Food detection and preferences of the nudibranch mollusc Hermissenda crassicornis: experiments in a Y-maze.

Hermissenda crassicornis (Eschscholtz, 183 I), a nudibranch mollusc, is being reared in our laboratory as a biomedical research model for learning and memory studies. The aim of the present study was to determine whether Hermissenda, a generalist carnivore (l), is able to detect its prey chemotactically, and whether it displays a preference for certain foods. Previous studies on other nudibranchs demonstrated their ability to detect prey from a remote distance (2, 3). Experiments were carried out in a 35 cm wide X 70 cm long X 12 cm deep Y-maze with a total equilibrated flow of 15002500 ml/min of ambient seawater (water depth: 3 cm each arm: 17 cm wide). A uniform, unidirectional flow was achieved with an upstream collimator made of soda straws. Dye tests confirmed the uniformity of the currents (i.e., stimulus) across the full width of both arms of the maze. The hydroid Tubularia crocea, the tunicate Ciona intestinalis, and the mussel Mytilus edulis were the foods tested. These species have all been used successfully in the past to feed Hermissenda in the laboratory (1, 4); Ciona was presented with the tunic removed, and mussels were presented without shells. Food was placed on one side of the Y-maze, with the other side remaining empty. For Y-maze experiments, Tubularia was presented whole; Ciona in three ways: whole, viscera only, and viscera cut into pieces; mussels were whole but without shells. In trials testing for a preference between stalks and polyps of Tubularia, the pieces of the hydroid were contained in 500~pm mesh bags, one on each side of the Y-maze. Healthy Hermissenda, in groups of 3-6, were starved overnight before being tested. Individuals were selected at random from the starved group for each trial ( l-3 non-consecutive trials/specimen). The nudibranch was placed on the centerline of the Ymaze, where it received stimuli from both sides (distance from the food barrier: 40 cm; 1215 cm from the divider). The trial ended when Hermissenda was within 3 cm of the food barrier or after 30 min. The Y-maze was thoroughly cleaned between trials to remove the mucous trail of the previous specimen, and the maze was reconditioned by flushing for 10 min with ambient seawater between trials. Food was switched to the opposite side of the maze after 3-5 trials, and 33-56 trials were done with each food item. At least 50 Hermissenda individuals were used. About the same number of trials were done with the stimulus on the left and on the right; no significant bias was detected (x2). Results (number of times the side containing food was chosen vs. number of times the empty side was chosen) were compared by Chi-squared analysis. Results of the Y-maze experiments are shown in Table I. In trials with Tubularia, when a choice was made, the nudibranchs went to the hydroid side in 72.5% of the trials (or 59% of the trials including no-choice data). When approaching Tubularia, a tendency for the pedal locomotion to accelerate in the last few centimeters was observed. The other foods gave much larger percentages of negative results (no choice made), whereas the results in which a choice was made were split about evenly between the presence and absence of food. When animals failed to make a choice, they crawled aimlessly or entered a quiescent state. For all foods tested, the average time for the nudibranchs to reach the finish line (when they made a choice) was between 6 and 16 min. Additional experiments with Tubularia were carried out at higher water temperatures; they also gave a large proportion (69%) of no-choice results. When stalks and polyps