Viability of pico- and nanophytoplankton in the Baltic Sea during spring
Phytoplankton cell death is an important process in marine food webs, but the viability of natural phytoplankton communities remains unexplored in many ecosystems. In this study, we measured the viability of natural pico- and nanophytoplankton communities in the central and southern parts of the Baltic Sea (55°21′ N, 17°06′ E–60°18′ N, 19°14′ E) during spring (4th–15th April 2016) to assess differences among phytoplankton groups and the potential relationship between cell death and temperature, and inorganic nutrient availability. Cell viability was determined by SYTOX Green cell staining and flow cytometry at a total of 27 stations representing differing hydrographic regimes. Three general groups of phytoplankton (picocyanobacteria, picoeukaryotes, and nanophytoplankton) were identified by cytometry using pigment fluorescence and light scatter characteristics. The picocyanobacteria and picoeukaryotes had significantly higher cell viability than the nanophytoplankton population at all depths throughout the study area. Viability correlated positively with the photosynthetic efficiency (Fv/Fm, maximum quantum yield of photosystem II) as measured on the total phytoplankton community. However, an anticipated correlation with dissolved organic carbon was not observed. We found that the abiotic factors suggested to affect phytoplankton viability in other marine ecosystems were not as important in the Baltic Sea, and other biotic processes, e.g. processes related to species succession could have a more pronounced role.
KeywordsBaltic Sea Spring bloom Phytoplankton viability Flow cytometry SYTOX Green
Grazing by zooplankton and sinking have traditionally been considered the main loss processes for phytoplankton populations. Cell death is a third loss factor, but its quantification in marine systems, and aquatic systems in general, remains rare compared to the quantification of sinking and grazing losses. Phytoplankton cell death can be caused by pathogens (Bramucci and Case 2019; Schieler et al. 2019) or physiological stress, and a handful of studies indicate that a considerable proportion of phytoplankton cells may not be viable (e.g. Brussaard et al. 1995; Veldhuis et al. 2001; Agustí 2004; Berman-Frank et al. 2004; Rychtecký et al. 2014). In addition to external factors, also cell-intrinsic factors (senescence) can result in reduced viability among phytoplankton (Veldhuis et al. 2001; Franklin et al. 2006; Bidle 2015). Microalgae can undergo programmed cell death under unfavourable environmental conditions (Berges and Falkowski 1998; Bidle and Falkowski 2004; Jiménez et al. 2009; Gallo et al. 2017). Cell death can also be induced by allelochemicals produced by other phytoplankton. For example polyunsaturated aldehydes (PUAs) produced by marine diatoms reduce growth and viability among other phytoplankton species (Casotti et al. 2005; Ribalet et al. 2007, 2014). Recently, it has been shown that also some nano- and picoplankton taxa produce PUAs (Vidoudez et al. 2011a; Morillo-García et al. 2014).
Although time-scales differ, phytoplankton cell death can result in cell lysis, thereby providing dissolved organic matter to the pelagic microbial food web (Franklin et al. 2006, Thornton 2014). The supply of DOM affects energy transfer to higher trophic levels, and therefore cell death can have an impact distinct from other population loss factors such as grazing and sinking. The way a phytoplankton cell dies thus influences the biogeochemical cycling of organic matter (Kirchman 1999). Dissolved organic carbon (DOC) is the largest reservoir of organic carbon in the ocean and plays an important role in marine ecosystems as the primary energy source for heterotrophic bacteria. DOC is therefore considered one of the main components of aquatic food webs (Packard et al. 2000; Gustafsson et al. 2014). In coastal environments, such as the Baltic Sea, DOC can have several origins, with riverine runoff often being a substantial source (Kuliński and Pempkowiak 2008; Hoikkala et al. 2015). Terrestrial DOC is mostly retained in river estuaries of the Baltic Sea and has its greatest influence on the coastal Bothnian Sea while the open-sea area of the western Gulf of Finland (GoF) and the Baltic Proper (BP) show primarily autochthonous origin of DOC (Hoikkala et al. 2015). Other minor DOC sources are sloppy feeding by phytoplankton grazers and DOC diffusion from faecal pellets (Lignell et al. 1993; Saba et al. 2011). Certain phytoplankton may also release excess dissolved organic material during growth or lose organic compounds passively into the surrounding water (Bjørrisen 1988, Thornton 2014).
Loss of phytoplankton cell viability can be caused by suboptimal trophic conditions, temperature and UV radiation (Berges and Falkowski 1998; Agustí and Duarte 2000; Agustí and Sánchez 2002; Agustí 2004; Llabrés and Agustí 2006). How phytoplankton viability is affected by varying abiotic stressors depends on the taxa; some phytoplankton have a wider tolerance range than other coexisting taxa (Alonso-Laita and Agustí 2006; Rychtecký et al. 2014). Such taxa could survive better in rapidly changing environments, whereas more sensitive taxa succumb to external stressors and show lower viability. The Baltic Sea is enriched with inorganic nutrients due to anthropogenic loading, and eutrophication is an ongoing process in most parts of the Baltic Sea (Fleming-Lehtinen et al. 2008). At the same time, climate change is causing structural and functional shifts in the communities of aquatic ecosystems (Li et al. 2009; Kahru et al. 2016), with potential implications for sedimentation (Tamelander et al. 2017) and biogeochemical cycles (Spilling et al. 2018) in the Baltic Sea. Patterns of group-specific phytoplankton cell viability may thus be changing as viability is affected by both abiotic factors such as inorganic nutrient availability and temperature, as well as a host of biotic factors.
Phytoplankton cell viability is poorly understood in marine environments and has not been investigated in the Baltic Sea. The aim of this study was to increase our understanding of how pico- and nanophytoplankton viability varies spatially in the Baltic Sea and to elucidate its relationship to phytoplankton community composition, physiological state (as assessed by measurements of photosynthetic efficiency) and a range of abiotic factors. We focused on the pico- and nanophytoplankton because of their importance to overall productivity, and because cell populations which mostly correspond to these size classes could be easily determined with flow cytometry allowing for rapid assessment of abundance and viability. Speed and efficiency were essential for successful viability measurement from multiple depths per site.
The specific objectives were to identify (1) potential differences in the viability of different phytoplankton groups, and (2) test for correlations between cell viability and abiotic and biotic factors including temperature, nutrient concentration, DOC concentration (that is affected by the release from lysing phytoplankton cells), abundance of larger phytoplankton (especially diatoms as potential PUA producers) and photosynthetic efficiency (Fv/Fm) of the phytoplankton community.
Materials and methods
Seawater samples were collected using Niskin bottles on a Rosette sampler in combination with a SeaBird SBE911 Plus CTD probe. Samples from 3 m depth were taken at every station. To examine the depth variation of phytoplankton communities, depth profiles from 19 stations were sampled at 1, 3, 10, 30 and 60 m (GoF and BP) and from 3, 10, 20 and 50 m (ÅS) (Supplementary Table 1). Water temperature, salinity, and concentrations of chlorophyll a (Chl a), inorganic nutrients (NO3− + NO2−, NH4+, PO43−) and DOC were measured at each station. Photic zone depth was calculated from Secchi depth according to Luhtala and Tolvanen (2013).
Chl a concentration was determined by filtration in duplicates onto GF/F filters (Whatman). The Chl a was extracted in 10 mL ethanol (Jespersen and Christoffersen 1987) and stored in a freezer (− 20 °C). Samples were placed at room temperature for 24 h to ensure that all Chl a was extracted before measurement with a fluorescence spectrophotometer using 450 nm excitation and 670 nm emission wavelength with 10 nm slit with (Cary Eclipse, Agilent Technologies) calibrated against Chl a standards (Sigma-Aldrich) by linear regression (n = 6).
Inorganic nutrients, NO3− + NO2−, NH4+ and PO43−, were determined using standard colorimetric methods (Grasshoff et al. 1983) directly after sampling. Limits for accurate measurements were 0.1 µmol L−1 for NO3− + NO2− and NH4+, and 0.05 µmol L−1 for PO43−. The DOC samples (20 mL) were filtered through 0.2-µm polycarbonate syringe filters into acid washed and pre-combusted vials, then 80 µL of 2 M HCl was added and the vials placed in a freezer (− 20 °C). The samples were placed at room temperature overnight before the DOC was determined by a high-temperature catalytic oxidation (HTCO), using a Shimadzu TOC-V CPH carbon and nitrogen analyser (Benner et al. 1993).
For microscopy, 200 mL was preserved with acidic Lugol’s solution and prepared using the settling method of Utermöhl (1958). Samples were enumerated under an inverted microscope (Leitz DM IRB), and a defined area of the counting chamber was viewed at three different magnifications (125 ×, 250 ×, 500 ×). The software EnvPhyto phytoplankton counting program was used and the data stored directly into the Hertta database (Finnish Environment Institute, Helsinki). Calculations of abundance, biovolume and carbon biomass were done automatically by the software according to Olenina et al. (2006), the biovolume list of HELCOM Phytoplankton Expert Group (PEG) (http://helcom.fi/helcomat-work/projects/phytoplankton) and Menden-Deuer and Lessard (2000).
A more detailed description of the phytoplankton enumeration method can be found in Lipsewers and Spilling (2018). Microscopy was used for determining phytoplankton community composition, whereas flow cytometry was used for counting the small phytoplankton and for dividing them into size categories (see below).
The photochemical efficiency, the ratio between variable and maximum Chl a fluorescence (Fv/Fm), was determined for all samples after dark acclimation (15 min) using the fluorescence induction (OJIP) curve (AquaPen fluorometer, Photon Systems Instruments) with 450 nm excitation light.
Flow cytometric analyses
Phytoplankton enumeration and viability assessment were conducted with flow cytometry (Partec Cube 8, Sysmex Partec GmbH, Goerlitz, Germany). Flow cytometry allowed the analysis of multiple depths rapidly after sampling and thereby minimized the artefacts potentially generated by sample storage, i.e. enclosure effects. Microscopic analysis of viability would not have been possible within the schedule of the cruise. Flow cytometry allows for easy and detailed analysis of pico- and nano-sized phytoplankton that are difficult to analyse microscopically.
The seawater samples were kept cold (in situ temperature) in darkness until they were split into subsamples for flow cytometric measurements. These measurements were conducted within 1 to 7 h (on average 2 h 27 min) after sampling, except for a 9 h delay at BY32 and BY15 due to harsh weather conditions. In total, four subsamples of 800 μL were taken from each sampling depth to determine phytoplankton cell viability and cell abundance. One of the four subsamples was kept unstained to estimate the green background fluorescence, and the other three were stained with 4 µL SYTOX Green to a final concentration of 0.5 μM (Veldhuis et al. 2001). The stained and gently mixed subsamples were incubated in cold and dark from 10 min (minimum staining time based on recommendations of manufacturer) to 30 min prior to flow cytometry measurements. Viability did not differ systematically between samples measured after 10 and 30 min suggesting that the stain incorporation within this time range was uniform (data not shown).
A major uncertainty associated with SYTOX Green staining, as well as with most cellular stains, is that uniform response among different cell types (i.e. phytoplankton species, in this case) cannot be guaranteed, as discussed by, e.g. Veldhuis et al. (2001) and Peperzak and Brussaard (2011). Therefore, we used killed control samples to assess the comprehensiveness of SYTOX Green staining within the total phytoplankton community at each site. One subsample from 3 m at each station was killed by keeping the sample tube in a hot water bath (80 °C) for 10 min. The heat-treated samples were used as a positive control, with all cells in the sample assumed to be dead (Franklin et al. 2009). The same staining procedure described above was conducted with the heat-treated samples to test possible differences in staining intensity of the dead cells among different phytoplankton groups (Peperzak and Brussaard 2011). Heat-treating altered the scatter and fluorescence properties of the cells (Fig. 2c) in a way that they could not reliably be divided into the same groups as the non-heated samples (Fig. 2b), which prevented direct comparison of differences in staining. Therefore, we calculated the abundance ratio of each group to all other groups and to total cell abundance (flow cytometry results) in non-heated samples and compared these ratios against the viability of the heat-treated cells at each site. We also compared flow cytometry derived total abundance and Chl a concentration against the viability of the heat-treated cells to retrospectively test if the used SYTOX Green concentration had been sufficient to stain the maximum amount of dead cells.
All statistical analyses were done using R version 3.5.1 (R Core Team 2018). Differences in viability among the three phytoplankton groups and vertical differences in viability and cell abundance of phytoplankton communities were analysed using Welch-ANOVA. A nonparametric method was chosen due to unequal sample sizes and heteroscedasticity among the three flow cytometer-based phytoplankton groups. Games–Howell post hoc test was applied on the significant Welch-ANOVA results. Before analyses, the percentage values of viability were logit transformed, which is a transformation commonly used for proportions. Differences were considered significant at a p value < 0.001. Detailed results for Welch-ANOVA tests are presented in Supplementary Table 2.
A generalized linear model (GLM) with beta distribution was used to investigate relationships between phytoplankton viability versus environmental variables and large phytoplankton (i.e. species counted with microscopy) abundance. Beta distribution was chosen because of its applicability to analysing proportions. Different model selections were conducted between viability and abiotic variables and between viability and large phytoplankton abundance to avoid collinearity issues between abiotic variables and large phytoplankton. GLM with beta distribution was also used for investigating the relationship between Fv/Fm and viability. We used a negative binomial GLM to investigate the relationship between phytoplankton abundance and environmental variables and large phytoplankton because negative binomial GLM can be used to analyse count data and to deal with overdispersion. Linear regression was used when the response variable was neither proportion nor count data. Model selection for GLM was done using Akaike information criterion. Data exploration was performed according to the protocol of Zuur et al. (2010) as closely as possible. Individual regressions were conducted for abundances and viabilities for G3 and G2, as the two groups were assumed to occupy different ecological niches. These regression models only apply to 3 m depth as deeper samples were not taken at every station. G1 were excluded from the regression analyses because this group was present in too low abundance at many stations. Stations BOSEXC1 and BY15 in the BP were excluded from the statistical analyses due to missing values of temperature and salinity at BOSEXC1 and a laboratory error in DOC measurement at BY15. Detailed regression model parameters are presented in Supplementary Table 3.
Physicochemical properties of the water
Mean values and standard deviations (± SD) of temperature (T) (°C), salinity (Sal), NO3− + NO2− (µmol L−1), NH4+ (µmol L−1), PO43− (µmol L−1), Chl a (µg L−1), DOC (mg L−1) and photochemical efficiency (Fv/Fm) in the Gulf of Finland, the Baltic Proper, and the Åland Sea at 3 m
NO3 + NO2
Gulf of Finland
2.5 ± 0.6
5.1 ± 0.1
1.3 ± 2.0
0.1 ± 0.0
0.5 ± 0.1
19.6 ± 9.0
16.0 ± 9.4
0.63 ± 0.07
4.4 ± 0.8
7.0 ± 0.5
0.1 ± 0.4
0.1 ± 0.0
0.4 ± 0.1
5.9 ± 3.3
8.4 ± 2.5
0.65 ± 0.08
3.0 ± 0.2
5.6 ± 0.1
0.1 ± 0.1
0.1 ± 0.0
0.2 ± 0.0
9.2 ± 1.1
8.1 ± 2.9
0.62 ± 0.05
Abundance and viability of pico- and nanophytoplankton
Cell abundances (cells mL−1) of G1, G2 and G3 in the Gulf of Finland, the Baltic Proper and the Åland Sea at depths 1, 3, 10, 20, 30, 50, 60 m
Gulf of Finland
1155 ± 130
1394 ± 192
1215 ± 166
287 ± 37
365 ± 61
14,473 ± 383
21,350 ± 6617
19,571 ± 6146
2288 ± 527
1618 ± 247
3158 ± 248
3310 ± 305
2867 ± 443
805 ± 372
547 ± 196
211 ± 85
302 ± 153
236 ± 130
161 ± 100
217 ± 80
10,769 ± 4228
11,390 ± 4013
11,178 ± 3649
4846 ± 3190
1495 ± 621
2043 ± 1037
1747 ± 634
1801 ± 878
1045 ± 984
269 ± 144
866 ± 99
785 ± 175
583 ± 408
130 ± 30
31,556 ± 2939
27,833 ± 2738
18,510 ± 12,525
1653 ± 413
1679 ± 323
1350 ± 110
913 ± 403
238 ± 28
DOC concentration values were log-transformed for regression analyses to avoid issues caused by two stations with atypically high DOC concentrations. We did not find a relationship between viability and DOC, but there was a slight correlation between DOC concentration and G3 abundance (linear regression, r2 = 0.18, F1, 23 = 6.28, p = 0.020, Fig. 6h). The Fv/Fm correlated with G2 viability (beta regression, pseudo-r2 = 0.44, p < 0.001, Fig. 6g).
Total viability in the heat-killed samples varied from 3 to 28% with an average of 12% and standard deviation of 6%. There was no relationship between the viability in the heat-treated controls and the abundance ratios of any of the flow cytometry-based phytoplankton groups. There was a significant positive relationship between Chl a concentration and viability in the heat-treated controls (beta regression, pseudo-r2 = 0.37, p < 0.001, data not shown). There was no significant difference in flow cytometry total event counts between unstained and stained samples. Usually the event count in dead controls was somewhat lower than in non-heated samples and higher only at three stations. However, at these stations the event count was much higher; in particular at the station BY7 there were more than twice as many counts in the dead controls than in non-heated samples.
The current study is to our knowledge the first to examine natural phytoplankton viability over a large spatial scale in the Baltic Sea. Total viability of the phytoplankton community measured by flow cytometry did not vary much, but our analyses revealed high variability among different phytoplankton groups and relatively low viabilities for nano-sized cells (G3). Flow cytometric studies of phytoplankton simultaneously provide information on abundance of a given size range and on the physiological state of individual cells within the community. The results from this study clearly emphasize the importance of studying viability at the single-cell level (Davey and Kell 1996), as restricting the viability analyses to the community level would have completely missed the difference in viability between pico- and nano-sized fractions.
At some stations, e.g. in the BP, non-viable cells accounted for more than half of the G3 population. The viability of G1 and G2 was significantly higher than the viability of G3, which could be a true observation or caused by different sensitivity to SYTOX Green staining between the groups. Since the total viability of the heat-killed phytoplankton samples did not differ regardless of the proportion of individual groups within the community we can cautiously assume that the staining sensitivity of all the groups is equal. Of course, this is only a very approximate test of equal staining and does not rule out differences in the staining sensitivity among the individual species within the broad flow cytometry-based groups. Also at some stations, the total event count in dead controls was much higher than in non-heated samples which further complicates the interpretation of dead controls. If we exclude these stations and assume that the heat treatment is fatal to all phytoplankton cells, then an average of 12% of the dead phytoplankton would not express SYTOX Green fluorescence when killed by heat treatment. If this staining anomaly carries on to natural environment and other causes of death, then our analysis slightly overestimates the viability. This overestimation would generally be low, and because there are no clear trends between staining of heat-killed cells and the abundance ratios of the phytoplankton groups, we argue that the lower viability of G3 is a true observation and not entirely caused by differences in staining sensitivity. The positive relationship between viability in heat-treated samples and the Chl a concentration might be a more serious source of error with our method. This suggests that the SYTOX Green concentration we used was not sufficient to stain all the phytoplankton in dead controls when their density was high, which might lead to lower fluorescence intensity and thus to overestimation of viability in such situations. Viabilities > 15% in dead controls started to appear when the Chl a concentration was > 8 µg L−1 although also low viability values persisted with such high Chl a concentrations. This possible overestimation of viability cannot be proven nor removed retrospectively. Instead we emphasize the uncertainty of viability estimates at high Chl a concentrations and suggest SYTOX Green concentration to be adapted to Chl a concentration.
Our results are in line with other studies that report high variability in phytoplankton viabilities among different phytoplankton taxa (e.g. Veldhuis et al. 2001; Hayakawa et al. 2008; Rychtecký et al. 2014). Hayakawa et al. (2008) quantified phytoplankton viability with an enzymatic membrane permeability test and found that eukaryotic phytoplankton (< 10 µm) had significantly lower viability compared to Synechococcus sp. in the northwest Pacific Ocean. Also, Veldhuis et al. (2001) found the highest percentages of viable cells in Synechococcus sp., with a viability range of 75–95% during spring. Similarly, in our study, the viability of picocyanobacteria (G1), mainly represented by Synechococcus spp. in the Baltic Sea (Kuosa 1991; Motwani and Gorokhova 2013), varied from 77 to 91%. However, Peperzak and Brussaard (2011) reported poor staining by SYTOX Green of Synechococcus sp. If Synechococcus sp. globally stain poorly with SYTOX Green and, as a result, the green fluorescence intensity of some dead Synechococcus sp. cells stays below the five times background fluorescence of the sample, then it is possible that the viabilities reported for G1 in this study are overestimated. Also G2 and G3 might contain significant amounts of phytoplankton species with poor staining response to SYTOX Green. This is an inherent limitation of SYTOX Green method and with our data we cannot assess the responses of individual species included in the flow cytometry-based phytoplankton groups. Yet, SYTOX Green is a commonly used viability probe, which functioned well with most of the species tested by Peperzak and Brussaard (2011) including small local species such as Rhodomonas baltica, and we assume that our results are comparable with other studies investigating phytoplankton viability at the community level.
The depth-dependent variation in viability could possibly be explained by stratification of the water column. G2 viability at 60 m (18 stations) was slightly higher at stations where the halocline was deeper and the 60 m sample was retrieved above the halocline. Temperature above the halocline was mostly uniform suggesting that the water column above the halocline was well mixed at most stations. This may result in uniform average viability of the phytoplankton when the cells are retained within the mixed layer and assumed to be regularly brought into the photic layer. However, the photic layer was always much shallower than the halocline (minimum distance between euphotic zone and upper limit of halocline varied between 15 and 64 m). Thus, phytoplankton may be exposed to extended periods of darkness even within the mixed layer, which may explain the decreased G2 viability in the deep end of the mixed layer (most 30 m measurements). If the cells end up below the mixed layer (as seen in 8 out of 16, 60 m depth measurements), and therefore permanently beyond the photic zone and the compensation depth, even higher decrease in viability could be expected, as was seen for all flow cytometry-based phytoplankton groups. Any changes in viability caused by light intensity (Agustí 2004) could therefore be expected to be influenced by the mixed layer depth.
We detected lower abundances of all flow cytometry-based phytoplankton groups in the warmer southern stations compared to the sampling sites in the north (Table 2). In addition, our regression models suggest a negative relationship between temperature and G2 abundance (Fig. 6c) which indicates that phytoplankton might encounter higher grazing pressure in warmer waters where growth rate of zooplankton is higher (Sommer et al. 2007). Higher grazing pressure in the south could also be suggested based on the anomalous spatial distribution of G1 (including Synechococcus spp.) in our study area. In general, even the cold-adapted clades of Synechococcus spp. are more abundant in warmer waters (Paulsen et al. 2016), but in our study, G1 (including Synechococcus spp.) had the highest abundance in the colder northern Baltic Sea and was either low or absent at most of the stations in the south, which could be an indication of top down control.
Nutrient limitation (Agustí 2004; Alonso-Laita and Agustí 2006; Rychtecký et al. 2014) and temperature (Alonso-Laita and Agustí 2006) have been shown to determine phytoplankton viability in the field. At many stations in the BP, the NO3− concentration was not detectable, indicating that the phytoplankton community had already consumed most of the NO3− available and entered an N-limited physiological state. In our results, the only correlation between abiotic factors and viable cells was the weak correlation between PO43− concentration and G2 viability implying a decoupling between nutrient concentration and viability. Somewhat surprisingly, G2 abundance correlated negatively with PO43− concentration (Fig. 6b). Concurrently, G3 abundance correlated positively with NH4+ concentration (Fig. 6a), but G3 viability did not. It seems, therefore, that the environmental variables controlling phytoplankton abundance cannot directly be used to predict phytoplankton viability. For example, abundance may be affected by grazing and sinking, whereas viability might not. However, there is uncertainty in the regression analysis involving NH4+ concentration, because at some stations the measured concentrations were below the accurate detection limit.
Our results demonstrate that inorganic nutrient concentration cannot per se be used to evaluate the physiological state of phytoplankton, even though phosphate concentration seemed to explain a small fraction of the variation in G2 viability. Nutrient affinity is tightly linked to size as the surface to volume ratio changes with a 2/3 power exponent, and the smaller sized picophytoplankton satisfy their nutritional needs at a much lower nutrient concentration (Irwin et al. 2006). In addition, with rapid nutrient turnover, the phytoplankton cells might not experience nutrient stress even at very low inorganic nutrient concentrations. This might in part explain the high G2 abundance in low PO43−concentration, as especially the small phytoplankton gain competitive advantage against larger cells by efficiently using the recycled PO43− in nutrient depleted environment (Irwin et al. 2006). The photochemical efficiency (Fv/Fm) is a better proxy for physiological state as the photosynthesis is rapidly downregulated during stress conditions, e.g. depletion of inorganic nutrient(s), but there is also a taxonomic component affecting the Fv/Fm (Suggett et al. 2009). In this study, Fv/Fm explained cell viability better than inorganic nutrient concentration. Especially G2 viability correlated with Fv/Fm (Fig. 6g), which could be due to their high abundance throughout the sampling area. A more comprehensive viability assessment of natural phytoplankton communities might reveal how well Fv/Fm and membrane integrity-based viability assessments are aligned. This, however, can be relatively complicated because as, e.g. Franklin et al. (2009) demonstrate, high Fv/Fm value might not be a clear sign of absence of dead cells, although cells with reduced viability likely have lower photosynthetic efficiency (Veldhuis et al. 2001). A drop in Fv/Fm might also be a transient response to stress, as phytoplankton continuously acclimates to their surroundings (Halsey and Jones 2015), and during the spring bloom in the Baltic Sea, the primary production output per Chl a unit (the assimilation number) is not affected by the inorganic nutrient concentration (Spilling et al. 2019). Therefore, it might be better to consider measurements of Fv/Fm and viability as complementary assays for the physiological state of phytoplankton communities. Also Veldhuis et al. (2001) coupled viability analyses with a photosynthetic stage measurement. By using 14C incorporation as a determinant of cell physiological status, they found that populations of cells containing photopigments but possessing compromised membranes were, at least partially, capable of photosynthesis, but had lower 14C fixation rates. Bulk measures such as 14C fixation rates, while useful in overall population assessment, inevitably integrate physiological heterogeneity within microbial populations, meaning that correlations between the bulk measure, and single-cell measurement, are difficult to interpret (Davey and Kell 1996).
The question remains, what causes the presence of non-viable cells within the observed flow cytometry-based phytoplankton groups? Viability was occasionally fairly low even when Chl a concentration was high. This was especially pronounced at the station LL7S during both samplings, where Chl a was exceptionally high (23.5 and 26.1 µg L−1 on the first and the second sampling, respectively), but the viability of each flow cytometry-based group was comparable to stations with lower Chl a concentration. One possible explanation could be the allelopathic interactions among the members of the microbial community. Among such interactions is the release of PUAs, which have been shown to induce cell death among some phytoplankton species (Ribalet et al. 2007). PUAs can be produced by different phytoplankton species, but especially by diatoms. For example, Taylor et al. (2009) observed increased PUA production in Skeletonema marinoi during increased nutrient limitation in spring in the Baltic Sea. Skeletonema marinoi was not present in high numbers during the cruise, but at many stations diatom abundance was high (up to 5000 cells mL−1), and there was a clear negative correlation between total diatom biomass and G2 viability (Fig. 6e) which might indicate allelopathy, possibly mediated by PUAs. However, this interpretation is complicated by the low diatom biomass at several stations and by the positive correlation between diatom biomass and G2 abundance (Fig. 6d). A possible explanation for this observed conflict could be that the conditions were favourable for growth of both G2 and diatoms and PUA production started only at high cell densities at the onset of diatom bloom decline, as has been demonstrated for S. marinoi by Vidoudez et al. (2011b). Cózar et al. (2018) concluded, based on in situ measurements, that per cell release of PUAs increases with increased oligotrophy, presumably to enhance the bacterial remineralization rates of nutrients (Edwards et al. 2015). Most of the Baltic Sea is far from oligotrophic, but towards the end of the bloom the nutrient limitation might induce an increase in PUA production, which, given the high phytoplankton density, could result in a sufficiently high PUA concentration to induce a detectable reduction in viability in the measured fraction of the phytoplankton population. Since G2 viability also correlated negatively with the total abundance of flow cytometry-based phytoplankton (Fig. 6f), we cannot rule out the possibility of PUA mediated allelopathy among the small phytoplankton (Vidoudez et al. 2011a; Morillo-García et al. 2014). However, without measurements of PUA concentrations this remains speculation.
Dead phytoplankton cells have emerged as an important DOC source in many, mainly oligotrophic, marine environments (e.g. Kirchman 1999; Franklin et al. 2006; Agustí and Duarte 2013). In coastal seas, such as the Baltic Sea, autochthonous DOC from riverine sources may account for a substantial part of the DOC pool (Hoikkala et al. 2012). The Baltic Sea has various dissolved organic matter sources, and the influence of allochthonous DOC is strong (Sandberg et al. 2004; Alling et al. 2008; Kuliński and Pempkowiak 2008). Terrestrial origin of DOC could have been important also during our study, as we did not detect a relationship between viability and DOC. DOC concentration correlated slightly positively with G3 abundance (Fig. 6h) indicating that they might have produced a detectable increase in the DOC pool. This relationship was mainly caused by the two very high DOC values at LL7S and LL9 and was not significant anymore if these stations were excluded. Yet, these values are within the natural variation of DOC concentration in the area and were included in the analysis. Viability of G3 was on average lower than the viability of G2 which would support an interpretation that dying cells contribute to the DOC pool. However, the abundance of G3 was very low throughout the cruise; G2 was often 10 times more abundant and many large phytoplankton, such as large diatoms and dinoflagellates, which likely have been excluded from the flow cytometry-based G3 category, often coexisted in high abundances. With so low relative abundance, it seems unlikely that the DOC release from G3 would overshadow the DOC release from other, more abundant, phytoplankton groups. Our results therefore suggest that sources other than cell death might be more important for the DOC concentration, although DOC release from dying larger phytoplankton (Camarena-Gómez et al. 2018) cannot be excluded with our viability data that concentrated in the smaller size fractions. DOC may also originate from living phytoplankton (Thornton 2014) through passive diffusion (Bjørrisen 1988) or active release (Wear et al. 2015), or from heterotrophs (Steinberg and Landry 2017).
Essential complementary information on phytoplankton communities can be acquired by flow cytometry to address important ecological questions such as the distribution and fate of microalgal cells. By investigating the spatial patterns of phytoplankton viability and separately the different groups identified by flow cytometry, this study contributes to filling the gap of knowledge on the physiological condition of the phytoplankton communities across the Baltic Sea. We demonstrated that viability in natural phytoplankton communities in the Baltic Sea varied among different functional groups, and that non-viable cells were always present. Cell death therefore contributes to spring bloom dynamics where grazing and sinking traditionally have been regarded as the main loss factors. We found that abiotic factors that affect the viability of phytoplankton communities in other marine environments may not be as clearly associated with phytoplankton viability in the Baltic Sea during spring. We also showed that factors affecting the abundance of phytoplankton were not the same factors that affected their viability. Further studies assessing viability of larger phytoplankton taxa during other seasons are needed to understand their role in contributing to the Baltic Sea DOC pool.
Open access funding provided by University of Helsinki including Helsinki University Central Hospital. This study was funded by the Walter and Andrée de Nottbeck Foundation (MV, SE, TT), the Swedish Cultural Foundation in Finland (TT), and Academy of Finland (KS, decision no 259164). The study utilized SYKE marine research infrastructure as a part of the national FINMARI consortium. We kindly thank the crew and scientific staff of R/V Aranda for their support during the CFLUX16 cruise and Johanna Oja for doing the microscopy counts.
Compliance with ethical standards
Conflict of interest
The authors declare that they have no conflict of interest.
- Davey HM, Kell DB (1996) Flow cytometry and cell sorting of heterogeneous microbial populations: the importance of single-cell analyses. Microbiol Mol Biol Rev 60(4):641–696Google Scholar
- Grasshoff K, Ehrhardt M, Kremling K (1983) Methods of seawater analysis. Verlag Chemie, WeinheimGoogle Scholar
- Jespersen AM, Christoffersen K (1987) Measurements of chlorophyll a from phytoplankton using ethanol as extraction solvent. Arch Hydrobiol 109:445–454Google Scholar
- Jiménez C, Capasso JM, Edelstein CL, Rivard CJ, Lucia S, Breusegem S, Berl T, Segovia M (2009) Different ways to die: cell death modes of the unicellular chlorophyte Dunaliella viridis exposed to various environmental stresses are mediated by the caspase-like activity DEVDase. J Exp Bot 60:815–828PubMedPubMedCentralCrossRefGoogle Scholar
- Lipsewers T, Spilling K (2018) Microzooplankton, the missing link in Finnish plankton monitoring programs. Boreal Environ Res 23:127–137Google Scholar
- Olenina I, Hajdu S, Edler L, Andersson A, Wasmund N, Busch S, Göbel J, Gromisz S, Huseby S, Huttunen M, Jaanus A, Kokkonen P, Ledaine I, Niemkiewicz E (2006) Biovolumes and size-classes of phytoplankton in the Baltic Sea. Balt Sea Environ Proc 106. HELCOM, Helsinki, pp 142Google Scholar
- Olson RJ, Zettler ER, DuRand MD (1993) Phytoplankton analysis using flow cytometry. In: Kemp PF et al (eds) Handbook of methods in aquatic microbial ecology. CRC Press, FloridaGoogle Scholar
- R Core Team (2018) R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna. https://www.R-project.org/. Accessed 1 Sept 2018
- Smith SR (2009) Flow cytometric analysis of phytoplankton viability in Elkhorn Slough, California. MSc thesis. San Jose State University, San JoseGoogle Scholar
- Spilling K, Fuentes-Lema A, Quemaliños D, Klais R, Sobrino C (2019) Primary production, carbon release, and respiration during spring bloom in the Baltic Sea. Limnol Oceanogr 64:1779–1789Google Scholar
- Utermöhl H (1958) Zur Vervollkommnung der quantitativen phytoplankton-methodik. Mitteilungen der Internationale Vereinigung für Theoretische und Angewandte. Limnologie 9:1–38Google Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.