\received
DD MMMM YYYY \acceptedDD MMMM YYYYBehavioural trials
Following the final 12-week rearing period, all trout were anaesthetised with a solution of benzocaine, measured between the tip of the snout and the fork of the tail (i.e. ‘fork length’: 71 ± 8 mm, mean ± s.d.), and marked with a pattern of variously coloured visible implant elastomer tags (Northwest Marine Technology Inc., Anacortes, WA, USA). Sixty groups of three trout each were established. Within each triad, trout were size-matched, in which the largest was larger by < 5% of the fork length of the smallest, to minimise possible effects of size on dominance (Huntingford et al., 1990; Johnsson and Åkerman, 1998 ). Additionally, individuals were unfamiliar with one another, having never shared a rearing tank. Each triad consisted of a dyad from different replicates of a treatment group, plus an ‘observer’ from another replicate. Across the 60 triads, all combinations of replicates and treatment groups were represented with the same approximate frequency.
Behavioural tests were held as dyadic trials in successive stages between groups of three. Triad members were kept alone and separate from one another before and between behavioural trials in identical 30 L glass aquaria, adorned with an air stone and one plastic plant set in a corner, for 24 hours and fasted. For the initial behavioural trials (naïve trials ), a dyad was placed simultaneously into a replica of the fasting aquaria. This trial tank was open at the top and illuminated by a 26 W, 1750 lm ceiling lamp. The sides were visually blocked, except the front, to allow observation by the researcher, and one side, to allow observation by the adjacent tank inhabitant. In an identical aquarium adjacent, the third trout of the size-matched triad was given the opportunity to observe the first two. This aquarium was shaded to prevent the dyad from observing the observer, and it was visually blocked on all sides except that facing the trial tank (Fig. 2).