4.3 CONSTRAINTS ON PREDATOR DESIGN AND BEHAVIORONLY ON THE TROPHIC...

20.4.3 Constraints on predator design and behavior

only on the trophic level immediately below it, especially as the

prey there are likely to be larger, fiercer and rarer than species

at lower levels. Overall, theoretical explorations (Hastings &

There may also be evolutionary constraints on the anatomy

or behavior of predators that limit the lengths of food chains. To

Conrad, 1979) suggest that an evolutionarily stable food chain

length (one that would be optimal for predator fitness) would be

feed on prey at a given trophic level, a predator has to be large

around two (three trophic levels). Such arguments, however, have

enough, maneuverable enough and fierce enough to effect a

rather little to offer by way of explanation for the variations in

capture. In general, predators are larger than their prey (not true,

food chain length.

though, of grazing insects and parasites), and body size tends

Thus, there are complete answers to neither of our original

to increase (and density to decrease) at successive trophic levels

questions (see p. 595). The constraints on predators are likely to

(Cohen et al., 2003). There may well be a limit above which

design constraints rule out another link in the food chain. It

set some general upper limit on the lengths of many food chains.

Food chains are likely to be atypically short in especially unpro-

may be impossible to design a predator that is both fast enough

ductive environments. Food chain length seems to increase with

to catch an eagle and big and fierce enough to kill it.

increases in productive space, but it is unclear whether this is an

Also, consider the arrival in a community of a new carnivore

association with the total energy available in an ecosystem or with

species. Would it do best to feed on the herbivores or the

ecosystem size alone – and if the latter, it is unclear precisely how

size comes to determine food chain length. The two longest estab-

40

Briand webs

lished hypotheses – energy per unit area and dynamic fragility

30

Glacial lakes

– have, if anything, the least support.

Finally, it is important to note that,

20

are the data simply

as with connectance, estimates of food

10

not good enough?

chain length are sensitive to the degree

of taxonomic resolution. This may be

0

6

2 4 8 12

why many of the more recently documented webs have longer

Average degree of omnivory

than average chain lengths ranging from five to seven (Hall

& Raffaelli, 1993). Moreover, if a well-resolved large web is

Figure 20.17 The prevalence of omnivory in glacial lakes in

progressively simplified by lumping taxa together (in a manner

northeast North America (Sprules & Bowerman, 1988) is much

analogous to earlier studies), the estimate of food chain length

greater than that observed in Briand’s set of food webs (see

declines (Martinez, 1993). There is clearly a need for rigorous

Figure 20.9a). The degree of omnivory in a web is quantified as

studies of many more food webs before acceptable generaliza-

the number of closed omnivorous links divided by the number of

tions can be reached.

top predators. A closed omnivorous link exists when a feeding

path can be traced to a prey more than one trophic level away,

and from that prey back to the predator through at least one