Refer to the Methods section of Senft-Batoh, et al. (2015).

Phytoplankton Culturing:
The dinoflagellates, Alexandrium fundyense (toxic strain BF-5, isolated from Bay of Fundy, Canada) and Alexandrium tamarense (nontoxic, isolated from Mumford Cove, Groton, Connecticut), were grown in semi-continuous culture in F/2 medium without silicate. Cultures were maintained in, and all experiments were conducted in, an environmental chamber with fluorescent lighting set to a 12 h:12 h light:dark photoperiod (100 μmol m⁻² s⁻¹ photosynthetically active radiation) and 18°C. The Alexandrium fundyense strain was used in grazer-enhanced toxin production, feeding-selection, and algal alarm-cue assays, while Alexandrium tamarense was used in the feeding-selection and alarm-cue assays. Aside from production of paralytic shellfish toxin, A. tamarense and A. fundyense are nearly identical in shape, size, and carbon and nitrogen content per cell.

Copepod Collection and culturing:
The calanoid copepod Acartia hudsonica was collected from Casco Bay, Maine (43°39′N, 74°47′W; historically co-occurring with toxic Alexandrium, and Little Egg Harbor, Tuckerton, New Jersey (39°63′N, 74°33′W). Triplicate cultures for each copepod population were maintained with a mixed phytoplankton food medium replenished thrice weekly. Copepods were cultured for at least three generations (~3 months) prior to experiments to eliminate maternal and environmental effects. Animals (eggs to adults) in cultures were transferred monthly to new containers. Prior to assays, adult, female copepods were acclimatized to experimental conditions for 24 h and starved during that period to ensure complete gut evacuation.

Direct and indirect induction of toxin production by copepods:
Direct and indirect mechanisms of toxin induction were tested simultaneously using experimental cages. 1 L polycarbonate beakers with bottoms made of 10 µm mesh were nested within another 1 L beaker containing 500 mL of toxic Alexandrium fundyense (300 cells mL⁻¹). The mesh isolated these cells from materials within the cage. Adult female Acartia hudsonica (15 individuals) from either Maine or New Jersey were added to each cage and offered a diet of toxic A. fundyense (300 cells mL⁻¹) or were starved (no addition of algal food). Triplicate treatments (n=3) of the combinations of copepods and algal food within the cages were: 1) Maine copepods fed toxic algae; 2) Maine copepods starved; 3) New Jersey copepods fed toxic algae; 4) New Jersey copepods starved. Control cages (n=3) contained 300 cells mL⁻¹ of toxic algae and no copepods. Assays were run for 72 h, long enough to ensure induction of toxin production, and incubation conditions were identical to those of the algal cultures. Cages were lifted every 12 hours to ensure exchange of cues between compartments. At the termination of the assay, cells of Alexandrium fundyense within cages (where applicable) and below cages were collected from treatments and controls for toxin analysis (see below). Cells within cages (treatments and controls) were enumerated microscopically, before and after incubation, to calculate copepod ingestion rates (Frost 1972). Differences in ingestion rate between the populations were assessed by a t-test.

Induction of toxin production by algal alarm cue:
To determine if alarm cues released by Alexandrium fundyense, and a congener (but non-toxic) species, Alexandrium tamarense, could induce toxin production in a culture of A. fundyense, extracts of sonicated, conspecific cells or cells of A. tamarense (equivalent to 50,000 cells; complete lysis confirmed microscopically), were added daily, over a 3 day period, to triplicate treatments (n=3 for A. fundyense and A. tamarense extracts) of 500 mL, nutrient-replete (F/2) cultures of A. fundyense (300 cells mL⁻¹). The daily re-inoculation of cue-receiving cultures with sonicated extracts ensured that potentially labile alarm cues were maintained throughout the duration of the experiment. Extracts were not added to control cultures (n=3).

Toxin analysis:
Cells from treatments and controls were collected on a 10 µm mesh and resuspended in filtered seawater in a 50 mL centrifuge tube. Replicate subsamples (1 mL) were taken from each tube, and cells were enumerated microscopically to determine the number of cells present in each extract (50,000-150,000 cells, depending on the assay). Cells were centrifuged at 4,000 x g for 20 minutes. The seawater supernatant was decanted, and the cell pellet was resuspended in 1 mL of 0.1 M acetic acid. Cells were lysed using a probe sonic dismembrator. Sonication was conducted with the tube immersed in ice to prevent heating of the samples. Samples were then centrifuged again at 4,000 x g for 20 minutes to remove cell debris. The acetic acid supernatant (extract) was filtered through a 0.45 µm ultracentrifuge filter cartridge to remove any remaining particles. Samples were stored at -80°C until analysis.

Concentrations of saxitoxin (STX), neosaxitoxin (NEO) and gonyautoxins 1 through 4 (GTX 1-4), as well as the sulfamate congeners, C1 and C2 were measured using high-performance liquid chromatography with fluorescence detection (Oshima 1995) after calibration with standards (Certified Reference Materials Program, NRC Institute for Marine Biosciences, Canada) and expressed as mass of STX equivalents per cell according to conversion factors of Oshima (1995).