Parasites and Sexual Selection: a Macroevolutionary Perspective

University of Nebraska - Lincoln DigitalCommons@University of Nebraska - Lincoln

Faculty Publications from the Harold W. Manter Laboratory of Parasitology Parasitology, Harold W. Manter Laboratory of

9-1991

Parasites and Sexual Selection: A Macroevolutionary Perspective

Deborah A. McLennan University of Toronto

Daniel R. Brooks University of Toronto, [email protected]

Follow this and additional works at: https://digitalcommons.unl.edu/parasitologyfacpubs

Part of the Parasitology Commons

McLennan, Deborah A. and Brooks, Daniel R., "Parasites and Sexual Selection: A Macroevolutionary Perspective" (1991). Faculty Publications from the Harold W. Manter Laboratory of Parasitology. 250. https://digitalcommons.unl.edu/parasitologyfacpubs/250

This Article is brought to you for free and open access by the Parasitology, Harold W. Manter Laboratory of at DigitalCommons@University of Nebraska - Lincoln. It has been accepted for inclusion in Faculty Publications from the Harold W. Manter Laboratory of Parasitology by an authorized administrator of DigitalCommons@University of Nebraska - Lincoln. VOLUME 66, No.3 SEPTEMBER 1991 THE QUARTERLY REVIEW of BIOLOGY

PARASITES AND SEXUAL SELECTION: A MACROEVOLUTIONARY PERSPECTIVE

DEBORAH A. McLENNAN AND DANIEL R. BROOKS Department ofZoology, University of Toronto Toronto, Ontario, Canada M5S lAl

ABSTRACT The Hamilton-Zuk hypothesis postulates a causal link between parasitism and theevolution of epigamic traits by intersexual selection. Oversimplified assumptions aboutbasic parasite biology, ambiguous formulation of thehypothesis, andpoorcommunication between ethologists andparasi tologists havehampered its testing. The hypothesis is supported at themicroevolutionary level iffe males showsignificantpreferencefor lightlyoruninfected males, if intensityof infection reflects host resistance toparasites thatdepress hostfitnessbycausingdisease, andif intensity ofinfectionisrelated tothedegree ofepigamic development. It must beshownthatparticularparasites cause disease, that thehostpopulation is polymorphic for resistance to infection by those species, and thatfemale hosts are capable ofdistinguishing malehosts with lowparasite loads duetoheritable aspects ofhostresis tancefrom males thatare uninfected duetochance. The macroevolutionary prediction ofthehypothe sis, thatspecies displaying strongly developed epigamic characters shouldhost"more parasites" than species with weaklydeveloped epigamic traits, contradicts themicroevolutionary dynamic of thehy pothesis, and is too ambiguous. * propose a macroevolutionary prediction based on understanding theevolutionary origin of epigamic traitsand theevolutionary origin of each host-parasite associa tion. Associations originating in the ancestor in which theepigamic traitappeared corroborate the hypothesis moststrongly; those originating prior to the evolution ofthe epigamic trait corroborate it weakly; those beginning aftertheoriginof theepigamic traitcouldnot havebeen involved in the originand spread of the epigamic trait.

INTRODUCTION cited in Ruse, 1979: 209). Since that initial dis pute, numerous researchers have demonstrated HE INFLUENCE offemale choice on the that intersexual selection is an important com Tevolution of male epigamic characters has ponent in the mating system of a variety ofspe been a controversial issue since Darwin first cies (see references in Bradbury and Anders proposed his theory of intersexual selection son, 1987). After establishing the existence of (Darwin, 1871). Indeed, in a letter to Wallace, such a mechanism, these researchers then who had argued against the existence of female turned their attention to the question of how choice, Darwin concluded"we shall never con that mechanism operated. Fisher (1930) pro vince each other" (Darwin and Seward, 1903; posed that the development of the male charac-

255 256 THE QUARTERLY REVIEW OF BIOLOGY VOLUME 66 ter and female preference for the character fense and characters involved in host defense, would "advance together" with ever increasing produces cycles of coadaptation. This tem speed until the process was checked by "severe porally varying selection pressure maintains counterselection" against males bearing the a significant level of additive genetic variabil exaggerated trait. At the very heart of the ity for viability in the host population (Hamil Fisherian hypothesis lies the assumption that ton, 1982; Eshel and Aiken, 1983; Eshel and the relationship between female choice and the Hamilton, 1984; Anderson and May, 1985). preferred male character is arbitrary in terms Second, parasites adversely affect the health ofmale viability. That is, females who choose (viability) oftheir hosts. Third, the condition males with bigger tails, brighter colors, or more of male epigamic characters is an accurate and vigorous displays get nothing out ofthe inter direct reflection of the bearer's health (a "re action other than the production of male off vealing handicap;' Maynard Smith, 1985). spring with the preferred character (see, e.g., Consequently, the interaction between the os O'Donald, 1962, 1967; Lande 1981; Kirk cillating force of natural selection on host via patrick, 1982, 1986). The generality of the bility and female choice for a male character Fisherian process was challenged by propo that accurately communicates underlying via nents ofthe "good genes" school (e.g., Trivers, bility results in a form of directional intersex 1972; Zahavi, 1975, 1977; Borgia, 1979; An ual selection that does not exhaust genetic vari dersson, 1982, 1986; Nur and Hasson, 1983; ability in the population (see also Tomlinson, Kodric-Brown and Brown, 1984) who argued 1988). that there is an association between expression Kirkpatrick (1986) investigated the theoret of the epigamic character and male viability. ical implications of the Hamilton-Zukhypoth Because of this association, females who choose esis by performing a series of444 simulations mates with bigger tails, brighter colors, or more based on 29 sets of parameter values, includ vigorous displays also get mates with "better ing viabilities of genotypes, selection coeffi genes;' and thus produce offspring who pos cients, and recombination rates. He concluded sess the preferred character and are more vig that the mechanism, on its own, was not suffi orous than offspring of nonpreferred males. cient to guarantee the spread of alleles for the Unlike the Fisherian mechanism, then, the male character and female preference, partic "good genes" hypothesis unites superiority in ularly when both are rare in the population. attracting mates and superiority in survival He noted, however, that once these alleles, in ability into a mutually reinforcing force of in fluenced by other evolutionary forces such as tersexual selection. drift or selection on the pleiotropic effects of The chief objection to the good-genes hy preference, had reached some threshold fre pothesis comes from population genetics. Iffe quency, then the effects of the Hamilton-Zuk male choice of genetically "superior" males is mechanism could cause both the preference a strong enough directional force to influence and the male character to spread in a popula the evolutionary elaboration of the character tion. In other words, the interaction between advertising this superiority, how could genetic hosts, parasites and the origin offemale prefer variability, the materials of selection, be main ence will not initiate the process of intersexual tained in the population (Maynard Smith, selection, but may reinforce that process once 1978; Borgia, 1979; Taylor and Williams, 1982; it has begun. Pomiankowski (1987a,b) reexam Kirkpatrick, 1986)? Hamilton (1982)and Ham ined the problem by incorporating more com ilton and Zuk (1982) attempted to answer this plex viability assumptions into his model. objection by exploring the relationship among U sing this new set of parameters, he was able parasite fitness, host fitness, and the degree of to demonstrate that an exaggeration of the male host sexual dimorphism (see also Freeland, character and the female mating preference 1976; Zuk, 1984). Their proposal, based upon could occur if the development of the epigamic studies of parasite population structure, rests trait was a direct revealer of viability. Although upon three assumptions. First, parasites and he disagreed with Kirkpatrick about the gen their hosts are involved in a microevolution eral importance of the Hamilton and Zuk ary arms race. That is, the time lag between mechanism, he also concluded that "The'hand changes in characters involved in parasite of- icap principle' by itselfcannot initiate increases SEPTEMBER 1991 PARASITES AND SEXUAL SELECTION 257 in female preference when the handicap is rare. hosts and parasites, and the details of sexual It only works when a threshold value of fe selection (see also Endler and Lyles, 1989). male preference is exceeded, and Fisher's feed Although researchers disagree about the back process operates.... Both Zahavi's hand generality of this mechanism, they are united icap and the revealing handicap act as one-way in their proclaimed interest in the "phylogenetic directional forces which, in concert with Fisher's effects" revealed by Read (1987), Read and Har feedback process, lead to the runaway exagger vey (1989a,b), and Read and Weary (1990) (see, ation of male secondary sexual characters." e.g., Cox, 1989; Hamilton and Zuk, 1989; Zuk, After uncovering a potential role for the 1989). We share in this enthusiasm, and in this Hamilton-Zuk mechanism on theoretical paper we will show how the Hamilton-Zuk hy grounds, Kirkpatrick (1986) called for the reso pothesis can be examined from this phylo lution of questions surrounding the importance genetic perspective. Rather than attempting of such a mechanism in natural populations. to identify and remove phylogenetic effects so Researchers have sought such a resolution on they do not confound statistical analyses (e.g., both a micro- and macroevolutionary level of Clutton-Brock and Harvey, 1984; Cheverud et analysis. Microevolutionary studies investigate aI., 1985; Felsenstein, 1985; Pagel and Harvey, the effects ofparasite prevalence and intensity 1988; Gittleman, 1989; Gittleman and Kot, on the development of a male character and 1990), we will concentrate on the identification resultant female choice within one species of of these effects in order to incorporate them into host. Results of such studies have been equivo the formulation of the Hamilton-Zuk hypoth cal and attest to the difficulties involved in as esis. We will begin with a discussion of the as sessing the variables influencing parasite-host sumptions about parasite biology that under and male-female interactions (Table 1). Mac lie the hypothesis, and move from there to a roevolutionary studies focus upon comparisons macroevolutionary examination of the origin of hosts across taxonomic levels. Hamilton and and interaction of host-parasite associations Zuk (1982) analysed 108 passerine species in and epigamic characters. Ultimately, a fusion their search for a positive correlation between of both the microevolutionary and phylogenetic number ofblood parasites and degree of male approaches, based on empirical input from host "showiness" (see Ward, 1988 for a similar parasitology and ethology, will be required to study with freshwater fishes). Their discovery produce a more rigorous set of predictions, and of a weak but significant association was at first from this a more robust theory of the relation corroborated by Read (1987, 1988) and later ship between parasitism and intersexual selec challenged by Read and Harvey (1989a,b) and tion will be developed. We hope to demonstrate Read and Weary (1990), who cautioned that herein that there is enough overlap in these per data from related taxa are not independent, so spectives to initiate mutually beneficial cross persistent phylogenetic effects will confound communication. statistical tests of the hypothesis. When the effects of phylogeny were controlled in the Hamilton-Zuk database, the strength of the QUESTIONS ABOUT BASIC BIOLOGY correlation between parasite prevalence and The term "parasite" encompasses a variety male showiness was discovered to vary among of living organisms, running the gamut of different avian groups. Interestingly, this re diversity from viruses through bacteria to in sult mirrors the conclusions drawn by Kirk vertebrates and plants (Price, 1980). Like fish, patrick and the researchers working at the parasites are not a monophyletic group, and microevolutionary level: although the mecha hence cannot be diagnosed by the presence of nism proposed by Hamilton and Zuk may not characters shared by all members of the group. be generally applicable to all sexually selected No ichthyologist, however, would ever equate systems, it may be im portant in specific cases. a salmon morphologically, ecologically, phys The importance, in turn, is dependent upon iologically, or behaviorally with a pipefish, nor a variety of factors, including the life-history would any biologist treat all fishes as inter characteristics and population structure of the changeable entities. Unfortunately, the recog parasites in question, the biology of host resis nition of species individuality accorded to most tance, the coevolutionary interactions between free-living organisms has rarely been given to 258 THE QUARTERLY REVIEW OF BIOLOGY VOLUME 66

TABLE 1 Results of microevolutionary studies investigating the relationships amongparasite intensity, development of male character andfemale choice based upon that character* System Female choice Experimental results

Field crickets/gregarines Preferentially attracted to No correlation between level of gregarine (Zuk, 1987a,b; 1988) older males; older males are infection and number of females attracted; less heavily parasitized than however, there was a negative correlation younger males. between level of infection and spermatophore production that may influence a male's ability to successfully fertilize his mate. Satin bowerbirds/feather Preferentially attracted to Intensity of infection negatively correlated with lice (Borgia, 1986; males with bowers; choose male age. Bower holders had fewer parasites Borgia and Collins, among bower holders based than nonbower holders; within the bower holders 1989) upon quality and (the breeding population) there was a difference decoration of bowers and in intensity of infection, positively correlated there is a correlation with mating success in 1 of 3 years (in the other between male dominance/ 2 years there was no difference among bower aggression and bower holders in either degree of infection or mating quality. success based upon degree of infection). Intensity of infection had no effect on male condition, survivorship, bower quality or plumage brightness. Guppies/nematodes Females preferentially Among males matched for size, tail length and (Kennedy et al., 1987) attracted to males based color, there is a strong negative correlation upon a variety of traits between intensity of infection and (1) male such as tail length, color display rate and (2) the time a female spends and display frequency. responding to male courtship (one indicator of female choice). Fence lizards/malaria Not examined. Among males matched for size, and thus age, (Schall et al., 1982; infected males displayed more ventral black Schall and Dearing, coloration, spent less time engaged in social 1987; Schall and Sarni, activity, and were more likely to be subordinate 1987; Ressel and Schall, than uninfected males. The authors estimated 1989) that females using color as an indicator of malarial infection would only marginally im prove their chance of choosing an uninfected male over random choice. Three-spined Preferentially attracted to Parasitization caused a significant decrease in stickleback!ciliate more intensely colored (red) the intensity of a male's color signal and physical (Milinski and Bakker, males. condition. 1990) Drosophila/nematodes Preferentially mate with Not examined. Suggested that parasitism may (jaenike, 1988) unparasitized males. reduce courtship vigor or color cues important in Male-male interactions may female choice of a mating partner. also playa role.

* For additional studies see papers from the American Society of Zoologists' symposium on Parasites and Sexual Selection (Am. Zool., 30:225-352, 1990); this symposium was summarized by Pomiankowski, 1989. parasites. The general perception of these crea parasites are subject to, and constrained by, the tures is that they are evolutionarily plastic, same rules that govern the evolution ofall bio structurally degenerate carriers of disease. In logical systems, and that the biology of these the following section we hope to demonstrate diverse organisms is not only fascinating, but that "parasite" does not equal "disease;' that also is a crucial component to any study of the SEPTEMBER 1991 PARASITES AND SEXUAL SELECTION 259 interaction between hosts, parasites and inter age, the coevolutionary course can be towards sexual selection. essentially zero virulence, or to very high vir There are three aspects of parasite-host in ulence, or to some intermediate grade." teractions that are of interest to students of sex What does the empirical evidence tell us ual selection: the effects of parasites on the about the effects of parasites on host health? health of their hosts, host resistance, and para There are a plethora of studies detailing the site prevalence and intensity. Since parasitol pathological effects of medically and veterinar ogists have been investigating these questions ily important parasites, in both the natural host for many years, both the conceptual framework and the laboratory-sculptured model host. U s and the empirical database are well developed ing these systems as evidence for the general in each area. Rather than attempting to sum pathology of parasites, however, is in effect marize the discoveries and advances over the a self-fulfilling prophecy, since only species past 50 years of parasitological research, we will that cause disease are intensely scrutinized. present a brief outline of the current wisdom Nevertheless, the results of this research can in each area and provide the interested reader provide us with information about the host with additional references. Given the complex parasite relationship at the "virulent" end of An ity of these questions, however, we would like derson and May's evolutionary scale; natural to reemphasize our belief that any investiga examples of highly lethal host-parasite exam tion of sexual selection via parasite-host inter ples do exist, albeit rarely, in nature (see, e.g., actions requires the input of both ethologists Crump and Pounds, 1985). Within this cate and parasitologists. gory, details of invasion and infection by the widespread, clinically important and tiny Do Parasites Affect the Health of Their Hosts? nematode Trichinella spiralis have been exam The question ofwhether parasites affect the ined at length in both human beings and their health of their hosts has been widely debated laboratory equivalent, the ubiquitous white among parasitologists. Until recently, most bi mouse. ologists accepted the intuitive theoretical per The life cycle can be briefly described by fol spective that the most evolutionarily success lowing the life of one female juvenile worm: af ful parasite species were the least pathogenic. ter being ingested by a suitable host (almost The reasoning behind this assumption is any mammal), she wanders in sinusoidal waves straightforward: parasites causing less severe through the cells of the small intestinal floor damage to their host are at a selective advan feeding on host tissue (Wright et aI., 1987), ma tage because they are able to reproduce for a tures and reproduces in the intestinal epithe longer period of time than their more patho lium, and produces about 1500 juveniles over genic brethren. The benefits to the host of be 4 to 16 weeks. These juveniles work their way ing infected by less pathogenic individuals are into the hepatic portal system, are carried obvious. During the lengthy course of coevo throughout the host's body, feed on and encyst lution, then, the relationship between the host within skeletal muscle tissue, and are released and its parasite should move toward a type of when the infected host is eaten by another commensalism in which the "cost" of parasit mammal. Although Homo sapiens is not a nor ism, although still present, is not unmanage mal definitive host, invasion by the worm, a side able for the host (Schmidt and Roberts, 1985). effect ofhunting (e.g., bears and seals) and as Anderson and May (1982) questioned the sim sociation with domestic animals (e.g., pigs), has plistic assumption that interactions between become an important medical problem. The parasite pathology and reproductive output al effects of parasites on the mouse have been well ways drive the host-parasite association down documented on all levels of the host-parasite an evolutionary pathway toward minimal vir interaction. Genes have been identified that in ulence. They concluded that "the coevolution fluence both the rate at which adult worms are ary trajectory followed by a particular host expelled from the mouse's intestine and the parasite association will ultimately'depend on fecundity of females (Dick et aI., 1988). Dur the way the virulence and the production of ing the intestinal stage ofthe infection, oxygen transmission stages of the parasite are linked consumption, body temperature (Kilgore et al., together: depending on the specifics of this link- 1988), and food uptake from the intestine (Cas- 260 THE QUARTERLY REVIEW OF BIOLOGY VOLUME 66 tro and Olson, 1967; Castro et aI., 1967) are tylus salmonis (Cone and Odense, 1984) that all decreased relative to the levels of infection increased fry mortality significantly in Salveli in the host. These physiological changes are nusfontinalis (Cusack and Cone, 1986). Addi mirrored in behavioral changes; during this tionally, experimental investigation frequently period the activity of infected individuals involves exposing individuals to levels of para decreases dramatically (Kilgore et aI., 1988). sitism that are much higher than those docu Once the larvae have encysted in muscle tis mented in natural populations (parasites are sue, the behavioral abnormalities become more generally overdispersed in the host population; pronounced (Zohar and Rau, 1986). Infected Anderson and May, 1978). Since the effects of females are less likely to be courted by males, parasites on host health have rarely been rig reject that courtship more often (Edwards and orously examined in the field, it may be mis Barnard, 1987), and pay less attention to their leading to extrapolate from the extremities of young (Rau, 1985) than uninfected females. the laboratory to the compJexities of nature. Pups born to and raised by infected mothers, Overall, the results of both laboratory and although uninfected themselves, weigh less, are field investigations are equivocal. In some sys more hyperactive (Rau, 1985), and show a de tems, parasite presence negatively affects some lay in development of reflexive, locomotor and parameter ofhost fitness such as growth, lon memory processes (Nagy et aI., 1983) relative gevity' or reproductive output [e.g., dipterans to uninfected pups. This, in turn, may affect and water mites (Lanciani, 1986); tapeworms the future dominance status of males and the and oligochaetes (Courtney and Christensen, development of normal parental behavior in 1987)]. In other systems, the presence of the females. In addition, infected males are subor parasite has no detectable effect on the host dinate to uninfected males, and alpha males [(e.g., rainbow trout and monogeneans (Cu frequently lose their dominance ranking after sack, 1986); tapeworms and oligochaetes (Cal becoming infected (Rau, 1983, 1984). There entine, 1967; Courtney and Christensen, 1987); is thus a large body ofevidence accumulating hematozoans and birds (Fallis et al., 1974; Fallis that supports the hypothesis that T spiralis has and Desser, 1977)]. And in still other systems, a powerful adverse influence on the fitness of the presence of the parasite may actually be both infected individuals and their offspring beneficial to the host under certain conditions (Rau, 1985; Zohar and Rau, 1986; for an ex (see, e.g., Lincicome, 1971; Davies et aI., 1980; cellent review of research in this area, see M unger and Holmes, 1988). Let us investigate Ewald, 1983 and references therein). this question further by examining the effects Compared to the intense activity in the med ofparasites on two sexually dimorphic species, ical and veterinary fields, few studies have fo the three-spined stickleback, Gasterosteus aculea cused their attention upon the effects ofpara tus, and the iguanid lizard, Anolis limifrons. sites on their natural hosts, either in wild or Three-spined sticklebacks, Gasterosteus aculea in laboratory populations (Anderson, 1976; tus, find copepods to be attractive food items. Bennett et aI., 1976; Rand et aI., 1983). Re Unfortunately, since these crustaceans are the search in this area is bounded by several cav first intermediate host in the life cycle of the eats because the effects of a given parasite spe avian tapeworm Schistocephalus solidus, this din cies may vary with host age (Khan and Lee, ing preference exposes the diners to infection. 1989), host size (Loker et aI., 1987), host sex Once eaten, the larval tapeworm migrates into (Evans et aI., 1985), host stress level (Esch et the skeletal muscles, and develops and grows aI., 1975), and amongclosely related hosts (Cu into a largejuvenile, forestalling sexual matu sack, 1986). The latter effect is particularly im ration until released in its definitive avian host portant because it emphasizes the dangers of (usually a heron). In this system, the stickle generalizing from an association between host back is simply a handy, transitory vehicle for A and parasite A to an association between host the parasite to move from its birth in the water B and parasite A. For example, Salmogairdneri to its home in the bird, so it is not surprising and S. trutta survived an association with the that the relationship between the fish and the same Gyrodactylus sp. that killed Saloelinusfon parasite is a pathological one. Infected indi tinalis(Embody, 1924), while Salmogairdneri fry viduals, easily recognized by their extremely were not affected by the infections of Gyrodac- bloated abdomens, are physiologically com- SEPTEMBER 1991 PARASITES AND SEXUAL SELECTION 261 promised, slower swimmers, less successful for and infecting new RBCs, and other individu agers, and reproduce at a reduced rate (Arme als form gametocytes that remain quiescent un and Owen, 1967; Pennycuick, 1971; Milinski, til ingested by the invertebrate host, in which 1984; but see McPhail and Peacock, 1983). fertilization and embryogenesis occur (see More importantly, from the tapeworm's per Rand et al., 1983 for a description). Like the spective, these individuals are willing to take relationship between the tapeworm, stickleback more risks during foraging, and demonstrate and heron, the parasite must be transmitted an increased oxygen requirement that drives to a second host species to complete its life cy them to the water's surface, gasping at the air cle. The dynamics of transmission, however, water interface (Lester, 1971).Both of these be are different from the preceding example be havioral changes increase the likelihood that cause the lizard need only be located and bit the vehicle will be eaten by a heron, transport ten, not consumed, by that host (reducing se ing the parasite to its definitive host, where it lection pressure on the lizard to avoid the matures and resides in relative harmony for a parasite), and the parasite multiplies asexually short time. The coevolutionary associations be within the lizard, thus increasing the proba tween the parasite and its intermediate host are bility of transmission (reducing the selection under strong selection pressure because the pressure for profound host behavioral altera evolutionary stakes are higher at this stage of tions by the parasite). All other things being the life cycle. If the parasite does not reach its equal, then, we would predict that this associ definitive host, it cannot reproduce, so the ef ation should coevolve toward a balance at the fects of nontransmission are absolute on para lower end of Anderson and May's virulence site fitness. From the host's perspective, infec scale. Rand et al. (1983) investigated the hy tion reduces reproductive output and longevity, pothesis by collecting 1,287 blood smears from but since the amount of this reduction depends 735 Panamanian anoles over a two-year period. upon factors such as age of infection, intensity This extensive study revealed that neither the ofinfection and intensity ofpredation, the ef presence nor intensity ofmalarial infection af fects on host fitness are relative, not absolute. fected lizard weight, food intake, time of ac All other things being equal, then, the associ tivity, female reproductive condition (mea ation should coevolve toward a balance at the sured as appearance of oogenesis, number and high end of Anderson and May's virulence scale size of eggs), growth, survivorship, or preda (see Dobson, 1988 for an extensive investiga tion. The authors cautioned that the effects of tion of the consequences ofbehavior manipu infection may be felt more strongly early in liz lation by parasites on the population dynamics ard development, but once an individual has of their hosts; see also Holmes and Bethel, 1972; reached maturity "malaria does not have a Minchella, 1985; Moore and Lasswell, 1986; marked effect on the anoles that it infects." In Quinn et aI., 1987 for excellent discussions). this particular system, we have evidence that In this particular system we have evidence that the parasite is not affecting the health of its adult the parasite is affecting the health of its host; host, so all reproductively active males are however, because the interaction is so strongly projecting the same message: "I made it biased in favor ofthe parasite, it is difficult to through the juvenile bottleneck." Since males detect variation in male response to that para who did not make it through the bottleneck are site. This, like many associations between para not a part of the breeding population, the mes sites and their intermediate hosts, therefore sage is of limited value to a "choosy" female. appears to be an unsuitable system for investi Additionally, there are at least two types of gations of the relationship between parasites, reproductively active males in this popula hosts, and sexual selection. tion - those who were infected asjuveniles and Infection with Plasmodium balliand members survived the infection, and those who have of the P tropiduri species group in Anolislimijrons never been exposed to the parasite, and may follows the usual malarial course: parasites in or may not have survived, so the message is an jected into the bloodstream of a suitable lizard ambiguous one in terms of overall male "vigor:' host eventually enter red blood cells, where So, like the stickleback example but for differ some individuals undergo cycles of intense ent reasons, this would not be a suitable sys asexual reproduction, bursting out of the old tem for investigating the relationship between 262 THE QUAR TERL Y REVIEW OF BIOLOGY VOLUME 66 parasites and sexual selection. There are two predicted outcomes of this bio The current database provides us with an logical "arms race." Either the host is resistant important insight into host-parasite associa and effectively parries the invadingparasite, or tions: never assume a priori that the parasite the host is susceptible and the parasite becomes is adversely affecting the health of its host. The established in its new home. According to this host-parasite association can travel down a perspective there are two host phenotypes, "sus number of evolutionary pathways toward re ceptible" and "resistant;' and two parasite duced or increased pathogenicity, and this, in phenotypes, "virulent" and "benign;' the devel turn, will affect the suitability of the associa opment ofwhich is influenced by genetical and tion for studies concerned with documenting physiological factors. Hamilton (1982) and the interaction between host, parasites and fe Hamilton and Zuk-(1982) used this concept of male choice among reproductively active males coadaptive cycling between hosts and parasites (Table 2). as the foundation for their theory. Although this traditional perspective of host resistance was What Is Involved in the Evolution of Host retained by biologists outside of parasitology, "Resistance" to Parasites? it was being challenged within that discipline The interaction between a parasite and its prior to Hamilton and Zuk's proposal. Price host has traditionally been viewed as a constant (1980) outlined the emerging new viewpoint struggle between the parasite's attempt to in thus: "The selection pressure exerted by highly vade the host and the host's attempt to evade specialized organisms on host populations can the invader (Mode, 1958; Ehrlich and Raven, produce the most unexpected results. We are 1964; Feeny, 1976; Dawkins and Krebs, 1979). not yet sensitive enough to appreciate even a

TABLE 2 Summary of the associations between the strength of a parasite's adverse effect on its host's health, and the suitability of the systemfor a sexual selection investigation * This simplistic analysis is based upon the assumption that all males are equally exposed to parasites. As discussed previously, the reliability of the message conveyed in the degree ofepigamic development will be decreased if all males are not equally exposed to parasitic invasion. Strength of adverse effect on host health Suitability for sexual selection study

1. No effect on adults or juveniles Unsuitable 2. Strong effect on juveniles- no effect on adults Unsuitable: only "resistant" males make it through the juvenile bottleneck; however, once mature, there is no difference among breeding males based on the early host-parasite association. 3. Moderate effect on juveniles- no effect on adults Suitable: if effects of parasitism during development are translated into differences between breeding males, this provides "truthful" information about those males. 4. No effect on juveniles- moderate effect on adults Suitable: if the effects of parasitism in the adult are translated into intermale differences in the degree of epigamic development, there is potential for a female to use this information as an indicator of male "vigor." 5. Strong effect on adults Unsuitable: pathogenicity is too high, infected individuals so physiologically or behaviorally disrupted that they are effectively removed from the breeding population and are thus not part of a female's choice based upon differences among reproductively active males.

* See also Hamilton and Zuk, 1982; Endler and Lyles, 1989. SEPTEMBER 1991 PARASITES AND SEXUAL SELECTION 263 small fraction ofthem...." The resurgence of fitness, there should be strong selection for any interest in using parasites as model systems for change that counters this decrease if thecost of studying the evolution of biological associations host resistance is less than the cost of being infected. has prompted many authors to reexamine the Minchella proposed that these alternative problem ofhost resistance. These studies have strategies in the evolution ofhost resistance de indicated that, contrary to the traditional per pended upon the interaction among four fac spective, there may be numerous evolutionary tors: intraspecific variability in host resistance outcomes to the struggle between hosts and at the genetic level, the strength ofdirectional parasites. For example, research in a variety selection against susceptibility as measured by of systems has uncovered intraspecific genetic the effect of the parasite on host fitness, the variability in both parasite infectivity (see, e.g., probability of host infection by the parasite Dick, 1983; LoVerde et aI., 1985; Tanuri et aI., (roughly equivalent to the prevalence of the 1985; Collins et aI., 1987; Nadler and Honig parasite in the host population), and the cost berg, 1988) and host susceptibility (see, e.g., of resistance, stemmingfrom possible compen Richards, 1975, 1976; Wakelin, 1978; Sitepu satory changes associated with increasing re et aI., 1986; Wassom et aI., 1986; Mulvey et aI., sistance such as disadvantageous changes in 1987; Leighton et aI., 1989). Additionally, the fecundity (Wright, 1971) or an increase in the development of an immune response may be energy input required to combat the invader. dependent upon a complex polygenic interac The cost of resistance also depends upon tion (e.g., between a variety of "resistance whether the host is simply biochemically un promoting" and"disease-promoting" cells: Moll suitable for parasite development (passive re and Mitchell, 1988), and resistance may be sistance) orwhether the host reacts to the para transferred from a mother to her offspring, at site's presence (active resistance). Theoretically, least in mammals (Knopfand Coghlan, 1989). active resistance, which involves the physiolog And finally, the old parasitological maxim that ical chain reaction "recognize invader, immobi the prevalence and intensity of infection within lize invader and destroy invader;' should be a a host population represents the end point of more costly strategy and thus more difficult to an interaction between the genetical, physio evolve, than passive resistance. Interestingly, logical, behavioral, and ecological character this prediction has been confirmed for one of istics of both the host and the parasite, is gain the most intensely studied parasitological sys ing increasing importance in studies of host tems to date; the relationship between the para resistance. As we will discuss in the next sec sitic flatworm Schistosoma mansoni, the perpetra tion, host infection may vary with age, sex, size, tor of human schistosomiasis or Bilharzia and season, so assessments ofthe relationship disease, and its intermediate snail host Biom between number of parasites and the degree phalariaglabrata (see discussion and references ofmale character development cannot be con in Minchella, 1985; for additional discussions ducted in the absence ofthis information. This see Rollinson and Southgate, 1985; Richards caveat is not new to biology. No ethologist would and Shade, 1987). Since the host's "strategy" include sexually immature individuals in an for dealing with parasite invaders is determined analysis of courtship behavior. Overall, then, by the outcome of the interactions among the as suggested by Price, these studies have dem precedingfour factors, Minchella argued that onstrated that the relationship between offen the evolution ofhost resistance will occur only sive parasite and recalcitrant host is far more if the probability ofinfection multiplied by the complex than previously thought (for a game cost of being infected is greater than the cost theoretical discussion of this problem see Co of host resistance. hen and Newman, 1989). Ifthe cost ofhost resistance is high, and the Minchella (1985) suggested that a variety of probability of being infected is low and un alternative strategies, based upon changes in predictable, Minchella proposed that the hosts resistance and changes in life-history tactics, and parasites might travel down an alternate are available to hosts faced with invading para evolutionary pathway involving changes in life sites. He extended the classical argument for history variables. Although there are numer the evolution of host resistance by arguing that ous examples of parasites affecting the behavior if parasite infection is causinga decrease in host of their hosts (see references in Minchella, 264 THE QUARTERLY REVIEW OF BIOLOGY VOLUME 66

1985), the evolution ofhost counteradaptions of host-parasite associations and the directional has rarely been studied beyond the level of phys nature of female choice was not sufficient to iological resistance. Some tantalizing glimpses explain the evolutionary elaboration of the into host-parasite interactions are available, male character and did not increase popula however, once again, from studies of Biom tion fitness. Overall, then, theoretical informa phalariaglabrata. Four to six weeks after infec tion from population genetics and empirical tion by Schistosoma mansonimiracidia (the first data from host-parasite studies are consonant larval stage), the snail's reproductive output be in their conclusion that a refinement of the gins to decrease, and eventually most of the in mechanism is required. Does this condemn fected individuals are physiologically castrated. field biologists to inactivity within a modelless This example leaves no doubt about the nega limbo until the next theoretical advance occurs tive effects a parasite can have on host fitness. in this area? Ofcourse, the answer to this ques Now, we already know that the evolution of ac tion is "no, with some caveats." In the best of tive resistance to these larval invaders has not all possible worlds, there would be a profusion evolved very often, if at all, in these snails. Are of host-parasite systems for which the details the parasites winning the arms race in this sys of resistance and infectivity had been exam tem? In order to answer this question, research ined on genetical, physiological, ecological, and ers focused their attention on the host's be behavioral levels. In the real world, most parasi havior during the interval between penetration tologists have focused their attention upon con by the miracidia and the beginning of repro trolled, laboratory investigations of medically ductive interference. These studies revealed a and veterinarily important parasite species. fascinating response, now termed "fecundity Therefore, it becomes even more critical that compensation." Immediately following pene a parasitologist is included in any parasites-and tration, the host embarks upon a spurt ofegg sexual-selection project, maximally to inves laying, continuing along this intensified path tigate the host-parasite association, minimally way until the physiological impact of the para to provide best guesses based upon knowledge site's presence begins to affect its reproductive of the parasite species in particular, and the dy output. Overall, this increase is hypothesized namics of parasite-host interactions in general. to compensate, at least to some de gree, for Although it is not perfect, this strategy is still expected future reproductive losses (see also preferable to one that proceeds from a preferred Crews and Esch, 1986 for a similar discussion host type to an associated parasite, chosen at of a different host-parasite system). random or because it is easy to count, to con The preceding discussion, although brief, sultation of papers concerningthe general biol has shown that the evolutionary interaction be ogy of the chosen parasite. tween parasite infectivity and host resistance is more intricate than previously thought. Factors AjJecting Parasite Prevalence There are a number of evolutionary pathways To clarify first a point ofterminological con available to host-parasite associations, and the fusion, note that the terms prevalence, inci direction taken depends upon the interplay be dence and intensity are used by parasitologists tween the ecological and genetical constraints to describe the distribution of one species of and flexibilities inherent to each individual sys parasite within a host population (for defini tem. If this single-species interaction is mag tions see Margolis et aI., 1982). It is important nified by the actual number ofdifferent para to rememher that, unless otherwise stated, the site species associated with any given host, the word "parasites" refers to conspecifics, not an problem of "host resistance" takes on almost amalgam of every parasite species found in the overwhelming complexity. Such complexity host. While keeping this in mind, let us return does not invalidate Hamilton and Zuk's origi to the problem of parasite prevalence. The nal concept about a selective interplay between Hamilton-2 uk hypothesis requires that resis parasites, hosts, and female choice; it does, tant males court more vigorously, develop however, call into question the specific mecha brighter colors, orlonger tails, and thus truth nism proposed to explain how this interplay fully advertise their"genetic superiority" to their might occur. Kirkpatrick (1986) concluded that nonresistant, parasite-stressed conspecifics. In the interaction between the oscillating nature order to postulate that the dynamical interac- SEPTEMBER 1991 PARASITES AND SEXUAL SELECTION 265 tion among parasite load, male quality and fe son, 1982; Dobson, 1988), weakening the as male choice produces a strong enough selec sumption that differential distribution of para tive force to affect the evolution of the male sites within a host species reflects differences epigamic trait, we must also assume that all in resistance to infection among equally ex males are equally exposed to parasite infection. posed individuals. The assumption is further Is this assumption valid? Parasite population weakened by the observation that host age structure is generally overdispersed or conta (Smith et aI., 1985; Loker et aI., 1987; Court gious (see, e.g., Anderson and May, 1978; Price, ney and Christensen, 1988), host sex (Whit 1980; Anderson, 1982; Esch, 1983; Dobson, lock, 1937; Pickeringand Christie, 1980; Smith 1988; Yanez and Canaris, 1988), meaning that et aI., 1985; Lanciani, 1988), host social struc there is a very high variance to mean ratio when ture (Freeland, 1979; Moore et aI., 1988; abundance or relative density (number of para Blower and Roughgarden, 1989), and changes sites per host) is documented. From the para in environmental factors such as season, tem site's point ofview, this allows the host popula perature and water level either singly, or in tion to carry a larger total number of parasites combination, can influence parasite prevalence for any given prevalence (percent of hosts ac and intensity in a host population (see, e.g.; tually infected with at least one parasite) than Crews and Esch, 1986; Gleason, 1987; Mar would be possible if infection followed a Pois cogliese and Esch, 1989; for two excellent long son distribution. Typically, more than 50 per term studies ofthe influences ofenvironmen cent of the parasites occur in less than 5 per tal factors on parasite population dynamics, see cent of the hosts; hence, there are relatively few Janovy and Hardin, 1988, and Riggs and Esch, heavily infected organisms in the host popula 1987 and references therein). tion at any given time. It has been tempting Overall, the general observation that there to explain these heavily infected hosts as being is a high variability in prevalence and inten evolutionarily unfit for some reason, and as sity ofinfection both among hosts and within sume that their heavy parasite load indicates an individual host over time can be attributed strong selection against them. to the interaction of a variety of factors, only Anderson and May (1979) examined this as one of which need be heterogeneity in host re sumption for one system, larval schistosomes sistance (Anderson and May, 1979; Anderson, and their molluscan intermediate host, and 1982; Anderson and Gordon, 1982; Scott and concluded that the extant population models Gibbs, 1986; Wallace and Pence, 1986; Moore were poor predictors of many aspects of the as et aI., 1987). Hence, the assumption that all sociation because the biological assumptions host individuals have been equally exposed to were unrealistically simplified. They empha parasitism is unfounded, introducing an ele sized that, outside of the laboratory, parasites ment of chance into an intersexual communica and their hosts do not inhabit a spatially and tive interchange based upon degree ofparasit temporally uniform world. For example, para ism. Without information about the factors site distribution patterns are determined, in affecting parasite distribution on the host of in part, by the probability ofhost contact with in terest, we cannot assume that the absence or fective stages of the parasite. Depending upon low numbers of a parasite is an accurate reflec the system, this probability will be influenced tion ofan individual host's resistance. Heavily by host interactions (direct transmission from infected hosts may be either unfit or unlucky, host to host), the availability and population an uninfected host may be either resistant or dynamics of intermediate hosts, and the dis lucky. Consequently, ifhost health is adversely tribution of infective stages (e.g., Riggs and affected by increasing parasite load, then a fe Esch, 1987). If, as more studies are revealing, male will be exposed to, minimally, four male this dIstribution is clumped in the environ "health messages": males with high parasite ment, then the probability of any host in burdens (nonresistant), truthfully advertising dividual being infected is strongly subject to their poor health; males with high parasite bur its chance contact with a clump of infective den (resistant or nonresistant and unlucky), parasite stages (Anderson and Gordon, 1982). ambiguously advertising their "genetic qual An overdispersed population structure is thus ity"; males with no or low parasite burdens who expected on purely stochastic grounds (Ander- have been exposed to parasitism (resistant), 266 THE QUARTERLY REVIEW OF BIOLOGY VOLUME 66 truthfully advertising their good health and prediction in more detail. "good genes"; and males with no or low para The Hamilton-Zuk hypothesis was origi site burdens who have not been exposed to par nally formulated to explain the evolution of epi asitism (resistant or nonresistant and lucky), gamic characters, not to explain the macro ambiguously advertising their "genetic qual evolutionary distribution patterns ofparasites ity?' Depending upon the frequency of female among their hosts. Incorporation of these dis exposure to the second and, more importantly, tribution patterns into our picture of host and the fourth male message, the presence of am parasite interplay and sexual selection, how biguous information about the mating system ever, uncovers two im portant weaknesses in the will weaken the directional association between macroevolutionary prediction as it is currently male character development, degree of para stated. The first problem is a methodological sitism, and female choice. one. If we uncover a correlation between the Although it is subject to the caveats discussed number of parasite species inhabiting a host above, the microevolutionary prediction of the and degree of epigamic development in the Hamilton-Zuk hypothesis is straightforward: host, we cannot accept that correlation as evi within a species, females should prefer males dence of causality without determining that the with fewer parasites. The macroevolutionary factors involved in the observed asymmetrical prediction, on the other hand, is more ob distribution of parasite species could not, in scurely worded, "animals that show more turn, be involved in the evolution ofthe sexu strongly developed epigamic characters should ally selected character (Harvey and Partridge, be subject to a widervariety of parasites" (Hamil 1982). For example, based upon a correlation ton and Zuk, 1982, p. 385, our italics). Confir between parasite load and the degree of male mation of this prediction is provided by un showiness in passerine birds, Hamilton and covering a strong correlation between some Zuk concluded that the evolution of showiness measure ofparasite load and the degree of epi was being driven by the interaction between gamic development across host taxa. When a host-parasite arms race and female choice for searching for this correlation, however, para a male character that accurately reflected the site species are often combined to produce one ou tcome ofthat race. Although this is one pos "infestation" value regardless of the differences sible explanation for the observed correlation, in pathology that may be caused by different it is not the only one. Kirkpatrick and Suthers species ofparasites, and prevalence, incidence (1988) examined blood smears from 59 species and intensity data are used interchangeably of North American passerines in an attempt, with each other and with more nebulous terms among other things, to uncover the intrinsic applied to the distribution of more than one and extrinsic factors associated with increased parasite species (parasite load or burden). or decreased risk of infection by members of Therefore, it is initially unclear whether the parasite genera Haemoproteus, Plasmodium Hamilton and Zuk's prediction ofa "wider va and Leucocytozoon. They discovered that passer riety" of parasites refers to more parasites ab ine species with ground feeding habits were sig solutely, or to more species of parasites. Authors nificantly more likely to be infected with blood have criticized the first interpretation, arguing parasites in general, and Leucocytozoon spp. in that the model could also make the opposite particular, than passerine species feeding above prediction: "species with bright males should the ground. Bennett et al. (1978) and Kirk be the least parasitized if female selection for patrick and Smith (1988) reported that the prev male brightness has lowered the level of infection alence of blood parasites in colonially nesting in the population" (Borgia, 1986, our italics) birds was significantly higher than in non and "ifhowever, parasite loadis reduced as a con colonial nesters. And finally, various authors sequence offemale choice for resistant mates, have suggested that susceptibility to infection showiness might become negatively correlated by insect-borne blood parasites may be in with parasite load" (Read and Harvey, 1989b, fluenced by the height of the nest location our italics). So, this leaves us with the second preferred by a given avian species (Greiner et interpretation: sexually showy species should aI., 1975; but see Kirkpatrick and Suthers, host more species of parasites than their less 1988). So, in this system, the number of para flamboyant relatives. Let us investigate this site species associated with a host is strongly SEPTEMBER 1991 PARASITES AND SEXUAL SELECTION 267 correlated with foraging, nesting and social of parasite species regardless of the "suscepti habits of the birds, anyone of which might, bility" of the hosts. Because the distribution of in turn, be associated with the evolution of sex parasites among potential host species is more ual dimorphism. Ifthis is the case, then the ob complex thanjust an extrapolation ofpo pula served correlation between parasite load and tion-level host resistance, there is no a priori sexual dimorphism is an artifact of the causal reason to believe that a sexually showy species interaction between ecological factor z and will always be associated with more parasite parasite load and ecological factor z and sex species than its less flamboyant relative. ual dimorphism (see also Read, 1987; Scott and We can approach this problem from another Clutton-Brock, 1989). direction by restating the Hamilton-Zuk pre The second problem is more fundamental diction in the following manner. The existence and it is inherent in the structure of the mac of female choice in species A does not cause spe roevolutionary prediction as it is currently for cies A to have more parasites than species B; mulated. From the observation that parasite however, the fact that species A has more para species are not equally distributed among host sites than species B may cause the evolution of species, Hamilton and Zuk extrapolated the epigamic characters viafemale choice in species conclusion that some hosts were more suscep A. Now, consider a hypothetical group of birds tible to parasite attack than others. Equating comprised of two sister groups, the ABCidae "susceptible" with "less capable of fending off and DEFidae (Fig. 1). All the members of the parasitic invasion;' however, returns us to the ABCidae are sexually monomorphic and host resistance paradox: within a species, preferred five species ofparasites. All the members of the males are more resistant to parasitism than DEFidae are sexually dimorphic and also host nonpreferred males, but species containing five species of parasites. Experimental investi these preferred individuals are less resistant to gations have demonstrated that females in the parasitism than species without this type of DEFidae prefer males with brighter feathers, choice dynamic. Once again, the results of and these males, in turn, have lower numbers parasitological research can help us resolve this of parasites than their nonpreferred conspeci problem, not by revealing some magical an fics. The microevolutionary study confirms swer but by demonstrating that the paradox that a Hamilton-Zuk mechanism is operating does not really exist. Macroevolutionary pat in the DEFidae. The macroevolutionary anal terns of parasite distribution are determined ysis, however, does not confirm this result be by a complex interaction among a variety of cause it contradicts the predicted relationship factors, only one of which may be changes in between the degree of sexual showiness and the species-specific host"resistance." For example, number of parasite species. Is there a way out the distribution of helminth parasites in frogs of this conflict? is determined, in part, by both parasite biol Let us return to our initial formulation of ogy and host biology. The infective stages of the macroevolutionary prediction and refine the majority of helminth species are generally transmitted to the primary host in the water, and as adults, some anurans (e.g., toads and # of parasite species 5 5 5 5 5 5 their relatives) are terrestrial orfossorial, while sexual dimorphism D D D •• other anurans (e.g., leopard frogs, green frogs, A B CD E •F bullfrogs) are aquatic. Because of the con straints imposed by the biology of the hosts and

parasites, we would expect the aquatic ranids origin of sexually to host more helminths than terrestrial bu selected trait fonids. Not surprisingly, studies by parasitol 5 host/parasite associations ogists have all shown a strong positive relation ship between the number of parasitic helminths FIG. 1. DISTRIBUTION OF PARASITES AND infecting anurans and the aquatic habits of the SEXUAL DIMORPHISM AMONG A CLADE anurans (Brandt, 1936; Prokopic and Kri OF HYPOfHETICAL BIRDS vanec, 1975; Brooks, 1976). In this system, White boxes, sexual dimorphism absent; black then, there will be a differential distribution boxes, sexual dimorphism present. 268 THE QUARTERLY REVIEW OF BIOLOGY VOLUME 66 it. The existence of female choice in species A some systems can also be made. Overall, then, does not cause species A to have more para the Hamilton-Zuk hypothesis does not make sites than species B; however, the fact that spe a prediction about the relationship between the cies A has more parasites than species B may number of parasite species and the degree of cause the evolution of epigamic characters via epigamic development in the hosts, aside from female choice in species A once that traitappears the trivial prediction that at some time in their in thepopulation.The presence of parasites does history the dimorphic species must have been not cause the sexually selected trait to appear associated with parasites. It is important to em in the population. Ifit did, evolution would be phasize here that this does not invalidate the Lamarckian, not Darwinian. Dependingupon hypothesis; it only forces us to rethink the mac the details of the genetical-developmental sys roevolutionary prediction. This suggests that tem of the host, once the trait has appeared, the macroevolutionary outcome ofthe Hamil the presence of parasites may influence its ton-Zuk mechanism is qualitatively different evolutionary success and elaboration. Given from, rather than being an extrapolation of, this time lag, there is no reason to believe that the microevolutionary dynamic based upon a relationship should exist between the num parasite numbers. We will now examine this ber of parasite species and the degree of epi question further, and present what we believe gamic development. In the preceding exam to be a more robust macroevolutionary predic ple, the presence of an ancestral host-parasite tion for the Hamilton-Zuk hypothesis. association was sufficient to promote the spread of an epigamic characterwhen it appeared (Fig. 1). We can also make a case for sexually di QUESTIONS ABCUT EVOLUTIONARY ORIGINS morphic species hosting fewer parasites than The Hamilton-Zuk hypothesis is concerned their monomorphic relatives of the interaction. with uncovering correlations between the ex For example, the interaction between the di tent of male epigamic development and para morphic species and their parasites may be a site prevalence. This, in turn, is an extension more pathological one. On the other hand, as of the general adaptationist program, in which depicted in Figure 2, the monomorphic hosts observations of correlations among traits in ex may have continued to pick up parasite spe tant species are used as the basis for hypothe cies, while the dimorphic hosts remained as ses concerning the evolutionary relationship sociated with parasites that had been originally between the traits. There are, however, three involved in the host-parasite sexual selection components to character evolution: origin, di dynamic. A case for sexually selected species versification and maintenance. Microevolu having more parasites than their relatives in tionary studies concentrate on the maintenance of traits in current environments where the processes shapingthe interactions between the organism and its environment can be observed # of parasite species 9 6 6 5 5 5 and measured directly. After untangling this sexual dimorphism D D D •• complicated web, these researchers then ex A B CD E •F trapolate backwards to the processes involved 3 parasites added in the character's initial appearance in, and sub sequent spread through, the ancestral species. A macroevolutionary analysis complements 1 parasite origin of sexually added selected trait these studies by providing direct estimates of phylogeny that can be used as a template for 5 host/parasite associations reconstructing the historical patterns ofcharac ter origin and diversification. Such a template FIG. 2. ASSOCIATION BETWEEN SEXUAL can help biologists to focus their search for the DIMORPHISM AND NUMBER OF PARASITES processes underlying these evolutionary pat In this scenario, members of the ABCidae have terns. For example, consider an interesting(but continued to pick up parasites, while members of improbable) group of species with the follow the DEFidae have retained the ancestral host-para ing characteristics (Table 3) and phylogenetic site associations. relationships (Fig. 3). In this group, the rela- SEPTEMBER 1991 PARASITES AND SEXUAL SELECTION 269

TABLE 3 phylogenetic methods. Therefore, we will in Distribution of three characters amongfour members troduce each section with a brief, step-by-step of a monophyletic group (species A + B + C + D) discussion ofa very simple and unambiguous and the outgroup (taxon X) example. Species The Origin of the Sexually Selected Character Trait X AB C D in the Host Group Color blue blue red red red One cardinal rule in a phylogenetic analy Shape square square square round round sis is that one should never bias an analysis by Size small small small small big using the ecological information about to be studied to build a phylogenetic tree. Etholo gists who are interested in determining the ori gin of a sexually selected trait must therefore begin their explorations with two pieces of in formation: a phylogenetic tree for the group tionship between a character and the environ and information about the development ofthe ment in which it originated can be investigated epigamic character throughout the group. The for only one taxon and one character state, best supported (most parsimonious) sequences namely, the evolutionary increase in size ("big") of evolutionary transformations for the epi in species D. Researchers who are interested gamic character, either binary or m ultistate, in the processes involved in characterorigin and can then be determined with reference to the adaptive success, if any, should concentrate phylogenetic tree. This method, called charac their efforts on this species. The presence of ter optimization (Farris, 1970), is best illus small, blue and square characters in species A; trated by data from the Drosophila of ethology, small, red and square characters in species B; stickleback fishes. small, red and round characters in species C; Within the family Gasterosteidae, the three and red and round characters in species Dare spined stickleback has been extensively stud all ancestral legacies. Studies of these species ied from almost every morphological, ecologi will uncover the processes involved in charac cal and behavioral angle because of one bio ter maintenance. By combining the results logical attribute. In the spring males undergo from all these studies we will be provided with a wondrous transformation from inconspicu a more direct estimate of the relationship among ous silver green fishes to flamboyant mosaics the processes underlyingthe origin, spread and of flame scarlet bodies and flashing aquama maintenance of potentially adaptive traits. The rine blue eyes. When the relatives of Gasterosteus incorporation of both the patterns of the past and the processes ofthe present into our frame work of evolutionary explanations will thus x A B c D strengthen our hypotheses of evolutionary change and interactions. In the following section, we will discuss methods for uncovering the origin of the epi gamic character and the origin of the associa tion between, and coevolutionary interactions among, hosts and their parasites. It is not our intention to present an in-depth discussion of the assumptions, strengths and weaknesses of small the methods. We hope only to demonstrate blue that, once unveiled, the macroevolutionary pat square terns can provide important information about evolutionary processes. In order to fully ap FIG. 3. EXAMINING ADAPTATIONIST HYPOTHESES preciate the relationship between pattern and Phylogenetic tree for the improbable species with process, however, it is important to understand the distribution of the three characters mapped the fundamental mech~anics of the various onto it. 270 THE QUARTERLY REVIEW OF BIOLOGY VOLUME 66

TABLE 4 and G. wheatlandi, both of which exhibit charac The distribution of male nuptial coloration among ter b, we generalize that the nodal (ancestral) species within the family Gasterosteidae state is b (node 1). The argument is the same Whole-body nuptial for the node connecting Pungitius and Culaea Species coloration (node 2). Node 3 is the ancestor ofnodes 1 and 2, both of which are b; therefore the state at Spinachia spinachia absent this node is also b. Node 4 connects an a branch Apeltes quadracus absent (Apeltes) with a b node (node 3); therefore, the Pungitius pungitius present state is ambiguous for a or b (a,b). Finally, we Culaea inconstans present Gasterosteus wheatlandi present assign a value ofato the ingroup node because Gasterosteus aculeatus present it connects an a,b node and an a branch (Spinachia), so a wins out over b by majority vote (the principle of parsimony). It is important to note here that any ambiguity at the ingroup node may be resolved by reference to out aculeatus are examined for the presence of groups. whole-body nuptial coloration in breeding Step 2. Predicting up the tree (Fig. 6): Move males, an interestingpattern appears (Table 4). from the ingroup node (Fig. 5) up the tree, We can optimize the character "male nuptial resolving any ambiguity (i.e., node 4, a,b) by color: present or absent" onto a phylogenetic comparing the value of the ambiguous node tree for the stickleback fishes in the following with the value of the node directly below it (in manner. Figure 4 depicts the phylogenetic tree group node). In this example only node 4, for the Gasterosteidae (Paepke, 1983; McLen designated a,b in Figure 5, is ambiguous. Since nan et aI., 1988). The distribution of the bi the value of the node below it is a, node 4 is nary epigamic character is mapped at the ends reassigned state a. All nodal states have now ofthe branches, and the nodes are labeled for been unambiguously resolved on the tree. reference in Figures 5 to 7. Character optimization thus provides the fol Step 1. Generalizing down the tree (Fig. 5): lowing evolutionary hypothesis for this binary Label the two nodes that are farthest from the character (Fig. 7): (1)the absence ofwhole-body ingroup node in the following manner: (1) la male nuptial coloration (state a) is a persistent bel the node a if the two closest nodes or ancestral condition in two of the six terminal branches are either both a, or a and a,b; (2) taxa (plesiomorphy); (2) there was a change label the node b if the two closest nodes or from absent to whole-body, male nuptial col branches are either both b, or band a,b; (3) oration (a to b) in the ancestor ofthe Pungitius if the closest branches or nodes have different + Culaea + Gasterosteus wheatlandi + G. aculea labels (one a and the other b), then label the tus clade. node a,b. Continue working toward the in Although we have now identified the evolu group node in this manner. Thus, beginning tionary origin of the sexually dimorphic charac with the node connecting Gasterosteus aculeatus ter, it is still necessary to determine the mech-

( IlARACTEH CHARACTER SIArE STATE Spmachia Apeltes Pungitlus Culaea G wheatlandi G aculeatus Spmachla Apeltes Pungrtius Culaea G wheatlandI G acule.uus

b b

mgroup node

FIG. 4. PHYLOGENETIC TREE FOR THE FIG. 5. STEP 1 IN FARRIS OPTIMIZATION OF A GASTEROSTEIDAE BINARY CHARACTER: GENERALIZE a, body nuptial coloration absent; b, body nup DOWN THE ThEE tial coloration present. Nodes are labeled with numbers. SEPTEMBER 1991 PARASITES AND SEXUAL SELECTION 271

CIIARACTE', CHARACTER SlATE STATE Spmachia Apeltes Pungitius CuIaea G wheatlandI G aculeatus Spmachia Apeltes Pungttius Culaea G wheatland I G aculc.uus

b b

FIG. 6. STEP 2 IN FARRIS OPTIMIZATION OF A FIG. 7. MACROEVOLUTIONARY HYPOTHESIS BINARY CHARACTER: PREDICT Up THE ABOUT THE ORIGIN AND ELABORATION TREE OF MALE NUPTIAL COLORATION IN THE GASTEROSTEIDAE

anism underlying its appearance. Whole-body nan, 1991 for a more detailed explanation of coloration occurs in the monophyletic group basic optimization procedures). In some sys comprising Pungitius, Culaea, Gasterosteus, and tems, additional biological information (e.g., their ancestor. Since its distribution is cor developmental or biochemical data) can be related with the phylogenetic relationships of brought to bear on the problem. In other sys these taxa, its presence in these species rep tems, we can attempt to evaluate the conflict resents the persistence of an ancestral trait. ing hypotheses of character evolution by direct Consequently, female choice experiments may experimental studies in both the field and the uncover valuable information concerning the laboratory. Although we would like nature to maintenance of sexual dimorphism under cur provide us with a perfect record of evolution, rent environmental conditions; however, on this does not always happen. In such cases, we their own, such experiments do not address expect that phylogenetic optimization will tell why that character was originally successful. us exactly where the ambiguity lies, even if In order to explain this, we need an assessment it cannot provide an unambiguous interpre of that ancestor's biology in its environment. tation. Needless to say, since this assessment requires information about the social interactions in The Origin of Parasite-Host Association a (presumably) extinct ancestor, we cannot From a macroevolutionary perspective, there measure it directly. We can, however, use the are two explanations for the occurrence of a optimization methods just described for un given parasite in a given host; either the para covering the origin of sexual dimorphism, to site inhabits the host because its ancestor was determine whether female choice based upon associated with the host's ancestor (association that dimorphism originated at the same time. bydescent; Mitter and Brooks, 1983) or because For example, McLennan and McPhail (1990) it evolved in association with a different host demonstrated that G. aculeatus females prefer group and colonized the host of interest subse more intensely colored males when given a quent to its evolution(association bycolonization; choice between two suitors vying for their at Mitter and Brooks, 1983). Any parasite spe tention. Ifthis preference is also demonstrated cies inhabiting two or more hosts may exhibit by female Pungitius, Culaea and G. wheatlandi, both types of associations. Clades that show his then we would have strong evidence that the torical association with other clades exhibit initial elaboration of male color in the ances cospeciation (Brooks, 1979) patterns. A phylo tor of the clade was driven by female prefer genetic systematic method(Brooks, 1981,1985, ence for more intensely colored males. 1988, 1990; see also Cracraft, 1988; Kluge, Like many aspects of the natural world, op 1988; Mayden, 1988; Wiley, 1988a,b) has been timization does not always produce unambig developed to distinguish association by descent uous results (Swofford and Maddison, 1987; from association by colonization in cospecia Donoghue, 1989; see also Brooks and McLen- tion studies. In general, it is important to dis- 272 THE QUARTERLY REVIEW OF BIOLOGY VOLUME 66

2 3 4 5 6 7 8 2 3 4 5 6 7 8

FIG. 8. PHYLOGENETIC TREE FOR THE FIG. 9. PHYLOGE~ETIC TREE FOR EIGHT SPECIES AMPHILINIDEA OF AMPHILINID FLATTvVOR~fS WITH 1, Amphilinafoliacea; 2, A. japonica; 3, Gigantolina INTERNAL BRANCHES NU~BERED FOR elongata; 4, G. magna; 5, Schizochoerus paragonopora; 6, COSPECIATION ANALYSIS S. [anickii; 7, S. liguloideus; and 8, S. africanus. 1, Amphilina foliacea; 2, A. japonica; 3, Gigantolina elongata; 4, G. magna; 5, Schizochoerus paragonopora; 6, S. janickii; 7, S. liguloideus; 8, S. africanus; and 9-15, ancestral species.

tinguish these two components in the origins cies offreshwater turtle. The first step is recon of any ecological association if we are to un struction of the phylogenetic relationships of derstand the mechanistic implications of his the organisms. Phylogenetic systematic anal torical versus nonhistorical interactions (Brooks ysis ofthe amphilinids, based on 46 morpho and McLennan, 1991). We hope to show that logical characters, produced a single tree with by understanding the evolutionary origins of a consistency index of 87.5 percent (Bandoni host-parasite associations, one can distinguish and Brooks, 1987) (Fig. 8). We then list the am those parasites that are relevant to the Hamil philinid species with their hosts (Table 5). The ton-Zuk hypothesis from those that are not. phylogenetic relationships of the eight am A simple example applies this method to an philinid species can now be treated as if they esoteric group ofparasitic flatworms, the Am were a completely polarized multistate trans philinidea. Amphilinids, the sister group of the formation series, in which each taxon and each species-rich true tapeworms, are a small (eight internal branch of the tree is numbered (Fig. known species) but widespread group ofpara 9). Each amphilinid species now has a code that sites that live in the body cavities offreshwater indicates both its identity and its common an and estuarine ray-finned fishes and in one spe- cestry. For example, the code for Amphilina

TABLE 5 TABLE 6 List of hosts and species of amphilinids that Data matrix listing binary codes for the eight species inhabit them of amphilinidflatworms Parasite Parasite Host no. Parasite name Parasite name no. Binary code

Aci penseriformes 1 Amphilina foliacea Amphilina foliacea 1 100000001000001 Aci penseriformes 2 Amphilina japonica Amphilina japonica 2 010000001000001 Perciformes 3 Gigantolina elongata Gigantolina elongata 3 001000000100011 Chelonia 4 Gigantolina magna Gigantolina magna 4 000100000100011 Siluriformes 5 Schizochoerus paragonopora Schizochoerus paragonopora 5 000010000000111 Osteoglossiformes 6 Schizochoerus janickii Schizochoerus janickii 6 000001000001111 Osteoglossiformes 7 Schizochoerus liguloideus Schizochoerus liguloideus 7 000000100011111 Osteoglossiformes 8 Schizochoerus africanus Schizochoerus africanus 8 000000010011111 SEPTEMBER 1991 PARASITES AND SEXUAL SELECTION 273

Acrpenseriforrnes Osteoglossiformes Silunfonnes Pe rciforrnes Che lorua Ch elorna Acrpenserrtormes Osteoglosslformes Silunformes Pe rciforrnes

6 7 8 11. 12

15 ...= host SWItch

FIG. 10. HOST CLADOGRAM BASED ON FIG. 11. PHYLOGENETIC TREE FOR THE HOSTS OF PHYLOGENETIC RELATIONSHIPS OF THE EIGHT AMPHILINID SPECIES AMPHILINID FLATWORM PARASITES

Incongruencies between the host cladograms of Figure 10 and the phylogenetic tree of Figure 11 are hypothesized to represent instances of host switching. Congruent portions of the two trees represent in stances of cospeciation between the host and the parasite.

japonica (Parasite No.2) is (2,9,15) and the code upon characteristics of the hosts themselves. for Schizochoerus africanus is (8,11,12,13,14,15). Congruence between the two cladograms in These codes can be represented in a data ma dicates the portion ofthe host-parasite system trix in w-hich the presence of a number in the that can be explained by phylogenetic associa species code is listed as "1" and the absence of tion (i.e., their ancestors were associated), a nurnber in the species code is listed as "0" whereas incongruence highlights cases ofhost (Table 6). The phylogenetic relationships of the switching. For example, the amphilinids in amphilinids can thus be represented by binary habiting acipenseriforms are a monophyletic codes. Now we replace the narne ofeach para group, as are those inhabitingosteoglossiforms. site species in Table 6 with the name of its The turtle Chelodina longicollis, however, is not associated host group (Table 7). Finally, we the sister group ofperciform teleostean fishes, construct a host cladogram based on the phylo so we must interpret the presence of Giganto genetic relationships of the parasite species linamagna(character 4 in Fig. 10) in C. longicollis (Fig. 10). This produces a picture of the his as the result ofa host switch. In addition, con torical association between the parasites and trary to the current phylogenetic analysis ofthe their hosts. actinopterygians, this cladogram places silu We can now compare the host cladogram riform fishes with the osteoglossiforms rather reconstructed from the phylogenetic relation than with the perciforms; therefore, the pres ships ofthe parasites with the cladogram based ence of Schizochoerus paragonopora (character 5 in Fig. 10) in a siluriform host must be the re sult of a host switch as well. So, the evolution TABLE 7 ofthe eight extant amphilinids has involved six Matrix of hostsfor amphilinidflatworms and the binary codes for the parasites and their cases of phylogenetic association with their phylogenetic relationships hosts and two cases ofhost switching (Fig. 11). We now turn our attention specifically to Host Parasite no. Binary code the Hamilton-Zuk hypothesis. Since this hy Acipenseriformes 1 100000001000001 pothesis is about mechanisms of evolution Acipenseriformes 2 010000001000001 ary change, it should make predictions about Perciformes 3 001000000100011 the evolutionary origins of, and associations Chelonia 4 000100000100011 among, characters through time. In this case, Siluriformes 5 000010000000111 the characters are male epigamic traits and Osteoglossiformes 6 000001000001111 host-parasite associations. Unfortunately, there Osteoglossiformes 7 000000100011111 are very few phylogenies available for parasite Osteoglossiformes 8 000000010011111 groups, and ofthese, none are for parasites that 274 THE QUARTERLY REVIEW OF BIOLOGY VOLUME 66

HOST Sp Ap Pu Cu Gw Ga

1 2 3 4 5 6

( a )

R Pu en Gw Ga Gw Ga 12 13 14 15 16 Q 21 22 23

( b ) (C)

FIG. 12. PHYLOGENETIC TREES FOR THREE HYPOTHETICAL, BUT MONOPHYLETIC, GROUPS OF PARASITES (a) Tapeworms living in the upper intestine of gasterosteid fishes. (b) Monogeneans living on the body of the monophyletic group R and some members ofthe Gasterosteidae. (c) Nematodes living in the intes tine of the monophyletic group Q and some members of the Gasterosteidae. Hosts: R, a monophyletic group of theoretical trout species; Q, a monophyletic group of theoretical blenny species; Sp, Spinachia spinachia; Ap, Apeltes quadracus; Pu, Pungitius pungitius; Cu, Culaea inconstans; Gw, Gasterosteus wheatlandi; and Ga, G. aculeatus.

have interested students ofthe Hamilton-Zuk (characters 1-11) originated with the ancestor hypothesis. For the sake of illustration, let us of the stickleback fishes (association by de consider the Gasterosteidae and their associa scent), and speciated along with their host tions with a hypothetical group oftapeworms, group (cospeciation). The monogeneans (char monogeneans and nematodes. The phyloge acters 12-20) are present because of a host netic relationships of these parasites, numbered switch (association by colonization) by ances for phylogenetic analysis, are depicted in Fig tor 20 from the monophyletic group R to the ure 12. This produces the host cladogram ancestor ofthe Pungitius + Culaea + Gasterosteus shown in Figure 13. clade. Once the host switch had occurred the According to this analysis, the tapeworms monogenean ancestor speciated, producing SEPTEMBER 1991 PARASITES AND SEXUAL SELECTION 275 descendant 19, and the remainder ofthe para cospeciation. Although researchers arejust be site phylogeny is congruent with the gaster ginning to ask questions about the relationship osteid speciation pattern (cospeciation). Fi between parasite virulence and the age of the nally, the presence of the nematodes (characters host-parasite association, the working assump 21-25) also resulted from a host switch, in this tion is that tightly cospeciated groups will fall case from the monophyletic group Qto the an at the low end ofthe Anderson and May viru cestor of Gasterosteus wheatlandi and G. aculeatus. lence scale. Because ofthis, by the time the sex When that ancestor (24) speciated, so did the ually dimorphic character appeared in the an ancestral nematode, producingdescendants 22 cestor of the Pungitius + Culaea + Gasterosteus and 23 in their respective hosts. Thus, each of clade, the tapeworm associated with this an the three parasite groups began its association cestor would be hypothesized to have only ami with gasterosteids at a different time. Now, con nor effect on host health. Past this point, the sider the origin ofthe three stickleback-parasite host-parasite interaction would continue to associations in relation to the origin ofthe sex move toward minimal virulence, so by the time ually dimorphic character "whole-body nup Pungitius, Culaea, and Gasterosteus appeared, the tial coloration" as shown in Figure 14. dynamic between host, parasite, male color, There are three types ofmacroevolutionary and female choice would be further weakened, associations delineated in this figure. First, the rendering it less likely that a female choosing host-parasite association appears before the ori males based upon color would also be choos gin of the sexually selected character (tape ing mates demonstrating a superior ability to worms and sticklebacks). In this scenario, the resist invasion and damage from the appropri host and parasite have shared a long history of ate tapeworm species. If these assumptions

R Q Sp Ap Pu OJ. Gw Ga

FIG. 13. PHYLOGENETIC TREE FOR THE HOSTS RECONSTRUCTED FROM THE PARASITE DATA Numbers refer to the three parasite phylogenies depicted in Figure 12: tapeworms (characters 1-11), monogeneans (characters 12-20) and nematodes (characters 21-25). Bold numbers indicate instances of presumptive host switching. Hosts: R, a monophyletic group of theoretical trout species; Q, a mono phyletic group of theoretical blenny species; Sp, Spinachia spinachia; Ap, Apeltes quadracus; Pu, Pungitius pungitius; Cu, Culaea inconstans; Gw, Gasterosteus wheatlandi; and Ga, G. aculeatus. 276 THE QUARTERLY REVIEW OF BIOLOGY VOLUME 66

R Q Spinachia Apeltes Pungittus Culaea Gw G. aculeatus

origin of the nematode/ stickleback association

origin of the monogenean/ stickleback association

origin of the tapeworm/ stickleback association

origin of male nuptial coloration

FIG. 14. ORIGI~S OF !--IOST-PA~ASITEASSOCIAT:OKS AND MALE ~UPTIAL COLORATlON IN GASTEROSTEIDAE, Gw, Gasterosteus w 'ieatlandi

about parasite pathology and the age of the Overall, there are four potential macroevolu host-parasite association are correct, then this tionary patterns, two of which support [Figs. type ofpattern provides, at best, weak support 15(b)and 15(d)], and two ofwhich refute [Figs. for the Hamilton-Zuk hypothesis. Second, the 15(a) and 15(c)] the hypothesis that there is host-parasite association originated in the same an evolutionary interaction between the pres species in which the sexua.ly dimorpi:lic trait ence of parasites and the development of sex arose (monogeneans and sticklebacks). In this ual dimorphism in the study group. Since the scenario, the appearance ofa new parasite spe sequence ofcharacter origin cannot be deter cies in a host species (by a host switch) is ac mined without a phylogeny, a macroevolution companied by the appearance of the sexually ary analysis is a critical first step in distinguish selected character in the host. Since host ing between systems that will, and those that switches are presumed to be accompanied by will not, provide a strong test ofthe Hamilton more virulent adaptive-coadaptive cycling be Zuk hypothesis, For example, suppose a re tween the parasite and its new host (Stock and searcher decided to investigate a host species Holmes, 1987), both the macroevolutionary because it was "sexually dimorphic" and "had pattern and the biological interactions assumed parasites." If this species was like Y and Z in to underlie such a pattern provide strong sup Figure 15(c), this investigator would either (1) port for the Hamilton-Zuk hypothesis. And uncover a significant association between para third, the host-parasite association originates site burden and degree of male epigamic de afterthe sexually selected character(nematodes velopment and incorrectly conclude that the and sticklebacks). This pattern refutes the hy Hamilton and Zuk mechanism was involved pothesis that the interaction between the host in the initial elaboration ofthe male character, parasite dynamic and female c'ioice fo::' a "'TIale or (2) uncover no association and correctly con character reflecting that dynamic is driving the clude that the hypothesis was not supported for initial elaboration of the male character. Al this group. If, however, phylogenies had been though this tells us nothing about the mecha available prior to the experiments (and this is nism of sexual selection, it does not rule out currently a large "if"), the researcher could have the possibility that this interaction is positively determined that the epigamic trait originated reinforcing the mechanism once it occurs. prior to the host-parasite association, rejected SEPTEMBER 1991 PARASITES AND SEXUAL SELECTION 277 the hypothesis on macroevolutionary grounds, the influence of host switching on host-parasite and saved the time and expense of running the interactions, only one ofwhich involves a spe experiment. cific cycling of mutual modification, or recipro The strongest macroevolutionary evidence cal adaptation, of the population ecology or for the Hamilton-Zuk hypothesis comes from population genetics of the ecologically as cases in which a host switch coincides with the sociated species. Resource tracking (or coloni origin of the sexually selected trait. The pro zation) coevolutionary models are based on the posal that host switches are an important com concept that hosts represent patches of neces ponent of the hypothesis complies with the as sary resources that associates have "tracked" sumption that the interaction between the host through evolutionary time (Kethley andJohn and parasite represents a coevolutionary arms ston, 1975). In this case, the diversification of race dynamic. There are, however, two general the associated taxa occurs independently, since classes of coevolutionary models that address this host switching involves the "tracking" of a

vV x y z w x y z

( a ) ( b )

vV x y z Vv x y z

FIG. 15. EXAMINING PHYLOGENETIC RELATIONSHIPS BETWEEN Two DIFFERENT CHARACTERS Black boxes, origin of the host-parasite association; white boxes, origin of the sexually selected trait. (a) The host-parasite association and the sexually selected trait arose independently in different species; the traits neither co-originate nor co-occur. (b) The host-parasite association arose before the sexually di morphic trait; the traits do not co-originate but they do co-occur in some members of the group. (c) The sexually selected trait arose before the host-parasite association; the traits do not co-originate but they do co-occur in some members of the group. (d) Both characters originated in the same ancestor; the traits co-originate and co-occur. Patterns (a) and (c) refute the Hamilton-Zuk hypothesis, pattern (b) provides weak support and pattern (d) provides strong support for the hypothesis. 278 THE QUARTERLY REVIEW OF BIOLOGY VOLUME 66 w x x x x x y cally [Fig. 16(a)] or convergently [Fig. 16(b)] widespread. The evolutionary arms race is the classical coevolution model developed by Ehrlich and Raven (1964) for phytophagous insects and plant systems (see also Mode, 1958; Feeny, 1976; Berenbaum, 1983; Futuyma and Slatkin, 1983). The primary assumption in "arms-race" models is that coevolving ecological associa tions are maintained by mutual adaptive re sponses. For example, it is possible that dur ing the course of evolution, novel traits arise ( a ) RESOURCE IS PLESIOMORPHIC that "protect" the host from the effects of the parasite. It is also possible that traits counter ing such "defense mechanisms" may evolve in the parasite lineage. Evolutionary arms-race models generally assume that the time scale on x which the "defense" and"counterdefense" traits originate in response to reciprocal selection pressure is longer than the time between speci ation events. As a result, we might expect to find macroevolutionary patterns in which the parasite group is missing from most members of the host clade characterized by possession ofthe "defense" trait (Fig. 17). A second possi-

H ( b ) RESOURCE IS CONVERGENT

FIG. 16. SEQUENTIAL COLONIZATION MODEL G .. counterdelense (a) Target area for colonization sequences (en F .. closed within the box) if the resource being used is plesiomorphic and widespread. (b) Target area E .. for colonization sequences if the resource being used is convergent and widespread (from Brooks and D McLennan, 1991). c

B resource that is shared among a group ofhosts A that do not form a clade. For example, the se HOST PARASITE quential colonization model (Jermy 1976, 1984), originally designed to explain insect FIG. 17. EVOLUTIONARY ARMS-RACE TYPE I plant coevolution, proposes that the diversifi The host and parasite phylogenies are congruent cation of phytophagous insects took place af (enclosed within the box) up to the point at which ter the radiation of their host plants. The in the defense trait appears. If the origination of a sects are hypothesized to have colonized new counterdefense lags behind this, and if the hosts con tinue to speciate, parasite species will be prohibited host plants many times during their evolution. from interacting with any of the new host species In each case the colonization was the result of (marked with an asterisk). Once a counterdefense the evolution ofinsects responding to a partic appears, the host and parasite phylogenies rejoin ular biotic resource that already existed in at (parasite species 5 switches from host species D to least one plant species. That resource, in turn, host species H) and the cycle continues (from Brooks is postulated to have been either plesiomorphi- and McLennan, 1991). SEPTEMBER 1991 PARASITES AND SEXUAL SELECTION 279

H tionary evidence supportingthe Hamilton-Zuk hypothesis. The interaction between a host Gilt parasite arms-race dynamic and female choice F for a male character reflecting that dynamic is transmitted from ancestor z to three of its E descendants, species D and the ancestor of spe cies E + F, and species E, maintaining sexual dimorphism in those species. A novel defense c character appears in species F, allowing the host to win the arms race with the parasite, for the B time being. This, in turn, should eventually dis

A rupt the parasite-driven sexual selection dy namic as less information about male quality HOST PARASITE is incorporated in the male signal. The time FIG. 18. EVOLUTIONARY ARMS-RACE OUTCOME II required fo-~ this decay, however, will depend Once the counterdefense trait has appeared in upon a variety of factors, including the initial the parasite lineage, continued speciation will pro strength of the interaction (i.e., costs versus duce new species capable ofcolonizing host species benefits ofchoice for both males and females), bearing only the old defense. The situation depicted the degree ofdevelopment ofthe male charac here is an intermediate version of the large num ter, and the genetic-developmental constraints ber of possible patterns which may be produced; ranging from all parasites bearing the counterde upon changes in the male character. Since this fense trait (species 5, 6 and 7) back colonizing on process will be very system dependent, we must all members ofthe host lineage, to no back coloni consider two alternative explanations for the zation (from Brooks and McLennan, 1991). existence of a parasite-free, sexually dimorphic species F: (1) somethingotherthan the Hamil ton-Zuk dynamic is maintaining sexual dimor phism in this species, or (2) insufficient time ble macroevolutionary patternresults when one has passed for us to see the predicted decay in or more relatively plesiomorphic members of the relationship between the intensity of the a host clade are colonized by more recent!y de male signal and male quality. rived members of the parasite group bearing the"counterdefense" trait. In this case, host and F iItiit parasite phylogenies will demonstrate some de defense 2 gree ofincongruence and we would expect to find evidence that some parasites have "back E colonized" hosts in the clade diagnosed by the o counterdefense 1 presence of the "defense" trait (Fig. 18). defense 1 It is important to realize that one macro evolutionary outcome of an arms-race dynamic cilt is the production of parasite-free hosts, i.e., hosts that "won" the arms race. The discovery Bilt of parasite-free, but sexually dimorphic host species would not appear, at first glance, to sup A port the Hamilton-Zuk hypothesis. However, HOST PARASITE consider the situation depicted in Figure 19. In this scenario, the appearance ofa new para FIG. 19. ONE POSSIBLE OUTCOME OF AN site counterdefense character is associated with EVOLUTIONARY ARMS-RACE a host shift by the new parasite species from SCENARIO *, parasite absent because this host developed de the old host (species A) to a new host (ancestor fense 1 trait, and was never back colonized by the z). This host switch, in turn, is associated with parasite species bearing the counterdefense 1 trait; the appearance and elaboration of a sexually **, parasite absent because this host developed de selected character in ancestor z. At this point fense 2 trait and there are no extant parasite spe in the phylogeny we have strong macroevolu- cies bearing the appropriate counterdefense. 280 THE QUARTERLY REVIEW OF BIOLOGY VOLUME 66

In summary, any given episode of associa and (1) the age ofthe host-parasite relationship tion by colonization could be due to an arms and (2) the type ofhost switch. This informa race dynamic if phylogenetic patterns of host tion is crucial to the development of a strong switching are correlated with the origins of par coevolutionary theory, and to the understand ticular defense and counterdefense traits or ing of how particular coevolutionary events such episodes could be due to a resource-track may interact with the otherevolutionary forces ing dynamic if the required resource exhibits operating on that system. a plesiomorphic or convergent distribution SUMMARY amonghosts (see Brooks and McLennan, 1991 for a more extensive discussion ofmacroevolu It has not been our purpose in this paper to tionary correlates of coevolutionary models). discuss the validity of the Hamilton-Zuk hy It is important to distinguish between the two pothesis. Like any scientific idea, that question types ofhost switchingbecause they carry very can only be resolved by the collection oflarge different implications for the theoretical foun databases, and at the moment those databases dations ofthe hypothesis. Ifa macroevolution are limited. We have, instead, taken the stand ary analysis demonstrates that the host switch that the hypothesis is interesting and impor correlated with the appearance ofthe sexually tant, and thus deserves the most critical, rig selected trait fits a resource-tracking model, orous evaluation available to researchers. Such then we cannot assume a priori that there is an evaluation will enable us to uncover systems a cycle ofadaptation or coadaptation occurring in which the Hamilton-Zuk dynamic has in the host-parasite association. On its own, played an important role in the evolution of epi then, this type ofpattern does not provide sup gamic characters. This, in turn, will allow us port for the Hamilton-Zuk hypothesis. If, on to begin assembling a picture ofthe relative in the other hand, the host switch represents an fluence ofboth Fisherian and "good genes" in instance ofthe evolutionary arms race, then we tersexual selection on such evolution. have strongmacroevolutionary support for the In conclusion, we provide a protocol that existence of the tight coadaptive interaction be depicts the critical components in any study tween host and parasite required by the hy ofthe Hamilton-Zuk hypothesis. Such studies pothesis. will require both time and the cooperation of Here we would point out that there are two ethologists, parasitologists and systematists. important gaps in our knowledge. First, there The first requirement is not new to ethology, is a profound asymmetry between research on since all good studies require a healthy dollop the patterns and the processes of coevolution. oftime and patience. The second requirement Most of the models of coevolution are based will send one upon a quest for the golden fleece on microevolutionary (population genetical of science, cooperation among researchers. and population ecological) processes. This state Hamilton and Zuk have woven the threads of ofaffairs is hardly surprising, given the wealth many research programs into a general hypoth ofexperimental and field data available at that esis, the utility of which depends upon the level; compared to these studies, the number strength ofeach thread. We need phylogenies ofstudies investigating macroevolutionary pat for the host and parasite groups, evidence for terns comes in a distant, but nevertheless op intersexual sexual selection, and integrated timistic, second. This asymmetry is beginning studies ofthe patterns and processes ofcoevo to disappear as the methods for uncoveringthe lution. The result of such a quest will be, macroevolutionary patterns and for delineat minimally, the identification of systems con ing the relationship between these patterns and forming to Hamilton and Zuk's predictions, the underlying processes are developed. Sec and maximally, the development of a more pro ond, research at the population level continues found understanding of the unifying concept to center around the arms-race model while in biology: evolution. studies investigating the dynamics of the re source tracking or cospeciation components of A Hierarchical Protocolfor Identifying an evolutionary association are vanishingly Appropriate Systems for A nalysis of the small. At the moment, we have very little in Hamilton-Zuk Hypothesis formation about the difference, if any, in the (1) Find sets of sister groups, one of which is, relationship between the degree of pathology and the other of which is not, sexually di- SEPTEMBER 1991 PARASITES AND SEXUAL SELECTION 281

morphic. Perform experimental tests to de (5) For the subset ofthe species demonstrating termine the selective regime maintaining a positive correlation between male para that sexual dimorphism (i.e., establish the site load, epigamic development and female existence of female choice based upon the choice, determine: sexually dimorphic character). (a) Demography of resistance (prevalence (2) For the subset displaying female choice for and intensity data), the appropriate character, perform a phylo (b) Heritability of resistance, genetic analysis ofthe host group to estab (c) Mechanistic basis of resistance, lish the patterns of origin and diversifica (d) Relationship between parasite load and tion of the male epigamic trait. host health and vigor. (3) Perform a phylogenetic analysis of the host Steps (4) and (5) identify species in which parasite associations to determine which as the interaction between hosts and parasites sociations began before, during and after has been incorporated into an intersexual se the origin ofthe male epigamic trait. Iden lection dynamic. These species will provide tify episodes of cospeciation, resource track the strongest test systems for investigating ing and the evolutionary arms race. the mechanisms underlying this dynamic. Steps (1) to (3) identify the host-parasite sys tems that will provide the most rigorous test ACKNOvVLEDGMENTS of the hypothesis. We would like to thank Sherwin S. Desser, whose (4) Having identified species in which the epi interest in the original pap~r by Hamilton and Zuk gamic trait and host-parasite association co sparked an enthusiastic discussion group and pro originate, perform the appropriate experi vided us with a forum in which to present our ideas. ments to determine whether there is a Thanks are also due Sherwin Desser (again), Eric correlation between the intensity of para Hoberg, and Marlene Zuk for their thoughtful and sitism and the degree of development ofthe positive comments and criticisms of the original male character. Notice that we have already manuscript. This research was supported by a Nat demonstrated in Step 2 that females choose ural Sciences and Engineering Research Council of mates based upon the degree of epigamic Canada operating grant (No. A7696) to D. R. B. development.

R.EFERENCES

Anderson, R. C. 1976. Helminths. In L. A. Page --, and --. 1985. Epidemiology and genetics (ed.), Wildlife Diseases. Section II. Impact of Para in the coevolution of parasites and hosts. J.DroC. R. siticDiseases on WildlifePopulations, pp. 35-43. Ple Soc. Lond. B Bio!. Sci., 219: 281-283. num Press, New York. Andersson, M. 1982. Sexual selection, natural se Anderson, R. M. 1982. Host-parasite population lection and quality advertisement. Biol. J Linn. biology. In D. F.Mettrick and S. S. Desser(eds.), Soc., 17: 375-393. Parasites-Their World and Ours, pp. 303-312. El --. 1986. Evolution of condition-dependent sex sevier Biomedical Press, Amsterdam. ornaments and mating preferences: sexual selec Anderson, R. M., and D. M. Gordon. 1982. tion based on viability differences. Evolution, 40: Processes influencing the distribution of para 804-816. site numbers within host populations with spe Arme, C., and R. Wynne Owen. 1967. Infections cial emphasis on parasite-induced host mortali of the three-spined stickleback, Gasterosteus aculea ties. Parasitology, 85: 373-398. tusL., with the plerocercoid larvae of Schistocepha Anderson, R. M., and R. 1\1. May. 1978. Regula lussolidus (Muller, 1776), with special reference tion and stability of host-parasite population in to pathological effects. Parasitology, 57: 301-314. teractions. I. Regulatory processes.] An£m. Ecol. , Bandoni, S. M., and D. R. Brooks, 1987. Revision 47: 219-247. and phylogenetic analysis of the Amphilinidea --, and--. 1979. Prevalence of schistosome in Poche, 1922 (Platyhelminthes: Cercomeria: Cer fections within molluscan populations: observed comeromorpha). Can. J Zool., 65: 1110-1128. patterns and theoretical predictions. Parasitology, Bennett, G. F.,]. Blancou, E. M. White, and N. A. 79: 63-94. Williams. 1978. Blood parasites of some birds --,and--. 1982. Coevolution of'hosts and para from Senegal. J Wildl. Dis., 14: 67-73. sites. Parasitology, 85: 411-426. Bennett, G. F., E. C. Greiner, and W. Threlfall. 1976. 282 THE QUARTERLY REVIEW OF BIOLOGY VOLUME 66

Protozoans. In L. A. Page (ed.), WildlifeDiseases. ger. 1985. The quantitative assessment ofphylo Section II. ImpactofParasitic Diseases onWildlifePopu genetic constraints in comparative analyses: sex lations, pp. 25-33. Plenum Press, New York. ual dimorphism in body weight among primates. Berenbaum, M. R. 1983. Coumarins and caterpil Evolution, 39: 1335-1351. lars: a case for coevolution. Evolution, 37: 163-179. Clutton-Brock, T. H., and P. H. Harvey. 1984. Com Blower, S. M., and]. Roughgarden. 1989. Parasites parative approaches to investigating adaptation. detect host spatial pattern and density: a field ex In]. R. Krebs and N. B. Davies (eds.), Behavioural perimental analysis. Oecologia, 78: 138-141. Ecology: An Evolutionary Approach. 2nd ed., pp. 7 Borgia, G. 1979. Sexual selection and the evolution 29. Sinauer Associates, Sunderland. of mating systems. In M. S. Blum and M. A. Cohen, ]. E., and C. M. Newman. 1989. Host Blum (eds.), SexualSelection andReproductive Com parasite relations and random zero-sum games: petition in Insects, pp. 19-80. Academic Press, New the stabilizing effect of strategy diversification. York. Am. Nat., 133: 533-552. --. 1986. Satin bowerbird parasites: a test of the Collins, W. E., ] .. C. Skinner, M. Pappaioanou, bright male hypothesis. Behav. Ecol. Sociobiol. , 19: N. Shui-Fong Ma,]. R. Broderson, B. B. Sut 355-358. ton, and P. S. Stanfill. 1987. Infection of Aotus Borgia, G., and K. Collins. 1989. Female choice for vociferans (Karyotype V) monkeys with different parasite-free male satin bowerbirds and the evo strains of Plasmodium vivax.]. Parasito!., 73: 536 lution of bright male plumage. Behav. Ecol. So 540. ciobiol., 25: 445-454. Cone, D. K., and P. H. Odense. 1984. Pathology of Bradbury,]. W., and M. B. Andersson (eds.). 1987. five species of Gyrodactylus Nordmann, 1932 SexualSelection: 'Jesting theAlternatives. ] ohn Wiley (Monogenea). Can. ]. ZooI., 62: 1084-1088. and Sons, London. Courtney, C. C., and B. M. Christensen. 1987. Host Brandt, B. B. 1936. Parasites of certain North Caro parasite relationships of caryophyllaeid cestodes lina salientia. Ecol. Monogr., 6: 491-532. and aquatic oligochaetes: I. Host longevity and Brooks, D. R. 1976. Parasites of amphibians of the parasite intensity.]. Parasitol., 73: 1124-1132. Great Plains, II. Platyhelminths of amphibians --, and --. 1988. Host-parasite relationships in Nebraska. Bull. Univ. Nebr. State Mus., 10: of caryophyllaeid cestodes and aquatic oligo 65-92. chaetes: II. Effects of host age and mixed infec --. 1979. Testing the context and extent of host tions.]. ParasitoI., 74: 573-581. parasite coevolution. Syst. Zool., 28: 299-307. Cox, F. E. G. 1989. Parasites and sexual selection. --. 1981.Hennig's parasitological method: a pro Nature, 341: 289. posed solution. Syst. Zoo!., 30: 229-249. Cracraft, ]. 1988. Deep-history biogeography: --. 1985. Historical ecology: a new approach to retrieving the historical pattern of evolving con studying the evolution of ecological associations. tinental biotas. Syst. ZooI., 37: 221-236. Ann. Mo. Bot. Gard., 72: 660-680. Crews, A. E., and G. W. Esch. 1986. Seasonal dy --. 1988. Macroevolutionary comparisons of host namics of Halipegus occidualis (Trematoda: Hemi and parasite phylogenies. Annu. Rev. Ecol. Syst., uridae) in Helisoma anceps and its impact on the 19: 235-259. fecundity of the host snaiL]. ParasitoI., 72: --. 1990. Parsimony analysis in historical biogeog 646-651. raphy and coevolution: methodological and the Crump, M. L., and]. A. Pounds. 1985. Lethal par oretical update. Syst. Zool., 39: 14-30. asitism of an aposematic anuran (Atelopus varius) Brooks, D. R., and D. A. McLennan. 1991. Phylog by Notochaeta bufonivora (Diptera: Sarcophagidae). eny, Ecology and Behavior: A Research Program in J Parasitol., 71: 588-591. Comparative Biology. The University of Chicago Cusack, R. 1986. Development of infections of Gy Press, Chicago. rodactylus colemanensis Mizelle and Kritsky, 1967 Calentine, R. L. 1967. Larval development of four (Monogenea) and the effect on fry of Salmogaird caryophyllaeid cestodes. Proc. IowaAcad. Sci., 72: neri Richardson. J ParasitoI., 72: 663-668. 418-424. Cusack, R., and D. K. Cone. 1986. Studies on Gy Castro, G. A., and L.]. Olson. 1967. Relationship rodactylus salmonis (Yin and Sproston, 1948) between body weight and food and water intake parasitizing fry of Salvelinus fontinalis (Mitchill). in Trichinella spiralls-infected guinea pigs.J Para J Wildl. Dis., 22: 209-213. sito!., 53: 589-594. Darwin, C. 1871. The Descent ofMan, and Selection in Castro, G. A., L.]. Olson, and R. D. Baker. 1967. Relation to Sex. John Murray, London. Glucose malabsorption and intestinal histopa Davies, A.]. S.,]. G. Hall, G. A. T. Targett, and thology in Trichinella spiralis-infected guinea pigs. M. Murray. 1980. The biological significance of ]. Parasitol., 53: 595-612. the immune response with special reference to Cheverud,]. M., M. M. Dow, and W. Leuteneg- parasites and cancer.]. Parasitol., 66: 705-721. SEPTEMBER 1991 PARASITES AND SEXUAL SELECTION 283

Dawkins, R., and]. R. Krebs. 1979. Arms races be fense. In]. W. Wallace and R. L. Mansell (eds.), tween and within species. Proc. R. Soc. Lond. B Biochemical Interaction between PlantsandInsects, Vol. Bioi. Sci., 205: 489-511. 10, pp. 1-40. Plenum Press, New York. Dick, T. A. 1983. Species and interspecific variation. Felsenstein,]. 1985. Phylogenies and the compara In W. C. Campbell (ed.), 7fichinellaand Trichina tive method. Am. Nat., 125: 1-15. sis, pp. 31-73. Plenum Press, New York. Fisher, R. 1930. The Genetical Theoryof Natural Selec Dick, T. A., D. A. Dougherty, and D. L. Wassom. tion. Clarendon Press, Oxford. 1988. Tiichinella spiralisinfections of inbred mice: Freeland, W.]. 1976. Pathogens and the evolution genetics of the host response following infection of primate sociality. Biotropica, 8: 12-24. with different Trichinella isolates.] ParasitoI. , 74: Futuyma, D.]., and M. Slatkin (eds.). 1983. Coevo 665-669. lution. Sinauer Associates, Sunderland. Dobson, A. P. 1988. The population biology of Gittleman,]. L. 1989. The comparative approach parasite-induced changes in host behavior. Q Rev. in ethology: aims and limitations. In P.P. G. Bate si«, 63: 139-165. son and P. H. Klopfer (eds.), Perspectives in Ethol Donoghue, M.]. 1989. Phylogenies and the analy ogy, Vol. 8, pp. 55-83. Plenum, London. sisof evolutionary sequences, with examples from Gittleman,]. L., and M. Kot. 1990. Adaptation: seed plants. Evolution, 43: 1137-1156. statistics and a null model for estimating Edwards,]. C., and C.]. Barnard. 1987. The ef phylogenetic effects. Syst. ZooI., 39: 227-241. fects of Trichinella infection on intersexual inter Gleason, L. N. 1987. Population dynamics of Pom actions between mice. Anim. Behav. , 35: 533-540. phorhynchus bulbocolli in Gammaruspseudolimnaeus. Ehrlich, P. R., and P. H. Raven. 1964. Butterflies ] ParasitoI., 73: 1099-1101. and plants: a study in coevolution. Evolution, 18: Greiner, E. C., G. F. Bennett, E. M. White, and 586-608. R. F. Coombs. 1975. Distribution of the avian Embody, G. C. 1924. Notes on the control of Gy hematozoa of North America. Can.J. ZooI., 53: rodactylus on trout. Trans. Am. Fish. Soc. , 54: 48-53. 1762-1787. Endler,]. A., and A. M. Lyles. 1989. Bright ideas Hamilton, W. D. 1982.Pathogens as causes of genetic about parasites. 7fendsEcol. & Evol. ,4: 246-248. diversity in theirhost populations. In R. M. An Esch, G. W. 1983. The population and community derson and R. M. May (eds.), Population Biology ecology of cestodes. In C. Arme and P. Pappas ofInfectious Diseases, pp. 269-296. Springer-Verlag, (eds.), Biologyof theEucestoda, Vol. 1, pp. 80-137. New York. Academic Press, New York. Hamilton, W. D., andM. Zuk. 1982. Heritable true Esch, G. W., ]. Whitfield Gibbons, and]. E. fitness and bright birds: a role for parasites? Bourque. 1975. An analysis of the relationship Science, 218: 384-387. between stress and parasitism. Am. Midi. Nat., --,and--. 1989. Hamilton and Zuk reply. Na 93: 339-353. ture, 341: 289-290. Eshel, I., and E. Aiken. 1983. On the evolutionary Harvey, P. H., and L. Partridge. 1982. Bird colora instability of inner Nash solutions of the coevolv tion and parasites -atask for the future? Nature, ing populations.] Math. si«, 18: 123-133. 300: 480-481. Eshel, I., and W. D. Hamilton. 1984. Parent Holmes,]. C., and W. M. Bethel. 1972. Modifica offspring correlation in fitness under fluctuating tion of intermediate host behaviour by parasites. selection. Proc. R. Soc. Lond. B Bioi. Sci., 222: 1-14. In E. V. Canning and C. A. Wright (eds.), Be Evans, W. S., M. Novak, and A. Basilevsky. 1985. haviouralAspectsof Parasitic Transmission, pp. 123 Effects of environmental temperature, sex, and 143. Academic Press, New York. infection with Hymenolepismicrostoma on the liver ] aenike,].]. 1988. Parasitism and male mating suc and bile duct weights of mice. J. ParasitoI., 71: cess in Drosophila testacea. Am. Nat., 131: 774-780. 106-109. ]anovy, ]., ]r., and E. L. Hardin. 1988. Diversity Ewald, P. W. 1983. Host-parasite relations, vectors, of the parasite assemblage of Funduluszebrinusin and the evolution of disease severity. Annu. Rev. the Platte River of Nebraska.] ParasitoI., 74: Ecol. Syst., 14: 465-485. 207-213. Fallis, A. M., and S. S. Desser. 1977. On species of ]ermy, T. 1976. Insect-host plant relationships Leucocytozoon, Haemoproteus, and Hepatocystis. In coevolution or sequential evolution? Symp. Bioi. ]. P. Kreier (ed.), Parasitic Protozoa, Vol. 3, pp. Hung., 16: 109-113. 239-266. Academic Press, New York. --. 1984. Evolution of insect/host plant relation Fallis, A. M., S. S. Desser, and R. A. Kahn. 1974. ships. Am. Nat., 124: 609-630. On species Leucocytozoon. Adv. Parasitol. ,12: 1-67. Kennedy, C. E.].,]. A. Endler, S. L. Poynton, and Farris,]. S. 1970. Methods for computing Wagner H. McMinn. 1987. Parasite load predicts mate trees. Syst. ZooI., 19: 83-92. choice in guppies. Behav. Ecol. Sociobiol., 21: Feeny, P. 1976. Plant apparency and chemical de- 291-295. 284 THE QUARTERLY REVIEW OF BIOLOGY VOLUME 66

Kethley,]. B., andD. E.]ohnston. 1975. Resource Loverde, P. T.,]. Dewald, and D.]. Minchella. 1985. tracking patterns in bird and mammal ectopara Further studies of genetic variation in Schistosoma sites. Misc. Publ. Entomol. Soc. Am., 9: 231-236. mansoni. J Parasitol., 71: 732-734. Khan, R. A., and E. M. Lee. 1989. Influence of Ler Marcogliese, D. ]., and G. W. Esch. 1989. Altera naeocera branchialis (Crustacea: Copepoda) on the tions in seasonal dynamics of Bothriocephalus growth rate of Atlantic cod, Gadus morhua. J. acheilognathi in a North Carolina cooling reser ParasitoI. , 75: 449-454. voir over a seven-year period. J Parasitol., 75: Kilgore, M. W., G. L. Stewart, and N.]. Smatresk. 378-382. 1988. Oxygen uptake in mice infected with Margolis, L., G. W. Esch, ]. C. Holmes, A. M. Trichinella spiralis. J ParasitoI. , 74: 721-724. Kuris, and G. A. Schad. 1982. The use of eco Kirkpatrick, C. E., and T. B. Smith. 1988. Blood logical terms in parasitology (report of an ad hoc parasites of birds in cameroon. J. Parasitol., 74: committee of the American Society of Parasitol 1009-1013. ogists). J. ParasitoI. , 68: 131-133. Kirkpatrick, C. E., and H. B. Suthers. 1988. Mayden, R. L. 1988. Biogeography, parsimony, and Epizootiology of blood parasite infections in pas evolution in North American freshwater fishes. serine birds from central New]ersey. Can. J Zool. , Syst. Zoo!., 37: 329-355. 66: 2374-2382. Maynard Smith,]. 1978. TheEvolutionofSex. Cam Kirkpatrick, M. 1982. Sexual selection and the evo bridge University Press, Cambridge. lution of female choice. Evolution, 36: 1-12. --. 1985. Sexual selection, handicaps, and true --. 1986. Sexual selection and cycling parasites: fitness. J Theor. Bio!., 115: 1-8. a simulation study of Hamilton's hypothesis. J McLennan, D. A., D. R. Brooks, and]. D. McPhail. Theor. si«, 119: 263-271. 1988. The benefits of communication between Kluge, A. G. 1988. Parsimony in vicariance biogeog comparative ethology and phylogenetic sys raphy: a quantitative method and a GreaterAn tematics: a case study using gasterosteid fishes. tillean example. Syst. Zool., 37: 315-328. Can. J ZooI. , 66: 2177-2190. Knopf, P. M., and R. L. Coghlan. 1989. Maternal McLennan, D. A., and]. D. McPhail. 1990. Ex transfer of resistance to Schistosoma mansoni. J perimental investigations of the evolutionary sig ParasitoI. , 75: 398-404. nificance of sexually dimorphic nuptial coloura Kodric-Brown, A., and]. H. Brown. 1984. Truth tion in Gasterosteus aculeatus (L.): the relationships in advertising: the kinds of traits favored by sex between male colour and female behaviour. Can. ual selection. Am. Nat., 124: 309-323. J. Zool., 68: 482-492. Lanciani, C.-A. 1986. Reduced survivorship in Dasy McPhail,]. D., and S. D. Peacock. 1983. Some ef helea mutabilis(Diptera: Ceratopogonidae) para fects of the cestode (Schistocephalus solidus) on sitized by the water mite 7jrrellia circularis reproduction in the threespine stickleback (Gas (Acariformes: Limnesiidae), J. Parasitol., 72: terosteus aculeatus): evolutionary aspects of a host 613-614. parasite interaction. Can. J. Zoo!., 61: 901-905. --. 1988. Sexual bias in host selection by para Milinski, M. 1984. Parasites determine a predator's sitic mites of the mosquito Anopheles crucians (Dip optimal feeding strategy. Behav. Ecol. Sociobiol., tera: Culcidae). J. Parasitol., 74: 768-773. 15: 35-37. Lande, R. 1981. Models of speciation by sexual se Milinski, M., and T. C. M. Bakker. 1990. Female lection on polygenic traits. Proc. Nat. Acad. Sci. sticklebacks use male coloration in mate choice USA, 78: 3721-3725. and hence avoid parasitized males. Nature, 344: Leighton, E. A., K. D. Murrell, and L. C. Gasbarre. 330-333. 1989. Evidence for genetic control of nematode Minchella, D.]. 1985. Host life-history variation in egg-shedding rates in calves. J. Paras itoI. , 75: response to parasitism. Parasitology, 90: 205-216. 498-504. Mitter, C., and D. R. Brooks. 1983. Phylogenetic Lester, R.]. G. 1971. The influence of Schistocepha aspects of coevolution. In D. ]. Futuyma and Iusplerocercoids on the respiration of Gasterosteus M. Slatkin (eds.), Coevolution, pp. 65-98. Sinauer and a possible resulting effect on the behavior Associates, Sunderland. of the fish. Can. J. Zool., 49: 361-366. Mode, C.]. 1958. A mathematical model for the Lincicome, D. R. 1971. The goodness of parasitism: co-evolution of obligate parasites and theirhosts. a new hypothesis. In T. C. Cheng(ed.), TheBiol Evolution, 12: 158-165. ogy of Symbiosis, pp. 139-227. University Park Moll, H., and G. F. Mitchell. 1988. Analysis ofvari Press, Baltimore. ables associated with promotion of resistance and Loker, E. S., D. F. Cimino, G. A. Stryker, and L. A. its abrogation in T cell-reconstituted nude mice Hertel. 1987. The effect of size of M line Biom infected with Leishmania major. J ParasitoI. , 74: phalariaglabrata on the course of development of 993-998. Echinostoma paraensei. J. ParasitoI. , 73: 1090-1098. Moore, ]., and]. Lasswell. 1986. Altered behavior SEPTEMBER 1991 PARASITES AND SEXUAL SELECTION 285

in isopods (Armadillidium vulgare) infected with the Brno., 9: 1-48. nematode Dispharynx nasuta. ]. ParasitoI. , 72: Quinn, S. C., R. J. Brooks, and R. J. Cawthorn. 186-189. 1987. Effects of the protozoan parasite Sarcocystis Moore, J., J. Lasswell., M. Freehling, D. Horton, rauschorum on open-field behaviour of its inter and D. Simberloff. 1987. Host age and sex in re mediate vertebrate host, Dicrostonyx richardsoni.]. lation to intestinal helminths of Bobwhite quail. Parasitol., 73: 265-271. ] ParasitoI. , 73: 230-233. Rand, A. S., S. Guerrero, and R. M. Andrews. 1983. Moore,J., D. Simberloff, and M. Freehling. 1988. The ecological effects of malaria on populations Relationships between bobwhite quail social of the lizard Anolis limijrons on Barro Colorado group size and intestinal helminth parasitism. Island, Panama. In A. G. J. Rhodin and K. Mi Am. Nat., 131: 22-32. yata (eds.), Advances in Herpetology and Evolution Mulvey, M., T. M. Goater, G. W. Esch, and A. E. ary Biology, pp. 455-471. Museum of Compara Crews. 1987. Genotype frequency differences in tive Zoology, Cambridge. Halipegus occidualzs-infected and uninfected Heli Rau, M. E. 1983. Establishment and maintenance soma anceps.] Paras itoI. , 73: 757-761. of behavioural dominance in male mice infected Munger, J. C., andJ. C. Holmes. 1988. Benefits of with Trichinella spiralis. Parasitology, 86: 319-322. parasitic infection: a test using a ground squirrel --. 1984. Loss of behavioural dominance in male trypanosome system. Can.] ZooI. ,66: 222-227. mice infected with Trichinella spiralis. Parasitology, Nadler, S. A., and B. M. Honigberg. 1988. Genetic 88: 371-373. differentia tion and biochemical polymorphism --. 1985. The effects of Trichinella spiralis infec among trichomonads.] Parasitol., 74: 797-804. tion of pregnant mice on the future behavior of Nagy, Z. M., K.J. Porada, andJ. A. Anderson. 1983. their offspring.]. Parasitol., 71: 774-778. Under-nutrition by rearing in large litters delays Read, A. F. 1987. Comparative evidence supports the development of reflexive, locomotor, and the Hamilton and Zuk hypothesis on parasites memory processes in mice.]. Compo Physiol. Psy and sexual selection. Nature, 328: 68-70. chol., 91: 682-696. --. 1988. Sexual selection and the role of para Nur, N., and O. Hasson. 1983. Phenotypic plastic sites. Trends Ecol. & Evol., 3: 97-102. ity and the handicap principle.] Theor. Biol., 110: Read, A. F., and P. H. Harvey. 1989a. Reassessment 275-298. of comparative evidence for Hamilton and Zuk O'Donald, P. 1962. The theory of sexual selection. theory on the evolution of secondary sexual Heredity, 17: 541-552. characters. Nature, 339: 618-620. --. 1967. A general model of sexual and natural --, and--. 1989b. Read and Harvey reply. Na selection. Heredity, 22: 499-518. ture, 340: 105. Paepke, H. J. 1983. Die Stichlinge. A. Ziemsen Ver Read, A. F., and D. M. Weary. 1990. Sexual selec lag, Wittenberg. tion and the evolution of bird song: a test of the Pagel, M. D., and P. H. Harvey. 1988. Recent de Hamilton-Zuk hypothesis. Behav. Ecol. Sociobiol. , velopments in the analysis of comparative data. 26: 47-56. Q Rev. si«, 63: 413-440. Ressel, S., andJ.J. Schall. 1989. Parasites and showy Pennycuick, L. 1971. Quantitative effects of three males: malarial infection and color variation in species of parasites on a population of three fence lizards. Oecologia, 78: 158-164. spined sticklebacks, Gasterosteus aculeatus.]. ZooI. Richards, C. S. 1975. Genetic factors in suscepti Lond., 165: 143-162. bility of Biomphalariaglabrata for different strains Pickering, A. D., and P. Christie. 1980. Sexual differ of Schistosoma mansoni. Parasitology, 70: 231-238. ences in the incidence and severity of ectopara --. 1976. Variations in infectivity for Biomphalaria sitic infestation of the brown trout, Salmo trutta. glabrata in strains of Schistosoma mansonifrom the ] Fish si«, 16: 669-683. same geographical area. Bull. WHO, 54: 706-707. Pomiankowski, A. 1987a. Sexual selection: the hand Richards, C. S., andP. C. Shade. 1987. The genetic icap principle does work-sometimes. Proc. R. variation of compatibility in Biomphalaria glabrata Soc. Lond. B BioI. Sci., 231: 123-145. and Schistosoma mansoni.] ParasitoI. ,73: 1146-1151. --. 1987b. The costs of choice in sexual selection. Riggs, M. R., and G. W. Esch. 1987. The supra ]. Theor. ei«, 128: 195-218. population dynamics of Bothriocephalus acheilog --. 1989. Choosing parasite-free mates. Nature, nathiin a North Carolina Reservoir: abundance, 338: 115-116. dispersion, and prevalence.] ParasitoI. , 73: 877 Price, P. W. 1980. Evolutionary Biology of Parasites. 892. Princeton University Press, Princeton. Rollinson, D., and V. R. Southgate. 1985. Schisto Prokopic,J., and K. Krivanec. 1975. Helminths of some and snail populations: genetic variability amphibians, their interaction and host-parasite and parasite transmission. In D. Rollinson and relationships. ActaSci. Nat. Acad. Sci. Bohemoslov. R. M. Anderson (eds.), Ecology andGenetics ofHost- 286 THE QUARTERLY REVIEW OF BIOLOGY VOLUME 66

Parasite Interactions, pp. 91-109. Academic Press, Wallace, B. M., and D. B. Pence. 1986. Population London. dynamics of the helminth community from Ruse, M. 1979. The Darwinian Revolution. The migrating blue-winged teal: loss of helminths University of Chicago Press, Chicago. without replacement on the wintering grounds. Schall,J.J., andM. D. Dearing. 1987. Malarial par Can. ]. ZooI., 64: 1765-1773. asitism and male competition for mates in the Ward, P. I. 1988. Sexual dichromatism and parasit western fence lizard, Sceloporus occidentalis. Oecolo ism in British and Irish freshwater fish. Anim. Be gia, 73: 389-392. hav., 36: 1210-1215. Schall, J. J., and G. A. Sarni. 1987. Malarial para Wassom, D. L., T. A. Dick, N. Arnason, D. Strick sitism and the behavior of the lizard, Sceloporus land, and A. W. Grundmann. 1986. Host occidentalis. Copeia, 1987: 84-93. genetics: a key factor in regulating the distribu Schall,J.J., A. F. Bennett, andR. W. Putnam. 1982. tion of parasites in natural host populations. ] Lizards infected with malaria: physiological and ParasitoI., 72: 334-337. behavioral consequences. Science, 217: 1057-1059. Whitlock, S. C. 1937. An apparent case of sexual Schmidt, G. D., and L. S. Roberts. 1985. Founda difference in resistance to parasite infections.] tionsofParasitology. 3rd ed. Times Mirror/Mosby ParasitoI., 23: 426. College Publishing, St. Louis. Wiley, E. O. 1988a. Vicariance biogeography. Annu. Scott, D. K., and T. H. Clutton-Brock. 1989. Mat Rev. Ecol. Syst., 19: 513-542. ing systems, parasites and plumage dimorphism --. 1988b. Parsimony analysis and vicariance bio in waterfowl. Behav. Ecol. Sociobiol. , 26: 261-273. geography. Syst. ZooI., 37: 271-290. Scott, M. E., and H. C. Gibbs. 1986. Long-term Wright, C. A. 1971. Flukesand Snails. Allen & Un population dynamics of pinworms (Syphacia ob win, London. velata and Aspiculuris tetraptera) in mice.] Parasitol. , Wright, K. A., E. Weidman, and H. Hong. 1987. 72: 652-662. The distribution of cells killed by Tiichinella spiralis Sitepu, P., C. Dobson, and P. J. Brindley. 1986. in the mucosal epithelium of two strains of mice. Nematospiroides dubius: multiple infections in mice ]. ParasitoI., 73: 935-939. bred for immune responsiveness.] Parasitol., 72: Yanez, D. M., and A. G. Canaris. 1988. Metazoan 559-564. parasite community composition and structure Smith, R. A., M. L. Kennedy, and W. E. Wilhelm. of migrating Wilson's phalarope, Steganopus tri 1985. Helminth parasites of the raccoon (Procyon color Viellot, 1819(Aves), from EI Paso County, lotor) from Tennessee and Kentucky.]. ParasitoI., Texas.]. ParasitoI., 74: 754-762. 71: 599-603. Zahavi, A. 1975. Mate selection -a selection for a Stock, T. M., andJ. C. Holmes. 1987. Dioecocestus handicap.]. Theor. si«, 53: 205-214. asper(Cestoda: Dioecocestidae): an interference --. 1977. The cost of honesty.] Theor. Biol., 67: competitor in an enteric helminth community. 603-605. J. ParasitoI., 73: 1116-1123. Zohar, A. S., and M. E. Rau. 1986. The role of mus Swofford, D. L., and W. P. Maddison. 1987. Recon cle larvae of Trichinella spiralis in the behavioral structing ancestral character states underWagner alterations of the mouse host.]. ParasitoI., 72: parsimony. Math. Biosci., 87: 199-229. 464-466. Tanuri, A., P. P. de Andrade, and D. F. de Almeida. Zuk, M. 1984. A charming resistance to parasites. 1985. Tiypanosoma cruzi: isolation of cloned strains Nat. Hist., 93: 28-34. and characterization of their infectivity. ]. --. 1987a. Variability in attractiveness of male ParasitoI., 71: 397-402. field crickets (Orthoptera: Gryllidae) to females. Taylor, G., and G. Williams. 1982. The lek paradox Anim. Behav., 35: 1240-1248. not resolved. Theor. Populo si«, 22: 392-404. --. 1987b. The effects of gregarine parasites, body Tomlinson, I. P. M. 1988. Major-gene models of sex size, and time of day on spermatophore produc ual selection undercyclicalnatural selection. Evo tion and sexual selection in field crickets. Behav. lution, 42: 814-816. Ecol. Sociobiol., 21: 65-72. Trivers, R. L. 1972. Parental investment and sexual --.1988. Parasite load, body size, and age of wild selection. In B. Campbell (ed.), Sexual Selection caught male field crickets (Orthoptera: Grylli andtheDescent ofMan, 1871-1971, pp.136-179. AI dae): effects on sexual selection. Evolution, 42: dine, Chicago. 969-976. Wakelin, D. 1978. Genetic control of susceptibility --. 1989. Validity of sexual selection in birds. Na and resistance to parasite infection. Adv. Parasitol. , ture, 340: 104-105. 16: 219-308.