The 2016 “Godzilla Red Tide” event caused the most catastrophic episode in the history of the Chilean aquaculture due to the convergence of two potent toxic microalgae, the dictyochophyte Pseudochattonella verruculosa and the dinoflagellate Alexandrium catenella. This massive event produced the largest fish farm mortality ever recorded worldwide (export loss of USD 800 million), which when combined with PSP shellfish toxicity (~15% reduction in harvest compared to 2015), resulted in major social unrest and rioting in the south of Chile (^{Trainer et al., 2020}). Historical fishkilling harmful algal bloom (HAB) events have, however, been reported along the southern Chilean fjords since the development of the salmon farming industry in 1982 (Table 1). Fish mortalities have been often associated with the rapid growth of harmful microalgal species, but this noxious effect can also be associated with harmless species because high biomass of any phytoplankton can generate oxygen depletion in the water column at the end of bloom due to bacterial degradation, as well as gas-bubble trauma due to extreme oxygen saturation from algal photosynthesis (^{Lutz, 1995}). However, some microalgal species can kill fish at low cell concentrations via a high impact on the gill tissues. Two main mechanisms for fish gill damage by microalgae have been proposed: 1) damage due to the presence of penetrating or cutting micro-structures, and 2) due to the production of the so-called “ichthyotoxins” (^{Hallegraeff et al., 2017}). Research on the latter has been an enormous scientific challenge because the production of these metabolites often differs from the well-studied paralytic, diarrheic, amnesic and neurotoxic phycotoxins.

The novel application of a standardized and highly sensitive, repeatable and reproducible rainbow trout Oncorhynchus mykiss RTgill-W1 cell line assay (^{Fischer et al., 2019}) has become a popular screening tool for assessing toxicity of algal blooms in several impacted coasts (^{Tanneberger et al., 2013}; ^{Dorantes-Aranda et al., 2015}; ^{Natsch et al., 2018}). Through this cell line assay, ichthyotoxic compounds produced by Chilean strains of A. catenella and P. verruculosa have been initially explored (^{Mardones et al., 2015}, ²⁰¹⁸, ²⁰¹⁹). Synergism between the long-chain (≥C20) polyunsaturated fatty acids (PUFA) docosahexaenoic acid (DHA, 22:6ω3; 16-20% of total fatty acids) and reactive oxygen species (ROS -as superoxide anion) has been suggested as a primary ichthyotoxic mechanism in A. catenella, but cannot be explained by paralytic shellfish toxins (^{Mardones et al., 2015}, ²⁰¹⁸). Furthermore, the role of mucocysts has been pointed for a more in-depth research insight, which could provide a greater understanding as to the harmful mode of action of P. verruculosa (^{Mardones et al., 2019}).

Ichthyotoxic microalgae is a major concern for the Chilean Government, and thus highly expensive monitoring programs are conducted along the southern Patagonian coast. Despite such monitoring efforts, many of the fish-killing microalgae species that occur in the Chilean fjords are practically unstudied concerning their potential toxicities. As a result, it is frequent that Chilean monitoring programs alert fish farmers about the presence of a “putative” ichthyotoxic microalgae species only based on their ichthyotoxic characteristics reported in other coastal areas.

This study aims to assess the ichthyotoxic potency of the main microalgae species that has been reported in fish-killing events in the historical Chilean monitoring programs, in order to rank their potential impact on the local aquaculture.

Nine potentially ichthyotoxic microalgae strains and one non-toxic species (Teleaulax sp. as a control treatment) were obtained from the Harmful Algal Collection of the Centro de Estudios de Algas Nocivas in Chile (CREAN/IFOP) (Table 2). Algae were grown in F/2 medium using seawater at 30 of salinity and kept at 17°C under 16:8 h light:dark cycle at 100 μmol photons m^-2 s^-1 of light intensity (cool white fluorescent lamp). The gill cell line RTgill-W1, originally from rainbow trout (^{Bols et al., 1994}), was obtained from the American Type Culture Collection (CRL-2523, ATCC). Cells were cultured at 17°C in the dark in Leibovitz’s L-15 medium (L1518 Sigma), supplemented with 10% (v/v) fetal bovine serum (FBS, 12003C, Sigma), and an antibiotic-antimycotic solution (A5955, Sigma) containing amphotericin B (25 mg mL^-1), streptomycin (10 mg mL^-1) and penicillin (10,000 units mL^-1) in 25 cm² culture-treated flasks (3100-025, Iwaki). 0.25% trypsin-0.02% EDTA in Hank’s balanced salt solution (59428C, Sigma) was used to detach cells that grew as an adherent monolayer at the bottom of the flask. Subcultures were normally established twice per week at a ratio of 1:2 with L-15 medium renewal.

The cytotoxicity assay with the 10 microalgae strains was carried out using conventional multi-well microplates, according to ^{Dorantes-Aranda et al. (2011)}. Cultures with confluent gill cells were trypsinized (59428C, Sigma) for detachment, counted using a hemocytometer and adjusted to a concentration of 2×10⁵ cells mL^-1 in L-15 medium. Subsequently, gill cells were seeded in quadruplicate in 96-well flat-bottom microplates (3860-096, Iwaki, Japan), using a volume of 100 μL per well. After 48 h at 17°C in the dark for gill cell attachment, L-15 medium was discarded, the cells rinsed with PBS and exposed to 100 μL of lysed cells and supernatant medium from the 10 strain cultures at 30 of salinity at 17°C in the dark. Supernatant and lysed cell treatments were prepared from all cultures in the exponential growth phase at 2,000 cells mL^-1. The lysed cell suspension was prepared by sonication of diluted samples for 10 min at an amplitude of 10 μm peak to peak at 17°C and filtered using a syringe with a nylon filter (0.22 μm). After two hours exposure, the viability of the gill cells was determined using L-15/ex medium (^{Schirmer et al., 1997}), a modified version of the L-15 medium, containing 5% of the indicator dye alamarBlue (DAL1025, Invitrogen) (^{Pagé et al., 1993}). The medium containing the indicator dye was added to all cell-seeded wells and incubated for one hour in the dark (^{Dayeh et al., 2005}). Using a microplate reader (FLUOstar OMEGA, BMG Labtech), the fluorescence signal of alamarBlue was detected using excitation and emission filters of 540 and 570 nm, respectively. The viability of the gill cells was expressed as a response percentage of the treatments relative to the controls (% of control). Analysis of variance (ANOVA) from simple linear regression models on gill cell viability against intra- and extracellular compounds was performed, to explore the ichthyotoxic of the 10 microalgae strains. Normality and homogeneity of variances were assessed by the Kolmogorov-Smirnov method and Levene’s test. A post-hoc analysis using a Tukey HSD test was performed to determine differences among treatments. The null hypothesis (no difference in responses) was rejected in all statistical analyses if the respective P-value was <0.05. These analyses were performed using the R software 3.0.1 (^{Ihaka & Gentleman, 1996}). Algal biovolume was calculated according to ^{Hillebrand et al. (1999)}.

Historical monitoring data in Chile shows that after the startup of the salmon industry in 1982, the first harmful algal blooms (HABs) producing fish-kills were related to the occurrence of flagellates of the class Dinophyceae and Raphydophyceae. During the 90’s, salmon mortalities were mainly linked to harmful diatoms of the genus Leptocylindrus and Chaetoceros, whereas since 2000, there has been an increase in the intensity, duration and geographic coverage of ichthyotoxic flagellates (see the type of harmful effect in Table 1). This study tested the ichthyotoxic potency of the most recurrent phytoflagellates producing fishkilling events in the last 36 years in southern Chile (Fig. 1, Table 2). After two hours exposure of microalgae cultures against fish gill cells, all algae cultures except the control (Teleaulax sp.) and one Karenia strain (KS_01), showed significant differences in the gill cell viability response between the supernatant and lysed cell algae treatments (Figs. 2a-b; ANOVA, P < 0.05), and no correlation with algal biovolume (Fig. 2c; R² < 0.001). This result agrees with several studies that have shown that ruptured microalgae cells exhibit consistently higher toxicity than supernatant against RTgill-W1 cells (^{Dorantes-Aranda et al., 2011}, ²⁰¹⁵; ^{Mooney et al., 2011}; ^{Mardones et al., 2015}). While specific lytic compounds produced by some phytoflagellates are released into the surrounding aquatic environment, the most active portion remains cell-bound. Based on these findings, mitigation approaches that are currently used in salmon farms during bloom events, where the main target is to destroy the harmful algae cells (by rupturing), are likely to make effects worse.

Figure 1 Microalgae species used in the study. a) Vicicitus globosus, b) Pseudochattonella verruculosa, c) Heterosigma akashiwo, d) Teleaulax sp., e) Alexandrium ostenfeldii, f) Alexandrium catenella, g) Prorocentrum micans, and h) Karenia selliformis. Scale bar: 10 μm. 

Figure 2 Loss of viability of RTgill-W1 cells after 2 h exposure to a range of algal cultures, arranged from most ichthyotoxic (left) to harmless (right). a) supernatant, b) lysed cells, and c) effect of supernatant (blue circles) and lysed cell (black circles) treatments on gill cell viability estimated from total algal biovolume. The trend line and equation of linear adjustment are presented for all data. The nontoxic species Teleaulax sp. was used as a negative control. VG_01: Vicicitus globosus, PV_01 and PV_06: Pseudochattonella verruculosa, HA_01: Heterosigma akashiwo, TL_01: Teleaulax sp., AO_01: Alexandrium ostenfeldii, AC_01: Alexandrium catenella, PM_01: Prorocentrum micans, KS_01 and KS_02: Karenia selliformis. 

As observed in the lysed-cells treatment (the most toxic fraction), the most cytotoxic species at 2,000 cells mL^-1 -in decreasing order- were Karenia selliformis, A. catenella, Vicicitus globosus, Alexandrium ostenfeldii, P. verruculosa, Heterosigma akashiwo, and Prorocentrum micans (Fig. 2b). Except for K. selliformis and A. catenella, the cytotoxic potency observed in all surveyed species is low, considering that 2,000 cells mL^-1 could be accepted as an intense HAB event in natural oceanic conditions. Assuming that ichthyotoxins are equivalent to allelopathic compounds, our results could be a consequence of in vitro monoclonal cultures, where the microalgae do not require the release of toxic compounds under no competence nor predation pressure. This outcome might partially explain the observed mismatch in ichthyotoxic potency between natural algae blooms (higher toxicity) and laboratory studies on target species (lower toxicity).

Coupling historical fish-killing events (Table 1) and results in this study regarding the ichthyotoxic potency of algae species, two main considerations should be taken into account by the Chilean salmon industry. First, Karenia selliformis has been observed to bloom in offshore areas (2-10 NM), barely affecting inshore salmon farming activities (<2 cells mL^-1). Exceptional outbreaks of this toxic dinoflagellate, in inner channels, might, however, be even more damaging than the Pseudochattonella bloom recorded in 2016 (^{Mardones et al., 2019}); thus, early warning systems should be activated when the presence of this dinoflagellate species. Second, according to the present historical analysis, there is an apparent increment in the occurrence of phytoflagellates and their ichthyotoxic potency during the last decade in southern Chile (see the ranking in Table 1). This fact might be the result of enhanced monitoring efforts (especially from the salmon farming industry) and improved expertise in taxonomic identification of cryptic toxic flagellates species, but also due to the observed resilience of some toxic flagellate to projected climate variability in the eastern Pacific coast (^{Mardones et al., 2016}). Expected climate change or exceptional warmer environmental conditions might increase the occurrence of ichthyotoxic microalgae, as reported by ^{Trainer et al. (2020)}.

In conclusion, this study has shown that the application of an in vitro fish gill assay is an effective method to carry out sensitive screening tests for the toxicity of marine microalgae. An important output to consider by salmon farmers is that ichthyotoxic flagellates are more toxic to fish gill cells upon rupture and that Karenia spp. are more toxic than previously thought. These results have important implications for the way the salmon industry should manage and mitigate algal bloom impacts.