THE DILEMMA OF FISHERIAN SEXUAL SELECTION: MATE CHOICE

FOR INDIRECT BENEFITS DESPITE RARITY AND OVERALL

WEAKNESS OF TRAIT-PREFERENCE GENETIC CORRELATION

Michael D. Greenfield1, Sylvain Alem2, Denis Limousin3,4, and Nathan W. Bailey5

Institut de recherche sur la biologie de linsecte (IRBI), CNRS UMR 7261, Universit

Franois
Rabelais de Tours, Parc de Grandmont, 37200 Tours, France
2

Research Centre for Psychology, School of Biological and Chemical Sciences, Queen

Mary
University of London, Mile End Road, London E1 4NS, United Kingdom
3

UPMC Univ Paris 06, UMR 1272, Physiologie de l'Insecte Signalisation et Communication,
F-78026, Versailles, France

INRA, UMR 1272, Physiologie de l'Insecte Signalisation et Communication, F-78026,

Versailles,
France
5

Centre for Biological Diversity, School of Biology, University of St Andrews, Fife KY16

9TH,
United Kingdom

for : Evolution

This article has been accepted for publication and undergone full peer review but has not been through the copyediting,
typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of
Record. Please cite this article as doi: 10.1111/evo.12542.
This article is protected by copyright. All rights reserved.

Abstract

Fishers mechanism of sexual selection is a fundamental element of evolutionary theory. In

it non-random mate choice causes a genetic covariance between a male trait and female
preference for that trait and thereby generates a positive feedback process sustaining
accelerated coevolution of the trait and preference, is a fundamental element of evolutionary
theory. Numerous theoretical models of Fishers mechanism have confirmed its
mathematical underpinnings, yet biologists have often failed to find evidence for traitpreference genetic correlation in populations where the mechanism was expected to
function. We undertook a survey of the literature to conduct a formal meta-analysis probing
the incidence and strength of trait-preference correlation among animal species. Our metaanalysis found significant positive genetic correlations in fewer than 20% of the species
studied and an overall weighted correlation that is slightly positive. Importantly, a significant
positive correlation was not found in any thorough study that included multiple subgroups.
We discuss several ways in which the dynamic, multivariate nature of mate choice may
reduce the trait-preference genetic correlation predicted by Fishers mechanism. We then
entertain the possibilities that Fisherian-like processes sometimes function without genetic
correlation, and that mate choice may persist in a population as long as genetic correlation,
and therefore Fishers mechanism, occurs intermittently.

KEY WORDS: female choice, genetic covariance, meta-analysis, runaway selection

Short title: The dilemma of Fisherian sexual selection

This article is protected by copyright. All rights reserved.

Fishers mechanism of sexual selection (Fisher 1915; 1930; 1958) has dominated biological
thought on the genetics of mate choice over the past 40 years. The mechanism is central to
most sexual selection models (Prum 2010) and has been formally modelled on numerous
occasions beginning in the early 1980s (ODonald 1980; Lande 1981; Kirkpatrick 1982;
Bateson 1983; Andersson 1994; Mead and Arnold 2004; Andersson and Simmons 2006;
Kuijper et al. 2012). Fisher described the process by outlining several discrete steps: First,
specific female preferences and male traits originate in a population due to genetic drift and
founder events or because they are favored in a context outside of sexual selection. The
resulting non-random matings generate an accumulating additive genetic covariance (COVA)
between genes that affect the preferred male trait and the female preference. When strong,
this covariance can generate unstable evolutionary conditions that cause rapid exaggeration
(or diminution) of both trait and preference. Fisher explained this result as a geometric
progression and termed it runaway selection. He emphasized that runaway selection
would not be expected to produce infinite trait elaboration, and that the process must
eventually be checked by opposing selection pressures.
Trait-preference genetic covariance may simply represent an emergent property of
sexual selection, as all that is required for its appearance is additive genetic variance (VA) for
traits that bias mating success and for preferences for such traits (Pomiankowski and
Sheridan 1994a; Bakker and Pomiankowski 1995). Nonetheless, COVA can play a special
role in the Fisher process by counteracting natural selection that prevents exaggeration of
the trait and preference, thereby determining whether evolving populations experience
unstable runaway dynamics (Lande 1981, Kirkpatrick 1982). While some authors suggested
that COVA is rarely strong enough to overcome direct selection on mating preferences (e.g.
Kirkpatrick and Barton 1997), others claimed an important role for COVA even when
preferences are costly (Pomiankowski et al. 1991; Day 2000; Hall et al. 2000). Because
COVA is generally implicated in modulating the evolutionary dynamics of sexual selection
under the Fisherian mechanism, detecting and measuring COVA merits special attention:
The presence and strength of trait-preference COVA represents a key test of whether the

This article is protected by copyright. All rights reserved.

Fisherian mechanism is functioning in a population (Lande 1981).

We expect that COVA will persist during periods of stable equilibrium and brief
episodes of instability as long as trait and preference VA, and a link between preference and
elevated reproductive success of males bearing the preferred trait, remain. Even when
runaway selection does proceed long enough that fixation of trait and/or preference alleles,
and the loss of COVA between those specific alleles, occur (Kirkpatrick 1982), the Fisher
process may be expected to regenerate itself anew at some point in the future. However,
the regeneration would likely involve different alleles and possibly new directions of trait and
preference evolution (Iwasa and Pomiankowski 1995; Pomiankowski and Iwasa 1998).
Thus, cycles of unstable and stable phases per se do not necessarily remove COVA, nor do
so permanently. These expectations are largely consistent with a recent report (Roff and
Fairbairn 2014) whose simulations predict low ( 0.20) but positive inter-sexual genetic
correlations. Higher correlation values may occur where females typically choose among
several males, for example in lekking species.
Whereas empirical work now confirms the existence of genes influencing male traits
and female preferences as well as VA and inheritance of traits and preferences in many
species (Andersson 1994; Boake 1994; Bakker and Pomiankowski 1995), evidence for traitpreference COVA has remained equivocal (Ritchie et al. 2005; Zhou et al. 2011). Several
initial studies on genetic correlation focused on a character, mating speed in Drosophila,
easily measured in both sexes (Gromko 1992; Stamencovic-Radak et al. 1992). Mating
speed, however, can represent a trait more indicative of choosiness than female
preference. Moreover, the intersexual genetic correlations reported in these studies were
challenged as being artifacts of the selection regime imposed during the experiments
(Nichols and Butlin 1992; Butlin 1993; Gray and Cade 1999a; cf. Stamencovic-Radak et al.
1993). However, in the following years three studies on stickleback fish (Bakker 1993),
guppies (Houde 1994), and stalk-eyed flies (Diopsidae) (Wilkinson and Reillo 1994) gave the
appearance of having avoided the pitfalls of earlier work. These studies reported positive

This article is protected by copyright. All rights reserved.

genetic correlations between sexually-selected male traits and female preferences and
collectively generated considerable enthusiasm for the Fisherian mechanism (Pomiankowski
and Sheridan 1994a). But again, suggestions of problems owing to a correlated response to
selection experimental procedure, small sample sizes, and equivocal results (Breden et al.
1994, Breden and Hornaday 1994) and rebuttals (Pomiankowski and Sheridan 1994b)
surfaced quickly. Despite these concerns, the reports vindicated the Fisherian mechanism
for many biologists, and the trait-preference graphs whose trajectory slopes reflect the
strength of COVA remained a fixture in behavioral ecology.
Several factors compel us to revisit the classical problem of trait-preference COVA.
First, nearly 20 additional studies of intersexual genetic correlation have been conducted on
various insects, fish and birds since the original reports, and these have used breeding
designs or pedigree analyses of wild populations (Table 1). Second, we now have a more
mature appreciation for mate choice as a dynamic, multivariate process (Chenoweth and
Blows 2006; Chenoweth and McGuigan 2010). This characteristic of mate choice may
greatly reduce genetic correlation in spite of the presence within a population of VA for a
male trait and a female preference. A third, and perhaps most pressing, reason for calling
attention to the genetic correlation issue is that it is likely to be confused with recent findings
on trait-preference correlation within the context of species recognition and speciation (Shaw
et al. 2011; Bousquet et al. 2012; Wiley et al. 2012; McNiven and Moehring 2013). These
findings relied on molecular genetic investigations, and they revealed either pleiotropy or
tight physical linkage between trait and preference genes. Such relationships are strong
genetic correlations, but they do not necessarily arise nor are they specifically maintained by
the linkage disequilibrium created by non-random mate choice within a species.
We undertook a formal meta-analysis of the studies that estimated trait-preference
COVA over the past 30 years. Our main objective was to evaluate the prevalence and levels
of COVA among animal species and thus clarify the potential operation of Fishers
mechanism in the evolution of sexually selected traits. To avoid potential confusion, we
distinguish between the general Fisherian mechanism, as described above, and the subset

This article is protected by copyright. All rights reserved.

of this mechanism where COVA is strong enough to generate an unstable, runaway situation.
While we acknowledge that runaway selection may arise temporarily on certain occasions,
our main focus is on the general Fisherian mechanism as a means of generating and
sustaining mate choice for genetic benefits. We found that the overall incidence and level of
trait-preference COVA are weak. The global pattern of COVA is consistent with publication
bias, although our analysis cannot definitively conclude that it is caused by publication bias.
These results present evolutionary biologists with a significant dilemma, as female
preferences for exaggerated male traits are regularly observed in nature. Moreover,
heritability of these traits and preferences, as well as evidence for indirect benefits (e.g.
Gilburn and Day 1994; Jones et al. 1998; Rundle et al. 2007; Prokop et al. 2012), has often
been documented. We address this dilemma by discussing several explanations that may
account for the curious absence of COVA where it is predicted by the Fisher process, as well
as the apparent presence of the Fisher process in cases where COVA is not detected.

Methods
STUDY SELECTION AND INCLUSION CRITERIA
We initiated our survey of studies that addressed trait-preference COVA by searching the
Web of Science using genetic correlation and mate choice, genetic correlation and
sexual selection, genetic covariance and mate choice, and genetic covariance and
sexual selection as topic combinations. We supplemented this survey with a similar inquiry
using the Google Scholar search engine to find studies presented only in books, book
chapters, dissertations, and other sources not included in the Web of Science. This latter
search yielded one additional result, a recent doctoral dissertation. In both searches we
retained only those studies that pertained to inter-sexual trait-preference correlation, and we
then checked the references cited by each retained study as well as later articles that cited
the retained studies. These checks revealed two studies that were not detected by our four
combinations of search topics. Additional searches using the terms Fisherian selection,

This article is protected by copyright. All rights reserved.

Fisherian mechanism both in combination with genetic correlation or genetic covariance

and intersexual genetic correlation did not yield any additional pertinent references. We
then eliminated studies that did not provide a quantitative estimate of trait-preference COVA
(n = 7) or that were found to address species recognition rather than mate choice (n = 6). In
further screening we chose to remove those studies that assessed trait-preference COVA by
evaluating hybrid crosses between different species or populations, or by evaluating the traitpreference correlation across multiple populations (n = 4). These criteria were applied to
ensure that our survey was restricted to measures of COVA within a single population. We
also removed studies (n = 2) that measured female choosiness, evaluated as the reciprocal
of mating speed, as opposed to female preference for certain males over others. In three
cases, we contacted authors and obtained more details on experimental design, sample
size, and standard error of COVA estimates than was available in their published article and
thereby made a more informed decision on whether to retain the study and how to represent
its effect size.

META-ANALYSIS
Our screening ultimately retained 20 studies distributed among insects (13), fish (6) and
birds (1) (Table 1). Most of these studies reported estimates of trait-preference genetic
correlation as opposed to COVA, and collectively they evaluated different sorts of traits that
were measured by various units. Thus, in our meta-analysis we only determined summary
statistics for the genetic correlation and did not conduct other tests. Six of the retained
studies relied on the correlated response to selection approach, all of which were completed
prior to 1995. Thirteen studies computed the trait-preference genetic correlation directly
from a breeding design, and one study computed the correlation from pedigree analysis of a
wild population. We analyzed the genetic correlations reported in these studies via
Comprehensive Meta Analysis version 2 (Biostat, Inc.; Englewood, New Jersey, U.S.A.; see
Borenstein et al. 2009), which estimated an overall correlation value, calculated as a
weighted mean, and its 95% confidence limits. Effect sizes of the individual studies,

This article is protected by copyright. All rights reserved.

represented by either standard error or sample size, provided the weights for determining
the overall correlation value. We applied the programs random effects model (DerSimonian
and Laird 1986), rather than its fixed effects model, for estimating means and confidence
limits. The fixed effect option assumes equal effect sizes across all studies considered.
Because we had no reason to assume a priori that effect sizes would be homogeneous in
our dataset, we deemed it more appropriate to implement the random effects option which
allows effect sizes to vary from study to study (Borenstein et al. 2009). We note that overall
correlation coefficients determined as a simple average generally underestimate the true
mean when it is positive (Silver and Dunlap 1987). To reduce this bias, we converted all
correlation coefficients with Fishers z transformation, determined the mean and confidence
limits of these transformed values, and then back-transformed these Fishers z values to an
overall correlation coefficient and its confidence limits. Because the Fisherian mechanism
specifically predicts a positive genetic correlation between a trait and a preference, we paid
particular attention to the direction of the correlation in the each study. In our meta-analysis
we listed the correlation direction of a study as positive if it was coincident with the Fisherian
prediction and negative if it was opposite. For example, in the beetle Onthophagus taurus,
Simmons and Kotiaho (2007) reported a negative genetic correlation between sperm length
and size of the female spermatheca. But larger spermatheca favour shorter sperm in this
species, and we therefore designate the inter-sexual genetic correlation as positive in
direction (Figure 1a).
Table 1 lists one study, that on the seaweed fly Coelopa frigida by Gilburn and Day
(1994), not retained for the formal meta-analysis but which deserves consideration. Data on
traits and preferences in C. frigida were collected from only three genotypes (karyotypes),
making the effect size very low. However, five different populations were studied, and flies
were sampled extensively in each population. We discuss the findings relevant to traitpreference genetic correlation in the Results below.

This article is protected by copyright. All rights reserved.

Nine of the studies retained in our meta-analysis included two or more subgroups that either
represented different populations of the same species, different generations of the same
population, replicates of the same population reared under the same or different
environmental conditions, different scoring protocols during experimentation, or correlated
responses to selection imposed in different directions. As in the estimation of the overall
correlation, we applied the random effects model in combining subgroups within a study.
We determined four basic estimates of the overall genetic correlation: 1) an estimate that
treated all 19 studies as distinct samples; 2) an estimate that grouped separate studies of
the same species as a single sample, a measure that reduced the total number of samples
to 16; 3) an estimate treating all studies as distinct samples but not including two studies
(species) in which the reported VA or heritability (hs2) for female preference was very low,
effectively rendering trait-preference COVA negligible by definition; and 4) an estimate similar
to the previous one (3) except that separate studies of the same species were grouped and
treated as a single sample.
We considered an additional estimate in which retained studies would have been
adjusted for phylogeny (Chamberlain et al. 2012) based on the assumption that samples of
very closely related species may not be independent. Whereas three species of Drosophila
(D. mercatorum, D. montana, D. simulans) appeared among our retained studies, these
three are not situated on the same or even neighboring branches in current phylogenies of
Drosophila (e.g. Van der Linde et al. 2010). Thus, we would not have been justified in
grouping the reported data (Ikeda and Maruo 1982; Ritchie et al. 2005; Ingleby et al. 2013a)
as if they represented separate studies of the same species, or in modifying their relative
weights by some comparative formula. Because our other 16 retained studies are
distributed widely among insects and vertebrates, we judged that other adjustments for
phylogeny were neither warranted nor possible.

This article is protected by copyright. All rights reserved.

PUBLICATION BIAS
We examined whether the results of each of our four meta-analyses reflected potential
publication bias by implementing two methods from Comprehensive Meta-Analysis version
2, Duval and Tweedies trim and fill (Duval and Tweedie 2000) and Eggers regression
intercept (Egger et al. 1997). For the first method, correlation values in each study were
converted to Fishers z statistic, and a funnel plot was then generated showing standard
error vs. Fishers z statistic for each of the studies included in our analysis. If this plot of
observed studies was not symmetrical around the overall z statistic, imputed studies were
added to make the funnel symmetrical, and the overall z statistic and its standard error were
recalculated using both the observed and imputed studies in the new funnel plot. Again, the
programs random effects model was applied to determine the imputed studies and the new
overall z statistic and standard error. We applied Eggers regression intercept to determine
the significance of asymmetry observed in the original funnel plot. Publication bias analysis
was conducted because of the likelihood that studies finding negative results, i.e. an
absence of trait-preference genetic correlation, were not published, particularly if their effect
size was small.

Results
For all four estimates of the overall trait-preference genetic correlation, our meta-analyses
revealed low values that were nonetheless significantly greater than 0 (Table 2; Fig. 1a).
However, our subsequent analyses in which we examined the possibility that publication
bias had influenced these estimates showed that none of them could be confirmed as
significantly positive values following adjustment by Duval and Tweedies method (Table 2;
Fig. 1b) and corroboration of this method by Eggers regression intercept (Table 2). We
acknowledge that these latter analyses can only indicate that results are consistent with
publication bias. That is, asymmetry in our funnel plots of standard error vs. Fishers z

This article is protected by copyright. All rights reserved.

10

statistic values may conceivably reflect other biases inherent in the studies retained from the
survey.
Consideration of the study on the seaweed fly Coelopa frigida by Gilburn and Day
(1994),
listed in Table 1 but not retained in the formal meta-analysis, is consistent with the overall
findings presented above. A positive regression of female preference on the male trait
among karyotypes was found in only one of the five populations sampled. These results
suggest that an overall trait-preference genetic correlation in C. frigida, given sufficient
sampling of genotypes within the populations, would most likely not be positive.
We note that a significantly positive genetic correlation was found in only three of the
species retained in the survey (meta-analysis 2; overall correlation for species was
determined by combining the separate studies using the random effects model noted
above). These three showed no taxonomic trend, as they include a moth (Utetheisa
ornatrix), a beetle (Onthophagus taurus), and a fish (Gasterosteus aculeata). Two additional
species exhibited a positive genetic correlation in one population (study) but not another, or
in one generation but not the subsequent one. This latter case, female preference for male
song length in a field cricket, merits further discussion: A broad-sense genetic correlation
determined from animals that had mated in a natural population in the field was positive and
significant (0.51 0.17; Gray and Cade 1999b), whereas a narrow-sense genetic correlation
determined from the following generation who were paired in the laboratory was negative
and non-significant (-0.20 0.42). A 50% reduction in trait-preference COVA each
generation is expected once laboratory breeding wherein the experimenter decides who
pairs with who is started (Bakker and Pomiankowski 1995; Bakker 1999; Gray and Cade
1999a), but the observed shift to a negative value is not. In summary, no thorough study,
having a large effect size owing to replicated sampling via multiple subgroups, exhibited a
significantly positive trait-preference genetic correlation.

This article is protected by copyright. All rights reserved.

11

Discussion
The low incidence of significant trait-preference genetic correlation, and the estimates
adjusted for publication bias showing that overall correlation values might not be significantly
different from 0, are surprising in light of theoretical expectations (Roff and Fairbairn 2014).
Either the measurement of trait-preference COVA is fraught with so much difficulty that
biologists have been unable to detect its presence, or the absence we report is real.
Accurate estimates of trait-preference genetic correlation are admittedly difficult to attain, yet
many biologists have recognized their importance for the Fisherian mechanism, and the
studies in our meta-analysis show that more than a few have made careful determinations of
the parameter. Since 1995, studies have avoided the correlated response to selection
approach, and most have also measured genetic correlations in natural populations or in the
laboratory immediately after animals were collected from the field. Moreover, our metaanalysis shows that the more thorough studies were less, not more, likely to report positive
findings on trait-preference genetic correlation.
Many, if not all, of the traits in our analysis may be under natural as well as sexual
selection. This probability should not be surprising given that even landmark traits long
associated with Fisherian runaway selection, e.g. male peacock plumage (Petrie 1994; cf.
Callaway 2011), have been shown to be associated with a survival advantage. We
emphasize that the Fisherian mechanism and other sexual selection mechanisms are not
mutually exclusive. Whereas the Fisherian mechanism is sufficient to explain rapid evolution
of traits with no direct benefits, it may also be both a cause and a consequence of other
mechanisms of sexual selection, e.g. good genes, compatible genes, or direct benefits.
The above points suggest that the inconsistency observed between empirical
findings and the theoretical expectation is neither an artifact of limited sampling nor results
from studying traits where Fisherian selection would not be expected. In the following
sections we consider several explanations that may account for the apparent absence of
trait-preference COVA.

This article is protected by copyright. All rights reserved.

12

MATE CHOICE AS A DYNAMIC, MULTIVARIATE PROCESS

Mate choice as a dynamic process implies that what is popular today and in region A might
not be in vogue tomorrow and in region B. Migration can be the easiest way of envisioning
the problem that changes in traits and preferences can pose for experimenters evaluating
predictions of the Fisherian mechanism. To the extent that resident females accept and pair
with immigrant males, the genetic correlation between preference and trait will be perturbed
and reduced in the subsequent generation. Thus, the building up of a genetic correlation
between a strong preference and a particular form of trait exaggeration can be interrupted
whenever migration occurs (Greenfield and Rodriguez 2004). Problems analogous to those
caused by migration will arise if environmental conditions change over time and generations
overlap. Thus, trait-preference genetic correlation might be an irregular event within a
population, one that is interrupted by immigration, generation overlap, and environmental
heterogeneity over space and time. As such, detecting trait-preference genetic correlation
can depend on when sampling is done during a populations trajectory of migration and
environmental conditions, as well as episodes of unstable runaway evolution. This issue
could even influence the study of laboratory populations subject to subtle changes in
environmental conditions and genetic drift (Blows 1999).
A particularly insidious quandary for the Fisherian mechanism that involves migration
can occur when traits and preferences exhibit phenotypic plasticity along an environmental
gradient and are therefore described by reaction norms (Fig. 2; see Greenfield and
Rodriguez 2004; Kokko and Heubel 2008). The likelihood of problems associated with
phenotypic plasticity is indicated by work on the katydid Neoconocephalus triops
(Orthoptera: Tettigoniidae), which revealed seasonal influences on male song features and
thermal influences on female preferences for those calls (Beckers and Schul 2008).
Temporal influences on both male plumage characteristics and female preferences for
plumage traits have similarly been documented in the lark bunting, Calamospiza
melanocorys (Chaine and Lyon 2008).

This article is protected by copyright. All rights reserved.

13

The multivariate nature of the mate choice process can interfere further with traitpreference genetic correlation. Male traits normally include several parameters that
influence females, a complexity that is quite evident in chemical signals in Drosophila spp.,
where a bouquet of cuticular hydrocarbons (CHCs) often influences courtship success.
Female response to CHCs does not include separate preference traits for each compound
but rather appears to be represented by one, or only a few, composite preference functions
vectors in n-dimensional CHC space (McGuigan et al. 2008). Importantly, this composite
female preference function might not align with the composite function representing the suite
of male CHCs in the population (McGuigan et al. 2008; Fisher et al. 2009). Disagreement
between preference and trait functions possibly reflects other factors influencing CHC
composition, such as the contribution of these compounds to desiccation resistance (Ingleby
et al. 2013b). Additionally, multiple signals transmitted in various modalities ultimately
contribute to a males mating success. Unless males with the most preferred chemicals also
excel in acoustic, visual, and tactile signaling, a trait-preference genetic correlation for CHCs
will be unlikely.

CAN FISHER-LIKE PROCESSES' OCCUR WITHOUT ADDITIVE GENETIC VARIANCE?

Several theoretical explanations formulated in recent years suggest implicitly that Fisher-like
processes might proceed in the absence of strong trait-preference COVA. An indirect
genetic effect (IGE) model of sexual selection (Bailey and Moore 2012) found that when the
expression of female preferences is socially flexible, the genetic conditions under which trait
evolution, including the runaway process, is favored are relaxed. This situation occurs if
genes expressed in males exert indirect effects on the expression of female preferences
through such social interactions as copying, imprinting, or learning through experiences with
mates. That social interactions can influence mating decisions is well known in birds (Galef
and White 2000; Slagsvold et al. 2002), fish (Kozak and Baughman 2009; Tinghitella et al.
2013), and insects (Dukas 2005; Mery et al. 2009; Svensson et al. 2010; Bailey et al. 2010;
Fowler-Finn and Rodriguez 2012). Studies of IGEs on sexually selected CHC blends in

This article is protected by copyright. All rights reserved.

14

Drosophila serrata have uncovered evidence that they can alter the dynamics of sexual
selection, in addition to themselves evolving (Petfield et al. 2005; Chenoweth et al. 2010).
Social effects are also likely to shape the expression of female choice, as shown by several
studies on field crickets (Judge 2010; Bailey and Zuk 2012; Kasumovic et al. 2012;
Tinghitella 2014). Such social effects on female mating decisions can generate IGEs that
affect the coevolutionary dynamics of sexually selected traits and preferences and reduce
the importance of genetic covariance in setting up unstable runaway dynamics.
A similar model developed around the theme of cultural niche construction (Rendell
et al. 2011) suggests that the Fisher process is an example of a general class of phenomena
in which positive feedback arises between two traits within a population and leads to the
rapid evolution of both. In this model one trait that is culturally transmitted is statistically
associated and coevolves with a non-culturally transmitted one. Within the context of sexual
selection, these traits might be a culturally acquired or imprinted female preference and
alleles for a male ornament, respectively. Importantly, the process can select for a costly
non-cultural trait if that trait confers an enhanced ability to deal with the environment created
by the cultural trait; e.g. an expensive male ornament that is nonetheless favored by the
prevailing, and culturally transmitted, female preference (cf. Danchin et al. 2013).

IMPLICATIONS
The dynamic, multivariate nature of mate choice implies that trait-preference genetic
correlation ought not to be strongly predicted, which is consistent with our meta-analysis
findings. However, mate choice without tangible, direct benefits is often observed in natural
populations, and a recent report suggests that the Fisherian mechanism may frequently
operate in such situations (Prokop et al. 2012). Therefore, we are forced to reconcile these
general observations of mate choice with an incidence of COVA that is quite low and an
overall COVA value that is small and possibly insignificant. Other than the species in which
IGEs or cultural niches might operate and permit Fisher-like processes to function in the
absence of strong COVA, how does mate choice occur if the Fisherian mechanism does not

This article is protected by copyright. All rights reserved.

15

regularly function? In many species mate choice may simply represent orientation toward
those signals and ornaments that evoke greater afferent activity in female receivers, the socalled sensory bias mechanism. However, where trait and preference VA but not
intersexual COVA are observed (e.g. Ritchie et al. 2005; Zhou et al. 2011), other likely
explanations are interference due to sexual conflict (Ritchie et al. 2005) or that the Fisher
process does occur but only intermittently. An irregular occurrence of the Fisher process
would be consistent with observations in several studies (e.g. Gilburn and Day 1994; Zhou et
al. 2011) in Table 1 wherein genetic covariance was found in one subgroup, representing a
particular geographic location or environmental rearing condition, but not elsewhere.
Our findings and interpretations point toward areas that demand further attention.
More information on the potential influences of migration and environmental fluctuation on
sexual selection, and how the intensity and direction of mate choice may change over time
and between neighboring populations (e.g. Gosden and Svensson 2008) would definitely be
helpful. Similarly, we need more extensive incorporation of these potential influences and
changes in modeling the Fisher process. At present it is not clear whether irregular
occurrences of the Fisher process and the resulting spikes in COVA in between migration
events, or prior to and during episodes of runaway selection would be sufficient to sustain
exaggerated traits and preferences over the long term. The role of migration is now often
considered in models of the sexual selection process (e.g. Getty 2014; Holman and Kokko
2014; Woolf et al. 2014). Additionally, we note that several earlier models incorporating
migration and environmental heterogeneity reported that these features would favor the
evolution of mate choice by conserving the trait, and preference, VA that are necessary for
sexual selection to function (Day 2000; Proulx 2001). But these models also specified that
mate choice evolution may be highest at intermediate levels of migration and heterogeneity.
Thus, their predictions may not be at odds with the perspective presented here (Fig. 2) in
which a high level of migration and heterogeneity can disrupt linkage disequilibrium, thereby
reducing COVA and weakening sexual selection.

This article is protected by copyright. All rights reserved.

16

A major outcome of our meta-analysis is that trait-preference COVA in mate choice

appears to be strikingly different from traits and preferences that afford species recognition,
which are linked genetically via pleiotropy or tight physical linkage in some cases. In
accounting for this difference, it is tempting to propose that where traits and preferences
happen to be influenced pleiotropically, allelic differences tend to favor population
divergence, and potentially speciation (Nosil 2012), rather than maintaining VA and COVA in
the mate choice process within an interbreeding population.

ACKNOWLEDGEMENTS

We thank A. Gilburn, A. Houde and N. Ratterman for answering our queries on details of
their studies, and F. Barbosa, M. Goubault, D. Gray, E. Herniou, M. Ritchie, V. Terwilliger
and B. Visser for their input on previous versions of the manuscript. Four anonymous
referees and the reviewing editor all provided detailed and thoughtful feedback that improved
the manuscript. Many ideas expressed here were inspired by research funded by the
Agence Nationale de la Recherche of France (contrat ANR-07-BLAN-0113-01; project
GENARCH), the U.S. National Science Foundation (grant IOB-0516634) and the U.K.
Natural Environment Research Council (NE/G014906/1).

This article is protected by copyright. All rights reserved.

17

LITERATURE CITED

Andersson, M. 1994. Sexual Selection. Princeton Univ. Press.

Andersson, M., and L.W. Simmons. 2006. Sexual selection and mate choice. Trends Ecol.
Evol. 21: 296-302.
Bailey, N.W., and A.J. Moore. 2012. Runaway sexual selection without genetic correlations:
Social environments and flexible mate choice initiate and enhance the Fisher
process. Evolution 66: 2674-2684.
Bailey, N.W., B. Gray, and M. Zuk. 2010. Acoustic experience shapes alternative mating
tactics and reproductive investment in male field crickets. Curr. Biol. 20: 845-849.
Bailey, N.W., and M. Zuk. 2012. Socially-flexible female choice differs among populations of
the Pacific field cricket: geographic variation in the interaction coefficient psi ().
Proc. R. Soc. Lond. B. 279:3589-3596.
Bakker, T.C.M. 1993. Positive genetic correlation between female preference and preferred
male ornament in sticklebacks. Nature 363: 255-257.
Bakker, T.C.M. 1999. The study of intersexual selection using quantitative genetics.
Behaviour 136: 1237-1266.
Bakker, T.C.M., and A. Pomiankowski. 1995. The genetic basis of female mate
preferences. J. Evol. Biol. 8: 129-171.
Bateson, P. 1983. Mate Choice. Cambridge University Press, Cambridge, UK.
Beckers, O. M., and J. Schul. 2008. Developmental plasticity of mating calls enables
acoustic communication in diverse environments. Proc. R. Soc. Lond. B. 275: 12431248.
Blows, M.W. 1999. Evolution of the genetic covariance between male and female
components of mate recognition: an experimental test. Proc. R. Soc. Lond. B 266:
2169-2174.
Boake, C.R.B. 1994. Quantitative Genetic Studies of Behavioral Evolution. Univ. Chicago
Press.

This article is protected by copyright. All rights reserved.

18

Borenstein, M., L.V. Hedges, J.P.T. Higgins, and H. Rothstein. 2009. Introduction to metaanalysis. John Wiley & Sons, Ltd., Chichester, U.K.
Bousquet, F., T. Nojima, B. Houot, I. Chauvel, S. Chaudy, S. Dupas, D. Yamamoto, and J.F.
Ferveur. 2012. Expression of a desaturase gene, desat1, in neural and nonneural
tissues separately affects perception and emission of sex pheromones in Drosophila.
Proc. Natl. Acad. Sci. U.S.A. 109: 249-254.
Breden, F., H.C. Gerhardt, and R.K. Butlin. 1994. Female choice and genetic correlations.
Trends Ecol. Evol. 9: 343.
Breden, F., and K. Hornaday. 1994. Test of indirect models of selection in the Trinidad
guppy. Heredity 73: 291-297.
Butlin, R.K. 1993. A comment on the evidence for a genetic correlation between the sexes
in Drosophila melanogaster. Anim. Behav. 45: 403-404.
Callaway, E. 2011. Size doesnt always matter for peacocks. Nature News (18 April 2011)
doi:10.1038/news.2011.245
Chaine, A.S., and B.E. Lyon. 2008. Adaptive plasticity in female mate choice dampens
sexual selection on male ornaments in the lark bunting. Science. 319: 459-462.
Chamberlain, S.A., S.M. Hovick, C.J. Dibble, N.L. Rasmussen, B.G. Van Allen, B.S. Maitrier,
J.R. Ahem, L.P. Bell-Dereske, C.L. Roy, M. Meza-Lopez, J. Carrillo, E. Siemann, M.J.
Lajeunesse, and K.D. Whitney. 2012. Does phylogeny matter? Assessing the
impact of phylogenetic information in ecological meta-analysis. Ecol. Lett. 15: 627636.
Charalambous, M., R.K. Butlin, and G.M. Hewitt. 1994. Genetic variation in male song and
female song preference in the grasshopper Chorthippus brunneus (Orthoptera,
Acrididae). Anim. Behav. 47: 399-411.
Chenoweth, S.F., and M.W. Blows. 2006. Dissecting the complex genetic basis of mate
choice. Nature Rev. Genet. 7: 681-692.
Chenoweth, S.F., and K. McGuigan. 2010. The genetic basis of sexually selected variation.
Annu. Rev. Ecol. Evol. Syst. 41: 81-101.

This article is protected by copyright. All rights reserved.

19

Chenoweth, S.F., H.D. Rundle, and M.W. Blows. 2010. Experimental evidence for the
evolution of indirect genetic effects: changes in the interaction coefficient, psi (),
due to sexual selection. Evolution 64: 1849-1856.

Danchin, E., B. Pujol, and R.H. Wagner. 2013. The double pedigree: a method for studying
culturally and genetically inherited behavior in tandem. PLoS ONE 8(5): e61254.
doi:10.1371/journal.pone.0061254
Day, T. 2000. Sexual selection and the evolution of costly female preferences: spatial
effects. Evolution 54: 715-730.
DerSimonian, R., and N. Laird. 1986. Meta-analysis in clinical trials. Controlled Clinical
Trials 7: 177-188.
Dewinter, A.J. 1992. The genetic basis and evolution of acoustic mate recognition signals
in a Ribautodelphax planthopper (Homoptera, Delphacidae). 1. The female call. J.
Evol. Biol. 5: 249-265.
Dukas, R. 2005. Learning affects mate choice in female fruit flies. Behav. Ecol. 16: 800804.
Duval, S., and R. Tweedie. 2000. A nonparametric trim and fill method of accounting for
publication bias in meta-analysis. J. Am. Stat. Ass. 95: 89-98.
Egger, M., G.D. Smith, M. Schneider, and C. Minder. 1997. Bias in meta-analysis detected
by a simple, graphical test. Br. Med. J. 315: 629-634.
Fisher, H.S., S.J. Mascuch, and G.G. Rosenthal. 2009. Multivariate male traits misalign
with multivariate female preferences in the swordtail fish, Xiphophorus birchmanni.
Anim. Behav. 78: 265-269.
Fisher, R.A. 1915. The evolution of sexual preference. Eugenics Rev. 7: 184-192.
Fisher, R.A. 1930. The Genetical Theory of Natural Selection. Clarendon Press, Oxford.
Fisher, R.A. 1958. The Genetical Theory of Natural Selection, 2nd edition. Dover
Publications, New York.

This article is protected by copyright. All rights reserved.

20

Fowler-Finn, K.D., and R.L. Rodriguez. 2012. Experience-mediated plasticity in mate

preferences: mating assurance in a variable environment. Evolution 66: 459-468.
Galef, B.G., and D.J. White. 2000. 'Culture' in quail: social influences on mate choices of
female Coturnix japonica. Anim. Behav. 59: 975-979.
Getty, T. 2014. GEIs when information transfer is uncertain or incomplete. Pp. 19-40 in J.
Hunt and D. Hosken, eds. Genotype-by-environment interactions and sexual
selection. John Wiley & Sons, Ltd. Chichester, U.K.
Gilburn, A.S., and T.H. Day. 1994. Evolution of female choice in seaweed flies. Fisherian
and good genes mechanisms operate in different populations. Proc. R. Soc. Lond. B
255: 159-165.
Gosden, T.P., and Svensson, E.I. 2008. Spatial and temporal dynamics in a sexual
selection

mosaic. Evolution 62: 845-856.

Gray, D.A., and W.H. Cade. 1999a. Correlated-response-to-selection experiments

designed to test for a genetic correlation between female preferences and male traits
yield biased results. Anim. Behav. 58: 1325-1327.
Gray, D.A., and W.H. Cade. 1999b. Quantitative genetics of sexual selection in the field
cricket, Gryllus integer. Evolution 53: 848-854.
Greenfield, M.D., and R.L. Rodriguez. 2004. Genotype environment interaction and the
reliability of mating signals. Anim. Behav. 68: 1461-1468.
Gromko, M.H. 1992. Genetic correlation of male and female mating frequency: evidence
from Drosophila melanogaster. Anim. Behav. 43: 176-177.
Hall, D.W., M. Kirkpatrick, and B. West. 2000. Runaway sexual selection when female
preferences are directly selected. Evolution 51: 1862-1869.
Holman, L., and H. Kokko. 2014. Local adaptation and the evolution of female choice. Pp.
41-62 in J. Hunt and D. Hosken, eds. Genotype-by-environment interactions and
sexual selection. John Wiley & Sons, Ltd. Chichester, U.K.
Houde, A.E. 1994. Effect of artificial selection on male color patterns on mating preference
of female guppies. Proc. R. Soc. Lond. B 256: 125-130.

This article is protected by copyright. All rights reserved.

21

Ikeda, H., and O. Maruo. 1982. Directional selection for pulse repetition rate of the
courtship sound and correlated responses occurring in several characters in
Drosophila mercatorum. Jap. J. Genet. 57: 241-258.
Ingleby, F.C., J. Hunt, and D.J. Hosken. 2013a. Genotype-by-environment interactions for
female mate choice of male cuticular hydrocarbons in Drosophila simulans. PLoS
One 8(6): e67623.
Ingleby, F.C., D.J. Hosken, K. Flowers, M.F. Hawkes, S.M. Lane, J. Rapkin, I, Dworkin, and
J. Hunt. 2013b. Genotype-by-environment interactions for cuticular hydrocarbon
expression in Drosophila simulans. J. Evol. Biol. 26: 94-107.
Iyengar, V.K., H.K. Reeve, and T. Eisner. 2002. Paternal inheritance of a female moths
mating preference. Nature 419: 830-832.
Iwasa, Y., and A. Pomiankowski. 1995. Continual change in mate preferences. Nature 377:
420-422.
Jones, T.M., R.J. Quinnell, and A. Balmford. 1998. Fisherian flies: benefits of female choice
in a lekking sandfly. Proc. R. Soc. Lond. B 265: 1651-1657.
Judge, K.A. 2010. Female social experience affects the shape of sexual selection on males.
Evol. Ecol. Res. 12: 389-402.3
Kasumovic, M.M., M.D. Hall, and R.C. Brooks. 2012. The juvenile social environment
introduces variation in the choice and expression of sexually selected traits. Ecol.
Evol. 2: 1036-1047.
Kirkpatrick, M. 1982. Sexual selection and the evolution of female choice. Evolution 36: 112.
Kirkpatrick, M., and N. Barton. 1997. The strength of indirect selection on female mating
preferences. Proc. Natl. Acad. Sci. USA 91:1281-1286.
Kokko, H., and K.U. Heubel. 2008. Condition-dependence, genotype-by-environment
interactions and the lek paradox. Genetica 132: 209216.
Kozak, G.M., and J.W. Boughman. 2009. Learned conspecific mate preference in a species
pair of sticklebacks. Behav. Ecol. 20: 1282-1288.

This article is protected by copyright. All rights reserved.

22

Kuijper, B., I. Pen, and F.J. Weissing. 2012. A guide to sexual selection theory. Annu. Rev.
Ecol. Evol. System. 43: 287-311.
Lande, R. 1981. Models of speciation by sexual selection on polygenic traits. Proc. Natl.
Acad. Sci. U.S.A. 78: 3721-3725.
McGuigan, K., A. Van Homrigh, and M.W. Blows. 2008. Genetic analysis of female
preference functions as function-valued traits. Am. Nat. 172: 194-202.
McNiven, V.T.K., and A.J. Moehring. 2013. Identification of genetically linked female
preference and male trait. Evolution. 67: 2155-2165.
Mead, L.S., and S.J. Arnold. 2004. Quantitative genetic models of sexual selection. Trends
Ecol. Evol. 19: 264-271.
Mery, F., S.A.M. Varela, E. Danchin, S. Blanchet. D. Parejo, I. Coolen, R.H. Wagner. 2009.
Public versus personal information for mate copying in an invertebrate. Curr. Biol.
19: 730-734.
Muhlhauser, C., and W.U. Blanckenhorn. 2004. The quantitative genetics of sexual
selection in the dung fly Sepsis cynipsea. Behaviour 141: 327-341.
Nichols, R.A., and R.K. Butlin. 1992. Sexual preferences and genetic correlations. Trends
Ecol. Evol. 7: 29-30.
Nosil, P. 2012. Ecological Speciation. Oxford Series in Ecology and Evolution. Oxford
Univ. Press.
ODonald, P. 1980. Genetic Models of Sexual Selection. Cambridge Univ. Press.
Petfield, D., S.F. Chenoweth, H.D. Rundle, and M.W. Blows. 2005. Genetic variance in
female condition predicts indirect genetic variance in male sexual display traits. Proc.
Natl. Acad. Sci. USA. 102: 6045-6050.
Petrie, M. 1994. Improved growth and survival of offspring of peacocks with more elaborate
trains. Nature 371: 598-599.
Pomiankowski, A., Y. Iwasa, and S. Nee. 1991. The evolution of costly mate preferences I.
Fisher and biased mutation. Evolution 45: 1422-1430.

This article is protected by copyright. All rights reserved.

23

Pomiankowski, A., and Y. Iwasa. 1998. Runaway ornament diversity caused by Fisherian
sexual selection. Proc. Natl. Acad. Sci. USA. 95: 5106-5111.
Pomiankowski, A., and L. Sheridan. 1994a. Linked sexiness and choosiness. Trends Ecol.
Evol. 9: 241-244.
Pomiankowski, A., and L. Sheridan. 1994b. Female choice and genetic correlations. Reply.
Trends Ecol. Evol. 9: 343.
Prokop, Z.M., L. Michalczyk, S.M. Drobniak, M. Herdegen, and J. Radwan. 2012. Metaanalysis suggests choosy females get sexy sons more than good genes. Evolution
66: 2665-2673.
Proulx, S.R. 2001. Female choice via indicator traits easily evolves in the face of
recombination and migration. Evolution 55: 2401-2411.
Prum, R.O. 2010. The Lande-Kirkpatrick mechanism is the null model of evolution by
intersexual selection: implications for meaning, honesty, and design in intersexual
signals. Evolution 64: 3085-3100.
Qvarnstrm, A., J.E. Brommer, and L. Gustafsson. 2006. Testing the genetics underlying
the co-evolution of mate choice and ornament in the wild. Nature 441: 84-86.
Rendell, L., L. Fogarty, and K.N. Laland. 2011. Runaway cultural niche construction. Phil.
Trans. R. Soc. B 366: 823-835.
Rick, I.P., M. Mehlis, and T.C.M. Bakker. 2011. Male red ornamentation is associated with
female red sensitivity in sticklebacks. PLoS One 6(9): 6.
Ritchie, M.G., M. Saarikettu, and A. Hoikkala. 2005. Variation, but no covariance, in female
preference functions and male song in a natural population of Drosophila montana.
Anim. Behav. 70: 849-854.
Roff, D.A., and D.J. Fairbairn. 2014. The evolution of phenotypes and genetic parameters
under preferential mating. Ecology and Evolution. doi: 10.1002/ece3.1130
Rundle, H.D., A. Odeen, and A.O. Mooers. 2007. An experimental test for indirect benefits
in Drosophila melanogaster. BMC Evol. Biol. 7(36) DOI: 10.1186/1471-2148-7-36.

This article is protected by copyright. All rights reserved.

24

Shaw, K.L., C.K. Ellison, K.P. Oh, and C. Wiley. 2011. Pleiotropy, sexy traits, and
speciation. Behav. Ecol. 22: 1154-1155.
Silver, N.C., and W.P. Dunlap. 1987. Averaging correlation coefficients: should Fishers z
transformation be used? J. Appl. Psychol. 72: 146-148.
Simmons, L.W. 2004. Genotypic variation in calling song and female preferences of the
field cricket Teleogryllus oceanicus. Anim. Behav. 68: 313-322.
Simmons, L.W., and J.S. Kotiaho. 2007. Quantitative genetic correlation between trait and
preference supports a sexually selected sperm process. Proc. Natl. Acad. Sci.
U.S.A. 104: 16604-16608.
Slagsvold, T., B.T. Hansen, L.E. Johannessen, and J.T. Lifjeld. 2002. Mate choice and
imprinting in birds studied by cross-fostering in the wild. Proc. R. Soc. Lond. B 269:
1449-1455.
Stamencovic-Radak, M., L. Partridge, and M. Andjelcovic. 1992. A genetic correlation
between the sexes for mating speed in Drosophila melanogaster. Anim. Behav. 43:
389-396.
Stamencovic-Radak, M., L. Partridge, and M. Andjelcovic. 1993. Genetic correlation
between the sexes in Drosophila melanogaster: a reply to Butlin. Anim. Behav. 45:
405.
Svensson, E.I., F. Eroukhmanoff, K. Karlsson, A. Runemark, A. Brodin. 2010. A role for
learning in population divergence of mate preferences. Evolution 64: 3101-3113.
Tinghitella, R.M., E.G. Weigel, M. Head, and J.W. Boughman. 2013. Flexible mate choice
when mates are rare and time is short. Ecol. Evol. 3: 2820-2831.
Tinghitella, R. M. 2014. Male and female cricket modulate their courtship behaviour
depending on female experience with mate availability. Anim. Behav. 91: 9-15.
Van der Linde, K., D. Houle, G.S. Spicer, and S.J. Steppan. 2010. A supermatrix-based
molecular phylogeny of the family Drosophilidae. Genetics Research 92: 25-38.

This article is protected by copyright. All rights reserved.

25

Van der Sluijs, I., O. Seehausen, T.J.M. Van Dooren, and J.J.M. van Alphen. 2010. No
evidence for a genetic association between female mating preference and male
secondary sexual trait in a Lake Victoria cichlid fish. Curr. Zool. 56(1): 57-64.
Van der Sluijs, I., P.D. Dijkstra, C.M. Lindeyer, B. Visser, A.M. Smith, T.G.G. Groothuis,
J.J.M. van Alphen, and O. Seehausen. 2013. A test of genetic association among
male nuptial coloration, female mating preference, and male aggression bias within a
polymorphic population of cichlid fish. Curr. Zool. 59: 221-229.
Wiley, C., C.K. Ellison, and K.L. Shaw. 2012. Widespread genetic linkage of mating signals
and preferences in the Hawaiian cricket Laupala. Proc. Roy. Soc. Lond. B 279:
1203-1209.
Wilkinson, G.S, and P.R. Reillo. 1994. Female choice response to artificial selection on an
exaggerated male trait in a stalk-eyed fly. Proc. R. Soc. Lond. B 253: 1-6.
Woolf, J.B., N.J. Royle, and J. Hunt. 2014. Genotype-by-environment interactions when the
social environment contains genes. Pp. 63-97 in J. Hunt and D. Hosken, eds.
Genotype-by-environment interactions and sexual selection. John Wiley & Sons, Ltd.
Chichester, U.K.
Zhou, Y., J.K. Kelly, and M.D. Greenfield. 2011. Testing the fisherian mechanism:
examining the genetic correlation between male song and female response in
waxmoths. Evol. Ecol. 25: 307-329.

This article is protected by copyright. All rights reserved.

26

Table 1. Studies retained for meta-analysis of intersexual trait-preference additive genetic correlation

Reference

Method ; Sample size

Trait (preference) ; Heritability 4

Chorthippus brunneus

Charalambous et al. 1994

Correlated response to selection; 44

females total sampled from high and low
lines

Syllable length in male song; VA for

trait and preference indicated as
present but not specified

Gryllus texensis
(formerly G. integer)

Gray & Cade 1999b

Analysis of F1 and F2 siblings; 78 families

Pulses per trill in male song;

H2trait: 0.370.13 and 0.400.14 (h2);
H2pref.: 0.320.15 and 0.470.21 (h2)

1,2

Teleogryllus oceanicus

Simmons 2004

Full-sibling breeding design ; 3

populations, 12 sires per population

Length of long chirp in male song;

h2trait: -0.0010.08 and 0.170.14;
h2pref.: 0.00

DeWinter 1992

Correlated response to selection ; 27

males tested from high line and 44 males
tested from low line

Inter-pulse interval in male call;

h2trait: 0.640.05 and 1.020.05

Group ; species
Insects, Orthoptera

Homoptera
Ribautodelphax imitans

Coleoptera

This article is protected by copyright. All rights reserved.

28

Onthophagus taurus

Simmons & Kotiaho 2007

Half-sibling breeding design; 12 sires, 3

dams per sire

Sperm length (spermatheca size in

female) ;
h2trait: 1.140.05;
h2pref.: 0.280.03 and 0.680.15

Achroia grisella

Zhou et al. 2011

Half-sibling breeding design; 2 populations

and 4 environmental conditions tested per
population, 22 sires with 3 dams per sire in
each of the 8 population x environment
combinations

Pulse rate in male song (pulse rate

threshold in female response) ;
h2trait: 8 values from 0.100.06 to
0.360.09; h2pref.: 8 values from
0.080.06 to 0.450.13

Utetheisa ornatrix

Iyengar et al. 2002

Father-daughter correlation; 44 sires, 6

daughters per sire

Male body size ;

h2pref.: 0.510.11

Coelopa frigida

Gilburn & Day 1994

Regression analysis of karyotype means; 5

populations, 3 karyotypes per population

Male body size;

H2trait: 0.50

Cyrtodiopsis dalmanni

Wilkinson & Reillo 1994

Eye span in male

Drosophila mercatorum

Ikeda & Maruo 1982

Correlated response to selection; 18 tests

of females sampled from 2 high line
replicates and 25 tests of females sampled
from 1 low line
Correlated response to selection; 3
different selected lines, 6 different female
preference tests with 300 females per test

Lepidoptera
1

Diptera
1,3

This article is protected by copyright. All rights reserved.

29

Pulse repetition rate in male song;

h2trait : 3 values from 0.140.09 to
0.320.04

Drosophila Montana

Ritchie et al. 2005

Analysis of family means; 27 families, 3

males and 3 females per family

Carrier frequency in male song;

H2trait: 0.35-0.65 ; VA pref.: indicated
as present but not specified

Analysis of female isolines; 20 isolines

tested under 2 environmental conditions

Male cuticular hydrocarbon profile;

H2pref. : 0.85 and 1.15

Muhlhauser & Blanckenhorn

2004

Half-sibling breeding design; 17 sires, 2

dams per sire

Male body size; h2trait = 0.17;

h2pref.: 0.18

Bakker 1993

Father-daughter correlation; 6 sires

Red coloration on male ;

H2trait: 0.230.27; H2pref.: 0.430.37

Rick et al. 2011

Father-daughter correlation; 25 sires

Red coloration on male

Houde 1994

Correlated response to selection; 4

selected lines, 6 males and 6 females
tested per line over 3 generations

Orange coloration on male

Breden & Hornaday 1994

Correlated response to selection; 4

selected lines, 39 males and 39 females
tested per line

Orange coloration on male;

h2pref.: 4 values from 0.047 to 0.33

Van der Sluijs et al. 2010

Mother-son correlation; 12 dams; 3

different male color scores

Van der Sluijs et al. 2013

Father-daughter correlation; 2 sires, 5

dams per sire

Coloration on male;
h2trait: 3 values from 0.31 to 0.82;
h2pref,: indicated as high but not
specified
Coloration on male

Ingleby et al. 2013a

Drosophila simulans

Sepsis cynipsea

Fish
Gasterosteus aculeata

Poecilia reticulate

Pundamilia pundamilia;

This article is protected by copyright. All rights reserved.

30

Birds
2

Ficedula albicollis

Study included two or more subgroups.

Study reported low VA for female preference and therefore was removed from meta-analyses 3 and 4.

Study considered and analyzed only 3 genotypes (karyotypes) and was therefore not included in formal meta-analysis

Qvarnstrm et al. 2006

Animal model analysis of natural

population pedigree; 3715 individuals
studied over a 23-year period

Forehead patch size on male;

h2pref.: 0.0260.01

because of restricted sampling.

4

h2 and H2 are narrow-sense and broad-sense heritability estimates, respectively; values are given as estimate SE when available.

This article is protected by copyright. All rights reserved.

31

Table 2. Results of meta-analyses of intersexual trait-preference additive genetic correlation in species listed in Table 1.
Meta-analysis

Number of
distinct
samples

Overall genetic correlation

with 95% confidence limits
in parentheses

Overall genetic correlation with 95% confidence limits

following application of Duval and Tweedies trim and
fill method to correct for possible publication bias ;
number of samples trimmed.
1
Q statistic ; p-value.

Eggers regression
intercept with 95%
confidence limits ;
p-value, 2-tailed

1. All studies treated

as distinct samples

19

0.132 (0.054 0.209)

0.040 (-0.049 0.129) ; 6

Q = 113.29 ; p < 0.01

1.417 (0.494 2.339) ;

p = 0.005

2. Separate studies of
the same species
treated as part of one
distinct sample

16

0.129 (0.049 0.207)

0.034 (-0.056 0.124) ; 5

Q = 107.47 ; p < 0.01

1.496 (0.359 2.632) ;

p = 0.013

3. All studies treated

as distinct samples ;
studies with low VA
for female preference
not retained

17

0.153 (0.070 0.233)

0.038 (-0.053 0.130) ; 7

Q = 114.67 ; p < 0.01

1.689 (0.747 2.630) ;

p = 0.001

4. Separate studies of
the same species
treated as part of one
distinct sample ;
studies with low VA for
female preference not
retained

14

0.150 (0.066 0.232)

0.033 (-0.060 0.126) ; 6

Q = 108.53 ; p < 0.01

1.871 (0.691 3.051) ;

p = 0.004

This article is protected by copyright. All rights reserved.

32

Q statistic measures heterogeneity of Fishers Z statistic as a function of standard error. See Fig. 1b.

An intercept significantly different from 0 indicates an asymmetrical funnel plot ; p-value is 2-tailed probability that intercept = 0. See Fig. 1b.

This article is protected by copyright. All rights reserved.

33

Fig. 1. a) Forest plot showing point estimates (Xs) for trait-preference intersexual genetic
correlation in 19 retained studies (meta-analysis 1), a point estimate for the overall genetic
correlation reflecting all 19 studies using a random effects model, 95% confidence limits for
these point estimates, and the relative weights (effect sizes, indicated by black boxes) of
each study used to estimate the overall genetic correlation. Details on the studies are given
in Table 1.
Figure 1a

This article is protected by copyright. All rights reserved.

34

b) Funnel plot showing standard error and Fishers Z statistic for each of 19 retained studies
(meta-analysis 1). Observed values are indicated by open circles ; values imputed by Duval
and Tweedies trim and fill method using a random effects model are indicated solid circles.
Point estimate and its 95% confidence limits for the overall Z statistic, which is very similar to
the genetic correlation for values close to 0, are shown at the bottom of the plot; open
diamond reflects observed values only; solid diamond reflects observed and imputed values.

Figure 1b

This article is protected by copyright. All rights reserved.

35

Fig. 2. Phenotypic plasticity, migration and trait/preference covariance. Scenario 1 (a-d):

Plastic trait and plastic preference. a) Individuals in population A are subject to
environmental condition I, which favours trait development in males of the t1t1 genotype but
retards trait development in males of the t3t3 genotype. Reverse effects occur in population
B, which is subject to environmental condition II. Solid red and dashed blue lines are the
reaction norms for trait development in males of these two genotypes. b) A similar plasticity
occurs for trait preference in females across environments I and II. Long dash-dot violet and
dotted green lines are the reaction norms for preference expression in females of the p1p1
and p3p3 genotypes. Females of the p1p1 genotype prefer males of the t1t1 genotype;
females of the p3p3 genotype prefer males of the t3t3 genotypes. c) We now consider two
individuals, a female of genotype t1t1p1p1 in population A and a male of genotype t3t3p3p3 in
population B. Owing to environmental condition II that prevails in population B, the t3t3p3p3
male expresses a superior level of trait development in population B. The t3t3p3p3 male now
migrates (dashed black arrow), as a fully developed adult, to population A where he
continues to express his superior trait, which is attractive to the t1t1p1p1 female resident in
that population. d) The t3t3p3p3 male and the t1t1p1p1 female mate; male offspring of this pair
will have only an intermediate level of trait development in environmental condition I.
Scenarios 2-4 are similar to scenario 1 (a male migrates from population B to A) except for
the shape of the reaction norms. Scenario 2 (e): Plastic trait and constant preference: female
preferences are constant across environments. In environment II, because of non-random
mating males possessing t3 alleles tend to possess p1 alleles. Scenario 3 (f): Constant trait
and plastic preference: trait development in males is constant across environments. In this
scenario, because of non-random mating males possessing t1 alleles in environment II tend
to possess p3 alleles. Scenario 4 (g): Plastic trait and preference with genotype- dependent
plasticity: only genotype p3p3 shows plasticity across environments. Because both p1 and p3
genotype females prefer t3 genotype males in environment II, migrant males tend to possess
both p1 and p3 alleles. In environment I male offspring of migrant males and resident t1t1p1p1
females are either of t1t3p1p1 or t1t3p1p3 genotype. In every scenario, male offspring of the
This article is protected by copyright. All rights reserved.

36

resident t1t1p1p1 female and the immigrant male will have a lower level of the additive genetic
covariance between t1 and p1 alleles that had been accumulating in the population because
of non-random mating between t1t1p1p1 females and t1t1p1p1 males (Greenfield and
Rodriguez 2004).

Figure 2

This article is protected by copyright. All rights reserved.

37