Review of Data on Evolutionary Arms Races in Predator-prey Systems

Joseph E. Boxhorn

Copyright © 1995 by Joseph E. Boxhorn   All rights reserved.

Introduction

The rapid development and improvement of military hardware during the Cold War between the capitalist West and the socialist East was described as an arms race. During this period both NATO and the Warsaw Pact designed and deployed several generations of increasingly sophisticated weapons. The alliances often justified building these new munitions by suggesting that they were falling behind in the race. They feared that the penalty for falling too far behind in this interaction was extinction. Thus they stated that they needed the new technologies to keep the relative positions of the competing alliances constant. The leaders also nursed hopes that their blocs might escape the race by pulling far ahead of their competitors and driving them to extinction.

Given this historical context, it isn't surprising that the metaphor of an arms race has been used over the last couple of decades to describe coevolution among interacting species of organisms. It has been suggested that the evolution by one species of a feature that increases the likelihood that the interaction will have a favorable result places a selective pressure favoring countermeasures on other species. The evolution of a countermeasure, in turn, places a selective pressure favoring counter-countermeasures on the first species. A "tit for tat" feedback loop results, leading to progressive evolutionary change within each interacting species. Because individuals who are "too far behind" tend to be eliminated by the interaction, each species evolves increasingly sophisticated adaptations for dealing with the interactions with the others. But because all of the species are evolving in response to the same interaction, the relative position of each remains nearly constant.

Much theoretical attention has been paid to adaptational arms races as two species interactions. It has been asserted that each involved species acts as an agent of selection on the other and that each responds adaptively to the selection imposed by the other. As a result these responses provoke further adaptations in each species. This poses a central question of: can adaptational arms races occur in two species predator-prey systems?

The arms race idea has great theoretical importance. It underlies much current thinking about coevolution. In an influential paper, Janzen (1980) defined coevolution as "an evolutionary change in a trait of the individuals of one population in response to a trait of the individuals of a second population, followed by an evolutionary response by the second population to the change in the first." By this definition, coevolution involves an arms race. An example is given in Thompson 1989.

Since then, the word [coevolution] has been used more consistently as reciprocal evolutionary change in interacting species. The key word is reciprocal, because the central problem in coevolutionary studies is to understand the ecological and genetic conditions that permit interacting species to undergo repeated bouts of reciprocal genetic change specifically because of the interaction.

Clearly, the assumption of the existence and importance of adaptational arms races is implicit in this definition of coevolution.

The validity of this assumption may be the key to resolving a long standing controversy in community ecology. Historically ecologists have argued over two very different perspectives on the nature of biological communities. One, originally derived by drawing an analogy between communities and individual organisms, views communities as whole, discernible entities (Forbes 1887; Clements 1936). In its most extreme statement, communities are regarded as 'superorganisms'. Current statements of this position view communities as coevolved entities. In a coevolved community, arms races and other types of coevolution have shaped the constituent species to have roles within the community analogous t o the roles of organs in the body of an animal. The alternative to this view regards a community as a collection of organisms that happen to be in the same place at the same time (Gleason 1926). The apparent structure of the community is actually an epiphenomenon of the life histories, environmental tolerances and dispersal abilities of the individual species.

Whether adaptational arms races occur has great bearing upon this controversy. Communities can exist as coevolved entities only if arms races occur among interacting species. A finding that adaptational arms races cannot occur or are not important in nature would rule out this perspective.

Definitions

This review examines coevolutionary arms races in predator-prey systems. It will be helpful to identify three types of interactions between organisms that could possibly produce arms races: predation, grazing and parasitism. The definitions of these interactions are taken from Thompson (1982) and, for predators and grazers, are similar to those presented by Williamson (1972). In the following discussion, I use the term lifetime to mean the time it takes an organism to complete the current stage of its life cycle and produce either the next stage or the next generation.

A predator is an organism which uses many "hosts" during its lifetime. It usually kills its hosts, reducing host fitness to zero. It may eat the entire host or only a portion of the host. Examples of predation include an animal killing and eating another animal. Herbivory can also be predatory. Suspension feeding of phytoplankton by zooplankton is predation. Similarly, grass carp feeding on aquatic macrophytes is also predation.

A grazer also uses many hosts during its lifetime. It usually eats only a portion of the host. In contrast to predation, the host usually survives the encounter. The host may or may not experience a lowering of its fitness. A deer browsing on leaves is an example of grazing. A mosquito taking a blood meal from a mammal is another example. A third example is a cichlid grazing on the scales of other fish.

Unlike a predator or grazer, a parasite uses only one host during its lifetime. It may or may not kill its host or substantially affect its host's fitness. One example of a parasite is a trematode worms that is infecting a snail. Lytic and lysogenic bacteriophages are additional examples. Rusts and smuts that infect cultivated plants are further examples.

I would note that these definitions are somewhat different from the usage in much of the literature. They emphasize the nature of the interaction rather than the taxonomic statuses of the interacting organisms. All three of these processes involve organisms using other organisms as a food source. The distinction between predation and grazing has often been drawn on the basis of the identity of the victim. Predation has been said to occur when the victim is an animal or animal-like protist. Grazing has been said to occur when the victim is a plant or plant-like protist. Whether the victim dies or survives has often not been considered relevant to these definitions.

This taxonomically based distinction between predation and grazing obscures something fundamental about the nature of these interactions. The effect of the ecological process on the fitness of the victim may depend on whether the victim normally survives the encounter. Under the above definition, a predatory encounter which is successful from the perspective of the predator may involve either the death or survival of the victim. The same can be said for a grazing encounter. The effect upon the fitness of the victim probably doesn't depend on whether the victim is a plant or animal. The taxonomic definition thus confuses examples of two different processes.

Predation is a different interaction from grazing and parasitism. The reliance of a parasite upon a single individual makes parasite-host systems fundamentally different from predator-prey or grazer-host systems. This is reflected in the close correspondences between the phylogenies of some parasites and their hosts (Brooks 1979; Hafner and Nadler 1988; Demastes and Hafner 1993; but see Page 1993). The looser coupling between predator and their prey and grazers and their hosts presents the potential for these organisms to become channeled genetically into requiring a mixed diet. Predators and grazers also have the potential to learn which victims to eat and which to avoid. These considerations should act to decrease the extent to which particular pairs of species in the latter two interactions change through coevolution as compared to parasites and their hosts (Thompson 1982). From the victim's perspective, predation is a more severe interaction than grazing. An unsuccessful victim may (or may not) suffer a reduction in its fitness due to grazing, but rarely will this reduction take future fitness to zero. An unsuccessful victim will nearly always have its future fitness reduced to zero as a result of predation. Because of this, predation should impose stronger selection upon its victims than grazing.

Literature Review

Arms race hypotheses in predator-prey systems have been examined both theoretically and empirically. Theoretical development has focused on both verbal and mathematical models. Empirical evidence comes from paleontological, field and laboratory studies.

Theoreticians have examined two very different classes of arms race hypotheses: hypotheses which focus on an entire community and hypotheses which focus on two species interactions. The first class is known as the Red Queen Hypothesis (Van Valen 1973). It addresses the coevolution of an entire community. It is based on the notion that an improvement in one species in a community constitutes a degradation of the environment for all the other species. The result of this zero-sum assumption is that only two "coevolutionary stable states" can exist. In the first, all species in the community evolve at the same rate, everyone is running as hard as the can to stay in the same place (Van Valen 1973). It has also been noted that this could lead to a stable state where, in the absence of environmental change, no- one evolves (Maynard Smith 1976). The second class of hypotheses is the simpler case of two interacting species. This will be the focus of this review.

Theoreticians have focused on three broad questions related to the arms race analogy. Can adaptational arms races occur? Can a participant in this type of interaction pull ahead of its opponent or win? How does an adaptational arms race end? Both verbal and mathematical arguments have been made on positions on all three questions.

Can adaptational arms races occur, that is does an adaptive change by one species in an interaction provoke a counter adaptation by the other species, that at least partially offsets the original adaptation? The power of both the arms race metaphor and the conflicting adaptive goals of predator and prey has been viewed as sufficiently obvious that some authors have simply assumed it is true (Gould 1977: Dawkins and Krebs 1979; Roughgarden 1979). Some mathematical results do suggest that arms races occur in predator-prey systems. One model (Schaffer and Rosenzweig 1978) examined how coevolution affects the population dynamics of two interacting species. The conclusion from this work was that the system will go into a "coevolutionary steady state" in which both species will be evolving at the same rate. At steady state, the prey will be abundant enough to deplete its resource. Greenwood (1984) notes that Schaffer and Rosenzweig's conclusion depends on the assumption that any adaptation in one of the species will reduce the absolute success rate of the other species. He argues that data on the effect of clumping by prey as antipredation behavior in several species shows that this assumption isn't universally valid. While clumping by prey can reduce the probability that any individual is predated, it can also act to increase the absolute success of the predator (references given in Greenwood 1984). There is also some evidence supporting the arms race hypothesis (see below).

On the other hand, several authors have argued that the asymmetries in predator-prey interactions are such that arms races either shouldn't be expected to occur or should not last for many rounds. One issue bearing on this is the question of exactly who (or what) is racing against whom (or what)? Dawkins and Krebs (1979) make explicit what many workers assume, that one lineage races against another. Thus a specific predator is the main selective force in predation related adaptations for a specific prey and vice versa. This view has been criticized as misplacing where natural selection acts. Greenwood (1984) argues that individual selection will make the race between genotypes within each lineage and not between lineages. He cites an old joke to illustrate his point.

The importance of this subtle distinction may be illustraed by the story of the two trappers cowering in their tent in Greenland while a polar bear, Ursus maritimus, prowled outside. One, a native Greenlander, began to remove his boots. 'What are you doing?' whispered his Danish companion, 'You know that even without your boots you can't run faster than a polar bear'. 'Yes' replied the Greenlander 'But without my boots I can run faster than you'.

This suggests that competition within a species may have a greater effect on the evolution of the species than interactions between species.

Demonstrating that both species in an interaction evolved in response to the interaction may be problematic (Thompson 1989). This is especially apparent when we consider that improved predation-related abilities of predators and prey should be favored by natural selection independent of changes in the other species (Abrams 1986b). Alternative selection pressures may also prevent one species in an interaction from responding (Vermeij 1982). Most predators feed on more than one prey (Williamson 1972; Futuyma 1979) and most prey face more than one predator. This will tend to lessen the effect that any one opponent will have on the evolution of a species. A final verbal argument states that an increase in prey density caused by an antipredation adaptation might make up for the reduced catchability of the adapted prey (Abrams 1986a). The result of this would be that the predators' food supply would remain unchanged with little selective pressure for countermeasure.

Mathematical arguments have been made in addition to these verbal ones. Abrams (1986a) analyzed coevolution in 24 types of model predator-prey systems which differed in their assumptions on issues such as the shape of the predator functional response, predator and prey densities, multiplicative versus additive combinations of predator and prey investment, etc. Adaptational arms races were possible occurrences in 14 of the model systems, but inevitable in none. In 16 of the systems, prey always increased its investment in the interaction in response to an increase in predator investment. In no system did the predator always increase its investment in response to an increase in prey investment. There were 16 systems in which the predator sometimes increased its investment in response to an increase in prey investment. He concluded that there was a functional asymmetry between predator and prey. Thus theory is equivocal on whether arms races can occur.

Much has been written about whether a participant in an adaptational arms race can pull ahead of its opponent or even win. This could be rephrased as asking whether one side in an arms race has some sort of inherent evolutionary advantage in the interaction, however; it isn't at all clear from the discussion what "pulling ahead" or having an "inherent advantage" actually means. It may be that after a species evolves a predation- related adaptation, it experiences a period of grace before the other species in the interaction develops a countermeasure (Dawkins and Krebs 1979). During this grace period the organism would experience a reduction of adverse effects from the interaction. We could say that if one side consistently enjoys a longer period of grace than the other, it is keeping ahead. But it isn't at all clear what the appropriate units are for measuring this. Should we use an objective unit of time (eg. years)? Should we use generations? If we use generations, how do we account for differences in the generation times of interacting organisms? I'm not sure whether this question has much meaning.

Assuming for the moment that there is some meaning in the question posed above, some authors have suggested that the prey should be able to pull ahead in an arms race with a predator. Dawkins and Krebs (1979) suggest that prey may have an inherent advantage in an arms race due to what they call the "Life-dinner Principle". Paraphrasing Aesop, they suggest that "The rabbit runs faster than the fox, because the rabbit is running for his life while the fox is only running for his dinner". By this they mean that the interaction places stronger selection pressure on the prey because failure results in total loss of fitness while failure for the predator results in a slight loss of fitness. The stronger selection on the prey should keep the prey ahead of the predator in the race. This reasoning has been challenged on a number of points. A predator that is sufficiently poor at catching prey will starve to death (Abrams 1986a). This suggests that the asymmetry in the selection pressures may be less than the life-dinner principle's proponents believe it is. It also ignores the differences in the relative frequencies that predator and prey encounter each other. To survive, an individual predator must encounter and catch prey more often than an individual prey must escape from predators (Greenwood 1984). This 'rare enemy effect' would tend to suggest that the predator may have an advantage. Dawkins (1982) has accepted that the circumstances of the particular case will determine which of these two 'principles' will be stronger.

Three fossil data sets have been examined to determine whether arms races occur in predator-prey systems. None gives an unequivocal answer to the question.

Jerison (1970, 1973) described a trend of increasing brain size in ungulates and carnivores from the Tertiary to the present. He stated that carnivores had larger brains relative to body size than ungulates. He concluded that there was a feedback or selective pressure where increased relative brain size in carnivores led to increased relative brain size in ungulates and vice versa, with the carnivores always remaining a bit ahead. Though he did not use the term, he claimed to have found an evolutionary arms race between predator and prey species. These conclusions are controversial. Both the evidence Jerison cited and the assumptions underlying his conclusions have been challenged. Radinsky (1978) pointed out that the carnivores in each of the assemblages examined by Jerison come from higher stratigraphic levels than the ungulates. Thus the ungulates in the study are older than the carnivores. Jerison also doubled the body weight of ungulate species he considered to be of "heavy habitus". This introduced bias in the direction of ungulates having relatively smaller brains (Gould 1975). Jerison's method for assigning weights to fossils has also been questioned (Holloway 1974). Jerison used the same equation for estimating body weight from body length for both groups, even though the equations used for making these estimates in modern members of these groups have significantly different slopes (Radinsky 1978). Jerison also assumed that larger relative brain size indicates greater intelligence and that greater intelligence confers superior fitness. There is little evidence to support either of these beliefs (Some of the reasoning behind these assumptions is critiqued in Holloway 1966).

In another study, Vermeij (1982, 1983) examined the evolution of predator and prey species in shallow-water marine communities. He concluded that although prey species evolved greater ability to avoid predation, predator species probably did not evolve responses. It has been pointed out that his data set seems to contain less information on predators than on prey and that this may have affected his conclusions (Abrams 1986).

Finally, Bakker (1983) studied leg length and related characteristics in ungulates and their carnivores. On the basis of limb characteristics, the ability of the carnivores to capture their prey appears to have decreased over the last 60 million years. These data do not address whether predators have evolved prey capture adaptations that did not involve limb changes.

These three studies illustrate the difficulties inherent in using paleontological approaches to examine arms race hypotheses. We can study only the material that is preserved in the fossil record -- and the fossil record is subject to preservational biases. Organisms with hard parts are preserved much better than other organisms. But it goes further than this. The fossil record is biased as to which features of organisms we can study. The evidence a fossil can give regarding the morphology of a creature is much clearer than the evidence it can give about the creature's physiology. A change in how efficiently an organism processes its food, for example, may very well be invisible to the fossil record. In addition to this, only some morphological features are accessible in fossils. Though we can often make inferences about soft tissues, we only rarely observe them in the fossil record. This is reflected in the debate about "heavy habitus" ungulates surrounding Jerison's data. Much environmental data is inaccessible in the fossil record, especially at fine scales. Finally, the temporal resolution of the fossil record is too coarse to determine whether a change in one organism followed a change in another or occurred at the same time. Thus we often cannot resolve whether the changes in the organisms were the result of an interaction between the two or responses to a similar set of environmental circumstances. It is likely that the most we will be able to glean from the fossil record is an indication whether adaptive trends in the species of some communities are consistent with Van Valen's Red Queen hypothesis.

I was able to find only one field study examining arms races in predator-prey systems. A few studies have looked at other systems.

There are considerable data examining gene-for-gene coevolutionary relationships in plants and plant parasites. Gene-for-gene coevolution could occur when parasites and hosts have complementary genetic loci for virulence and resistance. This model assumes that each gene for determining virulence in the parasite population has a corresponding gene for determining resistance in the host population. An arms race in a gene-for-gene situation would appear as changes in allele frequencies at these loci in the interacting species. One possible scenario involves the appearance and increase of a particular allele for virulence in the parasite population being followed by an increase in the specific allele for resistance in the host. This would be followed by an increase in the frequency of another virulence allele in the predator population, ad infinitum. Other plausible scenarios exist.

Some of the best evidence for this sort of relationship in natural populations comes from work on interactions between wild oats (Avena spp.) and Puccinia rusts (Burdon, et. al. 1983; Oates, et. al. 1983) and between soybean rust (Phakopsora pachyrhizi) and Glycine canescens (Burdon 1987). Gene-for-gene interactions have been found in studies involving some degree of formal analysis of the genes for resistance and avirulence for at least 12 agricultural plant-pathogen associations. In addition, in at least 29 other plant-pathogen interactions, gene-for-gene relationships have also been postulated based on studies of phenotypic patterns in the hosts and parasites and studies of host genetics (references given in Thompson and Burdon 1992). Some evidence for gene-for-gene relationships has also been reported in grazer-host systems involving plants and phytophagous insects and nematodes. The best example of this occurs in the wheat- Hessian fly system. Twenty genes for resistance to this fly are known in wheat and the fly has 11 biotypes that differ in their ability to attack wheat varieties with these resistance genes (Weller, et al. 1991). It has been argued that these data may exaggerate the frequency of gene-for-gene relationships in nature. The high number of these relationships found in crop plants may reflect agricultural breeding techniques which select for resistant varieties. This may magnify single gene effects (Barrett 1985; Diehl and Bush 1984). Quantitative genetic analyses of plant-pathogen interactions may prove illuminating here, but few have been done.

The studies discussed above suggest that gene-for-gene relationships may be common. It is not at all clear, though, whether they also imply that adaptational arms races are occurring. A recent study of variation in hosts and pathogens in several interacting populations of Linum marginale and Melampsoral lini in Australia suggests that gene-for-gene interactions may not always lead to arms races between the interaction species (Burdon and Jarosz 1991, 1992; Jarosz and Burdon 1991; Burdon et al. 1990). This study found little correlation between the genetic structure of pathogen populations and the genetic structure of the co-occurring host populations (Jarosz and Burdon 1991). Stochastic fluctuations, local extinction of pathogen genotypes, and gene flow among pathogen demes seem to be the major influences on the genetic structure of pathogen populations (Burdon and Jarosz 1991). Changes in the relative frequencies of resistant host lines showed little correlation with changes in the virulence of the associated parasites. Thus, at least at the level of individual populations, the arms-race metaphor does not seem to apply to this system.

A second study also examined arms races in parasite-host systems. Using field data and simulation modeling, Weis et. al. (1989) examined an apparent escalating arms race between a gallmaker fly, Eurosta solidaginis, and its wasp parasitoid, Eurytoma gigantea. Their results suggest that phenotypic plasticity in the parasite can drive genetic change in the host. They use the facts that ovipositor length can limit the size of a gall that a wasp can penetrate for oviposition (Weis and Abrahamson 1985) and that larval nutrition influences the size of the adult wasp (Weis et. al. 1989) to argue that nutritionally based increases in parasite size drive genetic changes in host size. Though their results do not address whether an arms race is actually occurring between these species, they give us a warning to beware of the confounding effects of phenotypic plasticity.

A third field study examines a grazer-host system. The scale-eating cichlid Perissodus microlepis in Lake Tanganyika has a mouth that opens to one side due to an asymmetrical joint of the jaw to the spensorium (Liem and Stewart 1976). In some individuals, the jaw opens to the right and in others it opens to the left. These fish feed on scales of other living fish. Field experiments and analysis of scales in the stomachs of these fish indicate that right handed (dextral) fish always take scales from the left side of their prey and left handed (sinsitral) fish always take scales from the right side of their prey (Hori 1993). The frequency of handedness in the population oscillates about 0.5 with a period of about 5 years. Analysis of denuded spots on the flanks of adult Cyathopharynx furicifer, a common prey fish, suggests that prey preferentially guard against attack from the side that the most common form of the grazer attacks. This imposes selection favoring the rare morph. Analysis of the phenotype ratios of sinsistral to dextral morphs in broods of P. microlepis suggest that handedness is determined by a simple Mendelian one-gene, two-allele system; however, these data are confounded by an unusual aspect of the biology of these fish. P. microlepis farms its fry out to the care of other breeding pairs (Yanagisawa 1985). This makes it impossible relate parent to offspring or to tell if a brood represents the result of a single mating.

The last field study that I'll discuss examines a possible arms race in a predator-prey interaction. Brodie and Brodie (1990) noted that evidence of an arms race between two species should be observable in isolated populations. They argued that if an adaptational arms race is occurring between two species, the level of the one species' abilities relative to the other's should be correlated with the level of the second species' abilities relative to it. Populations of prey organisms with less effective antipredation adaptations should most likely be sympatric with populations of predators that have less effective countermeasures. Similarly, greater resistance to predation in a prey population should be accompanied by more effective countermeasures in the predator population. Brodie and Brodie (1991) claim to have found crude evidence for this sort of correlation in an interaction between the newt Taricha granulosa and its garter snake predator Thamnophis sirtalis. They examined three populations of the snake for resistance to a toxin produced by the newt. Snakes from a population that lives in a region without the newt are very sensitive to tetrodotoxin (TTX). Snakes from a second population show slight resistance to TTX. The newts that co-occur with this population exhibit low levels of TTX toxicity. Snakes from a third population show high levels of TTX resistance. The newts that live in this region produce high levels of TTX. The authors argue that these data provide support for the idea of coevolution of prey defense and predator exploitation ability.

The last two studies illustrate why a field approach alone will probably not answer the question of whether adaptational arms races occur in predator-prey systems. I suspect that the results of most field studies will resemble those of Brodie and Brodie, that is they will be correlational and give us just one point in evolutionary time. Field studies just can't give unequivocal evidence of this type of evolutionary change in real time. They will always contain some ambiguities, both within the data and due to our inability to control for other influences in the field. It is my suspicion that Hori's study represent about the best obtainable from a field approach. But even this work is riddled with ambiguities. Behavioral eccentricities of the study organism prevent a clear determination of the genetic basis of mouth asymmetry in P. microlepis. Because of the difficulties associated with observing attacks in the field, Hori's argument that an individual grazer always attacks its host from the same side is based on gut-content analysis. But Hori could not determine which side of the victim 20% of the scales found in grazer guts came from. Hori's data follows the dynamics of only one population through more than half a complete cycle of one oscillation. Thus we really can't tell whether the results of this interaction reflects a quirk of these populations or a general evolutionary principle. To reduce ambiguity and difficulty, field studies require that the organisms examined should be fairly conspicuous. But this requirement will often mean that the study organisms will take relatively long to complete their life cycles. This means that to document the dynamics of an arms race, the study will either have to run for a long period or document what happens to only a few generations. Thus a real-time examination of an arms race in the field is impractical.

A few experimental studies have examined arms-race hypotheses. Those which looked at predator-prey systems have found that antipredation adaptations can evolve rapidly in prey species. Observing predator counteradaptations has been more elusive.

Shikano et al. (1990) observed a major morphological change in an unidentified gram negative bacterium when grown in semicontinuous culture with a predator. This bacterium is normally an amotile, short (up to 1.5-µm) rod. The bacteria were grown in 200 ml of medium. The ciliate Cyclidium sp. was added to some flasks. Every seven days 1 ml of the culture was transferred to new medium. After 8 to 10 transfers long cells (up to 20 µm) appeared in the cultures with the ciliate. These type-L cells lacked crosswalls. They coexisted with a short morph (type S). They looked at the size distributions of 30 colonies derived from one of their experimental cultures. The frequency distributions of cell lengths from type S colonies were indistinguishable from those of the controls and the parental strain. The frequency distribution of cell lengths of the type-L cell is considerably broader. The fact that daughter colonies derived from type-L colonies show the same distributions of cell length as the type-L colonies suggests that this change in morphology reflects a genetic change. Feeding experiments showed that the ciliate preferentially grazes upon the shorter cells. After the appearance of the long form, the density of ciliates in the experimental flasks declined. No evidence of a counteradaptation by the predator was seen.

Holen (personal communication) produced a similar predator induced shift in Escherichia coli. When grown in a chemostat with the predatory flagellate Poterioochromonas malhamensis a long form of the bacterium appeared within about five days. These filaments were as long as 100 µm and lacked septa. Many were so long that the flagellate could not completely ingest them.

A similar antipredation adaptation also appeared in E. coli grown in continuous culture with the ciliate Tetrahymena thermophila (Nakajima and Kurihara 1994). Long-form cells appeared in these cultures within 15 days of innoculation. Unlike Holen's results, the long form was comprised of chains of normal-sized cells. Few nonseptate filaments were observed. The mean sizes of three clones of these filamentous cells are shown in table 1.

The means for the three clones may underestimate the lengths of the filaments due to cutting of filaments at arbitrary sites during cultivation. This morphological change persisted through several platings on agar. No changes were reported in the predator.

Van den Ende (1973) introduced the ciliate predator Tetrahymena pyriformis into a chemostat containing the bacterium Klebsiella aerogenes in steady state. He observed a number of changes over about 1,000 hours. The densities of both populations went through a damping series of five oscillations. During the period 140 hours to 200 hours following innoculation of the culture with the ciliate, the bacteria's colony morphology changed from normal mucoid appearance to a glassy appearance. This reflected a change in the bacteria from a capsulate to a noncapsulate morphology. Bacteria began to adhere to the walls of the culture vessel at this time. No wall growth was seen in the controls. (After subsequent innoculation of the control with Tetrahymena, wall growth of bacteria began after about 240 hours). These changes occurred while the bacterial density was increasing and approaching the peak of the first oscillation. Van den Ende suggested that the morphological change represents an adaptation that allows the bacteria to utilize a refuge from predation: growth on the vessel wall. He did not state whether the change in bacterial morphology persisted in the absence of the predator. Some changes also occurred in the predator. At ì innoculation the ciliates showed a distribution of lengths ranging from 40-200 µm. After 800 hours few ciliates exceeded 60 µm in length. Van den Ende attributed this size change to the effects of starvation. His data do not address whether this is an effect of growth, phenotypic plasticity or a genetic change.

Boraas (1983) was able to induce multicellularity in Chlorella vulgaris by introducing the flagellate Ochromonas sp. into a chemostat in which the alga was growing. The colonial form appeared within five days (Boraas 1983, Fig. 1). It rapidly came to dominate the culture. Initially, colony size ranged from 4-32 cells. Eventually this stabilized at colonies of 8 cells. When they are produced, these colonies are barely small enough for the flagellate to ingest. After a short period of growth they are too big. Multicellularity in this new strain has persisted in the absence of the predator for over a decade.

The development of resistance of a bacterium to a parasitic bacterium has also been observed. Varon (1979) introduced the parasite Bdellovibrio into a chemostat with the luminescent bacterium Photobacterium leiognathi growing in steady state. Within six days a new strain of the host appeared which was resistant to attack by the parasite. This mutant coexisted in the culture with a form similar to the original strain. Normally P. leiognathi grows as pairs of rod-shaped cells and forms translucent colonies. The mutant strain grew as chains of oval cells and formed opaque colonies. Plaque assays showed that the efficiency of plating of Bdellovibrio suspension on lawns of the mutant was at least 10^-7 times lower than on the original strain or on the wild-type cells from the culture. Examination of mixed suspensions of parasite and host using phase-contrast microscopy showed that wild type cells were attacked immediately upon mixing by Bdellovibrio, while the mutant remained untouched. Batch-culture studies showed that under similar culture conditions, the mutant has a lower growth rate than the wild-type bacterium. The mechanism of resistance to Bdellovibrio is not known. No counteradaptation of Bdellovibrio was observed.

While predator responses to prey adaptations have not been unequivocally observed in laboratory experiments, parasite responses to host adaptations have been. Interactions between E. coli B and several T series bacteriophages growing in chemostats have been studied extensively. E. coli is readily able to develop resistance to five of the seven T phages (Lenski 1984 and references therein). This resistance can show up in as short a time as 20 hours (Horne 1970). While counteradaptations have not been observed to develop by phages T4 (Horne 1970; Lenski and Levin 1985) and T5 (Lenski and Levin 1985), mutants with wider host ranges have appeared for T2 (Lenski and Levin 1985; Paytner and Bungay (1969) also observed the appearance of mutant T2 phage after the development of resistance in E. coli, but did not assay for changes in virulence or host range), T7 (Chao et al. 1977; Lenski and Levin 1985) and possibly T3 (Horne 1970, see discussion in Lenski 1987). In at least one instance, a further round of escalation of defenses occurred. Bacteria developed resistance to the host range mutant of phage T7 (Chao et al. 1977). At first glance, this seems to offer support for the notion of a continuing arms race between E. coli and its phage parasites. The experimenters, though, interpret this as evidence against the idea. Lenski and Levin (1985) note that none of the phage T2, T4, T5 and T7 could be supported by the last resistant bacteria produced. They estimate that if any further host-range mutations exist, they must occur at rates on the order of 10^-11 per virus replication or less. Lenski (1987) argues that these studies give support to the idea that there is a functional asymmetry in the coevolutionary relationship between bacteria and virulent phage. He suggests that this may be due to the phage adsorption sites on the bacterial surface placing structural constraints on the phage adsorption process that are more severe than the physiological constraints they place on resource assimilation by bacteria (Lenski 1984). Thus, there seem to be bacterial mutations conferring resistance for which no (or only extremely rare) viral host-range mutations exist. Thus after a few rounds of escalation of the arms race, bacteria seem able to win by escaping from parasitism.

The studies discussed above give a good summary of the empirical status of arms race hypotheses. Of the three fossil data sets which address predator-prey systems, one seems consistent with Van Valen's Red Queen hypothesis. The other two suggest that prey may pull ahead of predators. None of these three, however, yield clear cut conclusions due to methodological problems and the limitations of the fossil record. Field studies present considerable data suggesting that gene-for-gene relationships may occur among parasites and their hosts, though these findings may instead reflect artifacts introduced by agricultural breeding techniques. In any case, the lack of correspondence between the genetic structure of parasite populations and that of host populations suggests that a gene-for-gene relationship may not necessarily reflect the occurrence of an evolutionary arms race. Other field studies in parasite-host systems and grazer-host systems show an evolutionary response in one organism to changes caused by phenotypic plasticity in another. The only field study that examines predator-prey systems shows a correlation that is suggestive of an arms race, but is hardly definitive. Five laboratory studies show the development antipredation adaptations by a prey organism in real time. In four cases the prey were bacteria, in the fifth the prey was an alga. In none of these cases was the development of predator countermeasures observed, three observations do suggest that predators have the capacity to change evolutionarily in real-time (see below). There is considerable evidence that bacteria can develop antiparasite adaptations in real-time. There is evidence that viral parasites can develop countermeasures, though this evidence suggests that arms races in these systems may not be sustainable for very many rounds.

References

Abrams, P. A. 1986a. Adaptive responses of predators to prey and prey to predators: The failure of the arms-race analogy. Evolution 40:1229-1247.

Abrams, P. A. 1986b. Is predator-prey coevolution an arms race? TREE 1:108-110.

Bakker, R. T. 1983. The deer flees, the wolf pursues: Incongruencies in predator-prey evolution, pp. 350-382. In D. J. Futuyma and M. Slatkin (eds.), Coevolution. Sinauer, Sunderland, Mass.

Barrett, J. 1985. In: Rollinson, D. and R. M. Anderson (eds.) Ecology and genetics of host-parasite interactions. Academic Press, New York.

Boraas, M. E. 1983. Predator-induced algal evolution in chemostat culture. EOS Transactions of the American Geophysical Union. 64:1102.

Brodie, E. D., III and Brodie, E. D., Jr. 1990. Tetrodotoxin resistance in garter snakes: An evolutionary response of predators to dangerous prey. Evolution. 44:651-659.

Brodie, E. D., III and Brodie, E. D., Jr. 1991. Evolutionary response of predators to dangerous prey: Reduction of toxicity of newts and resistance of garter snakes in island populations. Evolution. 45:221-224.

Brooks, D. R. 1979. Testing hypotheses of evolutionary relationships among parasitic helminths: The digeneans of crocodilians. Amer. Zool. 19:192-203.

Burdon, J. J. 1987. Phenotypic and genetic patterns of resistance to the pathogen Phakospora pachyrhizi in populations of Glycine canescens. Oecologia 73:257-267.

Burdon, J. J., A. H. D. Brown and A. M. Jarosz. 1990. The spatial scale of genetic interactions in host-pathogen coevolved systems. In J. J. Burdon and S. R. Leather (eds.) Pests, Pathogens and Plant Communities. pp. 233-247. Blackwell, Oxford.

Burdon, J. J. and A. M. Jarosz. 1991. Host-pathogen interactions in natural populations of Linum marginale and Melampsora lini: I. Patterns of resistance and racial variation in a large host population. Evolution. 45:205-217.

Burdon, J. J. and A. M. Jarosz. 1992. Plant Path. 41:165-179.

Burdon, J. J., J. D. Oates and D. R. Marshall. 1983. Interactions between Avena and Puccinia species. I. The wild hosts: Avena barbatra Pott ex Link, A. Fatua L., A. Ludoviciana Durieu. J. Appl. Ecol. 20:571-584.

Chao, L., B. R. Levin, and F. M. Stewart. 1977. A complex community in a simple habitat: An experimental study with bacteria and phage. Ecology. 58:369-378.

Clements, F. E. 1936. Nature and structure of the climax. J. Ecol. 24:2532-284.

Dawkins, R. 1982. The extended phenotype. Oxford University Press, Oxford.

Dawkins, R. and J. R. Krebs. 1979. Arms races between and within species. Proc. R. Soc. Lond. B. 205:489-511.

Demastes, J. W. and M. S. Hafner. 1993. Cospeciation of pocket gophers (Geomys) and their chewing lice (Geomydoecus). J. Mamm. 74:521-530.

Diehl, D. R. and G. L. Bush. 1984. Ann. Rev. Entomol. 29:471-504.

Forbes, S. A. 1887. The lake as a microcosm. Bulletin of the Peoria Scientific Association, pp. 77-87. Reprinted in the Bulletin of the Illinois State Natural History Survey 15(1925):537-550.

Futuyma, D. J. 1979. Evolutionary Biology. Sinauer, Sunderland, Mass.

Gleason, H. A. 1926. The individualistic concept of the plant association. Bull. of the Torrey Botanical Club. 53:7-26.

Gould, S. J. 1975. A direct assault upon the citadel itself. Paleobiology 1:125-129.

Gould, S. J. 1977. Ever since Darwin. Norton. New York.

Greenwood, J. J. D. 1984. The evolutionary ecology of predation. pp. 233-273 in: B. Shorrocks (ed.) Evolutionary Ecology. Blackwell, Oxford.

Hafner, M. S. and S. A. Nadler. 1988. Phylogenetic trees support the coevolution of parasites and their hosts. Nature. 332:258-259.

Holloway, R. L. 1966. Cranial capacity and neuron number: A critique and proposal. Am. J. Phys. Anthrop. 25:305-314.

Holloway, R. L. 1974. On the meaning of brain size. Science 184:677-679.

Hori, M. 1993. Frequency-dependent natural selection in the handedness of scale-eating cichlid fish. Science. 260:216-219.

Horne, M. T. 1970. Coevolution on Escherichia coli and bacteriophages in chemostat culture. Science. 168:992-993.

Janzen, D. H. 1980. When is it coevolution? Evolution 34:611-612.

Jarosz, A. M. and J. J. Burdon. 1991. Host-pathogen interactions in natural populations of Linum marginale and Melampsora lini: II. Local and regional variation in patterns of resistance and racial structure. Evolution. 45:1618-1627.

Jerison, H. J. 1970. Brain evolution: New light on old principles. Science 170:1224- 1225.

Jerison, H. J. 1973. Evolution of the brain and intelligence. Academic Press, New York.

Lenski, R. E. 1984. Coevolution of bacteria and phage: are there endless cycles of bacterial defenses and phage counterdefenses? J. theor. Biol. 108:319-325.

Lenski, R. E. 1987. Dynamics of interactions between bacteria and virulent bacteriophage. Adv. Microbial. Ecol. 10:1-44.

Lenski, R. E. and B. R. Levin. 1985. Constraints on the coevolution of bacteria and virulent phage: A model, some experiments and predictions for natural communities. Am. Nat. 125:585-602.

Liem, K. F. and D. J. Stewart. 1976. Evolution of scale-eating cichlid fish of Lake Tanganyika: A generic revision with a description of a new species. Bull. Mus. Comp. Sool. 147:319-350.

Maynard Smith, J. 1976. What determines the rate of evolution? Am. Nat. 110:331-338.

Nakajima, T. and Y. Kurihara. 1994. Evolutionary changes of ecological traits of bacterial populations through predator-mediated competition. I. Experimental analysis. Oikos. 71:24-34.

Oates, J. D., J. J. Burdon and J. B. Brouwer. 1983. Interactions between Avena and Puccinia species. II. The pathogens: Puccinia coronata CDA and P. graminis Pers. F. sp. avenae Eriks. & Henn. J. Appl. Ecol. 20:585-596.

Page, R. D. M. 1993. Parasites, phylogeny and cospeciation. Int. J. Parasitol. 23:499-506.

Paynter, M. J. B. and H. R. Bungay III. 1969. Dynamics of coliphage infections. In D. Perlman, (ed.) Fermentation Advances. Academic Press, New York. pp. 323-335.

Radinsky, L. 1978. Evolution of brain size in carnivores and ungulates. Am. Nat. 112:815-831.

Roughgarden, J. 1979. Theory of population genetics and evolutionary ecology: An introduction. Macmillan, New York.

Schaffer, W. M. and M. L. Rosenzweig. 1978. Homage to the Red Queen. I. Coevolution of predators and their victims. Theor. Pop. Biol. 14:135-157.

Shikano, S., L. S. Luckinbill and Y. Kurihara. 1990. Changes of traits in a bacterial population associated with protozoal predation. Microbial Ecology. 20:75-84.

Thompson, J. N. 1982. Interaction and coevolution. Wiley, New York.

Thompson, J. N. 1989. Concepts of coevolution. TREE 4:179-183.

Thompson, J. N. and J. J. Burdon. 1992. Gene-for-gene coevolution between plants and parasites. Nature. 360:121-125.

van den Ende, P. 1973. Predator-prey interactions in continuous culture. Science. 181:562-564.

Van Valen, L. 1973. A new evolutionary law. Evolutionary Theory. 1:1-30.

Varon, M. 1979. Selection of predation-resistance bacteria in continuous culture. Nature. 277:386-388.

Vermeij, G. J. 1982. Unsuccessful predation and evolution. Am. Nat. 120:701-720.

Vermeij, G. J. 1983. Intimate associations and coevolution in the sea, pp. 311-327. In D. J. Futuyma and M. Slatkin (eds.), Coevolution. Sinauer, Sunderland, Mass.

Weis, A. E. and W. G. Abrahamson. 1985. Potential selection pressures on a plant-herbivore interaction. Ecology. 66:1261-1269.

Weis, A. E., K. D. McCrea and W. G. Abrahamson. 1989. Can there be an escalating arms race without coevolution? Implications from a host-parasite simulation. Evolutionary Ecology. 3:361-370.

Weller, S. J., H. W. Ohm, F. L. Petterson, J. E. Foster and P. L. Taylor. 1991. Crop Sci. 31:1163-1168.

Williamson, M. 1972. The analysis of biological populations. Edward Arnold, London.

Yanagisawa, Y. 1985. Parental strategy of the cichlid fish Perissodus microlepis with particular reference to intraspecific brood 'farming out' Environ. Biol. Fishes. 12:241- 249.