*5.1. Biological Material*

#### 5.1.1. Neuroblastoma Cell Line Culture

Mouse neuroblastoma (N2a) cell line (CCL-131) was purchased from the American Type Culture Collection (ATCC, Manassas, VA, USA) (Table S4). The complete growth medium consisted of RPMI medium 1640 with HEPES and without L-Glutamine, supplemented with 10% fetal bovine serum (FBS), 2 mM GlutaMAX, 1 mM sodium pyruvate, 50 μg/mL streptomycin with 50 units/mL penicillin and 2.5 μg/mL Amphotericin B. This cell line was routinely maintained at 37 ◦C in a 5% CO2 humidified atmosphere by splitting cell culture by 1:10 (every two days) or 1:30 (every three days) by rinsing the cell layer with Dulbecco's Phosphate Bu ffered Saline without CaCl2 and MgCl2 (DPBS-1X), followed by a dissociation step using 0.05% Trypsin-EDTA. The N2a cell line used in this study showed a stabilized growth after 300 P. All experiments presented in this study were done using cell passages ranging from 383 to 810 P.

All reagents are listed in Supplementary Materials (Table S5). N2a cell line was sub-cultured in 25 cm<sup>2</sup> or 75 cm<sup>2</sup> tissue culture flasks (Nunc ™ Easy-Flask, ThermoFisherScientific, Kamstrupvej, Danemark), while 96-well (0.32 cm<sup>2</sup>/well) flat bottom microplates (Falcon ®, Corning Brand, New York, NY, USA) were used for CBA-N2a experiments. The list of equipments and materials are presented in Supplementary Materials (Table S6).

#### 5.1.2. Reagents and Toxin Standards

Ouabain octahydrate (O3125), Veratridine (V5754), MTT, DMSO and MeOH are listed in Supplementary Materials (Table S4). Aqueous stock solutions of Ouabain and Veratridine were prepared at 20 mM in pure water and 5 mM in pH2 pure water, respectively.

Four toxin standards were used to characterize the mode of action of two toxin families using the CBA-N2a (Table S4): (i) two VGSC activators for which certified reference material is not available as yet, i.e., Pacific ciguatoxin P-CTX3C obtained from the bank of standards of Institut Louis Malardé [77,78,141] and brevetoxin PbTx3 (ref. L8902) purchased from Latoxan (Valence, France). Stock solutions of P-CTX3C and PbTx3 were prepared in DMSO at 20 ng/mL and 100 μg/mL, respectively. Non volatile DMSO solvent was chosen to ensure stable toxin concentrations over long periods of storage; (ii) two VGSC inhibitors for which certified reference materials were purchased from the National Research Council Canada (NRCC, Halifax, NS, Canada), i.e., saxitoxin STX (ref. CRM-STX-f) supplied at a concentration of 66.3 μM in aqueous hydrochloric acid 3 mM and decarbamoylsaxitoxin dc-STX (ref. CRM-dc-STX-b) supplied at a concentration of 65 μM in aqueous hydrochloric acid 3 mM. Concentrations of the STX and dc-STX standard solutions correspond to 24.7 and 21.4 μg/mL, respectively.

## 5.1.3. Fish Samples

The four fish samples tested in this study were collected from two ciguatera-endemic areas of French Polynesia, namely Tikehau Island (Tuamotu Archipelago) and Mangareva Island (Gambier Archipelago), and conditioned in the form of fillets kept at −20 ◦C until further CTX extraction. For some of them, their toxicity was also previously characterized by means of the fluorescent Receptor Binding Assay (fRBA) and/or Liquid Chromatography coupled with Tandem Mass Spectrometry (LC-MS/MS) [73,120]. Table 8 details their geographic origin, species, trophic status, as well as ciguatoxic status.


**Table 8.** List of fish samples tested for the revisited CBA-N2a.

\* NT: not tested; \*\* Fluorescent RBA expressed in ng P-CTX-3C eq/g fish flesh; \*\*\* Unpublished data.

#### *5.2. Extraction Procedure*

Fish fillets were carefully homogenized in a blender (Groupe SEB, Lourdes, France) waste disposal unit for 1–2 min [113]. For each fish sample, a portion of 10 g was processed following the protocol described in Darius et al. 2018 [77]. Briefly, a liquid/liquid partition was applied followed by a purification step using Sep-Pak C18 cartridges (360 mg sorbent per cartridge; Waters®, Saint-Quentin, France) leading to three distinct liposoluble fractions, i.e., LF70/30, LF90/10 and LF100. Fractions LF90/10 likely to contain CTXs were further dried in a SpeedVac concentrator (ThermoFisherScientific, Waltham, MA, USA) and weighed with a Sartorius Micro balance (model MC 410 S, Sartorius, Göttingen, Germany) with a reading accuracy of 0.1 mg (Table 9). The resulting dry extracts were resuspended in pure methanol at a concentration of 10 mg/mL and kept at −20 ◦C until further CBA-N2a analysis. Prior to dosing by N2a cells, fish extracts were brought to room temperature.

**Table 9.** Comparison of LF90/10 dry extract weights prepared from 10 g (fresh weight) of fish samples tested with CBA-N2a.


#### *5.3. Neuroblastoma Cell-Based Assay (CBA-N2a)*

#### 5.3.1. Characterization of N2a Cell Growth and Initial Viability

The initial viability of N2a cells is a function of the cell density reached in wells after 26 h growth, their physiological state and the time allocated for the metabolization of MTT by viable cells.

First, in order to determine the maximum cell density supported by each of the wells (representing a surface of 0.32 cm2) of 96-well Falcon® microplates, the growth curve of the N2a cell line used in this study was examined over a period of 4 days. For more convenience, growth experiment was conducted in 25 cm<sup>2</sup> culture flasks. Twelve culture flasks were seeded with 335,000 cells resuspended in 11 mL 10% FBS culture medium solution, and left to incubate for 22 h as described in Section 5.1.1. Cell densities were then monitored by sacrificing one flask every 6 h starting from 22 to 76 h. The two last flasks were sacrificed at 85 and 100 h, respectively. Cell enumeration was achieved by manual counts (*n* = 10/flask) using KOVA Glasstic slides (Hycor), and the results plotted against time. Cell density data were normalized to a surface of 0.32 cm<sup>2</sup> and used as a proxy of cell densities in 96-well microplates: e.g., using this rationale, the initial cell density in a 96-well microplate was estimated at 4288 cells per well.

Second, the initial viability of N2a cells was examined under two different % FBS (5% and 10% in culture medium) and six different MTT incubation times (15–65 min). Cell numbers before seeding were determined from *n* = 5 counts performed on an aliquot of the cell stock suspension and apply for all further experiments. To this end, twelve microplates (six microplates per % FBS) were seeded with 200 μL of ten different cell seeding densities ranging from 10,000 to 100,000 cells/well (*n* = 6 wells per cell density). After 26 h growth, cell viability was assessed via the MTT assay conducted by sacrificing one microplate at 15 min and the remaining every 10 min. MTT and DMSO conditions were used as originally described [19]. The MTT protocol followed the one described in Darius et al. [77]. Briefly, after removing the culture medium from the microplate, the 60 inner wells were filled with 60 μL of MTT solution at 0.83 mg/mL prepared in 2% FBS culture medium. After MTT incubation at 37 ◦C in a 5% CO2 incubator, the wells were emptied and the 60 inner wells and 12 outer wells (rows 1 and 12) were filled with 100 μL of DMSO. Following cell layer lysis with DMSO and manual homogenization, absorbance data were measured at 570 nm using an iMark ™ microplate reader (Biorad, Marnes la Coquette, France). All viability data were expressed in absorbance data.

#### 5.3.2. Characterization of N2a Cell Final Viability in the Absence and under O/V Treatment

Ouabain and Veratridine (O/V) treatment is required for the successful detection by CBA-N2a of toxins active on VGSCs. The N2a cell final viability, i.e., in the absence of O/V treatment (OV− conditions) or following O/V treatment (OV+ conditions) was characterized as follows.

To characterize the final viability of N2a cells in OV− conditions, six microplates were seeded with 200 μL of a 5% FBS culture medium at ten di fferent cell densities ranging from 10,000 to 100,000 cells/well (*n* = 6 wells per cell density) and left to grow for 26 h. One microplate, defined as the RCV control, was used to check the initial viability of N2a cells after 26 h growth and measured by MTT assay, while the five remaining microplates were treated as follows: culture medium was discarded under sterile conditions by overturning, and wells were dried by tapping each microplate on an ethanol-sterilized absorbent paper to remove any residual liquid. The 60 inner wells of the five microplates then received 200 μL/well of growth medium supplemented with 1%, 2%, 3%, 4% or 5% FBS, respectively, while the peripheral wells received the same volume of sterile distilled water. Culture microplates were further incubated overnight (Section 5.1.1) for an additional 19 h. Cell final viability was assessed using the MTT assay after 45 min incubation with MTT.

To characterize the final viability of N2a cells in OV+ conditions, two microplates were seeded with 200 μL/well of a 5% FBS culture medium at an initial cell density of 50,000 ± 10,000 cells/well, and left to grow for 26 h. After cell layer settlement, one microplate served as RCV control and measured by MTT assay. For the second one, 20 mM (O) and 5 mM (V) solutions were first diluted in 2% FBS culture medium to obtain a 360/36 μM O/V stock solution, which was further used to prepare serial dilutions ranging from 340/34 to 20/2 μM (with a decrease of 20/2 μM between each dilution). After removal of the used culture medium as previously described, each of the 60 inner wells received 200 μL/well of culture medium at nineteen distinct O/V concentrations ranging from 0/0 to 360/36 μM (*n* = 3 wells per O/V treatment conditions, except for 0/0 μM condition for which *n* = 6). Peripheral wells received the same volume of sterile distilled water. Culture plates were then incubated overnight for 19 h. Cell final viability was assessed using the MTT assay after 45 min incubation with MTT. Five independent experiments were performed.

#### 5.3.3. Characterization of the Unspecific E ffects of Solvent and Dry Extract on N2a Cell Viability

The e ffects on N2a cell viability of two solvents commonly used to resuspend toxin standards and dry extracts, i.e., MeOH and DMSO, were examined in the absence vs. under non-destructive O/V treatment conditions. In this experiment, the 60 inner wells of three 96-well microplates were seeded with 200 μL/well of a 5% FBS culture medium at an initial cell density of 50,000 ± 10,000 cells/well, and left to grow for 26 h. One microplate served as RCV control and was measured by MTT assay while the two remaining ones were treated as follows: after the complete removal of the used culture medium as previously described, the 30 inner wells on the upper half of the microplate received 200 μL/well of a 2% FBS culture medium (OV− conditions), while the 30 inner wells on the bottom half of the microplate received 200 μL/well of a 105/10.5 μM O/V treatment (OV+ conditions). In parallel, 100 μL of eight-points serial dilution at 1:2 of each solvent were prepared in the same culture medium using a U-bottom 96-well microtiter plate. Then, 10 μL of each solvent dilution were directly added in triplicate

wells and tested under OV− conditions versus OV+ conditions (100/10 μM final concentrations). Hence, the final concentrations tested ranged from 0.037 to 4.762%. Addition of 10 μL of 2% FBS culture medium in triplicate wells under OV− and OV+ conditions (COV− and COV+ controls, respectively), allowed to check final viability in solvent-free growth medium. Peripheral wells received the same volume of sterile distilled water. Culture plates were left to incubate overnight for 19 h, and the cell final viability determined using the MTT assay after 45 min incubation with MTT.

In order to determine the MCE of the four fish samples, increasing concentrations of fish extracts were tested in OV− conditions only, as no activity on VGSCs is expected to occur in this condition of treatment [81]. In most CBA-N2a studies however, these concentrations are established based on the fresh tissue weight of biological samples. Here, the tested concentrations were estimated based on the DEW instead, since dry extracts interact directly on cell layers and substantial di fferences in DEWs are often observed between extracts prepared from a same amount of tissue sample. In order to determine the MCE of the four fish samples used in this study, the 60 inner wells of three 96-well microplates were seeded with 200 μL of a 5% FBS culture medium at an initial cell density of 50,000 ± 10,000 cells/well, then left to grow for 26 h. One microplate served as RCV control and was measured by MTT assay, while the two remaining ones were treated as follows: first, 5% FBS growth medium was renewed by addition of 200 μL of a 2% FBS culture medium in wells. In parallel, a 1:10 dilution of the LF90/10 dry extract of four fish samples was prepared in 2% FBS culture medium using a U-bottom 96-well microtiter plate, followed by a nine-points serial 1:2 dilution (v = 100 μL per concentration). Then, 10 μL of each concentration was directly added in triplicate wells, leading to final dry extract concentrations that ranged from 186 to 47,619 pg/μL, i.e., 0.517 to 132.3 μg fish flesh equivalent/μ<sup>L</sup> for Cmic02, 0.433 to 110.7 μg/μ<sup>L</sup> for Cmic19, 0.600 to 153.6 μg/μ<sup>L</sup> for Emer05 and 0.689 to 176.3 μg/μ<sup>L</sup> for Emer13. Two fish samples were tested per plate. Controls wells (COV−) were also established by addition of 10 μL of 2% FBS culture medium in the absence of dry extract. Peripheral wells received the same volume of sterile distilled water. Culture plates were left to incubate overnight for 19 h, and the cell final viability determined using the MTT assay after 45 min incubation with MTT. Three independent experiments were performed.

#### 5.3.4. Detection of VGSCs Activators and Inhibitors by CBA-N2a

In this experiment, 60 inner wells of five 96-well microplates were seeded with 200 μL of a 5% FBS culture medium at an initial cell density of 50,000 ± 10,000 cells/well, and left to grow for 26 h. One microplate served as RCV control and was measured by MTT assay, while the four remaining ones were treated as follows: first, the growth medium was renewed by the addition of 200 μL of 2% FBS culture medium in OV− conditions (upper half of the microplate) versus 200 μL of culture medium in OV+ conditions (bottom half). The initial O/V concentrations in wells were 105/10.5 and 284/28.4 μM when detecting VGSC activators and VGSC inhibitors, respectively. Further, a nine-points serial 1:2 dilution of each toxin standard stock solution was prepared in 2% FBS culture medium (v = 100 μL per concentration) using a U-bottom 96-well microtiter plate, then 10 μL of each toxin concentration were directly added in triplicate under OV− and OV+ conditions (Section 5.3.3). The final concentrations of toxins tested ranged from 0.074 to 19.048 fg/μ<sup>L</sup> for P-CTX3C, 372 to 95,238 fg/μ<sup>L</sup> for PbTx3, 368 to 94,095 fg/μ<sup>L</sup> for STX and 1592 to 407,619 fg/μ<sup>L</sup> for dc-STX. The final O/V concentrations in wells were 100/10 and 270/27 μM when detecting VGSC activators and VGSC inhibitors, respectively. Appropriate controls in both conditions of O/V treatment, COV− and COV<sup>+</sup>, were established to verify the cell layer viability and the e ffect of O/V treatment, respectively, in the absence of toxins (Figure S7). The resulting full dose-response curves were used to characterize the response typical of a given standard toxin and for further toxin quantification. For each toxin standard, one microplate in three independent experiments was examined for the purpose of inter-assay variability comparison.

#### 5.3.5. Detection of VGSC Activators in Fish Samples by CBA-N2a

This experiment was conducted as previously described in Section 5.3.4, except that the initial concentrations of O/V in wells was set to 90/9 μM. Toxin detection and quantification in biological matrix of unknown varying toxicity require the implementation of additional quality check controls (QC) in OV− and OV+ conditions, namely QCOV− and QCOV<sup>+</sup>, to check for the validity of further toxicity results (Figure S2). Practically, in these controls, a known concentration of a VGSC activator is tested, whose effect on N2a cell viability has been pre-established. The QCOV− and QCOV+ were established by adding 10 μL of 0.1 μg /mL of PbTx3 in triplicate, to reach a final concentration of 4760 fg/μ<sup>L</sup> of PbTx3 in wells (PbTx3 was preferred to P-CTX3C as it is commercially available). An eight-points serial 1:2 dilution of P-CTX3C and fish dry extracts were prepared (v = 100 μL per concentration) using a U-bottom 96-well microtiter, then 10 μL of each concentration were directly added in triplicate under OV− and OV+ conditions (85.7/8.57 μM final concentrations). Hence, the final concentrations of P-CTX3C tested ranged from 0.099 to 12.70 fg/μ<sup>L</sup> and from 74.4 to 9523.8 pg of dry extracts/μL for Cmic02 and Emer13, 14.9 to 1904.8 pg of dry extracts/μL for Cmic19 and 18.6 to 2381 pg of dry extracts/μL for Emer05. The full sigmoidal dose–response curves obtained for each toxin standard and fish samples when tested in parallel in the same experiment were used to determine EC50 values and infer toxin content in fish samples (Section 5.3.6). Three microplates in one experiment (*n* = 3) and one microplate in three independent experiments (*n* = 3) were examined for the purpose of intra- and inter-assay variability comparison, respectively.

Validation of the CBA-N2a results by means of appropriate viability controls are presented in Supplementary Materials (Table S4).

#### 5.3.6. Absorbance Data and Toxin Analysis

First, for each experimental plate, all raw absorbance data were corrected by deducting the corresponding mean DMSO control absorbance data (*n* = 12) to obtain net absorbance data. Dose-response curves were then established by plotting net absorbance values vs. pure toxins or dry extract concentrations tested, using GraphPad Prism software version 8.1.2 (GraphPad, San Diego, CA, USA) based on a four parameter logistic regression model (4PL) according to the following equation:

$$\text{Y} = \text{Bottom} + (\text{Top} - \text{Bottom}) \left( 1 + 10^{\circ} ((\text{Log(EC}\_{50}) - \text{Log(X)}) \* \text{Filllsslope}) \right) \tag{2}$$

In which Y is the net absorbance data and X is the concentrations tested (fg/μL), and EC50 (fg/μL) represents the effective concentration of dry extract inducing a viability half way (50%) between the basal (Bottom) and the maximal (Top) values of the curve (EC50). This parameter is used to establish the toxic potency of each toxin standard.

The EC80 and EC20 values of toxin standards were inferred from dose-response curves and correspond to the effective concentration of dry extract inducing a viability 80% and 20% between the basal (Bottom) and the maximal (Top) values of the curve, respectively.

The limit of detection (LOD) and quantification (LOQ) of the CTX-like toxicity in fish samples were determined according to the following equations:

$$\text{LOD} = \text{(EC}\_{80} \text{/MCE)} \tag{3}$$

$$\text{LOQ} = \text{(EC}\_{50} \text{/MCE)} \tag{4}$$

where EC80 and EC50 are the values obtained for P-CTX3C toxin standard, with values expressed in ng P-CTX3C eq/mg of dry extract.

For more convenience, LOD and LOQ values can be expressed in the same unit as the one used in the advisory level recommended by the EFSA and US FDA. Calculations are based on the fresh weight of flesh tissue extracted (FW) and the corresponding dry extract weight (DEW) (Table 9), and use the following equations:

$$\text{LOD} = \text{(EC}\_{80} \text{/MCE)} \times \text{(DEN/FW)}\tag{5}$$

$$\text{LOQ} = \text{(EC}\_{\text{50}} \text{(MCE)} \times \text{(DEN/FW)}\tag{6}$$

In which LOD and LOQ of CTXs in biological matrix are expressed in ng P-CTX3C eq/g fish flesh. In the same way, quantification of the composite toxicity in fish dry extracts (T), expressed in ng P-CTX3C eq/mg, was determined by comparing the EC50 values of P-CTX3C and fish dry extracts determined in the same experiment, using the following equation:

$$\text{IT} = \text{EC}\_{50} \text{ of P-CTX3C/EC}\_{50} \text{ of dry extract} \tag{7}$$

The composite toxicity in biological samples (Q), expressed in ng P-CTX3C eq/g of fish flesh, is determined using the following equation:

$$\mathbf{Q} = \mathbf{T} \times \text{(DEF/FW)}\tag{8}$$

5.3.7. Data Analyses

Variabilities of CBA-N2a data were examined using the mean ± SD of three microplates tested in one experiment on the same day (intra-assay comparisons) or the mean ± SD of one microplate tested in three independent experiments (inter-assay comparisons). Statistical analyses were performed by means of the Wilcoxon test with significant di fferences considered at *p*-values < 0.05, using RStudio software version 1.0.153 Version 1.0.153– © 2009-2017 (RStudio, Inc., Boston, MA, USA).

First, the intra- and inter-assay variabilities of the five viability controls (Table S7), RCV, COV<sup>−</sup>, COV<sup>+</sup>, QCOV− and QCOV+ controls were assessed from inter-assay experiments conducted at 100/10 μM run at 662-663-666 cell passages and intra- and inter-assay experiments conducted at 85.7/8.57 μM run at 810 and 795-797-798 cell passages, respectively of CBA-N2a experiments presented in Sections 2.4.2 and 2.4.3 and Table S1.

Second, the Wilcoxon test was applied to search for variability between EC80 and EC50 of P-CTX3C and PbTx3 obtained from inter-assay experiments conducted at 100/10 μM and run at 662-663-666 cell passages, and intra- and inter-assay experiments conducted at 85.7/8.57 μM and run at 810 and 795-797-798 cell passages, respectively (Table 4, Table S2).

For intra- and inter assay comparison of LOD and LOQ values of the CTX-like toxicity in fish dry extracts, three di fferent P-CTX3C EC80 or EC50 values were compared to a unique MCE value, i.e., 10,000 pg/μ<sup>L</sup> applying for all fish samples. Then, three LOD and LOQ values were obtained giving a mean ± SD with *n* = 3 for intra- and inter-assay for all fish samples (Table 5).

For intra- and inter assay comparison of LOD and LOQ values of the CTX-like toxicity in fish flesh, one specific MCE value (converted from the DEW/FW ratio) was obtained for each fish sample. Then, three di fferent P-CTX3C EC80 or EC50 values were compared to a unique MCE value per fish sample. In the same way, three LOD and LOQ values were obtained for each fish giving a mean ± SD with *n* = 3 for intra- and inter-assays (Table 5). Additionally, the Wilcoxon test was successfully applied to search for possible variability between LOD and LOQ values among the four fish samples (Table S3).

For intra-assay comparison of toxin content in fish sample, three di fferent P-CTX3C EC50 values obtained from *n* = 3 microplates were compared to three di fferent fish EC50 values obtained from *n* = 3 microplates of a given fish sample run the same day. Then, cross calculation (three di fferent P-CTX3C EC50 values x three di fferent fish EC50 values) were done giving nine toxin content values and a mean ± SD with *n* = 9 per fish sample (Table 7).

For inter-assay comparison of toxin content in fish sample, one P-CTX3C EC50 value was compared to one fish EC50 value for each fish sample per independent experiment run at di fferent times. Then, three toxin content values were obtained from the three independent experiments giving a mean ± SD with *n* = 3 for each fish sample (Table 7).

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2072-6651/12/5/281/s1, Figure S1: Examples of pictures obtained from the revisited CBA-N2a, Figure S2: Suggested layouts on a 96-wells microplate at day 2 of the CBA-N2a, Figure S3: Detection (DL) and confirmation (CL) limits adopted in qualitative screening analysis of samples tested for the presence of VGSC activators (under non-destructive O/V treatment), Figure S4: Detection (DL) and confirmation (CL) limits adopted in qualitative screening analysis of samples tested for the presence of VGSC inhibitors (under destructive O/V treatment), Figure S5: Typical dose-response curve expected with a sample positive for VGSC activators, tested at eight distinct concentrations (C1–C8) adjusted to fall below the MCE, Figure S6: Typical dose-response curve expected with a sample positive for VGSC inhibitors, tested at eight distinct concentrations (C1–C8) adjusted to fall below the MCE, Table S1: Comparison between assays of the absorbance data of five viability controls, Table S2: Comparison between assays of EC50 and EC80 values of P-CTX3C and PbTx3 standards under two non destructive O/V treatments (100/10 μM vs. 85.7/8.57 μM), Table S3: Comparison between assays of LOD and LOQ values (ng P-CTX3C equivalent/g fish flesh) under non-destructive O/V treatment at 85.7/8.57 μM), Table S4: List of biological materials, Table S5: List of reagents, Table S6: List of equipments and materials, Table S7: Expected results for viability controls to enable validation of the assay.

**Author Contributions:** Conceptualization, J.V.; Data curation, J.V. and H.T.D.; Formal analysis, J.V. and H.T.D.; Funding acquisition, M.C. and H.T.D.; Methodology, J.V. and H.T.D.; Project administration, M.C.; Resources, M.C. and H.T.D.; Supervision, M.C.; Validation, J.V., M.C. and H.T.D.; Visualization, J.V., M.C. and H.T.D.; Writing—original draft, J.V., M.C. and H.T.D.; Writing—review & editing, J.V., M.C. and H.T.D. All authors have read and agreed to the published version of the manuscript.

**Funding:** The present work was supported by funds from the countries of France and French Polynesia in the framework of the CARISTO-Pf" (no. 7937/MSR/REC of 4 December 2015 and Arrêté no. HC/491/DIE/BPT of 30 March 2016) and FLUOTRACK-CIGUATERA" (no. 023-15 of 19 February 2015) research programs.

**Acknowledgments:** The authors greatly acknowledge Philippe CRUCHET for the chemical extraction of fish samples and Sébastien LONGO for statistical analysis, as well as the three anonymous reviewers whose comments greatly helped improve the manuscript.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.
