2. Locally persistent pest–enemy interactions
3. Locally nonpersistent systems
4. Ecological theory and biological control
Biological control—defined here as the suppression of insect pests by other insects that attack them—has been pursued by entomologists for more than a century, in part because it is typically cheap and can yield very large economic returns on investment. Over this period, the empirical record is one of often spectacular successes mixed with rather more failures. It is also a history of trial and error. In an extreme example, entomologists introduced about 50 species of enemy insects before achieving great success in controlling California red scale on citrus, a case discussed below. Such applied population dynamics has naturally attracted the interest of ecologists who, together with many of the entomologists themselves, have worked to develop theory that would explain the essential features underlying success. This theory and its connection to real biological control comprise the subjects of this chapter. The theory’s domain, however, is much broader—it is the dynamics of interacting resource populations and the consumer populations that attack them.
consumer–resource interactions. These include interactions between populations of predators and prey, parasitoids and hosts, indeed any interaction in which one species depends on another for sustenance. These terms are used interchangeably here.
density-dependent processes. These cause the per-head rate of increase of the population to decrease when the population’s density increases.
equilibrium. Density at which the population will remain, once it is reached, if the population is not perturbed.
parasitoid. An insect that parasitizes another insect (the host species) by laying egg(s) in or on the host, which is eaten by the immature parasitoid.
scale insects. Plant-sucking bugs that stay attached to the plant for almost their entire life history.
stable equilibrium. An equilibrium is stable when the population tends to return to it after the population is perturbed.
unstable equilibrium. The population moves away from the equilibrium following a perturbation. The result may be cycles in abundance, extinction, or chaos, in which the densities are always bounded, but there are no repeated sequences of abundance.
It is useful to distinguish two settings in which biological control occurs. First, most successful biological control has occurred in relatively long-lived agricultural systems such as orchards and, less successfully, forests. It is known for at least some cases that the pest and enemy populations persist together at a small spatial scale such as a single orchard. This is the kind of dynamics for which the bulk of ecological consumer–resource theory has been developed.
In the majority of successes in this class, the pest is an alien species as, almost always, is its successful enemy. Typically, the entomologist faced with an introduced pest travels to its place of origin, searches for enemy species in that environment and, after some preliminary laboratory work, releases the enemy, which usually confines its attacks to the pest species—it is effectively a specialist in the introduced agricultural ecosystem. Again, most consumer–resource theory is developed for specialist consumers.
The second class of control occurs in temporary crops, an environment where success is more elusive, at least in part because enemy species (and the pest) usually do not persist locally. Pest species in these situations may be native or introduced. Less is known about control here; it is likely to involve a number of enemy species, some of which may be introduced and some of whom are likely to be generalist feeders. There is also less relevant theory available.
N. J. Mills has shown that the greatest number (and proportion) of successful cases of biological control are against plant-sucking bugs (Homoptera), for example, scale insects and whiteflies. Caterpillars of moths and butterflies (Lepidoptera) constitute the second largest group, although the number of successes and especially the success rate are lower. The majority of successful control agents are parasitoids, although predatory insects have also been successful.
The central assumption in classical biological control, in which an introduced pest is suppressed by an introduced specialist enemy species, is that the natural enemy controls the pest by maintaining it locally at a low stable equilibrium density. This is because most theory for population dynamics is equilibrium centered, and such a framework provides the most straightforward way to account for long-term persistence of the interaction. Extirpation of the pest is generally thought to be logistically unfeasible, although it has sometimes been achieved, for example in isolated situations on islands.
Successful biological control in this context exemplifies a fortiori a famous problem in predator–prey dynamics—the so-called paradox of enrichment. Basic mathematical predator–prey theory, regardless of its formulation, predicts that the more a predator suppresses its prey below the capacity of the environment to support the prey, the more the interaction will be unstable. Predicted instability takes the form of large-amplitude cycles in population abundance. Successful enemies often suppress the pest well below 1% of the pest’s carrying capacity, so we should see large-amplitude oscillations in density or even population extinction. The dilemma is sharpened by the fact that mechanisms added to models to stabilize the interaction, with few exceptions, cause prey density to increase. Obviously, in successful, persistent, locally stable biological control situations, we see severe pest suppression without instability. This is a general mismatch between theory and observation and is also observed in natural ecosystems.
A number of well-documented cases establish that successful biological control can indeed operate through the creation of a low stable equilibrium, although there have been few efforts to confirm that this is the general case. The best evidence that this is generally true is that later visits to successful instances of biological control frequently find both pest and enemy rare but present. Scale insects (which are plant-sucking bugs that stay attached to the plant for almost their entire life history) have experienced many successful biological control efforts; a large fraction of efforts in this group have been successful, and they show the best evidence for control via establishment of a stable equilibrium. One thoroughly studied case, described below, shows a nice match with theory.
Insects may reproduce only once, and have only one generation per year (so generations do not overlap), or they breed more than once and have multiple, often overlapping, generations per year. The illustrative case discussed below is in this category. It is a Homopteran (the group showing greatest success), and within that is an armored scale, a group that alone accounts for two-thirds of the biological control cases considered to be complete successes. The enemy is a specialist parasitoid, a feature that greatly increases its chances of success. This example thus represents a large fraction of, and many aspects that typify, successful biological control.
California red scale (Aonidiella aurantii), a worldwide pest of citrus, was accidentally introduced into California about 100 years ago from China via Australia. When sufficiently abundant, it can kill trees and, on several occasions in the last century, almost destroyed the California citrus industry. It has spread to most citrus-growing areas of the world.
Among the many natural enemies introduced for control, three were in the genus Aphytis. These are tiny (less than 1 mm long) parasitic wasps (parasitoids). The female (males do not kill hosts) lays eggs in most of the juvenile scale stages, feeds on the smaller stages to get nutrients for maintenance and future egg production, and lays male eggs on smaller and female eggs on larger immature scale hosts. Adult scale are not attacked. The scale passes through two or three generations per year, and Aphytis has about three times as many generations as does the scale. The pest and parasitoid generations are overlapping, so that different life stages coexist at the same time. As a result, vulnerable stages of the pest are present when adult parasitoids are present and searching. This parasitoid genus has been used successfully throughout the world on many different pests of citrus and other crops.
In California, one Aphytis species was present for most of the twentieth century and apparently had little effect on scale densities. Aphytis lingnanensis was introduced in the late 1940s and reduced scale densities but not quite enough for economic control. Finally, Aphytis melinus was introduced around 1959. It was spread throughout much of southern California, immediately both reduced scale densities below the economic threshold and displaced lingnanensis, and has remained a spectacular success for almost 50 years except when sprayed, for example by DDT to which the scale were resistant. All aspects of this history are satisfactorily explained by ecological theory, and indeed, this example provides one of the strongest demonstrations that consumer–resource theory can explain population dynamics in the field.
Over about 20 years, W. Murdoch and colleagues showed that none of the existing notions that had been proposed to explain stability and control in general applied in this system. Field observations and experiments showed stability was not caused by parasitoids concentrating their attacks in a small fraction of hosts or by the existence of a spatial refuge from the parasitoid. Ecologists have come to suspect that spatial processes are frequently the key to the stability of field populations, but these workers also established that spatial processes, i.e., movement of pests or parasitoids among trees, played no role in control and stability. Thus, the mechanisms of control and stability operated on a local spatial scale.
Finally, these workers developed mathematical “stage-structured” models to explore and test the possibility that control and stability are explained by life-history features of the pest and how the parasitoid responds to them. Such models are designed precisely for organisms such as insects that have distinct life stages (e.g., egg, larvae of different sizes, adult) and that reproduce more or less continuously so that generations overlap. They are written in the form of delay-differential equations and keep track of the numbers entering and leaving each pest and parasitoid stage, of losses in each stage to “background” deaths and those imposed on the pest by the parasitoid, and of reproduction of pests surviving to be adults. Equilibrial densities can be calculated analytically, as can the local stability behavior of these equilibria. Dynamics following large perturbations of density far from equilibrium are investigated by computer simulation guided by results from local stability analysis.
A detailed stage-structured model of the red scale–Aphytis interaction was developed in which the main stabilizing mechanisms are the invulnerable adult stage of scale, a potentially powerful stabilizing force, and the fact that the parasitoid goes through three generations in the time it takes the pest to pass through one generation. This model was independently parameterized and was then able to predict with astonishing accuracy the control by Aphytis of experimental scale outbreaks on individual trees in a lemon orchard. Control by the resident Aphytis population in each tree was very rapid—the scale population was effectively brought under control in approximately a single scale generation.
This case speaks to the larger issue in ecological theory raised above, namely the almost universal tradeoff in models between stability and prey suppression: almost all mechanisms that stabilize the predator–prey equilibrium also increase the prey equilibrium. The model shows that stability in this case hinges on two main mechanisms: the invulnerable adult stage of the pest and the much shorter development time of the parasitoid relative to that of the pest. An invulnerable class of prey involves the stability–suppression tradeoff. But when the invulnerable stage is a highly fecund adult stage, it requires relatively few adults to maintain the interaction, so overall suppression is consistent with stability. The second mechanism, unlike virtually all other stabilizing processes, does not involve a tradeoff: stability is more likely, and pest suppression is stronger the shorter the parasitoid’s development (or generation) time is in comparison with that of the pest. We return to these properties below.
The Aphytis – red scale model is also able to explain the subtle life-history features that account for the rapid displacement of Aphytis lingnanensis by melinus, itself a classic example of “competitive displacement.” Aphytis melinus is able to produce female offspring from smaller immature hosts than is lingnanensis, and the model shows that this leads to increased suppression of scale and rapid displacement of lingnanensis. There is thus quite strong evidence that this model contains the essential features of the pest–parasitoid interaction.
The features of Aphytis’s control of red scale appear to be shared by many other successful cases of biological control. As noted above, of the two large groups of pest insects, the greatest number of successes and the greatest rate of success per unit effort are in the Homoptera (plant-sucking bugs); efforts on Lepidoptera have been much less successful. Mills reviewed many cases of biological control and showed that the Homoptera typically reproduce continuously, have several overlapping generations a year, and their parasitoids have markedly shorter generation times than the pest species. Lepidoptera, by contrast, typically have fewer, discrete, nonoverlapping generations per year, and the generations of their parasitoids are also discrete and synchronized with those of the pest, so they have equal development times. Mills showed, furthermore, that, among Homopteran control efforts, short parasitoid generation times were more prevalent among successes than among failures. An invulnerable adult stage is the norm in parasitoid–host relationships.
One or Multiple Enemy Species?
As noted, the displacement of lingnanensis by melinus illustrates ecology’s classical competition theory. Where there is a resource population maintained by a consumer at a stable equilibrium, theory states that that consumer species will win that suppresses the resource to such a low density that it cannot support less efficient, competing, consumer species. This is how melinus displaced lingnanensis in both theory and reality. Mills notes that there have now been numerous documented examples of competitive displacement, leading in all cases to improved biological control. (As an aside, exotic ladybugs and other enemies introduced for pest control have also been shown in some cases to competitively displace native species in natural habitats.)
Theory also says, however, that competing consumers can coexist if they interfere with their own increase more than they do with that of their competitors (here there is another, analogous trade-off between pest suppression and coexistence of enemy species). Such coexistence would not seem, a priori, to be a recipe for successful biological control, and arguments have raged for decades over releasing the most efficient single enemy species (if such can be found) versus releasing multiple species. Of course, if the “best” species wins, the issue is moot. But it is possible in theory to release an enemy that wins by, for example, directly attacking the otherwise most efficient enemy and thereby inducing a higher pest density. Whether this is in reality an important issue is not clear.
In fact, although Mills found, in biological control systems apparently near equilibrium over long periods, that a single, successful, enemy species was a frequent occurrence, it is also common to observe several species of natural enemies coexisting. Obviously there are differences among such species, such as exploiting the pest on different parts of the plant, that may explain their coexistence. Typically, however, we do not know if the inevitable differences among species actually explain coexistence, nor do we typically know the roles played by the different species, if any, in control. The presence of “extra” generalist predators (which attack species other than the pest) does not need explanation because they can coexist by being supported by other resources. More vexing is the coexistence of effectively specialist parasitoids attacking the same pest.
In the red scale example, in addition to Aphytis, there is usually both one rare generalist predator and one rare specialist parasitoid. The above experiment and model established that neither was important in control of red scale, and there is evidence that a single species also dominates in other similar examples of control of scale. Thus, the mere presence of multiple enemy species does not imply they are all necessary or even useful for control. By contrast, long-running control of olive scale in northern California was known to require two specialist parasitoids, each of which was effective in a different season. As in ecology in general, we do not yet have a complete explanation for, or explication of the roles of, competitors coexisting at equilibrium.
Mills (2006a) examined the record of hundreds of efforts to introduce natural enemies. The overall message is that as more enemy species are introduced, a smaller proportion is established—and by far the most common outcome is that only one enemy species is established. On the other hand, the fraction of introductions that led to successful control was independent of the number of enemy species established.
Much equilibrium-centered theory for biological control, in contrast to the systems discussed above, is written for a species whose adults are present and breed in a single episode at one time of year—they are discrete breeders, and the theory is formulated in discrete time (difference equations). Insects with such a life history are typically found at higher latitudes. Unfortunately, there is no deeply studied example that explores experimentally the mechanisms of control and applicability of the theory. Recent work by D. N. Kimberling suggests that success rates are lower in such systems.
Perhaps the best example is the winter moth in North America (mainly Nova Scotia and British Columbia), where it was a severe pest of hardwood forests and urban shade trees. This Lepidopteran was introduced from England, where it was studied in some detail. Of several parasitoid species introduced to control the moth, a fly (Cyzenis) and a wasp (Agrypon) parasitoid became established. The system has been most thoroughly studied in British Columbia, where J. Roland and D. G. Embree have shown, interestingly, that although initial suppression of the pest appears to have been induced by the parasitoid Cyzenis, maintenance of persistent low populations appears to rely on high and density-dependent mortality induced by native predators, mainly beetles, attacking pupae in the soil. Both the moth and fly populations in British Columbia have persisted at low densities for around 20 years except for small, local, and short-lived increases in the moth.
If control of these winter moth populations is indeed induced by the generalist beetle predators, there is likely to be no trade-off between prey suppression and stability. Theory does suggest, however, that unless the pest is strongly preferred by the generalist predators, there is an omnipresent danger that favorable conditions for the pest will allow it to escape their control.
An interesting situation, intermediate between those above and the locally nonpersistent system discussed next, is one in which the pest may be well regulated at a spatial scale larger than the local level, at which there is instability or extinction. This is the ecological notion of metapopulation dynamics. Apparent examples include the famous cottony-cushion scale controlled by the predatory beetle Rodolia. Both species were introduced to California citrus more than a century ago, and successful control continues to this day. Rodolia is seen to drive the pest density to zero on a spatial scale of at least individual trees, but movements among different trees or small groups of trees have allowed both species to persist. A similar pattern has been seen in some mite populations in apple orchards, and some greenhouse pests.
Many pest–enemy interactions in agriculture appear not to persist at the local spatial scale of interest—a tract of forest, an orchard, a field, or a rice paddy. This is true especially in seasonal field crops, where the pest and its enemies do not persist even in a local collection of crop units. Instead, regional persistence requires that the pest and/or the natural enemy population invade from some other habitat each growing season, and the pest may or may not be kept below the density at which it causes economic damage before harvesting, tilling, etc. drive it locally extinct or to very low numbers.
We would not expect the insights of classical consumer–resource theory from the previous section—in particular the efficacy of a dominant specialist enemy species—to be applicable where local dynamics does not fit the equilibrium paradigm and the pest is known to arrive at the crop from “elsewhere.” Because an enemy cannot rely on the pest as a year-round resource, we expect generalist predators to be important. Because the pest is mobile, we expect enemy population movements to be a central feature of successful control. Because local dynamics does not go to a persistent equilibrium state, multiple species of enemies are likely to persist, and local competitive exclusion will not occur. Because local extinction of the pest is consistent with global control and persistence of generalist enemies, an invulnerable pest stage is not a requirement for persistence and may interfere with control. These expectations are borne out in the rather few cases where we know biological control is successful and where mechanisms, or at least dynamics, have been investigated.
Aphid pests exemplify mobile and often nonpersistent pests and, probably as a result, provide few examples of successful biological control. There is, however, one well-studied case in a temporary crop, and it is consistent with the above expectations. W. E. Snyder and A. R. Ives have studied pea aphid (Acyrthrosiphon pisum), an introduced pest in alfalfa fields in Wisconsin, which is under control by a range of natural enemies. The enemies include an introduced specialist parasitoid, Aphidius ervi, and a wide range of mainly native generalist predators, including Nabis and Orius (bug species), ladybirds, and carabid beetles. As expected, the specialist parasitoid is relatively ineffective and is unable to control the aphid on its own. Generalist predators are essential for control, although in some instances they probably decrease the effectiveness of the specialist parasitoid.
The generalists’ adult densities are determined mainly in other habitats, where they feed on other prey. The predators exert control by moving into alfalfa fields and both feeding there and producing predatory larvae, even though the latter may not even complete their development before harvest. Most of the generalists probably eat all stages of aphids—there is no invulnerable stage. In alfalfa fields, however, the predator–prey interactions are not self-maintaining— the enemies exist in the crop only because of processes at a much larger scale. Short generation time is no longer relevant, and indeed, the generalists have much longer development times than the prey. L. E. Ehler has uncovered a remarkably similar story, also in alfalfa fields, but this time in the control of an introduced moth (the caterpillar is the beet armyworm) in California.
These examples appear to illustrate a general situation. Thus, it has long been known that multiple enemies coexist in a number of successful cases of pest control in temporary crops. Indeed, historically, entomologists working in temporary crops have often maintained populations of generalists through various cultural practices, such as interplanting and judiciously timed cutting of noncrop vegetation as a replacement for spatially dispersed other habitats.
There is extensive ecological theory that incorporates spatial processes into models of consumer–resource dynamics. There is also some theory for generalist predators. There is, however, little theory that combines these two features of temporary crop systems, and, so far, biological control in such crop systems has not provided the same rich opportunity for testing general ecological theory. As the work on aphids described above illustrates, however, there is reason to be sanguine about future progress in this area.
Successful biological control of pest insects, especially in long-lived crops, has been a fertile field for ecological theorists. It has been a substantial inspiration for much, ever-more-sophisticated consumer–resource theory that has been shown to be powerful in explaining the dynamics of many natural and nonagricultural systems. Great unexploited opportunities remain for exploring and testing dynamic theory because successful control occurs in simple and species-sparse systems—features that facilitate experimental manipulation. Although these are nonnatural systems, some of them still exhibit, for example, dynamic stability, which is a central ecological phenomenon to be explained.
Benefit has flowed less conspicuously in the opposite direction: biological control has remained a largely empirical, trial-and-error process. This is to some extent inevitable—whether an enemy species will succeed or fail must depend to some extent on local and particular contingencies. Nonetheless, as we saw above, potentially useful guidelines have emerged, and the recent analyses by Mills and others of large sets of historical cases have established that these guidelines apply in a wide range of real pest control cases. Further probing along these lines is likely to be rewarding.
Ehler, L. E. 2007. Impact of native predators and parasites on Spodoptera exigua, an introduced pest of alfalfa hay in northern California. Biological Control 52: 323–338. Establishes that control of this pest is by native generalist predators.
Kimberling, D. N. 2004. Lessons from history: Predicting successes and risks of intentional introductions for arthropod biological control. Biological Invasions 6: 301–318. A statistical analysis of the factors affecting success and failure of biological control using information from historical examples.
Mills, N. J. 2006a. Interspecific competition among natural enemies and single versus multiple introductions in biological control. In J. Broeder and G. Boivin, eds. Trophic and Guild Interactions in Biological Control. Berlin: Springer, 191–220. An analysis of many historical cases of biological control.
Mills, N. J. 2006b. Accounting for differential success in the biological control of homopteran and lepidopteran pests. New Zealand Journal of Ecology 30: 61–72. An analysis of the life-history features that influence success in biological control, using numerous case studies.
Murdoch, W. W., C. J. Briggs, and R. M. Nisbet. 2003. Consumer–Resource Dynamics. Princeton, NJ: Princeton University Press.
Murdoch, W. W., C. J. Briggs, and S. Swarbrick. 2005. Host suppression and stability in a parasitoid–host system: Experimental demonstration. Science 309: 610–613. Presents the key experiment and model establishing the mechanisms causing control and stability in the red scale example.
Snyder, W. E., and A. R. Ives. 2003. Interactions between specialist and generalist natural enemies: Parasitoids, predators, and pea aphid biocontrol. Ecology 84: 91–107.