Chemical alarm cues alert aquatic prey to the presence of an actively foraging predator. There is a large literature based upon responses to alarm cues derived from skin extract, because it is anticipated that prey skin is damaged when prey are attacked by a predator. However, many predators feed by suction feeding whereby prey are quickly drawn into the buccal cavity and swallowed whole with little, if any, direct contact between the teeth of the predator and the prey. Here, we test if predation by suction feeding releases chemical information in sufficient quantity to elicit an antipredator response in conspecific prey. In tests of individual zebrafish Danio rerio, we found that odor of crushed zebrafish produced a clear antipredator behavioral response, but water collected immediately adjacent to staged predation events between a largemouth bass Micropterus salmoides (122-145 mm TL) and adult zebrafish (39 mm TL) did not elicit alarm behavior, and did not differ from behavioral responses to blank water or bass odor (on a diet of earthworms). In a second experiment, zebrafish swallowed by largemouth bass, then retrieved seconds later through gastric lavage, produced zebrafish that were alive and completely intact with minimal epidermal damage. Published relationships between bass length, gape size and the geometry of suction feeding suggest that in a hypothetical population of largemouth bass feeding on adult zebrafish, or fathead minnows, the majority of predation events by piscivorous fish probably would not release detectable levels of prey alarm cue. Accounting for the role of feeding mechanics by fish predators requires a recalibration of the literature on risk assessment by small prey fishes. Chemically-mediated antipredator behaviors against suction-feeding predators may occur primarily via post-ingestion dietary cues, or disturbance cues released near the moment of attack.

Experimental design
There were four treatment groups: (1) alarm cue (positive control), (2) deionized blank water (negative control), (3) water collected from a holding tank containing bass kept on a diet of earthworms Lumbricus terrestris to control for the possibility of pre-existing recognition of bass odor and (4) water from where a bass had eaten an adult zebrafish 15 s beforehand that should contain detectable levels of alarm cue.

Cue preparation
To create alarm cue we euthanized 20 zebrafish (TL = 39.0 ± 0.89 mm) by an overdose of tricaine methanesulfonate (MS-222), added them to 210 ml of deionized water and homogenized them for 60 s using a blender, then filtered the solution through a loose wad of polyester fiber and aliquoted the filtrate into 21, 10-ml doses and froze them at -20° C until needed. The same number of doses of blank deionized water cue was aliquoted into 10-ml doses and frozen at -20° C until needed. 

Bass-earthworm-diet odor was collected from a 75-L aquarium filled with 32 L of dechlorinated water that held three largemouth bass (122, 127 and 145 mm TL) that had been maintained on a diet of earthworms. Fish were held for 24 h then tank water was collected and aliquoted into 60-ml doses in whirlpac® bags and frozen at -20° C until needed.

Bass-zebrafish alarm cue was collected by moving each bass to separate 37-L aquaria. Each bass was restricted to one half of the aquarium using a coroplast® barrier to limit diffusion and dilution of any alarm cue released during predation. Filtration was turned off during collection of alarm cue to cease all water flow in the aquarium. A live zebrafish from the same stock from which the alarm cue was prepared (ca. 39 mm TL) was dropped into the side of the tank containing the bass. The bass struck and ingested the zebrafish. We waited 15 s to allow time for alarm cue to diffuse from the bass, then we immersed a 1-L beaker with a gloved hand to collect water from the site of predation adjacent to the bass. We repeated this procedure for each of the three bass. The three 1-L water samples were combined together in a 20-L liter pail, from which 20, 60-ml doses were aliquoted into whirlpac® bags and frozen at -20° C until needed.

Experimental apparatus
Twelve 37-L tanks were arranged on a shelf in the laboratory with the short side facing the aisle. Tanks were filled with dechlorinated tap water. Each tank was fitted with an air-powered sponge filter and a shelter object (8 × 8 cm ceramic tile with a supporting leg on each corner approximately 6 cm tall). Black coroplast® dividers were placed between tanks to visually isolate test subjects. Airline tubing approximately 2 m in length was inserted into the lift tube of the sponge filter to allow for surreptitious injection and rapid dispersal of test cues. Gridlines approximately 5 x 5 cm were marked on the front pane of each tank for scoring activity (number of lines crossed) and vertical distribution (horizontal rows numbered 1 (bottom) through 5 (surface) (Wisenden 2011).

Experimental protocol
Individual zebrafish were arbitrarily selected from stock tanks and placed into each 37-L tank and allowed to acclimate for 24 h. Prior to the start of each trial, 60 ml of tank water was withdrawn through the injection tubing and discarded to rinse the tubing. A second 60 ml was withdrawn and retained to be used as a flush to fully dispense cue treatments into test aquaria. Trials were completed in groups of four representing one of each treatment type, with the order of the treatment type and the order of tanks being tested both randomized. A dose of each treatment type was thawed in a cup of warm deionized water and drawn into a 60 ml syringe. Overhead ceiling room lights were shut off leaving only the above-tank lights on to decrease the visibility of observers to test subjects. Each zebrafish was observed for 5 min before cue was introduced (pre-treatment observation period). We recorded (1) Activity, as the total number of grid lines crossed by the fish over each 5-min observation period, (2) Vertical Distribution as the horizontal row occupied by the fish every 10 s, and (3) Time in Shelter as time (s) spent under the shelter object or hidden behind the sponge filter. When the pre-stimulus observation period was over, one of the four test cues was gently injected into the aquarium followed by a flush of 60 ml of previously-retained tank water. Post-stimulus behavioral observation commenced immediately for another 5 min, recording activity, vertical distribution, and time in shelter as before. When each trial was finished, the zebrafish was moved to a separate 190 L holding tank so that each fish was tested only once. Test tanks were drained, rinsed, and refilled with dechlorinated water between subsequent trials. Injection tubes were replaced after each trial. We conducted 15 replicates of each treatment, for a total of 60 trials.