Rapid changes in social behaviour in medaka (Oryzias latipes) caused by selection for competitive and non-competitive growth.

Behaviour is a major component of growth rate variation in fish, but the genetic relationship between behaviour and growth is largely unknown. In particular, it is not known how behaviour responds to natural or artificial selection on growth (as during domestication of aquaculture stocks). I selected medaka (Oryzias latipes) for high and low growth rate, in two environments that differed in the intensity of interactions. In the high interaction environment food was provided to excess inside a floating cork ring, which limited access to the food, and allowed fish to attempt to monopolize the resource. In the low interaction environment, an equivalent amount of food was distributed over the container's surface. In each regime fish were selected within family for fast and for slow growth rate during two generations (G$\sb1$ and G$\sb2)$. Four selected lines were established after the first generation of selection: fast and slow growth rate lines in the high interaction environment, and fast and slow growth rate lines in the low interaction environment. In the third generation, half the number of broods in each line was raised in a high interaction, and half in a low interaction environment, bringing the number of sublines to eight. The response was measured in the third generation (G$\sb3)$ on the directly selected trait (growth) and on several correlated behavioural traits: (a) agonistic behaviour; (b) tolerance and locomotor activity; (c) schooling behaviour. The results were as follows: (1) Growth rate was modified as a consequence of selection, and the response was more pronounced in the high interaction than in the low interaction environment. (2) Selection for rapid growth in a high interaction environment, with food provided ad lib., induced a decrease in the intensity of agonistic behaviour. These results show how agonistic behaviour changes during domestication selection, and confirm the predictions of Doyle and Talbot's (1986) game theoretic model. In particular, growth rate selection in medaka favours fish that are indifferent to the presence of conspecifics, and not those that are most aggressive. (3) The genetic correlation between growth and social tolerance, and between growth and activity were rapidly modified as a result of selection for fast and slow growth in a high interaction environment. These results provide experimental support for the concern about the robustness of evolutionary inferences derived under the constant covariance assumption (e.g. Turelli 1988), and suggest at least some of the variability of these behavioural traits may be controlled by a few genes of major effect. (4) The school cohesiveness in the presence of a predator was also modified as a result of selection, indicating that schooling behaviour in medaka is influenced by genetic factors related to growth.