Laboratory experiments have shown that several environmental factors and biotic parameters influence the production and release of particulate and dissolved PUA [12,14]. We undertook mesocosm experiments that allowed us to access and manipulate a natural system, mimicking natural conditions during typical field situations including plankton blooms [22]. In mesocosm A, where fjord water was enclosed without further manipulation, a relatively constant background level of S. marinoi was observed that did not exceed 100 cells mL⁻¹ (Figure 1A). This mesocosm did not differ from the outside fjord waters in any of the monitored parameters and confirmed earlier studies in this mesocosm set-up that demonstrate that enclosure does not alter the plankton community significantly (data not shown, [22]). When nitrate and phosphate but no silicate were added (mesocosm B), the S. marinoi density increased to a maximum of 0.6·10³ cells mL⁻¹ at day 10 (Figure 1B). When the enclosed natural plankton population was fertilized with nitrate, phosphate and silicate (mesocosm C) the cell densities of S. marinoi increased to a maximum of 1000 cells mL⁻¹ at day 10 (Figure 1C).

Inoculation of a fully fertilized mesocosm (D) with 100 cells mL⁻¹ S. marinoi did not result in a noticeable increase in S. marinoi cell count compared to mesocosm C (Figure 1D). In contrast, inoculation with 400 and 1000 S. marinoi cells mL⁻¹ (mesocosm E and F, respectively) led to a rapid bloom development and maximum densities were reached at day 8 (∼17·10³ cells mL⁻¹) and day 7 (∼70·10³ cells mL⁻¹)(Figure 1E and F). In all fertilized mesocosms, we also observed a bloom of Phaeocystis sp. with maximum cell counts at day 7–10 (Figure 1B–F). No Phaeocystis was detected in mesocosm A. In the mesocosms with nutrients but no induced S. marinoi bloom (mesocosms B–D), Phaeocystis sp. reached a relatively stable stationary phase with concentrations up to 1 10⁵ cells mL⁻¹, while in mesocosms E and F, concentrations started to decline from 5 10⁴ cells mL⁻¹ towards the end of the experiment (Figure 1). Blooms of different intensities of the PUA-producing diatom S. marinoi could be triggered by manipulating nutrients (nitrate, phosphate and silicate) as well as the cell counts of S. marinoi cultures used for inoculation (see also [23]). Since S. marinoi was also present in the mesocosms that were not inoculated with the starting culture, we can conclude that this species was a member of the natural plankton community during the field conditions and its abundance was only enhanced in mesocosms B–F. The isolate used for the inoculations was obtained from the Fjord water before and even if we cannot exclude genetic variations we can state that the observed blooms consisted of algae that are native to the environment. Despite the manipulations, we still observed a complex and diverse plankton community in all mesocosms throughout the experiment. Besides the two dominant phytoplankton species discussed here, ciliates as well as heterotrophic and autotrophic dinoflagellates were also detected in varying amounts. These contributed to the overall biomass in the size fraction >15 μm roughly equal to the summed carbon of Phaeocystis sp. and S. marinoi [23]. Other phytoplankton <15 μm that could potentially also contribute to PUA were observed by flow cytometry (Figure 2). These were coccolithophores, cryptophytes and other un-identified small species. However, none of these cells reached cell abundances near the concentrations observed for S. marinoi and Phaeocystis sp.. Phaeocystis and S. marinoi population development scaled well with PUA while the underlying phytoplankton background did not display a similar pattern (Figures 1 and 2). This observation argues in favor of Phaeocystis sp. and S. marinoi as the main source of PUA. In this mesocosm set-up, we deliberately decided against replicates but rather performed a gradual shift of the natural plankton composition to an induced intense bloom in several steps. This set-up allowed evaluating the contribution of different phytoplankton species to the overall PUA production under different scenarios.

Particulate PUA were detected at all time points in the fertilized mesocosms. Analysis of the mass spectra and comparison with authentic standards revealed the presence of three PUA (2,4-heptadienal, 2,4-octadienal and 2,4-decadienal) in variable amounts. Cells from mesocosm A did not produce any particulate heptadienal or octadienal, and decadienal was recorded only at four sampling days (day 0, 1, 12, 13; Figure 1A). In all other mesocosms, all three PUA were detected (Figure 1B–F).

Heptadienal and octadienal - In mesocosm B, low concentrations of particulate heptadienal were detected (Figure 1B). A similar picture arose in mesocosms C and D, but slightly higher concentrations of these particulate PUA were found (Figure 1C, D). In mesocosm B, particulate heptadienal increased until day 10, which was also the day of the highest S. marinoi density. Up to 0.56 nmol heptadienal was produced by the totality of all cells recovered by filtration of 1 L (referred to as nmol cells⁻¹ of 1 L from here on)(Figure 1B). In mesocosms C and D, particulate heptadienal reached a maximum at day 8, well in accordance with the S. marinoi density. Total concentrations reached 0.88 and 0.69 nmol cells⁻¹ of 1 L, respectively (Figure 1C, D). In mesocosm E, where a S. marinoi bloom was induced, particulate heptadienal reached 5.96 nmol cells⁻¹ of 1 L at day 7, one day before the maximum of S. marinoi cell density. In mesocosm F, where an even more pronounced bloom was triggered, up to 6.8 nmol heptadienal cells⁻¹ of 1 L were found at day 8, one day after the maximum of S. marinoi (Figure 1E and F). If normalized only to the S. marinoi cell counts, the particulate heptadienal was similar in all mesocosms (0.3–1.3 fmol S. marinoi cell⁻¹). This value is in good accordance with PUA production levels of this diatom observed in lab studies [4]. The production rates of total PUA from the strain used for the inoculations in this experiment were varying between 0.2 and 1.7 fmol cell⁻¹ dependent on culture conditions in an independent lab experiment. Octadienal was detected only when the S. marinoi density was highest, in mesocosm E and F (Figure 1E and F). The particulate concentrations reached a maximum at day 8, measuring 0.17 and 0.21 nmol cells⁻¹ of 1 L respectively (10 and 6 attomol if normalized per S. marinoi cells). It is interesting to note that in the lab studies higher proportions of octadienal were observed. Since both, the overall level of PUA production as well as the relative ratio of heptadienal:octadienal is dependent on the culture conditions it is possible that a shift towards heptadienal occurs under near natural conditions [14]. The particulate heptadienal was correlated to S. marinoi cell density when all mesocosm data were considered and confirmed the connection between this PUA producer and the detected heptadienal (Spearman correlation factor between S. marinoi cell density and particulate heptadienal: 0.86, p < 0.001; between Phaeocystis cell density and particulate heptadienal: 0.66, p < 0.001). Best subset regression showed that Phaeocystis cell density was not a significant variable in predicting cell potential of heptadienal production (S. marinoi cell densities only independent variable, r² = 0.59, p < 0.001, S. marinoi and Phaeocystis as independent variables, r² = 0.60, p < 0.001 for S. marinoi variable, p = 0.4 for Phaeocystis variable). A contribution of the other species in the mesocosm to the overall detected heptadienal and octadienal cannot be fully ruled out. But since the production rates per S. marinoi cell match well with lab values and since the other members of the plankton community were not very abundant, we favor the interpretation that 2,4-heptadienal and 2,4-octadienal can be attributed to the induced bloom of the G4 strain of S. marinoi that was used for the inoculation.

Decadienal—Decadienal was detected in all investigated mesocosms. This PUA is no typical diatom metabolite and, despite numerous surveys, has only been reported once as a natural product from Thalassiosira rotula [8]. An alternative source of decadienal could be Phaeocystis sp. that always occurred when this PUA was found. In laboratory studies, decadienal was detected from Phaeocystis pouchetii from northern Norway water [5]. Indeed it was observed in mesocosm B that the particulate decadienal gradually increased until reaching 0.8–0.9 nmol L⁻¹ at day 6. Then, at day 7 when Phaeocystis sp. entered stationary phase, particulate decadienal jumped to greater than 1.5 nmol cells⁻¹ of 1 L, and stayed relatively stable until day 12. Similar behavior was observed in mesocosms C and D where S. marinoi did not bloom (Figure 1C and D). If we assume that Phaeocystis sp. is the only decadienal producer, the potential production of PUA would be equivalent to 18–38 attomol decadienal cell⁻¹of this algae in mesocosm B. In mesocosms E and F particulate decadienal increased to a maximum at day 8, reaching concentrations of 8.7 and 5.7 nmol cells⁻¹ of 1 L, respectively (quantity of this PUA produced by the cells recovered from filtration of 1 L) (Figure 1E and F). The per cell decadienal potential in mesocosms E–F was always higher than in the other mesocosms. If normalized only to Phaeocystis sp. cell counts, the detected particulate PUA would correspond to 300 and 290 attomol Phaeocystis cell⁻¹, respectively.

Particulate decadienal was best correlated with Phaeocystis cell density if compared to S. marinoi cell density, when data from all mesocosms were taken into account (Spearman correlation factor between Phaeocystis cell density and decadienal potential: 0.823, p < 0.001; between S. marinoi cell density and decadienal potential: 0.65, p < 0.001). Best subsets regression analysis, however, showed that the regression model was better if S. marinoi cell density was also taken into account (Phaeocystis as only independent variable, r² = 0.117, p < 0.001, Phaeocystis and S. marinoi as independent variables, r² = 0.512, p < 0.001 for both variables). However, when the data from each mesocosm were considered separately, decadienal correlated with Phaeocystis cell density, with influence from S. marinoi (Figure 3).

The detection of decadienal in plankton communities requires attention and future experimental work, since it is the most active of the known PUA [9]. The particulate decadienal production of Phaeocystis sp. cells increases in all manipulated mesocosms where S. marinoi are present in higher proportions (Figure 1C–F). Little is known about the regulation of PUA production in Phaeocystis, but it could be hypothesized that an induced PUA production in the presence of a competitor could give a selective advantage to Phaeocystis sp. Alternatively, the altered environmental conditions caused by the higher overall cell counts in the mesocosm could result in a non-specific physiological up regulation of the particulate PUA in the prymnesiophyte.

The controlled mesocosm studies presented here allowed us to follow the PUA potential of at least two major bloom forming members of the phytoplankton community that develop in parallel. The activity of PUA depends on the α,β,γ,δ-unsaturated aldehyde structure element found in all detected PUA of this study [9,24]. It can thus be assumed that PUA from different sources can have additive effects if, for example, their toxicity is considered. If only the particulate PUA released upon wounding of the cells are considered, such an additive effect would require feeding of herbivores on both S. marinoi and Phaeocystis sp. With few exceptions [25] most previous studies focused on the influence of diatom-derived PUA on other members of the plankton community [3,10]. Our survey, however, shows that PUA from other sources, such as Phaeocystis, can be dominant in mixed bloom conditions. The fact that apparently other phytoplankton members than diatoms can be the major PUA producers in plankton should be considered if diatom effects are discussed in the light of the action of PUA [26]. This is also supported by our recent survey of a developing S. marinoi bloom in the Adriatic Sea that revealed additional PUA-producers with smaller cell sizes compared to the diatom [1].

The values of dissolved PUA were very variable, despite that the employed method was previously validated for a unialgal culture of S. marinoi [12]. The high variability might arise from interfering biological material in the mesocosms that might absorb partially the released PUA. This is also observed for the internal standard added. The concentration of dissolved PUA was very low (<0.025 nM) or undetectable in mesocosm A (Figure 4A). The dissolved heptadienal concentration remained low in mesocosm B, increasing only up to ∼0.1 nM towards the end of the experiment (Figure 4B). In mesocosms C and D, dissolved heptadienal concentrations increased over the first 8–10 days, reaching comparably high values when S. marinoi cell density was starting to slowly decrease (Figure 4C and D). In the mesocosms where S. marinoi bloomed, dissolved heptadienal concentrations reached their maximum (0.4 nM and 1 nM for mesocosms E and F, respectively) one day before S. marinoi cell densities reached their maxima. In both mesocosms the concentration dropped again on the day of the maximum S. marinoi cell density and only subsequently increased again to 0.2–0.3 nM on the first day of the S. marinoi decline. The concentrations of heptadienal at the end of the experiments were lower compared to those on day 6–9 (Figure 4E and F). Only very low levels of dissolved octadienal were detected in the mesocosms, without clear trends. In mesocosm B, decadienal was the most abundant dissolved PUA. Its concentration in the water remained very low until day 7 and then increased to reach a concentration of 0.28 ± 0.19 nM at day 8 (Figure 4B). Decadienal concentrations then decreased gradually over the following days with one exception on day 13 (Figure 4B). In the other two mesocosms where nutrients were added but S. marinoi did not bloom, the same trend for decadienal was observed (Figure 4C and D). In mesocosms E and F, dissolved decadienal was only found in very low concentrations (Figure 4E and F).

Release of dissolved PUA from S. marinoi was previously observed for laboratory cultures. There, PUA release preceded the transition of S. marinoi from the stationary to the declining phase [12]. Our mesocosm observations are in good accordance with this release pattern (Figure 4E, F). A role of PUA as signal in stress surveillance of diatoms has previously been proposed and our findings support a possible involvement of released PUA as potential signals [11]. In addition, a delayed second maximum of dissolved PUA was observed in these mesocosms, which was most likely due to the release of PUA during cell lysis of the declining bloom [13]. Dissolved decadienal, however, did not follow such distinct patterns. In mesocosms C and D, where S. marinoi only reached low cell counts, dissolved decadienal was clearly detectable. In mesocosms E and F, where Phaeocystis cell counts were similar to those in mesocosms C and D, dissolved decadienal was only detected on a few occasions and in low quantities. Further follow-up studies are required to decipher if this is due to absorption or detoxification processes or if the decadienal release is reduced in the presence of the competitor.

Mesocosm experiments were conducted from week 16 to 18 (15–28 April) in 2008 at the Espegrend marine biological field station by the Raunefjord, Western Norway (latitude 60.161°N; longitude: 5.138 °E). For a general description of the location, mesocosm design, filling procedures, and water column mixing see [22]. Six mesocosms (denominated A–F), each 11 m³, 2 m diameter, open to air, were filled with 10 m³ fjord water from 4 m depth within 5 m of the mesocosm set-up. The transparent polyethylene enclosures were fixed on a frame and kept floating in the surrounding Fjord water. Throughout the experiment mesocosms were kept at natural illumination and bubbled with air. Mesocosm A was kept as a control. At day 1, B was supplemented with 0.4 μM phosphate and 4.24 μM nitrate, while mesocosm C was supplemented with the same concentration of phosphate and nitrate, and 3.61 μM silicate. Mesocosms D–F were supplemented with the same nutrients as C, but in addition, aliquots of a S. marinoi culture in exponential phase were added. The diatom culture is maintained at the University of Bergen, Department of Biology (S. marinoi strain G4 isolated on 13 June 2006 at the permanent station in the Raunefjord, one nautical mile off the field station, 60°16′N and 5°11′E). The cell density of this diatom reached ∼100 cells mL⁻¹ (mesocosm D), ∼400 cells mL⁻¹ (mesocosm E) or ∼1000 cells mL⁻¹ (mesocosm F) directly after inoculation. Cells were counted using a Cytobuoy™ scanning flow cytometer (CytoBuoy b.v. Woerden, 136 The Netherlands) and a FlowCAM II™ (Fluid Imaging Technologies, Yarmouth, ME, USA) fitted with black and white camera and additionally identified using light microscopy [23].

Sampling of mesocosms was performed daily at 9:00 am with 5 L plastic canisters that were transported immediately to the lab. Due to the constant bubbling, the water in the mesocosms was mixed and surface water sampling was sufficient to obtain information about the average situation in the respective mesocosms [22]. Canisters were stored for a maximum of 4 h in a cold room adapted to outside water temperatures until filtration. 0.5–5 L were filtered on GF/C filters under moderate vacuum (ca. 500 mbar) for quantification of particulate PUA and 1 L was used for quantification of dissolved PUA as described below.

A GCT premier orthogonal reflectron time-of-flight (oTOF) mass spectrometer (MS) (Waters, Manchester, UK), coupled to an Agilent 6890N gas chromatograph (GC) equipped with a DB-5ms 30 m column (0.25 mm internal diameter, 0.25 μm film thickness, with 10 m Duraguard pre-column, Agilent, Waldbronn, Germany) was used for GC-EI-MS measurements. Helium 5.0 was the carrier gas with a constant flow of 1 mL min⁻¹. The electron energy was 70 eV. Samples were injected with a 7683B autosampler (Agilent) equipped with a 10 μL tapered, fixed needle, PTFE-tipped plunger syringe (23–26s/42, Agilent). Calibration of the MS parameters (beam steering, focusing lenses, dynamic range extension (DRE)) was performed just before the analysis of the samples.

High grade methanol (Chromasolv © Plus >99.9%, Sigma-Aldrich) and hexane (Suprasolv, Merck, via VWR) were used for all analyses. O-(2,3,4,5,6-Pentafluorobenzyl)hydroxylamine hydrochloride (PFBHA, derivatization grade >99%, ABCR, Karlsruhe, Germany) was used for derivatization.

The determination of particulate PUA was done according to the modified protocol described in [1], based on [27]. In short, filtered cells on GF/C filters were suspended in 1 mL of a 25 mM PFBHA solution in Tris-HCl 100 mM, pH 7.2 containing 5 μL of internal standard (benzaldehyde, 1 mM in methanol, Sigma-Aldrich). The cells were disrupted by 1 min of pulsed sonication (0.5 s ultrasound–0.5 s break cycle) to initiate the PUA production. After incubation for 4 h at room temperature the samples were frozen and kept at −20 °C until further analysis. The samples were then extracted with hexane and methanol, and the hexane phase was analyzed by GC-MS.

Dissolved PUA were quantified using the modified protocol described in [1], based on [12]. In short, three replicate 1 L samples of each mesocosm were extracted on PFBHA preloaded EASY^® solid phase extraction cartridges (Macherey-Nagel, Düren, Germany). After elution with methanol, the samples were kept at −20 °C until further analysis. The samples were partitioned with water and hexane, and the hexane phase was analyzed by GC-MS.

The analysis and quantification of the PUA was conducted as described in [1]. Statistical analyses were performed with Sigmaplot (version 11.0, Systatsoftwares).