Download Evolutionary mismatch of mating preferences and male colour

Anim. Behav., 1997, 53, 343–351
Evolutionary mismatch of mating preferences and male colour patterns
in guppies
ANNE E. HOUDE & MELISSA A. HANKES
Department of Biology, Lake Forest College
(Received 31 July 1995; initial acceptance 11 October 1995;
final acceptance 4 June 1996; MS. number: 7375)
Abstract. Female mating preferences and the male colour pattern characteristics upon which preferences
are based vary both within and between guppy, Poecilia reticulata, populations from Trinidad. A
general trend is for the degree of preference for orange coloration to match the degree of expression of
orange coloration in each population, resulting in a positive correlation between the means of the two
traits across populations. This study documented an exception to this trend. The mating preferences of
female guppies from Trinidad’s Yarra and Paria Rivers were compared using standard behavioural
assays. Because Yarra males show less orange coloration in their colour patterns than do Paria males,
results of previous studies led to a prediction in the present study of a corresponding difference in
mating preferences between Yarra and Paria females. Both Yarra and Paria females, however, showed
a strong preference for more orange individuals when responding to courtship by Paria males. The
expression of preference in Yarra females therefore does not match the expression of orange coloration
as predicted. Several hypotheses are suggested that may account for this apparent mismatch.
?
Theoretical and empirical work (Lande 1981;
Kirkpatrick 1982; Kirkpatrick & Ryan 1991;
Houde 1993, 1994; Gilburn & Day 1994;
Wilkinson & Reillo 1994) has shown that the
mating preferences of females and the traits of
males on which preferences are based are likely to
evolve in parallel. This parallel evolution can
come about for two main reasons. The simplest is
that mating preferences can result in sexual selection leading to the evolution of corresponding
male traits. Under Fisherian and good genes
models, the two characters also become genetically correlated due to non-random mating, when
females with a particular preference mate with
males with corresponding trait development, and
both characters are inherited by offspring. This
genetic correlation can then result in evolution of
female mating preferences as a correlated response
to selection on the preferred male trait. In either
case, the evolutionary relationship between the
two characters may lead to correlated patterns of
variation across species or populations (Houde
Correspondence: A. E. Houde, Department of Biology,
Lake Forest College, 555 North Sheridan Road, Lake
Forest, IL 60045, U.S.A. (email: [email protected]).
0003–3472/97/020343+09 $25.00/0/ar960399
?
1997 The Association for the Study of Animal Behaviour
1993). The prediction of a correspondence
between female preferences and male colour
patterns across populations depends on several
assumptions. First, correlated evolution of male
traits and female preferences within a population
can only occur if there is sufficient genetic variation in both traits. Second, male traits and
female preferences may correspond well only in
populations that have reached evolutionary equilibrium for both characters. Finally, correlated
evolution of the two traits requires that females
in nature are actually expressing the mating preferences that we measure behaviourally in the
laboratory, so that mate choice leads to sexual
selection on the male trait. Given that these
assumptions might not be met in all populations,
it is of interest to continue to examine the predicted correspondence between female preferences
and male secondary sexual traits, as we have done
here.
Studies of sexual selection and mating preferences in guppies have produced evidence that
female mating preferences and male colour patterns evolve in parallel (Houde 1988a; Houde
& Endler 1990; Houde 1994; Endler & Houde
1995; but see Breden & Hornaday 1994). Female
1997 The Association for the Study of Animal Behaviour
343
344
Animal Behaviour, 53, 2
guppies from many natural populations show a
preference for males with large amounts of orange
coloration (e.g. Kodric-Brown 1985; Houde 1987;
Houde & Endler 1990). Some populations, however, do not show the strong preference for orange
coloration, and the expression of orange in males
from these populations tends to be relatively low
(Houde 1988a; Houde & Endler 1990; Endler &
Houde 1995). A survey of 11 populations showed
a significant association between the expression of
orange in male colour patterns and the expression
of female mating preferences in laboratory behaviour assays (Houde & Endler 1990; Endler &
Houde 1995). These comparative results indicate
that the evolution of mating preferences and
colour patterns can be interdependent. Further
evidence for this interdependence comes from
artificial selection experiments in which selection
for increased and decreased orange coloration was
imposed within a population (Houde 1994; but see
Breden & Hornaday 1994). In some selection
lines, a strong response in the orange character
was accompanied by a corresponding shift in
female mating preferences, apparently as a correlated response to artificial selection. This kind
of correlated response to selection could explain
the correlation between preferences and colour
patterns across populations.
The correspondence between expression of preferences for orange in females and orange coloration in males is not perfect. Some populations
show less expression of orange than would be
predicted for their level of preference, and others
show an opposite pattern (Figure 3 in Endler &
Houde 1995), but previous data were not
adequate to determine whether this variation was
due to experimental error or to other factors
affecting the sexual selection system. Most,
although not all, of the guppy populations that
show strong preferences and strongly expressed
orange coloration come from north-flowing
streams (notably the Paria River) in the Northern
Range mountains of Trinidad, and most of the
populations with weak or no preferences and
weakly expressed orange coloration come from
south-flowing streams on the other side of the
range (Houde & Endler 1990; Endler & Houde
1995). In this study, we attempted an additional
test of the hypothesis that the evolution of preferences and colour patterns are correlated by
comparing mating preferences between two populations from north-flowing streams that differ in
expression of orange coloration and predation
regime.
We examined the mating preferences of female
guppies from a population from the lower Yarra
river of Trinidad, which shows the least expression of orange coloration of all populations
studied from north-flowing streams in Trinidad
(Houde & Endler 1990; Endler & Houde 1995).
The reduced expression of orange is probably due
to selective predation on brightly coloured males
by large fishes present in the lower Yarra river
(Endler 1983; Reznick et al. 1996). Lower Yarra
guppies also show life-history traits consistent
with high levels of predation on adults (Reznick
et al. 1996). This population is also the most
genetically divergent of all guppy populations
examined to date (Magurran et al. 1995). It does
not appear to be closely related to other north
slope populations or to south slope populations.
The Yarra population therefore provides an
additional test of the hypothesis that colour patterns and mating preferences are evolutionarily
linked. Specifically, given that Yarra males have
less orange in their colour patterns compared with
Paria males, which show the greatest expression
of orange of all studied populations (Houde &
Endler 1990; Endler & Houde 1995), we predicted
that Yarra females should show a correspondingly
weaker preference or no preference for orange
coloration compared to Paria females.
MATERIALS AND METHODS
Experimental Fish
Guppies (Poeciliidae) are live-bearing fishes
native to streams and rivers of Trinidad and
adjacent parts of South America. Males show
elaborate colour patterns consisting of variously
coloured spots; females are uncoloured. Colour
patterns are genetically variable within and
between populations, depending in part on the
risk of predation by piscivorous fishes in each
stream (e.g. Endler 1978, 1983). Female mating
preferences also vary within and between populations (Endler & Houde 1995; Godin & Dugatkin
1995). Our experiments were conducted using
guppies from stocks descended from individuals
captured in the Paria and Yarra rivers of Trinidad
approximately four generations previously. All
fish were reared and maintained under standard
conditions (Houde 1987), with Paria and Yarra
Houde & Hankes: Guppy mating preference mismatch
stocks randomized throughout the laboratory
facility. Given the common garden rearing procedure, any differences in behaviour and colour
pattern between the populations can be attributed
to genetic differences (Houde 1988a).
Virgin females for the experiments were
obtained by separating them from males at
approximately 6 weeks of age, and thus before
sexual maturity. Virgin females were housed in
separate sections of aquaria and were visually
isolated from males. Experimental groups of
males were composed of unrelated individuals
that had not been reared together. The variation
in orange coloration within the experimental
groups was maximized by intentionally including
males with both large and small amounts of
orange coloration. To ensure normal behaviour,
males used in the experiments were allowed to
interact with other females for several days prior
to experimental tests. All behavioural observations commenced within 1 h of the time when
aquarium lights came on each morning.
Female Mating Preferences
To measure female mating preferences, we
observed guppies in groups consisting of six Paria
males and six Paria or Yarra females, using a
standard behavioural assay in which we recorded
the responses of females to each male’s sigmoid
courtship displays (Houde 1987, 1988a, b; Houde
& Endler 1990; Endler & Houde 1995). We
observed the responses of both Paria and Yarra
females to each of five groups of Paria males in
separate observation trials (total N=30 males).
Groups of Paria and Yarra males and females
were matched for age and size. Each group of
males was tested with females from the two populations on successive days, in random order. This
yielded a direct comparison of the responses of
Paria and Yarra females to each male in the
experiment.
The responsiveness of females to the test males
was measured using focal observations of each
male. Individual males were identified by their
unique colour patterns. The courtship displays of
each male in turn were recorded during 5-min
observation sessions. Male displays were recorded
only if they were directed towards a particular
female and if the display was not interrupted by
another male (interruptions were rare, however).
The order of observation of the males was rand-
345
omized. Once all six males had been observed, a
second round of 5-min observation sessions was
conducted, for a total of 10 min of observation on
each male. For each male display, a female sexual
response was scored if the female stopped her
current activity, oriented towards the male and
glided (sensu Liley 1966; see also Houde 1987)
unambiguously towards him. A male’s attractiveness was then measured by the fraction of his
displays that resulted in such a female response.
The effect of variation in phenotypic characters
such as colour pattern elements on attractiveness
can then be analysed to determine the overall
pattern of preference of females in a population. Houde (1988b) found that this measure
of attractiveness predicts mating success under
laboratory conditions. Previous work with this
measure of attractiveness (Houde 1987, 1988a)
has shown that it is normally distributed and can
be statistically analysed without transformation.
Our method of assaying female preferences
using groups of males and females does not eliminate the possibility of male–male interactions, but
their effects were minimized by eliminating interrupted displays from the data, by using test males
that had not been reared together and by using
large test aquaria with 1:1 sex ratios. Although
male–male interactions can affect mating success
under some experimental conditions (KodricBrown 1992, 1993), male–male interactions and
male behaviour in general appear to have little
effect on attractiveness and mating success as
measured under experimental conditions similar
to ours (Houde 1987, 1988b).
Male Colour Patterns
We measured the extent of orange coloration in
the colour patterns of the experimental males by
recording both sides of each male on videotape,
capturing the images onto computer disk, and
using an image analysis program to calculate
the relative area of orange spots. Males were
anaesthetized using MS-222 (0.04 g tricaine
methanesulfonic acid dissolved in 150 ml of
water). The relative orange area was calculated as
the total area of all orange spots (including those
on the tail) divided by the total body area of the
fish in side view (excluding fins), averaged between
the right and left sides. The tails of the guppies
were not always fully extended in the video
images, so we did not include them in the body
346
Animal Behaviour, 53, 2
area measurements. Because previous studies (e.g.
Houde & Endler 1990) included tail area as part
of body area, we adjusted the data by multiplying
body area by 1.25 (the average ratio of body-plustail area to body area from measurements of Paria
guppies), providing a standardized body area for
the calculation. We measured the extent of orange
in all Paria males involved in the mating preference tests and also in a sample of Yarra males of
similar age to document the difference in mean
orange area. The orange area of test males ranged
from 0.05 (5%) to 0.28 (28%) orange. The most
orange and least orange males within a test group
differed in orange area by a minimum of 0.12 and
a maximum of 0.19.
Data Analysis
Based on results of previous studies, (e.g.
Houde 1987), we predicted that the responsiveness
of Paria females to Paria males would depend
strongly on orange area of the males. We further
predicted that, in tests with the same males, the
sexual responsiveness of Yarra females would
depend less strongly on orange or would be
independent of orange coloration. Our analysis is
similar to that used by Houde (1988a).
We examined the effect of extent of orange
coloration on male attractiveness by calculating
the linear regression slope with fraction responses
to displays as the dependent variable and orange
area as independent variable. A significant positive slope indicates a preference for orange coloration. The fraction of responses by Paria and
Yarra females to each test male can be directly
compared. Therefore, we tested the hypothesis
that there is a difference in degree of preference
between them by calculating the slope of the linear
regression with the difference in fraction of
responses of Paria versus Yarra females as the
dependent variable. This method of evaluating the
paired data is analogous to a paired sample t-test
for means. We also compared the two independent regression slopes, using an analysis of covariance in which a significant population by orange
area interaction term would indicate a difference
in slopes and thus a difference in the preferences
of Paria and Yarra females.
Strictly speaking, the potential for male–male
interactions (Kodric-Brown 1992, 1993) and
female copying (Dugatkin 1992; Dugatkin &
Godin 1992) in our experimental tests may violate
the assumption of independence necessary for
the linear regression analyses we performed. The
effects of male–male interactions were probably
minimal, however, as argued above, and previous
studies have demonstrated the robustness of
regression analyses for similar data (Houde 1987,
1988a; Houde & Endler 1990; Endler & Houde
1995).
RESULTS
Yarra males showed significantly less orange in
their colour patterns than did Paria males (Yarra
orange area: X&=0.106&0.055, N=30; Paria
orange area: 0.178&0.058, N=30; significant
difference: t58 =4.9, P<0.001). In addition to the
difference in area of orange coloration, the orange
spots of Yarra males appear paler and yellower
than those of Paria males. These differences are
comparable to the differences between Paria and
Aripo guppies in a previous study in which differences in preference were documented (Houde
1988a).
The test males (all from the Paria population)
displayed somewhat less frequently to Yarra
females than to Paria females (Fig. 1a; paired
t29 =2.25, P=0.032). The mean fraction of male
displays eliciting a response was similar for Yarra
and Paria females, however (Fig. 1b; paired
t29 =0.70, P=0.49). There was a significant negative correlation between display rate and orange
area when males were displaying to Paria females
(r= "0.43, P=0.02, N=30), but not when males
were displaying to Yarra females (r=0.10,
P=0.61, N=30).
Both Yarra and Paria females showed a significant preference for more orange males (Fig. 2;
Paria: F1,28 =16.55, P<0.001, R2 =0.37; Yarra:
F1,28 =9.20, P=0.005, R2 =0.25). The slopes of the
regressions of fraction female response against
male orange area were similar in magnitude for
Paria and Yarra females (non-significant interaction of population and orange area in analysis
of covariance: F1,56 =0.14, P=0.70). Moreover,
there was no significant relationship between the
difference in Paria and Yarra females’ fraction
responses to the same males and male orange area
(linear regression: F1,28 =0.11, P=0.74). Overall,
these results indicate that there was no difference
in the degree of preference for more orange males
by Paria and Yarra females.
Houde & Hankes: Guppy mating preference mismatch
1.00
(a)
Paria female preference
20.00
15.00
10.00
5.00
0.00
Paria
Fraction female responses
0.60
Yarra
(a)
0.80
0.60
0.40
0.20
(b)
0.50
0.00
1.00
0.40
0.30
0.20
0.10
0.00
Paria
Yarra
Females
Figure 1. (a) Mean& male display rates of Paria males
courting Paria and Yarra females (displays per 10 min).
(b) Mean fraction of male displays eliciting a female
response in observations with Paria and Yarra females
(displays per 10 min).
DISCUSSION
Our results showed a mismatch between the
expression of female preferences and the expression of orange coloration in Yarra guppies. Our
initial prediction was that difference in the expression of orange coloration between the Paria and
Yarra guppy populations should be matched by a
corresponding difference in the mating preferences
of females. Instead, we found that both Paria and
Yarra females showed significant, but similar,
preferences for more orange males. The sample
sizes used should have been adequate to detect
differences in preferences similar to those
documented in other studies (e.g. Houde 1988a).
In addition to showing similar preferences for
individual males based on orange coloration,
Paria and Yarra females were similar in overall
responsiveness to male displays. Nevertheless, the
males courted the Paria females slightly more
Yarra female preference
Male display rate
25.00
347
0.05
0.25
0.30
0.10 0.15 0.20 0.25
Male orange area
0.30
0.10
0.15
0.20
(b)
0.75
0.50
0.25
0.00
0.05
Figure 2. Effect of relative area of orange coloration on
the fraction of male displays eliciting sexual responses
from (a) Paria females and (b) Yarra females. Leastsquares regression lines: Y=0.15+2.31X (Paria), and
Y=0.17+1.98X (Yarra).
frequently than they did the Yarra females. This
difference in courtship rate might represent reluctance of Paria males to court females from a
different population, or it could reflect differences
in behaviour of the females. Yarra guppies are
notably more responsive to attempts to capture
them and generally seem more skittish than Paria
guppies. A greater rate of movement by Yarra
females might reduce the frequency of opportunities for display by males. Males with greater
orange area did not owe their attractiveness to a
higher rate of courtship relative to males with
a smaller orange area. Indeed, orange area was
negatively correlated with courtship rate when
348
Animal Behaviour, 53, 2
males were courting Paria females, but was uncorrelated with courtship rate with Yarra females.
The negative correlation is consistent with a previous observation (Houde 1988b) that attractive
males have lower rates of courtship, which may
also reflect an influence of female behaviour on
male courtship behaviour. It is not clear why the
relationship of orange area and male courtship
rate differed for Paria and Yarra females, but
again, subtle differences in the behaviour of Yarra
females may have contributed. Despite the differences in male behaviour when courting Paria
and Yarra females, the similar pattern of mating
preferences in the two populations is clear.
Our initial prediction of parallel variation
between female preferences and male colour patterns was based on theoretical expectations (e.g.
Lande 1981) and on empirical findings from other
guppy populations (Houde 1988a; Houde &
Endler 1990; Endler & Houde 1995). Theory
predicts that colour patterns should evolve
to match levels of preference in populations and
also that preferences may undergo correlated
responses to selection on colour patterns. Both of
these arguments are based on the assumption that
females mate non-randomly, with males having
phenotypes corresponding to their preference. If
genetic variation is present in both colour patterns
and female preferences, then this non-random
mating can result in evolutionary change in the
male trait and also in establishment of genetic
correlation of male colour patterns with female
mating preferences. Empirically, our prediction of
a correspondence between female preferences and
male coloration seemed warranted, because colour patterns and mating preferences seemed to
vary in parallel in other populations of guppies
(Houde 1988a; Houde & Endler 1990; Endler &
Houde 1995), and because artificial selection
experiments provided some evidence for a genetic
correlation between the two traits (Houde 1994).
On the other hand, a mismatch similar to that
documented here has been found in other studies.
Females show a preference for the ‘chuck’ component of the male call in the frogs Physalaemus
pustulosus and P. coloradorum, but the ‘chuck’ is
produced only by P. pustulosus males (Ryan et al.
1990; Ryan & Keddy-Hector 1992). Females from
both platyfish, Xiphophorus maculatus, and swordtail, X. helleri, species show a preference for male
tails with ‘swords’, but the sword is present only in
swordtails (Basolo 1990a, b). Basolo (1995) has
documented a similar female preference for a
sword in Priapella, the sister genus of Xiphophorus, in which swords are not expressed. As in
the guppies, Physalaemus frogs and Priapella and
Xiphophorus fishes all show a mismatch between
female mating preferences and male secondary
sexual traits, suggesting that the evolution of these
two traits does not always proceed in parallel.
The Physalaemus, Xiphophorus and Priapella
cases provide evidence that female mating preferences may have their origins in pre-existing
sensory biases that predate the evolution of the
preferred male traits. This inference is based on
detailed phylogenetic information about the taxa
involved. For example, Basolo’s (1995) results
imply that a preference for swords may have
originated as a sensory bias in a common ancestor
of swordtails, platyfishes and Priapella, and that
the sword trait itself evolved as a later response to
this preference in some but not all species of this
clade. Our data do not permit us to make a similar
argument for the evolution of mate preferences in
guppies, however, because not enough is known
about the phylogeny of guppy populations.
Although a preference for orange coloration
could have been present in a common ancestor
of the Paria and Yarra populations, data from
allozyme studies (Magurran et al. 1995) indicate
that the Paria population is more closely related
to populations in which the orange preference is
not expressed than it is to the Yarra population.
The evolutionary history of mating preferences in
guppies is therefore likely to be more complicated
than the relatively straightforward pattern seen in
Physalaemus and Xiphophorus–Priapella. Further
details of the phylogeography of guppy populations in conjunction with data on expression of
female mating preferences and male coloration are
needed to determine the evolutionary history of
sexual selection in guppies. Nevertheless, our
major conclusion from the present study remains
that not all guppy populations show a correspondence between female mating preferences and
male coloration, despite a general tendency for
these characters to vary in parallel in this species.
The theoretical basis for our initial prediction
that mating preferences should match the expression of male colour patterns is compelling, and
appears to have some validity in a number of
guppy populations. How, then, can we explain the
observed mismatches seen in guppies and other
species? As noted above, our prediction was based
Houde & Hankes: Guppy mating preference mismatch
on at least three major assumptions, each of which
might not be met under certain conditions.
First, the colour pattern of Yarra males may
not have responded to sexual selection for
increased orange imposed by female mating preferences because of a lack of heritable variation in
colour patterns. Given the extremely high levels of
heritable variation for colour patterns seen in
some guppy populations (Houde 1988b, 1992), it
would be very surprising to find that heritability
of orange area in the Yarra population was low
enough to constrain evolutionary response to
sexual selection. Considerable phenotypic variation in colour pattern is present in this population (see Results), and heritability of orange
area is currently under investigation.
Second, the Yarra and Paria populations may
not have been diverging for as long as indicated
by allozyme comparisons (Magurran et al. 1995),
and the mismatch could reflect a non-equilibrium
situation in the Yarra population. Magurran et al.
(1995) noted that divergence time could be overestimated if a founder effect contributed to the
genetic differentiation between Yarra and other
populations. For example, the Yarra population
might have been founded relatively recently by
immigrants from a low-predation population.
Direct selection by predators in the Yarra river
could have produced a relatively rapid evolutionary response in male colour patterns with a slower
correlated response (Fisher 1930; Lande 1981) to
selection in female mating preferences. Therefore,
the correlated evolution of mating preferences
may have lagged behind and may not yet match
the expression of colour patterns. Additional
population genetic and quantitative genetic data
from the Yarra population would be needed to
assess this possibility.
Finally, Yarra females in the wild may not be
mating with males corresponding to the preferences shown in the laboratory. When females
are not able to mate with preferred males, there
are two evolutionary consequences. First, sexual
selection will not lead to evolution of male traits
that correspond to the female preference. Second,
a genetic correlation between the preference and
the male trait will not be established, and correlated shifts in female preferences and male secondary sexual traits will not occur as predicted
by models (e.g. Fisher 1930; Lande 1981). We
suggest, therefore, that there may be a behavioural difference between the Paria and Yarra
349
populations that affects the ability of Yarra
females to mate with preferred males.
What might constrain the ability of females to
exercise mate choice? One possibility is that there
is plasticity in the expression of mating preferences, so that the behavioural preferences shown
by females under laboratory conditions are not
always expressed under field conditions. The presence of large predators in the Yarra river, for
example, might reduce the discrimination of
females there. Godin & Briggs (1996) and Gong &
Gibson (1996) have documented reductions in
female selectivity in the presence of a predator.
Another possibility is that social interactions
within and between the sexes may affect the ability
of female guppies to choose mates, and that the
nature of social interactions may differ between
populations, in particular, in response to differences in predation regime. Populations under
different predation regimes differ in the rate of
‘sneak copulation’ attempts by males (Magurran
& Novack 1991; Magurran & Seghers 1994;
Godin 1995), and in the tendency of males to
‘harass’ females and of females to avoid the
attentions of males (Magurran & Novack 1991;
Magurran & Seghers 1994). All of these behaviour
patterns could affect the ability of females to
exercise choice. Courting males are sometimes
interrupted by other males, and copulations are
sometimes even ‘stolen’ by an intruding male
(A. E. Houde & M. A. Hankes, personal observations). These kinds of male–male interactions
are rare in the laboratory, but might be more
common under some natural conditions. Differences in density of fish, sex ratio or visibility of
courting pairs to other individuals could all affect
the rate of interference in courtship, which in turn
could affect the likelihood of females mating with
their preferred mates. More generally, the relative
importance of male–male interactions and female
choice to the ability of males to obtain copulations could differ between populations (Houde
1988b; Kodric-Brown 1992, 1993). Finally, the
tendency of females to copy the mate choice of
other females (Dugatkin 1992; Dugatkin & Godin
1992) could affect the expression of mating
preferences and patterns of male mating success.
Any of these patterns of social interaction could
differ between populations as a consequence of
variation in predation regime or for other reasons.
Such differences could account for an evolutionary mismatch in which female mating preferences
Animal Behaviour, 53, 2
350
have not led to the evolution of corresponding
secondary sexual traits in males, as we have documented here. The potential ways in which female
choice might be constrained and their effects on
male mating success need to be investigated.
Theoretical and empirical studies have made it
clear that mating preferences can have predictable
effects on the evolution of male secondary sexual
traits. Our finding that the expression of male
traits does not always match female mating preferences, and similar results from other studies,
indicate that the relationship of mating preferences to the expression of male traits is not always
a simple matter of cause and effect.
ACKNOWLEDGMENTS
We thank Rebecca Bordeau, Kirsten Findlay,
Leslie Malmgren and Jonathan Weiland for
assistance and discussion, and Joe Benz, Jean-Guy
Godin, Michael McManus and an anonymous
referee for comments on the manuscript. This
study was supported by the Keck Foundation,
a faculty research stipend from Lake Forest
College, and NSF grant IBN-9396137 to A.E.H.
REFERENCES
Basolo, A. 1990a. Female preference for male sword
length in the green swordtail, Xiphophorus helleri
(Pisces: Poeciliidae). Anim. Behav., 40, 332–339.
Basolo, A. 1990b. Female preference predates the
evolution of the sword in swordtail fish. Science, 250,
808–810.
Basolo, A. 1995. Phylogenetic evidence for the role of a
pre-existing bias in sexual selection. Proc. R. Soc.
Lond. Ser. B, 259, 307–311.
Breden, F. & Hornaday, K. 1994. Test of indirect
models of selection in the Trinidad guppy. Heredity,
73, 291–297.
Dugatkin, L. A. 1992. Sexual selection and imitation:
females copy the mate choice of others. Am. Nat., 139,
1384–1389.
Dugatkin, L. A. & Godin, J.-G. J. 1992. Reversal of
female mate choice by copying in the guppy. Proc. R.
Soc. Lond. Ser. B, 249, 179–184.
Endler, J. A. 1978. A predator’s view of animal color
patterns. Evol. Biol., 11, 319–364.
Endler, J. A. 1983. Natural and sexual selection on color
patterns in poeciliid fishes. Environ. Biol. Fishes, 9,
173–190.
Endler, J. A. & Houde, A. E. 1995. Geographic variation in female preferences for male traits in Poecilia
reticulata. Evolution, 49, 456–468.
Fisher, R. A. 1930. The Genetical Theory of Natural
Selection. Oxford: Clarendon Press.
Gilburn, A. & Day, T. 1994. Evolution of female choice
in seaweed flies: Fisherian and good genes mechanisms operate in different populations. Proc. R. Soc.
Lond. Ser. B, 225, 159–165.
Godin, J.-G. J. 1995. Predation risk and alternative
mating tactics in male Trinidadian guppies (Poecilia
reticulata). Oecologia (Berl.), 103, 224–229.
Godin, J.-G. J. & Briggs, S. E. 1996. Female mate choice
under predation risk in the guppy. Anim. Behav., 51,
117–130.
Godin, J.-G. J. & Dugatkin, L. A. 1995. Variability and
repeatability of mating preference in the guppy. Anim.
Behav., 49, 1427–1433.
Gong, A. & Gibson, R. M. 1996. Reversal of a female
preferences after visual exposure to a predator in the
guppy, Poecilia reticulata. Anim. Behav., 52, 1007–
1015.
Houde, A. E. 1987. Mate choice based upon naturally
occurring color pattern variation in a guppy population. Evolution, 41, 1–10.
Houde, A. E. 1988a. Genetic difference in female choice
between two guppy populations. Anim. Behav., 36,
510–516.
Houde, A. E. 1988b. The effects of female choice and
male–male competition on the mating success of male
guppies. Anim. Behav., 36, 888–896.
Houde, A. E. 1992. Sex-linked heritability of a sexually
selected character in a natural population of Poecilia
reticulata (Pisces: Poeciliidae) (guppies). Heredity, 69,
229–235.
Houde, A. E. 1993. Evolution by sexual selection: what
can population comparisons tell us? Am. Nat., 141,
796–803.
Houde, A. E. 1994. Effect of artificial selection on male
colour patterns on mating preference of female
guppies. Proc. R. Soc. Lond. Ser. B, 256, 125–130.
Houde, A. E. & Endler, J. A. 1990. Correlated evolution
of female mating preferences and male color patterns
in the guppy Poecilia reticulata. Science, 248, 1405–
1408.
Kirkpatrick, M. 1982. Sexual selection and the evolution
of female choice. Evolution, 36, 1–12.
Kirkpatrick, M. & Ryan, M. J. 1991. The evolution of
mating preferences and the paradox of the lek.
Nature, Lond., 350, 33–38.
Kodric-Brown, A. 1985. Female preference and sexual
selection for male coloration in the guppy. Behav.
Ecol. Sociobiol., 17, 199–205.
Kodric-Brown, A. 1992. Male dominance can enhance
mating success in guppies. Anim. Behav., 44, 165–
167.
Kodric-Brown, A. 1993. Female choice of multiple male
criteria in guppies: interacting effects of dominance,
coloration and courtship. Behav. Ecol. Sociobiol., 32,
415–420.
Lande, R. 1981. Models of speciation by sexual selection
on polygenic characters. Proc. natn. Acad. Sci. U.S.A.,
78, 3721–3725.
Liley, N. R. 1966. Ethological isolating mechanisms in
four species of poeciliid fishes. Behaviour Suppl., 13,
1–197.
Magurran, A. E. & Novack, M. A. 1991. Another battle
of the sexes: the consequences of sexual asymmetry in
Houde & Hankes: Guppy mating preference mismatch
mating cost and predation risk in the guppy, Poecilia
reticulata. Proc. R. Soc. Lond. Ser. B, 246, 31–38.
Magurran, A. E. & Seghers, B. H. 1994. Predator
inspection behaviour covaries with schooling tendency amongst wild guppy, Poecilia reticulata, populations in Trinidad. Behaviour, 128, 121–134.
Magurran, A., Seghers, B., Shaw, P. & Carvalho, G.
1995. Evolutionary basis of the behavioural diversity of the guppy, Poecilia reticulata, populations in
Trinidad. Adv. Study. Behav., 24, 155–202.
Reznick, D. N., Rodd, F. H. & Cardenas, M. 1996. Life
history evolution in guppies (Poecilia reticulata:
351
Poeciliidae). IV. Parallelism in life-history phenotypes. Am. Nat., 147, 319–338.
Ryan, M. J. & Keddy-Hector, A. 1992. Directional
patterns of female mate choice and the role of sensory
biases. Am. Nat., 139, S4–S35.
Ryan, M. J., Fox, J. H., Wilczynski, W. & Rand, A. S.
1990. Sexual selection for sensory exploitation in the
frog Physalaemus pustulosus. Nature, Lond., 343,
66–67.
Wilkinson, G. & Reillo, P. 1994. Female choice response
to artificial selection on an exaggerated male trait in a
stalk eyed fly. Proc. R. Soc. Lond. Ser. B, 255, 1–6.
352
Animal Behaviour, 53, 2