Observations were made on three fish species (banded killifish (Fundulus di
aphanus), golden shiner (Notemigonus crysoleucas) and white sucker (Catosto
mus commersoni)) in a temperate lake (New Brunswick, Canada) in order to in
vestigate the relationship between shoal choice behaviour of individual fis
hes and shoal composition. Encounters between shoals were observed to take
place every 1.1 min per shoal and an encounter lasted 3.7 s on average. The
duration of shoal encounters was influenced by shoal size but not by diffe
rences between shoals in either body length or species. Conversely the outc
ome of shoal encounters (i.e. whether or not an individual changes shoal) w
as influenced by body length and species differences but not by shoal size.
Together, these results suggest that encounter duration itself is unlikely
to have an important influence on encounter outcome. The collection of ten
entire fish shoals showed that they were assorted by species and body leng
th. A simulation model demonstrated that individual shoal choice behaviour
alone could account for the generation and maintenance of the observed leve
ls of size assortedness of shoals without invoking the existence of other s
orting mechanisms such as differential swimming speeds. However, the genera
tion of species assortedness was not predicted by the model. Furthermore, o
ur data suggest that fish density acts as a constraint on shoal choice, inf
luencing both shoal size and composition. This work has implications for st
udies on information transfer and reciprocal altruism within populations.