, Volume 655, Issue 1, pp 89–97

Characterization of paralytic shellfish toxins in seawater and sardines (Sardina pilchardus) during blooms of Gymnodinium catenatum


    • IPIMAR - National Institute for Biological Resources
  • Maria João Botelho
    • IPIMAR - National Institute for Biological Resources
  • Kathi A. Lefebvre
    • NOAA FisheriesNorthwest Fisheries Science Center, Marine Biotoxins Program
Primary research paper

DOI: 10.1007/s10750-010-0406-5

Cite this article as:
Costa, P.R., Botelho, M.J. & Lefebvre, K.A. Hydrobiologia (2010) 655: 89. doi:10.1007/s10750-010-0406-5


The re-emergence of Gymnodinum catenatum blooms after a 10 year hiatus of absence initiated the present investigation. This study aims to evaluate the exposure of small pelagic fishes to paralytic shellfish toxins (PST) during blooms of G. catenatum. Sardines (Sardina pilchardus) were selected as a representative fish species. In order to assess toxin availability to fish, both intracellular PSTs (toxin retained within the algal cells) and extracellular PSTs (toxin found in seawater outside algal cells) were quantified, as well as toxin levels within three fish tissue matrices (viscera, muscle and brain). During the study period, the highest cell densities of G. catenatum reached 2.5 × 104 cells l−1 and intracellular PST levels ranged from 3.4 to 398 ng STXeq l−1 as detected via an enzyme linked immunosorbent assay (ELISA). Measurable extracellular PSTs were also detected in seawater (0.2–1.1 μg STXeq l−1) for the first time in Atlantic waters. The PST profile in G. catenatum was determined via high performance liquid chromatography with fluorescence detection (HPLC-FLD) and consisted mostly of sulfocarbamoyl (C1+2, B1) and decarbamoyl (dcSTX, dcGTX2+3, dcNEO) toxins. The observed profile was similar to that reported previously in G. catenatum blooms in this region before the 10-year hiatus. Sardines, planktivorous fish that ingest a large number of phytoplankton cells, were found to contain PSTs in the viscera, reaching a maximum of 531 μg STXeq kg−1. PSTs were not detected in corresponding muscle or brain tissues. The PST profile characterized in sardine samples consisted of the same sulfocarbamoyl and decarbamoyl toxins found in the algal prey with minor differences in relative abundance of each toxin. Overall, the data suggest that significant biotransformation of PSTs does not occur in sardines. Therefore, planktivorous fish may be a good tracer for the occurrence of offshore G. catenatum blooms and the associated PSTs produced by these algae.


Gymnodinium catenatumVectorSardinesSaxitoxinsParalytic shellfish poisoningDissolved and particulatePlanktivorous fish


Many events leading to large-scale mortalities of marine fauna have been associated with harmful algal blooms (Landsberg, 2002). Harmful algae can cause mortality and affect marine fauna through both physical and chemical mechanisms (Smayda, 1997), including the production of toxic substances such as the paralytic shellfish toxins (PSTs). In the marine environment they are produced by select dinoflagellate species in the genera Gymnodinium, Alexandrium and Pyrodinium (Hallegraeff, 1993). More than 30 related tetrahydropurine compounds belong to the PST family (Shimizu, 2000; Llewellyn, 2006). In general, these toxins can be divided into three groups based on their side chain chemical structure: the carbamoyl, the decarbamoyl and the sulfocarbamoyl toxins. Toxicity occurs via the blockage of voltage-gated sodium channels and can result in death by respiratory arrest and cardiovascular shock in mammals (Lagos & Andrinolo, 2000). However, due to side chain variability, each toxin has a different binding affinity to voltage-gated sodium channel receptors, which results in differing toxicities (Genenah & Shimizu, 1981). The carbamoyl group, which includes saxitoxin (STX), neosaxitoxin (NEO) and the gonyautoxins (GTX 1-4), contains the most toxic compounds. The sulfocarbamoyl group is the least toxic and includes C1-4 and also B1 (GTX5) and B2 (GTX6). The decarbamoyl group has an intermediate toxicity and includes the decarbamoyl derivatives of STX (dcSTX), GTX (dcGTX) and NEO (dcNEO) (Oshima, 1995).

The toxin-producing dinoflagellate, Gymnodinium catenatum, was associated with paralytic shellfish poisoning (PSP) outbreaks in Portuguese waters in the late 1980s and early 1990s (Franca & Almeida, 1989). However, after 1995, blooms of G. catenatum were not observed until September of 2005 (Moita et al., 2006). Since the 10-year hiatus, the occurrence of G. catenatum blooms has been persistent, leading to frequent closures of shellfish harvesting due to PST contamination.

Planktivorous marine organisms such as bivalves and small pelagic fish have two main roles in the food web in relation to PST exposure and impact. These organisms may act as vectors of PSTs to higher trophic levels, as victims of the toxins or as both. To date, the role of bivalves as vectors and victims of PST toxicity and exposure have been the most well studied (e.g. Shumway, 1990; Bricelj & Shumway, 1998). Examples of shellfish vectoring incidents due to accumulation of PSTs have been reported worldwide (IPCS, 1984; Gessner & Middaugh, 1995; Sampayo et al., 2001; Batoréu et al., 2005; Garcia et al., 2005). Additionally, bivalve molluscs are themselves susceptible to PSTs. It has been shown that PST exposure can alter burrowing behaviour and shell valve closure, produce histopathological lesions, impair immune responses and impair signal transmission between neurons (Bricelj et al., 1996, 2005; Galimany et al., 2008).

Although not as well documented, fish have also been shown to be impacted by PSTs (White, 1984). Exposure of fish to these toxins has been shown to affect adult and larval survival, as well as impair sensorimotor function (White, 1981; Robineau et al., 1991; Lefebvre et al., 2005). Fish kills associated with PST-producing dinoflagellates have been reported in several countries including the United Kingdom, Canada and Argentina (Adams et al., 1968; White, 1977; Montoya et al., 1996). Planktivorous fish can accumulate PSTs by direct ingestion of toxic phytoplankton or by the consumption of zooplanktonic organisms that have fed on toxic algae (White, 1984; Samson et al., 2008). Interestingly, farm raised salmonids, fed an artificially administered diet, have also experienced mortality during blooms of PST-producing dinoflagellates (Mortensen, 1985). PSTs were not detected in liver tissue or the digestive tract of the affected farmed salmon, and the cause of mortality was attributed to direct exposure to toxic cells and/or to soluble toxins released during the bloom (Cembella et al., 2002).

In addition to highly visible PST toxic events in fish, such as those described above involving mass mortalities, fish can also accumulate or transfer toxins without outward signs of toxicity. Under these conditions, fish can play a potent vectorial role and transfer toxins to piscivorous predators, leading to toxicity events in top predators such as marine mammals (Geraci et al., 1989; Castonguay et al., 1997; Reyero et al., 1999).

The goal of the present study was to characterize the relationship between G. catenatum cell density, toxin content and fish contamination. To this end, a field study was performed to determine both the levels and profiles of PSTs present in sardines (Sardina pilchardus), as well as in G. catenatum cells (intracellular PSTs) and seawater (extracellular PSTs). The toxin profiles observed in algal prey and fish were compared to determine if significant biotransformation occurs in fish. Additionally, the tissue distribution of PSTs was examined in fish muscle, viscera and whole brain for a better understanding of toxicological implications.

Materials and methods

Collection and preparation of fish samples

Portuguese coast, located between 37° and 42°N, is at the northern limit of the North Atlantic Upwelling System. Sardine samples were obtained by pelagic trawling (depth range: 30–55 m) along the northwest (NW) and the south (S) coasts of Portugal between October 27th–November 3rd and November 11th–16th 2007, respectively, on board the R/V Noruega of IPIMAR (Fig. 1). Fifteen specimens were obtained from each trawl, and immediately frozen separately in bags at −20°C. Within 2 weeks of collection, samples were divided into three sets of five specimens each. The body weight and total length of each animal was recorded. The brain, white muscle and visceral tissues were dissected under partial defrost conditions to minimize toxin leakage between tissues. These tissues were weighed and homogenized. A 5-g portion was taken for toxin extraction procedures.
Fig. 1

Sampling stations (1–10) for seawater and sardine (Sardina pilchardus) along the Portuguese coast (bathymetric contour line = 100 m). Stations 1–7 represent the Northwest (NW) Portuguese coast and stations 8–10 represent the South (S) Portuguese coast

Collection of seawater samples

Surface seawater samples (depth = 3 m) were collected by a submersible pump in each sampling station during each sardine trawling. Water samples passed through a 200 μm plankton net to exclude large plankton and were prepared as described in Lefebvre et al. (2008). Briefly, for analysis of intracellular PSTs, 1.5 l of seawater was filtered under light vacuum pressure (100 mm Hg) on 0.45 μm pore size HA membrane filters (Durapore Millipore Inc.) and frozen at −20°C. An aliquot (5 ml) of the 0.45 μm filtrate was collected and frozen at −20°C for subsequent determination of extracellular PSTs. Leakage of cellular-bound PSTs into the extracellular media is not expected via this filtration process, based on vacuum studies with dinoflagellate Alexandrium spp. (Lefebvre et al., 2008).

The collected seawater samples (250 ml) were preserved in a 3% formalin solution for subsequent quantification of G. catenatum cell density using light microscopy.

Quantification of phytoplankton cell density

Cell densities of G. catenatum were quantified via manual cell counts. Cells in subsamples (100 ml) of preserved seawater were concentrated by settling (24 h) and counted in a Palmer-Maloney chamber under a Zeiss light microscope (×400). The detection limit was 20 cells l−1 (Utermöhl, 1931).

Determination of intracellular and extracellular PST levels via ELISA

In order to quantify intracellular PST levels, frozen plankton samples harvested in HA membrane filters as described above were extracted in 5 ml of 0.05 M acetic acid by sonication for 4 min at 25 W, 50% pulse duty cycle (Branson Sonifier 450) in an ice bath. The extract was then filtered (0.45 μm) to remove physical particles and analyzed for determination of intracellular PSTs via an enzyme linked immunosorbent assay (ELISA) and for determination of the toxin profile via HPLC-FLD. The protocol provided by the ELISA manufacturer (Abraxis LLC, USA) was followed; however, samples were diluted up to 1:250 in order to bring the samples within the working range of the assay and to eliminate matrix effects. After performing a dilution series with a control sample, a minimum dilution of 1:50 was established to eliminate matrix effects for intracellular PST samples. In order to quantify extracellular PST levels, frozen filtrates from seawater samples were thawed and analyzed via ELISA after a 1:10 dilution.

The Abraxis ELISA has a detection limit of 0.05 nM for STX and cross-reactivities of <0.2% for GTX1+4, 0.6% for dcNEO, 1.3% for NEO, 23% for GTX2+3, 23% for GTX5, 29% dcSTX and 100% for STX, as is stated by the manufacturer.

Characterization of PST profiles in algal and fish samples via HPLC-FLD

Extractions were carried out according to the method approved by the Association of Official Analytical Chemists (AOAC) for determination of PSTs in shellfish (Lawrence et al., 2005). Toxins from fish homogenate were heat-extracted in 1% acetic acid and toxins from algal samples were extracted as reported above for ELISA. Both extracts followed a solid-phase extraction (SPE) clean-up with an octadecyl bonded phase silica (Supelclean LC-18 SPE cartridge, 3 ml, Supelco, USA). Periodate and peroxide oxidations of PSTs were carried out and non-hydroxylated toxins (STX, C1+2, B1, dcSTX, GTX2+3) were immediately quantified by high-performance liquid chromatography with fluorescence detection (HPLC-FLD). Samples containing N-1-hydroxylated PSTs (NEO, dcNEO, GTX1+4, C3+4 and B2) were further purified using an SPE ion exchange cartridge with carboxylic acidsiliane (COOH) bonded to silica gel (Bakerbond COOH, 3 ml, J.T. Baker, USA).

The HPLC-FLD equipment consisted of a Hewlett-Packard/Agilent Model 1050 quaternary pump, Model 1100 in-line degasser, autosampler, column oven and Model 1200 fluorescence detector. The PST oxidation products were separated using a reversed-phase Supelcosil LC-18, 15 × 4.6, 5 μm column (Supelco, USA). The mobile phase gradient consisted of 0–5% B (0.1 M ammonium formate in 5% acetonitrile, pH = 6) in the first 5 min, 5–70% B for the next 4 min and back to 0% B in the next 2 min. Before the subsequent injection, 100% mobile phase A (0.1 M ammonium formate, pH = 6) was used for 3 min. Flow rate was 1 ml/min and the detection wavelength set to 340 nm for excitation and 395 nm for emission. Certified calibration solutions for PSTs were purchased from the Certified Reference Materials Program of the Institute for Marine Biosciences, National Research Council, Canada (STX-e, NEO-b, GTX2&3-b, GTX1&4-b, dcSTX, dcGTX2&3, GTX5-b (B1), C1&2 and dcNEO-b). The instrumental limits of detection (LOD) ranged from 0.002 μM for C1 + 2, B1 and STX to 0.02 μM for dcNEO. The toxicity factors stated by Quilliam (2007) were used for calculation of PSTs in terms of saxitoxin dihydrochloride equivalents (Quilliam, 2007). Since these are the units commonly used for regulation of shellfish consumption they were chosen in this study to express concentration of PSTs in sardines. For toxins that co-eluted and were measured together (C1+2; dcGTX2+3; GTX1+4 and GTX2+3), the highest toxicity factor of the group was used. Based on decisions of the European Community Reference Laboratory for Marine Biotoxins, the toxicity factor of dcSTX was used for dcNEO.


The non-parametric Mann–Whitney test was applied for comparison between toxin levels in G. catenatum and sardines.


Gymnodinium catenatum cell densities and toxin determination in seawater via ELISA

Gymnodinium catenatum was found in all samples collected from the NW coast (stations 1–7), with cell densities ranging from 220 to 24,500 cells l−1. However, cells of G. catenatum were not observed in seawater samples collected from the S coast (stations 8–10). Intracellular PSTs (0.45–200 μm) were detected in each sample containing G. catenatum cells, with values ranging from 3.4 to 398 ng STXeq l−1. A strong correlation between cell densities and particulate PSTs was found (r2 = 0.917, n = 10) (Fig. 2). Additionally, dissolved PSTs, as measured via ELISA, were detected in each sample collected from the NW coast (Fig. 3). Dissolved PSTs ranged from 200 to 1080 ng STXeq l−1 and did not show a clear relationship to G. catenatum cell densities. No dissolved PSTs were detected in samples from the S coast.
Fig. 2

Correlation between G. catenatum cell density (103 cell l−1) and intracellular paralytic shellfish toxins (PSTs) as determined via ELISA (y = 16.695x + 9.79, r2 = 0.917, n = 10)

Fig. 3

Intracellular and extracellular paralytic shellfish toxins (ng STXeq l−1) as determined via ELISA in seawater samples collected from southward (station 7) to northward (station 1) along the NW Portuguese coast. Open square intracellular, filled square extracellular. Values over bars represent G. catenatum cell count

Toxin profile produced by Gymnodinium catenatum

The highest intracellular PST concentrations were detected in samples collected from stations 3 to 6, and those samples were used for determination of the profile of PSTs via HPLC-FLD. The profile consisted of the sulfamate congeners C1+2 and B1, and the decarbamoyl congeners dcSTX, dcGTX2+3 and dcNEO (Fig. 4). The suite of PSTs determined in G. catenatum in decreasing order of relative abundance was: C1+2 > dcSTX ≥ B1 > dcGTX2+3 ≥ dcNEO (Fig. 4). Notably, G. catenatum also contained two other PST congeners. Based on elution after SPE-COOH clean up, they were identified as sulfocarbamoyl B2 and C3+4 toxins. Due to the lack of commercially available standards they were not quantified. Nevertheless, in terms of fluorescence response they exhibited a pronounced peak in the chromatographic analysis (data not shown).
Fig. 4

Profiles of paralytic shellfish toxins (% molar fraction) characterized in G. catenatum (stations 3–6) and sardines (stations 1, 2 and 5–7). Means (±SD) marked with asterisk (*) represents significant differences (P < 0.05) after applying the Mann–Whitney test. Filled square G. catenatum, Open square sardines

PST profile characterized in sardines

Paralytic shellfish toxins (PSTs) were found in sardine viscera in all samples collected from the NW coast; the highest viscera level detected was 531 μg STXeq kg−1. Toxins were not detected in any of the corresponding brain or muscle tissue samples (Table 1). The same suite of PSTs detected in G. catenatum was also found in sardine viscera (Fig. 4); however, differences in their relative abundance were observed between algae and fish. This was primarily due to dcNEO, which was significantly higher (P < 0.05) in sardine viscera. The sulfocarbamoyl B2 and C3+4 toxins were also detected in sardine viscera, as in G.catenatum, but were not quantifiable due to the lack of standards.
Table 1

Sardina pilchardus. Body weight and body length (mean (SD)) and concentrations (mean (SD)) of paralytic shellfish toxins (PSTs) determined in sardine viscera, muscle and brain (n = 3 with 5 pooled fish each)


Station coordinates

Body weight (g)

Body length (mm)

PSTs (μg STXeq kg−1)







94.1 (7.1)

20.9 (0.4)

211 (95)






70.7 (4.3)

19.6 (0.3)

63 (30)






76.4 (7.0)

20.2 (0.6)

177 (47)






73.5 (6.4)

20.1 (0.5)

16 (5)






60.3 (4.8)

18.7 (0.4)

331 (71)






30.2 (2.6)

15.5 (0.4)

531 (115)






64.8 (4.0)

19.0 (0.3)

364 (69)






60.8 (5.2)

18.7 (0.4)







59.3 (4.8)

18.9 (0.5)







62.2 (3.8)

19.0 (0.3)





The presence and distribution of G. catenatum have been previously studied along the Portuguese coast during the 1980s and 1990s (Moita et al., 2003), and although information on species dynamics is available, relatively little is known regarding toxin production and baseline toxin levels in wild populations of G. catenatum (Gárrate-Lizárraga et al., 2005). In the present study, cell densities of G. catenatum in seawater samples collected in 2007 reached levels similar to those observed by Moita et al. (2003). The highest intracellular PST level of 398 ng STXeq l−1 coincided with the maximum cell abundance of 2.5 × 104 cells l−1 suggesting a positive relationship between cell densities and toxin levels. Intracellular toxins can be released into the surrounding seawater via excretion or lysis of phytoplankton cells, thereby posing an additional threat of exposure to aquatic fauna via direct absorption from the water by aquatic animals. The concentration of extracellular PSTs was consistently higher than intracellular toxin values. To the best of our knowledge, this is the first data on extracellular PSTs in Atlantic waters. Extracellular PSTs have been measured in the marine environment in Puget Sound, Washington State, USA, where toxin concentrations reaching levels similar to our study were measured via the same ELISA test (Lefebvre et al., 2008).

Extracellular PST values were not correlated with cell densities in this study; however, a general increase in extracellular PST concentrations was observed from south to north along the NW coast (Fig. 3). According to the national monitoring program for toxic marine phytoplankton, the G. catenatum bloom events observed in 2007 initiated in the nearby area of Lisbon Bay and developed northwards (unpublished data), following the pattern proposed by Moita et al. (2003). The concentrations of extracellular PSTs were in accordance with bloom progression, having lower values detected in samples collected where blooms were most recently rising. While intracellular PSTs can be considered a real-time value, extracellular PSTs may represent a time-integrated indication of PST occurrence in seawater. However, extracellular PSTs may also be affected by upwelling events or river plumes and may not persist long at seawater pH unless stabilization is achieved by complexation with other substances (Shimizu, 2000).

Blooms of G. catenatum spatially and temporally overlap with sardine spawning and recruitment areas along the Portuguese coast (Ré et al., 1990; Zwolinski et al., 2001) and thereby may pose a threat to survival and recruitment of early non-feeding planktonic larval stages, as has been suggested for Pacific herring (Clupea harengus pallasi) populations in Puget Sound, Washington, USA (Lefebvre et al., 2005). High levels of extracellular PSTs have been shown to inhibit sensorimotor function in marine herring larvae and cause morphological abnormalities in zebrafish (Danio rerio) larvae in laboratory studies (Lefebvre et al., 2004, 2005). In the Portuguese coastal region, sardines (Sardina pilchardus) are the most abundant planktivorous fish and a major component of the marine foodweb (Silva, 1999; Santos et al., 2007). Sardine diets are comprised of a high number of phytoplankton cells, including species in the same size range as toxic G. catenatum (Costa & Garrido, 2004). The maximum level of PSTs (531 μg STXeq kg−1) measured in sardine viscera in the present study was similar to values measured previously in Pacific sardines (Sardinops sagax), northern anchovies (Engraulis mordax) and Pacific herring (Clupea pallas) in Monterey Bay, CA, USA (Jester et al., 2009).

Overall toxin compositions were similar in G. catenatum and sardine samples, and consisted of sulfocarbamoyl (C1+2 and B1) and decarbamoyl toxins (dcSTX, dcGTX2+3 and dcNEO). However, there was a statistically significant difference (P < 0.05) between sardine viscera and G. catenatum samples, with sardines having a larger proportion of dcNEO in the gut. In previous studies with fish consuming zooplankton or fish vectors of PSTs, differences were observed in toxin compositions when compared between the algal source and the fish. For example, Montoya et al. (1998) identified different toxin profiles in anchovies (Engraulis anchoita) and in the PSTs-producing dinoflagellate Alexandrium tamarense in the Argentinean coast. Microscopic analysis of anchovy stomach contents revealed the predominance of copepods and cladocerans, while abundance of cells of A. tamarense was highly variable and often in low levels (Montoya et al., 1996, 1998). The Atlantic mackerel (Scomber scombrus), which is not a direct consumer of the toxin producer but a predator of young herring, also showed different PST profiles than the PST causative agent (A. fundyense) in a study developed by Haya et al. (1990) on the Canadian coast. These data suggest that the length of the vector food chain may influence toxin composition.

In summary, the suite of toxins produced by recent samples of G. catenatum is identical to the suite of toxins found in strains isolated during the 1990s, as reported by Franca et al. (1996) and consists primarily of sulfocarbamoyl and decarbamoyl toxins. The same suite of PSTs is present in sardines, suggesting they are a good tracer for the occurrence of offshore G. catenatum blooms and their intracellular toxins. In addition to intracellular toxins, sardines and other marine organisms can be exposed to extracellular toxins released by G. catenatum. The first field data of extracellular PSTs in Atlantic waters is reported, and is relevant to studies aiming to investigate the susceptibility of marine organisms exposed to PSTs during different stages of ontogeny.


Authors greatly appreciated the helpful assistance of Ms Bich-Thuy Eberhart (Seattle-NOAA) in ELISA determinations, and Ms Lourdes Dias (Lisbon-IPIMAR) for seawater sampling. The Portuguese Foundation for Science and Technology supported this study through the research grant PTDC/MAR/78997/2006 and post-doctoral grant to P.R. Costa, grant number: SFRH/BPD/27002/2006.

Copyright information

© Springer Science+Business Media B.V. 2010