Host species:
Host fishes selected for testing were French grunt (Haemulon flavolineatum: Haemulidae), lane snapper (Lutjanus synagris: Lutjanidae), longspine squirrelfish (Holocentrus rufus: Holocentridae), and longfin damselfish (Stegastes diencaeus: Pomacentridae). French grunt and lane snapper are carnivorous, nocturnal fish that forage in seagrass and sandy areas at night and school on reefs during the day. Both species are highly and similarly susceptible to gnathiids. French grunts are locally abundant and are known to be able to attract gnathiids with only chemical cues and therefore were used as a standard for the experiments (see below). Longspine squirrelfish are nocturnal invertivores that forage at night and shelter in or near refuges on reefs during the day. Finally, longfin damselfish are diurnal herbivores that are highly territorial. Longspine squirrelfish and longfin damselfish represent moderate and low susceptibility to gnathiids, respectively.

Field collection:
Juvenile Gnathia marleyi and adult host fishes were collected from John Brewers Bay, St. Thomas, United States Virgin Islands (18°20’ N, 64°58’ W) from May to August 2019 and January to March 2020. Gnathiids were collected using zooplankton light traps similar to those described by Artim and Sikkel (2016). They were held at the University of the Virgin Islands’ MacLean Marine Science Center in small colonies in 150-milliliter (ml) containers filled with filtered seawater. All host fishes were caught by free divers using cast nets and were held in a 2800-Liter tank with continuously flowing seawater, at 28 degrees C. Fish were kept for no longer than 24 hours and all were returned to the original capture site.

Behavioral assays:
To assess the ability of G. marleyi to detect and preferentially differentiate between hosts of different susceptibility an Atema choice flume was used. The Atema flume allows two separate sources of water to flow through a choice arena without mixing. Within the choice arena, an organism is able to move freely between the two sources. Water was fed at 100 milliliters per minute into the choice flume. Preliminary trials indicated that this flow rate was high enough to provide a laminar flow within the flume but low enough to allow a gnathiid to swim freely without being impeded by the water flow. A single gnathiid was placed in the center of a ‘choice arena’, allowing it to move freely throughout the arena and/or swim towards a preferred water source. Immediately after placing in the flume, a two-minute habituation period began followed by a three-minute testing period. During the testing period, the position (left or right side of the flume) was recorded using a Canon EOS Rebel T5i digital camera under near-infrared lighting. Due to recording limitations, only the larger, third-stage unfed, juvenile gnathiid isopods were used in behavioral assays. Preliminary experiments showed that gnathiid isopods did not react to and exhibited typical behavior patterns under infrared or near-infrared lighting. All trials were conducted between dusk and dawn to assure they were done during peak gnathiid activity. After the testing period, the gnathiid was removed for one minute while the flume was flushed, and the sources were exchanged. The habituation and testing periods were then repeated with the same individual to ensure a side preference was not being displayed.

To determine behavioral preferences toward specific chemical cues, filtered seawater (200-μm) was conditioned using host fish. Before conditioning, all fish were inspected to ensure that there were no parasites currently infesting the fish. To condition seawater with host chemical cues, a single host fish (French grunt: n = 8, size range: 51–128 millimeters (mm); lane snapper: n = 5, size range: 80–120 mm; longspine squirrelfish: n = 8, size range: 80–130 mm; longfin damselfish: n = 12; size range: 56–92 mm) was held in an aerated closed 15-liter aquarium for 60 minutes. When comparing the chemical cues of two fish species simultaneously, individuals of a similar size were paired together. Before assays, water types were randomly assigned to either side of the flume.

To confirm the ability of gnathiid isopods to detect host chemical cues, gnathiids (n = 15) were offered the choice between water conditioned with the chemicals of French grunt against untreated filtered seawater. To determine if gnathiids have a preference for hosts of higher susceptibility, gnathiids were offered a choice between French grunt and another fish. Trials conducted included: 1) French grunt versus lane snapper (high vs high susceptibility); 2) French grunt versus longspine squirrelfish (high vs medium susceptibility); and 3) French grunt versus longfin damselfish (high vs low susceptibility). In total, 45 (15 each) behavioral assays were conducted between the three comparison species. Each assay consisted of two 180-second trials with 360 seconds worth of behavioral recordings for each gnathiid.