the diet so long as their profitability exceeds the expected rate of long-term intake
obtained by specializing on all of the more profitable prey types.
Assuming that maximal acquisition of energy can improve the fitness of a forager
(improve reproduction or survival), we might expect something like the optimal
strategy to be favored by natural selection. Such selection does not necessarily mean
that we would expect every animal to act like a little computer, perfectly assessing
the implications of each behavioral decision it might make. Rather, a pattern of beha-
vior that approximates the optimal strategy should be more successful at producing
offspring than alternative patterns of behavior.
The optimal diet model makes a number of predictions that are testable by obser-
vation or, better still, experimental manipulation:
1 Foragers should rank food types in terms of their energetic profitability (energy
content divided by handling time).
2 Foragers should always include the most profitable prey, then expand their diet
to include less profitable prey when the expected rate of gain by specializing on more
profitable prey matches the profitability of poorer prey.
3 The decision to specialize or generalize should depend on the abundance of highly
profitable prey, but not on the abundance of less profitable prey.
4 An optimal forager should have an all-or-nothing response. By this we mean that
the perfect forager would either always accept alternative prey or never accept them,
depending on whether f(N 1 )>e 2 /h 2.
The optimal diet model has been tested in a variety of settings since it was first pro-
posed (MacArthur and Pianka 1966; Schoener 1971; Pulliam 1974; Charnov 1976a).
Despite its simplicity, the optimal diet model has proved remarkably successful in
predicting foraging behavior (Stephens and Krebs 1986). Sih and Christensen (2001)
reviewed the outcome of over 130 diet choice studies. They found that two out of
three of the species studied, ranging from invertebrate herbivores to mammalian
carnivores, showed foraging patterns qualitatively consistent with the optimal diet
choice model. Optimal diet models tend to perform particularly well in situations
where all of the relevant parameters have been accurately measured, enabling pre-
cise quantitative predictions.
A classic example is captive great tits (Parus major), trained to pick mealworms
off a conveyer belt as it passes in front of them (Krebs et al. 1977). The birds became
adept at choosing whether to specialize on one prey or to accept both prey indis-
criminately, in accordance with predictions (1), (2), and (3) of the model (Fig. 5.2).
However, the birds never mastered the all-or-nothing behavior that would be per-
fectly optimal. Instead, the foragers sometimes ate both prey types, and sometimes
only the more profitable prey, a pattern termed partial preference. Such partial
preferences are almost always observed, even in the most successful experiments, per-
haps because foragers cannot discriminate perfectly amongst prey, or because for-
agers need continually to “test” alternative prey to assess their relative profitability.
Initial doubts that optimal diet choice theory could be applied to complex field
situations with many prey species (Schluter 1981; Pierce and Ollason 1987) have been
allayed by successful field studies (Sih and Christensen 2001), but the theory does
require much effort in estimating many parameters. The optimal diet model has been
less successful with predators which utilize mobile prey (e.g. weasels feeding on rodents)
compared with those which utilize stationary prey (e.g. starfish feeding on mussels).
There are a variety of reasons for this difference (Sih and Christensen 2001). In nature,62 Chapter 5WECC05 18/08/2005 14:42 Page 62