. INTERESTINGLY, THE METHODS LEAN HEAVILY ON IDEAS FROM ARE IN FA...

2002). Interestingly, the methods lean heavily on ideas from

are in fact coupled. Thus, it may be that an intermediate degree

sociology, where the aim is to identify social cliques within a

of compartmentalization is the most stable.

This chapter closes, then, with a

broader society. An example is shown in Figure 20.19. Also, an

tone that has pervaded much of it:

answers are

alternative perspective has been to emphasize that what have

suggestive but uncertain. Further pro-

uncertain – but it is

been described as distinct food webs in different habitats may

important that we

often be linked by ‘spatial subsidies’ – crucial flows of energy and

gress, though, is essential. One standard

answer of ecologists to the layman’s

discover them

materials (Polis et al., 1997) – as, for example, when lake fish that

normally prey upon other fish in the pelagic (open water) food

question ‘What does it matter if we lose

that species?’ is, quite rightly, ‘But you must also consider the

web, switch to quite different prey in the benthic food web when

wider effects of that loss; losing that species may affect the whole

their preferred prey are scarce (Schindler & Scheuerell, 2002).

That is, what might seem to be separate webs are in fact com-

food web of which it is part’. The need for further understand-

partments within a larger web.

ing of those wider effects is intense.

Since no clear consensus has emerged that food webs are more

compartmentalized than would be expected by chance alone, it

Summary

would be inappropriate to argue that compartmentalization has

been ‘favored’ because compartmentalized webs persist. None

In this chapter, we shift the focus to systems that usually have

the less, since the earliest theoretical studies (e.g. May, 1972), a

consensus has emerged that communities will have increased

at least three trophic levels and with ‘many’ species.

We describe ‘unexpected’ effects in food webs, where, for

stability if they are compartmentalized, and it is easy to see

why this might be so. In the first place, a disturbance to a com-

example, the removal of a predator may lead to a decrease in

prey abundance.

partmentalized web tends to be contained within the disturbed

compartment, limiting the overall extent of the effects in the wider

The indirect effect within food webs that has received most

attention is the trophic cascade. We discuss cascades in systems

web. In addition, though, spatial subsidies between compartments

will tend to buffer individual compartments against the worst

with three and four trophic levels, and address the question of

excesses of disturbances within them. For instance, in the example

whether cascades are equally common in all types of habitat,

above, piscivorous fish, when their preferred prey are rare, may

requiring a distinction to be made between community- and

switch to the benthos rather than driving populations of those

species-level cascades. We ask whether food webs, or particular

preferred prey to extinction. The apparent contradiction between

types of food web, are dominated by either top-down (trophic

cascade) or bottom-up control. We then define and discuss the

these two justifications of the stabilizing properties of compart-

importance of keystone species.

mentalization can be resolved if we emphasize the first where a

2

Benthic producers

Bacteria <1 µm (small)

3

Bacteria >1 <2 µm (medium)

4

Bacteria >2 µm (large)

5

21

6

7

Microciliates

20

Macrociliates

8

Predaceous ciliates

9

25

10

(sea nettle)

24

11

(comb jelly)

12

31

26

13

Cladocera

14

Other zooplankton

45 19

15

(anchovy)

35

16

23

17

Fish larvae

18

43

22

(polychaete)

19

18 34

(oIigochaete)

Other polychaetes

(amphipod)

33

Other meiofauna

41

(Baltic clam)

(rosy clam)

27

(wedge clam)

28

29

(soft-shelled clam)

30

(blue crab)

32

27 4

(bay anchovy)

(croaker)

44

34

(hogchoaker)

1

36

37

(American shad)

38

40 42

(alewife)

39

(bIue-back herring)

40

(menhaden)

(white perch)

42

(striped bass)

(bluefish)

(flounder)

45

Figure 20.19 Pictorial representation of the results of an analysis of a food web from Chesapeake Bay (see also Figure 20.13) in which

interactions between the 45 taxa were quantified and the taxa assigned to compartments (the number of which was not predetermined)

in such a way as to maximize the differential between the connectance within compartments (in this case 0.0099) and that between

compartments (in this case 0.000087, more than two orders of magnitude lower). Food webs may be considered compartmentalized if that

differential is sufficiently large. Arrows represent interactions and point from predator to prey: solid color, within compartments; dashed

lines, between compartments. (After Krause et al., 2002.)

Any ecological community can be characterized by its structure,

Limitations and patterns in food chain length are discussed.

We examine the evidence that food chain length is limited by pro-

its productivity and its temporal stability. The variety of meanings

ductivity, by ‘productive space’ (productivity compounded by the

of ‘stability’ is outlined, distinguishing resilience and resistance,

local and global stability, and dynamic fragility and robustness.

extent of the community) or simply by ‘space’ – but that evidence

is inconclusive. We examine, too, the arguments that food chain

For many years, the ‘conventional wisdom’ was that more

complex communities were more stable. We describe the simple

length is limited by dynamic fragility (ultimately unconvincing)

or by constraints on predator design and behavior. There is a clear

mathematical models that first undermined this view. We show

how, in general, the effects of food web complexity on popula-

need for rigorous studies of many more food webs before

acceptable generalizations can be reached.

tion stability in model systems has been equivocal, whereas for

aggregate properties of whole model communities, such as their

We examine work linking the prevalence of omnivory and

its effect on food web stability, noting that earlier work found

biomass or productivity, complexity (especially species richness)

omnivory to be rare and destabilizing, whereas later work found

tends consistently to enhance stability.

it common and with no consistent effect on stability.

In real communities, too, evidence is equivocal at the popula-

Finally, we ask whether food webs tend to be more com-

tion level, including both studies that have examined the rela-

partmentalized than would be expected by chance. As long

tionships between species richness and connectance and those that

as habitat divisions are subtle, the evidence for compartments

have manipulated richness experimentally. Again, turning to the

aggregate, whole community level, evidence is largely consistent

is typically poor, and there are even greater difficulties in

in supporting the prediction that increased richness increases

demonstrating compartments (or the lack of them) within

stability (decreases variability). We stress, though, the importance

habitats. There is, though, a clear consensus from theoretical

of the nature, not just the richness, of a community in these regards,

studies that communities will have increased stability if they are

returning to the importance of keystone species.

compartmentalized.