Chapter 17 Food Webs

The ocean around Antarctica is one of the most productive marine environments on earth.  hytoplankton especially diatoms thrive in these frigid, windswept seas, where they are food for grazing zooplankton. One of the most important of these zooplankton are krill, shrimplike crustaceans named Euphausia superba. Krill are prey for a wide variety of larger plankton-feeding species, including crabeater seals, penguins, flying seabirds, and many species of fish and squid. The best known of the krill feeders are the baleen whales that once gathered in huge numbers to feed in Antarctic waters (fig.17.1).

FIGURE 17.1 A marine food web in action:feeding baleen whales and birds.

The krill-feeding fishes and squid are eaten by predaceous species, including emperor penguins, larger fish, and Weddell and Ross seals. Leopard seals, a highly carnivorous species, feed on penguins and the smaller seals. Finally, the ultimate predators in this community are the killer whales. which eat seals, including leopard seals, and even attack and consume baleen whales. Huge populations of organisms live in the oceans surrounding Antarctica, all bound together in a tangle of feeding relationships.

How can we go beyond a confusing verbal description to a useful and easily understood summary of the feeding relationships within communities? One of the earliest approaches to the study of communities was to describe who eats whom. Since the beginning of the twentieth century, ecologists have meticulously described the feeding relationships in hundreds of communities. The resulting tangles of relationships came to be called food webs. lfwe define a community as an association of interacting species, a moment's reflection will show that a food web, a summary of the feeding interactions within a community, is one of the most basic and revealing descriptions of community structure. A food web is, essentially, a community portrait (fig. 17.2).

FIGURE 17.2 The Antarctic pelagic food web.


l        A food web summarizes the feeding relations in a community.

l        The feeding activities of a few keystone species may control the structure of communities.

l        Exotic predators can collapse and simplify the structure of food webs.

CASE HISTORIES: community webs

A food web summarizes the feeding relations in a community.

The earliest work on food webs concentrated on simplified communities. In 1927, Charles Elton pointed out that the number of well-described food webs, which he called "food cycles," could be counted on the fingers of one hand. One of the first of those food webs described the feeding relations on Bear Island in the high Arctic (fig.17.3). Summerhayes and Elton (1923) studied the feeding relations there because they believed that the high Arctic, with few species, would be the best place to begin the study of food webs.

FIGURE 17.3 Simple food web of an Arctic island.

Summerhayes and Elton used a food web to present the feeding relations on Bear Island in a single picture. The primary producers in the Bear Island food web are terrestrial plants and aquatic algae. These primary producers are fed upon by several kinds of terrestrial and aquatic invertebrates, which are in turn consumed by birds. The birds on Bear Island are attacked by arctic foxes. Arctic foxes also feed on marine mammals that have washed up onto the beaches and on the dung of polar bears. The polar bears of Bear Island subsist on a diet of seals and beached marine mammals. Seabirds harvest food from the sea around Bear Island but enter the Bear Island food web because they are attacked by foxes, feed on beached marine animals and on the freshwater invertebrates of Bear Island, and contribute dung that fertilizes the primary producers of the island.

The work of Summerhayes and Elton revealed that even in these "impoverished faunas," feeding relations are complex and difficult to study. For instance, they failed to document several probable feeding relations in their Bear Island food web, which they indicated with dotted lines. However, the level of food web complexity increased dramatically as ecologists studied more diverse communities.

Detailed Food Webs Reveal Great Complexity

Now let's go from the high Arctic to the tropics, where Kirk Winemiller (1990) described the feeding relations among freshwater fishes. Winemiller studied the aquatic food webs at two locations in the savannas, or "llanos," of Venezuela and at two other sites in the lowlands of Costa Rica. His study sites supported from 20 to 88 fish species. One of Winemiller's least species-rich study sites was a mediumsized stream called Cano Volcan. This stream flows through the piedmont of the Andes and supports 20 fish species.

Winemiller represented the food webs at his study sites in various ways. In some he only included the "common" fish species whose aggregate abundance comprised 95% of the individuals in his collections. These common-fish webs excluded many rare species. He also drew "top-predator sink webs," food webs consisting of all prey consumed by the top predator in a community, all items consumed by the prey of the top predator, and so on down to the base of the food web. Third, Winemiller constructed food webs that excluded the weakest trophic links, those comprising less than 1% of the diet.

Let's look at the results from Cano Volcan, the simplest fish community. Figure 17.4 shows that even when only the 10 most common fish are included in the food web, it remains remarkably complex. The most comprehensible of Winemiller's food webs were those that focused on the strongest trophic links.

FIGURE 17.4, Food web representing the feeding relations of the 10 most  common fish species at Cano Volcan, Venezuela (data from Winemiller 1990).

Strong Interactions and Food Web Structure

Robert Paine (1980) suggested that, in many cases, the feeding activities of a few species have a dominant influence on community structure. He called these influential trophic relations strong interactions. Paine also suggested that the defining criterion for a strong interaction is not necessarily quantity of energy flow but rather degree of influence on community structure. We will revisit this topic later in the Case Histories section on keystone species, but for now, let's look at how recognizing interaction strength can simplify depictions of food webs.

Paine's distinction between strong and weak interactions within food webs has been used to model the interactions within at least one terrestrial food web. Teja Tscharntke (1992) has worked intensively on a food web associated with the wetland reed Phragmites australis. This reed grows in large stands along the shores of rivers and other wetlands. Tscharntke's study site was along the Elbe River near Hamburg in northwest Germany. Along the river, Phragmites is attacked by Giraudiella inclusa, a fly in the family Cecidomyidae, whose larvae develop within galls called "ricegrain" galls. At the study sites Phragmites is also attacked by Archanara geminipuncta, a moth in the family Noctuidae, whose larvae bore into the stems of Phragmites. Stem-boring by A. geminipuncta induces Phragmites to form side shoots, a response that provides additional sites for oviposition by the gall maker G. inclusa.

Tscharntke discovered that at least 14 species of parasitoid wasps attack G. inclusa. How can so many species attack a single host species and continue to coexist? Does this seem to violate the competitive exclusion principle (see chapter 13)? Tscharntke explains this apparent paradox by pointing out that each parasitoid species appears to specialize on attacking G. inclusa at different times and on different parts of Phragmites. In winter, blue tits, Parus caeruleus, move into stands of Phragmites, where they peck open the galls formed by G. inclusa and eat the larvae, causing mortality in this population as well as in its parasitoids.

Tscharntke represented these trophic interactions with a food web that captures the essential interactions among species in this community (fig. 17.5). Even though there are fewer interactions than in Winemiller's tropical fish webs (see fig.17.4), Tscharntke's web still contains plenty of complexity. However, figure 17.5 focuses the reader on the most important interactions in the community by distinguishing between strong, weaker, and weakest interactions by representing this gradient in interaction strength by red (strong), blue (weak), or green (weakest) lines.

FIGURE 17.5 Food web associated with Phragmites australis (data from Tscharntke 1992).

Figure 17.5 suggests that feeding by blue tits strongly influences the parasitoids Aprostocetus calamarius and Torymus arundinis and their host, G. inclusa, in large gall clusters on main shoots. The other series of strong interactions involves the parasitoids Aprostocetus gratus and Platygaster quadrifarius, which attack the G. inclusa that inhabit small gall clusters in side shoots of Phragmites. These side shoots are in turn stimulated by the stem-boring larvae of the moth A. geminipuncta. Notice that blue tits only weakly influence populations on this side of the web.

By distinguishing between weak and strong interactions, Tscharntke produced an easily understood food web to represent the study community. Identifying strong interactions allows us to determine which species may have the most significant influences on community structure. Those with substantial influence we now call keystone species.

CASE HISTORIES:keystone species

The feeding activities of a few keystone species may control the structure of communities.

Robert Paine (1966, 1969) proposed that the feeding activities of a few species have inordinate influences on community structure. He called these keystone species. Paine's keystone species hypothesis emerged from a chain of reasoning. First, he proposed that predators might keep prey populations below their carrying capacity. Next, he reasoned the potential for competitive exclusion would be low in populations kept below carrying capacity. Finally, he concluded that if keystone species reduce the likelihood of competitive exclusion, their activities would increase the number of species that could coexist in communities. In other words, Paine predicted that some predators may increase species diversity.

Food Web Structure and Species Diversity

Paine began his studies by examining the relationship between overall species diversity within food webs and the proportion of the community represented by predators. He cited studies that demonstrated that as the number of species in marine zooplankton communities increases, the proportion that are predators also increases. For instance, the zooplankton community in the Atlantic Ocean over continental shelves includes 81 species, 16% of which are predators. In contrast, the zooplankton community of the Sargasso Sea contains 268 species, 39% of which are predators. Paine set out to determine if similar patterns occur in marine intertidal communities.

Paine described a food web from the intertidal zone at Mukkaw Bay, Washington, which lies in the north temperate zone at 49°N. This food web is typical of the rocky shore community along the west coast of North America (fig.17.6). The base of this food web consists of nine dominant intertidal invertebrates: two species of chitons, two species of limpets, a mussel, three species of acorn barnacles, and one species of goose-neck barnacle. Paine pointed out that Pisaster commonly consumes two other prey species in other areas, a snail and another bivalve, bringing the total food web diversity to 13 species. Ninety percent of the energy consumed by the middle-level predator, Thais, consists of barnacles. Meanwhile the top predator, Pisaster, obtains 90% of its energy from a mixture of chitons (41%), mussels (37%), and barnacles (12%).

Paine also described a subtropical food web (31°N) from the northern Gulf of California, a much richer web that included 45 species. However, like the food web at Mukkaw Bay, Washington, the subtropical web was topped by a single predaton the starfish Heliaster kubinijii (fig. 17.6). However, six predators occupy middle levels in the subtropical web, compared to one middle-level predator at Mukkaw Bay. Because four of the five species in the snail family Columbellidae are also predaceous, the total number of predators in the subtropical web is 11. These predators feed on the 34 species that form the base of the food web. Despite the presence of many more species in this subtropical web, the top predator, Heliaster, obtains most of its energy from sources similar to those used by Pisaster at Mukkaw Bay. Heliaster obtains 74% of its energy directly from a mixture of bivalves, herbivorous gastropods, and barnacles.

FIGURE 17.6 Roots of the keystone species hypothesis: does a higher proportion of predators in diverse communities indicate that predators contribute to higher species diversity?

Paine found that as the number of species in his intertidal food webs increased, the proportion of the web represented by predators also increased, a pattern similar to that described by G. Grice and A. Hart (1962) when they compared zooplankton communities. As Paine went from Mukkaw Bay to the northern Gulf of California, overall web diversity increased from 13 species to 45 species, a 3.5-fold increase. However, at the same time, the number of predators in the two webs increased from 2 to 11. a 5.5-fold increase. According to Paine's predation hypothesis, this higher proportion of predators produces higher predation pressure on prey populations, which in turn promotes the higher diversity in the Gulf of California intertidal zone.

Does this pattern confirm Paine's predation hypothesis? No, it does not. First, Paine studied a small number of webs--not enough to make broad generalizations. Second, while the patterns described by Paine are consistent, with his hypothesis, they may be consistent with a number of other hypotheses. To evaluate the keystone species hypotheses, Paine needed a direct experimental test.

Experimental Removal of Starfish

For his first experiment, Paine removed the top predator from the intertidal food web at Mukkaw Bay and monitored the

response of the community. He chose two study sites in the middle intertidal zone that extended 8 m along the shore and 2 m vertically. One site was designated as a control and the other as an experimental site. He removed Pisaster from the experimental site and relocated them in another portion of the intertidal zone. Each week Paine checked the experimental site for the presence of Pisaster and removed any that might have colonized since his last visit.

Paine followed the response of the intertidal community for 2 years. Over this interval, the diversity of intertidal invertebrates in the control plot remained constant at 15, while the diversity within the experimental plot declined from 15 to 8, a loss of 7 species. This reduction in species diversity supported Paine's keystone species hypothesis. However, if this reduction was due to competitive exclusion, what was the resource over which species competed?

As we saw in chapter 11, the most common limiting resource in the rocky intertidal zone is space. Within 3 months of removing Pisaster from the experimental plot, the barnacle Balanus glandula occupied 60% to 80% of the available space. One year after Paine removed Pisaster, B. glandula was crowded out by mussels, Mytilus californianus, and goose-neck barnacles, Pollicipes polymerus. Benthic algal populations also declined because of a lack of space for attachment. The herbivorous chitons and limpets also left, due to a lack of space and a shortage of food. Sponges were also crowded out and a nudibranch that feeds on sponges also left. After 5 years, the Pisaster removal plot was dominated by two species: the mussel, M. californianus, and the goose-neck barnacle, P. polymerus.

This experiment showed that Pisaster is a keystone species. When Paine removed it from his study plot, the community collapsed. However, did this one experiment demonstrate the general importance of keystone species in nature? To demonstrate this we need more experiments and observations across a wide variety of communities. Paine followed his work at Mukkaw Bay with a similar experiment in New Zealand.

The intertidal community along the west coast of New Zealand is similar to the intertidal community along the Pacific coast of North America. The top predator is a starfish, Stichaster australis, that feeds on a wide variety of invertebrates, including barnacles, chitons, limpets, and a mussel, Perna canaliculus. During 9 months following Paine's removal of the starfish, the number of species in the removal plot decreased from 20 to 14 and the coverage of the area by the mussel increased from 24% to 68%. As in Mukkaw Bay, the removal of a predaceous starfish produced a decrease in species richness and a significant increase in the density of a major prey species. Again, the mechanism underlying disappearance of species from the experimehtal plot was competitive exclusion due to competition for space.

These results show that intertidal communities thousands of kilometers apart that do not share any species of algae or genera of invertebrates are influenced by similar biological processes (fig. 17.7). This is reassurance to ecologists seeking general ecological principles. However, the two communities are not identical. The New Zealand intertidal community includes a large brown alga, Durvillea antarctica, that vigorously competes for space with the mussel Perna. The mussel M. californianus does not face such a competitive challenge in the North American intertidal zone.

FIGURE 17.7 The effect of removing a top predator from two intertidal food webs (data from Paine 1966, 1971).

In a second removal experiment, Paine removed both the starfish Stichaster and the large brown alga Durvillea from two different study plots. The result was far more dramatic than when Paine had removed the starfish only. After only 15 months, Perna dominated the study area and excluded nearly all other flora and fauna, covering 68% to 78% of the space in the two removal sites. Paine's studies in North America and New Zealand provide substantial support for the keystone species hypothesis. Many other studies quickly followed the lead taken by Paine's pioneering work.

Consumers' Effects on Local Diversity

Jane Lubchenko (1978) observed that previous studies had indicated that herbivores sometimes increase plant diversity, sometimes decrease plant diversity, and sometimes seem to do both. She proposed that to resolve these apparently conflicting results it would be necessary to understand (1) the food preferences of herbivores, (2) the competitive relationships among plant species in the local community, and (3) how competitive relationships and feeding preferences vary across environments. Lubchenko used these criteria to guide her study of the influences of an intertidal snail, Littorina littorea, on the structure of an algal community.

Lubchenko studied the feeding preferences of Littorina in the laboratory. Her experiments indicated that algae fell into low, medium, or high preference categories. Generally, highly preferred algae were small, ephemeral, and tender like the green algae, Enteromorpha spp., while most tough, perennial species like the red alga Chondrus crispus were never eaten or eaten only if the snail was given no other choice.

Lubchenko also studied variation in the abundance of algae and Littorina in tide pools. She found that tide pools with high densities of Enteromorpha, one of the snail's favorite foods, contained low densities (4/m2) of snails. In contrast, pools with high densities of Littorina (233-267/m2) were dominated by Chondrus, a species for which the snail shows low preference. Lubchenko reasoned that in the absence of Littorina, Enteromorpha competitively displaces Chondrus. She tested this idea by removing the Linorina from one of the pools in which they were present in high density and introducing them to a pool in which Enteromorpha was dominant. Lubchenko monitored a third pool with a high density of the snails as a control.

FIGURE 17.8 Effect of Littorina littorea on algal communities in tide pools (data from Lubchenko 1978).

The results of Lubchenko's removal experiment were clear (fig.17.8). While the relative densities of Chondrus, Enteromorpha, and other ephemeral algae remained relatively constant in the control pool, the density of Enteromorpha declined with the introduction of Littorina. Meanwhile, Enteromorpha quickly increased in density and came to dominate the pool from which Lubchenko had removed the snails. In addition, as the Enteromorpha population in this pool increased, the population of Chondrus declined. Lubehenko began another addition and removal experiment in two other pools in the fall to check for seasonal effects on feeding and competitive relations. This second removal experiment produced results almost identical to the first. Where Littorina were added, the Enteromorpha population declined, while the Chondrus population increased. Where the snails were removed, the Chondrus population declined, while the Enteromorpha population increased.

How can we explain Lubchenko's results? Littorina prefers to feed on Enteromorpha, a species that can outcompete Chondrus in tide pools. So, in the absence of the snails, Chondrus is competitively displaced by Enterornorpha. However, where it is present in high densities, Littorina grazes  down  the Enteromorpha  population,  releasing Chondrus from competition with Enteromorpha.

What controls the local population density of Littorina ? Apparently, the green crab, Carcinus maenus, which lives in the canopy of Enteromorpha, preys upon young snails and can prevent the juveniles from colonizing tide pools. Adult Littorina are much less vulnerable to Carcinus but rarely move to new tide pools. Populations of Carcinus are in turn controlled by seagulls. Here again, we begin to see the complexity of a local food web and the influences that trophic interactions within webs can have on community structure.

So, within tide pools Enteromorpha can outcompete the other tide pool algae for space and Enteromorpha is the preferred food of Littorina. How might feeding by the snails affect the diversity of algae within tide pools? The relationship between the snails and the algal species they exploit is similar to the situation studied by Paine, where mussels were the competitively dominant species and one of the major foods of the starfish Pisaster.

Lubchenko examined the influence of Littorina on algal diversity by observing the number of algal species living in tide pools occupied by various densities of snails (fig. 17.9). As the density increased from low to medium, the number of algal species increased. Then, as the density increased further, from medium to high, the number of algal species declined.

How would you explain these results? At low density, the feeding activity by Littorina is not sufficient to prevent Enteromorpha from dominating a tide pool and crowding out some other species. At medium densities, the snail's feeding, which concentrates on the competitively dominant species, prevents competitive exclusion and so increases algal diversity. However, at high densities the feeding requirements of the population are so high that the snails eat their preferred algae as well as less preferred species. Consequently, intense grazing by snails at high density reduces algal diversity.

What would happen if Littorina preferred to eat competitively inferior species of algae? This is precisely the circumstance that occurs on emergent substrata, rock surfaces that are not submerged in tide pools during low tide. On these emergent habitats the competitively dominant algae are species in the genera Fucus and Ascophyllum, algae for which the snails show low preference. On emergent substrata, the snails continue to feed on ephemeral, tender algae such as Enteromorpha, largely ignoring Fucus and Ascophyllum. In this circumstance, Lubchenko found that algal diversity was highest when Littorina densities were low (fig.17.9).

FIGURE 17.9 Effect ofLittorina littorea on algal species richness in tide pools and emergent habitats (data from Lubchenko 1978).

What produces this reduction in diversity? Let's think first about the effect of competition by  Fucus and Ascophyllum. In the absence of disturbance, these two algae will gradually cover all emergent substrata, crowding out other algal species in the process. Feeding by snails accelerates elimination of the competitively subordinant species. In this case, competition by Fucus and Ascophyllum and exploitation by Littorina are both pushing the community toward reduced species richness.

Lubchenko's research improved our understanding of how trophic interactions can affect community structure. Her work demonstrated that the influence of consumers upon the structure of food webs depends upon their feeding preferences, the density of local consumer populations, and the relative competitive abilities of prey species. While Lubchenko moved the field well beyond the conceptual view held by ecologists when Paine first proposed the keystone species hypothesis, one basic element of the original hypothesis remained: Consumers can exert substantial control over food web structure; they can act as keystones.

Can predators act as keystone species in environments other than the intertidal zone? The next two examples concern keystone species in riverine and terrestrial environments.

Fish as Keystone Species in River Food Webs

Mary Power (1990) tested the possibility that fish can significantly alter the structure of food webs in rivers. She conducted her research on the Eel River in northern California, where most precipitation falls during October to April, sometimes producing torrential winter flooding. During the summer, however, the flow of the Eel River averages less than 1 m3 per second.

In early summer, the boulders and bedrock of the Eel River is covered by a turf of the filamentous alga Cladophora (fig. 17.10). However, the biomass of the algae declines by midsummer and what remains has a ropy, prostrate growth form and a "webbed" appearance. These mats of Cladophora support dense populations of larval midges in the fly family Chironomidae. One chironomid, Pseudochironomus richardsoni, is particularly abundant. Pseudochironomus feeds on Cladophora and other algae and weaves the algae into retreats, altering their appearance in the process.

FIGURE 17.10 Seasonal changes in hiomass and growth form of benthic algae in the Eel River, Caltfornia: (a) in early summer, June 1989, (b) in late summer, August 1989.

Chironomids are eaten by predatory insects and the young (known as fry) of two species of fish: a minnow called the California roach, Hesperoleucas symmetricus, and three-spined sticklebacks, Gasterosteus aculeatus. These small fish are eaten by young steelhead trout, Oncorhynchus mykiss. Steelhead and large roach eat predatory invertebrates, and. large roach also feed directly upon benthic algae. These interactions form the Eel River food web pictured in figure 17.11.

FIGURE 17.11 Food web associated with algal turf during the summer in the Eel River. California.

Power asked whether or not the two top predators in the Eel River food web, roach and steelhead, significantly influence web structure. She tested the effects of these fish on food web structure by using 3 mm mesh to cage off 12 areas 6 m2 in the riverbed. The mesh size of these cages prevented the passage of large fish but allowed free movement of aquatic insects and stickleback and roach fry. Power excluded fish from six of her cages and placed 20 juvenile steelhead and 40 large roach in each of the other six cages. These fish densities were within the range observed around boulders in the open riven.

Significant differences between the exclosures and enclosures soon emerged. Algal densities were initially similar; however, enclosing fish over an area of streambed significantly reduced algal biomass (fig.17.12). In addition, the Cladophora within cages with fish had the same ropy, webbed appearance as Cladophora in the open river.

FIGURE 17.12 The influence of juvenile steelhead and California roach on benthic algal biomass itt the Eel River (&ira from Power 1990).

How do predatory fish decrease algal densities? The key to answering this question lies with the Eel River food web (see fig. 17.11). Predatory fish feed heavily on predatory insects, young roach, and sticklebacks. Lower densities of these smaller predators within the enclosures increased predation on chironomids. Higher chironomid density increased the feeding pressure of these herbivores on algal populations. This explanation is supported by Power's estimate that enclosures contained lower densities of predatory insects and fish fry and higher densities of chironomids (fig. 17.13). By enclosing and excluding fish from sections of the Eel River, Power, like Paine and Lubchenko, who worked in the intertidal zone, demonstrated that fish act as keystone species in the Eel River food web.

FIGURE 17.13  Effect of juvenile steelhead and roach on numbers of  insects and young (fry) roach and sticklebacks (data from Power 1990).

All of the examples that we have discussed so far have been aquatic. Do terrestrial communities also contain keystone species? An increasing body of evidence indicates that they do.

The Effects of Predation by Birds on Herbivory

Let's move now from the Mediterranean climate of the Eel River basin to the boreal forests of northern Sweden. Where Ola Atlegrim (1989) studied the influence of birds on herbivorous insects and insect-caused plant damage. It appears that insectivorous birds may act as keystones in boreal forests through their effects on populations of herbivorous insects.

Atlegrim studied the food web associated with the bilberry, Vaccinium myrtillus, which is a dominant understory shrub in many boreal forests in northern Sweden. The insects that commonly feed on Voccinium include caterpillars of the moth families Geometridae and Tortricidae and the larvae of the Hymenoptera known as sawflies. The geometrid and sawfiy larvae feed on the bilberry from exposed positions, while the tortricid larvae bind leaves together with silk to form a shelter within which they feed. Because populations of these larvae can reach high densities, they can do considerable damage to Vaccinium. However, Atlegrim observed that larval insect densities peak when many insectivorous birds are feeding insects to their young and posed the following questions: (1) Do birds reduce the density of insect larvae feeding on Vaccinium? (2) Do birds have different effects on larvae feeding from exposed versus concealed positions? (3) Does predation by birds reduce larval insect damage to the shoots of Vaccinium ?

Atlegrim's study sites were located approximately 20 km northwest of Umea in northern Sweden. He established five study areas in forests ranging from 70 to 120 years old. At each study area, he established 10 study plots 4 m2 and built a bird exclosure over 5 of them. Exclosures consisted of 40 mm2 nylon mesh supported by a wooden frame.

Atlegrim took care to ensure that he could attribute any experimental effects to the exclusion of birds. His exclosures excluded birds but allowed small predaceous mammals, such as shrews, and predaceous invertebrates to move freely into and out of the study plots. He also kept track of the densities of these alternative predators by periodically sampling them with pitfall traps. Why was this an important aspect of Atlegrim's study? In the absence of predation by birds, higher densities of herbivorous insects might have attracted higher numbers of other predators, that is, produced a localized numerical response (see chapter 10). Atlegrim also measured the intensity of sunlight within his exclosures and in adjacent control plots. Why was this aspect of the study necessary? If the exclosures created significant shading, physical effects alone (see chapters 4-6) could have affected the distributions of herbivorous insects. Finally, Atlegrim measured the density of Vaccinium shoots to verify similar densities in exclosure and control plots. His measurements showed that levels of light, densities of nouavian predators, and densities of Vaccinium shoots were similar on exclosure and control plots.

Exclosures increased larval insect density an average of 63% across all study sites (fig. 17.14). So, the answer to Atlegrim's first question is yes. Insectivorous birds reduce the densities of herbivorous insect larvae feeding on Vacciniurn. However, as Atlegrim predicted, some herbivorous larvae are more vulnerable to insectivorous birds than are others. Sawfly and geometrid larvae, which feed in exposed positions, were significantly higher within exclosures, while the densities of tortricid larvae, which feed in their constructed shelters, showed no effects of bird exclusion. Higher densities of herbivorous insect larvae translated directly into higher levels of damage to Vacciniurn.

FIGURE 17.14 Effect of insectivorous birds on herbivorous insect populations on Vaccinium myrtiIlus (data from Atlegrim 1989).

What other piece of information might increase our confidence that the differences between exclosure and control plots were due to bird predation? One of the most significant bits of evidence would be direct observations of birds feeding on the control plots. Atlegrim observed three bird species feeding on control plots: Hazel hen chicks, Tetrastes bonasia, great tits, Parus major, and pied flycatcher, Ficedula hypoleuca.

Insectivorous birds also reduce insect populations and insect damage on plants in midlatitude forests in North America. Robert Marquis and Christopher Whelan (1994) used 3.8 cm white nylon mesh to exclude birds from 30 white oak, Quercus alba, saplings at the Tyson Research Center in Eureka, Missouri, during the growing seasons of 1989 and 1990. The bird community at this site includes several dozen species composed of a shifting mix of spring migrants and summer and spring residents.

The researchers sprayed another set of 30 white oak saplings each week with a pyrethroid insecticide. They also handpicked any remaining herbivorous insects from these trees. A third set of white oaks, the control, was not manipulated.

Marquis and Whelan's caged plants were populated by larger numbers of herbivorous insects and experienced significantly greater insect damage (fig. 17.15). These results are consistent with those of Atlegrim's earlier study. Marquis and Wbelan also measured the biomass of each of their trees in 1990 and 1991. They found that the biomasses of sprayed and uncaged white oaks were significantly higher than the biomass of caged white oaks. In other words, the higher densities of herbivorous insects on the trees from which birds were excluded reduced their growth rates. One implication of these results is that insectivorous birds increase the growth rates of temperate forest trees.

FIGURE 17.15 Effect of insectivorous birds on herbivorous inaect populations, leaf damage, and sapling growth in white oaks (data from Marquis and Whelan 1994).

Many studies of food webs and keystone species have been done since Robert Paine's classic study of the intertidal food web at Mukkaw Bay, Washington. The studies have revealed a great deal of biological diversity, which has prompted biologists to ask what characterizes keystone species. This reflection is necessary to avoid the possibility that the term may become so inclusive that it becomes meaningless. The conclusions reached by a conference designed to address this question are summarized in fgure 17.16 (Power et al. 1996). Keystone species are those that, despite low biomass, exert strong effects on the structure of the communities they inhabit. As we shall see in the following Case Histories example, those strong effects are not always positive.

FIGURE 17.16 What is a keystone species (data from Power et al. 1996)?

CASE HISTORIES: exotic predators

Exotic predators can collapse and simplify the structure of food webs.


Introduced Fish: Predators That Simplify Aquatic Food Webs

People have moved all sorts of species around the planet, but one of the most commonly introduced groups of organisms is fish. Introduced fish often substantially change the food webs of the water bodies where they are introduced. For instance, introduced fishes have devastated the native fishes of Lake Atitlan and Gatun Lake in Central America. In both these cases, introduced predaceous fish completely reshaped the lake food web, producing a less diverse community. The dramatic impact of exotic species on communities may result from their being outside the evolutionary experience of the prey populations in the local community.

Today we are witnessing what may be the greatest devastation ever wrought by an introduced predator. That predator is the Nile perch, Lates nilotica, and the aquatic system is Lake Victoria, one of the great lakes of East Africa. Lake Victoria is approximately 69,000 km2 and lies right on the equator. Despite its large surface area, the lake is relatively shallow. The deepest point is approximately 100 m, and most of the lake is less than 60 m deep.

Lake Victoria harbors one of the greatest concentrations of fish species in the world, and the Nile perch may be producing the greatest extinction of vertebrate animals to occur in modem times. The fish fauna of the lake, which included more than 400 species, is being rapidly reduced to a community dominated by a handful of species. Just three species now dominate the fish catches around Lake Victoria: the introduced Nile perch, the introduced Nile tilapia, Oreochromis niloticus, and a single native species, the omena, Rastrineobola argentea, a planktivorous minnow of the open waters of the lake.

Hundreds of fish species appear on their way to extinction because humans introduced one more fish species into a lake already containing over 400 species. The Nile perch is a predaceous fish native to East Africa, where it attains a length of nearly 2 m and a weight of 60 kg. It may have eliminated species from other East African lakes in the distant past, since lakes with Nile perch have fewer species of fish in the family Cichlidae than do lakes without it. Nile perch were introduced to Lake Victoria around 1954, along with several other fish species from surrounding rivers and lakes. However, Nile perch remained a minor component of the fish fauna for nearly two decades, and then in the early 1980s, its population exploded.

As the population of Nile perch grew, the populations of nearly all other fish species in the lake declined. The only native fish that increased during this period was the planktivorous omena. K. Ligtvoet and E Witte (1991 ) used food webs to represent the changes in the Lake Victoria fish community. As you can see in figure 17.17, the original food web consisted of a complex mixture of fishes belonging to several families and orders. Three species of fish dominate the present food web. Notice that Nile perch appear twice in the present food web, as a top predator and as a middle-level predator. In other words, there is a substantial amount of cannibalism going on in the community, with small Nile perch serving as a trophic link between insect larvae and large Nile perch. As impressive as the contrast shown in figure 17.17 may be, keep in mind that it is a gross simplification and shorthand for the trophic interactions of over 400 species of fish. For a comparison, look back at Winemiller's food webs for Cano Volcan, a food web with only 10 species (see fig. 17.4). The original Lake Victoria food web contained 40 times the number of fish species in the Cano Volcan food web.

FIGURE 17.17 Influence of an exotic predator. Nile perch, on the food web of Lake Victoria (data from Ligtvoet and Witte 1991)

The Nile perch has had a major effect on the food web of Lake Victoria and is contributing to a massive extinction. However, can we attribute all these changes in the biota of the lake to this one predator? Les Kaufman (1992) points out that the changes in the Lake Victoria fish community coincide with other changes in the ecosystem. For instance, concentrations of dissolved oxygen have declined significantly. Lake Victoria water was once oxygenated from top to bottom. Before 1978, organisms requiring oxygen lived in the deepest portions of the lake, which experienced periods of reduced oxygen concentration but were not anoxic. Now, however, the water below about 30 m is often essentially devoid of oxygen.

Depletion of oxygen has produced massive fish kills. Biologists had observed some fish kills at the surface of Lake Victoria but discovered many others only when they began exploring the bottom with a remote-operated vehicle. Images recorded by this vehicle in 1987 showed that there were no fish below about 30 m and that the bottom was littered with dead fish. Massive fish kills, including kills of Nile perch, have become a regular occurrence in Lake Victoria.

Maybe the Nile perch has not caused all of the changes in Lake Victoria. Perhaps the changes are ultimately a consequence of changes in the ecosystem, driven by heavy nutrient additions from the surrounding human population and consequent eutrophication of the lake (see chapter 3). However, as we shall see in chapter 18, predaceous fish like Nile perch can produce changes in ecosystem functioning that may in turn strongly influence populations and communities. Though we look for single causes of complex phenomena, ecological systems are affected by a complex interplay between biotic and abiotic factors. Some of these mutual influences will become more apparent in chapters 18 and 19 as we begin our discussion of the ecology of ecosystems.

APPLICATIONS AND TOOLS: humans as keystone species

People have long manipulated food webs both as a consequence of their own feeding activities and by introducing or deleting species from existing webs. In addition, many of these manipulations have focused on keystone species. Consequently, either consciously or unwittingly, people have, themselves, acted as keystone species in communities.

The Empty Forest: Hunters and Tropical Rain Forest Animal Communities

The current plight of the tropical rain forest is well known. However, Kent Redford (1992) points out that with few exceptions, most studies of human impact on the tropical rain forest have concentrated on direct effects of humans on vegetation, mainly on deforestation. Redford expands our view by examining the effects of humans on animals. The picture that emerges from this analysis is that humans have so reduced the population densities of rain forest animals in many areas that they no longer play their keystone roles in the system, a situation Redford calls "ecologically extinct."

Redford estimated that subsistence hunting, a major source of protein for many rural people, results in an annual death toll of approximately 14 million mammals and 5 million birds and reptiles within the Brazilian Amazon. He estimated further that commercial hunters, seeking skins, meat, and feathers, kill an additional 4 million animals annually. Consequently, the total take by hunters within the Brazilian Amazon is approximately 23 million individual animals. However, this figure underestimates the total number of animal deaths, since many wounded animals escape from hunters only to die. Including those fatally wounded animals that escape, Redford places the annual deaths within the Brazilian Amazon at approximately 60 million animals.

Hunters generally concentrate on a small percentage of larger bird and mammal species, howeven For instance, Redford estimated that at Cocha Cashu Biological Station in Manu National Park, located in the Amazon River basin in eastern Peru, hunters concentrate on 9% of the 319 bird species and 18% of the 67 mammal species. Because hunters generally concentrate on the larger species, this small portion of the total species pool makes up about 52% of the total bird biomass and approximately 75% of the total mammalian biomass around Manu National Park (fig. 17.18).

FIGURE 17.18 Highly selective hunting by Amazonian natives (data from Redford 1992).

As impressive as all these numbers are, there remains a critical question: Do hunters reduce the local densities of the birds and mammals they hunt? The answer is yes. Redford estimated that moderate to heavy hunting pressure in rain forests reduces mammalian hiomass by about 80% to 93% and bird biomass by about 70% to 94%.

There may be cause for concern, however, that goes beyond the losses of these immense numbers of animals. As you might expect, many large rain forest mammals and birds may act as keystone species (fig.17.19). If so, their decimation will have effects that ripple through the entire community. The first to suggest a keystone role for the large animals preferred by rain forest hunters was John Terborgh (1988), who presented his hypothesis in a provocative essay titled, "The Big Things That Run the World."

FIGURE 17.19 Large predators such as this jaguar may act as keystone species in tropical rain forests.

Terborgh's hypothesis has been supported by a variety of studies. He observed that in the absence of pumas and jaguars on Barro Colorado Island, Panama, medium-sized mammal species are over 10 times more abundant than in areas still supporting populations of these large cats.R.Dirzo and A. Miranda (1990) compared two forests in tropical southern Mexico, one in which hunting had eliminated most of the large mammals and one in which most of the large mammals were still present. The comparison was stark. In the absence of large mammals such as peccaries, jaguars, and deer, the researchers found forests carpeted with undamaged plant seedlings and piled with uneaten and rotting fruits and nuts, signs of a changing forest. Such observations prompted Redford to warn, "We must not let a forest full of trees fool us into believing all is well." Tropical rain forest conservation must also include the large, and potentially keystone, animal species that are vulnerable to hunting by humans.

Ants and Agriculture:

Keystone Predators for Pest Control

In 1982, Stephen Riseh and Ronald Carroll published a paper describing how the predaceous fire ant, Solenopsis gerninata, acts as a keystone predator in the food web of the corn-squash agroecosystem in southern Mexico. While "natural enemies" had been used to control insect pests for some time, Risch and Carroll put these efforts into a community context. They drew conceptual parallels between biological control of insects with natural enemies and studies of the influences of keystone species, citing studies of the influences of herbivores on plant communities and the effects of predators on intertidal communities. In their own experiments, Risch and Carroll demonstrated how predation by Solenopsis in the corn-squash agroecosystem reduces the number of arthropods and the arthropod diversity (fig.17.20). This study showed how Solenopsis could act as a keystone species to the benefit of the agriculturist.

FIGURE 17.20 Effect ofSolenopsis geminata on the arthropod populations on corn (data from Risch and Carroll 1982).

The conceptual breakthrough represented by the work of Risch and Carroll is impressive. However, their work had been anticipated, 1,700 years earlier, by farmers in southern China. H. Huang and P. Yang (1987) cite Ji Han, who, in A.D. 304, wrote "Plants and Trees of the Southern Regions" in which he included the following:

      The Gan (mandarin orange) is a kind of orange with an exceptionally sweet and delicious taste...In the market, the natives of Jiao-zhi [southeastern China and North Viemam] sell ants stored in bags of rush mats. The nests are like thin silk. The bags are all attached to twigs and leaves, which, with the ants inside the nests, are for sale. The ants are red- dish-yellow in color, bigger than ordinary ants. In the south, if the Gan n'ees do not have this kind of ant, the fruits will be damaged by many harmful insects and not a single fruit will be perfect.

Now, 17 centuries after the observations of Ji Han, we know this ant as the citrus ant, OecophyUa smaragdina. The use of this ant to control herbivorous insects in citrus orchards was unknown outside of China until 1915. In 1915, Walter Swingle, a plant physiologist who worked for the U.S. Department of Agriculture, was sent to China to search for varieties of oranges resistant to citrus canker, a disease that was devastating citrus groves in Florida. While on this trip, Swingle came across a small village where the main occupation of the people was growing ants for sale to orange growers. The ant was the same one described by Ji Han in A.D. 304.

Oecophylla is one of the weaver ants, which use silk to construct a nest by binding leaves and twigs together. These ants spend the night in their nest. During the day, the ants spread out over the home tree as they forage for insects. Farmers place a nest in a tree and then run bamboo strips between trees so that the ants can have access to more than one tree. The ants will eventually build nests in adjacent trees and can colonize an entire orchard.

The ants harvest protein and fats when they gather insects from their home tree, but they have other needs as well. They also need a source of liquid and carbohydrates, and they get these materials by cultivating Homoptera, known as soft-scale insects or mealy bugs, which produce nectar. The ants and soft-scale insects have a mutualistic relationship in  which the ants transport the insects from tree to tree and protect them from predators. In return the ants consume the nectar produced by the soft-scale insects. Because of this mutualism with the soft-scale insect, which can itself be a serious pest of citrus, several early agricultural scientists expressed skepticism that Oecophylla would be an effective agent for pest control in citrus. They suggested that the use of this ant could produce infestations by soft-scale insects.

Despite these criticisms, all Chinese citrus growers interviewed insisted that OecophyUa is effective at pest con trol and that the damage caused by soft-scale insects is minor. Research done by Yang appears to have solved this apparent contradiction. Comparing orange trees treated with chemical insecticides to those protected by Oecophylla, Yang recorded higher numbers of soft-scale insects in the trees tended by ants. However, these higher numbers did not appear to cause serious damage to the orange trees. When Yang inspected the soft-scale insects closely, he found that they were heavily infested with the larvae of parasitic wasps. He also found that the ants did not reduce populations of lacewing larvae and ladybird beetles, predators that feed on soft-scale insects. Huang and Yang concluded that Oecophylla is effective at pest control because while it attacks the principal, larger pests of citrus, it does not reduce populations of other predators that attack the smaller pests of citrus, such as soft-scale insects, aphids, and mites (fig. 17.21).

FIGURE 17.21  While pests in this North American orange orchard are controlled mainly by chemical insecticides, weaver ants have been used to control insect pests of orange orchards in China for over 17 centuries.

The association between Oecophylla and citrus trees seems similar to that between ants and acacias (see chapter 15). There is a difference, however. Humans maintain Oecophylla as a substantial component of the food web in citrus orchards. Not only have specialized farmers historically cultivated and distributed the ants, Oecophylla must also be protected from the winter cold. The ant cannot survive the winter in southeast China in orange trees. Consequently, farmers must generally provide shelter and food for the ants during winter.

The labor and expense of maintaining these ants through the winter may be reduced by mixed plantings of orchard trees. Farmers in Shajian village in the Huaan district of southeast China have successfully maintained Oecophylla over the winter in mixed plantings of orange and pomelo trees. During winter the ants are mostly in pomelo trees, which are larger and have thicker foliage than orange trees, characteristics that reduce cooling rates on winter nights. In this situation, farmem do not have to add new nests of OecophyUa each spring. Gradually, the ant has become integrated into the mixed citrus and pomelo orchards and requires little special care from the farmers.

The farmers of southeast China have employed Oecophylla as a keystone species in a complex citrus-based food web for a long time. However, the results would not be the same with just any ant species. The citrus growers required a species that acts in a particular way. One wonders how long farmers of the region had experimented with this species before Ji Han wrote his account of their activities in A.D. 304.


A food web summarizes the feeding relations in a community. The earliest work on food webs concentrated on simplified communities in areas such as the Arctic islands. However, researchers such as Charles Elton (1927) soon found that even these so-called simple communities included very complex feeding relations. The level of food web complexity increased substantially, however, as researchers began to study complex communities. Studies of the food webs of tropical freshwater fish communities revealed highly complex networks of trophic interaction that persisted even in the face of various simplifications. A focus on strong interactions can simplify food web structure and identify those interactions responsible for most of the energy flow in communities.

The feeding activities of a few keystone species may control the structure of communities. Robert Paine (1966) proposed that the feeding activities of a few species have inordinate influences on community structure. He predicted that some predators may increase species diversity by reducing the probability of competitive exclusion. Manipulative studies of predaceous species have identified many keystone species, including starfish and snails in the marine intertidal zone and fish in rivers. On land, birds exert substantial influences on communities of their arthropod prey. Jane Lubchenko (1978) demonstrated that the influence of consumers on community structure depends upon their feeding preferences, their local population density, and the relative competitive abilities of prey species. Keystone species are those that, despite low biomass, exert strong effects on the structure of the communities they inhabit.

Exotic predators can collapse and simplify the structure of food webs. Introduced fishes have devastated the native fishes of Lake Atitlan and Gatun Lake in Central America. Introduction of the Nile perch is rapidly reducing the species-rich fish fauna of Lake Victoria to a community dominated by a handful of species. The influence of the Nile perch on the fish community of Lake Victoria is enmeshed with massive changes in the lake's ecosystem.

Humans have acted as keystone species in communities. People have long manipulated food webs both as a consequence of their own feeding activities and by introducing or deleting species from existing food webs. In addition, many of these manipulations have been focused on keystone species. Hunters in tropical rain forests have been responsible for removing keystone animal species from large areas of the rain forests of Central and South America. Chinese farmers have used ants as keystone predators to control pests in citrus orchards for over 1,700 years.


1. You could argue that the classical food web of Bear Island    included several communities, each with its own food web. What were some of the different communities that    Summerhayes and Elton (1923) included in their web? On the other hand, because the Bear Island food web includes significant movement of energy (food) and nutrients between what many ecologists might consider to be separate communities, what does their food web say about the distinctness of what we call communities?

2. Winemiller (1990) deleted "weak" trophic links from one set of food webs that he described for fish communities in Venezuela (see fig. 17.4). What was his criterion for designating weak interactions? Earlier. Paine (1980) suggested that ecologists could learn something by focusing on "strong" links in communities. How did Paine's criterion for determining a strong link differ from Wineminer's?

3. What is a keystone species? Paine (1966, 1969) experimented with two starfish that act as keystone species in their intertidal communities along the west coast of North America and in New Zealand. Describe how the intertidal communities in these two areas are similar. Describe the differences between these two communities and the differences in the design of Paine's experiments in these two areas.

4. Explain how the experiments of Lubchenko (1978) showed that feeding preferences, population density, and competitive relations among food species all potentially contribute to the influences of "keystone" consumers on the structure of communities. What refinements did the work of Lubehenko add to the keystone species hypothesis?

5. When Power (1990) excluded predaceous fish from her river sites, the density of herbivorous insect larvae (chironomids) decreased. Use the food web described by Power to explain this response.

6. Using Tscharntke's food web (1992) shown in figure 17.5, predict which species would be most affected if you excluded the bird at the top of the web. Parus caeruleus. What species would be affected less? Assume that P.caeruleus is a keystone species in this community.

7. Atlegrim (1989) and also Marquis and Whelan (1994) showed that birds in high latitudes and temperate forests reduce insect populations. The results of this research suggests that birds act as keystone species in some communities. What else would we need to know about the birds in these communities before we could conclude that they are keystones in the strict sense? (Hint: Consider figure 17.16.)

 8. Notice that in the study by Marquis and Whelan (1994) the biomass of uncaged Q. alba was as great as that of sprayed individuals. In other words, spraying protected oak seedlings as well as birds. If spraying can control herbivorous forest insects, why rely on birds to improve tree growth? What advantages does predation by birds have over spraying?

 9. Some paleontologists have proposed that overhunting caused the extinction of many large North American mammals at the end of the Pleistocene about 11,000 and 10,000 years ago. The hunters implicated by paleontologists were a newly arrived predatory species, Homo sapiens. Offer arguments for and against this hypothesis.

10. All the keystone species work we have discussed in this chapter has concerned the influences of animals on the structure of communities. Can other groups of organisms act as keystones? What about parasites and pathogens?