Summary
Using plausible assumptions, we propose "transitive odds" and "intrinsic preferences" models of food preference transitivity to account for observed behaviour when animals are feeding on items of pairs of food types from a set of such types for which preferences differ. The latter model also accounts for feeding when items of a single food type are available. The two models describe well novel data collected from sunfish feeding on three food types presented alone or in pairs, as well as feeding data in the literature from largemouth bass. The estimated parameters reflect such features of feeding behaviour as the strengths of preferences for different food types and changes in these preferences over time
Methodology
Six pumpkinseed sunfish, from 13.5 cm to 14.5 cm in total length (year classes five and six) and seined from Lake Opinicon on the Rideau River system north of Kingston, were placed individually in living chambers in an experimental aquarium in a quiet, windowless room illuminated daily for 12 hours with fluorescent lights. Water temperature was controlled at 22.5 +_ 1C. The experimental aquarium consisted of a central feeding area (150 x 90 x 25 cm deep) with living chambers (28 x 22.5 cm) at each end. Preliminary studies determined appropriate values of various features of the experimental design, suggested that two different sizes of blowfly larvae and pieces of commercial redworms would serve as three suitable food types, and familiarized the fish with these food types.
The experimental design (Table 1) consisted of three repetitions (blocks) of a six-treatment Latin square design balanced for residual effects (i.e., effects carrying over from feeding on the preceding day; see Cochran and Cox, 1957, p. 133ff). Over 6 days, in each block, each food type alone, and each of the three pairwise combinations of the food types, were presented to each of the six fish. Days 1, 7, and 13 allowed for complete balancing of residual effects for the three blocks. The six treatments were: A, large larvae alone; B, small larvae alone; C, worms alone; D, small larvae and worms; E, large larvae and worms; and F, large and small larvae.
At the beginning of the daily feeding bout for each fish, at a fixed time of day, either forty items of the single food type or twenty items of each of the two food types to be presented were distributed irregularly over the bottom of the central feeding area. The fish was then released from its living chamber and permitted to feed for between 50 and 90 seconds, the exact time for each fish, constant over the experiment, being adjusted so that all fish ate approximately the same amount. An observer, equipped with counters and sitting behind a blind, recorded the number of items of each food type eaten. The data, ignoring day 1, are given in Table 2.