Fish only have a three second memory? Learning and information use in Gambusia holbrooki

Kieran MacKenzie (Supervisors: Associate Professor Ashley Ward and Associate Professor Ross Coleman)


Learning, defined as a change in behaviour with experience, has been shown to influence many facets of fish behaviour including foraging, breeding, predator recognition and attack abatement. Possessing the ability to learn can therefore be critical to the survival of an individual animal as negative fitness consequences can result from inability to adapt behaviour in a variable environment.

Broadly speaking, individual animals learn through two mechanisms, through personal sampling of the environment (asocial learning), or indirectly, through interpreting social cues produced intentionally or inadvertently by other individuals (social learning). Generally, information acquired through asocial learning is more accurate, but expensive to acquire, and information acquired through social learning is less reliable, but cheaper to acquire.


Fish behaviour varies considerably between species, populations and even between individuals. I therefore aimed to investigate whether individual variation also exists in the learning ability of shoaling fish, using the mosquitofish Gambusia holbrooki as a model species.

Animals can also learn from others, and evolutionary game theory and population genetic models predict that individuals should be selective with respect to who they should learn from due to the potential risk of acquiring maladaptive information. I therefore aimed to examine whether there was a difference in the propensity for individuals in a group to learn socially from male demonstrators and female demonstrators, using the mosquitofish Gambusia holbrooki as a model species.


As key ecological features will seldom be completely predictable, due to changing weather conditions, the behaviour of other organisms and other factors that are out of the direct control of the animal, animals require the ability to reduce uncertainty in a changing environment by gathering information through learning. So in order to make informed decisions, such as where to forage and with whom to mate, animals require accurate information in order to maximise their fitness benefits. Therefore, studying how animals acquire information and the strategies they employ to maximise accuracy is fundamentally important.


The efficiency of the display of a learnt response from an associative learning task for a shoal of mosquitofish was dependent upon shoal size and shoal composition, where larger shoals and juveniles display a learnt response in fewer associative learning trials than smaller shoals and adults, respectively. However, sex did not significantly influence the rate of learning. In addition, mosquitofish showed evidence for social learning where they displayed a propensity to use the social cues provided by female demonstrators, but not from male demonstrators. This provides support for the opinion that the transmission of information through a population of mosquitofish is not random but is in fact biased by a preference for female demonstrators over male demonstrators. These findings have implications for the diffusion of novel information, where information may be transmitted though certain subsections of groups or populations at different rates as a result of individual differences in propensities to learn individually and learn from others.