Fish don't read textbooks: juvenile salmon prove ignorant of foraging theory

The rules governing the selection of feeding patches by foraging animals is an area of intense interest. Much work has focussed on the development of theoretical models that predict when individuals should switch patches. Tests of these models have often been conducted in laboratory environments, but it is not clear how much influence patch‐switching decisions have on population‐level parameters such as growth and distribution in more complex natural environments. We used juvenile Atlantic salmon as a model species to investigate the effects of randomly fluctuating food levels on growth and site selection. We used PIT technology to monitor in detail individuals' patterns of patch use and activity in an artificial stream, at natural densities. This allowed us experimental control of food supplies and sufficient replication, while retaining many features of a natural system. Only a few individuals of high social rank switched patches as predicted by an appropriate foraging model; otherwise, although frequent, patch‐switching was not related to food availability. Thus, while laboratory experiments indicate that this species has the potential to choose foraging sites on the basis of food availability, it is unlikely that this behavioural mechanism is of great importance in natural systems; further tests of foraging models under natural conditions are essential if we are to understand their effects at the level of populations.