Familiarity Facilitates Social Learning of Foraging Behaviour in the Guppy

Previous studies have shown that guppies, Poecilia reticulata, can learn the route to a food source by shoaling with knowledgeable conspecifics, and prefer to shoal with experienced foragers and familiar fish. We tested the hypothesis that guppies would learn more effectively from (1) familiar than unfamiliar demonstrators and (2) well-trained than poorly trained demonstrators. Demonstrator fish were given experience in swimming a route to a food source and then introduced into shoals of untrained observer guppies; the spread of this foraging skill was recorded over 15 trials. The demonstrators were either familiar or unfamiliar to the observers and either well trained or poorly trained. Observers performed significantly better when the demonstrators were familiar. The training of the demonstrators made no overall difference to the performance of naïve observers. However, whilst observers in shoals exposed to well-trained demonstrators did better initially than those with poorly trained ones, the latter learned the route to the feeder faster. Our results suggest that familiarity may generate a form of directed social learning in guppy shoals, in which fish learn more effectively from familiar conspecifics. An analysis of who follows whom suggests that well-trained demonstrators can provide a ‘tip-off’ as to the location of the hole but poorly trained demonstrators were more likely to be followed. The results suggest that while observers are able to shoal with poorly trained demonstrators, well-trained demonstrators swim the maze route too quickly to be followed, but may attract attention to the maze route. Social learning refers to learning that is influenced by observation of, or interaction with, other individuals (Galef & Giraldeau 2001). It is thought to be adaptive because it allows individuals to acquire pertinent information by exploiting the experience and knowledge of conspecifics, without incurring risks that might be associated with asocial learning or having to devise novel behaviour. Social learning processes may result in behavioural innovations spreading rapidly within populations, as the cost of acquisition of a socially learned behavior for each individual is relatively low. Laboratory studies have established several ways that animals can learn socially, only some of which rely on complex cognitive abilities (Galef 1988; Whiten & Ham 1992; Heyes 1994). Most cases of social learning in animals appear to result from very simple processes (Galef 1988) such as ‘local enhancement’ (Thorpe 1956) and stimulus enhancement (Spence 1937), when the behaviour of one animal draws the attention of a second to a particular locality or stimulus in the environment (Spence 1937). Often what is transmitted between demonstrator and observer amounts to little more than a tip-off, but experimental studies have found that this can frequently lead to learning in the observer, and to the diffusion of a behaviour through a population (Galef 1988). Galef (1988) has suggested that local enhancement may result from a propensity on the part of naı ̈ve individuals to approach conspecifics, or alterations they have made in the environment. Shoaling in fish represents such a tendency, and may instigate the transmission of adaptive information between individuals (for example forage patch copying). Local enhancement may be responsible for the shoaling traditions of reef fish (Helfman & Schultz 1984; Warner 1988), as well as the findings of experimental studies of social learning in fish (Laland & Williams 1998). Several studies on social learning have been carried out with the guppy, Poecilia reticulata, principally because of the ease of maintaining aquarium populations, and the practical advantages of fish for population-level analyses where many populations are required. In the guppy, social transmission of information influences female mate choice (Dugatkin & Godin 1992, 1993), avoidance task learning (Sugita 1980) and foraging behavior (Laland & Williams 1997, 1998; Lachlan et al. 1998). In the wild, guppies typically forage in small, loosely organized shoals, on a variety of food sources, such as benthic algae, aquatic insect larvae and diatoms (Dussault & Kramer 1981; Magurran et al. 1995). Shoaling appears to be a response to predation pressure (Seghers 1974), but may have other functions (Pitcher & Parrish 1993; Magurran et al. 1995), including the transmission of foraging information between fish (Laland & Williams 1997). Guppy shoals are not random aggregations, and numerous factors influence the choice of shoaling partners, including shoal size (Lindström & Ranta 1993; Lachlan et al 1998), fish size (Lachlan et al. 1998), sex (Lindström & Ranta 1993), demonstrator experience (Lachlan et al. 1998) and familiarity (Magurran et al. 1994). Lachlan et al. (1998) suggested that, given the importance of shoaling to social learning in guppies, factors affecting shoaling preferences might also affect the transmission of novel behaviour within shoals. There is evidence that familiar fish form more cohesive shoals (Chivers et al. 1995), and while this has antipredator benefits, Lachlan et al. (1998) suggested that familiarity might also provide greater opportunities for social learning. If this is the case, familiarity could affect the transmission of foraging information in shoals, and social learning may play a role in guppies’ shoaling preferences for familiar conspecifics. A key assumption behind theoretical models of social learning is that social interactions occur at random within the population. Recently however, Coussi-Korbel & Fragaszy (1995) discussed the possibility of ‘directed social learning’, in which transmitted information is restricted to, or directed through, a subset of individuals as a result of various processes influencing social interaction. Reader & Laland (1999) implicated directed social learning in a study of social foraging in the guppy, which found that sex, age and hunger level strongly affected the foraging performance of fish. Therefore if the knowledgeable members of a group are not of uniform foraging ability, naïve individuals may learn at different rates from differentially skilled individuals. Lachlan et al. (1998) suggested that if certain guppies within a shoal were knowledgeable about local foraging opportunities, naïve fish that chose to shoal with them would increase their own foraging success through social learning about food sources. This might imply that individuals are likely to learn more effectively from skilled demonstrators. However, several studies in birds have produced the counterintuitive opposite result, with nonproficient demonstrators being better tutors than proficient ones. Biederman & Vanayan (1988) showed that pigeons, Columba livia, learned discrimination tasks better from nonproficient tutors and Beauchamp & Kacelnik (1991) found that in zebra finches, Taeniopygia guttata, observers learned faster if their demonstrator was less knowledgeable. We investigated whether the learning of a foraging task in shoals of guppies is affected by the demonstrator’s (1) level of performance and (2) familiarity to the observers. We adopted a diffusion study design, establishing shoals composed of both skilled demonstrator and naïve observer fish. In half of the shoals the demonstrators were proficient, and in the other half they were less proficient. In addition, the shoals of naïve and trained fish were composed of either familiar or unfamiliar fish.