METHODS, EXCERPTED FROM ROBINSON ET AL (2019):

Permits: All protocols and experiments, described below, followed institutional guidelines and were approved by the University of Hawaii Institutional Animal Care & Use Committee (IACUC protocol number 2099).

Larval fish rearing protocol:
All experiments were performed on A. ocellaris fish larvae from two consecutive broods from one breeding pair, hatched in June and July, 2015. Late-stage eggs of A. ocellaris attached to the inside of a clay flower pot were obtained from a local fish breeder (K. Brittain) the afternoon prior to hatching and transported in an aerated container to the laboratory. Fish larvae hatched the same evening, within two hours of the transfer to the laboratory. The clay pot and any unhatched eggs were removed the next morning within 12 hours of hatching. Up to 200 fish were transferred to a rectangular aquarium (51 x 28 x 33 cm, length x width x height) containing 30 L of seawater, which was gently aerated and on a 12:12 h light:dark cycle from two 20W fluorescent bulbs located above the tank. Water temperature was kept between 24–26° C. Daily maintenance included siphoning detritus and dead fish from the bottom of the tank each morning and exchanging 10% of the water in the tank with UV sterilized seawater. Ammonia and nitrates/nitrites were monitored three times per week using a saltwater test kit (Mardel) and remained at 0–5 ppm throughout the experiments.

Cultivation of copepods and rotifers:
All larval fish were fed live prey, which was added to the rearing tank twice daily. Larval fish were fed two prey species: rotifers (Brachionus plicatilis), and mixed developmental stages of a calanoid copepod, Parvocalanus crassirostris. A different copepod species was used for daily feeding (P. crassirostris) than for the experiments (Bestiolina similis) to provide novel prey during experiments. Both calanoid species were originally isolated from Kaneohe Bay, Hawaii, Oahu, and have been in continuous culture since 2008 (VanderLugt & Lenz 2008, VanderLugt et al. 2009, Jackson 2011). Rotifers (B. plicatilis) were obtained from stock cultures maintained by K. Brittain (Kaneohe, HI). Copepods and rotifers were cultured in 21- L containers under similar temperature, salinity, and light:dark conditions as the fish. Copepod and rotifer cultures were fed every other day by adding 100 to 300 mL of mature cultures of Tisochrysis lutea (106–108 cells per mL) (formerly known as Isochrysis galbana Tahitian strain). Nitex™ sieves were used to isolate B. plicatilis (>20 μm), P. crassirostris nauplii (<80 μm), copepodites (80 μm < x <123 μm), and adults (>123 μm) from cultures to feed the larval fish as needed. The proportion of each prey species, and stage, provided to the fish was adjusted with fish age to accommodate their growing nutritional needs as follows: 300 prey per day at 1 day post-hatch (dph) and up to 1,000 prey per day at 14 dph. Fish consumption rates were monitored twice per day by counting the prey in a 25 mL water sample from the rearing tank. Prey were then added to maintain a target concentration, and were delivered to the rearing tank through an array of four feeding tubes that reduced agitation of the water.

To produce copepods of a specific developmental stage for behavioral experiments, cohorts of B. similis were raised from eggs. Approximately two thousand adults were sieved from the stock culture and transferred into 2 L of aerated, UV-sterilized seawater in a container with 40 mL of mature T. lutea stock culture. After 4 hours, the contents of the container were sieved (123 μm) to remove the adults, and eggs were allowed to hatch and cohorts were cultured as described above. These cohorts were harvested according to the copepod developmental stage needed for an experiment: nauplii at 24 hours (NIII–NIV); early copepodites at 120 hours (CII–CIII); and late copepodites after 165 hours (mostly CV). Adult copepods (CVI) were sieved from stock cultures using a 156 μm mesh to remove earlier developmental stages. Prior to the introduction of copepods to the observation chamber, harvested copepods were checked and photographed under the microscope to confirm developmental stages.

Behavioral observations:
Video set-up:
A system for tracking freely-swimming fish was used to obtain high-speed, high-resolution recordings (500 frames per second – fps; 1024 x 1024 pixels) of both larval fish (3–8 mm length) and individual copepods (100–500 μm length). The observation chamber consisted of an acrylic ring (20 cm inner diameter, 2.5 cm height) sealed onto a glass plate with aquarium cement. The glass plate was held in place by a metal frame supported by four posts fastened to an optical table. A high-speed camera (Photron FastCAM SA4) was mounted on a vertical optical rail to view organisms from above. The high-speed camera was fitted with a Nikon micro-NIKKOR 60 mm lens and a 36 mm extension tube, providing a field of view of 35 x 35 mm. A video monitor connected to the camera provided feedback to allow manual tracking of individual fish in the aquarium and focus adjustment. The experimental chamber was illuminated by a dark field, ring light (Fiber-Lite MI-150 high-intensity illuminator, Dolan-Jenner) positioned below the aquarium. Thus, the relative position of camera and light was fixed, and they moved as a single unit. This unit (vertical rail, camera, and light) was mounted on a manually-operated, linear positioning slide (Automation Gages Inc.) affixed to the optical table. The same aquarium background paper surrounding the rearing tank was used around the experimental chamber.

Experimental protocol:
Two types of predator-prey experiments were conducted during the planktonic phase of the fish larvae, which lasts for approximately 14 days with first feeding starting at 1 dph (Wittenrich & Turingan 2011). A preliminary feeding experiment was conducted to determine at what age fish attempted to capture adult B. similis copepods. A previous study by Jackson & Lenz (2016) using a similar copepod species (P. crassirostris) found that A. ocellaris did not capture adults in mixed copepod assemblages until 8 dph. To determine this milestone for B. similis, feeding experiments were conducted with fish between the ages of 1 and 14 dph. One hundred adult copepods were isolated under the microscope with a pipette and transferred into 100 mL of UV-sterilized seawater in a cylindrical, glass dish (6.5 cm inner diameter). Then two fish that had been removed from the rearing tank and left without food for 4–6 hours were transferred into the glass dish and allowed to forage for 1 hour. After removal of the fish, the remaining copepods were counted. All feeding experiments were conducted at 25 °C under 8 W m-2 illumination with experimental containers surrounded by background paper that was also used around the rearing aquarium.

During behavioral experiments, four different developmental stages of B. similis were used: NIII–IV (nauplii), CII–CIII (early copepodites), CV (late copepodites), and CVI (adults). Nauplii were tested with fish ages 1 to 9 dph, early copepodites with fish ages 1 to 12 dph, and adult copepods with fish ages 6 to 14 dph (Fig. 2). Late-stage copepodites were tested with fish larvae ages 3, 6 and 9 dph. No observations were made at 10 dph. On each day of the behavioral experiments, 12–16 fish were removed from the rearing tank before morning feeding, transferred into a 7- L clear plastic container with UV-sterilized seawater without food for 4–6 hours. For each trial, the experimental chamber was filled with 700 mL of UV-sterilized seawater (2.2 cm depth) and 0.2–0.7 copepods mL-1 of a single developmental stage class of B. similis: nauplii (ca. 500), early copepodites (ca. 250), late copepodites (ca. 250), or adult copepods (ca. 100). Subsequent mention of copepodites will refer to early- stage copepodites; late copepodites will be specified. Two of the isolated clownfish were placed into the chamber and allowed to acclimate and feed for 10 minutes. During the subsequent observation period, the camera was manually positioned to keep the tracked fish in the field of view and focused. When a fish-copepod interaction was observed, the high-speed camera was triggered and recorded video was saved onto a computer as a sequence of digital images. Both successful and unsuccessful fish strikes were recorded. An experimental trial was terminated after six interactions were recorded, or after one hour. Experimental trials were run between 1200 and 1900 h.