Copyright © 1995 Adrianna Smyth.
Do not use or copy without permission of the author.



Adrianna Smyth

IB 160: Evolution

20 November, 1995

I. Introduction

The concept of sexual selection was first presented by Charles Darwin (1859, 1871) to explain the existence of traits, seen primarily among male animals, that appear to contradict natural selection. Such traits include, for example, the exaggerated tail feathers of male peacocks and the large horns and antlers of many artiodactyls. The evolution of these traits is difficult to explain using natural selection; the tail of peacocks presumably increases their risk of predation, while the large horns of a ram are energetically expensive but serve no obvious purpose with respect to survival. Darwin suggested that such traits arise by means of a second form of selection, sexual selection, which results when individuals (usually males) compete for reproductive access to the opposite sex. He identified two forms of sexual selection. Intrasexual selection involves fighting or other aggressive interactions between males, as well as the development of elaborate structures to serve as weapons. This mechanism for the evolution of exaggerated male traits is generally accepted by evolutionary biologists. The second form of sexual selection suggested by Darwin is intersexual (or epigamic, Huxley 1938) selection. Darwin observed that sexual selection is often driven by female preference for particular males within the breeding population, creating variability in male reproductive success. He proposed that competition among males to attract the attention of females leads to the elaboration of structures or behaviors that attract females, and assumed that females show a sexual preference for males exhibiting a particular type of ornamentation or behavior.

Table 1: Proposed mechanisms for the evolution of mating preferences in polygynous animals (Kirkpatrick and Ryan 1991).

1. Direct selection of preference
A. Males provide resources to females or offspring
B. Costs of searching for mates
C. Selection against hybridization
D. Males differ in sperm fertility
E. Pleiotropic effects of preference genes*
F. Disease or parasite transmission

2. Indirect selection of preferences
A. Runaway processes
B. Good genes
(i) Host-parasite coevolutionary cycles
(ii) unconditionally advantageous mutations
(iii) unconditionally deleterious mutations
C. Genetic epistasis and dominance
D. Social system
E. Mutation pressure of trait

3. Other mechanisms
A. Random genetic drift
B. Group selection
C. Mutation pressure on preference

*including pre-existing sensory bias

Though little interest was invested in sexual selection during the century following Darwin's work (with notable exceptions such as Fisher, 1930, 1958) this topic has experienced a resurgence of interest among evolutionary biologists in the last several decades. A central issue emerging from this work is the evolution of female preferences for specific traits, particularly in systems that are not resource-based. In recent years, detailed mathematical models have allowed the identification of a variety of evolutionary forces that could potentially cause females to prefer exaggerated male traits; these are listed in Table 1 (after Kirkpatrick and Ryan 1991). Of the mechanisms given in Table 1, three have received the vast majority of attention: direct selection on female preferences, indirect selection of preferences via a "runaway process," and indirect selection of preferences due to "good genes." Direct selection on mating preferences arises whenever the preference leads to increased female survival or fecundity. Females preferring bright or otherwise conspicuous males, for example, may spend less time and energy searching for mates (Kirkpatrick and Ryan 1991). Theoretical studies indicate that direct selection will favor those preferences leading to an increase in the average fitness of females, regardless of impacts on male fitness (Kirkpatrick and Ryan 1991). Indirect selection via a runaway process was first proposed by Fisher (1930, 1958). This mode of evolution depends on the evolutionary exaggeration of a character that is genetically correlated to the mating preference. Evolution of this character then causes the preference to become exaggerated as a side-effect. The development of the male character and female preference then advance together ever more rapidly until checked by natural selection. At the heart of this mechanism is the assumption that the relationship between female choice and the preferred male character is arbitrary with respect to male viability, and that the only advantage to the female is that her offspring will have the preferred trait. Models of indirect selection via good genes are based on the assumption that preferences for males with genes that enhance viability are favored by evolution. By mating with a male having a high fitness, a female gains an evolutionary advantage by passing those genes on to her offspring (Trivers 1972, Zahavi 1975). Good genes models in general pose a serious theoretical problem because directional selection for male traits should eventually eliminate heritable variation in those traits (e.g. Falconer 1987, but see Heywood 1979, and Pomiankowski 1987). One specific good genes model, however, the parasite hypothesis (Hamilton and Zuk 1982), provides a possible mechanism for the maintenance of heritable variation. According to this model, the expression of particular male traits is associated with resistance to parasites or other pathogens.

Females thus use these male traits as indictors of heritable parasite resistance. This establishes a genetic correlation between preference and resistance genes so that the evolution of greater parasite resistance also causes the evolution of more extreme preferences, and the male trait will become exaggerated as a result. The critical component of this hypothesis that differentiates it from other good genes models is that the male trait and female preference never reach an equilibrium because genes for resistance constantly change in response to parasite adaptation. In other words, because parasites and their hosts coevolve, the most fit host genotype may oscillate between generations, creating variability in male traits associated with resistance.

Though a growing body of empirical research regarding the relative importance of the models discussed above is now available, the primary factors responsible for the evolution of female choice remain unclear and controversial (Kirkpatrick and Ryan 1991). In this paper I will investigate potential explanations for these difficulties using the Hamilton-Zuk parasite hypothesis as an example. Since its inception, this model has generated more interest and controversy than any other mechanism of female preference. Despite the attention it has garnered, however, I will demonstrate in this paper that tests of this hypothesis have thus far not contributed significantly to our understanding of sexual selection. Studies to date have either found no or very limited and equivocal support for this hypothesis. They have, however, served to highlight the major issues requiring resolution if progress is to be made in understanding the evolution of female preferences. The purpose of this paper is to provide a critical synthesis of the major problems characterizing studies of the role of parasites in the evolution of female preference. In particular, I will:

€ briefly review theoretical and empirical tests of the Hamilton-Zuk hypothesis

€ critically evaluate the state of our current knowledge regarding the role of parasites in the evolution of female preference

€identify the major issues that need to be resolved before conclusive results can be obtained

€suggest an empirical framework to guide future studies of the evolution of female preference

While other reviews of the parasite hypothesis have been published (e.g. Kirkpatrick 1987, Read 1988, Moller 1990, Kirkpatrick and Ryan 1991), they have tended to be narrow in focus and have not clearly and completely elucidated future research needs. In this paper, I hope to place the Hamilton-Zuk hypothesis into the more comprehensive framework of alternative models of the evolution of female preferences and suggest an avenue of research that should prove more productive than those investigated thus far.

II. Fundamental predictions and assumptions of the Hamilton-Zuk
As described above, the parasite hypothesis assumes that: 1) females choose mates on the basis of secondary sex characters; 2) the full expression of these characters is limited by parasite infection; 3) females choose males with exaggerated secondary traits in order to obtain resistance genes for their offspring; and 4) heritable variation in parasite resistance is maintained due to coevolution between parasites and hosts (Hamilton and Zuk 1982). Two major predictions arise from this hypothesis. According to the intraspecific prediction, females should preferentially mate with the brightest or most ornate males available, and these males should carry lower parasite loads than less showy males. The interspecific prediction states that across a variety of species there should be a positive correlation between parasite load and male showiness, since the opportunity for selection will be higher in those species likely to be parasitized in the first place (Hamilton and Zuk 1982). Both of these predictions have been investigated both theoretically and empirically, and the results of these investigations are presented below.

III. Tests of the Hamilton-Zuk hypothesis
Empirical tests of the Hamilton-Zuk hypothesis have generally evaluated either the intra- or interspecific prediction of the model separately. The following section will briefly summarize the studies of each type undertaken to date. Before addressing empirical results, however, I will review the results of genetic models developed to determine the conditions under which, if any, the Hamilton-Zuk model might operate. These results are important because they identify a number of constraints that are likely to severely limit the generality of this model, and thus provide an important backdrop for the empirical findings.

A. Genetic Models
In modeling the parasite hypothesis, researchers have sought to evaluate the feasibility of two basic components of the model. First, they've tried to determine whether or not host-parasite cycles actually can maintain heritable variation. Second, they've attempted to demonstrate that characters revealing a male's genetic ability to resist parasites will spread through a population because of the evolution of female preferences for males with those characters (Read 1988).

With respect to the first issue, a number of models have shown that host-parasite genotype fluctuations are one possible, but not necessarily common, outcome of host-parasite coevolution (Hamilton 1980, 1982, Eschel and Hamilton 1984). It is important to note that simply showing that coevolution may result in genotype fluctuations is necessary but not sufficient for the maintenance of variation. For example, the rate of allelic fixation, which in turn will vary with the strength of selection, as well as non-deterministic factors, may be very high relative to the period of genotype fluctuation. Only under specific conditions, then, will fluctuating selection pressure maintain heritable variation (Hamilton 1980, 1982, Eshel and Hamilton 1984, Kirkpatrick 1986). With respect to the likelihood of a male trait spreading through the population as a result of female preference, models presented by Kirkpatrick (1986) and Pomiankowski (1987) have shown that heritable variation, as discussed above, increases the likelihood that the trait will be spread via a Fisherian runaway process. This is an important result, as it indicates that runaway and good gene models are not mutually exclusive, but that a good genes scenario, such as that depicted in the parasite model, may initiate a runaway process. If this is a general result, field studies may reveal nothing about the initial role of parasites; indeed, this could explain many of the equivocal empirical results generated so far. Interestingly, one empirical study (Gilburn and Day 1994) has suggested that both models are in operation within different populations of a single species of kelp fly. Another somewhat problematic result of the Kirkpatrick and Pomiankowski models is that both require some additional mechanism, such as pleiotropy or genetic drift, to lift female preference above a threshold frequency within the population to begin the process of exaggeration. This problem is common to many models of sexual selection, and arises because there is little or no mating advantage to be gained by the possession of a secondary sexual character when the preference for it is rare (Read 1988). In summary, simple models show that the Hamilton-Zuk model can work, but only under fairly constrained conditions.

Additional modeling efforts have been directed specifically towards the intraspecific and interspecific predictions of the model. Poulin and Vickery (1993) investigated the intraspecific prediction and showed that only in cases where parasites are abundant and show low levels of aggregation with respect to distribution across hosts will there be sufficient variability to allow female choice of bright males to evolve. However, Anderson (1982) and Dobson and Merenlender (1991) have documented that aggregation of parasites among hosts is the usual situation, though levels of aggregation may vary in space and time. Furthermore, sufficient parasite-induced variability in brightness among breeding males will only occur in host-parasite systems in which the pathology of parasites is linearly related to the number of parasites infecting each host (Poulin and Vickery 1993).

These results may greatly limit the generality of the Hamilton-Zuk hypothesis and suggest that only a small fraction of host-parasite associations could promote the evolution of host mate choice for resistance based on brightness or showiness. With respect to the interspecific prediction of the model, Clayton et al. (1992) evaluated the relationship between the opportunity for selection (I = population variance in relative fitness) and two measures of parasite load (prevalence and intensity). Their results indicate that a positive correlation between I and parasite load (supporting the interspecific prediction) occurs under some conditions, but not others, depending on the exact nature of the host-parasite relationship. Clayton et al. concluded that the interspecific prediction itself is unwarranted, since the relationship between parasites and sexual selection, and thus the relationship between parasites and brightness or showiness, is highly dependent on the nature of the host parasite relationship, which is expected to vary considerably among species.

To summarize the modeling efforts directed towards the Hamilton-Zuk hypothesis, though the model itself and its individual predictions have been shown to be tenable, they have been shown to be so only within narrow ranges of conditions. Given the paucity of information available regarding the details of host-parasite relationships in most systems, it is impossible to evaluate the frequency with which these conditions are met in nature. While this does not preclude the possibility that this model operates at some times in some systems, its role as a general model must be questioned. At the very least, these results indicate that information regarding the nature of the host-parasite interaction must be carefully evaluated before a system is chosen for a study of the Hamilton-Zuk hypothesis.

B. Intraspecific Tests
A number of empirical tests of the intraspecific prediction of the Hamilton-Zuk hypothesis have been conducted, and these are summarized in Table 2. Thorough tests of the intraspecific prediction must address at least the following four premises: 1) that there is a negative correlation between an individual's fitness and its parasite load; 2) that there is heritable variation in parasite resistance (and then at those levels of parasitism found in natural populations and at which parasites have negative effects on hosts); 3) that the expression of secondary sexual characters varies with parasite burden; and 4) that preferred mates have fewer parasites. Evidence supporting 3) and 4) provides support for the hypothesis, but negative evidence can only be considered conclusive evidence against the model if 1) and 2) are true (Read 1988). The columns included in Table 2 indicate which of these assumptions are directly addressed in each published test of the intraspecific prediction. I have included several additional columns because I feel that while a satisfactory evaluation of the four premises listed above is necessary for a convincing test of the parasite hypothesis, it is not sufficient to conclude the hypothesis is correct. A variety of other considerations must be taken into account, and will be discussed in a later section of this paper. At this point, however, it's important to note that while much of the difficulty in interpreting tests of the Hamilton-Zuk hypothesis result from inherent difficulties in attempting to differentiate between predicted outcomes of runaway versus good genes models, the intraspecific model generates a unique prediction that allows this distinction to be made. According to the parasite hypothesis, showy traits should be disproportionately affected by parasite load compared with non-showy traits. Traits that are showy but not affected by parasite load are more likely to have resulted from runaway selection or an alternative mechanism (Endler and Lyles 1989).

TABLE 2. INTRASPECIFIC STUDIES OF THE HAMILTON-ZUK HYPOTHESIS. Blanks indicate assumption was not evaluated.
ReferenceOrganismParasite CostlyHeritable variationOrnament depends on parasiteFemales choose males with fewer parasitesUnique prediction testedParasite aggregation known
Zuk 1987, 1988cricketsYESNONONONO
Jaenike 1988fruitflyYESYESNONO
Kennedy et al 1987guppyYESYESYESNO
McMinn 1990guppyYESNO
Milinski and Baker 1990sticklebackYESYESYESNONO
Hausfater et al. 1990gray tree frogsYESYESNONO
Tinsley 1990spadefoot toadYESNONONO
Ressel and Schall 1989fence lizardsYESYESNONO
Hilgarth 1990 pheasantsYESYESYESYESNONO
Zuk et al . 1990jungle fowlYESYESYESYESNONO
Johnson and Boyce 1990sage grouseMAYBEYESYESNONO
Gibson 1990sage grouseNONONONONO
Clayton 1990rock dovesYESNONO
Moller1990barn swallowsYESYESYESYESNONO
Borgia1986; Borgia and Collis 1989bowerbirdsYESYESYESNONO
Pruett-Jones et al. 1990birds of paradiseYESYESNONO
ReferenceOrganismParasite CostlyHeritable variationOrnament depends on parasiteFemales choose males with fewer parasitesUnique prediction testedParasite aggregation known

It's apparent from Table 2 that many tests of the intraspecific prediction have failed to verify the fundamental premises of the prediction. It is also clear that researchers have generally ignored the novel prediction of the parasite model. In short, there is no single intraspecific system in which the Hamilton-Zuk hypothesis has shown to be operating.

C. Interspecific Tests
Table 3 summarizes the studies undertaken to test the interspecific prediction of the Hamilton-Zuk hypothesis The striking points that emerge from these results are that: 1) with regards to the basic test of a correlation between male ornamentation and any measure of parasite load, the data are quite mixed; 2) few studies have evaluated alternative factors such as ecological or social variables despite the fact that such considerations have been shown to be correlated with measures of parasite infection in other studies (Greiner et al. 1975, Bennett et al. 1978, Kirkpatrick and Smith 1988, Kirkpatrick and Suthers 1988); and 3) essentially no information regarding the biology of specific host parasite interactions has been included in these studies. Also, it is important to note that most studies have focused on prevalence (the proportion of individuals sampled showing any degree of infection) as the key measure of parasite load, and for the most part have ignored intensity (the number of parasites borne by infected individuals) and diversity (the number of parasite species present). Whether or not prevalence is actually the most appropriate measure of parasite load for the purpose of testing this model will be discussed in the next sections.

IV. A Synthesis of the Major Issues Needing Resolution
The studies surveyed in Tables 2 and 3 offer a tantalizing glimpse of a potential role for parasites in the evolution of female mate preference. Unfortunately, though the available data certainly implicate parasites to some degree, a variety of difficulties encountered in these studies make it impossible to assess the generality of the Hamilton-Zuk model. In this section I will pull together the problems characterizing tests of the Hamilton-Zuk model to date and point out those issues that are critical in developing a clear idea of the role of this mechanism in sexual selection.

A. Issues arising in intraspecific tests:
As is indicated by the number of empty spaces in Table 2, for no single system can we clearly say that the Hamilton-Zuk hypothesis is in operation. There are a number of reasons for this. First, researchers have simply failed to investigate all four basic premises of the intraspecific prediction, which as mentioned earlier represent a set of necessary (but I contend insufficient) conditions underlying the operation of the parasite model. In addition, even in those three studies that have shown these conditions to be met (i.e barn swallows, jungle fowl, and pheasants), the unique prediction of the Hamilton-Zuk hypothesis - that those traits used by females in mate choice should be more strongly affected by parasites than those not involved in mate choice - has not been tested, so it is impossible to exclude a runaway model of explanation for these systems. My assertion that the four underlying premises of the intraspecific model are not sufficient to prove the operation of the parasite mechanism stems in part from this need to exclude Fisherian models. It is also related, however, to the fact that it is impossible, on the basis of these four conditions, to determine whether or not parasites actually drive the evolution of female choice, or if they are simply correlated with some other trait that is actually crucial in determining female mate preference. This may be a trivial point in a sense, as such a system would probably still be explained by a good genes model, albeit one in which parasites per se aren't driving the system. Conceptually, however, it's important, since it indicates that the true mechanism cannot be discriminated on the basis of the four premises set forth in intraspecific tests. Similarly, given the results summarized in Table 2, there are considerable data indicating that parasites do, in fact, impact host fitness and that females tend to choose mates with lower parasite loads. This is cited as evidence supporting the Hamilton-Zuk hypothesis, which in fact it does. However, few of the papers cited point out that these results also support a much simpler model - direct selection on females (Kirkpatrick and Ryan 1991; see Table 1). If parasites are costly, females should increase their own fitness by recognizing males with high parasite loads and minimizing their own risk of infection by avoiding them. This hypothesis comes with its own set of assumptions, of course, none of which have been tested in this arena; the point is, however, that the data presented in the papers listed in Table 2 are consistent with both models but are unable to exclude either.

Another problem inherent in many tests of the intraspecific model relates to the ability of researchers to correctly identify those characters actually important in female mate choice. Endler and Lyles (1989) have pointed out that few studies have investigated the cues used by females to identify parasite-free males. In one system that actually provided support for all four of the basic conditions for the intraspecific hypothesis (Zuk et al. 1990) a subsequent study (Ligon and Zwartjes 1995) showed that bright male plumage coloration - a trait almost universally regarded as an important cue of mate choice among birds, and included in the 1990 study by Zuk et al. - was not important in female choice. In only 2 of the studies listed in table 2 was a trait investigated with respect to parasite load actually shown to be important in guiding female mate choice.

A number of problems pertaining to a lack of information regarding the details of host-parasite interactions plague both intra- and interspecific predictions. Fundamental to all tests of the Hamilton-Zuk hypothesis is the question of whether or not host-parasite coadaptaion cycles actually exist in nature. Though such cycles have been theoretically shown to be possible (Section III), only a single example has actually been demonstrated - that of the resistance cycle of a crop to a parasitic fungus (Barrett 1988). In addition, assuming that such cycles do exist on a general basis, the Hamilton-Zuk model will only work if the oscillating natural selection fundamental to the model is characterized by cycles having particular periodicities (Hamilton and Zuk 1982, Pomiankowski 1987). In no system studied so far have the details of the host-parasite interactions been worked out to allow an assessment of these factors. With regards to intraspecific tests in particular, a major issue needing further study involves the distribution of parasites across host individuals.

As shown theoretically by Poulin and Vickery (1993), the Hamilton-Zuk model will only work provided that parasites are abundant and not highly aggregated. The Hamilton-Zuk hypothesis implicitly assumes that all susceptible (nonresistant) individuals will become infected with parasites. This will not be the case, however, when parasites show a clumped distribution. Under such conditions, a female encountering a parasite free male has no way of assessing whether the male is actually resistant or simply hasn't been exposed to infection. The question of parasite distribution within the population has not yet been addressed in tests of the intraspecific prediction. Furthermore, the operation of the Hamilton-Zuk mechanism may depend highly on the type of parasite involved; i.e. the model only works for the "right kind" of parasites. Heritable variation will only be maintained when selection pressures on both the host and parasite is not so severe as to drive either extinct, and where there are only small differences in genotype viability within host and parasite populations so that no genotype goes to extinction. In addition, if parasite genotypes respond very rapidly to changes in host genotypes present resistance will not be a useful guide to future resistance. A lag between changes in host and parasite genotypes, such as that likely to be created by long-lived vectors or long periods of dormancy in the host is required. The parasites must, obviously, also affect the expression of secondary sex characters (Read 1988). Without more detailed information regarding host-parasite interactions, it is impossible to determine whether or not intraspecific studies have included the "right kind" of parasites.

Finally, in none of the tests of the Hamilton-Zuk hypothesis undertaken to date has information regarding the evolutionary relationship of the host-parasite association and the appearance of male secondary traits been available. This area is in critical need of investigation with regards to all tests of the Hamilton-Zuk hypothesis. If a particular male trait presumed to signal parasite resistance actually evolved before the initiation of the host-parasite association, the Hamilton-Zuk hypothesis cannot be invoked (McClennan and Brooks 1991). In addition, as mentioned earlier, it is theoretically possible that the Hamilton-Zuk mechanism has historically worked in concert with some type of Fisherian runaway mechanism. If this is true, only macroevolutionary analyses offer a hope of picking the two apart (McClennan and Brooks 1991, Ryan and Keddy-Hector 1992). There is thus a major need for a macroevolutionary, phylogenetic component to studies of the role of parasites in sexual selection, yet this area remains essentially untouched.

To summarize the contribution of tests of the intraspecific prediction toward an understanding of the relative importance of the Hamilton-Zuk hypothesis: there is nothing in the data presented so far to indicate that this mechanism is of general importance in the evolution and maintenance of female preferences. Indeed, based on the equivocal results of empirical studies and the narrow range of conditions imposed by theoretical considerations, it seems unlikely that this mechanism can explain a substantial proportion of observed cases of female preference. The issues that have been highlighted by these studies can be, however, resolved if future work focuses on those problems discussed above.

B. Issues arising in interspecific studies:
As mentioned above, some of the problems characterizing tests of the intraspecific prediction also apply to interspecific tests. However, in most cases, the consequences of these problems are much more serious when applied to the interspecific tests. For example, the fact that brightness traits are under many different selective pressures (aside from that generated by female choice) may be a confounding factor in all tests of the Hamilton-Zuk mechanism. Within a species, it may be possible to identify and perhaps control for alternative selective forces. When correlations are being drawn across a large number of species however, each with unique selective forces combining to mold male brightness traits, this confounding effect becomes impossible to evaluate, and would be expected to obscure whatever relationships do exist between parasites and male traits. Similarly, difficulties in assessing which trait or traits are actually used by females in selecting mates within a single species were discussed previously. Yet the interspecific prediction assumes that the same traits (e.g. plumage brightness) will be important across large groups of species (for example all passerine birds; Hamilton and Zuk 1982, Read 1987). Based on findings within species, this assumption seems extremely unjustified. In short, a fundamental, and I believe irresolvable, problem arising from interspecific tests is that analyses conducted among broad taxonomic groups are confounded by morphological, ecological, and historical differences that will obscure any existing relationships between male traits and parasites (Read and Harvey 1989). This problem is highlighted by findings that parasite loads have been found to vary significantly with ecological (Greiner et al. 1975, Bennet et al. 1978, Kirkpatrick and Suthers 1988) and social factors (Kirkpatrick and Smith 1988). Some studies (e.g. Johnson 1991, Pruett-Jones et al. 1991) have included ecological variables in their analyses; whether it's likely that such studies actually capture the key variables is highly questionable, however.

Another major, and again irresolvable, problem arising from the interspecific prediction is the inherent assumption that all host-parasite relationships are equal; all parasites are given equal weight regardless of the hosts position in the parasites life cycle, pathogenicity, or mode of transmission. As discussed previously, however, Clayton et al. 1993 have shown that the details of host-parasite interactions are crucial in determining the degree of correlation between heritable variation and male showiness, and Ewald (1987) has shown that modes of parasite transmission are important determinants of pathology. In general, it is clear that host-parasite interactions are complex and unique to individual associations. The assumption that all such interactions can be treated equally and compared with each other in a meaningful way seems highly unwarranted.

Thirdly, a variety of problems are associated with measures of parasite load. These problems are not unique to tests of the interspecific hypothesis, but seem more easily resolved within individual species. As evidenced by Table 3, measures of parasite prevalence are most commonly used in interspecific tests. Whether or not prevalence is the most appropriate measure of parasite load was not considered in any of the studies listed in Table 3. Endler and Lyles (1989) however, have pointed out that given the aggregated distribution of parasites typical of natural systems, a "wise" female should prefer males with low or intermediate parasite intensities, rather than one with no infection, since a male with a light parasite load is more reliably resistant than an uninfected male who has simply not yet been infected. Whether prevalence, intensity, or some other measure of parasite load is the most appropriate measure of parasitemia probably varies between species, however, based on differences in host-parasite systems, again leaving room for confounding results in an interspecific comparison.

Table 3. Interspecific Tests of the Hamilton-Zuk Hypothesis.

Correlation Observed Between Brightness And
ReferenceOrganismPrevalenceIntensityDiversityPhyloge netic Effects ContolledAlternative Factors ConsideredInfo On H-p Interaction
Hamilton and Zuk 1982North American passerines (plumage bright.and song complexity)YESNONONO
Read 1987 European passerines (plumage bright.)YESNONONO
Read and Harvey 1989North American passerines (plumage bright.)NOYESNONO
Read and Weary 1990North American passerines (song complexity)NOYESNONO
Zuk 1990Neotropic al birds (plumage characters)YESNONO
Pruett- Jones et al. 1991New Guinea birds (plumage characters)YESNOYESYESYESNO
Weathe rhead et al. Wood warblersNONONOYESNO
Ward 1988British fish (degree dimorph)YES
Cabana and Chandler 1991British + NA fish
Lefcort and Blaustein 1991Lizards (brightness)NO (neg correl)NOYESYESNO

Weatherhead et al. (1991) documented an additional complication when they found that substantial variation in parasite prevalence between geographic areas. Because most interspecific studies to date have drawn on reports of parasite loads collected over broad geographical areas, the results may be compromised by this type of variation. Unlike some of the problems discussed above, however, it should be possible to control for this problem by limiting the geographic range across which samples are collected, though the appropriate scale for inclusion is likely to be problematic.

Finally, an early criticism of interspecific tests of the Hamilton-Zuk hypothesis was that phylogenetic affects were not considered; that is, phylogenetically related species don't represent statistically independent samples (Felsenstein 1985, Pagel and Harvey 1988). Later studies controlled for this type of phylogenetic affect in a variety of ways (e.g. Read and Harvey 1988, Johnson 1991, Lefcort and Blaustein 1991, Pruett-Jones et al. 1991). However, another area of phylogenetic inquiry has been all but ignored. McClellan and Brooks (1991) suggested that a macroevolutionary approach was critical to clarifying the interspecific test of the Hamilton-Zuk hypothesis. They suggested that this approach be used to determine the likelihood that the parasite hypothesis has been historically important by creating phylogenies relating the appearance of showy brightness traits with patterns of novel host-parasite infections. Despite problems with phylogenetic approaches of this kind (e.g Oakes 1992), and with obtaining phylogenetic information pertaining to host-parasite associations, this approach, when combined with additional intraspecific studies, seems quite valuable, but has not yet been explored.

To summarize contributions offered by tests of the interspecific predication in clarifying the role of the parasite mechanism in the evolution of female preference, I assert that the interspecific prediction of this model has generated very little other than confusion. The few positive tests of this hypothesis that exist have been heralded as solid support for the Hamilton-Zuk hypothesis, yet the arguments given above should be sufficient to show that any correlation between parasite load and male traits observed across species almost certainly represents a spurious result. I therefore feel that future investigations of this model should not include tests of the interspecific prediction at all, unless it can be shown that all of the problems discussed in the section above can be accounted for - a condition highly unlikely to be met.

V. Conclusions: Putting the Hamilton-Zuk Mechanism in its Rightful Place:
I have shown in this paper that if the Hamilton-Zuk parasite model operates at all in nature, it probably only does so under very limited circumstances, and cannot thus be considered a general mechanism underlying the evolution of female mate preference in non-resource-based systems. I argue however, that the intraspecific and macroevolutionary components of the parasite hypothesis are still very worthy of study, as they have the potential to reveal a wealth of information concerning not only ways in which different models of female preference evolution may interact in time and space, but also details of the evolutionary relationships between hosts and parasites. What is required, however, is an integrated, systematic approach that will guide us towards those systems in which the Hamilton-Zuk hypothesis is most likely to be important and which will allow us to place this model in its proper place (with respect to relative importance) among the other mechanisms that have been suggested. Such an approach will also provide a framework for assessing the relative roles of other mechanisms so that we might someday be able to generalize about this phenomenon.

Figure 1 depicts an extremely simple conceptual model of how such an integrative approach might proceed. Of the mechanisms listed in Table 1, the first group, dealing with direct selection on females, has so far received the most empirical support. Because of this, and because of it's apparent simplicity, I suggest that this mechanism be the first to be evaluated in the study of a particular system. This model was not even considered in any of the studies listed in Tables 2 and 3, though it's easy to imagine how it could work to explain such traits as bright plumage; a female bird in the jungle might be better able to see a bright male more readily and could thus save time and energy in searching for a mate (for an elaboration of this argument see Gross and Reynolds, 1990). This idea has gained support from recent developments regarding the "pre-existing sensory bias" model put forth by Ryan (1990, Kirkpatrick and Ryan 1991). According to this hypothesis, male traits evolve to match pre-existing sensory biases that have evolved in females in response to direct selection. Several recent empirical studies (Ryan 1985, Bosolo 1990, Ryan 1990ab) have revealed the operation of pre-existing female bias, particularly in fish and amphibians. Combining these data with phylogenetic analyses of relationships between female preference and trait development has yielded the most concrete and exciting information yet to arise in this area.

Direct selection on females, including pre-existing sensory biases, thus seems like the simplest and most productive place to start in an evaluation of preference evolution. If direct selection on females can be conclusively rejected, it would seem reasonable to move on to consider runaway or good genes models. The above analysis of the problems characterizing tests of the Hamilton-Zuk hypothesis highlight considerations that would need to be undertaken at this point. A thorough analysis of traits actually used by females in choosing mates would be required, and optimally a genetic correlation between the chosen traits and female preference could be established (e.g. Gilburn et al. 1994, Houde 1994), as would be heritability of important male traits. If these conditions are met, unique predictions of the runaway and good genes models could be tested to differentiate between the two. The value of phylogenetic analyses has been shown clearly in studies by Ryan and his colleagues, and would be invaluable in each step of this evaluation. This simplistic model obviously doesn't include many of the other mechanisms suggested in Table 1, but as these are more strictly formalized they could easily be incorporated into this scheme. The essential idea is to evaluate the simplest, most parsimonious mechanisms first, before turning to more complex and difficult to test ideas.

Questions regarding the evolution of female mating preferences continue to provide an arena for the exploration of many fascinating evolutionary ideas. Following a systematic approach in the analysis of alternative models should serve to maximize the yield of new understanding resulting from each study, something sorely lacking in previous attempts to "prove" the Hamilton-Zuk hypothesis.

References Cited

Andersson, M. 1982. Female choice selects for extreme tail length in a widowbird. Nature 299: 818-820.

Arnold, S.J. 1985. Quantitative genetic models of sexual selection. Experientia 41: 1296-1310.

Barrett, J.A. Frequency-dependent selection in plant-fungal interactions. Phil. Trans. R. Soc. Lond. B319: 473-483.

Basolo, A. 1990. Female preference predates the evolution of the sword in swordtails. Science 250: 808-810.

Bennett, G.E., J. Blancou, E.M. White, and N.A. Williams. 1978. Blood parasites from some birds in Senegal. Jnl. Wild. Dist. 14: 67-73.

Borgia, G. and K. Collis. 1990. Parasites and bright male plumage in the satin bowerbird. Amer. Zool. 30: 279-285.

Boyce, M.S. 1990. The red queen visits sage grouse leks. Amer. Zool. 30: 263- 270.

Clayton. D.H. 1990. Mate choice in experimentally parasitized rock doves: lousy males lose. Amer. Zool. 30: 251-262.

Clayton, D.H., S.G. Pruett-Jones, and R. Lande. 1992. Reappraisal of the interspecific prediction of parasite-mediated sexual selection: opportunity knocks. J. Theor. Biol. 157: 95-108.

Darwin, C. 1871. The Descent of Man and Selection in Rlation to Sex. London, John Murray.

Endler, J.A., and A.M. Lyles. 1989. Bright ideas about parasites. Trends Ecol. Evol. 4: 246-248.

Eschel, J. and W.D. Hamilton. 1984. Parent-offspring correlation in fitness under fluctuating selection. Proc. R. Soc. Lond. B 122: 1-14.

Ewald, P.W. 1987. Transmission modes and evolution of the parasitism- mutualism continuum. Ann. N.Y. Acad. Sci. 503: 209-305.

Falconer, D.S. 1981. Introduction to Quantitative Genetics. Longman.

Felsenstein, J. 1985. Phylogenies and the comparative method. Am. Nat. 125: 1-15.

Fisher, R.A. 1930. The Genetical Theory of Natural Selection. Clarendon Press. Oxford.

Gilburn, A.S. and T.H. Day. 1994. Evolution of female choice in seaweed flies: fisherian and good genes models operate in different populations. Proc. R. Soc. Lond. B 255: 159- 165.

Greiner, E.C., G.F. Bennett, E.M. White, and R.F. Coombs. 1975. Distribution of the avian haematozoa of North America. Can. Jnl. Zool. 53: 1762-1787.

Hamilton, W.D. 1982. Pathogens as causes of genetic diversity in their host populations. In R.M. Anderson and R.M. May (eds.) Population Biology of Infectious Diseases. pp 269-296. Springer, New York

Hamilton, W.D. and M. Zuk. 1982. Heritable true fitness and bright birds: a role for parasites? Science 218: 384-387.

Hausfater, G., H.C. Gerhardt, G.M. Klump. 1990. Parasites and mate choice in gray treefrogs, Hyla versicolor. Amer. Zool. 30: 299-311

Hedrick, A.V. 1988. Female choice and the heritability of attractive male traits: an empirical study. Am. Nat. 132 267-276

Heywood, J. 1989. Sexual selection by the handicap mechanism. Evolution 43: 1387-1397.

Hillgarth, N. 1990. Parasites and female choice in the ring-necked pheasant. Amer. Zool. 30: 227-233.

Houde, A.E. 1994. Effect of artificial selection on male colour patterns on mating preference of female guppies. Proc. R. Soc. Lond. B 254: 125-130.

Jaenike, J. 1988. Parasitism and male mating success in Drosophila testacea. Amer. Nat. 131: 774-780

Kirkpatrick, C.E. and T.B. Smith. 1988. Blood parasites of birds in Camaroon. Jnl. Parasitol. 74: 1009-1013.

Kirkpatrick, C.E. and H.B. Suthers. 1988. Epizootiology of blood parasite infections in passerine birds from central New Jersey. Can. Jnl. Zool. 66: 2374-2382.

Kirkpatrick, M. 1986. Sexual selection and cycling parasites: a simulation model of Hamilton's hypothesis. Jnl. theor. biol. 119: 263-271.

Kirkpatrick, M. 1987. Sexual selection by female choice in polygynous animals. Ann. Rev. Ecol. System. 18: 43-70.

Kirkpatrick, M. and M.J. Ryan. 1991. The evolution of mating preferences and the paradox of the lek. Nature 350: 33-38.

Lefcort, H. and A.R. Blaustein. 1991. Parasite load and brightness in lizards: an interspecific test of the Hamilton-Zuk hypothesis. Jnl. Zool. 224: 491-499.

Ligon, J.D. and P.W. Zwartjes. 1995. Ornate plumage of male red junglefowl does not influence mate choice in females. Anim. Behav. 49: 117-125.

McClennan, D.A. and D.R. Brooks. 1991. Parasites and sexual selection: a macroevolutionary perspective. Quart. Rev. Biol. 66: 255- 286.

McMinn, H. 1990. Effects of the nematode parasite Camallanus cotti on sexual and non-sexual behaviors in the guppy (Poecilia reticulata). Amer. Zool. 245-249.

Milinski, M. and T.C.m. Bakker. 1990. Female sticklebacks use male coloration in mate choice and hence avoid parasitized males. Nature 344: 330-333.

Moller, A.P. 1990. Parasites and sexual selection: current status of the Hamilton and Zuk hypothesis. J. Evol. Biol. 3: 319-328.

Oakes, E.J. 1992. Lekking and the evolution of sexual dimorphism in birds: comparative approaches. Am. Nat 140: 665-684.

Pagel, M.D. and P.H. Harvey. 1988. Recent developments in the analysis of comparative data. Q. Rev. Biol. 63: 413-440.

Partridge, L. and P. Harvey. 1986. Contentious issues in sexual selection. Nature 323: 580-582.

Pomiankowski, A. 1987. Sexual selection: the handicap principle does work - sometimes. Proc. R. Soc. Lond. 231: 123-145.

Poulin, R., and W.L. Vickery. 1993. Parasite distribution and virulence: implications for parasite-mediated selection. Behav. Ecol. Sociobiol. 33: 429-436.

Pruett-Jones, S.G., M.A. Pruett-Jones, and H.I. Jones. 1990. Parasites and sexual selection in birds of parasites. Amer. Zool. 30: 287-298.

Pruett-Jones, S.G., M.A. Pruett-Jones, and H.I. Jones. 1991. Parasites and sexual selection in a New Guinea avifauna. Current Ornith. 7: 22-43.

Reynolds, J.D. and M.R. Gross. 1990. Costs and benefits of female mate choice: is there a lek paradox? Am. Nat. 136 230: 243.

Read, A.F. 1988. Sexual selection and the role of parasites. Trends Ecol. Evol. 3: 97-102.

Read, A.F., and P.H. Harvey. 1989. Reassessment of the comparative evidence for the Hamilton and Zuk theory on the evolution of secondary sexual characters. Nature 339: 618-619.

Ressel, S., and J.J. Schall. 1989. Parasites and showy males: malarial infection and color variation in fence lizards. Oecologia 78: 158-164.

Reynolds, J.D. and M.T. Gross. 1990. Costs and Benefits of female mate choice: is there a lek paradox? Am. Nat. 136: 230-243.

Ryan, M.J. 1983. Sexual selection and communication in a Neotropical frog, Physalaemus pustulosus. Evolution 37: 261-272.

_________.1985. The tungara frog, a study in sexual selection and communication. University of Chicago Press, Chicago.

_________.1990a. Signals, species and sexual selection. Amer. Zool. 78: 46-52.

_________. 1990b. Sexual selection, sensory systems, and sensory exploitation. Oxford Surveys in Evolutionary Biology 7: 157-195.

Ryan, M.J., and A. Keddy-Hector. 1992. Directional patterns of female mate choice and the role of sensory biases. Amer. Nat. S4 -S35

Tinsley, R.C. 1990. The influence of parasite infection on mating success in spadefoot toads, Scaphiopus couchii.. Amer. Zool. 30: 313-324.

Trivers, 1972. parental investment and sexual selection. In B. Campbell (ed.) Sexual Selection and the Descent of man.1971-1971. pp 136-179. Chicago.

von Schantz, T., G. Goransson, G. Andersson, I. Froberg, M. Grahn, A. Helgee, and H. Wittzell. 1989. Female choice selects for a viability-based male trait in pheasants. Nature 337:166-169.

Ward, P.I. 1988. Sexual dichromatism and parasitism in British and Irish freshwater fishes. Anim. Behav. 36: 1210-1215.

Weatherhead, P.J., G.F. Bennett, and D. Schluter. 1991. Sexual selection and parasites in wood warblers. Auk 108: 147-152.

Zuk, M., R. Thornhill, and J. David Ligon. 1990. Parasites and mate choice in red jungle fowl. Amer. Zool. 30: 235-244.