On the Term 'Search Image'

Abstract

The term 'search image' continues to be used uncritically by many evolutionary and behavioural ecologists. In the sense of Tinbergen (1960) and Dawkins (1971a) a search image is 'a perceptual change in the ability of a predator to detect familiar cryptic prey' . As a result of initial chance encounters with the cryptic prey, the predator 'learns to see', and selectively attends to those cues that enable it to distinguish the prey from the background (Croze 1970, Dawkins 1971a, b, Pietrewicz and Kamil 1979). The adaptive significance of search image formation might be that cryptic prey can be found more effectively when the predator concentrate on one prey type alone. It is assumed (although direct evidence is scarce) that a search image will be adopted only when the encounter rate is high and will be discarded when the encounter rate is either low or very high (Tinbergen 1960, Dawkins 1971a, b, Krebs 1973). Krebs (1973:83) pointed out that 'search image formation results in a preference, and the problem is to distinguish it from all the other reasons for having a preference'. He listed six types of learning that he and Dawkins (1971 a) emphasise must be excluded from any given case before evidence for search image can be accepted. These alternatives are: (a) learning to visit a particular place to find food, (b) learning to look in a particular type of place for food, (c) alteration of the search path to increase the chances of encountering prey, (d) learning to handle prey more effectively, (e) preference or avoidance of a prey over others that is independent of the predator's ability to see the different types and (f) learning of specialized hunting techniques by particular individuals. It would not be difficult to discover whether a predator under particular circumstances is increasing feeding efficiency by using a search image or one of its six alternatives (or any combination of the seven). We feel that it is important to identify the mechanisms) involved, because each can lead to a different expectation not only of what a predator does, but also of what a prey 'does' to avoid being eaten. For example, if the predator is 'learning to see' the evolutionary response of the prey might be quite different from the case where the main effect is of the predator learning 'where to hunt'. Details of the behavioural mechanisms involved can therefore be important even to those interested only in the effects of predation. Some authors still seem unaware of the distinction between search image and the other forms of learning that increase the efficiency of feeding on prey (or if they are aware, they presumably think the distinction trivial). Ricklefs (1980:878), for example, defines a search image as 'a behavioural selection mechanism that enables predators to increase searching efficiency for prey which are abundant and worth capturing'. Quite apart from the lack of reference to cryptic prey in this definition (a common failing), all six alternatives can also be regarded as 'behavioural selection mechanisms'. On page 613 of the same book, Ricklefs describes a search image as 'a preconception of what a given prey looks like and where it is found'; a similar definition is given by Owen (1976:123). In fact 'a preconception of what a given prey looks like' can result from category (e) above. This alternative explanation is frequently overlooked by those who claim that predators use 'search images' when they feed disproportionately more on prey that are common or familiar. Recent examples of authors who appear to have fallen into this trap include Cornell (1976), Jirvinen (1976), Hassell (1978), Rausher (1978), Hughes (1979), Atema et al. (1980) and Andr6n and Nilson (1981). These workers give no clear indication that the prey concerned were cryptic and therefore it is likely that the prey would have been seen immediately by their predators. Common conspicuous prey may be preferred simply because they are common. 'A preconception of where a given prey is found', the remaining part of Ricklefs second definition, is covered by alternatives (a) and (b) above. Pietrewicz and Kamil (1981:313) extend the true meaning of search image to include changes in the ability of predators to detect novel cryptic(?) prey or, in their own words, learning 'what the prey looks like'. These definitions fail to preclude a mechanism that belongs to category (e). To quote from Dawkins (1971a:567): 'we cannot conclude that a change of per-