All phytoplankton were grown in a culture cabinet at 15°C with a 16:8 h (Light:Dark) photoperiod, using seawater collected from the Western Channel Observatory monitoring site L4 off Plymouth, United Kingdom¹, amended with either F/2 (Rhodomonas sp.) or L1 (A. catenella) medium (Guillard and Ryther, 1962; Guillard and Hargraves, 1993). Acartia tonsa were purchased from Reefshotz, Cardiff, United Kingdom, where they have been cultured for >10 years, and were subsequently cultured in artificial seawater (Tropic Marin Sea salt, Classic Meersalz; salinity 31) at room temperature (21 ± 2°C) for >15 months prior to experimentation. Details of the culturing conditions for all phytoplankton species and for A. tonsa can be found elsewhere (Abdulhussain et al., 2020). The toxic A. catenella cultured was strain 1119/27 (CCAP, Loch Ainort, Scotland), and the non-toxic A. catenella was strain 1119/19 (CCAP Port Ellen, Islay, Scotland). A. tonsa cultures were fed Rhodomonas sp. (59.0 ± 7.8 pg C cell^–1) before experiments 1 and 2 and Isochrysis galbana (25.5 ± 0.3 pg C cell^–1) before experiment 3 at a concentration of 60,000 cells ml^–1 (2,000 ± 1,000 μg C L^–1) three times a week.

Experiment 1 examined how the survival of A. tonsa was affected by a 10-day exposure to either food deprivation (starved) or one of three different types of food: Rhodomonas sp. (RHO), toxic A. catenella (TAC) and non-toxic A. catenella (NAC). Prey cells were offered at a nominal concentration of 1,000 μg carbon (C) L^–1 to ensure that feeding conditions were always saturating (Berggreen et al., 1988; Abdulhussain et al., 2020). Individual adult females were incubated in 15 ml glass dram vials with 10 ml of sterile filtered L4 seawater (0.2 μm; SFSW hereafter) for 10 days at 19°C with a 12:12 h (L:D) photoperiod. Each of the four treatment levels was replicated fifteen times (total = 60 bottles). Nominal prey concentrations were achieved by determining the algal stock culture concentration using a Beckman Multisizer 3 Coulter Counter equipped with a 70 μm aperture and subsequently diluting it with an appropriate volume of SFSW. Each copepod was carefully transferred into a petri dish at the end of every 24 h incubation period and examined for survival, determined by movement after mechanical stimulus. Surviving copepods were subsequently pipetted into clean 15 ml vials of fresh SFSW containing the same species and concentration of prey cells that they had previously been exposed to. Algal concentrations in the stock cultures were counted every other day using the Coulter Counter. Triplicate 5 ml samples from the stock cultures of each of the three prey types (RHO, TAC, and NAC) were collected onto pre-combusted 13 mm glass fiber filters (Whatman GF/F) on the start of days 1 and 9 of the experiment, dried overnight at 40°C and stored in a desiccator prior to determining their carbon and nitrogen content (Carlo Erba EA-1108 elemental analyzer). Triplicate 30 ml samples of the TAC stock culture were collected at the outset of the experiment, centrifuged at 3,000 rpm for 20 min, and the resulting pellets were transferred into 2 ml Eppendorf tubes and stored frozen at −80°C until toxin profile analysis (see section “Toxin Analysis”).

Experiment 2 examined how the egg production rate of A. tonsa was affected by different types of food: RHO, TAC, and NAC. Prey cells were offered at a nominal concentration of 1,000 μg C L^–1. Four adults (three females and one male) were incubated in 150 ml LPDE plastic bottles with an inner mesh compartment of 120 μm (to separate the adult from the eggs produced during the experiment) filled with 100 ml of SFSW for 2 days at 19°C with a 12:12 h (L:D) photoperiod. Each of the four treatment levels (starved, RHO, TAC, NAC: 12 bottles) was replicated three times. The experiment was run over 2 days, with the first day being considered as an acclimation period (Breier and Buskey, 2007; Cohen et al., 2007; Jiang et al., 2009; Waggett et al., 2012; Abdulhussain et al., 2020). Hatching success was therefore only calculated for eggs produced on day 2. After the first day, the mesh compartments containing copepods were transferred to another LPDE plastic bottle with fresh SFSW containing the same species and concentration of prey cells that they had previously been exposed to. The eggs left in the LPDE plastic bottles from day 1 were counted under a stereomicroscope (to observe the response of A. tonsa to TAC during the acclimation period: see Supplementary Material). At the end of day 2, copepods in each treatment were carefully transferred into a petri dish and examined for survival (see section “Influence of Toxic A. catenella, Starvation, and Non-toxic Prey Alternatives on the Survival of A. tonsa”). Eggs produced on day 2 were counted and then transferred into petri dishes for 48 h at 19°C with a 12:12 h (L:D) photoperiod before counting the eggs hatched for each group, and calculating percentage hatching success (eq. 1):

Samples to determine the toxin- and elemental-content of the algal cultures were collected on the start of day 2 (see section “Influence of Toxic A. catenella, Starvation, and Non-toxic Prey Alternatives on the Survival of A. tonsa”).

Experiment 3 was conceptually similar to experiment 2, but examined how the egg production rate and hatching success of A. tonsa was affected by low (100 μg C L^–1) and high (1,000 μg C L^–1) concentrations of: RHO, TAC, and a diet where the carbon content was derived from 50% RHO to 50% TAC (MIX hereafter). Each of the seven treatment levels (starved, low RHO, high RHO, low TAC, high TAC, low MIX, and high MIX) was replicated three times (two times for low MIX; total = 20 bottles). Samples to determine the toxin- and elemental-content of the algal cultures were collected at the start of the experiment (after the acclimation period) (see section “Influence of Toxic A. catenella, Starvation, and Non-toxic Prey Alternatives on the Survival of A. tonsa”).

The frozen pellets of A. catenella from each experiment were mixed with 1.5 ml of 1% acetic acid before being vortex-mixed for 90 s at 2500 rpm. Solutions were subjected to probe sonication (Sonic Dismembrator, Fisher Scientific) set to 30% power for 1.5 min per sample. After sonication, vials were re-mixed, and 400 μL taken for desalting clean-up using carbon solid phase extraction (SPE). Details of the toxin extraction, column, mobile phases, chromatographic separation, and HILIC-MS/MS analysis conditions are described elsewhere (Turner et al., 2015, 2019; Abdulhussain et al., 2020).

The effects of 10 days of exposure to food deprivation, RHO, TAC, and NAC on the survival of A. tonsa were examined using Kaplan Meier survival analysis:

Where S is the Survival probability; t is time in days; N is the number of A. tonsa that survived; and D is the number of A. tonsa that died. Curve comparisons were achieved using the Log-rank test. Standard errors are presented to show the deviation from the population mean.

The influence of feeding conditions (starved, RHO, TAC, and NAC) on the egg production of A. tonsa in experiment 2 was examined using 1-way ANOVA. Egg hatching success was examined using a 1-way ANOVA. The influence of food type (starved, TAC, MIX) and food concentration (100 or 1,000 μg C L^–1) on the egg production of A. tonsa in experiment 3 were examined using 2-way ANOVA. Egg hatching success was examined using a 2-way ANOVA. All explanatory variables were treated as categories. Post hoc comparisons were conducted using Tukey’s test with alpha = 0.05. Standard errors are presented for the egg production and hatching experiments to show the variation within each treatment. All statistical analyses were carried out using Prism Graphpad (v.8.4) software.

The survival of female A. tonsa declined significantly across the 10-day duration of the experiment (Figure 1; Log-rank test, Chi square = 29, df = 3, p < 0.001). However, survival was only significantly reduced in the starving group, where there was 100% mortality by day 9 (Figure 1; Log-rank test, Chi square = 29, df = 3, p < 0.001). Survival of the females offered RHO (73%), TAC (80%), and NAC (60%) at nominal concentrations of 1,000 μg C L^–1 were statistically indistinguishable, with an average survival of 71 ± 10% across these treatments (Figure 1; Log-rank test, Chi square = 1, df = 2, p = 0.525).

The egg production rate of A. tonsa was affected by the type of food they received (Figure 2A; 1-way ANOVA, F(3,8) = 13.24, p ≤ 0.0115). The females offered RHO and NAC produced significantly more eggs (17.7 ± 4.9 and 14.1 ± 5.7 eggs female^–1 day^–1, respectively) than those offered TAC (2.4 ± 2.3 eggs female^–1 day^–1) and starved (1.0 ± 0.6 eggs female^–1 day^–1) (Figure 2A). The hatching success of eggs produced on day 2 was influenced by the maternal feeding conditions (1-way ANOVA, F(3,7) = 11.53, p = 0.0042). Hatching success of the eggs from females that had received RHO (60 ± 9.7%) and NAC (56 ± 23.9%) was higher than the eggs from females fed TAC (0%) or starved (7 ± 11.5%) (Figure 2B).

The egg production rate of A. tonsa was affected by food type and food quantity (Figure 3A, 2-way ANOVA, food type: F(3,15) = 40.0; food quantity F(1,15) = 25.0, p ≤ 0.002 in both cases). Copepods that were fed on MIX at 1,000 μg C L^–1 produced significantly fewer eggs (4.7 ± 1.2 eggs female^–1 day^–1) than in the RHO treatment at the same concentration (9.0 ± 1.3 eggs female^–1 day^–1) but significantly more than the females that received TAC (0.3 ± 0.3 eggs female^–1 day^–1) (Figure 3A). Egg production rates in the 100 μg C L^–1 treatments were all similarly low (Figure 3A). The hatching success of the eggs produced was influenced by the food type and food quantity (Figure 3B, 2-way ANOVA, F(3,15) = 15.39 and 6.32, respectively, p ≤ 0.0238 in both cases). The hatching success of eggs from females that were fed 1,000 μg C L^–1 of RHO (58 ± 14%) and MIX (56 ± 25%) were both higher than TAC (0%) and starved (0%) (Figure 3B), whereas the hatching success of eggs from females fed at 100 μg C L^–1 of RHO (67 ± 18%), MIX (60 ± 28%) and TAC (62 ± 28%) were all similar.

The elemental content, and composition and concentrations of toxins, of A. catenella (1119/27) were all variable (Tables 1, 2), despite being grown under near identical conditions. For example, the total toxin concentration was 501.1 STX eq cell^–1 for experiments 1 and 2, and 3575.5 STX eq cell^–1 for experiment 3. Some of the factors that can change the toxin levels within a HA bloom are turbulence (Juhl et al., 2001), salinity (Grzebyk et al., 2003), and nutrient conditions (Turner et al., 1998; John and Flynn, 2002; Leong et al., 2004). Toxin levels can also vary with growth phase in cultured algae (Hamasaki et al., 2001; Cook et al., 2010). We suggest that the differences in toxicity in our laboratory-based study were caused by differences in the nutrient concentrations of the seawater that was used for culturing TAC: Experiment 1 and 2 were conducted during January 2020, and experiment 3 during July 2019, when surface nitrate and phosphate concentrations were close to their highest and lowest, respectively². Moreover, in response to nutrient stress and chemical signals from copepods, HA can dramatically increase the development of PSP toxins (Wohlrab et al., 2010; Selander et al., 2015; Griffin et al., 2019). A group of predator cues called copepodamides (also known as taurine-containing lipids; Mayor et al., 2015), which are exuded by three species of calanoid copepods (Centropages typicus, Pseudocalanus sp., and Calanus sp.) have been found to trigger increased PSP toxins in dinoflagellates, as well as amnesic shellfish poisoning (ASP) toxins in diatoms (Selander et al., 2019). It is therefore possible that the toxin content of the prey cells increased during our incubations and the concentrations presented in Table 2 should be considered minimal. These differences highlight the importance of reporting toxin profile concentrations, particularly in the context of copepod survival, grazing, and egg production studies. This makes it easier to robustly compare between different studies.

Exposure of A. tonsa to a toxic strain of TAC for 10 days did not significantly affect their survival relative to the NAC treatment. Rather, increased mortality was only observed in the starved treatment (Figure 1). These findings suggest that A. tonsa is able to feed and survive on TAC over the likely duration of a typical bloom of TAC (Bresnan et al., 2008), regardless of their toxicity. Previous 24 h experiments showed that A. tonsa can feed upon TAC and survive blooms two orders of magnitude higher than the maximum densities of A. catenella recorded in natural systems (Abdulhussain et al., 2020), possibly by producing detoxification enzymes (Dutz, 1998; Costa et al., 2008) or by reducing food absorption efficiency (Asselman et al., 2012; Roncalli et al., 2016). Bloom concentrations of TAC are generally low and typically last for only a few days (Bresnan et al., 2005), although more persistent blooms are reported (Bresnan et al., 2008). Our findings therefore support the understanding that Acartia sp. tolerate exposure to TAC for the typical duration of naturally occurring blooms without negative impacts on the survival of adults (Liu and Wang, 2002; Estrada et al., 2008). However, exposure to high densities of HA for an extended period may cause substantial mortality within Acartia spp. populations (Prince et al., 2006; Barreiro et al., 2007; Jiang et al., 2009). Thus, the response of Acartia spp. appears to vary between different studies, possibly because of different toxin concentrations (section “A. catenella Toxin Composition and Concentrations”). Higher survival rates of A. tonsa in this study compared to other survival experiments could also be a result of reduced encounter rates between individuals as each experimental vial contained only one female per treatment replicate. However, in denser cultures, elevated encounter rates between individuals may increase levels of stress and cannibalism, potentially resulting in higher rates of mortality (Drillet et al., 2015).

The average egg production rate of A. tonsa feeding on RHO in our 2-day exposure experiments, ∼17 eggs female^–1 day^–1 (Figure 2), agrees with values previously reported for the genus Acartia sp. (range = 9–76 eggs female^–1 day^–1; Prince et al., 2006; Barreiro et al., 2007; Breier and Buskey, 2007; Jiang et al., 2009; Waggett et al., 2012). Our results reveal that egg production and hatching success were affected by the types of food available and the concentration at which it was offered (Figures 2, 3), with the observed values being significantly lower in the TAC treatments relative to those in the NAC and RHO treatments (Figure 2). Egg production rates by females fed RHO, NAC and MIX (50% RHO, 50% TAC) at 1,000 μg C L^–1 were also significantly higher than those fed TAC alone. By contrast, the rates of females fed TAC were not statistically different to those of the starved animals (Figures 2, 3). Furthermore, the percentages of eggs to successfully hatch from females offered mono-specific diets of toxic A. catenella at concentration of 1,000 μg C L^–1 were always close to zero. These results are consistent with the observation that reproduction in Acartia clausi is negatively affected when consuming toxic Alexandrium minutum, and that these effects can be alleviated by offering mixed diets (Barreiro et al., 2006, 2007).

Acartia tonsa can discriminate between different species of Alexandrium with different toxin contents, showing preference for the least toxic species (Xu et al., 2018; Abdulhussain et al., 2020). Copepods are thought to be capable of rejecting toxic prey cells (Cook et al., 2010; Abdulhussain et al., 2020) or consume them at lower rates than a similarly sized non-toxic cells (Xu and Kiørboe, 2018). It is therefore possible that the reduced egg production rate in the TAC treatment could be caused by starvation. However, we know that A. tonsa is capable of surviving and feeding on TAC when alternative prey are not available (Abdulhussain et al., 2020). This understanding is further supported by the current study, where survival of A. tonsa in TAC treatment was significantly higher than in the starved treatment (section “Survival of A. tonsa Exposed to Toxic A. catenella”). The significantly lower egg production rates observed in the TAC treatments of both experiments is therefore unlikely attributable to a starvation effect. The negative effects of TAC on the observed egg production rates and hatching success may have occurred because (a) the females were directly impeded by the toxins, (b) the females needed to produce detoxification enzymes, and therefore had to channel energy and other resources away from producing viable eggs, (c) the presence of toxins reduced the efficiency with which food is absorbed, and therefore reduced the amount of energy and other resources available for producing viable eggs, or (d) TAC were in some way nutritionally inferior to the other diets offered.

Nutritionally deficient food and the presence of toxins have both been suggested to cause reduced egg production in Acartia sp. (Dutz, 1998; Collumb and Buskey, 2004; Prince et al., 2006; Costa et al., 2008). The survival of females exposed to TAC was not significantly lower than females exposed to the non-toxic prey treatments, apparently suggesting that TAC toxins do not necessitate additional metabolic costs associated with the production of detoxification enzymes to the extent that they have a discernible, negative effect on the overall physiological performance of A. tonsa. However, copepods may increase their ingestion rates when feeding on toxic dinoflagellates in order to compensate for the increased energy required for detoxification (Dutz, 1998; Costa et al., 2008) or reduce the efficiency with which they absorb food (Asselman et al., 2012; Roncalli et al., 2016). For example, ingestion rates of A. tonsa were higher when fed on toxic A. catenella relative to animals fed on non-toxic Rhodomonas sp. at a food saturating concentration of 1,000 μg C L^–1 (Abdulhussain et al., 2020). Compensatory feeding may also be necessary if the nutritional value of HA is lower than other non-toxic food (Collumb and Buskey, 2004; Prince et al., 2006). Our survival experiments were conducted at a food saturating concentration of 1000 μg C L^–1 (Berggreen et al., 1988; Abdulhussain et al., 2020) and we cannot therefore exclude the possibility that the females increased their ingestion rates to offset any additional metabolic costs. Nevertheless, even if compensatory feeding did occur, it was clearly insufficient to mitigate the negative effects on reproduction associated with ingesting TAC. Egg production rates of females fed NAC, which we assume to have a highly similar nutritional make up to TAC, were indistinguishable from those fed on RHO, a food source that is widely assumed to support optimal reproductive output in Acartia spp. (Støttrup et al., 1986; Berggreen et al., 1988). We therefore suggest that the reduced rates observed in the TAC treatment (Figure 2) were not attributable to the absence of nutritive compounds; they occurred because of the presence of toxins, although the specific mechanism cannot be elucidated here.

The negative effects of consuming TAC were less apparent when prey cells were offered at 100 μg C L^–1, where egg production rates were similarly low across all treatments (Figure 3A). This agrees with the understanding that lower food concentrations limit growth, reproduction, and survival (Berggreen et al., 1988). Interestingly, the hatching success of the few eggs produced by females fed TAC at 100 μg C L^–1 was similarly high when compared to the values observed in the other, non-toxic treatments. This may suggest that there is a toxin-threshold level, beyond which eggs do not hatch.

Prey cell size can influence the outcome of copepod feeding experiments. However, in our study, differences between Rhodomonas (6–8 μm ESD) and I. galbana (4–6 μm ESD) were trivial, suggesting that these are unlikely to cause differences in the feeding rates of our experimental animals. Indeed, previous work has shown that female A. tonsa have an optimal prey size of 14.8 μm ESD, but are able to achieve high clearance rates when feeding on both Rhodomonas and on the dinoflagellate A. catenella (ESD = 20–40 μm) (Berggreen et al., 1988; Abdulhussain et al., 2020). Moreover, despite differences in the feeding histories of the animals in experiments 2 and 3, the outcomes of these two experiments were qualitatively and quantitatively the same: EPR and hatching success in the TAC and starved treatments in both experiments were similarly low (not different to zero); EPR and hatching success in the RHO treatments in both experiments were always higher than in the TAC and starved treatments (Figures 2, 3); ERP in the RHO treatments were not statistically different between the two experiments (Paired t-tests: t = 3.965, df = 2, p = 0.058). These results would not have occurred if any feeding history effects had not been accounted for within the 24 h acclimation period.

Finally, in the real environment, there are usually more than one species of algae available for copepods to feed on. The availability of alternative prey suggests that the impact of HA on copepod egg production may be very low due to availability of other algae. Nevertheless, it is clear that further work is required to understand the energetic and nutritional requirements of A. tonsa and their developing embryos, and how these are directly and indirectly affected when ingesting toxin-producing prey species.