"Hidden" reproductive conflict between mates in a wild bird population.
ABSTRACT Environmental conditions experienced by a female prior to reproducing may be influenced by her mate. Part of such an indirect effect of a male on his partner's reproduction may be genetic (indirect genetic effect). However, a female's direct and a male's indirect genetic effects need not align. We analyzed 10,652 records of seasonal timing of laying, an important reproductive trait in many organisms, of 1864 male and 1916 female common gulls Larus canus collected during 37 years. We show that there is both a direct (female) and an indirect (male) genetic effect (explaining 14.5% and 4.8% of the REML estimated variance in laying date, respectively), but these are significantly negatively correlated (-0.53+/-0.22 SE), indicating that genes for early laying in females are associated with genes for a delaying male effect on his partner's laying date (and vice versa). There is strong selection for laying early in this population, and these sexually antagonistic genetic effects may contribute in maintaining the variation in laying date. Our findings provide an empirical demonstration of a hitherto largely unstudied level of conflict between mates, with important ramifications for our understanding of evolutionary dynamics and mate choice in nature.
- SourceAvailable from: Lars Gustafsson[show abstract] [hide abstract]
ABSTRACT: The genetic benefits of mate choice are limited by the degree to which male and female fitness are genetically correlated. If the intersexual correlation for fitness is small or negative, choosing a highly fit mate does not necessarily result in high fitness offspring. Using an animal-model approach on data from a pedigreed population of over 7,000 collared flycatchers (Ficedula albicollis), we estimate the intersexual genetic correlation in Lifetime Reproductive Success (LRS) in a natural population to be negative in sign (-0.85+/-0.6). Simulations show this estimate to be robust in sign to the effects of extra-pair parentage. The genetic benefits in this population are further limited by a low level of genetic variation for fitness in males. The potential for indirect sexual selection is nullified by sexual antagonistic fitness effects in this natural population. Our findings and the scarce evidence from other studies suggest that the intersexual genetic correlation for lifetime fitness may be very low in nature. We argue that this form of conflict can, in general, both constrain and maintain sexual selection, depending on the sex-specific additive genetic variances in lifetime fitness.PLoS ONE 02/2007; 2(8):e744. · 3.73 Impact Factor
- [show abstract] [hide abstract]
ABSTRACT: Evolution of size and growth depends on heritable variation arising from additive and maternal genetic effects. Levels of heritable (and nonheritable) variation might change over ontogeny, increasing through "variance compounding" or decreasing through "compensatory growth." We test for these processes using a meta-analysis of age-specific weight traits in domestic ungulates. Generally, mean standardized variance components decrease with age, consistent with compensatory growth. Phenotypic convergence among adult sheep occurs through decreasing environmental and maternal genetic variation. Maternal variation similarly declines in cattle. Maternal genetic effects are thus reduced with age (both in absolute and relative terms). Significant trends in heritability (decreasing in cattle, increasing in sheep) result from declining maternal and environmental components rather than from changing additive variation. There was no evidence for increasing standardized variance components. Any compounding must therefore be masked by more important compensatory processes. While extrapolation of these patterns to processes in natural population is difficult, our results highlight the inadequacy of assuming constancy in genetic parameters over ontogeny. Negative covariance between direct and maternal genetic effects was common. Negative correlations with additive and maternal genetic variances indicate that antagonistic pleiotropy (between additive and maternal genetic effects) may maintain genetic variance and limit responses to selection.The American Naturalist 02/2006; 167(1):E23-38. · 4.55 Impact Factor
- [show abstract] [hide abstract]
ABSTRACT: Survival selection against individuals of inferior quality (measured as breeding success) has been proposed to account for the increase in average reproductive success with advancing age in presenescent birds. This so-called selection hypothesis relies on quality-dependent survival. In the present breeding performance study of common gulls, Larus canus, this assumption was not verified. In particular, omitting the last breeding year from the analysis resulted in the disappearance of the correlation between breeding success and survival. A positive correlation in the full dataset was thus solely based on the poor breeding success of ultimate breeders. Indeed, presenescent individuals were shown to have a specifically low breeding success in their terminal breeding event. The poor success of ultimate breeders thus reflects an abruptly declined condition rather than the birds' overall quality. A comparison of the survival of poor and good performers, involving last-time breeders, thus needs not to be a proper test of the selection hypothesis. Longitudinal analysis revealed a steady increase of individual breeding success until the tenth breeding year. The results suggest that an increase of breeding success with age often found in cross-sectional analyses is primarily a result of age-related improvements of competence and/or increased reproductive effort.Proceedings of the Royal Society B: Biological Sciences 11/2004; 271(1552):2059-64. · 5.68 Impact Factor
“HIDDEN” REPRODUCTIVE CONFLICT
BETWEEN MATES IN A WILD BIRD
Jon E. Brommer1,2and Kalev Rattiste3,4
1Bird Ecology Unit, Department of Biological and Environmental Sciences, P.O. Box 65 (Viikinkaari 1), FIN–00014 University
of Helsinki, Helsinki, Finland
3Institute of Agricultural and Environmental Sciences, Estonian University of Life Sciences, Riia 181, EE 51014, Tartu,
Received January 30, 2008
Accepted June 9, 2008
Environmental conditions experienced by a female prior to reproducing may be influenced by her mate. Part of such an indirect
effect of a male on his partner’s reproduction may be genetic (indirect genetic effect). However, a female’s direct and a male’s
indirect genetic effects need not align. We analyzed 10,652 records of seasonal timing of laying, an important reproductive trait
in many organisms, of 1864 male and 1916 female common gulls Larus canus collected during 37 years. We show that there is
both a direct (female) and an indirect (male) genetic effect (explaining 14.5% and 4.8% of the REML estimated variance in laying
date, respectively), but these are significantly negatively correlated (−0.53 ± 0.22 SE), indicating that genes for early laying in
females are associated with genes for a delaying male effect on his partner’s laying date (and vice versa). There is strong selection
for laying early in this population, and these sexually antagonistic genetic effects may contribute in maintaining the variation in
laying date. Our findings provide an empirical demonstration of a hitherto largely unstudied level of conflict between mates, with
important ramifications for our understanding of evolutionary dynamics and mate choice in nature.
KEY WORDS: Animalmodel,heritability,long-termdata,naturalselection,quantitativegenetics,sexualconflict,sexualselection.
Quantitative genetics describes the response to selection on a trait
(or complex of traits) (Fisher 1958; Falconer and MacKay 1996).
In the coarsest sense, variation in a trait is composed of genetic
and environmental effects, where only the former contributes to a
environment is often determined by its conspecifics, creating an
indirect effect of one individual (or group of individuals) on the
phenotype of a focal individual (Rossiter 1996), and part of such
indirect effects may be heritable (indirect genetic effect, Moore
et al. 1997; Moore and Pizzari 2005). Most empirical work on
indirect genetic effects concern parent–offspring relationships,
where the natal environment in which offspring develop is cre-
ated by the (related) parent and this environment has a genetic
component (Mousseau and Fox 1998; Wilson and Reale 2006).
Indirect genetic effects between unrelated individuals are rarely
studied under laboratory conditions (but see Wolf 2003; Mutic
and Wolf 2007) and are, to our knowledge, unexplored in wild
populations. The standard quantitative genetic approach largely
ignores indirect effects between unrelated individuals, or consid-
ers them as environmental effects only. However, if such indirect
effects are heritable and under selection, they will evolve, and
may either speed up or slow down evolution of a trait depending
on whether they are selected in the same or opposite direction as
the direct genetic effects (Moore et al. 1997; Wolf et al. 1998;
Wolf 2003). One potentially common indirect effect between un-
related individuals may occur if a male can improve his partner’s
C ?2008 The Author(s). Journal compilation C ?2008 The Society for the Study of Evolution.
J. E. BROMMER AND K. RATTISTE
prereproductive condition, for example by feeding her. Such in-
direct effects are likely to be of paramount importance for the
pair’s fitness. Part of a male’s indirect effect on his partner’s re-
production may be heritable, but, to our knowledge, no study to
date has documented male indirect genetic effects on his partner’s
The seasonal timing of reproduction (laying date) is a cru-
cial fitness-related trait in many organisms including birds (Rowe
et al. 1994), and this reproductive trait is expressed by the egg-
clutch size is invariant (three eggs), and laying date is thus pre-
sumably a key reproductive trait in this species. We view a male’s
potential indirect effect on laying date as a nonreciprocal social
interaction (Moore et al. 1997), because a male may influence a
female’s prelaying environment (and thereby her laying date), but
a female’s trait expression (laying the season’s first egg) cannot
ing environment by an (unknown) trait; for example his ability
to arrive early to the colony and quickly establish and defend a
breeding site, his ability for courtship feeding, and his timing of
spermatogenesis and copulation prowess. As a result of this in-
teraction, a female’s laying date is affected by the direct (female)
genetic effect and an environmental component consisting of en-
vironmental effects associated with the female, plus the effect
the male has on his partner’s prelaying environment. The latter
indirect effect may itself consist of a genetic and a nongenetic
component (Moore et al. 1997). Here, we introduce a REML
mixed model procedure to describe the direct (female) and the
indirect (male) effects on laying date, a female linked trait. To
explore the genetics of this indirect effect, we use information
from the population’s pedigree within the mixed modeling proce-
dure (animal model, Lynch and Walsh 1998). The animal model
approach allows to calculate both the indirect genetic effect and
its correlation with the direct genetic effect. Hence, evolutionary
insights into social interactions between mates can be achieved
on the genetic level, also in absence of knowledge of how males
would phenotypically exert an indirect effect on their partner.
We here analyze 37 years of data on laying dates recorded in
an individual-based study of common gulls breeding in Estonia.
rate of 67% (Rattiste and Lilleleht 1986) allows us to estimate the
effects each of the mates has on laying date. By using information
study the genetic architecture of and selection on this important
form of social interaction.
STUDY POPULATION AND METHODS
Data on common gulls were collected in 1968–1983 and 1986–
(detailed in Rattiste 2004). Laying date (date when a pair laid the
season’s first egg) was based on daily checks of the colonies. At
their first breeding event, adult birds were sexed and individu-
ally marked with a metal ring (in case they were not ringed as a
nestling) and a plastic ring with a clearly visible individual code.
Adults were identified in later years by direct observations from a
in 11,624 clutches of 2210 males and 2262 females. An individ-
ual had on average about two partners during its recorded breed-
ing life span (females 2.02 ± 0.03 [range: 1–12], males 2.08 ±
0.03 [range: 1–10]). There were 346 pairs in which neither the
male nor the female bred with another partner, forming 8.4%
from the analyses to reliably separate male and female effects on
laying date (thus using 10,652 laying dates of 1864 males and
1916 females for the analyses).
We assumed that the ith laying date could be described by the
di= μF+ year + f emale + male + εi,
= μF+ year + (af+ pef) + (am+ pem) + εi,
effect year estimates variance across years and random effects
female, and male specify deviations from the overall fixed-effect
εispecifying the residual error. Using an animal-model approach
(Lynch and Walsh 1998), the sex-specific phenotypic effects can
be broken down (eq. 1b) into random effects for a female’s (di-
rect) and a male’s (indirect) genetic effects (af and am, respec-
tively), and effects for the permanent environmental effects (pef
and pemfor female and male, respectively). The permanent envi-
are conserved across records but are not due to additive effects
(e.g., individual-specific environmental, maternal environment,
any nonadditive genetic effects, see Lynch and Walsh (1998) and
Kruuk (2004) for further details). We explicitly consider in some
detail maternal and common temporal environmental effects (see
below). The partitioning of phenotypic male effect into its ad-
ditive genetic and permanent environmental components can be
the partners of a male’s male relatives will be due to male additive
environment as the focal male. Equation (1) was solved using Re-
stricted Maximum Likelihood (REML) implemented in ASReml
(VSN International), which estimated the variance for each ran-
SEXUAL ANTAGONISM IN SEASONAL REPRODUCTIVE TIMING
(female) and indirect (male) genetic effects was estimated on the
basis of resemblance across related individuals of the opposite
sex. Note that of all the terms in equation (1), only the genetic
To test the statistical significance of the terms in the full
model (eq. 1b), we increased model complexity in steps. Statisti-
cal significance of entering each random effect was tested using
a likelihood-ratio test (LRT), calculated as −2 × the difference
in log likelihood between models with and without a particular
random effect. This likelihood ratio was tested as a chi-square
distribution in which the number of variance components added
were the associated degrees of freedom (always one in our case).
We started from a model with fixed effects and residuals only
(model 1). We then tested for the effect of year (model 2), fol-
lowed by the phenotypic effects of female (model 3) and male
(model 4). Thereafter we tested for the significance of parti-
tioning these effects into genetic and nongenetic components for
females (model 5) and males (model 6). Lastly, the genetic co-
variance between direct (female) and indirect (male) effects was
Additive genetic (co)variance was estimated using the pop-
ulation’s pedigree, which was based on all recruits recorded up
to and including 2006, and consisted of 1130 recruits with both
parents known, eight with only their father and four with only
their mother known. In total, relatedness was known between
46% (1731/3780) of individuals considered here. Due to sex bi-
ased dispersal in this species, 80% of the recruited offspring were
males, and there were 1017 male individuals linked to at least
one other male in the pedigree versus 391 females linked to at
least one other female. Across sexes, 623 females had at least
one male relative and 818 males had at least one female relative
in the pedigree. Genetic parentage of recruits was inferred on
the basis of the social status of the parents, which is correct for
females. Paternal links may contain some errors, because DNA
fingerprinting revealed that 3.6% of common gull chicks resulted
from extra-pair copulations in a Polish population (Bukaci´ nska
et al. 1998). However, such low rates of extra-pair paternities are
unlikely to bias estimates of additive genetic (co)variance com-
ponents (Charmantier and R´ eale 2005).
We included three fixed factorial effects. First, we coded for
which of the three islets the breeding occurred (colony effect).
Second, we included the status of a breeding pair, describing that
they were either a newly established pair or the same pair as in the
previous season, or unknown. Lastly, we included the breeding
experience of male and female parents. Breeding experience was
a count of the number of years an individual has been part of
the breeding population with first breeding counted as one (see
also Rattiste 2004). On average, males start to breed at 3.2 years,
and females at 3.7 years, but because the exact age could not be
determined for many individuals, we used breeding experience as
a proxy for age.
The proportion of phenotypic variance explained by direct
additive genetic effects (heritability h2) and indirect additive ge-
netic effects was calculated by using the sum of all REML es-
timates of variances as the phenotypic variance. Comparison of
studies is not meaningful when these studies differ in their fixed
effect structure (Wilson 2008). In particular, we accounted for an-
nual variation as a random effect such that variation across years
is included in the REML estimate of phenotypic variance. Many
other studies correct for annual variation in mean laying date by
including year as a fixed effect, which is likely to drastically re-
duce the REML estimate of phenotypic variance and produces,
all else being equal, a higher heritability than a model in which
across studies, we therefore also provide the observed phenotypic
variance in laying date, prior to conditioning on the fixed effects
(see Wilson 2008 for a discussion).
Selection was formally quantified following Lande and Arnold
(1983). The annual linear standardized selection gradients β?on
survival as the linear term in a least-squares linear regression of
phenotypic laying date standardized to zero annual mean and
unit annual standard deviation on relative recruitment or survival.
Annual nonlinear standardized selection gradients γ?were calcu-
lated as twice the coefficient of standardized laying date squared
in a linear regression model on relative recruitment or survival
that also included the linear term. Relative recruitment was the
observed number of recruits (offspring that bred later in life) of
both sexes produced in a breeding in a given year divided by
that year’s average recruitment (total number of recruits of both
sexes, divided by the number of breeding pairs). Relative survival
was calculated as an individual’s observed survival in a given
year (died  or survived ) divided by that year’s adult sur-
vival probability (fraction of same-sex individuals that survived).
All linear regression model describing the selection included a
On average, 94% (SD = 4.8%, range: 74%–99%) of nest
owners (both males and females) were identified. This value is
conservative because unidentified nest owners may have actu-
ally been identified later in the season in case they renested after
the initial clutch failed. Nest owners that could not be individ-
ually identified before clutch failure mainly were new individ-
uals that did not have a plastic ring and therefore needed to
be trapped. Given that there is selection for earlier laying (see
main text), selection estimates are conservative with respect to
J. E. BROMMER AND K. RATTISTE
Table 1. Hierarchical mixed models of 10,652 laying dates of 1916 common gull females and 1864 males forming 4157 pairs recorded over 37 years.
rA= −0.53 ± 0.22
The estimated variance with standard error in brackets is given for terms that were included in the model, with “–” indicating terms that were not included. Random effect terms are sequentially enteredand tested, where the significance of each term is based on the increase in log-likelihood (LogL), starting with the model with fixed effects and residuals only (model 1). Terms are the variance (V) for
year (V(year)), the sex-specific phenotypic variance for females (V(female)) and males (V(male)). The latter two were split into their components, the additive genetic variance (V(af) and V(am)) and
permanent-environmental variance (V(pef) and V(pem)), of females (f) and males (m), respectively. In model 7, in addition to all terms, the genetic covariance between direct (female) and indirect (male)
genetic effects (cov[af, am]), is included. We report the likelihood-ratio test (χ2statistic with associated df of one) between the given model and the model that is one hierarchical step higher (i.e., 2 vs. 1, 3
vs. 2, etc). Significance of LRT:
1P < 0.001;2P < 0.01;3P < 0.05. For the final model (model 7), the proportion p of variance due to the different causal factors calculated over the sum of all REML estimates of variance are presented with
their standard error. The fixed effect structure is the same in all models and is presented in Table 2. Phenotypic variance in laying date prior to conditioning on the fixed effects was 38.7. The proportion of
variance due to female additive genetic effects V(af) is the heritability (h2).
SEXUAL ANTAGONISM IN SEASONAL REPRODUCTIVE TIMING
unidentified adults, because unidentified individuals were indi-
viduals that tended to breed late in the season and whose clutch
failed or was destroyed before the individual could be caught.
We considered annual selection up to and including the year
2002 (total of 33 years) in order for recruits to be recorded by the
age of four years when most of them started their breeding career,
and to have as accurate survival estimates for adults as possible.
Temporal consistency in annual selection was tested by a sign
test. Statistical significance of annual selection was tested using a
generalized linear model (GLM) on the observed recruitment (0,
1, 2, or 3 offspring) and survival (survived, 1; died, 0) with either
recruitment and survival selection, respectively. Because a GLM
not reflect the effects size of the standardized selection gradients
β?and, in particular, γ?. We focused on phenotypic selection and
did not analyze selection on the predicted breeding value (PBV)
for direct (female) and indirect (male) genetic effects because
calculating selection on PBV is performing a statistical analysis
on model-derived statistics, which has been shown to be biased
We stepwise constructed the most parsimonious mixed model on
repeated observations of laying date recorded for 3780 individu-
als, where at least one of the mates of each pair bred with another
partner during its breeding career (Table 1). There were strong
annual effects on laying date (model 2). Both females (model 3)
and males (model 4) had highly significant phenotypic effects
on laying date. Using pedigree information from this population,
we partitioned the sex-specific phenotypic effects on laying date
components using an animal model approach. We found signif-
icant female (i.e., direct) genetic effects af (model 5) and male
(i.e., indirect) genetic effects am(model 6). Lastly, we included
ative (model 7). Hence, on the same genome, loci that in males
advance his partner’s laying date, delay laying when expressed in
females (and vice versa).
The fixed effects part of the final model (Table 2) showed
that laying date differed across colonies. Furthermore, chang-
ing a partner led to a delay in laying date of 1.8 days rela-
tive to a previously established pair. Lastly, the breeding expe-
rience (a proxy of age) of males and females affected laying
date. Differences across females were mostly due to direct (ad-
ditive) genetic effects rather than nongenetic individual-specific
Table 2. Fixedeffectsofmixed-model7inTable1(thesamefixed-
effect structure was used in the other models).
Categorical fixed effects were: the breeding status of a pair (new pair or
prior-established pair (or considered unknown in three cases)), breeding
colony, and breeding experience (the number of years individual was
part of the breeding population). Effect of “Status” is given relative to a
prior-established pair. F-tests are conditional Wald’s tests.
effects, with a heritability (h2) of 14.5%. Males had a small, but
significant, indirect additive genetic effect explaining 4.8% of
variance in laying date, in addition to a similarly sized nongenetic
COMMON ENVIRONMENTAL EFFECTS
Resemblance across relatives can—apart from additive genetic
effects—also stem from common environmental factors. The
covariance across relatives due to common environmental effects,
earlier that direct (female) genetic effects on laying date are not
affected by maternal effects and effects related to the year of
first breeding, and are not subjected to genotype-temperature in-
teractions (Brommer et al. 2008). We here considered whether
there is any resemblance across males due to shared temporal
environmental effects (year of first breeding and year of birth),
and due to shared maternal effects on indirect (male) genetic ef-
fects by adding these terms to the final model (model 7, Table 1).
There were no temporal effects due to the specific year of when
males started to breed (proportion of REML estimated variance
explained by year male started to breed: 0.0028 ± 0.0020). The
mother and the year of birth (cohort year) were known for 700
males (477 mothers in 35 different birth year). Setting the year
of birth for the remaining males as “unknown” revealed that a
low proportion of 0.0038 ± 0.0031 of REML estimated variance
was explained by year of birth. Coding the unknown mothers as
unknown showed little evidence of heterogeneity across moth-
ers (0.0037 ± 0.0041 of REML estimated variance). Because
practically all offspring in our study area were ringed, we alterna-
tively assumed that each male whose mother was not known had
a different mother because males with unknown mothers origi-
nated mainly from other colonies in the whole western Estonian
archipelago. This alternative coding produced qualitatively the
J. E. BROMMER AND K. RATTISTE
Table 3. Annual data on number of pairs identified (breed), the number of recruits produced (nrec), and the linear (β?) and nonlinear (γ?)
standardized selection gradients for survival (surv) and recruitment (rec).
For females andmales, wefurther present the number of individuals that survived (surv). The sign of the selection gradient is alsoindicated for verylow selec-
tion (−0.00). Statistically significant standardized selection gradients are indicated in bold. Significance was based on a generalized linear model with Poisson
(for recruitment) or binomial errors (for survival). Average over all years is presented in the last line of the Table, and a sign tests is presented in the main text.
same result. Including any of the common-environmental factors
that we could address did not significantly change the male addi-
tive genetic effect on female laying date (results not shown) and
we therefore conclude that the estimate of male indirect genetic
effect is not inflated due to any of these common-environmental
We calculated annual selection on laying date over 33 years
(Table 3). There was strong evidence of recruitment selection
for earlier laying, as the linear standardized selection gradients
were negative in 33 of 33 years (P < 0.001), and all but one
of the annual linear standardized selection gradients were signifi-
positive in 32 of 33 years (P < 0.001). The detected nonlinear
selection was due to curvature of the fitness map, rather than rep-
resenting true disruptive selection (Fig. 1). There was evidence
of survival selection in females (negative β?in 26/33 years, P =
0.001, significantly so in 10 years), but not in males (negative β?
SEXUAL ANTAGONISM IN SEASONAL REPRODUCTIVE TIMING
Figure 1. Recruitment selection for laying early in the common
gull. Plotted (black thick line) is the function for the average lin-
ear and nonlinear standardized selection gradients over 33 years,
describing how relative recruitment (observed number of recruits
divided by the annual average recruitment) depends on laying
date standardized to zero annual mean and unit standard devia-
tion. The plotted line is based on the average annual linear (β?) and
nonlinear (γ?) standardized recruitment selection gradients given
in Table 3, where the average constant (expected relative fitness
for the mean annual laying date) was 0.77. The average stan-
dardized recruitment selection gradient is plotted over the range
where relative recruitment is larger than zero. The observed an-
nual data over all 33 years is plotted in gray in the background to
aid in visual comparison.
years, P = 0.7), and in males (negative γ?in 16/33 years).
Probably one of the most common and important interactions
between two unrelated individuals occurs whenever opposite-sex
individuals form a pair and reproduce. We have here demon-
strated a male indirect genetic effect on the seasonal timing of his
partner’s reproduction. We have further shown that laying date is
a fitness-related trait. Common gulls are under strong and tem-
porally consistent recruitment selection for laying early in the
season. Hence, a male’s indirect genetic effect on his partner’s
laying date is clearly evolutionarily relevant. Previous studies on
birds that used a mixed model approach to test for female and
male-specific effects on avian reproductive traits found that male
birds have no significant effect on laying date and clutch size in
two species of passerine (Sheldon et al. 2003; McCleery et al.
2004; Gienapp et al. 2006) and one wader species (van de Pol
et al. 2006), although male mute swans Cygnus olor do have
a phenotypic effect on laying date (Charmantier et al. 2006).
Evolutionary significant male indirect genetic effects may occur
whenever males can contribute to their mate’s prereproductive
environment, which seems particularly likely to occur in long-
lived socially monogamous vertebrates, but may also occur in
invertebrates (e.g., where males present a nuptial gift, Gwynne
the genetics of such conflict is currently poorly understood (Arn-
qvist and Rowe 2005), traits that make a male successful in in-
creasing his partner’s reproduction may be generally detrimental
genetic level, where genes for an advancing direct (female) effect
on laying date are significantly negatively associated with genes
for a delaying indirect (male) effect on laying date. This negative
genetic association is probably not due to linkage disequilibrium,
because this would require the unlikely scenario of females with
genes for early laying consistently pairing with males with genes
for an indirect effect of delaying their partner’s laying date (and
vice versa). Instead, this negative genetic correlation is probably
due to antagonistic pleiotropy where genes that advance laying
date through a direct effect expressed in females have a delaying
effects occur between loci for direct and indirect effect (on the
same genome). We know of no other demonstration of such sex-
ually antagonistic genetic effects on a fitness-related trait in the
wild. However, negative genetic correlations between direct ge-
netic effects and maternal genetic effects (which are a particular
form ofindirect genetic effects) arefoundinseveral domesticated
(Wilson and Reale 2006), and one wild mammal species (Wilson
et al. 2005).
We have shown that an animal-model approach allows quan-
tification of the genetics of social interactions between mates,
even when the causal male trait affecting his partner’s laying
date is unknown. The negative genetic correlation between direct
(female) and indirect (male) genetic effects may create a force
balancing the otherwise temporally consistent selection for ear-
sexually antagonistic genetic effects on fitness traits, which has,
apart from a potential role in maintaining genetic variation in the
face of selection, also implications for mate choice (Chippindale
et al. 2001; Brommer et al. 2007; Foerster et al. 2007). In the
common gull, sexually antagonistic genetic effects will reduce
the indirect benefits of mate choice for both sexes, because se-
lecting a partner for genes for earlier laying (through either direct
or indirect effect) will have antagonistic effects in offspring of the
opposite sex. On the other hand, the relatively large nongenetic
variance component for the indirect (male) effect suggests that
direct (nonheritable) benefits of female mate choice may still be
In conclusion, we have demonstrated an empirical approach
that allows quantification of the genetics of social interactions
between mates. Intriguingly, we here find evidence of direct and
J. E. BROMMER AND K. RATTISTE
indirect genetic effects on a fitness-related life-history trait, but
also of a significantly negative relationship between these, im-
plying sexually antagonistic genetic effects. Such “hidden” in-
teractions and conflict between mates occurring on the genetic
level may be common in nature and may form a considerable
evolutionary force, only becoming visible when the traditional
evolutionary quantitative genetic framework is expanded to in-
clude indirect genetic effects between unrelated individuals.
of Finland, and KR by the Estonian Science Foundation (grant no. 7190).
M. Fred, P. Karell and A. Wilson are thanked for comments.
Arnqvist, G., and L. Rowe. 2005. Sexual conflict. Princeton Univ. Press,
intersexual genetic correlation for lifetime fitness in the wild and its im-
plications for sexual selection. PLoS one 2(8):e744. DOI: 10.1371/jour-
Brommer, J. E., K. Rattiste, and A. J. Wilson. 2008. Exploring plasticity in
the wild: laying date–temperature reaction norms in common gull Larus
canus. Proc. R. Soc. Lond. B 275:687–693.
Bukaci´ nska, M., D. Bukaci´ nski, J. T. Epplen, K. P. Sauer, and T. Lub-
juhn. 1998. Low frequency of extra-pair paternity in Common Gull
(Larus canus) as revealed by DNA fingerprinting. J. Ornithol. 139:413–
Charmantier, A., and D. R´ eale. 2005. How do misassigned paternities affect
the estimation of heritabilities in the wild? Mol. Ecol. 14:2839–2850.
Charmantier, A., C. Perrins, R. H. McCleery, and B. C. Sheldon. 2006. Evo-
lutionary response to selection in clutch size in a long-term study of the
mute swan. Am. Nat. 167:453–465.
Chippindale, A. K., J. R. Gibson, and W.R. Rice. 2001. Negative genetic
correlation for adult fitness between sexes reveals ontogenetic conflict
in Drosophila. Proc. Natl. Acad. Sci. USA 98:1671–1675.
Falconer, D. S., and T. F. C. MacKay. 1996. Introduction to quantitative
genetics, 4th ed. Longman, Harlow, U.K.
Fisher, R. A. 1958. The genetical theory of natural selection. Clarendon,
and L. E. B. Kruuk. 2007. Sexually antagonistic genetic variation for
fitness in red deer. Nature 447:1107–1110. doi:10.1038/nature05912.
Gienapp, P., E. Postma, and M. E. Visser. 2006. Why breeding time has not
responded to selection for earlier breeding in a songbird population.
Gwynne, D. T. 2008. Sexual conflict over nuptial gifts in insects. Annu. Rev.
Entomol. 53:83–101 (doi:10.1146/annurev.ento.53.103106.093423).
Hadfield, J. D. 2008. Estimating evolutionary parameters when viability se-
lection is operating. Proc. R. Soc. Lond. B. 275:723–734.
Kruuk, L. E. B. 2004. Estimating genetic parameters in natural populations
using the ‘animal model’. Philos. Trans. R. Soc. Lond. B 359:873–890.
characters. Evolution 37:1210–1226.
Lynch, M., and B. Walsh. 1998. Genetics and analysis of quantitative traits.
Sinauer, Sutherland, MA.
McCleery, R. H., R. A. Pettifor, P. Armbruster, K. Meyer, B. C. Sheldon, and
C. M. Perrins. 2004. Components of variance underlying fitness in a
natural population of the great tit Parus major. Am. Nat. 164:E62–E72.
Moore, A. J. and T. Pizzari. 2005. Quantitative genetic models of sexual
conflict based on interacting phenotypes. Am. Nat. S88–S97.
Moore, A. J., E. D. Brodie III, and J. B. Wolf. 1997. Interacting phenotypes
and the evolutionary process: I. direct and indirect genetic effects on
social interactions. Evolution 51:1351–1362.
Mutic, J. J., and J. B. Wolf. 2007. Indirect genetic effects from ecological
interactions in Arabidopsis thaliana. Mol. Ecol. 16:2371–2381.
Rattiste, K. 2004. Reproductive success in presenescent common gulls (Larus
canus): the importance of the last year of life. Proc. R. Soc. Lond. B
Rattiste, K., and V. Lilleleht. 1986. Some aspects of the demography of the
Common Gull Larus canus in Estonia. V˚ ar F˚ agelv¨ arld 11(Suppl.):179–
Rossiter, M. 1996. Incidence and consequences of inherited environmental
effects. Annu. Rev. Ecol. Syst. 27:451–476.
decline of avian clutch size. Am. Nat. 143:698–722.
Sheldon, B. C., L. E. B. Kruuk, and J. Meril¨ a. 2003. Natural selection and
inheritance of breeding time and clutch size in the collared flycatcher.
van de Pol M., D. Heg, L. W. Bruinzeel, B. Kuijper, and S. Verhulst. 2006.
ductive performance in oystercatchers (Haematopus ostralegus). Behav.
Wilson, A. J. 2008. Why h2 does not always equal VA/VP? J. Evol. Biol.
21:647–650; doi: 10.1111/j.1420-9101.2008.01500.x.
Wilson, A. J. and D. Reale. 2006. Ontogeny of additive and maternal genetic
effects: lessons from domestic mammals. Am. Nat. 167:E23–E38.
Wilson, A. J., D. W. Coltman, J. M. Pemberton, A. D. J. Overall, K. A. Byrne,
and L. E. B. Kruuk. 2005. Maternal genetic effects set the potential for
evolution in a free-living vertebrate population. J. Evol. Biol. 18:405–
Wolf, J. B. 2003. Genetic architecture and evolutionary constraint when the
environment contains genes. Proc. Natl. Acad. Sci. USA 100:4655–
Wolf, J. B., E. D. Brodie, III, J. M. Cheverud, A. J. Moore, and M. J. Wade.
1998. Evolutionary consequences of indirect genetic effects. Trends
Ecol. Evol. 13:64–69.
Associate Editor: M. Webster