Sibling cannibalism among juvenile vundu under controlled conditions. I. Cannibalistic behaviour, prey selection and prey size selectivity

Sibling cannibalism among vundu Heterobranchus longifilis larvae started at the age of 4 days, with the prey caught tail-first then swallowed up to th e head, which was eventually discarded (type I cannibalism). At 17 days old , this type of cannibalism vanished and was replaced by the ingestion of th e whole prey (type II cannibalism), which could only be exerted by predator s six times as heavy as their prey. Type IT cannibalism consisted of a seem ingly opportunistic ambush attack by a formerly passive predator towards a disorientated prey. It required no preliminary aggression or chase, or even contact with the prey, suggesting that the attack was not mediated by the tactile sense, and that cannibalism was independent of aggressive behaviour . When alternative food resources (formulated feed, live tilapia prey) were available, the intensity of cannibalism decreased but pellet-eaters or til apia predators always achieved lower growth rates than those feeding on con specifics, suggesting that cannibalism was the most advantageous foraging t actic. Losses to cannibalism among populations of 30-day old juvenile vundu with an initial ratio of 4% of cannibals were as high as 75.5-79.9% over 1 5 days. Predation peaked during the first days (up to 28 prey C-1 day(-1)), then vanished progressively as surviving prey grew faster than cannibals a nd escaped their predation. Cannibals preferred consuming the largest prey available with respect to the logistics of cannibalism (body weight ratio o f 6.0). This preference for large prey was interpreted both as a foraging t actic aiming to maximize the energetic return, and as foraging strategy ena bling the cannibals to exploit their prey as long as possible. Based on the se data, comprehensive models of the impact of cannibalism on vundu populat ions were developed. (C) 1999 The Fisheries Society of the British Isles.