Intoxicated copepods: ingesting toxic phytoplankton leads to risky behaviour

Abstract

Understanding interactions between harmful algal bloom (HAB) species and their grazers is essential for determining mechanisms of bloom proliferation and termination. We exposed the common calanoid copepod, Temora longicornis to the HAB species Alexandrium fundyense and examined effects on copepod survival, ingestion, egg production and swimming behaviour. A. fundyense was readily ingested by T. longicornis and significantly altered copepod swimming behaviour without affecting copepod survival or fitness. A. fundyense caused T. longicornis to increase their swimming speed, and the straightness of their path long after the copepods had been removed from the A. fundyense treatment. Models suggest that these changes could lead to a 25–56% increase in encounter frequency between copepods and their predators. This work highlights the need to determine how ingesting HAB species alters grazer behaviour as this can have significant impacts on the fate of HAB toxins in marine systems.

1. Background

Blooms of toxic phytoplankton, commonly referred to as harmful algal blooms (HABs), have devastating consequences for coastal economies and can pose a significant threat to human health [1]. As such, there has been a great deal of research in recent decades targeting mechanisms that control bloom formation, propagation and termination. A key factor responsible for bloom proliferation is the lack of adequate grazing pressure, reviewed by [2–5]. Therefore, understanding interactions between grazers and HAB species is of critical importance.

Copepods are important grazers of HABs as they can serve as vectors for toxin entry into pelagic food webs [6,7]. Therefore, much work has been done to determine the effects of harmful algae on copepod survival and fitness. Behavioural studies have consisted mostly of measuring copepods' immediate feeding response to HAB species, such as cell rejection, avoidance or incapacitation [8–10]. Fewer studies have documented the effects of harmful algae exposure on copepod swimming behaviour, but see [11,12]. Owing to the high search volume rates of marine copepods (approximately 100 l d⁻¹ [13]) and the patchy nature of HABs [14–17], copepods may experience intermittent exposure to patches containing HAB species followed by HAB-free areas. However, little is known about the lasting effects of HAB exposure on copepod swimming behaviour. Effects of HABs on copepod behaviour could be important, as swimming behaviour has a direct impact on copepod dispersal and encounter rates with predators [18–20], both of which determine the propensity of copepods to graze HABs and transfer HAB toxins through marine food webs.

Alexandrium fundyense is a neurotoxic dinoflagellate that causes HABs along northeastern United States and southeastern Canada. These bloom events have devastating impacts on the economy of these areas owing to shellfish fisheries closures and outbreaks of paralytic shellfish poisoning (PSP), which cause severe human health effects [21,22]. Furthermore, the frequency and intensity of A. fundyense blooms have increased in recent years [22,23]. Grazing interactions between Alexandrium sp. are variable, reviewed by [3,4]; outcomes of these interactions include copepods behaviourally rejecting Alexandrium sp. [9,10,24], becoming incapacitated after feeding [8,25] or showing no apparent adverse effect [26,27]. Furthermore, Alexandrium sp. can have detrimental [28] or positive [29,30] effects on egg production rate. The effects of Alexandrium sp., can vary between copepod species [26], gender [31], phenotype [29] and historical exposure [25,32].

In this study, we ask the following questions. (i) Does T. longicornis exhibit altered swimming behaviour in response to A. fundyense exposure? (ii) Is this behavioural change explained by nutritional inadequacies or toxicity of the phytoplankton? (iii) How does this behavioural change affect copepod encounter rates with predators?

2. Materials and methods

(a). Collection and culture of organisms

Our target grazer species, T. longicornis, is a calanoid copepod that co-occurs with A. fundyense in the Gulf of Maine. We collected T. longicornis by boat from the Damariscotta River estuary, Walpole, ME (43°56′ N, 69°35′ W) by obliquely towing a plankton net with a mesh size of 250 µm at approximately 30 m depth. Upon collection, animals were immediately transferred into 20 l containers of surface seawater and transported to a temperature-controlled room. Animals were transferred to 1 l polyethylene bottles within 24 h of collection. Bottles were packed in an insulated box containing ice packs and cushioning material to minimize stress and maintain a cool temperature (approx. 5–10°C). Animals were shipped overnight to the Georgia Institute of Technology, Atlanta, GA where they were diluted in artificial seawater and allowed to acclimate to their natural temperature over 24 h. Copepods were fed mixed cultures of Tetraselmis sp., and Isochrysis galbana. After a 24 h acclimation period, adult T. longicornis were sorted from mixed samples and placed in containers filled with filtered seawater at densities of fewer than 25 individuals l⁻¹.

Alexandrium fundyense was chosen as our ‘HAB’ species because it is known to contain PSP toxins. The strain we used (NCMA 1719) was obtained from the National Center for Marine Algae and Microbiota, Boothbay Harbor, ME. Cell biovolume (6.61 × 10³ µm³) was calculated treating the cell as a rotational ellipsoid. Cell dimensions 26.6 ± 1.9 µm, 21.8 ± 0.5 µm (mean ± s.e.) were measured from images using a FlowCAM (n = 12). Rhodomonas lens (NCMA 739, cell biovolume 3.36 × 10² µm³ calculated as a rotational ellipsoid from dimensions 13.5 ± 1.8 µm, 6.9 ± 0.3 µm [mean ± s.e.], n = 7) was used as our non-toxic ‘control’ algae because it is commonly fed to copepods in culture owing to its nutritional quality [33]. All phytoplankton species were cultured in L1 media, exposed to a 14 : 10 light : dark cycle and maintained at 14°C (for A. fundyense) and 21°C (for R. lens). Phytoplankton were kept in exponential phase during culturing by diluting with media every few days. Phytoplankton treatments were prepared by conducting triplicate cell counts (each count consisting of more than 200 cells) of our stock phytoplankton cultures. We then removed the appropriate volume with a serological pipette and diluted with 0.7 µm filtered artificial seawater. The diluted phytoplankton mixture was cooled to the experimental temperature of 14°C if necessary. We inverted the phytoplankton mixture several times to promote a homogeneous cell distribution before pouring into experimental vessels.

Carbon : nitrogen ratios for A. fundyense and R. lens were analysed by Micro-Dumas combustion at the University of Georgia, Athens, GA. Culturing and harvesting conditions were kept constant during all experiments to minimize any physiological differences within phytoplankton species. All phytoplankton cultures were in late exponential phase when harvested for experiments. Maximum culture densities were 4.8 × 10⁷ and 1.3 × 10⁸ cells l⁻¹ for A. fundyense and R. lens, respectively.

(b). Toxin analysis

To verify the toxicity of our A. fundyense stock culture, two samples were shipped overnight to Greenwater Laboratories in Palatka, FL. The first sample was analysed using a competitive enzyme-linked immunosorbent assay (ELISA) to detect the presence of saxitoxin (the most potent of the PSP toxin derivatives). However, ELISA is not reactive with all PSP toxin derivatives and often underestimates the total toxins present. Owing to logistical constraints, a detailed toxin profile was analysed only for the second sample. The toxins examined were N-sulfocarbamoyl-11-hydroxysulfate toxins (C1 and C2), gonyautoxin 1–4, saxitoxin, neosaxitoxin and other decarbamoyl derivatives. Toxins were analysed by high-pressure liquid chromatography (HPLC) with fluorescence detection following a pre-column oxidation method.

(c). Swimming behaviour

The aim of our behavioural experiment was to determine if the swimming behaviour of T. longicornis is altered by exposure to A. fundyense. Fifteen male and 15 female T. longicornis were placed in a 1 l glass tank containing filtered seawater and fed one of the following treatments: A. fundyense treatment that contained a mixture of 320 cells ml⁻¹ of A. fundyense + 1120 cells ml⁻¹ of R. lens or a control treatment that contained 5600 cells ml⁻¹ of R. lens. We calculated the diet composition using biovolume equivalents to ensure that copepods were given equivalent total biovolumes of food in both experimental groups. Copepods were starved for 12 h prior to experiments. We chose 320 cells ml⁻¹ for our A. fundyense treatment to ensure that copepods were not food limited. Although this concentration is quite high for A. fundyense blooms in the Gulf of Maine, where maximum cell densities are approximately 10 cells ml⁻¹ [34], this is within the range of A. fundyense cell densities measured in other areas [35]. Furthermore, we simultaneously offered T. longicornis an alternative food, R. lens, to avoid ‘force-feeding’ the copepods A. fundyense.

After exposing copepods to treatments for 2 h, we carefully transferred them to a tank with filtered seawater amended with Tetraselmis sp. under non-limiting conditions (ca 10³ cells ml⁻¹). Tetraselmis sp. has been fed to copepods during depuration periods in other A. fundyense grazing experiments [36]. Tanks were covered with parafilm and placed in a temperature-controlled room where copepods were allowed to feed for 15 h before being used in behavioural experiments. We repeated this exposure process with five replicate tanks for both treatments.

We chose a 2 h exposure time to mimic a brief encounter between a copepod and an A. fundyense patch but allow enough time for the copepods to fill their guts [37]. The purpose of a 15 h recovery period was to allow copepods to feed on control algae, so that changes in their behaviour would not be due to differences in hunger levels (caused by any differences in the nutritional value of A. fundyense versus R. lens). Fifteen hours is much longer than the average gut clearance time for marine copepods (approximately 1–2 h [38]), but less than the average residence time of saxitoxins harboured in copepod guts and tissues (approximately 33 h [36]).

Following the recovery period, the experimental tanks containing copepods were placed in a Schlieren optical system developed by Strickler & Hwang [39] and further described by Doall et al. [40]. Observations were conducted in the dark with a near-infrared laser used to illuminate the copepods. A green laser beam was directed down the centre of the vessel to attract copepods to the centre of the tank and reduce effects of vessel walls [40]. Preliminary observations revealed that the light encouraged copepods to swim in the centre of the tank, maximizing the number of swimming paths that could be analysed. Observations were recorded onto DVDs and digitized using Prism video converter 1.82 software and split into clips using SolveigMM video splitter. When necessary, clips were further processed in Adobe Premier Pro CS5.5 to enhance contrast. Swimming paths were analysed using LabTrack software. Our path selection criteria required that copepods were swimming in the centre of the tank for at least 8 s. For paths longer than 10 s, only the first 10 s were analysed. We then constructed three-dimensional tracks by matching the common z-position from superimposed mirror images of individual copepods.

We calculated instantaneous swimming velocity V (mm s⁻¹) of the animals using the distances between position in the x, y and z plane over a given time step t.

where x_t, y_t, z_t and x_{t +1}, y_{t +1}, z_{t +1} correspond to the copepod's positions at time t and time t + 1, respectively, and p is the frame acquisition rate of our camera (33 frames s⁻¹). Instantaneous velocities were averaged over an individual track to obtain a mean swimming velocity. Swimming speed has important consequences for T. longicornis distribution and survival, because increased swimming speed leads to greater dispersal distances, increased encounter rates with predators and increased conspicuousness [19,20,41]. We calculated the effect of changes in swimming velocity on encounter rates between T. longicornis and their predators (E) using the following equation [18]

where K is the perceptive distance of the predator and v_pred and v_prey are the swimming velocities of the predator and prey, respectively. The relative importance of the prey's swimming speed depends on type of predator encountered [19]. For a visual cruising predator such as a fish larvae, the encounter rate E increases proportionally as prey velocity v_prey increases [19]. However, for a rheotactic predator that relies on hydrodynamic cues from prey, the predator's detection distance (K) scales as

where (s) and (v_prey) are the prey's size and swimming velocity and (ω) is the sensitivity (threshold speed) that the predator can detect [19]. Therefore, a faster-swimming organism is more conspicuous to predators. In this case, the predator's encounter rate is roughly proportional to

Additionally, we calculated the net : gross displacement ratio (NGDR) for each swimming track. This is a commonly used metric that varies from 0 to 1 and describes the degree of path tortuosity [42]. An NGDR of 1 represents a perfectly straight path, whereas an NGDR of 0 describes Brownian motion. Because net : gross displacement ratios are inherently scale-dependent, we accounted for this scale-dependency by analysing paths of 8–10 s in length. High directional persistence of swimming paths indicates increased encounter rates between individuals [20,43].

(d). Survivorship experiments

To determine the effects of ingesting A. fundyense on copepod survival, we incubated T. longicornis in 250 ml beakers containing filtered artificial seawater amended with one of the following treatments: high food control, A. fundyense treatment, a low food control and a starved control (table 1). The percentages were based on biovolume equivalents to account for the different sizes of the phytoplankton cells.

Table 1.

Description of phytoplankton treatments administered during copepod behavioural, survival and reproductive experiments.

treatment	description	purpose
Alexandrium fundyense treatment	320 cells ml−1 of A. fundyense + 1200 cells ml−1 of Rhodomonas lens	contains both the HAB species A. fundyense and nutritious species R. lens
high food control	5600 cells ml−1 of R. lens	contains equal biomass of total phytoplankton as A. fundyense treatment
low food control	1200 cells ml−1 of R. lens	contains equal biomass of the nutritious phytoplankton, R. lens as A. fundyense treatment
starved	filtered seawater	contains no phytoplankton

To control for feeding history, T. longicornis were incubated in filtered seawater for 12 h prior to experiments. Three male and three female T. longicornis were randomly assigned to beakers (n = 10), which were covered with parafilm, placed in a 14°C incubator and exposed to a 14 : 10 h light : dark cycle. For all experiments, beakers were interspersed with respect to treatment during incubations. Every 24–25 h, copepods were visually inspected under a dissecting microscope and scored as either dead or alive. Live copepods were transferred to new beakers containing fresh phytoplankton and dead copepods were discarded. Experiments were conducted for 4 days.

(e). Ingestion experiments

We measured copepod ingestion rate in a subset of our beakers (n = 5) between 24 and 48 h of our survivorship experiment. Prior to the addition of animals, a 20 ml sample was removed from experimental beakers and preserved in Lugol's solution. The next day, copepods were removed from beakers, and another 20 ml sample was collected and preserved. To determine copepod ingestion rate, the number of cells removed was divided by the number of surviving copepods in the beaker and multiplied by the number of hours incubated (24 h).

Visual counts were performed using replicate 300 µl–1 ml preserved samples, so that a minimum of 500 cells were counted for each subsample. The rate of copepods ingesting each algal species was calculated from the formula described by Frost [44]

where C₁ is the cell concentration in the beaker just prior to adding copepods, C₂ is the cell concentration immediately after copepod removal, V is the cell biovolume and N is the number of live copepods at the end of the 24 h period. Animals that died during the incubation were assumed to have died halfway through the experimental period in calculations. We ran copepod-free controls with mixtures of A. fundyense and R. lens prior to grazing experiments to measure changes in cell density owing to growth or settling. We found no significant change in cell density over a 24 h period for either species (A. fundyense, p = 0.96, n = 4; R. lens, p = 0.68, n = 4, two-tailed t-test). Therefore, we did not include a growth parameter in our ingestion equation.

(f). Egg production and hatching success experiments

To determine the effects of A. fundyense on copepod fecundity, we incubated three male and three female T. longicornis in 250 ml beakers containing an ‘egg basket’ (n = 9). Egg baskets consisted of a 5.5 cm diameter and 10 cm height Plexiglas tube with 160 µm mesh attached to the bottom and their purpose was to create a false bottom through which copepod eggs could pass separating copepods from their eggs and thus preventing egg cannibalism [45]. To control for previous mating and feeding history, males and females were separately incubated for 24 h in filtered seawater prior to experiments. Copepods were individually added to beakers containing one of the following treatments: high food control, A. fundyense treatment, a low food control and filtered seawater control (table 1). Beakers were covered with parafilm and placed in a 14°C incubator and exposed to a 14 : 10 h light : dark cycle. Every 24–25 h, egg baskets containing copepods were moved to new beakers with fresh phytoplankton. Copepods found lying on the bottom of the egg basket were visually inspected under a dissecting microscope. Dead copepods were discarded. Experiments were conducted for 3 days.

We maintained a 1 : 1 sex ratio within our treatments to control for any bias in hatching success owing to either fertilization limitation or mate competition. Therefore, if a female died during the incubation, we removed a male from that replicate. Dead males were replaced with a new male from a supplemental stock culture that had been incubated in the appropriate treatment since the beginning of the experiment.

After moving copepods to new beakers, we gently transferred eggs to a Petri dish and counted them immediately. A subset of egg samples were covered with parafilm and placed in a 14°C incubator for 48 h (n = 4). After 48 h, we added a few drops of acetic acid to the Petri dishes to kill and stain the nauplii for enumeration. For the purpose of analysis, we pooled the total number of eggs and nauplii produced per female per replicate from 48 h and 72 h time periods (excluding the 0–24 h time period) to ensure that copepods had enough time to assimilate ingested phytoplankton [38].

(g). Statistical analyses

The number of trajectories analysed per trial depended on the number of trajectories that fit our selection criteria (i.e. copepod swimming near the centre of tank for at least 8 s), which yielded 10–15 swimming paths per trial. We averaged individual swimming speed and NGDR values within trial and consider experimental tank the unit of replication. Our analysis is based on a total of 64 trajectories consisting of 19 411 successive animal positions for the R. lens control and 56 trajectories (16 297 positions) for A. fundyense treatment (n = 5 replicate tanks per treatment). We used a Welch's t-test to detect statistical differences in swimming speed and path tortuosity between A. fundyense and R. lens treatments.

The total number of surviving copepods and number of eggs produced per female between phytoplankton treatments were analysed using an analysis of variance (ANOVA). Per cent hatching success was calculated from the total number of nauplii and eggs produced per replicate beaker. Differences in hatching success across treatments were analysed by a generalized linear model (GLM) with a logit-link function with a quasi-binomial distribution (to account for overdispersion).

Differences in total ingestion rate between the high food control and the A. fundyense treatment were compared using a Welch's t-test. Differences in the ingestion of R. lens between the low food control and the A. fundyense treatment were also compared using a Welch's t-test. All statistical tests were conducted using R (v. 2.14.1).

3. Results

After being exposed to phytoplankton treatments for 2 h, followed by 15 h of consuming non-HAB food, copepods exposed to the A. fundyense treatment swam faster than those exposed to the control phytoplankton (figure 1a t = 2.4, d.f. = 8, p-value < 0.05, Welch's t-test). Directional persistence was also significantly different (figure 1b; t = 2.6, d.f. = 8, p-value < 0.05, Welch's t-test), with copepods exhibiting straighter swimming paths (figure 2) following A. fundyense exposure.

Figure 1.

Effects of ingesting an Alexandrium fundyense treatment (containing 320 cells ml⁻¹ of A. fundyense + 1200 cells ml⁻¹ of Rhodomonas lens) versus a high food control (containing 5600 cells ml⁻¹ of R. lens on the swimming speed (a) and net : gross displacement ratio (b) of T. longicornis (mean ± s.e.). Copepods were incubated in treatments for 2 h and then fed nutritious food for 15 h prior to behavioural recordings. Significant differences were determined using a Welch's t-test to compare mean swimming speed, (t = 2.4, d.f. = 7.9, p-value < 0.05) and net : gross displacement ratio (t = 2.6, d.f. = 7.3, p-value < 0.05).

Figure 2.

Three-dimensional depictions of 8–10 s swimming paths of Temora longicornis after ingesting an Alexandrium fundyense treatment (containing 320 cells ml⁻¹ of A. fundyense + 1200 cells ml⁻¹ of Rhodomonas lens) (a) versus a high food control (containing 5600 cells ml⁻¹ of R. lens) (b). Plots represent a subset of the total number of trajectories analysed for visual clarity (n = 40 per treatment). Copepods were incubated in treatments for 2 h and then fed nutritious food for 15 h prior to behavioural recordings.

The average number of survivors did not differ among phytoplankton treatments (figure 3; F_3,36 = 1.42, p = 0.25, ANOVA). Furthermore, there was no difference between the amount of total food ingested by copepods in the A. fundyense treatment versus the high food control (figure 4; t = 0.84, d.f. = 7, p = 0.43, Welch's t-test). Additionally, T. longicornis consumed similar amounts of R. lens in the low food control versus A. fundyense treatment (figure 4; t = 1.3, d.f. = 7, p = 0.23, Welch's t-test), indicating that T. longicornis did not discriminate against or become incapacitated by ingesting A. fundyense.

Figure 3.

Average number of surviving Temora longicornis incubated in an Alexandrium fundyense treatment (containing 320 cells ml⁻¹ of A. fundyense + 1200 cells ml⁻¹ of Rhodomonas lens), a high food control (containing 5600 cells ml⁻¹ of R. lens), a low food control (control 1200 cells ml⁻¹ of R. lens) or filtered seawater (starved control). The total number of survivors at the end of 4 d was compared using an analysis of variance, ANOVA (F_3,36 = 1.42, p = 0.25).

Figure 4.

The biomass of phytoplankton ingested by Temora longicornis over a 24 h period when six individuals were incubated in an Alexandrium fundyense treatment (containing 320 cells ml⁻¹ of A. fundyense + 1200 cells ml⁻¹ of Rhodomonas lens), a high food control (containing 5600 cells ml⁻¹ of R. lens), a low food control (control 1200 cells ml⁻¹ of R. lens) or filtered seawater (starved control). The total amount of food consumed between the high food control and the A. fundyense treatment was compared using a Welch's t-test. The amount of R. lens consumed between the low food control and the A. fundyense treatment was also compared using a Welch's t-test. There were no significant differences detected in the total amount of food ingested (t = 0.84, d.f. = 7, p = 0.43) or the amount of R. lens ingested (t = 1.3, d.f. = 7, p = 0.23).

There was no difference in copepod egg production (figure 5a; F_3,32 = 0.62, p = 0.61, ANOVA) among the different food treatments. Hatching success was only significantly different between the high food treatment and starved treatment (figure 5b; t = −2.3, d.f. = 12, p < 0.05, GLM, quasi-binomial distribution). Both samples of our stock cultures contained PSP toxins (as verified by ELISA and HPLC). As detected by HPLC, our stock culture contained 2.9 pg total saxitoxin equivalents per cell. The carbon : nitrogen ratios for R. lens and A. fundyense were 7.16 ± 0.44 and 5.04 ± 0.17 (mean ± s.e., n = 2), respectively.

Figure 5.

The number of eggs produced per female (a) and the per cent hatching success (b) of Temora longicornis while incubated in an Alexandrium fundyense treatment (containing 320 cells ml⁻¹ of A. fundyense + 1200 cells ml⁻¹ of Rhodomonas lens), a high food control (containing 5600 cells ml⁻¹ of R. lens), a low food control (control 1200 cells ml⁻¹ of R. lens) or filtered seawater (starved control). The total number of eggs per female was analysed using an analysis of variance, ANOVA (F_3,32 = 0.62, p = 0.61). Per cent hatching was analysed using a generalized linear model with a quasi-binomial distribution and logit-link function (t =−2.3, d.f.= 12, p < 0.05). Different lower case letters indicate differences at the 0.05 alpha level.

4. Discussion

To assess the effects of A. fundyense exposure on T. longicornis, the following factors were examined: copepod swimming behaviour, survival, ingestion rate, as well as egg production rate and hatching success. Of these factors, the only significant effect observed was on the copepod's swimming behaviour. After being exposed to A. fundyense for 2 h and then allowed to depurate for 15 h, T. longicornis increased their swimming velocity and the straightness of their swimming path. This apparent stimulatory effect of A. fundyense on copepod swimming behaviour was an unanticipated finding.

The A. fundyense strain used in this experiment contains saxitoxins that can block sodium channels and interfere with nerve function, leading to incapacitation, paralysis or death in vertebrates [7,46] as well as copepods grazers [8,9]. However, T. longicornis swimming significantly faster and straighter after ingesting A. fundyense does not concur with a sodium channel-blocking event. It is possible that the population of T. longicornis tested in our experiments were resistant to A. fundyense toxins owing to their co-occurrence in the Gulf of Maine. Populations of the copepod Acartia hudsonica from regions exposed to Alexandrium blooms are resistant to adverse effects [32] and can even exhibit enhanced physiological effects such as increased ingestion rate [25] and increased egg production rate [47] when fed A. fundyense versus non-toxic diets. Furthermore, the addition of saxitoxins to a palatable alga diet have even been shown to stimulate feeding in amphipod grazers [48]. Therefore, our results add to a growing body of evidence that suggests saxitoxins and A. fundyense can have stimulatory effects on grazers in some cases.

Following A. fundyense exposure, T. longicornis increased its swimming velocity by 25%. The degree to which this increase in speed affects the theoretical encounter rate with T. longicornis predators depends on the predator's prey detection mode. Predators that detect copepod prey through visual (fish larvae) or tactile cues (gelatinous zooplankton) will experience an increased encounter rate that is roughly proportional to the increase in copepod swimming velocity [20]. The larger hydrodynamic signal created by an accelerating copepod further enhances encounter rates by extending the detection distance of rheotactic predators such as mysid shrimp [20,49]. Therefore, a 25% increase in swimming speed leads to a 56% increase in the theoretical encounter rate between copepods and their rheotactic predators [18,19].

Copepods having increased encounter rates with predators after ingesting HAB species could have important consequences for bloom maintenance and the fate of toxins in food webs. If copepods that have recently ingested A. fundyense are selectively removed due to increased predatory encounters, this could reduce grazing control of HABs. Whether this increase in encounter rates enhances toxin uptake by copepod predators depends on both the propensity of T. longicornis to accumulate PSP toxins and the susceptibility of copepod predators to those toxins.

Field surveys indicate that PSP toxins are accumulated by zooplankton in the Gulf of Maine [50–52] and can exceed acceptable levels of PSP toxins in shellfish for human consumption [53]. T. longicornis is often present in these multispecies field samples [50,53,54], but, to the best of our knowledge, no one has ever measured the PSP toxin levels accumulated by T. longicornis. Laboratory estimates reveal that although copepods can accumulate significant levels of PSP toxins, they generally accumulate only a small percentage of what is ingested [6,26,36] and therefore, are not as efficient in toxin uptake compared with other species such as shellfish. Nonetheless, detectable PSP toxin levels were accumulated in the tissues of Eurytemora herdmani and Acartia tonsa after only a few hours of grazing [26] and ingestion of toxin-laden copepods can have detrimental effects on predators [55].

Copepod predators exhibit different sensitivity levels to PSP toxins. According to encounter rate models, rheotactic predators such as mysid shrimp will experience the largest increase in encounter rates with copepods following A. fundyense ingestion [20]. Mysid shrimp can accumulate PSP toxins in their tissues [56] and suffer reduced survivorship and fecundity following direct exposure to Alexandrium tamarense [57]. However, it is unclear if indirect exposure through ingesting toxin-laden copepods would produce similar effects. Indirect effects of saxitoxins on fish larvae have been examined through an A. fundyense–copepod–fish larvae food chain. Results indicate that ingesting as few as 6–12 toxin-laden copepods is sufficient to cause mortality in fish larvae [55]. Therefore, fish populations would likely be detrimentally affected by increased encounters with toxin-laden copepods. Conversely, gelatinous predators could potentially benefit from these increased prey encounters since they possess sodium channels that are insensitive to saxitoxins [58,59]. Therefore, the indirect effects of saxitoxins on copepod predators is likely species-dependent and further studies are needed to determine the fate of these toxins in marine food webs.

It is important to note that encounter rate is not the sole determinant of predation rate. If ingesting A. fundyense alters the outcome of these encounters (i.e. by affecting escape behaviour), then this will likely modify the magnitude of these effects. Furthermore, if predatory encounters occur within an A. fundyense bloom, copepod predators could suffer direct effects of toxin exposure that may alter their foraging ability. Future studies of the direct and indirect effects of A. fundyense on predation rates and grazer removal are warranted.

In our study, we did not find any effect of A. fundyense on T. longicornis ingestion (figure 4). Similar to results for T. longicornis ingesting Alexandrium minutum and A. tamarense [60]. Our results indicate that T. longicornis do not discriminate against or become incapacitated by ingesting Alexandrium cells. Additionally, ingesting A. fundyense did not decrease copepod survivorship over a 4 day period. A. fundyense toxin concentration was much lower than in other studies [6,25]; therefore, copepod health may be affected by ingesting A. fundyense with higher toxin content. In our study, copepod mortality was 19–35% during the 4 days experiment, with the greatest mortality occurring in the starved treatment. There were no significant differences between any of the treatments. Because there were no differences detected between the starved and 100% R. lens treatments, we are limited in the conclusions we can make. We cannot conclude whether or not long-term exposure of A. fundyense could affect T. longicornis survivorship. However, we can conclude that the behavioural changes we observed in our experiments after 2 h exposure and 15 h depuration were not due to the T. longicornis being ‘on the verge of death’ because they survived well when exposed to A. fundyense continuously for 4 days.

Studies addressing the effects of HAB exposure on copepod kinematics are rare (but see [11]), and the results vary depending on the grazing species. Cohen et al. [11] found that ingestion of the red tide dinoflagellate Karenia brevis had negligible effects on swimming speed of Acartia tonsa but significantly reduced the swimming speed of Centropages typicus, despite the fact that K. brevis is a sodium channel activator [61]. It is noteworthy that zooplankton exposed to toxins and environmental pollutants often exhibit behavioural responses at odds with the ‘typical’ mode of action [62,63]. Furthermore, responses may depend on exposure time and dosage [11,12]. Future work addressing the behavioural effects of HAB species with contrasting modes of action administered under various exposure times and dosages would yield much needed insights into HAB–copepod interactions. This research highlights the need to address the behavioural consequences of ingesting harmful phytoplankton. Future work aimed at determining the outcome of elevated encounters with predators is needed to fully determine the impact harmful algal species can have on grazers.

Data accessibility

Data are publicly available at http://dx.doi.org/10.5061/dryad.1c2j6.

Authors' contributions

R.S.L.-R. conducted experiments, participated in data analysis, participated in the design of the study and drafted the manuscript; K.N. conducted experiments and participated in data analysis; A.A. conducted experiments and cared for experimental animals and phytoplankton cultures; J.Y. conceived of the study; participated in the design of the study; coordinated the study; and helped draft the manuscript. All authors gave final approval for publication.

Competing interests

We have no competing interests.

Funding

This work was supported by National Science Foundation grant no. OCE-0728238.