recent works 28 29 . Samples were collected in triplicate (1 ml) at each sampling station, fixed with 0.5% (final concentration) glutaraldehyde in the dark at 4°C for 15 min, quick frozen in liquid nitrogen and then stored at -80°C until analysis. After being quick thawed, samples were diluted (1:10) in 0.2 μm filtered TE Buffer stained with SYBR Green I solution (1:5000 dilution) and incubated at 80°C in the dark for 10 min 30 .

Phytoplankton biomass was estimated through chlorophyll a (Chl a) concentrations. Samples (50 to 100 ml) were filtered through glass-fibre filters (Whatman GF/C) and immediately deep frozen in liquid nitrogen until analysis. Chlorophyllous pigments were then extracted in 5 ml of methanol in the dark at 4°C during 24 h 31 . Concentration of Chl a (μg l^-1) was determined following Strickland and Parson 32 using a Turner 450 fluorometer previously calibrated with a pure Chlorophyll a solution (Anacystis nidulans extract, Sigma Chemicals, St Louis).

Photosynthetic picophytoplankton populations were identified and enumerated by flow cytometry (FCM) using a FACScanto flow cytometer (Becton-Dickinson) equipped with an air-cool argon laser (15 mW, 488 nm) with phosphate buffer saline (PBS) solution employed as a sheath fluid. Water samples (1 ml) were fixed in the field with 2% (final concentration) of paraformaldehyde, immediately deep frozen in liquid nitrogen and then stored at -80°C.

After being quick thawed, picoplankton cells were discriminated and enumerated by FCM according to their specific auto-fluorescence and light scatter properties 33 34 . Forward-angle light scatters (FSC), right-angle light scatter (SSC), red and orange fluorescence, were recorded for each sample. As the values of FSC are those most affected by density differences between the sheath fluid and the samples 21 , the values of FSC were not used to enumerate cells. Fluorescent beads 1 μm in diameter (Molecular Probes, Eugene, Oregon) were added to all samples as an internal standard. Working beads concentrations were estimated after each FCM session under epifluorescent microscopy to ensure reliability of the beads concentration and all FCM parameters were normalized to bead concentration and fluorescence. Finally, picophytoplankton populations were identified and enumerated using WinMDI 2.9 (^©Joseph Trotter) flow cytometry analysis software. Synechococcus sp., Prochlorococcus sp. and autotrophic pico-eukaryotic cells were discriminated in plots of SSC versus orange fluorescence (from phycoerythrin) and red fluorescence (from chlorophyll), according to standards protocols 33 34 . Synechococcus and Prochlorococcus cells can be discriminated by their flow cytometry scatter signal (SSC) related to their size, and their fluorescence emission when excited by a blue light. The phycobilins contained in Synechococcus emit a strong orange fluorescence, whereas Prochlorococcus harvest light mainly through chlorophyll a and b, and therefore emit only red fluorescence when excited by blue light 35 . In addition, Synechococcus cells are larger than Prochlorococcus cells (ca. 1-0.6 μm in diameter respectively) 4 . Pico-eukaryotes were identified by their larger size (SSC) and higher red fluorescence. Because, in the absence of genetic fingerprinting the identification of Synechococcus and Prochlorococcus cannot be warranted sensu stricto, the populations exhibiting the flow cytometric signatures of Synechococcus and Prochlorococcus as reported in the literature were referred to as Synechococcus-like and Prochlorococcus-like populations.

Comparisons between the two sampling depths were conducted using the Wilcoxon-Mann-Whitney U-test (U-test hereafter). The BIOENV 36 and BVSTEP 37 procedures (PRIMER version 6.0) were used to investigate relationships between environmental variables and picophytoplankton community's composition along the salinity gradient. Both analysis compare rank correlation between the matrices of environmental variables (based on normalised Euclidian distance) and the biotic similarity matrix of picophytoplankton variables (based on the Bray-Curtis similarity) using different permutations of the environmental variables. BIOENV compares different combinations of a specified number of variables, whereas BVSTEP uses a stepwise procedure to identify the best subset of variables. Spearman rank correlations between the biotic and abiotic similarity matrices were used to identify the best suites of environmental variables that best explained the distribution of picophytoplankton communities along the salinity gradient and the significance of the correlation was determined using a permutation procedure 38 . Environmental variables considered in the BIOENV/BVSTEP analysis were salinity, [DO], [SPM], [ ], [ + ], [ ] and VLP3 abundances. As temperature variability between stations was mainly related to the time of the day when the sampling occurred, this parameter was not considered in the analyses. Similarities between stations for picophytoplankton communities along the salinity gradient were inferred through a cluster analysis (i.e. hierarchical agglomeration using complete linkage cluster analysis performed on Euclidian distances) performed on the log (abundance + 1) data matrix 39 . This analysis was performed with STATISTICA version 8 software.

Water temperature ranged between 25.2°C and 27.7°C. Salinity increased from 1.77% in S₁ to 15.48% in S₂₀ (sampling site, referred to as S_i with i = 1 to 20, see Fig. 1, Table 1). At stations S₁ and S₂, salinity levels remained below 2.50%. Salinity then slowly increased from 2.75% to 5.03% between stations S₃ and S₁₁. In contrast, salinity sharply increased from station S₁₂ to reach 15.0% at station S₁₇ (Table 1).

Ammonium ([ ]) was by far the most abundant form of nitrogen and represented more than 80% of the total inorganic nitrogen pool throughout the salinity gradient (Table 1). [ ] concentrations increased with salinity and highest concentrations (i.e. > 110 μM) were observed from 13.35% (S16; Table 1). Phosphate concentrations [ ]) were relatively high below 3.32% (i.e. from S₁ to S₆) with values ranging between 20 to 40 μM, and decrease thereafter to reach very low levels (i.e. <4 μM) between 3.35% and 6.75% (i.e. from S₇ to S₁₂). Relatively high concentrations (i.e. 9 to 16 μM) were observed again from 8.30% to 13.35% (i.e. S₁₃ to S₁₆) and decreased thereafter to reach very low level (i.e. < 4 μM) in the hypersaline waters of the lagoon (i.e. salinity > 14.0%; Table 1).

Dissolved oxygen (DO) concentrations ranged between 1.4 and 5.5 mg l^-1 along the salinity gradient (Fig. 2A). DO concentrations remained below 3.0 mg l^-1 in the brackish (i.e. salinity < 2.5%) and hypersaline waters of the lagoon (i.e. salinity > 13%). Highest DO values, ranging between 4.2 and 5.5 mg l^-1, were observed between 2.74% and 10.72% (Fig. 2A). Concentrations of suspended particulate matter (SPM) increased exponentially along the salinity gradient, with values increasing from 30 mg l^-1 at 1.77% to 967 mg l^-1 at 15.01% (Fig. 2B).

Three virus-like particles (VLP) populations were identified along the salinity gradient (Fig. 3). Details of the pattern of VLP populations along the salinity gradient are fully described in Schapira et al. 40 . Briefly, the two first populations (VLP1 and VLP2) exhibit the same cytometric signature (SSC and green fluorescence) of viral population observed previously in seawater and identified as bacteriophages 33 . In contrast, the third population (VLP3) exhibits the same SYBR Green fluorescence (related to DNA content) level as VLP2 but a higher side scatter (i.e. size; Fig. 3), hence could represent a group of phytoplankton viruses 28 29 . VLP3 was only observed from 5.03% (S₁₁) to 10.00% (S₁₄) with abundances ranging between 1.9 × 10⁶ and 4.3 × 10⁶ VLP ml^-1.

Chlorophyll a (Chl a) concentration increased from 0.4 μg l^-1 to 14.1 μg l^-1 with salinity increasing from 1.77% to 13.36% (Fig. 4A). Chl a concentrations sharply decreased thereafter, for salinity greater than 14.0%, and remained below 4.5 μg l^-1 in the hypersaline part of the lagoon (Fig. 4A).

Picophytoplankton abundances were highly variable along the salinity gradient with values ranging between 8.3 × 10³ cells ml^-1 and 1.4 × 10⁶ cells ml^-1 (Fig. 4B). Concentrations were relatively low (i.e. ≤ 6.3 × 10⁵ cells ml^-1) where salinity remained below 7.0% (Fig. 4B). High abundances were observed for salinity between 8.0% and 11.0% with values ranging from 1.3 × 10⁶ cells ml^-1 to 1.4 × 10⁶ cells ml^-1 (Fig. 4B). Above 13.0% picophytoplankton abundance was very low with values remaining below 8.0 × 10⁴ cells ml^-1 (Fig. 4B). Samples were characterised by a highly complex community structure with multiple subpopulations of picophytoplankton throughout the salinity gradient. Flow cytometry analysis (FCM) revealed 2 distinct populations of pico-cyanobacteria (P-Cya) exhibiting fluorescence and side-scatter characteristics of Prochlorococcus sp. and Synechococcus sp. (referred hereafter as Prochlorococcus-like and Synechococcus-like), and 5 different populations of pico-eukaryotes (P-Eu), exhibiting consistently different side scatter (related to size) and red fluorescence. These different subpopulations were identified according to their differences in size and both red and orange fluorescence (Fig. 5). 1 μm beads were added in all occasion during flow cytometry run as a reference of size and the largest pico-eukaryote population was only slightly bigger than the beads (i.e. 1 μm).

Except in locations where salinity remained below 3.0%, P-Eu were by far the most abundant and contribute to more than 54% of the total abundances throughout the salinity gradient (Fig. 6A). Furthermore, the picophytoplankton community was only composed of P-Eu for salinities ranging from 4.5% to 14.0% (Fig. 6A). Below 3.0% (S₁ to S₃) P-Cya were much more abundant, contributing from 48% (S₂) to 81% (S₃) of the total community, with Synechococcus-like and Prochlorococcus-like populations respectively representing 23% to 34% and 22 to 47% of the total abundances (Fig. 6A). The Synechococcus-like population was observed from 1.7% to 4.5% (S₁ to S₉). In contrast, the Prochlorococcus-like population was observed for salinity ranging from 1.7% to 3.5% (S₁ to S₇) and for salinity greater than 14.0% (S₁₇ to S₂₀; Fig. 5, 6A). Five different sub-populations of P-Eu were discriminated (Fig. 6B). These populations consistently exhibited different side scatter and red fluorescence signatures (Fig. 5). P-Eu ₁ was by far the most abundant and contributed to more than 70% of the P-Eu abundances (Fig. 6B). P-Eu ₂ was the second most abundant population with relative contribution to the total P-Eu only occasionally exceeding 23% (Fig. 6B). Whilst P-Eu ₁ was identified throughout the salinity gradient P-Eu ₂ was not observed for salinity greater than 14.0%; P-Eu ₁ was also the only sub-population observed in these hypersaline waters of the lagoon (Fig. 6B). The relative abundances of populations P-Eu ₃, P-Eu ₄ and P-Eu ₅ remained below 2% and were locally observed mostly for salinity ranging from 3.2 and 3.5% (Fig. 6B). The cluster analysis performed on picophytoplankton abundance discriminated 2 main groups of stations based on their population richness (defined here as the cytometrically-defined richness, i.e. FCM richness): a high FCM richness group (i.e. richness ≥ 4) occurred for salinity lower than 3.5% (S₁ to S₇), and a low FCM richness group (i.e. richness < 4) included stations where salinity was greater than 3.5% (Fig. 7). In the later, two sub-groups of stations were identified: a sub-group where P-Eu (P-Eu ₁, P-Eu ₂ and P-Eu ₃) contributed to more than 93% to the total abundances occurring for salinity ranging between 3.5% and 14.0% (S₈ to S₁₆), and a sub-group characterized by a community composed exclusively by the P-Eu sub- population P-Eu ₁ and Prochlorococcus-like picocyanobacteria comprising stations where salinity was greater than 14.0% (S₁₇ to S₂₀; Fig. 7).

Multivariate correlations analysis, BIOENV showed that the environmental variable that best explained the picophytoplankton abundance pattern along the lagoon was salinity (ρ = 0.542; P < 0.01; Table 2). Salinity in combination with ammonium concentrations and abundance of viral population (VLP3) are the best subset of variables explaining the variability of picophytoplankton abundances observed along the salinity gradient (BVSTEP; Table 2). The Spearman's rank correlation coefficient for this analysis was 0.520 and was statistically significant (P < 0.01; Table 2).

Indicators of autotrophic biomass, including chlorophyll a concentrations and picophytoplankton cell numbers exhibited maxima at salinities ranging between 11.0-13.0% and 8.0% and 11.0% respectively (Fig. 4). This distribution of autotrophic biomass along the lagoon is consistent with previous observations conducted in solar salterns over comparable salinity ranges 21 22 41 42 . The maximal chlorophyll a concentration in the present work (i.e. 14.1 μg l^-1) was, however, well above values observed in these semi-artificial systems, which typically never exceeded 8 μg l^-1 21 41 42 . In addition, maximal picophytoplankton abundance observed in the lagoon was 1.4 × 10⁶ cells ml^-1. This is nearly one order of magnitude higher than, the maximum abundances estimated by flow cytometry which did not exceed 3.5 × 10⁵ cells ml^-1 at comparable salinities in the solar salterns of Bras del Port (Spain) 21 . These observations highlight the unique properties of the autotrophic communities found along a strong continuous salinity gradient, when compared to previous studies carried out in solar salterns.

An increase in primary producers' biomass was observed from salinity higher than 7.0% (Fig. 4) where dissolved oxygen concentrations were relatively high (Fig. 2A). This may be indicative of an enhancement of primary production in this part of the lagoon as previously observed under comparable salinity range in solar salterns 42 . An increase in primary production may be the result of a change in nutrient availability along the salinity gradient. This is consistent with the results of the BVSTEP analysis which highlighted a strong relationship between picophytoplankton abundances and ammonium concentrations (Table 2). The high abundances observed from salinity higher than 8.0% could be explained by the high ammonium concentrations found in the same part of the lagoon (Table 1). This is congruent with previous work showing the ability of picophytoplankton to efficiently utilize regenerated forms of nitrogen such as ammonium and urea 4 43 . A modification of the light regime along the salinity gradient may also have impacted the pattern of autotrophic organisms observed in the present study. The increase in suspended matter along the lagoon (Fig. 2B) is indeed likely to impact turbidity leading to a decline in light penetration in the water column in the high salinity area. However, the shallowness of the lagoon, as well as wind mixing and heating convection might prevent light limitation by ensuring sufficient turnover of the water column, as suggested by the absence of vertical stratification observed during our sampling.

Microzooplankton grazing, could also contribute to the observed pattern of autotrophic biomass along the salinity gradient. More specifically, considering the importance of microzooplankton grazing as a source of nutrients recycling in planktonic systems 44 and the high level of ammonium concentrations observed for salinity greater than 11.0% (Table 1), the sharp decrease in picophytoplankton abundances observed in the same area (Fig. 4A) may then be the result of an increase in grazing pressure. This observation is congruent with previous works reporting high microzooplankton grazing rates for salinity higher than 4.0% 42 as well as high abundances of heterotrophic nano-flagellates up to the highest salinities (i.e. > 30.0%) 45 46 . The decrease in chlorophyll a concentrations, representing the size fraction >1.2 μm, observed for salinity higher than 13.0% may also be the result of an increase in grazing pressure by large metazoans consumers, such as the brine shrimps (i.e. Artemia sp.) that were very abundant during the sampling experiment (i.e. 20-50 ind l^-1; Seuront, unpublished data) and known to survive up to the highest salinities 41 47 . Moreover, picophytoplankton population growth is tightly controlled by fast growing protozoans consumers under high nutrients conditions 48 49 whereas larger cells are temporally and/or locally, protected from predation by slow-growing Metazoa under the same conditions 50 51 . Therefore the decrease in picophytoplankton abundances and phytoplankton biomass (size fraction>1.2 μm) observed in different part of the lagoon could be explained by a difference in size-related grazing rate along the salinity gradient (Fig. 4). However, further work is needed to fully assess the role of grazing on the distribution of autotrophic organisms along the salinity gradient.

The peak in picophytoplankton abundance, observed for salinity ranging between 8.0% and 11.0%, was concomitant to the occurrence of the viral population VLP3 in the water column (from 5.0 to 10.0%). This viral population, exhibiting the typical cytometric signature of phytoplankton virus, was only observed in this narrow range of salinity. Considering the strong relationship between virus and their potential hosts, this result suggests a positive correlation between picophytoplankton populations and VLP3. This hypothesis is supported by the result of the BVSTEP analysis (Table 2). In addition, the abundances of VLP3 recorded in the lagoon (1.9 × 10⁶ - 4.3 × 10⁶ VLP ml^-1), were in the highest range of concentrations previously reported in marine waters, i.e. >10⁵ VLP ml^-1 13 . This observation suggests an important contribution of viral lysis to picophytoplankton losses in this part of the lagoon. However, as chlorophyll concentrations were also relatively high over this salinity range (8 to 11%), VLP3 could also be infecting the larger fraction (i.e. > 1.2 μm) of the phytoplankton community. While this was beyond the objectives of the present investigation, further work will be needed to specifically assess which fraction of the phytoplankton community was infected by VLP3. Viruses infecting both components of the picophytoplankton community (i.e. cyanobacteria and eukaryotes) have been previously reported 13 52 53 54 and the role of viral lysis on picophytoplankton mortality is now well established [e.g. 13 14 ]. However, the importance of phytoplankton viruses under this high salinity conditions has barely been investigated and further work is needed to confirm the identity of this viral population and evaluate the role of phytoplankton viruses along the salinity gradient.

The picophytoplankton cytometric-richness decreased along the salinity gradient, affecting both prokaryotic and eukaryotic picophytoplankton (Fig. 7). The existence of a decreasing trend in the number of phytoplankton species with increasing salinity has previously been observed in solar salterns ponds 21 22 23 41 42 and hypersaline lakes 24 55 . More specifically, a decrease in picophytoplankton cytometric-richness with increasing salinity has been reported in Bras del Port salterns 21 .

The most diversified community was observed for salinity lower than 3.5% (Fig. 6). In this habitat, Prochlorococcus and Synechococcus-like populations were abundant and 5 distinct populations of pico-eukaryotes were identified. This high cytometric richness coincided with relatively low total abundance (Fig. 4B). Favourable environmental conditions may have led to the establishment of highly diversified picophytoplankton community in this brackish-marine part of lagoon. In contrast, the number of cytometrically-defined populations was limited under higher salinity conditions (Fig. 6) where the highest abundances were observed (Fig. 4B). More specifically, the peak in picophytoplankton abundances observed for salinity ranging between 8.0% and 11.0% was largely dominated by the pico-eukaryotes P-Eu ₁ (Fig. 6). This is consistent with previous works reporting the dominance of one population of pico- eukaryote under the same salinity range in solar salterns 21 22 23 and hypersaline lakes 17 24 . The existence of a bloom of pico-eukaryotes, observed in such different saline systems suggests that salinity may be the main factor triggering the dominance of pico-eukaryotes over this particular salinity range. This hypothesis is supported by the results of the BIOENV/BVSTEP analysis (Table 2). The pico-eukaryote P-Eu ₁ may be dominant through a higher tolerance to high salinity and the subsequent decrease in competition within the reduced picoplankton community, may have allowed this salinity-tolerant population P- Eu ₁ to grow extensively and flourish in this part of the lagoon. The collapse of the pico- eukaryote bloom observed for salinity greater than 11.0% was followed by an increase in Prochlorococcus sp. concentration which contributed to more than 40% of the total abundance for salinity greater than 15.0% (Fig. 5 and 6). This observation is consistent with many studies highlighting the abundance of cyanobacteria under extreme saline conditions [e.g. 16 ]. It is also stressed that this is the first report of a Prochlorococcus-like population in such highly saline habitat. However, further work is therefore needed to confirm the identity of this population

The succession of picophytoplankton was only defined in the present work through their flow cytometric signature. Even if flow cytometry is a powerful tool to investigate the composition of picophytoplankton organisms, further work is needed to identify the species succeeding along the salinity gradient, especially considering the diversity existing among cyanobacteria and pico-eukaryotes 2 . However, our results provide new insight into the effect of salinity on picophytoplankton communities.

Salinity has been identified as the main factor triggering the succession of pico-autotrophs along the salinity gradient (BIOENV/BVSTEP analyses, Table 2). Salinity could act directly on picophytoplankton assemblages by selecting groups adapted to life at a particular salt concentration. Cyanobacteria are known to tolerate and acclimate to high salt concentrations 56 . However the different groups of cyanobacteria do not exhibit the same tolerance to salinity stress and have been consequently classified into 3 groups, i.e. stenohaline, halotolerant and extremely halotolerant 57 . The decrease in cyanobacteria abundances for salinity higher than 3.5%, suggests that the populations inhabiting the brackish-marine part of the Coorong may belong to the stenohaline group with a salinity tolerance range characteristic of estuarine and marine populations. Synechococcus species are known to be abundant in transitional and freshwater areas 58 whereas Prochlorococcus species are thought to be restricted to marine waters 59 . However, the observations of Prochlorococcus-like populations in the Rhône River 60 , in the Changjiang river estuaries 18 and in the present work in the low-salinity part of the Coorong tend to challenge this hypothesis.

In contrast, the occurrence and predominance of the pico-eukaryote P-Eu ₁ throughout the salinity gradient, suggest that this population may represent halotolerant organisms. The dominance of halotolerant pico-eukaryotes has been previously described for the same salinity range 17 23 24 . In particular, eukaryotic pico-autotrophs have been shown to be responsible for dense blooms in surface water of Mono Lake in California where salinity was around 8.5% 24 . The organism isolated from this lake and identified as a Picocystis spp., has been shown to exhibit high growth efficiency over the 0.2-15.0% salinity range, which is consistent with the range of salinity where the population P-Eu ₁ was found in the present work. However, this population was identified here through flow cytometry and further work is needed to unambiguously conclude on the identity of this population.

Independent of their intrinsic salinity tolerance, the succession of picophytoplankton organisms along the salinity gradient could be indirectly controlled by salinity. In particular, by controlling the diversity and abundance of microzooplankton and VLP, salinity could exert a control on the top-down processes including grazing and viral lysis. The presence of a large population of phytoplankton virus up to the highest salinity observed in the present work highlights the necessity to investigate the role of viral infection in regulating the community structure of picophytoplankton along this salinity gradient.