A linear regression model was developed to predict cannibalism in juve
nile snakehead Channa striatus. Based on morphological measurements of
mouth width, head width and body length, the model relates maximum pr
ey length (TL; mm) to predator TL (TL(prey) = 26.168 + 0.278 TL(predat
or)). This model was verified with a set of independent data, indicati
ng an underestimate of maximum prey size for a certain predator size.
The revised model should be: TL(prey) = 25.809 + 0.405 TL(predator) (m
m). Size differences substantially increased the rate of cannibalism,
but prey:predator predator TL ratios decreased with increased predator
TL. Cannibalism rate was 100% during a 5 day trial when the TL ratio
of smaller fish to larger fish was 0.35; but the cannibalism rate decr
eased to 43% when the TL ratio of smaller fish to larger ones increase
d to 0.64. Increased availability of formulated feed also reduced cann
ibalism. With no feed offered, cannibalism was 83%; but decreased to 4
3% when daily feeding rate was 15% of the larger fish's body weight. O
ur results indicate that cannibalism is unavoidable with this species,
but can be greatly reduced among juvenile fish by size grading and ad
libitum feeding.