16 and the Balance of

S.L. Pimm

16.1 What Biodiversity Is Good for

As we destroy biological diversity, what else are we doing to the environ­ ment, what is being changed, and how will those changes affect us? One part of the answers to these questions is provided by Lawton and Brown's consideration of redundancy (Chap. 12). Part of what the environment does for us involves " services" - the movement of energy and nutrients through the air, water, and land, and through the food chains (Ehrlich, Foreword). Just how much biological diversity we need to keep the movement at approximately natural levels is a question of critical im­ portance. Nonetheless, it is not a question that is commonly asked. A major synthesis of theories on the dynamics of nutrient cycling (DeAngelis 1991) devotes little space to the consequences of changes in the numbers of species per : It is the number of trophic levels that receives the atten­ tion. One might well conclude that, over broad limits, ecosystem services will continue to be provided, so long as there are some plants, some animals, some , and so on. Lawton and Brown conclude that numerous species are redundant. One cannot be confident about this conclusion, for the evidence one way or another is still very limited. Even if it were correct, it would not provide a complete answer to the questions of the value of biological diversity. The conclusion only deals with the average movement of material; ecologists have long recognised that diversity may be important in determining the variability and persistence of ecological processes and their resistance to external changes. Following the loss of diversity, one often hears ideas such as: the loss threatens ecological integrity; the environment becomes more fragile as diversity decreases; have some capacity to resist change or to recover from it, that is, there is some balance of nature or stability, and this balance is affected by the loss of diversity. So what do all these terms mean and what scientific principles do they encompass? One illustration of these concerns involves the fisheries in the Great Lakes of North America. For over a century and a half, these lakes have been subjected to pollution, intense fishing, the loss of several native species, E.-D. Schulze et al. (eds.), Biodiversity and Ecosystem Function © Springer-Verlag Berlin Heidelberg 1994 348 S.L. Pimm and the deliberate and accidental introduction of other fish species. For example, the Lake Erie fishery between 1920 and 1950 was dominated by blue pike (33% of the yield by weight), lake herring (14%), and yellow perch (15%); by 1979-1983 the first two species had been lost, and rainbow smelt accounted for 56% of the yield (Loftus et al. 1987). In contrast, the average total landings increased slightly (about 30%). Similar figures are noted for the other lakes. Simply, the secondary of these ecosystems has changed remarkable little during a period when the com­ position changed greatly. Considerable experience with the lakes and their fisheries (see the papers in Christie and Spangler 1987) reveals that there is no comfort in the similarity of the total yields. In part, this is because the blue pike are worth much more per kilogram than rainbow smelt. (In reality, the species do matter; we cannot reduce all to flows of nutrients.) More important is the sense that the ecosystems in the lakes no longer function in the same way as they did in the past. It is not always clear what these changes in function are, but they are of considerable concern. They involve the internal dynamics of the lakes and they are independent of such summary statistics as overall yield. So just what is diversity good for? It seems to have something to do with what we call stability. has a history, and I want to review it briefly to set the context of what follows.

16.2 A History of Ecological Stability

In his 1958 book, The Ecology of Invasions by Animals and Plants, Elton wrote: I will now try and set out some of the evidence that the balance of relatively simple communities of plants and animals is more easily upset than that of richer ones; that [the balance] is more subject to destructive oscillations in populations, especially of animals, and more vulnerable to invasions.... For if this can be shown to be anywhere near the truth, it will have to be admitted that there is something very dangerous about handling cultivated land as we handle it now, and [it will be] even more dangerous if we continue down the present road of 'simplification for efficiency'. Elton then discuses the evidence for his assertion that simple communities are less stable than complex ones. His first evidence comes from "mathe­ matical speculation about ." Simple models of a pred­ ator and its prey or a single host and its parasite have population densities that "fluctuate in numbers considerably." Such fluctuations would mean that in simple communities, populations would "never have constant population levels, but would be subject to periodic 'outbreaks' of each species". Elton's second line of evidence was from laboratory experiments that made the Biodiversity and the Balance of Nature 349 same point. The third piece of evidence is that the natural on small islands seem to be much more vulnerable to invading species than those of continents. His fourth argument was that "invasions and pest oubreaks most often occur on cultivated or planted land - that is, habitats and communities very much simplified by man". His fifth argument was similar. Insect out­ breaks, Elton claimed, were a feature of simple temperate forests but not of complex tropical forests. His sixth argument seems to recapitulate the earlier ones: insect pests are more likely to invade and become a nuisance in orchards, especially following spraying designed to eliminate the pests. Elton's arguments have received extensive discussion. The models and the laboratory experiments available to Elton looked only at simple systems: Models of, or experiments with, complex systems were not available as a control. The data on invasions were compelling, but comparative studies of population fluctuations were nonexistent. Moreover, Elton discussed a range of very different phenomena: population variability, the potential of a species, the rate at which a pest species might increase in numbers after the removal of its predators, how readily communities may be invaded, and the extent of damage the invading species cause. Yet this range is perhaps the most remarkable part of Elton's argument. Ecologists should be thinking about different kinds of stability: population variability, population recovery, ease of invasion, and consequences of invasion. All are ways of characterizing population and changes to facilitate comparisons. Moreover, Elton considers a factor called "complexity" that should determine "stability" and the importance of their interdependence. Notice something else: the effect of is to destroy stability, and with the loss of stability comes the loss of species, and so less complexity. Systems that are less complex are less stable, more likely to lose species, and so on.

16.2.1 Controversy

The relationships between stability and complexity have generated a large literature (Pimm 1982, 1984). Controversy arose because of the many meanings of "stability" and "complexity" (Pimm 1984). Elton and more recent theoretical and empirical ecologists have used the word "stability" to mean at least five different things (Pimm 1984, 1991b). Theoreticians have considered stability in its strict mathematical sense, where systems are either stable or not. So given May's (1974) argument that what we observe most often in nature will be stable systems, the theoretical results make predictions about the kind of communities we observe and those that we do not. The other four definitions of stability do permit comparisons of dif­ ferent existing systems, however, and these were the concerns of empirical ecologists. 350 S.L. Pimm

1.2

1.0 E .;:::J 0.8 .c 0.6 ::J 0'" 0.4 Q) E 0.2 o... -c: 0.0 f----;:=---=---=:a==.---.Ik-~tr_.....,.r___...-...._- 0 :;:; -0.2 III .~ -0.4 0 -0.6 0 2 4 6 8 10 Time (t)

140 40000

.c l/I 120 ::l 30000 III .c ::l t- iii Cl Q. C 100 ::l 0 al C/) 20000 "0 "0 80 ?: ?: 0; ·iii c c .. .. 10000 Q Q 60

40 0 10 20 Year (starting 1960 for Song Thrush, 1890 for Bupalus)

Fig. 16.1. Above, Two model populations returning to equilibrium density (set at zero) from the same initial displacement at +0.5. One population (open triangles) is much less resilient than the other (closed triangles) and returns more slowly to equilibrium. Below, Extent of year-to-year variabilities for different species: song thrush (open squares) in English farmlands (units of density relative to 100 in 1966) and the moth Bupalus (solid circles) in German forests (numbers per hectare of forest floor)

To complicate matters further, different studies looked at "stabilities" at different levels of ecological organization. Theoretical studies of whether interacting sets of species will be stable consider whether the densities of all those species return to equilibrium or not. To ask how long a community persists before it is invaded is to ask how long the community composition lasts. Such a discussion ignores fluctuations in the abundance of species and looks only at the species list itself. In contrast, McNaughton (1977; Chap. 17) placed his emphasis on how much the total changed and not on which species were involved in the changes. Therefore, we can discuss the stability of populations, community composition, or community biomass. Biodiversity and the Balance of Nature 351

The ecological literature also tried to relate the different definitions of stability to different features of community structure, "complexity" in par­ ticular. Complexity has been taken variously to mean the number of species, the degree of interspecific connectance, and the relative abundance of a species in the community. Controversial though Elton's ideas may be, what he does so effectively is to concentrate our minds on eco~ystem structure, the influence of that structure on the dynamics of the ecological community and its constituent species, and the overriding impact of our species, which is to simplify that structure. What Elton provides is a prescription for research into ecological stability; I will finish by considering some of the recent advances in this area.

16.3 The Stability of Populations

What happens to the dynamics of populations when we simplify the en­ vironment, either by causing species to become extinct, or by replacing complex, natural communities, with simpler, artificial communities such as crops, orchards, or tree farms? Two ways of characterising population stability are resilience and variability (Fig. 16.1) and both may be affected by such simplification.

16.3.1 Resilience: The Example of Pest Outbreaks

A high resilience means a quick return to normal conditions following some . For species that we deem desirable, we equate high resilience with "stability". There is an interesting paradox, however. Consider an insect population increasing toward a high density that involves far more pests and far too few plants for our liking. A species with a resilient population will reach unacceptably high levels faster than a species that is not resilient. Resilience can thus be a measure of instability and also a measure of how often we may have to control the pest by chemical means (Fig. 16.2). In a general way, it is their high resilience that makes insects such pests. Across animal species from aphids to elephants, a small body size is coincident with high population growth rates, and insects are small. The resilience of a population, however, depends on more than just features of the species itself. It depends on the ecosystem in which that species is embedded. When we disturb the densities of a set of interacting species, no species' density can return to its normal value until all the others have done so. The resilience of each of the species in the community will depend on the least resilient one. For example, suppose we reduce the density of 352 S.L. Pimm

Equilibrium

./' Times to reach ./ II) 100000 'uCII equilibrium/ a.CII 10000 II) >- +---- Without predators CII 1000 ...a. "0 100 ~ II) 10 c: CII With predators present 0 1 0 2 4 6 8 10 Time Fig. 16.2. Resilience and simplification. Systems may also be more resilient when they are simplified. Consider, for example, a pest species. Pest outbreaks will occur faster when we simplify the systems (by removing the pest's predators) than they would in the presence of predators. (The final density of the pest will probably be higher as well.) A way to evaluate these ideas may also give some insight into why insecticide applications often induce pest outbreaks. Insecticides clearly kill many pests, but many predators are often killed as well. The predators, moreover, cannot recover until the pests come back. This suggests that there will be a "time window" after insecticide application in which pests have very few predators and so, according to the theory, should increase more rapidly

phytoplankton in a lake. This will reduce the density of their zooplankton predators and, eventually, of the fish that feed on the zooplankton. The phytoplankton might have the capacity to return quickly to their former densities. But while the fish densities are recovering (slowly), zooplankton will be unusually abundant, and this will keep the density of the phyto­ down. Simple systems, therefore, may be more resilient than complex systems. Systems may also be more resilient when they are simplified. Indeed, pest outbreaks occur faster after the pest populations had been treated with an insecticide, because the predators are removed, too, and take longer to recover (Fig. 16.2).

16.3.2 Year-to-Year Variability in Densities

Highly variable species are more likely to occur at high enough densities for us to consider them pests, while for desirable species, a high variability may lead to the species encountering such low densities that they become extinct. Like resilience, variability depends on a number of factors. Population variability will be greater in those ecosystems where critical environmental variables, such as rainfall in deserts, are themselves more variable. Vari- Biodiversity and the Balance of Nature 353 ability depends on resilience, though the details of these relationships are very complicated (Pimm 1982, 1991a). Variability also depends on the struc­ ture of the ecosystem in which the species occurs and is altered when we simplify the system. MacArthur (1955) was one of the first to argue that more complex communities were also more stable. He defined complexity as the number of pathways energy took to reach a given population and stability as the extent to which the population density changed if one of those pathways failed. The argument is no more than the admonition, "don't put all your eggs in one basket." If you drop your only basket, you will break all your eggs; if there is only one energy pathway, the energy supply is highly vulnerable. Watt (1964) produced a diametrically opposite argument. He argued that species with ready access to more resources would be more variable. With the economically important exceptions of species in monocultures (which have access to large areas of the same resources), populations of poly­ phagous species (those feeding on many species) will be more variable than monophagous species (those feeding on a few species scattered among many unpalatable species). Both MacArthur's and Watt's arguments are sketched in Fig. 16.3, and there is field evidence to support both of them (Redfearn and Pimm 1988; Pimm 1991a). What of predator diversity: Increased predator diversity ought to reduce the variability of their prey species, because a diversity of predators provides a more reliable control of their prey's density. A species preyed upon by many species of predator may be much less likely to escape the attentions of those predators than a species exploited by only a very few species of predator. Increased predator diversity may reduce the likelihood of unusually high population densities. There is empirical support for this pattern in both insects and mammals (Pimm 1991a). In summary, the structure of a system affects the population variability of its component species, but in complex ways. The more generalized species are in their diets, the more or less variable they may be. Even with Watt's argument, however, ecological simplification that leads to near monocul­ tures will lead to more variability. The more predatory species a group of prey species suffer, the less variable they will be, so simplification leads to greater variability.

16.4 The Persistence of Communities

How long community composition lasts depends on how fast the community is losing species and how fast it is gaining species. Communities that gain and lose species the most slowly will be the most persistent. We must understand both the processes of extinction and invasion. 354 S.L. Pimm a

b

Fig. 16.3a,b. Two opposing arguments for the possible relationships between population variability and the degree of polyphagy. a MacArthur suggested that monophagous species (left) should be more variable than polyphagous species (right) because they will decrease most when one prey species declines in abundance. b Watt, in contrast, argued that polyphagous species may be more variable because they may be able to reach higher numbers during periods when their predators fail to control their densities. Circles represent species and lines, trophic interactions between species such that species higher on the page feed on species beneath them. The species for which we measure the population variability is shaded. Arrows indicate which trophic interactions are most important in determining the density of the species of interest. Compare two predatory species, one monophagous, the other polyphagous. (These "predators" may be feeding on their plant "prey" without any change in the argument.) In the kind of envisioned by MacArthur, each predator would feed on many different species of prey, with much overlap between predators. In a simple system, the food chains would be more or less distinct. The specialized (monophagous) species is likely to vary more than the more generalized (polyphagous) species. If one prey species of the polyphagous predator becomes rare, then individuals can switch to alternative prey. In contrast, the monophagous predator would become rare if it were deprived of its sole species of prey. In complete contrast, Watt (1964) argued that polyphagous species could be more variable than monophagous species. He suggested that species typically may be kept at levels below those set solely by the availability of their food supply, by the attentions of their predators, parasites, diseases, competitors, or generally inclement weather. Now, suppose that occasionally these various enemies are themselves rare or that the weather, for once, is favourable to the prey species we are considering. What will determine how abundant the species will become before the population declines to a more usual level? Two parameters seem preeminent: How fast the population can increase, i.e., its resilience, and how much of the the species can exploit. What affects how much of the habitat a species can exploit? A monophagous species can only exploit a large proportion of its habitat if the habitat is a monoculture of its food supply. A polyphagous species is likely to be able to exploit a large proportion of its habitat because of its diverse feeding habits. Therefore, populations that reach the highest relative densities will be either monophages in monoculture or polyphages Biodiversity and the Balance of Nature 355

16.4.1 Extinction

Rare species are particularly prone to extinction. One of the principal causes of extinction is that we either totally destroy species' habitats or leave fragments behind which house too few individuals for each localized popu­ lation to persist for long. There are many other factors that affect the rate of extinction, however. For example, species whose population sizes fluctuate greatly should be more likely to become extinct than those that vary little from year-to-year.

16.4.2 Invasions

Why do communities gain species at different rates? Physical features of the environment - such as how isolated the habitat is - play a part in deter­ mining the rate of invasion. Species differ greatly in their individual abilities of reach communities, and many species may fail because they land in places where the environment is unsuitable for them. Even when a few colonists find a potentially suitable environment, they may fail for one of the many reasons why very small populations become extinct. Persistence depends on features of the community as well and is likely to be altered significantly when the community is simplified. Models of how complex communities assemble from simple communities show a number of features (Fig. 16.4). Communities are easier to invade when they have few species, early in the assembly process, and when they have simple patterns of interspecific interactions. Moreover, model communities become pro­ gressively harder to invade as assembly proceeds, even when they are not changing their average numbers of species or connectance. Assembly itself increases community persistence, therefore, "old" communities are harder to invade than "new" ones. When we destroy communities, we substitute communities that are easily invaded by other, often alien species, much as Elton (1958) suggested. Testing to see whether "old" communities are easier to invade than "new" ones is difficult. Ecologists readily accept that disturbed communities are more readily invaded than undisturbed ones. While "disturbed" means several things, one meaning uses "disturbed" to describe communities that are recent assemblages of . In Hawaii, relatively species­ poor upland native (hence "old") bird communities appear harder to invade than the relatively species-rich lowland communities composed of largely introduced species (hence "new" communities; Pimm 1991a). An ubiquitous feature of community assembly models is the existence of alternative persistent states - different combinations of species, each of which is persistent in the presence of the other combinations. In the laboratory, these alternative communities seem very easy to obtain, and there is abundant, if circumstantial, field evidence of their existence (Drake 356 S. L. Pimm b 12

CIl 'u 10 C- CIl ~ g 8 Z

6 --~ 0 50 100 150 200 Total attempted invasions

a 60 (b) CIl CIl 50 " CIl'" 40 £ CIl c: 0 30 'iii co > .= 20 c; Q; .Q 10 E::s z c;20% 0 0 50 100 150 200 Total attempted invasions Fig. 16.4a,b. Sample results from simulations of community assembly as generatted by a recipe that starts with a fixed number of species, then tries to add species to a community. Some species make it into the community, and they may cause other species to become extinct when they do. Many species will fail to enter. a The number of species present increases quickly and then tends to level off as the number of total attempted invasions accumulates. b The number of invasions attempted before one is successful continues to rise. The two sets of results differ in how connected the models are. Models that have a high proportion of their species interacting, for the number of species present (dashed lines, open squares; connectance = 75%), tend to have fewer species but are harder to invade than those models that have a smaller proportion of the possible interspecies interactions (solid lines, solid squares; connectance = 20%). (Pimm 1991a, based on an extensive set of simulations in Post and Pimm 1983)

1990). Another feature of the models is that it may not always be possible to create persistent communities from just their constituent species: other species (catalysts, if you will) that become extinct in the process may be needed. There is a Humpty-Dumpty effect: Even with all the pieces, we cannot put the community back together again. If we are to restore damaged communities, we need to evaluate the importance of alternative Biodiversity and the Balance of Nature 357 persistent states that may prevent us from obtaining the desired persistent state. If there are Humpty-Dumpty effects, then when we destroy com­ munities, we may not be able to reassemble them, even if we have kept all the species alive in botanical gardens and zoos. In sum, simple communities are likely to be less persistent than complex ones, but simplified communities are very unlikely to persist - they are both simple and the product of a short assembly process. Complex communities, the end product of a long assembly, are likely to be the most persistent. We may not be able to reassemble damaged communities. When we try to restore communities, we may end up with another type of community, either because the sequence of the assembly matters or because there may have been species, now extinct, that were essential catalysts in forming the original community.

16.5 Resistance to Change

Given that we are causing extinctions directly and indirectly by deliberately or accidentally introducing alien species, there are obvious questions in­ volving resistance. Which invasions will cause the most damage, and which species in a community are the , that is, those species which, when lost, will cause the most secondary extinctions? Figure 16.5 suggests that complex communities should be most sensitive to the loss of species from the top of the food web because secondary extinctions propagate more widely in complex than in simple communities. In contrast, simple communities should be more sensitive to the loss of plant species than complex communities, because in simple communities the con­ sumers are dependent on only a few species and cannot survive their loss. Similarly, the introduction of trophically generalized species should have profound effects on community composition because such species can eliminate many of the species on which they feed, while the introduction of specialized species should effect fewer changes. There are examples that support these ideas. There were few extinctions, if any, following the loss of chestnut tree from North America, yet many bird extinctions in Hawaii appeared to follow the loss of a few keystone plant species. Presumably, the island communities had simpler food chains, with more trophically specialized species, and so were more vulnerable. Introductions of trophically generalized mammals to oceanic islands where the communities are relatively species-poor have had profound effects in a majority of cases; effects on introductions to continents are less noticeable, probably because there the introduced mammals were controlled by native enemies. In contrast to mammals, bird introductions are rarely reported as having effects. Birds are generally introduced for the purpose of brightening 358 S.L. Pimm

d

Total ~ loss

Fig. 16.5a-d. The effects of species removals depends on the complexity of the food web and from whence the species is removed. a A predator lost from a complex web may release its prey, the herbivore, and so lead to the loss of several species of plant. b In contrast, a plant lost from a more complex web may have little effect, because the herbivore has alternative food sources. c A predator removed from a simple may have no effect (because the herbivore cannot eliminate its only food supply), while d a plant removed from a simple food chain causes the entire food chain to be lost. (Pimm 1991a) our lives, rather than as a source of meat. Consequently, birds are intro­ duced in close proximity to us and thus placed into highly modified, species­ poor habitats where we should not expect them to have noticeable effects (Pimm 1991a). In summary systems differ in how resistant they are to the introduction of alien species or the loss of native species. Resistance depends on whether the system is simple or complex and from whence species are lost or where they are introduced. Often the consequence of the loss of one species is the loss of others.

16.6 Conclusions

Our affects on the environment are complicated but not incomprehensible. Certainly, when we destroy species, the simplified communities may more quickly gain new species. But these species may be undesirable, weedy species, the kind of tramp species that are commensals worldwide. In creating communities that are "new" and not the product of a long process of community assembly, we may create communities that will persist for only short periods of time, before new invasions take place. Invading species Biodiversity and the Balance of Nature 359 may cause the extinction of native species. Communities are often not resistant to extinctions: One extinction may cause a cascade of extinctions, leading to a simple community. When we lose communities, there is a good chance that we will not be able to restore them, even if we have kept all the species in zoos and botanic gardens. Similarly, species have the capacity to recover from losses, and that recovery appears to be faster in simpler systems. Sometimes, we do not want fast recovery (e.g. pests in simple, species-poor agricultural systems). Simplification can also make populations more variable, leading to some species becoming extinct, others becoming pests. The properties of natural systems may sometimes mitigate the damage to which we subject them, but all too often, that damage propagates widely.

References

Christie WJ, Spangler GR (eds) (1987) International symposium on stocks assessment and yield prediction. Can J Fish Aquat Sci 44 (Suppl 2) DeAngelis DL (1991) Dynamics of nutrient cycling and food webs. Chapman and Hall, London Drake JA (1990) Communities as assembled structures: do rules govern pattern? Trends Ecol Evol 5: 159-163 Elton CS (1958) The ecology of invasions by animals and plants. Chapman and Hall, London Loftus DH, Olver CH, Brown EH, Colby PJ, Hartman WL, Schupp DH (1987) Partioning fish yields from the Great Lakes. In: Christie WJ, Spangler GR (eds) International symposium on stocks assessment and yield prediction. Can J Fish Aquat Sci 44 (Suppi 2) pp 417-424 MacArthur, RH (1955) Fluctuations of animal populations and a measure of community stability. Ecology 36: 533-536 May, RM (1974) Stability and complexity in model ecosystems, 2nd edn. Princeton University Press, Princeton McNaughton, SJ (1977) Diversity and stability of ecological communities: a comment on the role of empiricism in ecology. Am Nat 111: 515-525 Pimm, SL (1982) Food webs. Chapman and Hall, London Pimm, SL (1984) The complexity and stability of ecosystems. Nature 307: 321-326 Pimm, SL (1991a) The balance of nature? Ecological issues in the conservation of species and communities. The University of Chicago Press, Chicago Pimm, SL (1991b) Human population growth and ecological integrity. In: Roy OJ, Wynne BE, Old RW (eds) Bioscience and society. John Wiley, Chichester Post, WM, Pimm SL (1983) Community assembly and food web stability. Math Biosci 64: 169-192 Redfearn A, Pimm SL (1988) Population variability and polyphagy in herbivorous insect communities. Ecol Monogr 58: 39-55 Watt KEF (1964) Comments on long-term fluctuations of animal populations and measures of community stability. Can Entomol 96: 1434-1442