untitled

the arctic island of Svalbard prefer patches of tall vegetation, even though it is nutri-

tionally inferior (Van der Wal et al. 2000), for reasons that are, as yet, unclear.

The marginal value theorem predicts that foragers should depart when the rate of

food intake in a patch (i.e. the functional response) equals the average rate of food

available elsewhere in the environment multiplied by a constant (proportional to the

travel time between patches). This implies that foragers should concentrate in areas

with above average prey abundance, ignoring areas with lower levels of prey abun-

dance. Incorporation of this behavior into models of predators and prey in patches

has a stabilizing influence on metapopulation dynamics (Chapter 7). Such behavior

reduces the degree of variability in abundance over time of both predators and prey

when averaged over all patches (Fryxell and Lundberg 1993; Krivan 1997). Average

abundance in a collection of patches tends to be stable when abundance in a single

patch at any given time tends to be independent of that in other patches (de Roos

et al. 1991; McCauley et al. 1993). This is more likely when predators abandon patches

with low prey abundance than when movement in and out of patches is unrelated

to resource abundance.

Many foragers are themselves at risk of being attacked by predators. Frequently

such risk is highest when the forager is actively searching for food, rather than safely

hidden away in a den or resting site. Incorporating predation risk is rarely straight-

forward in analyzing habitat use, yet we know from numerous empirical studies that

it is important (Lima and Dill 1990). For example, risk-sensitive habitat use by the

larvae of the aquatic insect Notonectahas been elegantly demonstrated in the labo-

ratory (Sih 1980). Large Notonectaindividuals often cannibalize smaller Notonecta

individuals. Sih set up an experimental arena where individual Notonectalarvae could

choose to feed in food-rich or food-poor patches. The larger Notonectaindividuals

selected food-rich patches, whereas smaller, more vulnerable, individuals foraged in

the poor patches. This seems to be a logical way to reduce the risk of predation, at

the cost of reduced food intake.

One of the most elegant examples of the complex effects of risk-sensitive foraging

is the series of experiments conducted by Schmitz and co-workers in small caged

populations of carnivorous spiders, herbivorous grasshoppers, and grasses and herbs

(Schmitz et al. 1997). The grasshoppers suffer high rates of mortality from spiders

under normal conditions. As a consequence, they tend to spend their time foraging

on herbs, which are less nutritious than grasses, but offer better cover from pre-

dators. At the spatial scale of a grasshopper, a single plant is a patch, so dietary

preferences are in fact habitat preferences. By gluing shut the mouthparts of spiders,

researchers were able to assess the demographic impact of perceived risk of pre-

dation versus real predation. Results showed that grasshoppers subject to perceived

risk of predation (but not actual predation) suffered mortality levels similar to those

of grasshoppers subject to real predation. Both treatment groups suffered consider-

ably higher mortality than grasshoppers in cages without predators, which quickly

learned to forage on the more nutritious grasses rather than the safer, but less nutri-

tious, herbs.

Bluegill sunfish (Lepomis macrochirus) have been shown to balance the risk of pre-

dation against foraging benefits in choosing habitats. Nearshore habitats offer dense

protective cover, but relatively poor feeding. Open water offers better feeding, but

more exposure. When predators are present, young bluegills tended to concentrate

THE ECOLOGY OF BEHAVIOR 69

5.4 Risk-sensitive habitat use

WECC05 18/08/2005 14:42 Page 69