The results show that there was no significant difference in temperature (F = 3.2, P = 0.09), Trichostatin A mw pH (F = 3.1, P = 0.09) or salinity (F = 0.1, P = 0.8) between the two sites during the study period. The surface water temperature at both sites increased gradually during the study period, whereas salinity decreased sharply until reaching the lowest level (26.5‰) on 3 June, coincident with the highest peak of H. akashiwo cells at site 1 ( Figure 3). The salinity rose again to more than 31‰ during the remaining part of the study period. In contrast, dissolved oxygen (F = 329.9, P < 0.001), NO3 (F = 2748.7, P < 0.001), NH4

(F = 1031, P < 0.001) and phosphate (F = 385.9, P < 0.001) concentrations varied significantly between the two sites. In general, nutrient concentrations (NH4, NO3 and PO4) were higher at the bloom site than at the non-bloom site ( Figure 4), indicating

their possible promotion of H. akashiwo bloom formation at the bloom site. The abundance of H. akashiwo at the bloom site increased markedly during the study, with the highest density (46 × 106 cells L− 1) obtained on 3 June ( Figure 4); it began to decline on 10 June and eventually crashed on 24 June, coinciding with the salinity buy JQ1 increase up to 40‰. The cell density of H. akashiwo correlated negatively with salinity (r = − 0.83) and pH (r = − 0.7), and positively with NH4 (r = 0.88), NO3 (r = 0.78) and PO4 (r = 0.86). The cell density of this alga was only weakly correlated with water temperature (r = 0.2), enough as the temperature did not vary significantly during the last three periods of the study ( Figure 3a). Chlorophyll a concentrations were higher at the bloom site than at the non-bloom site and correlated positively with H. akashiwo cell density (r = 0.87) at the bloom site. In addition to H. akashiwo cells, the bloom site contained 17 other algal species, but with low cell densities ( Table 1). Most of these algae are potentially toxic species of dinoflagellates (e.g. Alexandrium, Dinophysis, Gymnodinium), raphidophytes (e.g. Chattonella)

and cyanobacteria (e.g. Trichodesmium). Remarkably, all of these species except Chattonella had been recorded at this site before the H. akashiwo bloom appeared, and began to disappear gradually as the cell density of H. akashiwo increased ( Table 1). Thereafter, these species re-appeared at the site when the bloom collapsed on 24 June. In contrast, the raphidophyte Chattonella was associated with the Heterosigma bloom during the study period. During this study, the raphidophyte H. akashiwo was toxic to A. salina. As shown in Table 2, both the aqueous and methanol extracts of H. akashiwo blooms were toxic towards A. salina with a significant difference in LC50 values (F = 15.2–62.5, P = 0.01–0.001): the methanol extracts were more toxic (LC50 = 9.14–9.