Toxicity of Dinophysis Spp. in Relation to Population Density And

Harmful Algae 6 (2007) 218–231 www.elsevier.com/locate/hal

Toxicity of Dinophysis spp. in relation to population density and environmental conditions on the Swedish west coast Odd Lindahl *, Bengt Lundve, Marie Johansen The Royal Swedish Society of Sciences, Kristineberg Marine Research Station, SE 450 34 Fiskeba¨ckskil, Sweden Received 29 April 2006; received in revised form 14 May 2006; accepted 28 August 2006

Abstract The aim of this study in the field was to investigate whether there are differences between the outer archipelago (Gullmar Fjord) and a semi-enclosed fjord system (Koljo¨ Fjord) in occurrences of D. acuta and D. acuminata as well as in their content of diarrheic shellfish toxin (DST) per cell. When all data pairs of cell toxicity of D. acuminata and the corresponding number of cells l1 from the two sites were tested in a regression analysis, a statistically significant negative correlation became evident and was apparent as a straight line on a log–log plot ( p < 0.0001). Obviously, there was an overall inverse relationship between the population density of D. acuminata and the toxin content per cell. Plotted on a linear scale, all data-pairs of cell toxicity and cell number made up a parabolic curve. On this curve the data-pairs could be separated into three groups: (i) D. acuminata occurring in numbers of fewer than approximately 100 cells l1, and with a toxin content per cell above 5 rg cell1; (ii) cell numbers between 100 and approximately 250 cells l1 with a cell toxin content from 5 to 2 rg cell1; (iii) when the population became greater than 250 cells l1, the toxicity, with few exceptions, was less than 2 rg cell1. By applying this subdivision, some clear patterns of the distribution of the differently toxic D. acuminata became evident. When comparing the cell toxicity of the two sites, it was obvious that the D. acuminata cells from all depths from the Gullmar Fjord as a mean were significantly more toxic compared to the Koljo¨ Fjord samples. The results have demonstrated that approximately 100 high-toxicity cells in a low-density population at surface may lead to the same accumulation of DST in a mussel as the ingestion of 1500 low-toxicity cells from a high-density pycnocline population. # 2006 Elsevier B.V. All rights reserved.

Keywords: Dinophysis; Cell toxicity; Population density; Cell signalling; Diarrheic shellﬁsh poisoning (DSP)

1. Introduction 2002). The problem with toxic mussels has been highlighted in several studies during the last few Diarrheic shellfish poisoning (DSP) is a major decades, however, knowledge about the phenomena is problem for the shellfish industry around the world (Lee still sparse, and the ability to forecast the occurrence of et al., 1989; Haamer et al., 1990; Dahl and Johannessen, Dinophysis and the toxin content in the shellfish has not 2001; Pavela-Vrancic et al., 2002; Morono et al., 2003). been fully developed. On the Swedish west coast the toxic dinoflagellate Dinophysis spp. occurs in coastal waters all over the genus Dinophysis is the main contributor of the mussel world (Hallegraeff et al., 2003), including Scandinavia toxins DST (diarrheic shellfish toxins) (Godhe et al., (Edebo et al., 1988; Belgrano et al., 1999; Andersen et al., 1996; Aune et al., 1996; Godhe et al., 2002,). They vary in the content of toxin per cell and their * Corresponding author. ability to produce different kinds of toxins of different E-mail address: [email protected] (O. Lindahl). poisonouness (Tables 1 and 2). Diarrheic shellfish

Table 1 Summary of reported toxicity per cell (rg cell1)ofD. acuminata Country Area OA cell1 DTX-1 cell1 DTX-2 cell1 Reference Denmark Coastal 0–40 Andersen et al. (1996) The Wadden sea 0.3 Andersen et al. (1996) Limfjord 6.1 Andersen et al. (1996) East coast of Jutland 5.3 Andersen et al. (1996) Canada Gulf of St. Lawrence 25.5 Cembella (1989) France Le Havre 1.6 Lee et al. (1989) Japan Tokyo Bay J Trace Lee et al. (1989) Spain Rio de Pontevedra 1–37 0.3–0.6 ? Blanco et al. (1995) Sweden Gullmar Fjord 9.1 0 Johansson et al. (1996) Gullmar Fjord 0.04 0.02 Yasumoto (unpublished) Gullmar Fjord 0–17 Present paper Koljo¨ Fjord 0–2.6 Present paper poisoning (DSP) causes abdominal cramps, diarrhea, have been reported to cause most of the DST toxicity nausea, and vomiting (Daranas et al., 2001). The first (Godhe et al., 2002). D. acuminata has in many incident of illness was reported in the Netherlands in the countries in northern Europe been reported to produce 1960s, and the toxic substance was first isolated from OA (Carmody et al., 1996), while D. acuta has been the black sponge Halicondria okadai (Hallegraeff et al., reported to produce OA, DTX-1, or DTX-2 (Tables 1 2003). There are three kinds of DSP toxins: the okadaic and 2). In Sweden D. acuta has been observed to cause acid (OA) group, the pectenotoxin-group, and the OA in mussels in the outer archipelago, while the co- yessotoxin group. In Sweden the okadaic acid has occurrence of DTX-1 in mussels and algae was previously been the most commonly observed DST observed in confined fjord areas (Svensson, 2003a,b). toxin (Edebo et al., 1988). The OA group also includes In the review of Maestrini (1998) it was concluded DTX-1 and DTX-2 and several derivatives (Quilliam, that little is known about the taxonomy of Dinophysis 2003). These compounds are lipid-soluble long-chain- spp. and about how they thrive and survive in the linked polyether rings and are accumulated in the pelagic. Further, it was concluded that knowledge is hepatopancreas of the blue mussel (Mytilus edulis) especially limited concerning whether the species have (Yasumoto et al., 1985; Hallegraeff et al., 2003). several life stages, exactly what they take from the sea There are several species of DST-producing in order to live and grow, and to what extent dense, Dinophysis: for example, D. acuta, D. acuminata, D. vertically patchy populations result from active or norvegica, D rotundata, D. caudata, D. fortii, D. passive concentration, from growth, or from a mixture sacculus,andD. dens (Hallegraeff et al., 2003; Graneli of both processes. It is generally clear that at times of et al., 1997). In Sweden D. acuta and D. acuminata greatest cell density, cells of Dinophysis can be

Table 2 Summary of reported toxicity per cell (rg cell1)ofD. acuta Country Area OA cell1 DTX-1 cell1 DTX-2 cell1 Reference Spain Vigo 9.4 Lee et al. (1989) Norway Sogndal 4 4.2 Lee et al. (1989) Sweden Gullmar Fjord 20 Riisgaard (1991), Edler and Hageltorn (1990) Sweden Kulefjord 100–160 Haamer et al. (1990) Sweden Gullmar Fjord 0 6.6 Johansson et al. (1996) Sweden Gullmar Fjord 0.52 0.01 1.57 Yasumoto (unpublished) Ireland Southwest coast 58 78 Kevin et al. (1998) Spain Ria de Pontevedra 0.6–94 0.4–169 ? Blanco et al. (1995) Portugal Northwest coast Detected Vale and Sampayo (2000) Ireland Bantry Bay Detected Draisci et al. (1998) Gullmar Fjord Detected Present paper Koljo¨ Fjord 0.4–7.8 Present paper 220 O. Lindahl et al. / Harmful Algae 6 (2007) 218–231 concentrated in a layer of water that represents only a The aim of this study was to investigate whether small fraction of the water column. This could be a there are differences between the outer archipelago and result of the sinking of senescent cells accumulating in the semi-enclosed fjord system in occurrences of D. the pycnocline; active vertical migration; better growth acuta and D. acuminata as well as in their content of due to the occurrence of organic material at the DST per cell, which in turn may explain the observed pycnocline suitable for Dinophysis nutritional require- occurrences of DST in mussels. The most obvious and ments; or, finally, reduction or absence of grazing conspicuous result of the present study was the clear (Maestrini, 1998). negative correlation between the population density and DST occurrence in the blue mussel has been the toxicity per cell of D. acuminata. This relationship monitored along the Swedish west coast since the was more or less present regardless of site and depth end of the 1980s. After analysis of this data set, a pattern during a period of 2 months. This result seemed to be a in toxin occurrence has become apparent (Rehnstam- novel finding. Holm and Hernroth, 2005). Most obvious was the annual cycle of DST with peaks from late summer/early 2. Materials and methods autumn to winter (August–January), when DST levels occasionally were over the limit for human consump- 2.1. Study areas tion (160 mgkg1). This pattern has, at least partly, been thought to result from cold water temperatures and Water samples were taken from the Koljo¨ Fjord a lack of non-toxic phytoplankton (diatoms) during (588130.6N; 118330.4E), which is the northern part of the winter (Edebo et al., 1988). Along the Swedish west Orust-Tjo¨rn Fjord system (Fig. 1). This chain of coast the occurrence of DST in mussels has often connected fjords and sill basins is open at both ends, differed considerably between sites situated on the outer and there is a counterclockwise (=north-going) net archipelago (high toxic mussels) compared to the semi- current of 70 m3 s1 (Bjo¨rk et al., 2000). This net current enclosed fjord system (low toxic mussels) inside the is mainly driven by salinity differences in the surface large islands Orust and Tjo¨rn (Fig. 1)(Rehnstam-Holm water at the fjord system’s two entrances. The tide is and Hernroth, 2005). However, these differences have about 20 cm along the Swedish west coast, and tidal not been regularly appearing, and there have also been exchange is thus of relatively little importance for the periods when this pattern has been reversed. The Koljo¨ water exchange in the fjord system. Bjo¨rnsund, the Fjord area, which was previously regarded as a no- or northern sill between Koljo¨ Fjord and the coastal waters low-toxicity area, experienced a dramatic change in outside, is only 9 m deep and 40 m wide and thereby has 1999, when highly toxic mussels were observed for the limited water exchange. Water of high salinity (>28 psu) first time. Today mussels from this area may from time is not exchanged through Bjo¨rnsund, whereas surface to time have a toxicity above the limit for consumption, water is flowing through this passage. In the Koljo¨ Fjord but usually not at the same time that mussels outside the the mean residence time of the surface water has been fjord are highly toxic. estimated to be in the range of 6–14 weeks (Bjo¨rk et al.,

Fig. 1. Area of investigation and sampling sites. O. Lindahl et al. / Harmful Algae 6 (2007) 218–231 221

2000). The water body below the sill depth is renewed sample of 20 ml was taken from each of the three less than once a year, and the long residence times lead to replicates from each sampling depth and allowed to settle stagnant conditions and oxygen depletion in the basin for at least 4 h in sedimentation chambers according to water during periods of time. the Uthermo¨hl technique (1958). Dinophysis spp. were Water samples from the outer archipelago were identified and enumerated using an inverted microscope collected at the mouth area of the Gullmar Fjord (588150N at a magnification of 100 or 200. At least the equivalent of and 118250E). At this site, which is much more exposed to 1 l of seawater was analyzed in samples containing few off-shore dynamics than the Koljo¨ Fjord site, surface cells, and enumeration was terminated when one of the water residence time has been estimated to be about 12 species’ cell numbers reached more than 250. days (Rydberg, 1977), and physical forces like advection Hundred milliliters of seawater samples were and up-welling effects have been shown to strongly affect collected from each depth for analyze of the chlorophyll the occurrence of the phytoplankton populations concentration and filtered onto a GFF filter. The (Lindahl, 1987). The distance between the sites over chlorophyll a concentrations were measured in a Turner land was 9 km and by a vessel 8 nautical miles. Designs Fluorometer.

2.2. Sampling 2.4. Preparation of sample for DST analysis

The field sampling was carried out between 8 August A subsamples of 88 ml of each concentrated sample and 31 October 2001 at the Gullmar Fjord site and of 108 ml volume was used for the DST analyses between 8 August and 3 October at the Koljo¨ Fjord site. according to the method described by Quilliam (2003) Hydrographical profiles were taken twice a week from by using PDAM for derivatization and a silica column surface to the bottom with a portable CTD and secchi for cleaning the sample. A C18 HPLC-column was used depth was measured monthly with a 20 cm secchi disc. for separating the molecules. OA and DTX-1 standards Water samples for phytoplankton, chlorophyll and toxin were added in addition to an internal dichlorobenzene analyses were taken once a week at three depths, chosen (DCB) standard. For detection of the toxin, a Shimadzu according to the actual depth of the halocline RF-535 Fluorescence HPLC instrument was used with (=pycnocline). Thus, the samples were collected above an excitation at 340 nm and measuring emission at halocline (1–4 m depth), in the halocline (8–13 m 395 nm wavelength. The detection limit for measuring depth) and below the halocline (12–22.5 m depth). The OA by the HPLC protocol used was estimated to samplings were performed on 13 occasions at the outer 0.25 ng OA. archipelago site and on 9 occasions at the inner fjord site. 2.5. Toxicity per cell At each sampling occasions, triplicate seawater samples were collected from the three depths by using a The content of okadaic acid (OA) or DTX-1 per cell submersible pump with a capacity of 12 l/min. The of D. acuminata respective D. acuta was only calculated water samples were concentrated through plankton net when the species could be separated through occurrence of 20 mm mesh size. The salinity was checked during in time and/or space or through type of toxin. Further, pumping by the CTD-sond. Each sample was made up the toxicity per cell was only calculated when the cell of 108 l water, concentrated into 108 ml of filtered density of the sample exceeded 50 cells l1. As there sample, thereby 1 ml of sample equaled 1 l of seawater. was no DTX-2 standard available, it was not possible to Measurements of nutrient concentrations were detect or quantify this toxin by the used HPLC- carried out according to OSPAR standards during the technique. The toxin content per cell was then first week of each month in Gullmar Fjord by the calculated by dividing the measured amount of toxin Swedish national monitoring programme (http:// of the net sample with the corresponding cell density. A www.smhi.se/menyer/ind_ocean.htm) and in the Koljo¨ mean and the standard error of the three replicates were Fjord by The Water quality Association of the Bohus also calculated. Coast (http://www.bvvf.com/). 2.6. Data evaluation and statistical analyses 2.3. Phytoplankton cell counts and chlorophyll Each mean value is expressed with its standard error The phytoplankton samples were preserved with of mean (mean S.E.M.). Analysis of variance Lugol’s solution. One sub-sample of the concentrated (ANOVA) and t-tests were used to determine the 222 O. Lindahl et al. / Harmful Algae 6 (2007) 218–231 significance of the observed differences between 10 m. The halocline was present at 12–15 m, and the groups. Simple linear regression models were used to salinity at 20 m depth was about 30 psu. In late August a test the relation type between variables. All statistical significant upward movement of the halocline occurred, methods used (ANOVA, t-test, linear regression, and resulting in a sharp salinity gradient situated between canonical discriminant analysis) were designed under depths of 2 and 4 m and also in an up-lift of more saline the assumption that the data were normally distributed. water below the halocline. In early September no Tests were used to verify that the data constituted a pronounced halocline was observed, and the salinity random sample from a normal distribution, and the increased rather evenly from 22 psu at the surface to Shapiro and Wilk (1965) statistic (W) was used. When 32 psu at 20 m depth. During this period the tempera- data were not normally distributed or when hetero- ture was around 17–18 8C above the halocline and 16– scedasticity occurred, a logarithmic transformation of 17 8C below. In mid-September a conspicuous up- data was performed as indicated by Sokal and Rohl welling event was observed, resulting in a salinity of (1995). Analyses were performed using the Statistical 30 psu at surface and 34 psu at 10 m depth. At the same Analysis System (SAS Institute Inc., 1989). Matlab time an obvious decrease in temperature of 2 8C was (MathWorks) was used for processing CTD data and for measured throughout the water column. In early interpolating purposes, and contour plots were created October low saline water (20–24 psu) again dominated with the help of Surfer (Golden Software). at the surface, and the halocline was found rather deep at 20–25 m depth. The temperature dropped to about 3. Results 13 8C. During the rest of the studied period, the salinity remained low at the surface, but the halocline rose 3.1. Environmental conditions slowly to 10–15 m depth toward the end of October while surface water was cooled to about 11 8C. Hydro-meteorological processes, in principle Due to the long residence time of water of the fjord depressions passing southwest of Sweden, resulted in system, only small and slow movements of the halocline considerable movements of the halocline at the Gullmar was observed at the site situated in the Koljo¨ Fjord. The Fjord outer coastal site during the sampling period halocline varied between 10 and 15 m depth from the (Fig. 2A). During the first 2 weeks the salinity of the beginning of August until mid-September (Fig. 2B). surface water was 25 psu or less down to a depth of Thereafter it began to rise slowly and was found at about

Fig. 2. (A) Occurrence of D. acuminata (cells l1) at the Gullmar Fjord site during the period of investigation. (B) Occurrence of D. acuminata (cells l1) at Koljo¨ Fjord site during the period of investigation. The contour lines represent isohalines at 1 psu interval. O. Lindahl et al. / Harmful Algae 6 (2007) 218–231 223

10 m depth at the end of September. Surface salinity taken at the three sampling depths during the period from remained just below 22 psu until mid-September, where early August to the end of September (Fig. 2A). Only on after it slowly increased to 23.5 psu. The surface water three occasions, and then only in the surface sample, up to temperature was 18 8C at the beginning of the study and 600 cells l1 were found during this period. In October sank to 13 8C at the end. The temperature of the halocline up to 1200 cells l1 were found in the surface sample on varied between 14 and 12 8C. Below the halocline the two occasions, while cell numbers remained low in and conditions did almost not change during the period of below the pycnocline. During the studied period there study. The salinity was 26–27 psu and measured were significantly more cells l1 of D. acuminata temperature varied between 5 and 10 8C depending on (F = 19.77, p < 0.0001) in the surface samples than in from which depth the deep water samples were taken. the pycnocline and the below-pycnocline samples. The slow but consistent increase in salinity over time at The abundance of D. acuminata at the site in the Koljo¨ most depths during the second half of the study indicated Fjord differed considerably when compared to the that some exchange of water had occurred. Gullmar Fjord. During the first half of August, up to The secchi depth in Gullmar Fjord was 7 m in 2000 cells l1 were found above and in the pycnocline, August, 11 m in September, and 8.5 m in October, while and 6000–28,000 cells l1 in samples from about 20 m in Koljo¨ Fjord the secchi depth increased from 4.5 m in depth below the pycnocline (Fig. 2B). From late August August to 5.5 m in October. and until mid-September, large abundances of D. In the Gullmar Fjord nitrate concentrations were acuminata were found in the pycnoclinewith a maximum below the detection limit (<0.10 mmol l1) above the of 38,000 cells l1. Above and below the pycnocline the pycnocline from early August until early October. During cell densities mostly varied between 1000 and this period it was only at 30 m, where nitrate concentra- 2000 cells l1. During the end of September and early tions exceeded 1 mmol l1. In early November nitrate October less than 1000 cells l1 were mostly found and at ranged from 1.97 mmol l1 at 30 m depth to all sampling depths. During the period of study there 5.96 mmol l1 at the surface. Phosphate concentrations were significantly more cells l1 of D. acuminata were somewhat more variable, and at the surface and in (F = 8.67, p < 0.0005) in the pycnocline samples the pycnocline they ranged between 0.06 and compared to the surface and the below-pycnocline 0.24 mmol l1 from August to October. Values from samples. 0.27 up to 0.65 mmol l1 were found at below pycnocline A comparison between the two sites of all and through the whole water column in early November. occurrences of D. acuminata showed that overall there The N/P ratio varied between 1 and 3 from the surface were significantly more cells l1 at the Koljo¨ Fjord site down to 15 m depth for nearly the whole investigation; than at the Gullmar Fjord site (F = 91.58, p < 0.0001) the exception was in early November, when it was around during the study. 5 at all depths, except for the surface, where it was 17. In the Koljo¨ Fjord nitrate concentrations were also 3.3. Occurrence of D. acuta under the detection limit above the pycnocline during August and September. During the same time it was 5.3 In the Gullmar Fjord D. acuta was found in cell respective 1.3 mmol l1 at 15 m depth. In October nitrate numbers less than 300 cells l1 at all sampling depths in the surface water was around 0.3 mmol l1, 6.7 at 10 m during the sampled period, with one exception: in the and 10 mmol l1 at 15 m. Phosphate in the surface water surface samples up to 900 cells l1 were found during was just below 0.1 mmol l1 during the two first months the second half of October (Fig. 3A). There were and 0.4–1.3 mmol l1 at pycnocline and deeper. In significantly more cells l1 of D. acuta (F = 60.2, October 0.3 mmol l1 was found above the pycnocline, 1 p < 0.0001) in the surface samples compared to the in the pycnocline, and up to 2.5 in the almost anoxic water pycnocline samples and in the below-pycnocline below the pycnocline. N/P could not be calculated for the samples. surface water in August and September, but it was varied In Koljo¨ Fjord up to 3000 cells l1 of D. acuta were between 1 and 2 in October. It was 13 in August, 2 in found in the surface samples during August and until September, and 9 in October at 15 m depth. mid-September, after which the occurrence dropped to less than 1000 cells l1 (Fig. 3B). In and below the 3.2. Occurrence of D. acuminata pycnocline, cells of D. acuta were present during the whole sampling period, but in densities less than At the Gullmar Fjord site D. acuminata was mostly 1000 cells l1. During the whole sampling period found in low numbers (<300 cells l1) in the samples significantly more cells l1 of D. acuta (F = 100.13, 224 O. Lindahl et al. / Harmful Algae 6 (2007) 218–231

Fig. 3. (A) Occurrence of D. acuta (cells l1) at the Gullmar site during the period of investigation. (B) Occurrence of D. acuta (cells l1) at the Koljo¨ Fjord site during the period of investigation. The contour lines represent isohalines at 1 psu interval. p < 0.0001) were found in the surface samples however, calculated cell toxicity varied between 3.31 compared to the pycnocline and below-pycnocline (0.46) and 13.34 (8.10) rg cell1. From below the samples as well as when pycnocline was compared to pycnocline the toxicity per cell varied from 7.13 below-pycnocline samples. (1.99) to 16.63 (4.35) rg cell1, but only four A comparison between the two sites of all samples, mainly from mid-October, could be calcu- occurrences of D. acuta showed that there were lated. Over time there was no evident or consistent trend significantly more cells l1 at the Koljo¨ Fjord site than of the toxicity per cell of D. acuminata in the Gullmar at the Gullmar Fjord site (F = 91.58, p < 0.0001) data set. during the study. In the Koljo¨ Fjord the toxicity of D. acuminata varied from 0.46 (0.08) to 2.57 (0.93) rgcell1 at the 3.4. Toxicity of D. acuminata surface, while pycnocline sample toxicities were between 0.12 (0.07) and 1.65 (0.83) rg cell1 (Table 3B Both D. acuminata and D. acuta coexisted on some and Fig. 4B). From below-pycnocline no samples were sampling occasions and both OA and DTX-1 were available from September, and the toxicity of the found in most samples that exceeded 50 cells l1.On remaining period of time, mostly August, varied between many, but not all, occasions it was possible to separate 0.11 (0.02) and 0.97 (0.49) rg cell1. It was obvious the species and their toxins. OA per cell is reported for that the cell toxicity of the surface samples was D. acuminata but also the species combined OA content significantly higher (F = 14.07, p < 0.0001) compared when separation not was possible (Table 3A and B, to the pycnocline and below-pycnocline samples on most Fig. 4A and B). sampling occasions during a period of 2 months starting Due to the co-existence of D. acuminata and D. at the beginning of August (Fig. 5). acuta in the surface water in the Gullmar Fjord, the The amount of OA per cell was from 0.11 (0.02) to content of OA in cells of D. acuminata could only be 16.63 (4.53) rg cell1; a difference with a factor of calculated for one sampling occasion and was 1.21 150. When comparing the cell toxicity of all samples of (0.61) rg cell1 (Table 3A). The cell toxicity of the the two sites, it was evident that the D. acuminata cells pycnocline samples could not be calculated for all from all depths from the Gullmar Fjord as a mean were samples due to too-low cell numbers, and data from the significantly more toxic (F = 117.82, p < 0.0001) second half of September were mainly missing; compared to the Koljo¨ Fjord samples. Further, the O. Lindahl et al. / Harmful Algae 6 (2007) 218–231 225 0.64 3.41 0.49 0.21 toxicity of the surface cells of the two sites differed significantly (F = 6.26, p < 0.0165), with higher toxi-

31 October city in Gullmar Fjord. It was only in the last two samplings at the change of the months September– 4.72 x 2.63 9.22 October, after the large drop of toxicity in the Gullmar Fjord, that lower or equal toxicity was found compared 0.21 1.42 0.36 1.01

24 October to the Koljo¨ Fjord (Fig. 6). When all data pairs of cell toxicity of D. acuminata 1 4.35 9.94 5.53 6.19 and the corresponding number of cells l from both the Gullmar and Koljo¨ Fjords were tested in a regres-

17 October sion analysis, a statistically signiﬁcant negative correlation became evident and was apparent as a straight 0.17 0.94 0.22 2.32 4.68 11.66 line on a log–log plot (ln OA = 0.76 (0.05) 1.99 16.63

1 ln cell l + 5.18 (0.32), F = 240.36, p < 0.0001, r2 = 0.69) (Fig. 7). Obviously, there was an overall 7.13 12.18 10 October inverse relationship between the population density of D. acuminata and the toxin content per cell. A very 0.02 0.33 0.93 1.11 3 October similar relationship (ln OA = 0.75 (0.05) ln cell l1 +5.14 (0.32), F = 240.36, p < 0.0001, r2 = 0.65) appeared when the data pairs including D. acuta also was included in the regression analysis and 26 September the inverse relationship remained. 0.82 0.17 0.17 2.57 Plotted on a linear scale, all data-pairs of cell toxicity and cell number made up a parabolic curve 0.6773 2

19 September (y = 127.59x , r = 0.69) (Fig. 8). On this curve the data-pairs were separated into three groups: (i) D. acuminata occurred in numbers of fewer than approxi- 1 0.07 1.41 0.17 1.25 0.04 xmately x 100 cells l x, and the toxin content x per cell 0.97 was xxxx xxxx xxxxxx x x 12 September 1 always greater than 5 rg cell ; (ii) cell numbers between 100 and approximately 250 cells l1 and a cell toxin content from 5 to 2 rg cell1; and ﬁnally (iii) 1 5 September when the population became greater than 250 cells l 0.13 0.12 0.44 1.25 0.02 0.23 and the toxicity, with few exceptions, was less than 1

7.94 x

2.07 13.348.10 2 rg cell . By applying this subdivision, some clear patterns of the distribution of the differently toxic D.

29 August acuminata became evident. It was found that 90% of the high-and medium-toxicity cells and 26% of the low- 0.59 4.29 0.14 0.70 0.08 1.19 0.19 0.11 toxicity cells were found in the Gullmar Fjord. Consequently, only 10% of the high- and medium-

D. acuminata 22 August toxicity cells and 74% of the low-toxicity cells were found in the Koljo¨ Fjord.

=3)of Further, it was found that in the Gullmar Fjord 64% Fjord site 15 August n ¨ , 0.83 0.36 0.10 0.46 0.05 0.39

1 of the cells in the surface layer was of low toxicity, 33% was of medium toxicity, and only 3% was of high 0.61 x0.46 x x 4.44 x x

(B) Koljo 8 August 15 August 22 August 29 Augusttoxicity. 5 September In 12the September pycnocline 19 September 26 September only 3 October 4% was of low toxicity,

g OA cell 71% was of medium toxicity, and 25% was of high (A) Gullmar Fjord site 8 August r toxicity. Below the pycnocline, 15% was low toxic, 62% was medium, and 23% was high. Thus, the distribution of the cells based on their toxicity can in the Gullmar Fjord be described as: low-toxicity cells dominated at the surface, while medium-toxicity cells Pycnocline 1.65 Surface 0.49 Below pycno-cline x x x 15.63 SurfacePycnocline 3.31 1.21 Table 3 Calculated toxicity ( x means that toxicity per cell could not be calculated. Below pycno-cline 0.22 dominated at and below the pycnocline. Most of the 226 O. Lindahl et al. / Harmful Algae 6 (2007) 218–231

Fig. 4. (A) Combined toxicity per cell (rg OA cell1)ofD. acuminata and D. acuta at the Gullmar Fjord site during the period of investigation. (B) Combined toxicity per cell (rg OA cell1)ofD. acuminata and D. acuta at the Koljo¨ Fjord site during the period of investigation. The contour lines represent isohalines at 1 psu interval. It was estimated that it was only during the second half of October as D. acuta was present in such numbers that the calculation of the toxin content of D. acuminata may have been interfered. high-toxicity cells were found at and below the the toxicity per cell. In these samples it was found that pycnocline, and only a few were found at the surface. D. acuta produced DTX-1 by comparing the co- In the Koljo¨ Fjord the low-toxicity cells dominated at all variation of the DTX-1 concentration with the cell the three depth levels, especially at and below the concentration (F = 27.5, p < 0.005, r2 = 0.80) (Fig. 9). pycnocline, where 96% and 100% of the low-toxicity No co-variation between the OA-concentration and the cells were found, respectively. abundance of D. acuta was found. The calculated toxicity per cell of DTX-1 in the surface samples varied 3.5. Toxicity of D. acuta between 0.40 (0.40) and 7.83 (2.26) rg cell1 (Table 4). Of the pycnocline samples, only three values It was only in the Koljo¨ Fjord samples that the could be calculated, and they were between 0.54 abundance of D. acuta was large enough for calculating (0.44) and 7.32 (1.32) rg cell1.

Fig. 6. The toxicity of D. acuminata cells from the surface water was signiﬁcantly higher (F = 6.26, p < 0.0165) at the Gullmar Fjord site Fig. 5. In the Koljo¨ Fjord D. acuminata cells from the surface water (&) compared to the Koljo¨ Fjord (~) from August until September. It (~) were signiﬁcantly more toxic (F = 14.07, p < 0.0001) compared was estimated that the interference of D. acuta was negligible for this to cells from pycnocline or greater depths (&) during 6 weeks. comparison between the sites of D. acuminata cell toxicity. O. Lindahl et al. / Harmful Algae 6 (2007) 218–231 227

Fig. 8. Plotted on a linear scale, all data-pairs of cell toxicity and cell number made up a parabolic curve (y = 127.59x0.6773, r2 = 0.69). Cell toxicity was separated into three groups: (i) D. acuminata in numbers of fewer than approximately 100 cells l1, and the toxin content always greater than 5 rg cell1; (ii) cell numbers between 100 and approximately 250 cells l1 and a cell toxin content from 5 to 2 rg cell1; and ﬁnally (iii) cell numbers greater than 250 cells l1 and the toxicity, with few exceptions, less than 2 rg cell1.

Fig. 7. A signiﬁcant negative correlation (F = 240.36, p < 0.0001, r2 = 0.69) between the cell content of OA and the cell concentration of D. acuminata was found (ln OA = 0.76 (0.05) ln D. acuminata + 5.18 (0.32)). Cells from the Gullmar Fjord (&) were more toxic (1.2–16.6 rg OA cell1) but less abundant compared to cells from the Koljo¨ Fjord (~) (0.1–2.6 rg OA cell1). The open circles (*) and the dotted line represent a regression (r2 = 0.60) found between cell content of OA and the cell concentration of D. acuminata when data from Danish/German waters, reported by Andersen et al. (1996), was plotted. Fig. 9. In the Koljo¨ Fjord it was found in surface samples that D. acuta produced DTX-1 by comparing the co-variation of the cell concen- 4. Discussion tration (&) with the DTX-1 concentration (~). Bars show standard error (n = 3). 4.1. Population density and cell toxicity and OA production could not be made for the Gullmar In the literature it is stated that both D. acuta and D. Fjord surface samples, which consequently could not be acuminata can produce both OA and DTX-1 (Hallegraeff included in calculations of the OA content per cell of D. et al., 2003). In this study DTX-1 could clearly be coupled acuminata. This is also the explanation why cell toxicity to D. acuta based on the correlation shown in Fig. 9 andan of D. acuminata and D. acuta partly was combined. The analysis of hand-picked D. acuta. This lead to the discrimination of the toxin production between the conclusion that during the present study that the produc- species has although to be further investigated. tion of OA in the Koljo¨ Fjord was dominated by D. When comparing the toxin content per cell of D. acuminata.However,thisdiscriminationbetweenspecies acuminata and D. acuta of the present investigation

Table 4 Calculated toxicity (rg DTX-1 cell1, n =3)ofD. acuta at the Koljo¨ Fjord site Koljo¨ Fjord 8 15 22 29 5 12 19 26 3 August August August August September September September September October Surface 1.43 0.22 1.16 0.21 0.45 0.25 1.10 0.06 1.88 0.49 2.05 0.53 2.93 0.31 7.83 2.26 0.40 0.40 Pycnocline 0.54 0.54 x x x x x 3.94 2.27 x 7.32 1.35 Below pycno-cline x x x x x x x x x x means that toxicity per cell could not be calculated. 228 O. Lindahl et al. / Harmful Algae 6 (2007) 218–231 with that of other studies, it became evident that, despite warning when cell concentration exceed a given high variability in toxicity per cell of the species, the number. The apparent variability of toxin content per variable cell toxicities found in the Gullmar and Koljo¨ cell might partly explain the discrepancy between the Fjords during autumn 2001 fell well within the levels occurrence of D. acuminata in the water column and the reported in the literature (Tables 1 and 2). Further, toxicity of blue mussels (Andersen et al., 1996). Maestrini (1998) reviewed the bloom dynamics and the Different and persistent toxicity per cell between ecophysiology of Dinophysis and concluded that the surface (higher toxicity) and pycnocline populations most common feature in Dinophysis blooms appeared (lower toxicity) of D. acuminata was observed during 6 to be the formation of a dense, patchy population. The weeks in the Koljo¨ Fjord. The toxicity varied from less majority of all peak abundances was associated with the than 2 to as high as 15 rg cell1. The difference in pycnocline of a stratified water column and thus not toxicity of cells from the surface water at Gullmar and randomly distributed throughout the water column. The Koljo¨ Fjords taken the same day varied during the present investigation also found conspicuous accumu- sampling period by a factor between 1.3 and 16.8. The lations of D. acuminata in the pycnocline at the present results have demonstrated that approximately secluded Koljo¨ Fjord site but not at the hydrodynami- 100 high-toxicity cells in a low-density population may cally much more active Gullmar Fjord site. A likely lead to the same accumulation of DST in a mussel as the explanation for this difference could be the weaker ingestion of 1500 low-toxicity cells from a high-density stratification at the outer site compared to the inner site. population from the same water column. Thus, the The most obvious and conspicuous result of the product of cell abundance and the corresponding cell present study was the clear negative correlation between toxicity will be decisive for the DST content of mussels the population density and the toxicity per cell of D. from the two close sites, presupposing that the filtration acuminata. This relationship was more or less present rate of the mussels was the same. regardless of site and depth during a period of 2 months. SPATT, a new method of monitoring the occurrence of Surprisingly, no observations on any relationship toxic algal blooms in relation to shellfish contamination between population density and cell toxicity have been events, has recently been introduced by MacKenzie et al. described in the literature. Consequently, the results (2004). The problem with varying cell toxicity may be presented here seem to be novel. overcome with SPATT, since the method, involves the When published data from Danish and German waters passive adsorption of biotoxins onto porous synthetic (Andersen et al., 1996) were plotted, a rather similar resin-filled sachets and their subsequent extraction and relationship between population density and cell toxicity analysis. Field trials during D. acuminata blooms proved was found as in the present study (Fig. 7). The somewhat that the technique provides a means of forecasting lower toxicity per cell in the Danish/German data was shellfish contamination events and can predict the net probably due to a difference in the toxin analyze. It can accumulation of polyether toxins by mussels. The thus be concluded that a similar relationship between cell method is founded on the observation that during algal toxicity and population density may also be found in at blooms significant amounts of toxins are dissolved in the least Danish and German coastal waters. seawater (MacKenzie et al., 2004). The toxin content of In order to find an explanation for the ecological and/ the D. acuminata cells was not measured, and cell density or physiological processes behind the cell toxicity during these field trials varied by a factor of at least 10. If variation with the population density, a multi-variable it is presupposed that the cell toxicity varied with regression analysis on site (fjord), depth, temperature, population density as in the present investigation, it can salinity, nutrients, chlorophyll, Dinophysis abundance, be assumed that there was also a considerable span in and cell toxicity was carried out. This analysis did not cell toxicity, which the SPATT method seems able to produce any clear answer that could further explain the sample and measure. Consequently, the SPATT method, observed toxicity pattern. Consequently, it can at combined with cell toxicity measurements, may be a present only be speculated what mechanisms or way to further investigate the relationship between cell processes may have resulted in the varying cell toxicity toxin content and dissolved toxin concentration in the as a function of the population density. water.

4.2. Monitoring and cell toxicity 4.3. Cell growth and toxicity

The traditional way to forecast DST in shellfish is to From laboratory studies it is well known that the monitor the occurrence of the harmful algae and give a toxin production of toxic dinoflagellates vary with time O. Lindahl et al. / Harmful Algae 6 (2007) 218–231 229 during the growth and the toxicity usually increases possible to speculate about how the growth conditions with lengthening of culture time and then decreases throughout the water column may have influenced the (Hwang and Lu, 2000). However, Boyer et al. (1987) cell toxicity. and Anderson et al. (1990) found that the maximum Of special interest was the observed persistent amount of toxin produced was in the mid-exponential difference in cell toxicity in Koljo¨ Fjord between cells phase, while White (1978) and Oshima et al. (1989) from the surface layer and the pycnocline. Nutrient have reported that it was in the stationary phase of samples were taken once a month and at hydrographical growth. In laboratory studies of Alexandrium minutum standard depths. This low sampling frequency involved cell and total toxicity reached the maximum level at the that direct correlations between the nutrient concentra- post-stationary growth phase and decreased quickly tions in the water and the toxin content of the cells were (Hwang and Lu, 2000). Ogata et al. (1989) studied the not possible to perform. Johansson et al. (1996) effect of water temperature and light intensity on suggested that the greatest toxin production of D. growth rate and toxin production of three different toxin acuminata was observed when the growth was limited producing dinoflagellates. They found that the toxin by nitrogen. The nitrogen concentration in the surface content of the three species increased concomitantly water was below the detection limit during August and with the decrease of the growth rate when the growth September and very low (0.3 mmol l1) at the beginning was controlled by temperature. However, the toxin of October. At the same time, 5–10 mmol l1 of production decreased or did not increase significantly nitrogen were found at depths at or below 10 m. It is when the growth rate decreased by lowering the light thus possible that the difference in persistent toxicity intensity. It was concluded that the results indicated that per cell between the surface and pycnocline cells was a photosynthesis was essential for the toxin production of result of the different nitrogen concentrations or lower the species studied. N/P ratios in the surface water compared to greater For Dinophysis spp., species well known not to allow depths. Unfortunately, the data supply did not allow for cultivation in the laboratory, there are only field data a similar comparison for the Gullmar Fjord. available. Gisselsson et al. (2002) have shown that D. norvegica can grow and persist in the Baltic Sea 4.4. Cell toxicity signal system thermocline at a depth of around 20 m during the summer months, when it can attain growth rates of at The chemical signals that determine interactions or least 0.4 per day. Further, they found that photosynthesis reactions between unicellular microbes of the marine could support carbon to sustain growth rates of up to 0.1 environment are still poorly known (Wolfe, 2000). In per day at the depths of cell abundance; this lead to the particular, chemical signals between microbial pre- conclusion that Baltic D. norvegica was growing dators and prey contribute to food selection or primarily by heterotrophic means at high growth rates avoidance or to defense, factors that probably affect and that the reason for cell accumulation in the trophic structure and such large-scale phenomenon as thermocline could be to sequester inorganic nutrients. algal blooms. However, the physical constrains on the It is likely that the observed accumulation of D. transmission of chemical information, and strategies acuminata in and below the pycnocline of the present and mechanisms that microbes might use to send study could have a similar explanation as for D. chemical signals in a low Reynold number, viscosity- norvegica in the Baltic, since the secchi depth in the dominated physical environment are transferred by Koljo¨ Fjord was around 5 m, and thus available light at molecular diffusion and laminar advection. The signals the depth of the pycnocline was far below 1% of the may be perceived at nanomolar levels or lower. These light level at the surface. In that case, it can be assumed strategies include compartmented and activated reac- that the growth rate of D. acuminata was probably tions, utilizing both pulsed release of dissolved signals rather high; however, it seemed likely to suppose that and contact-activated signals at the cell surface. Many the small number of cells found in the surface layer marine protozoan grazers are known to selectively were autotrophic, slow growing and toxin producing, ingest or reject particular prey and different selective while the large number of cells found in the pycnocline feeding mechanisms can have very different conse- most likely were heterotrophic or mixotrophic, fast quences for the long-term persistence of prey popula- growing and more or less not toxin producing. The tions (Wolfe, 2000 and references therein). lack of information on the occurrence of organic In Maestrini’s review (1998) on bloom dynamics and nutritional sources (e.g., ciliates) at the depth of the ecophysiology of Dinophysis spp., it was concluded that pycnocline population of Dinophysis made it only there was no evidence for the existence of a true 230 O. Lindahl et al. / Harmful Algae 6 (2007) 218–231 allelopathic mechanism: i.e., the production of an Acknowledgements inimical substance acting at some distance from the producing cell. However, Carlsson et al. (1995) The authors would like to thank Dr. Lars-Ove Loo suggested the possible presence of a defense mechan- and Dr. Fredrik Noreń for statistical advice on the ism due to the delayed effect of the sequestered toxin. In sampling design and Dr. Samuel Dupont for statistical order for toxicity to evolve as a defense mechanism, it analyses and valuable comments to the manuscript. We would appear that the individual that produces are further grateful to Dr. Fredrik Noreń for comments intracellular toxic substances should not be eaten. to the manuscript and to BSc. Joakim Strickner for Nevertheless, it could be evolutionarily beneficial for a making the surfer-plots and to two unknown referees. population to lose some of its individuals in order to This work was supported by grants from EC-project decrease the grazing pressure on other cells. The HABES (contract EVK2-CT-2000-00092), Stiftelsen Dinophysis population could thus significantly diminish Birgit and Birger Wa˚hlstro¨ms minnesfond and Carl losses from grazing by following the strategy that some Trygger Foundation for Scientific Research.[SS] individuals are sacrificed in order to destroy or to deter predators (Carlsson et al., 1995). Further, laboratory References experiments that used a variety of microalgae, including a clone of the OA and DTX-1 producer Prorocentrum Anderson, D.M., Kulis, D.M., Sullivan, J.J., Hall, S., 1990. Toxin lima, studied the effects of DST toxins on marine composition variations in one isolate of the dinoflagellate Alex- organisms or their biological function (Windust et al., andrium fundyense. Toxicon 28, 885–893. Andersen, P., Benedicte, H., Emsholm, H., 1996. Toxicity of Dino- 1996). It was found that OA inhibited the growth of all physis acuminata in Danish coastal waters. In: Yasumoto, T., non-DST-producing test species at micromolar con- Oshima, Y., Fukuyo, Y. (Eds.), Harmful and Toxic Algal Blooms. centrations whereas the DST producing P. lima was not Intergovernmental Oceanographic Commission of UNESCO, effected. It was concluded that the DST toxins may play Paris, pp. 281–284. ¨ an allelopathic role. Aune, T., Strand, O., Aase, B., Weidemann, J., Dahl, E., Hovgaard, P., 1996. The Sognefjord in Norway, a possible location for mussel If the produced DST toxins in Dinophysis is simply farming? In: Yasumoto, T., Oshima, Y., Fukuyo, Y. (Eds.), not just a by-product, it can be assumed that the toxin Harmful and Toxic Algal Blooms. Intergovernmental Oceano- production costs the cell energy, and over-production is graphic Commission of UNESCO, Paris, pp. 73–76. costly. In order to reach toxin levels high enough to give Belgrano, A., Lindahl, O., Hernroth, B.,1999. North Atlantic Oscillation an allelopathic effect on the surroundings, few cells (NAO), primaryproductivityandtoxicphytoplanktonintheGullmar Fjord, Sweden (1985–1996). Proc. R. Soc. Lond. B 266, 425–430. have to be more toxic compared to many cells, which Bjo¨rk, G., Liungman, O., Rydberg, L., 2000. Net circulation and may be an explanation for the present findings of salinity variations in an open-ended Swedish Fjord System. Estu- inversed cell toxicity in relation to population density. aries 23, 367–380. Such an explanation involves some kind of signal Blanco, J., Fernańdez, M., Marino, J., Reguera, B., Miguez, A., system in which the individual cell could ‘‘measure’’ Maneiero, J., Cacho, E., Marinez, A., 1995. From Dinophysis spp. toxicity to DSP outbreaks: a preliminary model of toxin the actual population density. This is known from accumulation in mussles. In: Lassus, P., Arzul, G., Erard-Le Denn, bacteria and is termed quorum sensing (also called E., Gentien, P., Marcaillou-Le Baut, C. (Eds.), Harmful Marine autoinduction) and is an environmental sensing system Algal Blooms. Lavoisier Publishing, Paris, pp. 777–782. that allows the cells to monitor their own population Boyer, G.L., Sullivan, J.J., Harrison, P.J., Taylor, F.J.R., 1987. Effect of density (Brelles-Marinõ and Bedmar, 2001). Auto- nutrient limitation on toxin production and composition in marine dinoflagellate Protogonyaulax tamarensis. Mar. Biol. 96, 123–128. induction relies upon the interaction of small diffusible Brelles-Marinõ, G., Bedmar, E.J., 2001. Detection, purification and signal molecule with a transcriptional activator protein characterisation of quorum-sensing signal molecules in plant- to couple gene expression with cell abundance. associated bacteria. J. Biotechnol. 91, 197–209. Recent investigations in Norway have found by Carlsson, P., Grane´li, E., Finenko, G., Maestrini, S.Y., 1995. Copepod using the passive adsorption technique (SPATT) that grazing on a phytoplankton community containing the toxic dinoflagellate Dinophysis acuminata. J. Plankt. Res. 17, 1925–1938. when sampling toxic Dinophysis cells, okadaic acid and Carmody, E.P., James, K.J., Kelly, S.S., 1996. Dinophysistoxin-2: the other toxins were found dissolved in the seawater (Dr. predominant diarrhoetic shellfish toxin in Ireland. Toxicon 34, 351– Morten Sandvik, National Veterinary Institute, Norway; 359. pers. commun.). It can thus be hypothesized that the Cembella, A.D., 1989. Occurrence of okadaic acid a major shellfish toxins produced by the Dinophysis cells to some extent toxin in natural populations of Dinophysis spp. from the eastern coast of North America. J. Appl. Phycol. 1, 307–310. are released into the water and that, if the single cell is Dahl, E., Johannessen, T., 2001. Relationship between occurrence of experiencing a low toxicity in the environment, it Dinophysis species (Dinophyceae) and shellfish toxicity. Phyco- becomes more toxic and vice versa. logia 40, 223–227. O. Lindahl et al. / Harmful Algae 6 (2007) 218–231 231

Daranas, A.H., Norte, M., Fernandez, J.J., 2001. Toxic marine micro- Maestrini, S.Y., 1998. Bloom dynamics and ecophysiology of Dino- algae. Toxicon 39, 1101–1132. physis spp. In: Anderson, D.M., Cembella, A.D., Hallegraeff, Draisci, R., Lucentini, L., Giannetti, L., Marchiafava, C., James, K.J., G.M. (Eds.), Physiological Ecology of Harmful Algal Blooms, Kelly, S.S., 1998. Diarrhoeic shellfish toxin profiles in phyto- NATO-ASI, vol. 41. Springer-Verlag, Heidelberg, pp. 243–265. plankton and mussels from Italy and Ireland as determined by LC- Morono, A., Arevalo, F., Fernandez, M.L., Maneiro, J., Pazos, Y., MS.MS. In: Reguera, B., Blanco, J., Fernandez, M.L., Wyatt, T. Salgado, C., Blanco, J., 2003. Accumulation and transformation of (Eds.), Harmful Algae. Intergovernmental Oceanographic Com- DSP toxins in mussels Mytilus galloprovincialis during a toxic mission of UNESCO, pp. 495–498. episode caused by Dinophysis acuminata. Aquat. Toxicol. 62, Edebo, L., Lange, S., Li, X.P., Allenmark, S., Lindgren, K., Thomp- 269–280. son, R., 1988. Seasonal, geographic and individual variation of Ogata, T., Kodama, M., Ishimara, T., 1989. Effect of water tempera- okadaic acid content in cultivated mussels in Sweden. Apmis 96, ture and light intensity on growth rate and toxin production in toxic 1036–1042. dinoflagellates. In: Okaichi, T., Anderson, D.M., Nemoto, T. Edler, L., Hageltorn, M., 1990. Identification of the causative organ- (Eds.), Red Tides: Biology, Environmental Science and Toxicol- ism of a DSP-outbreak on the Swedish west coast. In: Graneli, E., ogy. Elsevier, New York, pp. 423–426. Sundstro¨m, B., Edler, L., Andersson, D.M. (Eds.), Toxic Marine Oshima, Y., Hirota, M., Yasumoto, T., Hallegraff, G.M., Blackburn, Phytoplankton. Elsevier Science Publishing Co. Inc., Lund, S.I., Steffensen, D., 1989. Production of paralytic shellfish toxin Sweden, pp. 345–349. by the dinoflagellate Alexandrium minutum Halim from Australia. Gisselsson, L.-A˚ ., Carlsson, P., Grane´li, E., Pallon, J., 2002. Dino- Nippon Suisan Gakkaishi 55, 925. physis blooms in the deep euphotic zone of the Baltic Sea: do they Pavela-Vrancic, M., Mestrovic, V., Marasovic, I., Gillman, M., Furey, grow in the dark? Harmful Algae 1, 401–418. A., James, K.J., 2002. DSP toxin profile in the coastal waters of the Godhe, A., Svensson, S., Rehnstam-Holm, A.S., 2002. Oceanographic central Adriatic Sea. Toxicon 40, 1601–1607. settings explain fluctuations in Dinophysis spp. and concentrations Quilliam, M.A., 2003. Chemical methods for lipophilic shellfish of diarrhetic shellfish toxin in the plankton community within a toxins. In: Hallegraeff, G.M., Anderson, D.M., Cembella, A.D. mussel farm area on the Swedish west coast. Mar. Ecol. Prog. Ser. (Eds.), Manual on Harmful Marine Microalgae. Monographs on 240, 71–83. Oceanographic Methodology. UNESCO Publishing. Graneli, E., Anderson, D.M., Carlsson, P., Maestrini, S.Y., 1997. Light Rehnstam-Holm, A.S., Hernroth, B., 2005. Shellfish and public and dark carbon uptake by Dinophysis species in comparison to health: a Swedish perspective. Ambio 34 (2), 139–144. other photosynthetic and heterotrophic dinoflagellates. Aquatic Riisgaard, H.U., 1991. Filtration rate in the blue mussel Mytilus edulis Microb. Ecol. 13, 177–186. Linneaus, 1758: dependence on algal concentration. J. Shellfish Haamer, J., Andersson, P.-O., Lindahl, O., Lange, S., Li, X.P., Edebo, Res. 10, 29–35. L., 1990. Geographic and seasonal variation of okadic acid content Rydberg, L., 1977. Circulation in the Gullmaren—a sill fjord with in farmed mussels, Mytilus edulis Linnaeus, 1758, along the externally maintained stratification. Inst. of Oceanography, Univ. Swedish west coast. J. Shellfish Res. 9, 103–108. of Go¨teborg, Rep. no. 23 (mimeo). Hallegraeff, G.M., Anderson, D.M., Cembella, A.D. (Eds.), 2003. SAS Institute Inc., 1989. SAS/STAT User’s Guide, vol. 1. SAS Manual on Harmful Marine Microalgae. Monographs on Oceano- Institute Inc., Cary, NC, p. 943. graphic Methodology. UNESCO Publishing. Shapiro, S.S., Wilk, M.B., 1965. An analysis of variance test for Hwang, D.F., Lu, Y.H., 2000. Influence of environmental and nutri- normality (complete samples). Biometrika 52, 591–611. tional factors on growth, toxicity, and toxin profile of dinoflagel- Sokal, R.R., Rohl, F.J., 1995. Biometry. W.H. Freeman & Co., New late Alexandrium minutum. Toxicon 38 (11), 1491–1503. York, p. 887. Kevin, J.J., Furey, A., Healy, B., Kelly, S.S., Twohig, M., 1998. Svensson, S., 2003a. Depuration of okadaic acid (diarrhetic shellfish Luminarin-3, a new fluorometric reagent for the HPLC determi- toxin) in mussels, Mytilus edulis, feeding on different quantities of nation of diarrhoeic shellfish poisoning toxins. In: Reguera, B., nontoxic algae. Aquaculture 218 (1–4), 277–291. Blanco, J., Fernandez, M.L., Wyatt, T. (Eds.), Harmful Algae. Svensson, S., 2003. Effects, dynamics and management of okadaic Intergovernmental Oceanographic Commission of UNESCO, pp. acid in blue mussels, Mytilus edulis. Thesis at Dep. of Zoology/ 519–520. Zoophysiology, Go¨teborg University. Johansson, N., Grane´li, E., Yasumoto, T., Carlsson, P., Legrand, C., Utermo¨hl, H., 1958. Zur Vervollkommnung der quantitativen Phyto- 1996. Toxin production by Dinophysis acuminata and D. acuta plankton-Methodik. Mitt. Int. Veriein. Limnol. 9, 1–38. cells grown under nutrient sufficient and deficient Conditions. In: Vale, P., Sampayo, M.A.D., 2000. Dinophysistoxin-2: a rare diarrhoeic Yasumoto, T., Oshima, Y., Fukuyo, Y. (Eds.), Harmful and Toxic toxin associated with Dinophysis acuta. Toxicon 38, 1599–1606. Algal Blooms. Intergovernmental Oceanographic Commission of White, A.W., 1978. Salinity effects on growth and toxin content of UNESCO, Sendai, Japan, pp. 277–280. Gonyaulax excavata, a marine dinoflagellate causing paralytic Lee, J.S., Igarashi, T., Fraga, S., Dahl, E., Hovgaard, P., Yamamoto, T., shellfish poison. J. Phycol. 14, 475–479. 1989. Determination of diarrhetic shellfish toxins in various Windust, A.J., Wright, J.L.C., McLachlan, J.L., 1996. The effects of dinoflagellate species. J. Appl. Phycol. 1, 147–152. the diarrhetic shellfish poisoning toxins, okadaic acid and dino- Lindahl, O., 1987. Plankton community dynamics in relation to water physistoxin-1, on the growth of microalgae. Mar. Biol. 126, 19–25. exchange in the Gullmar Fjord, Sweden. PhD Thesis. Department Wolfe, G.V., 2000. The chemical defence ecology of marine unicel- of Zoology, Stockholm University. lular plankton: constraints, mechanisms, and impacts. Biol. Bull. MacKenzie, L., Beuzenberg, V., Holland, P., McNabb, P., Selwood, 198, 225–244. A., 2004. Solid phase adsorption toxin tracking (SPATT): a new Yasumoto, T., Murata, M., Oshima, Y., Sano, M., Matsumoto, G.K., monitoring tool that simulates the biotoxin contamination of filter Clardy, J., 1985. Diarrhetic shellfish toxins. Tetradron 41, 1019– feeding bivalves. Toxicon 44, 901–918. 1025.