Clonal cultures of M. polykrikoides were used for clearance rate experiments along with cultured R. salina and G. aureolum. M. polykrikoides cultures that were isolated from bloom water samples included strain CP1 from Flanders Bay, NY, in 2006 (Gobler et al., 2008) and strain CPSB-1G from Shinnecock Bay, NY, in 2010. Both strains have been shown to be harmful to bivalve and fish larvae (Tang and Gobler, 2009a; Tang and Gobler, 2009b; Griffith and Gobler, 2016), but have not yet been used to study their impact on bivalve clearance. G. aureolum (strain GymMHC-23) was isolated from Meetinghouse Creek, a tributary within Peconic Bay, NY and confirmed via sequencing of the 28S rRNA gene, and R. salina CCMP 1319 was provided by the National Center for Marine Algae (NCMA). Gymnodinium aureolum (Hulburt) G Hansen 2000, is different and distinct from the ichthyotoxic North European Gyrodinium aureolum which is now Karenia mikimotoi. While R. salina, G. aureolum, and M. polykrikoides differ in cellular diameter (15, 40, and 35 µm, respectively), hard clams are capable of clearing particles greater than 5 µm in diameter (Riisgård, 1988) up to at least 200 µm (Riisgård and Larsen, 2010).

All cultures were maintained in sterile GSe medium with a salinity of ~32, made with boiled and 0.2 μm filtered seawater (Doblin et al., 1999), and held at 21° C in an incubator that provided a light intensity of ~100 µmol m^-2 s^-2 with a 12 h light: 12 h dark photoperiod. Cultures were allowed to grow until they reached exponential growth phase to avoid any differences in toxicity amongst strains (Tang and Gobler, 2009a). Concentrations of all species and strains were quantified via flow cytometry using a Beckman Coulter CytoFLEX, imaging light microscopy using a Keyence BZ-X800, or hand counts through light microscopy.

Wild populations of M. polykrikoides from blooms were also used in this study to compare to laboratory grown cultures. Bloom water was obtained from regions that have a history of M. polykrikoides blooms (e.g. Peconic Bay and Shinnecock Bay). Sampling for bloom water took place opportunistically from July through September of 2020 and 2021. At each site, 20L of surface bloom water was collected in polyethylene carboys and physical parameters (water temperature, salinity, and dissolved oxygen) were measured with a YSI ProPlus probe. Although not monospecific, M. polykrikoides blooms constituted a majority of M. polykrikoides microscopically identified cells in the samples (>75%). Bloom water was used within a day of collection and diluted with UV-sterilized 1-μm filtered water from Old Fort Pond, Shinnecock Bay if too dense (>3,000 cells mL^-1) for the purposes of these experiments. The strongly allelopathic nature of M. polykrikoides blooms reduces the presence of other, larger phytoplankton in bloom water (Tang and Gobler, 2010; Koch et al., 2014), minimizing their potential impact on experimental outcomes.

Juvenile clams (~10-20 mm) were hatchery spawned from wild brood stock. All individuals were held in a sea table with recirculating filtered (1 μm) Old Fort Pond seawater at experimental temperatures (~21°C) while being fed mixed live algae (M. muelleri, P. lutheri, R. salina, T. lutea) for at least one week before being used for clearance rate experiments (Helm et al., 2004). Bivalves were starved 24 h prior to conducting experiments to ensure some individuals opened during experiments (Shumway et al., 1985).

Juvenile bivalve clearance rates of M. polykrikoides, G. aureolum, and R. salina clearance rates were calculated using the clearance method (Riisgård, 2001). In the clearance method, clearance rate is measured as the volume of water cleared of suspended particles per unit of time. The reduction in the number of particles as a function of time is monitored by taking water samples at fixed time intervals and measuring the particle concentration. Clearance rate experiments were established using 250 mL glass beakers placed on multi-position magnetic stir plates (details below). For all experimental algal exposures (differing species and food levels), there were three control beakers with phytoplankton only and ten experimental beakers with bivalves of the same algal diet per treatment. Single species exposure experiments were completed to address question 1 where cultured and bloom M. polykrikoides concentrations represented bloom and sub-bloom concentrations: 1,000 cells mL^-1, 750 cells mL^-1, 500 cells mL^-1, and 250 cells mL^-1 (Tang and Gobler, 2009a; Tang and Gobler, 2009b). The highest concentration for experiments using cultured M. polykrikoides was 1,000 cells mL^-1 which was limited by the density of scaled-up 4-liter cultures (<3,000 cells mL^-1) that could be produced. Preliminary data (personal obs.) suggested that clams would cease to clear at 1,000 cells mL^-1 although was not observed throughout this study. To address question 2, clonal culture CPSB-1G was used for M. polykrikoides, R. salina mixed culture experiments. Past studies have demonstrated that the clonal culture, CPSB-1G was of similar toxicity when compared to the clonal culture CP1 (Yang et al., 2022). Mixed culture experiments normalized at the 1,000 M. polykrikoides cell mL^-1 biovolume level using five different proportions: (0 dinoflagellate, 0.25 dinoflagellate, 0.5 dinoflagellate, 0.75 dinoflagellate, or complete dinoflagellate) and R. salina was used to fulfill the remaining 0.75, 0.5, or 0.25 of cells present respectively. The experimental set up was identical for experiments with G. aureolum and R. salina with cell densities normalized to M. polykrikoides biovolume equivalent (i.e. μm³). To address question 3, several experiments using wild populations of M. polykrikoides at densities greater than 1,000 cells mL^-1 with R. salina present were completed to observe behavior and clearance rates of juvenile clams when exposed to denser blooms.

Experimental beakers (250 mL) used for clearance rate experiments were gently homogenized using a 1 inch stir bar rotating at 60 rpm. One bivalve was placed on a plastic mesh platform in each beaker, keeping them ~2 cm above the stir bar. Bivalves were allowed to clear for one hour once they opened and/or extended their siphon. Duplicate samples were taken from each beaker once the bivalve opened and after the one-hour period. Samples were used to quantify concentrations of each algal species present, one preserved with 10% glutaraldehyde for flow cytometric quantification and the other with Lugol’s iodine for imaging or hand quantification using microscopy. Clearance rates were calculated using the equation below (Coughlan, 1969):

Where V is the volume of water in the container, n is the number of bivalves present (n = 1), t is time (1 hour), C₀ is the initial concentration of algae, C_t is the concentration of algae after bivalves have been open for one hour, C₀′ is the initial concentration of algae for the control, and C_t′ is the concentration after one hour. Clearance rate quantification was normalized to the dry weight of each bivalve individual allowing for a weight-based comparison between experiment types (Harke et al., 2011). Dry weights were obtained by removing all soft tissue from individual bivalves and keeping them in a drying oven for at least 48 hours before weighing them again on a balance. Changes in the cell densities of the control beakers were used to adjust the clearance rate of each replicate correcting for any loss of cells in the vessel.

Algal concentration was quantified via flow cytometry using a Beckman Coulter CytoFLEX, imaging light microscopy using a Keyence BZ-X800, or light microscopy. Flow cytometry employed the allophycocyanin (APC) channel with a 660/20 bandpass filter to obtain counts for live and preserved G. aureolum, and R. salina cells. Samples for both M. polykrikoides strains and bloom water were counted via light microscopy and imaging light microscopy to obtain total counts and distinguish between single cells and double/triple/quadruple chains.

Statistical analyses were performed using RStudio^® statistical software (version 4.0.3). Outliers consisting of unrealistic clearance rates for juvenile clams (>50 L hr^-1 g^-1) were removed prior to analyses. To address unbalanced distributions and heterogenous variances bootstrap two-way and three-way ANOVAs were used to identify significant differences in clearance rates between varying densities, strains, and biovolumes of M. polykrikoides and species of algae. Beyond unbalanced distributions and heterogenous variances, bootstrapping requires minimal assumptions about normality and it allows statistical inferences to be drawn regarding whole population based on a limited data set (Efron, 1992). When significant differences were detected, a bootstrap Tukey’s honest significant difference test (Tukey’s HSD) was performed on the bootstrapped data to identify specific differences between groups. Additionally, bootstrap one-sample t-tests with a Bonferroni correction for type 1 errors were completed on clearance rate data for each group during all experiments to identify whether clearance rates were significantly different than zero. All assumptions were met for all statistical models. Bootstrap ANOVAs were performed using the lmboot package (Efron, 1992), bootstrap one-sample t-tests were completed using the MKinfer package (Efron, 1994), and Tukey’s HSD post-hoc comparison tests were completed using the boot package. Results for Tukey’s HSD tests are displayed in representative box plots via compact letter display where in groupings are based on significant differences (α = 0.05).

For the single algal species experiments, clearance rates were significantly different across the different phytoplankton populations (p<0.05; bootstrap two-way ANOVA, Figure 1 ). Hard clams cleared the clonal strain CPSB-1G faster than M. polykrikoides bloom water and the clonal CP1 strain (p<0.05; bootstrap Tukey’s HSD, Figure 1 ). Additionally, hard clams cleared R. salina faster than the CP1 strain (p<0.05; bootstrap Tukey’s HSD, Figure 1 ). Hard clam clearance rates of R. salina were not significantly different from clearance of CPSB-1G (p> 0.05; bootstrap two-way ANOVA, Figure 1 ) but were marginally faster than the clearance of M. polykrikoides bloom water (p=0.07; bootstrap Tukey’s HSD, Figure 1 ). Hard clam clearance rates of the clonal strain CP1 were not significantly different than the clearance of M. polykrikoides bloom water (p>0.05; bootstrap two-way ANOVA, Figure 1 ). Hard clam clearance rates of CP1(α′ = 0.0125, p=0.484; t-test) and bloom water were not significantly different than zero (α′ = 0.0125, p=0.186; t-test). Clearance rates for all single algal species exposures did not differ across cell densities examined (250 – 1,000 cells mL^-1; p=0.163; bootstrap two-way ANOVA, Figure 1 ).

Clearance rates were significantly different as a function of phytoplankton population present (p<0.05; bootstrap two-way ANOVA, Figure 2 ). During the M. polykrikoides bloom water/R. salina ( Figure 2 ) mixture experiments, clearance of M. polykrikoides bloom water was not significantly different than clearance rates of R. salina (p>0.05; bootstrap Tukey’s HSD, Figure 2 ), and clearance of M. polykrikoides bloom water was significantly greater than zero (α′ = 0.0125, p<0.0125; t-test). In contrast, clearance rates of strain CPSB-1G were significantly slower than those of R. salina when the two algae were present together (p<0.05; bootstrap Tukey’s HSD, Figure 2 ). During the G. aureolum/R. salina ( Figure 3 ) mixture experiments hard clam clearance of the two species was not significantly different (p>0.05; bootstrap two-way ANOVA, Figure 3 ). Similar to the single species exposures, there were no significant differences in clearance rates across the different proportions of dinoflagellates in the algal mixtures (25-75%; %; p>0.05; bootstrap two-way ANOVA, Figures 2 , 3 ).

Hard clam clearance rates of M. polykrikoides within bloom water and CPSB-1G clonal cultures during single and mixed species exposures were compared. Clearance rates differed as a function of single versus mixed species exposures (p=0.08), but not across differing M. polykrikoides densities (p=0.196) and algal populations (p=0.275; bootstrap three-way ANOVA, Figure 4 ). More specifically, clearance rates of mixed species were marginally higher than single species exposures (p=0.08; bootstrap three-way ANOVA, Figure 4 ).

Hard clam clearance rates of R. salina cells as a single species, mixed with CPSB-1G, and mixed within bloom water were compared. Clearance rates differed as a function of algal mixture (p<0.05; bootstrap two-way ANOVA, Figure 5 ) but not as a function of proportions of M. polykrikoides within the mixtures (p>0.05; bootstrap two-way ANOVA, Figure 5 ). Clearance rates of R. salina were significantly slower when it was mixed with bloom water compared to R. salina clearance rates when mixed with CPSB-1G or R. salina alone (p<0.05; bootstrap Tukey’s HSD, Figure 5 ). In contrast, clearance rates of R. salina as a uni-algal suspension or mixed with CPSB-1G were not significantly different ( Figure 5 ).

Clearance rates differed as a function of biovolume of phytoplankton (p<0.05; bootstrap three-way ANOVA, Figure 6 ) but not as a function of proportions of M. polykrikoides and differing algal species (p>0.05; bootstrap three-way ANOVA, Figure 6 ). Additionally, interactive effects between biovolume and proportions of M. polykrikoides led to significantly different clearance rates (p>0.05; bootstrap three-way ANOVA, Figure 6 ). Clearance rates at 1,000 cells mL^-1 were not significantly different from the other two biovolume experiments (p>0.05; bootstrap Tukey’s HSD, Figure 6 ). M. polykrikoides clearance rates at 1,500 cells mL^-1 and 3,000 cells mL^-1 were not significantly different than zero (α′ = 0.0083, p>0.0083; t-test, Figure 6 ). R. salina clearance rates were not significantly different than zero at 3,000 cells mL^-1(p>0.0083) but were significantly greater than zero at 1,500 cells mL^-1 (α′ = 0.0083, p<0.0083; t-test, Figure 6 ).

The dynamic nature of clearance rates of juvenile hard clams exposed to different strains of M. polykrikoides supports the need for multifaceted, unique approaches to understand the toxicity of this alga. M. polykrikoides is known to be toxic due to the production of ROS (superoxide anion, hydrogen peroxide, hydroxyl radical) and perhaps extracellular polysaccharides (Kim et al., 1999; Kim et al., 2002; Kim et al., 2009; Tang and Gobler, 2009a). Significant mortality events during HABs caused by M. polykrikoides have been observed across multiple taxa including finfish (Gobler et al., 2008; Dorantes-Aranda et al., 2009; Tang and Gobler, 2009a), shellfish (Kim et al., 2009; Tang and Gobler, 2009a; Tang and Gobler, 2009b; Li et al., 2012; Griffith et al., 2019c), reef building corals (Guzmán et al., 1990; Bauman et al., 2010; Richlen et al., 2010), copepods (Jiang et al., 2011), and shrimp (Pérez-Morales et al., 2017). Specifically, M. polykrikoides can also cause rapid mortality of larval stages of northwest Atlantic bivalves (Tang and Gobler, 2009b; Griffith and Gobler, 2016). In addition, it has been demonstrated that M. polykrikoides has species-specific effects on juvenile bivalves where a combination of bloom vertical migration, bloom duration, temperature, and inability for sessile fauna to evade blooms can lead to mortalities in Crassostrea virginica (eastern oyster; (Gobler et al., 2008; Li et al., 2012)), Argopecten irradians (bay scallop; (Gobler et al., 2008; Tang and Gobler, 2009a; Griffith et al., 2019c)), and Mytilus spp. (mussel species; (Curtiss et al., 2008)). In contrast, M. mercenaria (hard clam) has proven to be more resilient to this HAB, surviving dense blooms that have caused mortality or reduced growth rates in other bivalves such as M. virginica and A. irradians (Griffith et al., 2019c). Here, bivalve clearance rates were quantified to understand how M. polykrikoides might alter the behavior of these animals.

Bivalve clearance of HABs has been studied for a number of algal taxa with clearance rates varying between bivalve and algal species (Bricelj et al., 2001; Hegaret et al., 2007; Harke et al., 2011). Studies have been performed to address the effects of other toxins and harmful compounds, such as exposure to ROS produced by HABs on survival or clearance rate of bivalves (Jeong et al., 2004; Hegaret et al., 2007; Tang and Gobler, 2009b). Similar to the findings of this study, bivalves have been shown to clear M. polykrikoides at various rates (Jeong et al., 2004). Specifically, Jeong et al. (2004) demonstrated that mussel larvae can successfully clear M. polykrikoides cells from the water column after two weeks of development. Like all invertebrates, hard clams do not have acquired immunity and are limited to innate immunity through cellular response using hemocytes (Janeway, 1994). They also have an open circulatory system where particulates that are being taken in from the environment come into direct contact with clam hemolymph (Janeway, 1994; Hegaret, 2008). During a study by Hegaret (2008), the ROS-producing alga Heterosigma akashiwo displayed loss of shape and complexity and increased mortality as hard clam hemocytes engulfed cells through phagocytosis and use of an enzymatic chemical agent released by the hemocyte cells when hard clam hemolymph was exposed to H. akashiwo. Although H. akashiwo has cytotoxic properties that led to increased hard clam hemocyte death, suppressed phagocytosis, and increased apoptosis, hard clam hemocytes were the main drivers of harmful algal cell death in H. akashiwo and other HABs when compared to other bivalve hemocytes (Hegaret, 2008). M. polykrikoides cells may have a similar fate when cleared by juvenile hard clams.

The M. polykrikoides strain, CPSB-1G was cleared at rates similar to the nonharmful food source, R. salina, in single alga experiments while the CP1 strain and M. polykrikoides bloom populations were not cleared, suggesting possible hard clam resilience to a variety of M. polykrikoides strains and toxicities. Although hard clams have lower clearance rates than Eastern oysters and mussels (Bricelj and Malouf, 1984; Shumway et al., 1985; Leverone et al., 2007; Harke et al., 2011) they seem to display greater resistance to the effects of M. polykrikoides (Griffith et al., 2019c). Historically, hard clam populations have experienced mass die offs in response to some HABs, particularly those caused by Aureococcus anophagefferens (Gobler et al., 2005). In scenarios where hard clam populations have been restored to high densities or experimentally enhanced, this bivalve has been shown to clear the harmful brown tide alga at densities<10⁵ cells mL^-1 (Bricelj et al., 2001; Cerrato et al., 2004; Harke et al., 2011; Gobler et al., 2022).

While M. polykrikoides is a well-known toxic alga, there is variation in toxicity between strains of different origins (Kim et al., 2002; Tang and Gobler, 2009a; Wang et al., 2020; Yang et al., 2022), temperatures (Griffith and Gobler, 2016), and cellular growth stage (Tang and Gobler, 2009a; Griffith and Gobler, 2016). During this study, clams successfully cleared multiple strains of M. polykrikoides although the rates at which each strain was cleared varied, in some cases significantly. Clams cleared the clonal strain CPSB-1G faster than the CP1 strain and wild populations when exposed to each as monocultures. While in mixed assemblages with the non-harmful alga, R. salina, clearance of this non-harmful alga was faster than the clearance of CPSB-1G. Several possible scenarios regarding strain toxicity may be responsible for variation in clearance of M. polykrikoides strains and R. salina. The strain CPSB-1G has been shown to be similarly ichthyotoxic as strain CP1 (Yang et al., 2022). However, during this study, clams cleared CPSB-1G significantly faster than CP1, suggesting that clam clearance is not affected in the same manner as fish survival perhaps because the potentially toxic effects of this alga only manifest once a cell is physically in contact with the clam digestive tract (Rosa et al., 2018). We further note that there was large variance in the clearance rates of clams feeding on wild populations (e.g. Figure 6 ) and that this variation could have been driven by differing toxicity of bloom water collected from different sites on different dates ( Table 1 ). It has been established that particle selection by bivalves is partly mediated by cell-surface lectins which differ across differing algal species and strains (Espinosa et al., 2009; Espinosa et al., 2010). While this was not addressed in this study, it is possible that the noxious nature of M. polykrikoides and its production of ROS result in the post-consumption rejection of cells as pseudo-feces (Ward et al., 1998; Ward and Shumway, 2004; Mafra et al., 2009) with the subsequent fate of the algal cells (reemergence, cysts formation, death) unknown.

Beyond variation in toxicity due to genetic origin and biogeochemical parameters, the toxicity of clonal isolates of harmful algae can change over time (Villac et al., 1993; Bates, 1998; Bricelj et al., 2001; Furey et al., 2007). Although clonal strains are typically exposed to uniform conditions, toxin and or harmful compound concentrations can vary over time in culture as has been seen, for example, in the diatom Pseudo-nitzschia for which genetic changes, changes in the bacterial composition, and/or a lack of sexual reproduction (Villac et al., 1993; Bates, 1998) can cause a decrease in cellular toxin levels in as little as one year in culture (Villac et al., 1993). Aureococcus and Alexandrium are HABs that have also displayed decreases in toxicity in culture over time (Bricelj et al., 2004; Juhl et al., 2008; Harke et al., 2011). In contrast, regarding Karenia brevis, the “Wilson clone” has significantly higher brevetoxin production after six decades in of culture when compared to other more recently isolated strains (Lekan and Tomas, 2010). These toxicity changes in culture are consistent with the hypothesis that CPSB-1G and CP1 clones of M. polykrikoides may produce ROS at different rates and at rates that differ from prior studies (e.g. Tang and Gobler, 2009a; Yang et al., 2022)) and, therefore, were cleared at different rates by M. mercenaria (this study). This would be consistent with the more rapid clearance of strain CPSB-1G compared to wild populations and strain CP1, suggesting that overtime CPSB-1G had become less noxious compared to populations in the wild.

There were clear signs during this study that M. polykrikoides altered the feeding of M. mercenaria. While hard clam clearance of R. salina was not significantly different than CPSB-1G during single species exposure, clams cleared the cryptophyte significantly faster than M. polykrikoides during mixed algal treatments. This difference highlights the ability of hard clams to selectively clear non-harmful, beneficial algal species (Ward et al., 1998; Ward and Shumway, 2004; Mafra et al., 2009). In comparison, when the similar sized, nontoxic dinoflagellate G. aureolum was present with R. salina, hard clam clearance rates did not differ between the two species, emphasizing the noxious nature of M. polykrikoides relative to this non-harmful dinoflagellate. Similarly, the clearance of R. salina by the clams was significantly slower when mixed with M. polykrikoides bloom water compared to when the HAB was not present. The reduction in clearance rates of the cryptophyte may be due to higher exposure levels of ROS, extracellular polysaccharides, and other cytotoxic compounds mentioned previously (Kim et al., 2002; Hegaret, 2008; Yang et al., 2022). Finally, hard clams did not clear M. polykrikoides bloom water during single species exposures but when mixed with R. salina, they did successfully clear M. polykrikoides, highlighting possible facilitation of M. polykrikoides clearance by the nonharmful alga.

Allelopathy is the inhibition of one photosynthetic organism by another, and M. polykrikoides has been shown to be a strongly allelopathic HAB, likely due to the production of ROS (Tang and Gobler, 2010; Yang et al., 2022). Unlike toxicity amongst HABs that produce saxitoxin (Bricelj et al., 2005) and domoic acid (Bates, 1998), ROS are produced and exuded by actively growing M. polykrikoides cells and can have inhibitory effects on target algae in minutes-to-hours (Tang and Gobler, 2010). Prior studies on the allelopathic effects of M. polykrikoides have shown that R. salina is more resistant than other phytoplankton species but can still experience significant mortality after 24 h exposure to high densities of M. polykrikoides (2,400 cells mL^-1; Tang and Gobler (2010)). During the present study, R. salina cells were alive and swimming in co-cultures with M. polykrikoides over the course of the 1-hour clearance rate experiments although signs of deformation and aggregation of some cells were observed. While such morphological changes may have altered clearance rates in mixed cultures, more work is needed to understand the quantitative impact of such changes. Culture experiments with ROS scavenging enzymes could provide additional insight on hard clam clearance in the presence of multiple algal species. In addition, quantification and examination of pseudofeces could indicate the extent to which hard clams ingested M. polykrikoides cells, and expelled live cells, expelled dead cells, or induced cyst formation.

A higher seston load can induce a variety of food selection responses in bivalves (Bricelj and Malouf, 1984). Some species adjust to larger particle loads by producing pseudo-feces while others, such as hard clams, control their ingestion by reducing clearance rates (Bricelj and Malouf, 1984; Shumway et al., 1985). During this study, higher biovolume experiments were completed using M. polykrikoides bloom water and R. salina to identify a possible threshold for juvenile hard clam clearance. There was no noticeable difference with regard to whether clams would open or close between total biovolume, species, or proportion of M. polykrikoides. In contrast, clams cleared M. polykrikoides at significantly lower rates during the higher biovolume treatments of 1,500 and 3,000 M. polykrikoides cells mL^-1 compared to 1,000 M. polykrikoides cells mL^-1, a finding consistent with prior studies showing that M. mercenaria reduces its feeding rate under higher particles loads (Bricelj and Malouf, 1984; Shumway et al., 1985). Additionally, clams still cleared R. salina during the 1,500 M. polykrikoides cells mL^-1 biovolume experiments while not clearing M. polykrikoides highlighting possible selective clearance of the nonharmful algal as seen during the mixed culture experiments between CPSB-1G and R. salina. This finding also emphasizes that the impacts on clearance rates at high biovolume levels may be more a function of the toxicity of M. polykrikoides and/or total biovolume rather than total particle load given that the R. salina biovolume equivalent treatments had significantly higher cell densities. While M. mercenaria actively clears M. polykrikoides at densities of<1,000 cells mL^-1, these findings suggest as blooms intensify, the filtration provided by this bivalve will subside.

Bivalve fisheries represent a multi-billion-dollar industry in the US but hard clam populations in the northeast US, including NY, have suffered significant reductions partly due to HABs (Cerrato et al., 2004; Gobler et al., 2005). In turn, these declines may permit for the intensification of some HABs (Cerrato et al., 2004; Gobler and Sunda, 2012). Although clearance rates of M. mercenaria exposed to bloom water were significantly lower compared to exposure to one of the clonal cultures, hard clams cleared M. polykrikoides from bloom water when mixed with the nonharmful alga R. salina. M. polykrikoides bloom densities can vary from 10² mL^-1 to >10⁴ mL^-1 within dense bloom patches (Gobler et al., 2008; Kudela and Gobler, 2012). Such dense blooms are dynamic, migrating vertically daily and affecting surface and benthic fauna including bivalve shellfish (Park et al., 2001; Kim et al., 2010; Griffith and Gobler, 2016). Given this study explored the impacts of up to 3,000 cells mL^-1 and found a cessation of feeding at ≥ 1,500 cells mL^-1, it is likely extreme bloom densities would have been similarly impactful on M. mercenaria. Regardless, although M. polykrikoides is strongly toxic to many organisms including bivalves (Kudela and Gobler, 2012), hard clams are more resistant than other bivalve shellfish when exposed to blooms (Griffith et al., 2019c). Results herein demonstrate that at moderate M. polykrikoides densities (≤ 1,000 mL^-1), hard clams can clear this HAB from the water column. Given that adult hard clams are generally more resilient to environmental stressors than juvenile stages (Kraeuter and Castagna, 2001), we hypothesize that clearance rates of adult hard clams would be less inhibited than the juveniles studied here.

Restoration is a major part of estuarine research and coastal environmental protection across the globe (Duarte et al., 2020). One avenue of such efforts includes shellfisheries restoration such as hard clam spawner sanctuaries aiming to restore hard clam populations in regions where they may be recruitment limited (Doall et al., 2008; Gobler et al., 2022). Along the US East Coast, hard clams can represent a significant proportion of the benthic community and, in some cases, may help mitigate HABs (Cerrato et al., 2004). For example, M. polykrikoides blooms occurred regularly in Shinnecock Bay, NY, USA, from 2006 to 2017 (Gobler et al., 2008; Hattenrath et al., 2016). An active restoration of the M. mercenaria population in this estuary increased densities of these bivalves by more than 17-fold and reduced the turnover time due to hard clam filtration of the western half of Shinnecock Bay from >21 days to< 10 days from 2012 to 2017 (Gobler et al., 2022). Consequently, there has not been a bloom of M. polykrikoides in western Shinnecock Bay since 2017.