Content uploaded by Timothy A Linksvayer
Author content
All content in this area was uploaded by Timothy A Linksvayer on May 21, 2020
Content may be subject to copyright.
vol. 181, no. 2 the american naturalist february 2013
Genetic Constraints on Dishonesty and
Caste Dimorphism in an Ant
Luke Holman,
1,
* Timothy A. Linksvayer,
1,
†and Patrizia d’Ettorre
1,2
1. Centre for Social Evolution, Department of Biology, University of Copenhagen, Universitetsparken 15, 2200 Copenhagen, Denmark;
2. Laboratoire d’Ethologie Expe´rimentale et Compare´e (LEEC), University of Paris 13, 99 Avenue J. B. Cle´ment, 93430 Villetaneuse, France
Submitted May 14, 2012; Accepted September 8, 2012; Electronically published January 8, 2013
Online enhancement: appendix. Dryad data: http://dx.doi.org/dryad.cb0f6.
abstract: The ultimate causes of honest signaling remain a subject
of debate, with questions remaining over the relative importance of
costs and constraints. Signal costs may make dishonesty prohibitively
expensive, while genetic constraints could make it impossible. We
investigated honest signaling using full-sib analysis and parent-off-
spring regression in the ant Lasius niger, in which queens produce
a cuticular hydrocarbon-based pheromone that signals fertility and
inhibits worker reproduction and aggression. We found multiple lines
of evidence that cuticular hydrocarbon production is genetically cor-
related with oogenesis and that the queen pheromone 3-methylhen-
triacontane and other 3-methylalkanes have strong genetic links with
fertility relative to other cuticular hydrocarbons. These genetic cor-
relations may maintain honesty in the face of directional selection
on signaling and explain the putatively widespread use of cuticular
hydrocarbons in fertility signaling across the social insects. We also
found evidence for a positive genetic correlation for fertility between
the castes; that is, the most fertile queens produced especially fertile
workers. These results highlight that intercaste genetic correlations
could constrain the evolution of queen-worker dimorphism, such
that worker reproduction may sometimes reflect a nonadaptive “caste
load” rather than positively selected cheating.
Keywords: cuticular hydrocarbons, fertility signal, handicap, index,
Lasius niger, queen pheromone.
Introduction
Biological signals are actions or structures produced by a
sender that elicit a response in one or more receivers (May-
nard Smith and Harper 1995). By definition, signals have
been shaped by selection to increase the fitness of the
sender given the behavior of receivers, although these par-
* Corresponding author. Present address: Division of Ecology, Evolution and
Genetics, Research School of Biology, Australian National University, Can-
berra, Australian Capital Terriotories 0200, Australia; e-mail: luke.holman@
anu.edu.au.
†Present address: Department of Biology, University of Pennsylvania, Phil-
adelphia, Pennsylvania 19104.
Am. Nat. 2013. Vol. 181, pp. 161–170. 䉷2013 by The University of Chicago.
0003-0147/2013/18102-53845$15.00. All rights reserved.
DOI: 10.1086/668828
ties frequently have conflicting evolutionary interests
(Andersson 1980; Keller and Nonacs 1993; Brilot and
Johnstone 2003; Arnqvist 2006). It is therefore paradoxical
that most biological signals appear to convey information
reliably (sensu Searcy and Nowicki 2005) because senders
often stand to benefit by producing a misleading signal,
for example, one that exaggerates their hunger, fighting
ability, or quality as a mate. Evolutionary theory has pro-
vided a number of potential resolutions to this problem,
the best known of which is the handicap principle (Zahavi
1975; Zahavi and Zahavi 1997). This hypothesis states that
senders benefit by producing a signal whose strength is
proportional to the quality being advertised, because pro-
ducing a signal of higher intensity would carry fitness costs
that outweigh the benefit of the augmented signal. The
handicap principle rests on the assumption that senders
of differing qualities receive unequal fitness costs and/or
benefits per unit of investment in signaling (Grafen 1990;
Getty 2006).
Another hypothesis for the maintenance of signal hon-
esty is the index hypothesis, which states that some signals
(“indexes”) are inherently honest because of evolutionary
constraints that preclude dishonesty (Maynard Smith and
Harper 1995, 2003). A supposed difference with the hand-
icap principle is that dishonest signaling is impossible
rather than simply suboptimal. For example, small-bodied
acoustic signalers may be unable to produce calls with the
same acoustic properties as larger individuals because of
the physics of sound production, such that calls advertising
large size cannot be faked (Reby and McComb 2003; Forst-
meier et al. 2009; Charlton et al. 2011). Maynard Smith
and Harper (1995, 2003) postulated that novel traits im-
proving the relationship between sender quality and signal
strength (e.g., traits that make the caller sound larger, such
as a descended larynx) should quickly spread to fixation,
such that their effect on signal reliability would be tran-
sient. Rephrased in evolutionary genetic terms, selection
and genetic constraints might collude to exhaust genetic
This content downloaded from 130.091.128.242 on June 30, 2016 11:43:02 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
162 The American Naturalist
variation along the multivariate axis of selection (Blows
and Hoffmann 2005). For example, alleles that improve
the signal without negatively impacting other traits might
all be fixed, such that the only remaining genetic variation
in the signal shows antagonistic pleiotropy with nonsig-
naling traits, constraining any further response to selection
and preserving the trait-signal relationship (Clark and
Moore 1995; Van Homrigh et al. 2007; Hine et al. 2011).
Recently, Holman (2012) argued in favor of a pluralistic
model invoking both costs and constraints in the main-
tenance of signal honesty. He showed that the cost of
signaling optimally depends on the shape of the cost and
benefit functions of investing in signaling (see also Lach-
mann et al. 2001) and the extent to which these functions
are affected by the sender’s underlying quality (e.g., genetic
quality or phenotypic condition). Signals whose cost and
benefit functions are constrained strongly by the sender’s
quality are predicted to evolve to be cheap, while signals
whose expression is less dependent on sender quality
evolve to be expensive. One can therefore assess a signal’s
“strategic cost” (the cost over and above the minimum
required for adequate signal transmission; Guilford and
Dawkins 1991) by testing whether its expression is strongly
constrained by the sender’s quality. Constraints on signal
evolution are rarely studied explicitly, and more data are
needed to assess the proximate and ultimate mechanisms
responsible for signal honesty and to map and explain
variation in strategic signaling costs (which are impossible
to measure directly; Sza´mado´ 2011).
Here, we employ quantitative genetic methods (full-sib
analysis and parent-offspring regression in a cross-fostered
breeding design) to identify potential evolutionary con-
straints on signaling and queen-worker dimorphism, using
the queen pheromone of the black garden ant Lasius niger
as a model system (Holman et al. 2010a, 2010b; Holman
2012). Queens produce a blend of cuticular hydrocarbons
(waxy substances covering the body surface) that differs
from that of workers, and one queen-type hydrocarbon
(3-methylhentriacontane or 3-MeC
31
) was shown to be a
pheromone that inhibits worker reproduction and ag-
gression in experimental bioassays (Holman et al. 2010b).
Some hydrocarbons, particularly 3-MeC
31
, correlate with
queen maturity and fecundity (Holman et al. 2010a), sug-
gesting that hydrocarbons signal reproductive output or
overall quality. A similar set of queen-type hydrocarbons
predict whether a queen will be killed by workers in col-
onies containing supernumerary queens, and 3-MeC
31
was
again the strongest predictor (Holman et al. 2010a). Such
pheromone-dependent queen execution may be wide-
spread in ants (West-Eberhard 1983; de Biseau et al. 2004;
Wurm et al. 2010).
It is therefore puzzling why queens do not produce high
quantities of pheromone at all times. Recent experiments
showed that fecundity and 3-MeC
31
production are both
affected by juvenile hormone but in opposite directions,
demonstrating an endocrine-mediated trade-off between
the two traits and a possible strategic cost of pheromone
production (Holman 2012). This result also suggests that
fecundity and queen pheromone production may be un-
able to evolve fully independently, because of their co-
dependence on juvenile hormone. Estimating the genetic
correlation between queen pheromones and fecundity can
test this hypothesis: if two traits are genetically correlated,
one trait cannot evolve without changing the other, po-
tentially constraining both traits from reaching their op-
timal values. For example, if there were a positive genetic
correlation between queen pheromone production and fe-
cundity, queens would not be able to evolve increased
pheromone production without also increasing their fe-
cundity. Quantitative genetic parameters therefore provide
insight into the strength and nature of evolutionary con-
straints acting on the evolution of queen pheromones and
honest signals in general. Our results also allow us to assess
the extent to which queen and worker traits could evolve
independently under caste-specific selection (i.e., when
different phenotypes are selected for in queens and
workers).
Methods
Ant Collection and Cross-Fostering Design
Mated, wingless queen Lasius niger were collected during
a mating flight in Copenhagen, Denmark, on July 21, 2010,
and housed immediately in individual plastic cylinders (26
mm #38 mm) with a ball of moist cotton wool but no
food, mimicking natural colony foundation. Each queen
was randomly paired with a group of 10 separately housed
“foster workers”; all foster workers were collected from a
single large wild colony in Copenhagen and were kept in
8#6#5-cm plastic boxes with a moist plaster base and
ad lib. water, honey, and mealworms. Every 3–4 days, all
eggs were transferred from the queen containers to the
paired foster worker boxes using a soft paintbrush. Dead
foster workers were replaced within 0–2 days, keeping the
total number of foster workers constant.
The aims of this rearing procedure were to prevent post-
hatching maternal effects from influencing the offspring
phenotype and to standardize the biotic and abiotic rearing
environment across families as much as possible. Using
workers from a single colony as foster parents should
greatly diminish the variation in focal worker phenotype
caused by variation in the phenotype of the foster parents
(indirect genetic and environmental effects; Linksvayer
2006), particularly because mean relatedness among nest-
mate workers is very high in Scandinavian L. niger pop-
This content downloaded from 130.091.128.242 on June 30, 2016 11:43:02 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
Genetic Constraints on Queen Pheromones 163
ulations (close to 0.75; Corley and Fjerdingstad 2011).
Phenotypic covariation between maternal and offspring
traits should therefore be largely attributable to genetic
effects and prehatching maternal effects. A limitation of
using a standardized environment and foster worker ge-
notype is that we cannot be sure whether estimated quan-
titative genetic parameters would be the same in other
rearing environments. For example, genetic effects could
be plastic, depending on factors such as nutrition, colony
size, or the genetic composition of rearing workers.
All queens were freeze-killed for later trait measurement
40 days after the mating flight. Their offspring were al-
lowed to mature for an additional 23 days before being
frozen. The foster workers were marked at the start of the
experiment with a dot of enamel paint on the gaster in
order to distinguish them from workers hatched from the
lab-reared eggs, although this precaution proved unnec-
essary because all the lab-reared workers developed as nan-
itics (small workers often found in queens’ first broods;
Porter and Tschinkel 1986). Our experiment therefore in-
cidentally suggests that the nanitic phenotype is controlled
by a prehatching maternal effect, as opposed to a post-
hatching effect such as a queen-specific rearing behavior
(because the eggs were rapidly transferred to foster workers
obtained from a mature colony). For example, newly
mated L. niger queens might provision eggs with a hor-
mone that stunts growth, as young harvester ant queens
do to prevent their eggs developing into queens (Schwan-
der et al. 2008).
Trait Measurement
Worker ovaries were dissected out, and the degree of ovar-
ian activation was scored on a scale of 1–4 as in Holman
et al. (2010b), as follows: (1) completely empty, (2) 1–2
very small eggs and/or developing nurse cell material, (3)
1–3 developing eggs in both ovaries or a large egg in one
ovary, (4) well-developed eggs in both ovaries. The level
of activation of queen ovaries was quantified by counting
the number of full-sized eggs present in the ovaries. We
also measured head size, scored as the shortest distance
between the eyes, from digital photographs taken at #40
magnification.
The cuticular hydrocarbons (CHCs) of queens and their
offspring were analyzed by gas chromatography mass spec-
trometry (GC-MS) using similar methods to previous
studies (Holman et al. 2010a, 2010b; Holman 2012). Hy-
drocarbons were extracted from queens by placing the ant
in 150 mL of pentane in a 2-mL glass vial for 10 min, then
transferring the extract to a 200-mL glass insert and al-
lowing it to evaporate in a laminar flow cabinet. The ex-
tract was then diluted in 40 mL of pentane containing an
internal standard (icosane; 1 ng mL
⫺1
), which was later
used to determine the absolute mass of each CHC present
in the extract. Two microliters of queen extract was then
injected into an Agilent 6890N gas chromatograph coupled
to an Agilent 5375 mass spectrometer using an Agilent
7683B autosampler. Workers are much smaller, so they
were extracted in 10 mL of pentane (again containing a 1
ng mL
⫺1
icosane internal standard) in a 200-mL glass insert
for 10 min; 3 mL of this extract was then manually injected
into the GC-MS. The GC-MS parameters were the same
as in Holman et al. (2010a, 2010b), and the data were
analyzed with MSD Chemstation software. We included a
greater number of CHC peaks (46 peaks) than in previous
studies (Holman et al. 2010a; 2010b; Holman 2012) by
splitting previously used peaks that contained multiple
compounds with similar retention times into separate
peaks, in order to improve resolution of the chemical pro-
file. In particular, we avoided grouping hydrocarbons be-
longing to different functional groups wherever possible,
because chemical families may have differing genetic pa-
rameters (van Zweden et al. 2010).
The number of worker offspring successfully pheno-
typed per family was 1–9 (total: 85; mean per family:
2.66 Ⳳ0.41) and the number of families (and queens)
was 32. Seven data points (4 workers and 3 queens) were
removed from the hydrocarbon mass data set because the
internal standards could not be measured clearly. All dis-
sections and chemical analyses were performed in a ran-
dom order to prevent bias.
Statistical Analysis
Since the mechanisms of hydrocarbon perception are
poorly understood (e.g., the extent to which compounds
are perceived individually or as blends is largely unknown),
we investigated the hydrocarbon data set using two dif-
ferent approaches. First, hydrocarbon peaks were treated
individually, using the mass of the peak (as calculated by
comparison with the internal standard) as the response.
Second, the chemical profile was considered as a blend.
To this end, we transformed the raw peak area data by
, where A
i
is the area of the focal hydrocarbonln (A/g(A))
i
peak and g(A) is the geometric mean of all peak areas for
that individual (Aitchison 1986), then reduced it by prin-
cipal component analysis (PCA). Together, the first three
principal components explained 86.7% of the variance in
the original data set. We also tested for relationships be-
tween hydrocarbon family and certain genetic parameters
by assigning the 46 hydrocarbon peaks to one of the fol-
lowing six categories: alkanes, alkenes, 3-methylalkanes,
other methylalkanes, dimethylalkanes, and trimethyl-
alkanes.
We quantified the heritability of worker traits, and ge-
netic correlations among them, by comparing worker sib-
This content downloaded from 130.091.128.242 on June 30, 2016 11:43:02 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
164 The American Naturalist
ling groups. Heritability and its 95% confidence intervals
(CIs) were estimated by bootstrap resampling (10
5
repli-
cates) of the intraclass correlation coefficient (hereafter
ICC; Lessells and Boag 1987; Falconer and Mackay 1996).
The ICC is probably an overestimate of both broad-sense
heritability (i.e., the proportion of phenotypic variance
explained by additive and nonadditive genetic effects) and
narrow-sense heritability (the proportion of phenotypic
variance explained by additive genetic effects) because it
potentially encompasses prefostering maternal effects,
common environmental effects, dominance, and epistasis
in addition to additive genetic variance. In particular, hy-
drocarbons are transferred among cohabiting ants (van
Zweden et al. 2010; Bos et al. 2011), which should increase
the chemical similarity within families. The ICC is also
potentially biased downward, because the workers in our
study are siblings rather than clones. The ICC is therefore
likely to be positively correlated with genetic variance but
should not be equated with it.
Pearson product moment correlations among family
trait means were used to estimate the genetic correlation
among worker traits. Although these measures might also
over- or underestimate the true genetic correlation for
similar reasons and their magnitudes should be treated
with caution, it is improbable that the signs of significant
Pearson correlations are opposite to those of the true ge-
netic correlation (Agrawal et al. 2001; Astles et al. 2006).
For example, any significant positive trait correlations pro-
vide good evidence for a positive genetic correlation, but
the magnitude of the genetic correlation may be higher
or lower than the reported value.
In comparisons of worker and queen trait values, we
believe it is most biologically relevant to consider traits
expressed in different castes as separate traits (or as the
same trait expressed in a different environment), especially
because the characters in this study show strongly caste-
biased expression. We therefore regard trait covariance be-
tween queens and their worker offspring as an intercaste
genetic correlation, similar to the cross-sex genetic cor-
relation that is often used when comparing a trait ex-
pressed in males and females (Poissant et al. 2010). We
estimated the intercaste genetic correlation as the Pearson
product moment correlation between mean offspring and
maternal trait values. This method might underestimate
the true genetic correlation by a factor of two because
mothers share only half their genes with their daughters;
however, prefostering maternal effects could inflate (or
even deflate) the estimate. We therefore present the Pear-
son correlations as an estimate of the true genetic corre-
lations (as in Agrawal et al. 2001), illustrating their likely
signs.
We present an effect size statistic and its 95% confidence
intervals for all genetic estimates, illustrating the range of
estimates that is consistent with the data. Inspection of
the confidence intervals provides a measure of certainty
in whether or not the true value might be zero, obviating
the need to correct Pvalues for multiple testing (see Naka-
gawa 2004; Nakagawa and Cuthill 2007). These intervals
aid in the interpretation of both negative results (confi-
dence in a negative result grows as the upper and lower
bounds tend to zero) and positive results (confidence in
a positive result grows as the upper and lower bounds
move away from zero).
Results
Phenotypic Correlations among Queen Traits
The number of eggs in queens’ ovaries was not significantly
related to head size ( , 95%rp⫺0.31 CIs p⫺0.59
to 0.05, , ). In contrast to expectationstp1.76 Pp.09
30
based on previous work (Holman et al. 2010a; Holman
2012), queen hydrocarbons were not significantly corre-
lated with ovary development, whether peaks were treated
individually or by principal component analysis (all P1
, ). The number of queens sampled was sub-.053 df p30
stantially lower than in previous work focusing on queens
only (Holman et al. 2010a; Holman 2012) and we used a
different fertility measure, which may explain the
discrepancy.
Intraclass Correlation Coefficients of Worker Traits
The intraclass correlation coefficient was estimated as 0.36
(95% CIs: 0.14–0.60) for worker ovary development and
0.50 for worker head size (0.24–0.78) and was high for
the first three principal components of the hydrocarbon
data (PC1: 0.63, 0.44–0.79; PC2: 0.64, 0.43–0.79; PC3:
0.84, 0.76–0.91), suggesting that direct genetic effects, in-
direct sibling effects (e.g., sibling-to-sibling hydrocarbon
transfer) and prefostering maternal effects together explain
substantial variation in these traits. When considering the
masses of individual hydrocarbon peaks, we found some
evidence that the chain length of a hydrocarbon predicted
its ICC. There was a marginally nonsignificant negative
relationship between chain length and the ICC (fig. A1;
table A1, available online; , ). If thistp1.86 Pp.069
44
effect is genuine, it implies that long-chained hydrocar-
bons are less easily transferred among siblings sharing a
nest, have a weaker genetic component, and/or are less
strongly affected by prefostering maternal effects.
Linear alkanes also had a significantly lower ICC than
all other hydrocarbon families (table A1; planned contrast:
, ), consistent with a previous findingtp2.06 Pp.046
44
that linear alkanes have lower broad-sense heritability and
are less readily transferred among nestmates than other
This content downloaded from 130.091.128.242 on June 30, 2016 11:43:02 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
Genetic Constraints on Queen Pheromones 165
Genetic correlation with worker ovaries
0.0
0.1
0.2
0.3
0.4
0.5
Chain length of hydrocarbon
27 29 31 33 35 37
3-MeC31
-0.1
Figure 1: Chain length of a hydrocarbon significantly predicted the
strength of its genetic correlation with fertility (estimated by com-
paring 32 worker sibling groups). Almost all hydrocarbons were pos-
itively genetically correlated with ovaries, though these correlations
were not statistically significant for the majority of hydrocarbons.
The gray area shows 95% confidence limits on the regression line.
types of hydrocarbons in Formica rufibarbis ants (van
Zweden et al. 2010).
Genetic Correlations among Worker Traits
There was a significantly positive correlation between the
family mean trait values for worker head size and ovary
development ( , 95% ,rp0.26 CIs p0.036–0.45 tp
77
, ), implying that worker body size and fer-2.32 Pp.023
tility are genetically correlated.
The family mean values of hydrocarbon PC1, PC2, and
PC3 were all uncorrelated with ovary development (all
). However, the masses of seven individual peaksP1.35
30
were positively correlated with ovaries (table A2, available
online): these included 3-MeC
27
and 3-MeC
29
. The queen
pheromone 3-MeC
31
was not significantly related to ovary
development, though the statistically supported range of
effect sizes was predominantly positive, suggesting that we
have low confidence in the null hypothesis ( , 95%rp0.21
, , ). The other sig-CIs p⫺0.05–0.60 tp1.76 Pp.09
29
nificant peaks were dimethylalkanes and trimethylalkanes
(table A2). There was also a negative relationship between
the chain length of a hydrocarbon and the genetic cor-
relation between that hydrocarbon and ovary develop-
ment: in other words, shorter hydrocarbons showed evi-
dence of having a more positive genetic correlation with
ovaries (fig. 1; linear regression: , ).tp4.25 Pp.0001
44
Additionally, the estimated genetic correlations with ova-
ries were significantly higher for 3-methylalkanes than for
alkanes (contrast: , ) and other typestp2.15 Pp.038
40
of methylalkanes ( , ), and borderlinetp2.17 Pp.036
40
nonsignificantly higher than for alkenes ( ,tp1.99
40
). Lastly, the signs of 45/46 of the estimated cor-Pp.053
relations between individual hydrocarbons and ovaries
were positive (the exception was C
35:1
; table A2), further
suggesting that the production of eggs and hydrocarbons
are genetically correlated.
Intercaste Genetic Correlations for the Same Trait
The number of eggs in a queen’s ovaries predicted the
mean ovarian development of her worker offspring (fig.
2A; ; 95% , ,rp0.39 CIs p0.05–0.65 tp2.34 Pp
30
). This result is consistent with a positive intercaste.026
genetic correlation for ovary development and/or a pre-
fostering maternal effect. However, we note that queen egg
number was unrelated to worker size (see below), ruling
out the plausible possibility that fertile queens made larger
workers via a maternal effect (Schwander et al. 2008), in
turn increasing worker fertility (because larger workers are
more fertile; see above).
Queen head size did not significantly predict mean off-
spring head size, consistent with a weak or nonexistent
intercaste genetic correlation for head size ( , 95%rp0.21
,,).CIs p⫺0.15–0.51 tp1.17 Pp.25
30
The hydrocarbon profile of queens (as measured by
PCA) did not significantly predict that of their worker
offspring. However, inspection of the effect size 95% CIs
suggests that we cannot confidently rule out some degree
of parent-offspring similarity. Although neither result was
statistically significant, queens with high values of PC2
tended to produce workers with low values of PC2 (rp
; 95% , , ),⫺0.31 CIs p⫺0.59–0.05 tp1.76 Pp.088
30
while maternal and offspring values were positively related
for PC3 ( ; 95% , ,rp0.33 CIs p⫺0.025–0.60 tp1.89
30
). PC2 described the relative abundance of hy-Pp.068
drocarbons previously found to be overexpressed by
queens (Holman et al. 2010b) and augmented by juvenile
hormone treatment (Holman 2012), namely linear al-
kanes, 3-methylalkanes, and the alkene C
31:1
, while PC3
described the relative abundance of long-chained hydro-
carbons (table A3, available online). We therefore found
a nonsignificant trend for queens with particularly queen-
like hydrocarbons to produce workers with especially
worker-like hydrocarbons and for queens with long-chain
hydrocarbons to produce workers that also had long-chain
hydrocarbons.
When analyzing the mass of individual CHC peaks, only
two peaks were significantly positively correlated in moth-
ers and their offspring; these were 7,11,15-triMeC
31
and
C
35:1
(table A4, available online). However, the estimated
intercaste genetic correlations were significantly higher for
This content downloaded from 130.091.128.242 on June 30, 2016 11:43:02 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
166 The American Naturalist
Number of eggs in mother's ovaries
Mean mass of 3-MeC31 in offspring (ng)
6
10
14
02468
18
2
Number of eggs in mother’s ovaries
Mean ovary development of offspring
1.0
1.5
2.0
2.5
3.0
02468
A. B.
Figure 2: Queens with many eggs in their ovaries produced workers with significantly more well-developed ovaries (A) and a higher mean
mass of the queen pheromone 3-MeC
31
on the cuticle (B), consistent with a positive intercaste genetic correlation for fertility and genetic
links between fertility and queen pheromone production, respectively (np32 families). The large points in Ashow overplotting, with
points of increasing size representing 1, 2, 4, and 7 data points, respectively.
Chain length of hydrocarbon
Inter-caste genetic correlation with ovaries
0.0
0.1
0.2
0.3
0.4
27 29 31 33 35 37
-0.2
3-MeC31
0.5
Figure 3: There was a significant, nonlinear relationship betweenthe
chain length of a hydrocarbon and its intercaste genetic correlation
with ovary activation (as measured by comparing the number of eggs
in queens’ ovaries with worker hydrocarbon mass; np32 families).
Almost all hydrocarbons showed a positive genetic correlation with
ovaries, and the queen pheromone 3-MeC
31
had a particularly
strongly positive genetic correlation.
3-methylalkanes and alkenes relative to alkanes and di-
methylalkanes (table A4; contrasts: all ), againp!0.025
40
consistent with differences in the heritabilities of different
hydrocarbon families.
Intercaste Genetic Correlations among Different Traits
Queen egg number did not predict worker head size,
though there was a nonsignificant trend for a positive
correlation ( , 95% ,rp0.34 CIs p⫺0.011–0.61 tp
30
, ). For the reciprocal comparison, queen1.97 Pp.058
head size and worker ovarian development, there was no
relationship ( , 95% ,rp⫺0.11 CIs p⫺0.44–0.25 tp
30
,).0.61 Pp.54
The number of eggs in queens’ ovaries was significantly
positively correlated with the mass of eight cuticular hy-
drocarbon peaks in their worker offspring (table A5, avail-
able online). Of these, the queen pheromone 3-MeC
31
showed an especially strong correlation (fig. 2B;rp
, 95% , , ), sug-0.49 CIs p0.16–0.72 tp3.00 Pp.005
30
gesting a genetic correlation between fertility in queens
and queen pheromone production in workers. The other
peaks were 7,11-MeC
29
, 3,9- and 3,11-diMeC
29
, and five
trimethylalkanes. The signs of 45/46 of the estimated cor-
relations were positive (the exception was again C
35:1
; table
A5), providing further evidence that the production of eggs
and hydrocarbons is genetically correlated. In a similar
fashion to the worker data (cf. figs. 1, 3), the carbon chain
length of a hydrocarbon was a significant predictor of its
genetic correlation with ovaries (fig. 3; linear effect of chain
length: , ), though this time there wastp2.29 Pp.027
43
also a significant quadratic effect of chain length (tp
43
, ). This result suggests that hydrocarbons of2.41 Pp.021
intermediate chain length are more positively genetically
correlated with fertility than are long- and short-chained
hydrocarbons.
If we instead estimated the genetic correlations between
the mean trait value for worker ovaries and the mass of
This content downloaded from 130.091.128.242 on June 30, 2016 11:43:02 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
Genetic Constraints on Queen Pheromones 167
each CHC produced by the mother queen, only one cor-
relation was significantly different from zero (table A6,
available online). Maternal mass of 5,9,13-triMeC
33
was
negatively correlated with worker ovary development
( , 95% , ,rp⫺0.37 CIs p⫺0.64 to ⫺0.19 tp2.15
29
).Pp.04
Discussion
The data provide evidence that fertility and hydrocarbon
production are heritable but not genetically independent
in the ant Lasius niger. These genetic relationships may
impose evolutionary constraints on the evolution of both
fertility signaling and queen-worker dimorphism and
thereby contribute to the maintenance of honest signaling
and worker reproduction.
We found evidence for a significant genetic correlation
between ovary activation and the 3-methylalkanes 3-MeC
27
and 3-MeC
29
when comparing worker families; there was
also a nonsignificant trend for a positive genetic correla-
tion between ovaries and the queen pheromone 3-MeC
31
.
Additionally, we found a strong, significant intercaste ge-
netic correlation between the mass of the queen phero-
mone 3-MeC
31
and ovary development when comparing
maternal fertility with offspring 3-MeC
31
production. To-
gether, these findings suggest that 3-MeC
31
and other 3-
methylalkanes signal a genetic tendency for high fertility,
in addition to advertising an individual’s current fertility
level as found previously (Holman et al. 2010a).
A genetic correlation between queen pheromone pro-
duction and fertility could constrain the evolution of dis-
honest signaling. If queens were under directional selection
to produce more queen pheromone (West-Eberhard 1983),
for example, to reduce their chance of being executed
(Sommer and Ho¨lldobler 1995; Holman et al. 2010a;
Wurm et al. 2010), solicit a great share of workers’ atten-
tion (Hannonen et al. 2002), or more effectively inhibit
worker reproduction (Holman et al. 2010b), any response
to selection for increased queen pheromone production
would produce a correlated response in queen fertility. The
link between fertility and the pheromone would therefore
be maintained, preserving signal honesty, and continued
exaggeration of the pheromone would be limited by con-
current selection on fertility (Blows and Hoffmann 2005).
Our hypothesis that genetic constraints may maintain
signal honesty over evolutionary time is applicable to many
other taxa and signaling systems. For example, sexually
selected acoustic signals are strongly genetically correlated
with body size in zebra finches (Forstmeier et al. 2009)
and hissing cockroaches (Clark and Moore 1995), sug-
gesting that many polymorphic loci affecting song or hiss-
ing also affect size. Therefore, these signals cannot easily
evolve without a corresponding change in size, preserving
signal honesty. Likewise, Emlen et al. (2012) showed ex-
perimentally that rhinoceros beetles’ horns are especially
sensitive to perturbations of the insulin/IGF pathway,
making them strong predictors of overall condition. They
present the intriguing hypothesis that the proximate mech-
anism by which horn size was exaggerated in ancestral
small-horned beetles, namely elevated sensitivity of the
horn to growth factors, now ensures its honesty across the
population. This hypothesis also implies that there should
be little genetic variation for horn size that is independent
of condition, ensuring that condition and horn size remain
correlated over evolutionary time. A final example is pro-
vided by green beards, which are loci that produce three
traits: a signal, a preference for that signal, and a social
trait (reviewed in Gardner and West 2010). The “beard”
signal is an honest indicator of the social trait because a
single locus produces both traits, that is, there is a perfect
genetic correlation. Populations of green beards are there-
fore thought to be stable unless “false beards” that produce
the signal but do not perform the social behavior are able
to evolve (Gardner and West 2010). Therefore, genetic
constraints that prevent the decoupling of the beard and
the social trait are essential for long-term honesty. As in
the rhinoceros beetle example, such a constraint may arise
from the proximate mechanism. Two known green beard
loci in microorganisms produce cell adhesion proteins that
are able to bind to the same proteins expressed on other
cells (Queller et al. 2003; Smukalla et al. 2008). Therefore,
producing the signal (the adhesion protein) necessitates
that the bearer performs the social behavior (binding),
ensuring honesty.
The comparatively strong genetic link between 3-meth-
ylalkanes and ovaries may also explain why 3-MeC
31
is a
queen pheromone in L. niger, as opposed to some other
hydrocarbon. Tinbergen (1952) suggested that signals rep-
resent “ritualized” forms of preexisting cues associated
with a given behavior or trait. If some hydrocarbons were
better predictors of fertility than others at the origin of
sociality, for example, because their synthesiswas especially
affected by changes in reproductive physiology, ancestral
workers might have evolved to respond to these hydro-
carbons when performing behaviors that are only bene-
ficial in the presence of a fertile relative (e.g., deciding
whether to remain sterile). Senders might then be selected
to exaggerate the cue in order to increase the receiver’s
response, causing the cue to become a signal (by defini-
tion). The current data are consistent with the ritualization
hypothesis because the strong genetic correlations between
3-methylalkanes and fertility suggest that these traits are
especially genetically or physiologically related. Moreover,
Holman (2012) found that juvenile hormone treatment
affected both fertility and the production of the queen-
like hydrocarbons C
27
,C
29
, 3-MeC
29
, and 3-MeC
31
, sug-
This content downloaded from 130.091.128.242 on June 30, 2016 11:43:02 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
168 The American Naturalist
gesting either causal links between the production of eggs
and the expression of cuticular hydrocarbons or common
endocrine regulation of both traits (Fan et al. 2002; Blom-
quist and Bagne`res 2010).
We also found that the genetic correlation with ovaries
was highest for hydrocarbons with short or intermediate
chain length, both when contrasting worker families and
when comparing queens with their offspring. There was
also a marginally nonsignificant relationship between the
relative amount of long-chain hydrocarbons produced by
queens and their offspring. These results are consistent
with the presence of additive genetic variation for the elon-
gase enzymes that control hydrocarbon chain length
(Blomquist and Bagne`res 2010) and imply genetic links
between elongase enzymes and fertility. If shorter-chained
hydrocarbons were indeed the most genetically linked to
fertility across taxa, we might expect queens to have shorter
hydrocarbons than workers in most ant species. There is
mixed support for this prediction: queens possess more
relatively short-chained hydrocarbons than workers in the
ants Camponotus floridanus (Endler et al. 2004) and Lin-
epithema humile (de Biseau et al. 2004), but long-chain
hydrocarbons denote higher fertility in Harpegnathos sal-
tator (Liebig et al. 2000), and mean chain length increases
during reproductive development in queen L. niger
(Holman et al. 2010a). A negative genetic correlation be-
tween fertility and hydrocarbon chain length could also
influence the evolution of chain length; Kwan and Rundle
(2010) found that desiccation-stressed Drosophila melano-
gaster populations evolved longer hydrocarbons (suggest-
ing that long hydrocarbons are better for desiccation re-
sistance), and our results imply that selection on fecundity
might be one factor favoring short hydrocarbons in benign
conditions.
An unanticipated but important result of the study was
that we found evidence for an intercaste genetic correlation
for ovary development. This result suggests a possible con-
straint on the evolution of queen reproductive output:
selection for increased queen fertility would likely be ac-
companied by an increase in worker fertility, which is
thought to be detrimental to queen and colony produc-
tivity (reviewed in Wenseleers et al. 2004). Our data there-
fore suggest the existence of a “caste load,” comparable to
the “gender load” imposed by intralocus sexual conflict
(Prasad et al. 2007; Connallon et al. 2010; Innocenti and
Morrow 2010), in which a trait cannot be simultaneously
optimized in queens and workers because of intercaste
genetic correlations and caste-specific selection pressures.
Various authors have proposed that caste-biased gene ex-
pression should evolve in order to weaken the intercaste
genetic correlation for traits in which selection differs be-
tween castes (West-Eberhard 1996; Gadagkar 1997; Links-
vayer and Wade 2005), but our data suggest that selection
has not achieved total “genetic release” (Gadagkar 1997)
of queen and worker fertility in L. niger. Hints at the extent
of genetic release in L. niger were provided by a transcrip-
tomic study, which found that 16/369 of the examined
transcripts showed caste-specific expression (Graff et al.
2007). Queens had more of two transcripts putatively as-
sociated with oogenesis, though these transcripts were also
present in workers (e.g., 11 times lower for a vitellogenin
homolog).
The observed intercaste genetic correlation for ovary
development also supports the hypothesis that worker re-
production may partly result from maladaptation, in con-
trast to the widely accepted but largely untested hypothesis
that fertile workers are selfish “cheaters” (Smith et al. 2009)
maintained at a given level by selection (Wenseleers et al.
2004). Using a theoretical model, Van Dyken et al. (2011)
showed that a sizeable number of cheaters (e.g., repro-
ductive workers) can exist at mutation-selection balance
in a cooperative system even when cheating lowers the
cheater’s inclusive fitness. In light of our data, we propose
a related but distinct hypothesis, that worker reproduction
is maintained because it cannot be eliminated without
causing a costly decline in queen fecundity. That is, alleles
that cause a maladaptive increase in fertility when ex-
pressed in workers might experience net positive selection
if they have sufficiently beneficial effects on queen fertility.
Genetic variation along the putative multivariate axis of
selection (i.e., higher queen fertility and lower worker fer-
tility) might well be exhausted (Blows and Hoffmann
2005), especially because intragroup relatedness is high in
many social insects (including our study species; Corley
and Fjerdingstad 2011), strengthening the response to se-
lection on worker traits (see Van Dyken et al. 2011). The
proximate and ultimate causes of worker reproduction re-
main an exciting topic for future work. Empirical work
should attempt to evaluate the strength and generality of
the caste load in advanced and primitively eusocial species.
Theoretical models could benefit by incorporating (evolv-
able) genetic links between queen and worker traits, and
by considering nonadaptive explanations for patterns of
reproductive skew in animal societies.
Acknowledgments
We are grateful to M. W. Blows and the reviewers for
valuable comments, and C. Leroy and C. Olsen for tech-
nical support. L.H. and T.A.L. were supported by Marie
Curie fellowships, and all authors benefited from funding
to the Centre for Social Evolution, Copenhagen, from the
Danish National Research Foundation.
This content downloaded from 130.091.128.242 on June 30, 2016 11:43:02 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
Genetic Constraints on Queen Pheromones 169
Literature Cited
Agrawal, A. F., E. D. Brodie III, and J. Brown. 2001. Parent-offspring
coadaptation and the dual genetic control of maternal care. Science
292:1710–1712.
Aitchison, J. 1986. The statistical analysis of compositional data.
Blackburn, Caldwell, NJ.
Andersson, M. 1980. Why are there so many threat displays? Journal
of Theoretical Biology 86:773–781.
Arnqvist, G. 2006. Sensory exploitation and sexual conflict. Philo-
sophical Transactions of the Royal Society B: Biological Sciences
361:375–386.
Astles, P. A., A. J. Moore, and R. F. Preziosi. 2006. A comparison of
methods to estimate cross-environment genetic correlations. Jour-
nal of Evolutionary Biology 19:114–122.
Blomquist, G. C., and A. G. Bagne`res, eds. 2010. Insect hydrocarbons:
biology, biochemistry and chemical ecology. Cambridge University
Press, Cambridge.
Blows, M. W., and A. A. Hoffmann. 2005. A reassessment of genetic
limits to evolutionary change. Ecology 86:1371–1384.
Bos, N., L. Grinsted, and L. Holman. 2011. Wax on, wax off: nest
soil facilitates indirect transfer of recognition cues between ant
nestmates. PLoS ONE 6:e19435.
Brilot, B. O., and R. A. Johnstone. 2003. The limits to cost-free
signalling of need between relatives. Proceedings of the Royal So-
ciety B: Biological Sciences 270:1055–1060.
Charlton, B. D., W. A. H. Ellis, A. J. McKinnon, G. J. Cowin, J.
Brumm, K. Nilsson, and W. T. Fitch. 2011. Cues to body size in
the formant spacing of male koala (Phascolarctos cinereus) bellows:
honesty in an exaggerated trait. Journal of Experimental Biology
214:3414–3422.
Clark, D. C., and A. J. Moore. 1995. Genetic aspects of communi-
cation during male-male competition in the Madagascar hissing
cockroach: honest signalling of size. Heredity 75:198–205.
Connallon, T., R. M. Cox, and R. Calsbeek. 2010. Fitness conse-
quences of sex-specific selection. Evolution 64:1671–1682.
Corley, M., and E. Fjerdingstad. 2011. Mating strategies of queens
in Lasius niger ants: is environment type important? Behavioral
Ecology and Sociobiology 65:889–897.
de Biseau, J.-C., L. Passera, D. Daloze, and S. Aron. 2004. Ovarian
activity correlates with extreme changes in cuticular hydrocarbon
profile in the highly polygynous ant, Linepithema humile. Journal
of Insect Physiology 50:585–593.
Emlen, D. J., I. A. Warren, A. Johns, I. Dworkin, and L. C. Lavine.
2012. A mechanism of extreme growth and reliable signaling in
sexually selected ornaments and weapons. Science 337:860–864.
Endler, A., J. Liebig, T. Schmitt, J. E. Parker, G. R. Jones, P. Schreier,
and B. Ho¨lldobler. 2004. Surface hydrocarbons of queen eggs reg-
ulate worker reproduction in a social insect. Proceedings of the
National Academy of Sciences of the USA 101:2945–2950.
Falconer, D. S., and T. F. C. Mackay. 1996. Introduction to quanti-
tative genetics. Pearson, Harlow, Essex.
Fan, Y., J. Chase, V. L. Sevala, and C. Schal. 2002. Lipophorin-facil-
itated hydrocarbon uptake by oocytes in the German cockroach
Blattella germanica (L.). Journal of Experimental Biology 205:781–
790.
Forstmeier, W., C. Burger, K. Temnow, and S. Deregnaucourt. 2009.
The genetic basis of zebra finch vocalisations. Evolution 63:2114–
2130.
Gadagkar, R. 1997. The evolution of caste polymorphism in social
insects: genetic release followed by diversifying evolution. Journal
of Genetics 76:167–179.
Gardner, A., and S. A. West. 2010. Greenbeards. Evolution 64:25–
38.
Getty, T. 2006. Sexually selected signals are not similar to sports
handicaps. Trends in Ecology & Evolution 21:83–88.
Grafen, A. 1990. Biological signals as handicaps. Journal of Theo-
retical Biology 144:517–546.
Graff, J., S. Jemielity, J. D. Parker, K. M. Parker, and L. Keller. 2007.
Differential gene expression between adult queens and workers in
the ant Lasius niger. Molecular Ecology 16:675–683.
Guilford, T., and M. S. Dawkins. 1991. Receiver psychology and the
evolution of animal signals. Animal Behaviour 42:1–14.
Hannonen, M., M. F. Sledge, S. Turillazzi, and L. Sundstro¨m. 2002.
Queen reproduction, chemical signalling and worker behaviour in
polygyne colonies of the ant Formica fusca. Animal Behaviour 64:
477–485.
Hine, E., K. McGuigan, and M. W. Blows. 2011. Natural selection
stops the evolution of male attractiveness. Proceedings of the Na-
tional Academy of Sciences of the USA 108:3659–3664.
Holman, L. 2012. Costs and constraints conspire to produce honest
signalling: insights from an ant queen pheromone. Evolution 66:
2094–2105.
Holman, L., S. Dreier, and P. d’Ettorre. 2010a. Selfish strategies and
honest signalling: reproductive conflicts in ant queen associations.
Proceedings of the Royal Society B: Biological Sciences 277:2007–
2015.
Holman, L., C. G. Jørgensen, J. Nielsen, and P. d’Ettorre. 2010b.
Identification of an ant queen pheromone regulating worker ste-
rility. Proceedings of the Royal Society B: Biological Sciences 277:
3793–3800.
Innocenti, P., and E. H. Morrow. 2010. The sexually antagonistic
genes of Drosophila melanogaster. PLoS Biology 8:e1000335.
Keller, L., and P. Nonacs. 1993. The role of queen pheromones in
social insects: queen control or queen signal? Animal Behaviour
45:787–794.
Kwan, L., and H. D. Rundle. 2010. Adaptation to desiccation fails
to generate pre- and postmating isolation in replicate Drosophila
melanogaster laboratory populations. Evolution 64:710–723.
Lachmann, M., S. Sza´ mado´ , and C. T. Bergstrom. 2001. Cost and
conflict in animal signals and human language. Proceedings of the
National Academy of Sciences of the USA 98:13189–13194.
Lessells, C. M., and P. T. Boag. 1987. Unrepeatable repeatabilities: a
common mistake. Auk 104:116–121.
Liebig, J., C. Peeters, N. J. Oldham, C. Markstadter, andB. Ho¨ lldobler.
2000. Are variations in cuticular hydrocarbons of queens and
workers a reliable signal of fertility in the ant Harpegnathos saltator?
Proceedings of the National Academy of Sciences of the USA 97:
4124–4131.
Linksvayer, T. A. 2006. Direct, maternal, and sibsocial genetic effects
on individual and colony traits in an ant. Evolution 60:2552–2561.
Linksvayer, T. A., and M. J. Wade. 2005. The evolutionary origin and
elaboration of sociality in the aculeate Hymenoptera: maternal
effects, sib-social effects, and heterochrony. Quarterly Review of
Biology 80:317–336.
Maynard Smith, J., and D. G. C. Harper. 1995. Animalsignals: models
and terminology. Journal of Theoretical Biology 177:305–311.
———. 2003. Animal signals. Oxford University Press, Oxford.
Nakagawa, S. 2004. A farewell to Bonferroni: the problems of low
This content downloaded from 130.091.128.242 on June 30, 2016 11:43:02 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
170 The American Naturalist
statistical power and publication bias. Behavioral Ecology 15:1044–
1045.
Nakagawa, S., and I. C. Cuthill. 2007. Effect size, confidence interval
and statistical significance: a practical guide for biologists. Biolog-
ical Reviews of the Cambridge Philosophical Society 82:591–605.
Poissant, J., A. J. Wilson, and D. W. Coltman. 2010. Sex-specific
genetic variance and the evolution of sexual dimorphism: a sys-
tematic review of cross-sex genetic correlations. Evolution 64:97–
107.
Porter, S. D., and W. R. Tschinkel. 1986. Adaptive value of nanitic
workers in newly founded red imported fire ant colonies (Hy-
menoptera, Formicidae). Annals of the Entomological Society of
America 79:723–726.
Prasad, N. G., S. Bedhomme, T. Day, and A. K. Chippindale. 2007.
An evolutionary cost of separate genders revealed by male-limited
evolution. American Naturalist 169:29–37.
Queller, D. C., E. Ponte, S. Bozzaro, and J. E. Strassmann. 2003.
Single-gene greenbeard effects in the social amoeba Dictyostelium
discoideum. Science 299:105–106.
Reby, D., and K. McComb. 2003. Anatomical constraints generate
honesty: acoustic cues to age and weight in the roars of red deer
stags. Animal Behaviour 65:519–530.
Schwander, T., J.-Y. Humbert, C. S. Brent, S. H. Cahan, L. Chapuis,
E. Renai, and L. Keller. 2008. Maternal effect on female caste
determination in a social insect. Current Biology 18:265–269.
Searcy, W. A., and S. Nowicki. 2005. The evolution of animal com-
munication: reliability and deception in signaling systems. Prince-
ton University Press, Princeton, NJ.
Smith, A. A., B. Ho¨lldobler, and J. Liebig. 2009. Cuticular hydro-
carbons reliably identify cheaters and allow enforcement of altru-
ism in a social insect. Current Biology 19:78–81.
Smukalla, S., M. Caldara, N. Pochet, A. Beauvais, S. Guadagnini, C.
Yan, M. D. Vinces, et al. 2008. FLO1 is a variable green beard gene
that drives biofilm-like cooperation in budding yeast. Cell 135:
726–737.
Sommer, K., and B. Ho¨lldobler. 1995. Colony founding by queen
association and determinants of reduction in queen number in
the ant Lasius niger. Animal Behaviour 50:287–294.
Sza´ mado´, S. 2011. The cost of honesty and the fallacy of the handicap
principle. Animal Behaviour 81:3–10.
Tinbergen, N. 1952. Derived activities: their causation, biological
significance, origin and emancipation during evolution. Quarterly
Review of Biology 27:1–32.
Van Dyken, J. D., T. A. Linksvayer, and M. J. Wade. 2011. Kin se-
lection–mutation balance: a model for the origin, maintenance,
and consequences of social cheating. American Naturalist 177:288–
300.
Van Homrigh, A., M. Higgie, K. McGuigan, and M. W. Blows. 2007.
The depletion of genetic variance by sexual selection. Current
Biology 17:528–532.
van Zweden, J. S., J. B. Brask, J. H. Christensen, J. J. Boomsma, T.
A. Linksvayer, and P. D’Ettorre. 2010. Blending of heritable rec-
ognition cues among ant nestmates creates distinct colony gestalt
odours but prevents within-colony nepotism. Journal of Evolu-
tionary Biology 23:1498–1508.
Wenseleers, T., H. Helantera¨, A. G. Hart, and F. L. W. Ratnieks. 2004.
Worker reproduction and policing in insect societies: an ESS anal-
ysis. Journal of Evolutionary Biology 17:1035–1047.
West-Eberhard, M. J. 1983. Sexual selection, social competition, and
speciation. Quarterly Review of Biology 58:155–183.
———. 1996. Wasp societies as microcosms for the study of devel-
opment and evolution. Pages 290–317 in S. Turillazzi and M. J.
West-Eberhard, eds. Natural history and evolution of paper-wasps.
Oxford University Press, Oxford.
Wurm, Y., J. Wang, and L. Keller. 2010. Changes in reproductive
roles are associated with changes in gene expression in fire ant
queens. Molecular Ecology 19:1200–1211.
Zahavi, A. 1975. Mate selection: a selection for a handicap. Journal
of Theoretical Biology 53:205–214.
Zahavi, A., and A. Zahavi. 1997. The handicap principle: a missing
piece of Darwin’s puzzle. Oxford University Press, New York.
Associate Editor: J. Albert C. Uy
Editor: Judith L. Bronstein
Worker tends to Lasius niger queen about to lift off in search of a new colony. Photo by Stuart Hogton; www.hogton.com.
This content downloaded from 130.091.128.242 on June 30, 2016 11:43:02 AM
All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).