Male behaviour drives assortative reproduction during the initial stage of secondary contact

Abstract

Phenotypic divergence in allopatry can facilitate speciation by reducing the likelihood that individuals of different lineages hybridise during secondary contact. However, few studies have established the causes of reproductive isolation in the crucial early stages of secondary contact. Here we establish behavioural causes of assortative reproduction between two phenotypically divergent lineages of the European wall lizard (Podarcis muralis), which have recently come into secondary contact. Parentage was highly assortative in experimental contact zones. However, despite pronounced divergence in male phenotypes, including chemical and visual sexual signals, there was no evidence that females discriminated between males of the two lineages in staged interactions or under naturalistic free-ranging conditions. Instead, assortative reproduction was driven by male mate preferences and, to a lesser extent, male-male competition. The effects were more pronounced when the habitat structure promoted high lizard densities. These results emphasize that assortative reproduction can occur in the absence of female choice, and that male behaviour may play an important role in limiting hybridisation during the initial stages of secondary contact. This article is protected by copyright. All rights reserved.

Full text

Male behaviour drives assortative reproduction during the
initial stage of secondary contact
R. J. P. HEATHCOTE*
1
,G.M.WHILE*†,H.E.A.MACGREGOR*†,J.SCIBERRAS*,
C. LEROY‡,P.D’ETTORRE‡&T.ULLER*§
*Department of Zoology, Edward Grey Institute, University of Oxford, Oxfo rd, UK
†School of Biological Sciences, University of Tasmania, Sandy Bay, Tas., Australia
‡Laboratory of Experimental and Comparative Ethology, Sorbonne Paris Cit!
e, University of Paris 13, Villetaneuse, France
§Department of Biology, Lund University, Lund, Sweden
Keywords:
assortative mating;
hybridization;
male–male competition;
mate choice;
Podarcis muralis;
secondary contact.
Abstract
Phenotypic divergence in allopatry can facilitate speciation by reducing the
likelihood that individuals of different lineages hybridize during secondary
contact. However, few studies have established the causes of reproductive
isolation in the crucial early stages of secondary contact. Here, we establish
behavioural causes of assortative reproduction between two phenotypically
divergent lineages of the European wall lizard (Podarcis muralis), which have
recently come into secondary contact. Parentage was highly assortative in
experimental contact zones. However, despite pronounced divergence in
male phenotypes, including chemical and visual sexual signals, there was no
evidence that females discriminated between males of the two lineages in
staged interactions or under naturalistic free-ranging conditions. Instead,
assortative reproduction was driven by male mate preferences and, to a les-
ser extent, male–male competition. The effects were more pronounced
when the habitat structure promoted high lizard densities. These results
emphasize that assortative reproduction can occur in the absence of female
choice and that male behaviour may play an important role in limiting
hybridization during the initial stages of secondary contact.
Introduction
A fundamental requirement for speciation is that popu-
lations reproduce assortatively. This may be initiated by
morphological and behavioural divergence during geo-
graphical isolation, which limits interbreeding following
secondary contact (Mayr, 1963). Precopulatory mecha-
nisms such as heterospecific recognition and avoidance
are clearly pivotal in maintaining reproductive barriers
in currently sympatric lineages. Indeed, many species
that rarely hybridize in the wild will frequently do so
in captivity (McCarthy, 2006), suggesting precopulatory
mechanisms are important in maintaining species
boundaries over and above post-copulatory factors such
as hybrid sterility (Haldane, 1922; Hewitt et al., 1987).
The majority of studies examining how behavioural
mechanisms mediate gene flow are undertaken in
established (ancient) hybrid zones (Coyne & Orr, 2004;
Price, 2008; Abbott et al., 2013). These studies have
usually focussed on female choice, as the greater per
capita investment in reproduction by females should
result in their experiencing stronger selection against
costly hybrid matings compared to males (Wirtz, 1999;
Randler, 2002). This has led to the suggestion that
divergence in traits involved in female mate choice due
to sexual selection may play an integral role in specia-
tion, facilitating heterospecific mating avoidance and
reducing hybridization upon secondary contact (Lande,
1981; West-Eberhard, 1983; Panhuis et al., 2001).
Correspondence: R. J. P. Heathcote, Edward Grey Institute, Department of
Zoology, University of Oxford, Oxford OX1 3PS, UK.
Tel.: +44 1392 724 694; fax: +44(0)1392724623;
e-mail: r.j.p.heathcote@gmail.com
and
T. Uller, Department of Biology, Lund University, So
¨lvegatan 37, 223
62 Lund, Sweden.
Tel.: +46 46 222 30 94; e-mail: tobias.uller@biol.lu.se
1
Current address: Centre for Research in Animal Behaviour, College of
Life and Environmental Sciences, University of Exeter, Exeter EX4
4QG, UK.
ª2016 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY. J. EVOL. BIOL.
1
JOURNAL OF EVOLUTIONARY BIOLOGY ª2016 EUROPE AN SOCIETY FOR EVOLUTIONARY BIO LOGY
doi: 10.1111/jeb.12840

However, the mechanisms that mediate gene flow in
established hybrid zones may not be representative of
those important in the early stages of secondary con-
tact. This is because these mechanisms are themselves
likely to be the evolutionary consequences of costly
hybrid matings (i.e. reinforcement) (Brodsky et al.,
1988; Wirtz, 1999; Tynkkynen et al., 2004). Indeed,
studies from invasive species suggest initial female dis-
crimination is often weak due to a lack of historical
selection and that secondary contact can result in rapid
hybridization and loss of genetic integrity before rein-
forcement can occur (Echelle & Connor, 1989; Rhymer
& Simberloff, 1996; Huxel, 1999; Mu~
noz-Fuentes et al.,
2007). Speciation may therefore be a much less fre-
quent consequence of secondary contact than often
believed, with homogenization of populations and lin-
eage extinction being more common outcomes.
An important challenge in speciation research is
therefore to identify the behavioural processes that
can maintain barriers to gene flow before reinforce-
ment has occurred. However, empirical studies docu-
menting the phenotypic and ecological conditions that
mediate hybridization during recent secondary contact
are rare (see Grant & Grant (2009) for an exception).
The role of male behaviour in these circumstances is
in particular poorly understood, although there is a
growing awareness of its importance in mediating
hybridization between established species (e.g. Peter-
son et al., 2005; Svensson et al., 2007; Edward &
Chapman, 2011). Due to their strong influence on
intra- and interspecific interactions, environmental
conditions during secondary contact may also play an
important role in mediating hybridization (Doorn
et al., 1998; Boughman, 2001). For instance, hybrid
matings are more likely when conspecific mates are
rare (Hubbs, 1955; Pearson, 2000), implying that
environmental conditions that increase local popula-
tion density may contribute to the spatial and tempo-
ral variation in introgression often found in hybrid
zones (such as in studies by Klingenberg et al., 2000;
Pearson, 2000; Bronson et al., 2003). This factor may
therefore be particularly important in reducing intro-
gression during initial secondary contact, before
strong conspecific mate preference has evolved.
In this paper, we experimentally tested how precopu-
latory behaviour promotes assortative mating between
two phenotypically divergent lineages of the European
wall lizard (Podarcis muralis). These lineages, one native
to western Europe and the other native to central and
western Italy, have recently come into secondary con-
tact in several parts of Europe due to human introduc-
tions (Schulte et al., 2012a; Michaelides et al., 2013,
2015; While et al., 2015a). Such contact zones created
by human introductions provide excellent systems for
studying secondary contact because the mechanisms
and environmental factors mediating gene flow that are
normally lost to history can be directly observed (Sax
et al., 2007; Blackburn et al., 2009; Senn & Pemberton,
2009).
Our study had three main objectives. First, we aimed
to establish the extent to which reproduction is assorta-
tive using replicated, semi-natural enclosures that mim-
icked secondary contact. Second, we tested whether
assortativity depends on the spatial clustering of
basking sites, an important resource for heliothermic
(basking) lizard species that is likely to influence popu-
lation density. Third, we used analyses of behavioural
observations from our semi-natural enclosures com-
bined with controlled behavioural trials to establish the
relative contribution played by both male and female
precopulatory behaviours (including those influenced
by olfactory sexual signalling) in driving reproductive
assortativity upon secondary contact.
Materials and methods
Study species
The European wall lizard (Podarcis muralis) is a small
(up to 75 mm snout-to-vent length (SVL)) diurnal lac-
ertid native throughout western and southern Europe.
Over the last 30–150 years, multiple non-native popu-
lations have become established in Germany, North
America and the UK (Schulte et al., 2012a; Michaelides
et al., 2013). Several of the introduced populations in
Germany and the UK are composed of two main lin-
eages (which we refer to as ‘western European’ and
‘Italian’) that separated around two million years ago
(Schulte et al., 2012b; Michaelides et al., 2013, 2015).
All animals used in this study were from non-native
UK populations (detailed below; Table S1). The animals
we refer to as the western European lineage have haplo-
types from the western or eastern France subclades, and
the Italian lineage all have haplotypes from the Tuscan
or Venetian clade (as per Schulte et al., 2012a) and origi-
nate from the Tuscany or Bologna–Modena regions. The
lineages differ conspicuously in dorsolateral coloration
(brown in western Europe, green in Italy; Fig. 1) and
the Italian lineage exhibits exaggeration of characters
previously demonstrated to be under sexual selection in
lacertid lizards and Podarcis, including relative head size,
bite force, coloration and body mass (Bohme, 1986;
Bra~
na, 1996; While et al., 2015a; see below).
Phenotypic differences between lineages
Morphometrics
All lizards that were used in the experiments described
below had their SVL measured to the nearest 1 mm,
their mass measured to the closest 0.01 g, and the
width and length of the head measured to the nearest
0.1 mm. To determine how the lineages and sexes
differed in body and head size, we ran two separate
linear models, firstly with the first principal component
ª2016 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY. J. EVOL. BIOL. doi: 10.1111/jeb.12840
JOURNAL OF EVOLUTIONARY BIOLOGY ª2016 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY
2R. J. P. HEATHCOTE ET AL.

(PC) between head length and head width (‘head size’
from hereon) and secondly with the first PC between
SVL and mass (‘body size’ from hereon; see Table S2
for details on the PCs) as response variables with
lineage, sex and their interaction as predictors. In the
models for head size, SVL was also included.
Chemical composition of male scent marks
Femoral secretions are used in scent marking and
argued to play a major role in mate choice in lacertid
lizards (Mart!
ın & L!
opez, 2014a,b). To determine
whether the chemical composition of femoral secretions
differed between the two wall lizard lineages, we
caught 90 lizards (30 males and 15 female lizards of
each lineage) from seven introduced populations of
either Italian or western European origin (see supple-
mentary information). All lizards were captured by
noosing and transferred the same day to the facilities in
Oxford, where they were kept individually in cages
(800 9400 9500 mm (length 9width 9height)). All
cages contained a basking spot, two hides made from
bricks and tiles, and sand as substrate. Lighting and
heat was provided by a 60W spot bulb placed over a
brick on one end of the cage to provide a surface ther-
mal gradient that ranged from approximately 45 °C to
22 °C, an EXO-TERRA
TM
10.0 UVB fluorescent tube and
overhead fluorescent room lighting. All lizards were
provided with food (live crickets and mealworms) and
water ad libitum.
Femoral pore secretions were collected from the
males within 2 days of capture using sterilized forceps
and then stored in sterile glass containers at !20 °C
until chemical analysis (see Heathcote et al. (2014) for
details). The lipophilic compounds in secretions were
analysed by gas chromatography mass spectrometry
(GC-MS) using an Agilent Technologies 7890A gas
chromatograph coupled with an Agilent 5975 mass
spectrometer, after first placing secretions in 150 lL of
pentane. The GC peak areas were integrated using the
Agilent Chemstation software, and the relative com-
pound levels were calculated and then normalized with
Aitchison’s formula (Aitchison, 1986; Dietemann et al.,
2003). We identified compounds by initially comparing
them to the mass spectral library in NIST 2008 and
confirming these by comparison to pure synthetic
compounds. We report identified and unidentified com-
pounds by their retention time and characteristic ion
signature.
We characterized 28 different lipophilic compounds,
of which 15 were shared by all individuals (see
Table S3 for details on chemical compounds). We per-
formed a principal component analysis on the normal-
ized peak area from the 15 shared compounds. The first
seven PCs, which together explained 87% of total
chemical variation, included the main compounds
previously shown to be important in lizard olfactory
communication (Mason & Parker, 2010; see Table S4
for eigenvector loadings). These seven PCs were then
used to determine the overall difference in chemical
signature between the two lineages of males by run-
ning a MANOVA with lineage and population as the two
explanatory variables. After obtaining a significant
effect in the MANOVA s, we carried out further ‘Protected
ANOVAs’ on each principal component as a response vari-
able, with the same predictors as the preceding MANOVA.
Behavioural discrimination and mate choice
Female olfactory discrimination
We next used the same 60 males and 30 females detailed
above in a female discrimination and choice experiment.
Females were captured while gravid with their first
(a)
(c) (d)
(b)
Fig. 1 (a) Representative male lizard of
the ‘Italian’ lineage. (b) Representative
male lizard of the ‘western European’
lineage. (c, d) Example networks from a
single outdoor enclosure showing
courtship interactions (c) and the
parents of sired offspring (d). Line
thickness denotes the number of
interactions (c) or offspring (d) between
two individuals. Letter–number
combination for each node refers to
each lizard’s unique ID within that
enclosure. Italian animals are
represented by green nodes and
western European by brown nodes. Size
of node is proportional to lizard body
size.
ª2016 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY. J. EVOL. BIOL. doi: 10.1111/jeb.12840
JOURNAL OF EVOLUTIONARY BIOLOGY ª2016 EUROPE AN SOCIETY FOR EVOLUTIONARY BIO LOGY
Males drive assortative mating in lizards 3

clutch of the season, which they laid in the laboratory.
Following oviposition, each female underwent three dif-
ferent behavioural trials with the same pair of size-
matched (to within 3 mm) males (one from each lin-
eage) to determine (i) whether females could discrimi-
nate between the different lineages, (ii) whether females
preferred to associate and settle within the territories of
same-lineage males in outdoor enclosures and (iii)
whether courtship and mating were affected by the dif-
ferent male–female lineage combinations. These trials
began three to seven days after a female laid her first
clutch of eggs, which corresponds to the receptive period
under laboratory conditions (R.J.P. Heathcote, G.M.
While & T. Uller, unpublished data).
Tongue-flicking trials. Tongue flicking has been exten-
sively used as a proxy for olfactory discrimination and
mate choice in reptiles, particularly in lacertids (Cooper,
1994; Mart!
ın and L!
opez, 2014a). We therefore quanti-
fied a female’s ability to discriminate between the scent
marks of same- and different-lineage males by counting
the number of ‘tongue flicks’ directed at scented cotton
wool buds (Cooper, 1994; Barbosa et al., 2006). Due to
problems of functionally interpreting differential ton-
gue-flicking rates (Heathcote et al., 2014), we restrict
our interpretation of tongue flicks to evidence for dis-
crimination rather than mate choice. Scent was collected
from males in a standardized manner by rubbing a sterile
cotton bud along the length of the femoral pores of one
leg. These cotton buds were then immediately presented
to females. Each female underwent a separate tongue-
flicking trial for each of the two males at the same time
of day on two successive days. The presentation order of
the two treatments was randomized for each female, and
all trials were undertaken in the female’s own cage,
which was covered with one-way visual film to elimi-
nate observer disturbances. Each male-scented bud was
presented simultaneously 5 cm apart from a clean cotton
bud, which acted as a control. Both buds were attached
to a 30-cm-long wooden stick so they could be rested
under the basking spot without disturbing the female,
and thus removed any potential experimenter effects.
The number of tongue flicks directed to each cotton bud
was counted over a 30-min trial by an observer blind to
the treatment.
Female association preferences. We then tested
whether females were more likely to settle on sites
scent marked by males of their own lineage. Male wall
lizards are territorial, and a female’s proximity to a
male’s scent-marked territory has been used as a stan-
dard means of testing mate choice (e.g. Mart!
ın and
L!
opez, 2014a). Trials were undertaken two days after a
female completed the tongue-flicking trials. All trials
were run in large outdoor enclosures (~7m97 m).
Each enclosure had six basking sites, created from woo-
den pallets and breezeblocks. The basking sites in the
outdoor enclosures were artificially scent-marked using
the bricks, sand and hides obtained from the cage of
the same pair of males whose scent was presented to a
particular female during the tongue-flicking trials. The
spacing between each basking site was approximately
80 cm, which represents the mean distance between
male core territories in our semi-natural contact zone
experiment. Each male’s scent was randomly allocated
to two basking sites, leaving the last two basking sites
as unscented controls (which had sterile sand, bricks
and slates added). The female was introduced into the
enclosure, and her location was recorded every hour
between her first emergence until she retired in the
evening for four days, at which point she was captured
and returned to the laboratory.
Staged mating experiment. Finally, after the two
olfactory trials, the same female–male–male trio from
the tongue-flicking and settlement trials was used in a
staged mating experiment to determine lineage-based
differences in courtship behaviour indicative of beha-
vioural mate choice. All trials were carried out in clean
cages identical in dimensions to those used to house
the females. Each cage had a basking rock and shelter
at both ends and a divider in the middle. The female
and a randomly selected male from the trio were intro-
duced into separate sides of the divider and allowed to
acclimatize for 20 min. The divider was then lifted and
the trial commenced. Trials ran for 60 min per male (or
until a male mated), during which time we recorded
the latency until courtship (seconds), whether the pair
mated and the duration of mating (seconds). Once
mated, the male was removed and the second male
was immediately introduced. The order that males were
introduced was randomized with respect to lineage.
Cages were covered with one-way plastic to minimize
disturbance to the animals during behavioural
observations.
Male olfactory discrimination
Tongue-flicking trials. To determine whether males
can discriminate between females based on olfactory
cues, we captured an additional 90 lizards (30 females
and 15 males from each lineage) from four populations
(see supplementary information) in May 2012 and
brought them into the laboratory. Scent was collected
from females two days after oviposition of the first
clutch by running a cotton bud down the length of a
female’s abdomen and tail base. Each male was pre-
sented with the scent of a female in a similar protocol
to the female tongue-flicking trials. Because males (un-
like females) rapidly habituated to the scented buds
within a trial such that subsequent encounters with the
cotton wool bud after the initial exposure only elicited
very few tongue flicks, we only used the tongue flicks
from the first encounter with the scented cotton buds
ª2016 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY. J. EVOL. BIOL. doi: 10.1111/jeb.12840
JOURNAL OF EVOLUTIONARY BIOLOGY ª2016 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY
4R. J. P. HEATHCOTE ET AL.

within each trial in our analysis. Because females do
not have functional femoral pores and thus do not
actively scent mark territories, we did not run the same
association preference trials for males as detailed above
for females.
Statistical analysis of behavioural discrimination
For both male and female tongue-flicking trials, individ-
uals responded much more strongly towards the scented
buds than the control buds (females: 14.9 "1.7 (mean
"SE) tongue flicks per scented cotton bud vs 3.7 "0.8
for the control; likelihood ratio test: v
2
=51.62,
P<0.001; males: 35.0 "3.5 (mean "SE) tongue flicks
per scented cotton bud vs 8.9 "1.9 for the control; like-
lihood ratio test: v
2
=59.04, P<0.001). To test for dif-
ferences in responses to same-lineage vs. different-
lineage individuals, we therefore analysed the responses
to the scented cotton buds only using generalized linear
mixed models (GLMMs) with Poisson distributions and
log-link functions using R statistical software (R Core
Team, 2013). For both sexes, we included the total
number of tongue flicks directed towards the scented
cotton buds as a response variable, with male–female
lineage combination (i.e. a factor with four levels), trial
order (i.e. 1st vs 2nd), body size of the scented individ-
ual and an interaction between body size of the scented
individual and lineage combination. The focal individ-
ual’s ID was included as a random effect. Female loca-
tions at each pallet were pooled and analysed in two
separate Poisson GLMMs with observations on a treated
pallet as the response. The first included three levels of
treatment (same lineage, different lineage, control),
whereas the second omitted the control treatment to be
able to include male body size as a covariate. Female ID
was included as random effects in both models. For the
staged mating trials, latency to court was log-trans-
formed, and we ran separate mixed models for each of
our response variables (latency to court (LMM), mating
(yes/no) (binomial GLMM) and mating duration
(LMM)). In these models, we included lineage combina-
tion and female body size as predictors.
In all analyses using Poisson and binomial responses
distributions, we tested for, and if necessary, controlled
for overdispersion in our data by including an observa-
tion-level random effect, allowing the scale parameter
to be correctly modelled to validate a Poisson/binomial
distribution (Elston et al., 2001). The significance of fac-
tors in all mixed models was analysed using likelihood
ratio tests on models with and without the factor of
interest (Valdar et al., 2006;
€
Ockinger et al., 2010).
Replicated contact zone experiment
Enclosure and resource treatment set up
We used ten large outdoor enclosures (~7m97 m) to
simulate an artificial secondary contact zone between
lizards from the western European and Italian lineages.
Lizards populating these enclosures were caught in
May 2010 from nine UK populations that, according to
nuclear and mitochondrial genetic data, consist of pure
western European or Italian lineages (see supplemen-
tary information for populations). Each enclosure
housed eight males and eight females. All enclosures
had an equal number of males of both origins. Three
enclosures were stocked with an equal number of
females from both origins but, due to limited sample
size, five enclosures had five females of Italian and
three females of western European origin and one
enclosure was left with a single western European
female and seven Italian females.
Within each enclosure, we constructed nine separate
structures that made suitable basking resources, each
made from wooden pallets (each one 1 m
2
) and six
breezeblocks, providing lizards with areas to forage,
shelter and thermoregulate. To examine the extent to
which these structures may influence wall lizard beha-
viour, we varied their spatial distribution to create our
experimental treatment of resource distribution (‘treat-
ment’ from hereon). In five enclosures, these resources
were spaced equally apart in a 3 93 arrangement,
each pallet approximately 80 cm apart (‘dispersed treat-
ment’), and in the other five enclosures, the 3 93
pallet arrangement was clumped together into the cen-
tre of the enclosures (‘clumped treatment’). All individ-
uals had a tissue sample taken (approx. 3 mm tail tip)
for DNA analysis prior to release into the enclosures.
Lizards were marked with small (~7 mm diameter)
yellow and white tags (made from Tesa
!
tape) on their
dorsal surface, allowing them to be individually identi-
fied in the enclosure. Males were released into the
enclosures on the 8th May 2010 five to seven days ear-
lier than females to enable them to establish territories.
Females were introduced to the enclosures following
oviposition of their first clutch, with all females in a
given enclosure introduced at the same time. This pro-
cedure made sure that all females completed an entire
ovarian cycle in the enclosures (i.e. prereceptive, recep-
tive, post-receptive) and, because sperm storage does
not occur in this species (Pellitteri-Rosa et al., 2012),
only males from the enclosures could father any result-
ing offspring.
Behavioural data collection
Animals were observed daily for 26 days by three of
the authors (RJPH, HEAM and JS). Each observation
period for an enclosure lasted one hour before rotating
to the next enclosure, and observations started when
lizards first emerged in the morning and the last obser-
vation period finished after the last lizard retired to its
shelter in the evening, providing an accumulated obser-
vation period of ~510 h. During observations, we
recorded male and female identities in courtships and
prolonged close male–female associations (‘mate guard-
ing’, which allows males to monopolize mating access
ª2016 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY. J. EVOL. BIOL. doi: 10.1111/jeb.12840
JOURNAL OF EVOLUTIONARY BIOLOGY ª2016 EUROPE AN SOCIETY FOR EVOLUTIONARY BIO LOGY
Males drive assortative mating in lizards 5

to single females (Olsson, 1993; Gal!
an & Price, 2000)),
and male identities in agonistic encounters (win/lose;
defined by when an individual retreated from another
that was showing an agonistic display).
In addition to behavioural observations, we recorded
the locations of all visible lizards every hour through-
out their activity period each day. This location data
allowed us to calculate the core home range size in m
2
for each lizard using Ranges v8 (Anatrack Ltd, Ware-
ham, UK), with ‘core home ranges’ determined by cal-
culating the 50% kernel cores for each lizard, using a
least squares cross-validation smoothing parameter
(Powell, 2000). The 50% kernel was chosen as it
encompassed the majority of an animal’s social interac-
tions and delimited the onset of the plateau between
the kernel percentage isopleth and the home range
size (m
2
) correlation, indicating the lizards were spend-
ing a high concentration of time in this area. Using
the 50% kernel cores, we were able to determine the
number of overlapping core home ranges for each
individual lizard. We used this to determine whether
the spatial distribution of basking sites influenced the
density of lizards and whether this in turn influenced
patterns of courtship and reproductive assortativity.
We created four models, with males and females being
analysed separately. The response variables analysed
separately for each sex were core home range size in
m
2
and total number of overlapping individuals. Core
home range showed a left-biased skew and so was
logged and used in LMMs, whereas the overlap data
were analysed with Poisson GLMMs. For all four mod-
els, we included lineage, treatment and their interac-
tion as predictors, and in the overlap models, we
additionally included core home range size to control
for this covariate. All models included enclosure as a
random effect.
After the observational period, all lizards were cap-
tured and returned to the indoor facilities in Oxford.
Females were housed individually in cages as detailed
above. Eggs laid by females were incubated at 24 °C
with a 5:1 vermiculite: water volume ratio. Upon
hatching, tissue samples (tail tips) were taken of all
juveniles to use in the genetic analysis.
Molecular analysis and paternity assignment
All adults and their resulting offspring from the
enclosure experiment were genotyped at nine highly
variable microsatellite loci that amplify across both sub-
species as described in Heathcote et al. (2015). Paternity
assignment was conducted in CERVUS v3.2 (Kali-
nowski et al., 2007) using 10 000 simulations, with the
mother’s allele conformation being set as the known
parent and all eight males she shared her enclosure
with as the potential fathers. Six females nested in the
enclosures before removal and these clutches were
retrieved and assigned full parentage using CERVUS. A
total of 19 females did not produce a clutch following
return to the laboratory. Paternity of the 296 juveniles
from 61 clutches was confirmed at >99% confidence.
Social network and behavioural analyses
For each enclosure, we created ‘home range overlap’,
‘courtship’ and ‘genetic’ networks. Network ‘edges’ [i.e.
interactions or associations between individuals (Croft
et al., 2008)] for the home range data were created
between any two individuals that had overlapping core
home ranges, in the courtship network between indi-
viduals observed courting and in the genetic networks
between the parents of each offspring that was assigned
parentage. We calculated the lineage-based assortativity
for the three networks within each enclosure using the
R statistical package ‘assortnet’, which allows the calcu-
lation of Newman’s assortativity coefficients using
weighted networks (R Core Team, 2013; Farine, 2014).
To determine the statistical significance of assortativity,
we compared the observed coefficient values within
each network to a distribution of 10 000 randomly per-
muted matrices. To also test whether lineage-based
assortativity differed between treatments, we calculated
the t-statistic obtained by comparing the assortativity
coefficient between the clumped and dispersed experi-
mental treatments, and compared this to a random dis-
tribution of t-statistics generated from 10 000 permuted
matrices.
Using Mantel tests, we tested whether the courtship
and home range overlap networks correlated with the
genetic networks for each enclosure, which would indi-
cate that precopulatory male–female behaviours predict
resulting paternity patterns [which may not be the case
due to post-copulatory processes (Westneat, 1987;
Olsson & Madsen, 1998)]. Mantel t-statistics for the
correlation between networks for each enclosure were
calculated and then compared to 10 000 permuted
matrices to obtain our Pvalue. To determine whether
the enclosures differed in the degree to which the
courtship networks predicted the genetic networks
between the two treatments, we also compared the
Mantel t-statistics obtained for each enclosure between
the treatments using a t-test and compared this statistic
to 10 000 permuted matrices to obtain our Pvalue. We
assessed the overall support for lineage-based assortativ-
ity and correlations between the different types of
network by combining the results from each of the 10
enclosures using Fisher’s omnibus test (Haccou &
Meelis, 1992).
For the 58 males observed engaging in agonistic
interactions in the enclosure experiment, we calculated
their dominance using the David’s score in SOCPROG
analytical software (Whitehead, 2009). This was then
used as a response variable in a LMM with male lin-
eage, body size and their interaction as factors and
enclosure as a random effect. Because dominance
scores are nonindependent data, and thus violate the
assumptions of a LMM, we calculated Pvalues from
ª2016 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY. J. EVOL. BIOL. doi: 10.1111/jeb.12840
JOURNAL OF EVOLUTIONARY BIOLOGY ª2016 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY
6R. J. P. HEATHCOTE ET AL.

our models by comparing the model coefficients against
those of randomized data sets obtained through a quad-
ratic assignment procedure (where dominance scores
were shuffled between individuals within an enclosure)
with 10 000 permutations in R.
We determined the difference between the lineages
in their reproductive success by constructing two Pois-
son GLMMs with total number of females courted and
total number of offspring sired as separate response
variables. In the courtship and sired offspring models,
we included male lineage, treatment and their interac-
tion as predictors. Enclosure was included as a random
effect to control for the unbalanced number of females
of different lineages in the different enclosures. Number
of mate guardings were analysed in a Poisson GLMM
with male ID and enclosure as random effects, and with
male dominance score, the body size of the female he
was observed guarding and lineage combination (a fac-
tor with four levels) as the predictors. To determine
whether lineage-based differences in female traits might
explain the reproductive patterns in the enclosure
experiment, we analysed the number of courtships a
female received in a Poisson GLMM, with lineage,
treatment and body size as the predictors and enclosure
as a random effect.
Predictors of hybridization
In addition to the network analyses, we also performed
analyses of individual predictors of the number of
hybrid offspring produced separately for each lineage
and sex. For females, we only included animals that
had at least one offspring assigned. We analysed ani-
mals of Italian and western European origin separately
for two reasons. Firstly, because there were more Ital-
ian than western European clutches and the numbers
of each lineage differed across enclosures, the strong
assortativity would likely create spurious results for lin-
eage-dependent hybridization in males. Secondly, the
low incidence of hybridization for Italian females makes
individual predictors of hybridization largely uninfor-
mative for Italian females and western European males.
Broadly speaking, in the absence of male and female
preference for same-lineage mates, large dominant
males with high reproductive success with same-lineage
females should also produce more offspring with differ-
ent-lineage females. On the contrary, if males prefer
same-lineage females, smaller and less dominant males
should produce more offspring with females from dif-
ferent lineages due to being excluded from their pre-
ferred mates. In male models, the number of hybrid
offspring produced was therefore analysed in a Poisson
GLMM, with the number of same-lineage offspring,
dominance, core home range size, body size and treat-
ment as fixed factors and enclosure as a random effect.
For females, the number of hybrid offspring produced is
limited by her clutch size. Female models were there-
fore analysed as a binomial GLMM (total number of
hybrid offspring over total number of offspring). All else
being equal, larger females that receive more courtships
(i.e. generally preferred females) should have a higher
proportion of parentage with males of the dominant
lineages. In addition, females with larger home ranges
might hybridize with males from a less dominant
lineage due to higher encounter rates, in particular
with subdominant males. Our predictors for the female
models were therefore body size, core home range size,
number of courtships received and treatment, with
enclosure as a random effect.
First-generation contact in free-ranging lizards
As an additional test of the degree of parentage assorta-
tivity, we analysed paternity from a population of
lizards of mixed origin released as hatchlings in a dis-
used quarry on the Dorset coast. The primary aim of
the release was to investigate survival (see While et al.,
2015b), but here we analyse parentage data for eleven
clutches of first-time breeders. Briefly, hatchlings of
both origins were released close to existing lizard popu-
lations and recaptured in their second year after they
had reached maturity. We recaptured 18 males (14 Ital-
ian, four western European) and 24 females (13 Italian,
11 western European), of which eleven females carried
eggs at the time of capture. The females were brought
to the animal facilities in Oxford where their eggs and
offspring were processed as above. We analysed pater-
nity by including all possible fathers as described above;
all but one offspring could be assigned paternity with
>99% confidence.
Results
Phenotypic differences between lineages
Morphometrics
Italian lizards were larger overall than western Euro-
pean lizards (lineage: F
1,157
=6.23, P=0.014; sex:
F
1,157
=1.02, P=0.313; lineage9sex: F
1,156
=0.04,
P=0.841) with larger and more sexually dimorphic
heads (lineage: F
1,156
=31.23, P<0.001; sex: F
1,156
=
787.19, P<0.001; SVL: F
1,155
=194.22, P<0.001; lin-
eage9sex: F
1,155
=6.55, P=0.012).
Chemical composition of male scent marks
Western European and Italian males differed signifi-
cantly in the overall shared chemical composition of
their femoral secretions (MANOVA; lineage: F
7,44
=22.75,
P<0.001; Population: F
35,240
=1.05, P=0.005).
Specifically, the two lineages differed significantly in
values for PC1, which had strong positive loadings with
cholesterol, hexadecanal, heptadecene and octode-
canoic acid, and strong negative loadings with squa-
lene, cholecalciferol (provitamin D) and campesterol
(see Tables S3 and S4 for full chemical analysis results
ª2016 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY. J. EVOL. BIOL. doi: 10.1111/jeb.12840
JOURNAL OF EVOLUTIONARY BIOLOGY ª2016 EUROPE AN SOCIETY FOR EVOLUTIONARY BIO LOGY
Males drive assortative mating in lizards 7

and Table S5 for summary statistics). In addition, the
femoral secretions from 28 of 29 Italian males con-
tained high levels of a-tocopherol (vitamin E), compris-
ing 11.62 "1.12% (mean "SE) of their total chemical
signature. In contrast, only 3 of 28 western European
male secretions contained a-tocopherol, comprising
7.23 "1.93% (mean "SE) of the chemical composi-
tion from these three individuals.
Behavioural discrimination and mate choice
Female olfactory discrimination
Tongue-flicking trials. Despite the pronounced differ-
ences in scent mark composition, females did not
differ in the number of tongue flicks directed towards
same- and different-lineage femoral secretions
(Table S6).
Female association preferences. In the settlement tri-
als, females preferred to settle on male-scented pallets
over the controls (v
2
=13.30, P=0.020), but showed
no preference for visiting a pallet scent marked by
either same- or different-lineage males or for males of
different body sizes (same vs different lineage:
v
2
=0.17, P=0.676; male body size: v
2
=3.53,
P=0.061).
Staged mating experiment. There was no influence of
lineage or female body size on the latency to court (lin-
eage combination: v
2
=6.53, P=0.088; female body
size: v
2
=0.29, P=0.593). Lineage significantly pre-
dicted copulation likelihood, with Italian males being
more likely than western European males to copulate
with Italian females (lineage combination: v
2
=8.98,
P=0.029; female body size: v
2
<0.01, P=0.986; see
Table S7 for lineage-specific contrasts). For pairs that
copulated, western European males copulated for longer
with western European females than with Italian females
(45.8"1.7 vs 36.4"2.2; mean"SE seconds), and Italian
males copulated for longer with western European
females than with Italian females (39.2"2.8 vs 33.5"2.5;
mean"SE seconds) (lineage combination: v
2
=13.23,
P=0.004; female body size: v
2
=0.21, P=0.649; see
Table S8 for lineage-specific contrasts).
Male olfactory discrimination
In contrast to females, the number of tongue flicks by
males directed towards female scent was significantly
affected by the interaction between male–female
lineage combination and female body size. Whereas
Italian males tongue flicked less for larger females of
both lineages, western European males showed the
reverse response, and tongue flicked more overall when
presented with scent from females of their own lineage
(Table S6; Fig. S1).
Replicated contact zone experiment
Male core home range size showed a significant inter-
action between lineage and habitat structure, with
males of western European origin having substantially
larger home ranges in the dispersed habitat (Fig. S2).
There were no significant predictors of female home
range size. The experimental habitat manipulation
caused differing densities of lizards, as evidenced from a
greater home range overlap between lizards in the
clumped habitat when controlling for core home range
size (see Table S9 for all home range results).
Italian males were significantly dominant over west-
ern European males during territorial disputes (David’s
scores: Italian: 1.06 "0.41; western European: !1.26 "
0.45 (mean "SE) (lineage: P=0.006; body size:
P=0.255; lineage9body size: P=0.086). Of 58 interac-
tions observed between Italian and western European
males, we only observed five where the western Euro-
pean animal won (8.6%). Italian males also courted
more females and sired a greater number of offspring
than western European males, but there was no effect
of experimental treatment or an interaction between
lineage and treatment on either courtship or paternity
(females courted: treatment: v
2
=0.20, P=0.658;
lineage: v
2
=9.74, P=0.002; treatment9lineage:
v
2
=0.28, P=0.596; number of offspring: treatment:
v
2
=0.58, P=0.445; lineage: v
2
=8.33, P=0.004;
treatment9lineage: v
2
=0.51, P=0.477). Larger
females received more courtships than smaller females,
but there was no statistically significant difference
between lineages or treatment (treatment: v
2
=0.10,
P=0.757; lineage: v
2
=2.07, P=0.150; body size:
v
2
=4.91, P=0.027; Fig. S3). Mate guarding was pre-
dicted by high male dominance (v
2
=39.51, P<0.001),
large female body size (v
2
=4.73, P=0.029), but not
lineage combination (v
2
=0.54, P=0.911).
Social network analyses
Courtship networks were significantly assortative with
respect to lineage (v
2
=45.96, d.f. =20, P<0.001;
Fig. 1). Furthermore, courtships were more strongly
assortative towards the same lineage in the clumped
compared to the dispersed enclosures (P=0.044).
Courtship networks correlated strongly with the genetic
networks created from the paternity assignment
(v
2
=55.66, d.f. =20, P<0.001), showing that
observed courtship behaviours were important in deter-
mining resulting paternity. As such, genetic networks
were also highly assortative by lineage (v
2
=79.59,
d.f. =20, P<0.001; Fig. 1). Indeed, 83% (244 of 294)
of overall offspring were sired by males from the same
lineage. Assortativity in paternity networks did not dif-
fer between treatments (t =0.69, P=0.290). There was
a borderline-significant effect of courtship observations
being better predictors of paternity in the clumped treat-
ment enclosures compared to the dispersed (t =1.49;
ª2016 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY. J. EVOL. BIOL. doi: 10.1111/jeb.12840
JOURNAL OF EVOLUTIONARY BIOLOGY ª2016 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY
8R. J. P. HEATHCOTE ET AL.

P=0.058). Networks based on home range overlap
were not assortative by lineage (v
2
=16.41, d.f. =20,
P=0.681) and did not correlate with genetic networks
(v
2
=15.78, d.f. =20, P=0.810).
Predictors of hybridization
We were unable to identify individual male predictors
of the number of hybrid offspring produced, including
a lack of evidence that males that sired more offspring
with same-lineage females sired more offspring with
different-lineage females (Table 1). In the western
European lineage, larger females (those that received
more courtships) produced a higher number of hybrid
offspring, whereas this was not the case in Italian
females (Table 2).
First-generation contact in free-ranging lizards
The small sample size and low number of western
European males preclude statistical inference, but
parentage was highly assortative. Across the eleven
clutches, all offspring in three of four clutches of west-
ern European females were sired by western European
males (i.e. 73% of western European offspring were
from assortative matings), whereas all but one offspring
in one of seven clutches of Italian females were sired
by an Italian male (i.e. 97% of Italian offspring were
from assortative matings).
Discussion
The evolution of reproductive isolation is an important
step in the speciation process. Although many authors
suggest that phenotypic divergence during allopatry
facilitates assortative mating through female choice
(Wirtz, 1999; Price, 2008; Grant & Grant, 2009; Abbott
et al., 2013), very few studies have attempted to iden-
tify the mechanisms underlying assortative mating in
the initial stages of secondary contact, before reinforce-
ment has been able to occur.
The two lineages of wall lizard we discuss here have
been isolated for ~2 myr (Gassert et al., 2013) and have
subsequently diverged in their native range in many
traits known to be important in male–male competition
(body and head size, coloration, chemical signals, bite
force), female choice (male body size, chemical signals)
and male choice (female body size) in lizards (Tokarz,
1985; Olsson, 1993, 1994; Anderholm et al., 2004; Uller
et al., 2010; Mart!
ın and L!
opez, 2014a,b; While et al.,
2015a). The non-native populations used here retain
these differences. As predicted from this period of isola-
tion, paternity was highly assortative in our replicated,
semi-natural contact zone populations, with 83% of
offspring being produced by same-lineage parents,
results that were replicated in a mixed origin free-ran-
ging population. Below we discuss the relative contri-
bution played by male and female behavioural traits in
mediating these patterns.
Heterospecific mating avoidance is typically expected
to be driven by females due to their greater per capita
reproductive investment, which should lead to selection
for more intense mate discrimination (Wirtz, 1999).
Studies of natural hybrid zones have indeed shown that
female choice is a major cause of reproductive isolation
(e.g. Stein & Uy, 2006; Culumber et al., 2014). How-
ever, despite extensive phenotypic differences between
the wall lizard lineages, including the chemical compo-
sition of their scent marks (the principal modality of
female choice in lacertids, e.g. Mart!
ın & L!
opez, 2014b),
we failed to find any evidence that female choice con-
tributes to the maintenance of reproductive isolation.
Females did not appear to discriminate between males
based on olfactory cues; they did not prefer to settle in
the territories of same-lineage males based on these
olfactory cues; female home range overlap did not pre-
dict male paternity (as would be expected if females
preferred to associate spatially with same-lineage
males); and they were equally likely to accept males of
Table 1 Summary statistics of Poisson GLMM testing for
phenotypic predictors of the number of hybrid offspring produced
by males from the enclosure experiment. Statistics for
nonsignificant results are included at the point of their deletion
from the model.
Model Factor v
2
P
Italian male Treatment 0.06 0.814
Core home range 0.23 0.632
Body size 0.18 0.668
Conspecific offspring 0.45 0.503
Dominance 2.62 0.106
Western European male Dominance 0.19 0.872
Conspecific offspring 0.20 0.656
Body size 0.20 0.651
Core home range 1.54 0.215
Treatment 0.09 0.759
Table 2 Summary statistics of binomial GLMMs testing for
phenotypic predictors of hybridization in females from the
enclosure experiment. Significant factors are highlighted in bold.
Statistics for nonsignificant results are included at the point of
their deletion from the model.
Model Factor v
2
P
Italian female Body size 0.02 0.882
Received courtships 0.03 0.857
Core home range size 1.09 0.297
Treatment 0.56 0.455
Western European female Treatment 2.52 0.113
Core home range size 1.69 0.193
Received courtships 6.68 0.010
Body size 4.81 0.028
ª2016 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY. J. EVOL. BIOL. doi: 10.1111/jeb.12840
JOURNAL OF EVOLUTIONARY BIOLOGY ª2016 EUROPE AN SOCIETY FOR EVOLUTIONARY BIO LOGY
Males drive assortative mating in lizards 9

either lineage as mates in staged trials. Furthermore,
males always initiated courtships in our replicated sec-
ondary contact zone experiment, the patterns of which
strongly predicted resulting parentage. Taken together,
this lack of evidence for female discrimination is consis-
tent with the consensus view that precopulatory female
choice is very limited in lizards (Olsson, 1993; Olsson &
Madsen, 1995, 1998; Tokarz, 1995; Font et al., 2012).
In the absence of female mate choice, barriers to
hybridization should be weak as males are expected to
show relatively low degrees of mate rejection. This
should be particularly true in promiscuous species, such
as the wall lizard, where males do not provide parental
care and male per capita mating costs are assumed to be
relatively low (Wirtz, 1999; Servedio, 2007). Despite
this, levels of hybridization were low in both our semi-
natural contact zones and first-generation free-ranging
lizards. Our data from both naturalistic observations
and laboratory trials suggest that this reproductive
assortativity arises primarily from male preference for
females of the same lineage. Male wall lizards of both
lineages could distinguish between, and subsequently
preferentially courted, same-lineage females, and male
courtship patterns strongly predicted subsequent pater-
nity. Although we cannot be certain of the specific
female characters males use to assess potential mates,
our data suggest some component of this is size-depen-
dent since male courtship attention and olfactory dis-
crimination, as well as female hybridization patterns,
were predicted by female size. Because strong assorta-
tivity was also shown by the lizards that matured in
our mixed origin natural population, such mate prefer-
ences are unlikely to have been learnt (Verzijden et al.,
2012), which otherwise could have contributed to the
results from our enclosure experiments.
May other mechanisms have contributed to the
assortative mating patterns we describe? Male Italian
lizards are larger in size and exhibit greater exaggera-
tion of intrasexually selected secondary characters. As
a result, Italian males were competitively dominant
over their western European counterparts during terri-
torial disputes in the enclosures and hybridization pri-
marily occurred between Italian males and western
European females (see also While et al., 2015a).
Because large females were preferred by males of both
lineages, this competitive asymmetry likely explains
why Italian males were able to direct their courtship
attention towards and monopolize matings with large
females. This was particularly apparent in the mate
guarding observations, where dominant (mostly Ital-
ian) individuals guarded the largest females. Such
preference for larger and more fecund females is a
common observation in lizards (Olsson, 1993). As Ital-
ian females are generally larger than western Euro-
pean females, this mechanism should contribute to
reproductive assortativity even in the absence of other
lineage-specific mate preferences that restrict mating
between individuals of different lineages. In support of
this suggestion, larger western European females pro-
duced more hybrid offspring than smaller western
European females, whereas there was no effect of
female body size on incidence of hybridization in Ital-
ian females. As heterospecific mating avoidance may
often only evolve in response to selection associated
with costly hybrid matings (Brodsky et al., 1988;
Wirtz, 1999; Tynkkynen et al., 2004), hybridization
can be so rapid during initial secondary contact that
lineages can become completely introgressed before
reinforcement can occur (Echelle & Connor, 1989;
Rhymer & Simberloff, 1996; Huxel, 1999; Mu~
noz-
Fuentes et al., 2007). Correlated evolution of male and
female body size may therefore be an important factor
that reduces gene flow in the early stages of secondary
contact before strong conspecific mate choice has been
able to evolve. For example, size-mediated hybridiza-
tion has previously been shown in sticklebacks, where
hybridization between two species that differ in overall
body is most common between individuals in the
overlapping size range (Nagel & Schluter, 1998; Conte
& Schluter, 2013). However, in sticklebacks this
hybridization is the consequence of females preferring
similar-sized mates rather than through asymmetric
male–male competition.
Our second aim was to determine whether small-
scale differences in the clustering of suitable habitat
influence behavioural processes that promote repro-
ductive isolation. Such effects could partially explain
the geographical variation in introgression frequently
found in ‘mosaic hybrid zones’ (Barton & Hewitt,
1985; Sperling & Spence, 1991; Senn & Pemberton,
2009). In our experiment, courtships were more assor-
tative when the habitat was highly clustered and
home range overlap was subsequently high. This sug-
gests that high population density facilitates the ability
of males to choose their preferred females. Corroborat-
ing this, we found a borderline-significant effect of
courtship networks being more predictive of parentage
networks when resources were clumped. Considering
our statistical power was low, this result suggests the
possibility that precopulatory processes such as mate
guarding and male monopolization may be more
important in determining subsequent paternity when
population densities are high. Despite this, we found
no corresponding effect of habitat structure on assorta-
tivity in parentage nor did levels of hybridization differ
between treatments. However, statistical power is lim-
ited in the network analyses and we may simply be
unable to detect more modest effects. The contribution
of post-copulatory isolation in this system and other
secondary contact zones is worthy of further study
but, nonetheless, the strong correlation between the
courtship and genetic networks suggests that the
degree of reproductive isolation documented here
largely results from precopulatory processes.
ª2016 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY. J. EVOL. BIOL. doi: 10.1111/jeb.12840
JOURNAL OF EVOLUTIONARY BIOLOGY ª2016 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY
10 R. J. P. HEATHCOTE ET AL.

In summary, and in contrast to most studies that
emphasize the role of female choice in hybridization
(e.g. Wirtz, 1999; Servedio, 2007), our results suggest
that sexual selection acting on traits involved in male–
male competition and male choice for female quantita-
tive traits can promote assortative mating during initial
secondary contact, as previously hypothesized for
lizards (Barbosa et al., 2006; Font et al., 2012). Fur-
thermore, the reproductive skew and monopolization
of larger females by males of the dominant lineage in
the enclosure experiment suggests that divergence in
traits involved in male–male competition and corre-
lated changes in female body size may contribute to
assortative mating patterns during secondary contact.
In this particular system, however, it is important to
note that the degree of assortativity we report here is
still insufficient to prevent introgression (While et al.,
2015a). Whether or not stronger degrees of mate pref-
erence have evolved in naturally occurring zones of
secondary contact in this species warrants future
investigation.
Acknowledgments
We would like to thank Fiona Moultrie and James
Stroud for help collecting lizards and helpful discus-
sions, and Hal Whitehead and Dan Franks for advice
on the social network analyses. We thank Guillem
Perez i de Lanuza for the photograph of the male
French lizard in Fig. 1. This research was funded by the
British Ecological Society, the Royal Society of London,
NERC-NBAF (grant code: NBAF549) and the National
Geographic Society (all to TU); a BBSRC Studentship,
Pembroke College Research Grant (University of
Oxford) and Pembroke College Dean of Graduates
Award (to RJPH), and an FP7 Marie Curie Fellowship
(to GMW). TU is supported by the Royal Society of
London and the Knut and Alice Wallenberg Founda-
tions. All experiments carried out as part of this
research comply with UK laws and the work was
approved by Natural England (NNR 2009/002 and NNR
2012/0016 with extensions) and the University of
Oxford’s Local Ethical Review Process and the UK
Home Office (PPL: 30/2560).
Conflict of interest
The authors declare no conflict of interest.
References
Abbott, R., Albach, D., Ansell, S., Arntzen, J.W., Baird, S.J.E.,
Bierne, N. et al. 2013. Hybridization and speciation. J. Evol.
Biol. 26: 229–246.
Aitchison, J. 1986. The Statistical Analysis of Compositional Data.
Chapman and Hall Ltd, London.
Anderholm, S., Olsson, M., Wapstra, E. & Ryberg, K. 2004. Fit
and fat from enlarged badges: a field experiment on male
sand lizards. Proc. R. Soc. Lond. B 271: S142–S144.
Barbosa, D., Font, E., Desfilis, E. & Carretero, M.A. 2006.
Chemically mediated species recognition in closely related
Podarcis wall lizards. J. Chem. Ecol. 32: 1587–1598.
Barton, N.H. & Hewitt, G.M. 1985. Analysis of hybrid zones.
Annu. Rev. Ecol. Syst. 16: 113–148.
Blackburn, T.M., Lockwood, J.L. & Cassey, P. 2009. Avian Inva-
sions: The Ecology and Evolution of Exotic Birds. Oxford Univer-
sity Press, Oxford.
Bohme, W. 1986. Handbuch der Reptilien und Amphibien
Europas (Band 2/II Echsen (Sauria) III, Aula-Verlag Wies-
baden.
Boughman, J.W. 2001. Divergent sexual selection enhances
reproductive isolation in sticklebacks. Nature 411: 944–948.
Bra~
na, F. 1996. Sexual dimorphism in Lacertid lizards: male
head increase vs female abdomen increase? Oikos 75: 511–
523.
Brodsky, L.M., Ankney, C.D. & Dennis, D.G. 1988. The influ-
ence of male dominance on social interactions in black
ducks and mallards. Anim. Behav. 36: 1371–1378.
Bronson, C.L., Grubb, T.C. Jr, Sattler, G.D. & Braun, M.J.
2003. Mate preference: a possible causal mechanism for a
moving hybrid zone. Anim. Behav. 65: 489–500.
Conte, G.L. & Schluter, D. 2013. Experimental confirmation
that body size determines mate preference via phenotype
matching in a stickleback species pair. Evolution 67: 1477–
1484.
Cooper, W. 1994. Chemical discrimination by tongue-flicking
in lizards: A review with hypotheses on its origin and its
ecological and phylogenetic relationships. J. Chem. Ecol. 20:
439–487.
Coyne, J.A. & Orr, H.A. 2004. Speciation. Sinauer Associates,
Sunderland, MA.
Croft, D.P., James, R. & Krause, J. 2008. Exploring Animal Social
Networks. Princeton University Press, Princeton.
Culumber, Z.W., Ochoa, O.M. & Rosenthal, G.G. 2014. Assor-
tative mating and the maintenance of population structure
in a natural hybrid zone. Am. Nat. 184: 225–232.
Dietemann, V., Peeters, C., Liebig, J., Thivet, V. & Holldobler,
B. 2003. Cuticular hydrocarbons mediate discrimination of
reproductives and nonreproductives in the ant Myrmecia
gulosa.Proc. Nat. Acad. Sci. 100: 10341–10346.
Doorn, G.S.V., Noest, A.J. & Hogeweg, P. 1998. Sympatric spe-
ciation and extinction driven by environment dependent
sexual selection. Proc. R. Soc. Lond. B 265: 1915–1919.
Echelle, A.A. & Connor, P.J. 1989. Rapid, geographically
extensive genetic introgression after secondary contact
between two pupfish species (Cyprinodon, Cyprinodontidae).
Evolution 43: 717–727.
Edward, D.A. & Chapman, T. 2011. The evolution and signifi-
cance of male mate choice. Trends Ecol. Evol. 26: 647–654.
Elston, D., Moss, R., Boulinier, T., Arrowsmith, C. & Lambin,
X. 2001. Analysis of aggregation, a worked example: num-
bers of ticks on red grouse chicks. Parasitology 122: 563–569.
Farine, D.R. 2014. Measuring phenotypic assortment in animal
social networks: weighted associations are more robust than
binary edges. Anim. Behav. 89: 141–153.
Font, E., Barbosa, D., Sampedro, C. & Carazo, P. 2012. Social
behavior, chemical communication, and adult neurogenesis:
ª2016 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY. J. EVOL. BIOL. doi: 10.1111/jeb.12840
JOURNAL OF EVOLUTIONARY BIOLOGY ª2016 EUROPE AN SOCIETY FOR EVOLUTIONARY BIO LOGY
Males drive assortative mating in lizards 11

Studies of scent mark function in Podarcis wall lizards. Gen.
Comp. Endocrinol. 177:9–17.
Gal!
an, P. & Price, A. 2000. Females that imitate males: dorsal
coloration varies with reproductive stage in female Podarcis
bocagei (Lacertidae). Copeia 2000: 819–825.
Gassert, F., Schulte, U., Husemann, M., Ulrich, W., R€
odder, D.,
Hochkirch, A. et al. 2013. From southern refugia to the
northern range margin: genetic population structure of the
common wall lizard, Podarcis muralis.J. Biogeogr. 40: 1475–
1489.
Grant, P.R. & Grant, B.R. 2009. The secondary contact phase
of allopatric speciation in Darwin’s finches. PNAS 106:
20141–20148.
Haccou, P. & Meelis, E. 1992. Statistical Analysis of Behavioural
Data: An Approach Based on Time-Structured Models. Oxford
University Press, Oxford.
Haldane, J.B.S. 1922. Sex ratio and unisexual sterility in
hybrid animals. J. Genet. 12: 101–109.
Heathcote, R.J.P., Bell, E., D’Ettorre, P., While, G.M. & Uller,
T. 2014. The scent of sun worship: basking experience alters
scent mark composition in male lizards. Behav. Ecol. Sociobiol.
68: 861–870.
Heathcote, R.J.P., Dawson, D.A. & Uller, T. 2015. Characterisa-
tion of nine European wall lizard (Podarcis muralis)
microsatellite loci of utility across sub-species. Conserv. Genet.
Res. 7: 85–87.
Hewitt, G., Butlin, R. & East, T. 1987. Testicular dysfunction in
hybrids between parapatric subspecies of the grasshopper
Chorthippus parallelus.Biol. J. Linnean. Soc. 31: 25–34.
Hubbs, C.L. 1955. Hybridization between fish species in nat-
ure. Syst. Zool. 4:1–20.
Huxel, G.R. 1999. Rapid displacement of native species by inva-
sive species: effects of hybridization. Biol. Cons. 89: 143–152.
Kalinowski, S.T., Taper, M.L. & Marshall, T.C. 2007. Revising
how the computer program CERVUS accommodates geno-
typing error increases success in paternity assignment. Mol.
Ecol. 16: 1099–1106.
Klingenberg, C., Spence, J. & Mirth, C. 2000. Introgressive
hybridization between two species of waterstriders (Hemi-
ptera: Gerridae: Limnoporus): geographical structure and tem-
poral change of a hybrid zone. J. Evol. Biol. 13: 756–765.
Lande, R. 1981. Models of speciation by sexual selection on
polygenic traits. PNAS 78: 3721–3725.
Mart!
ın, J. & L!
opez, P. 2014a. Condition-dependent chemosignals
in reproductive behavior of lizards. Horm. Behav. 68: 14–24.
Mart
!
ın, J. & L!
opez, P. 2014b. Pheromones and chemical com-
munication in lizards. In: Reproductive Biology and Phylogeny of
Lizards and Tuatara (J.L. Rheubert, D.S. Siegel & S.E. Trauth,
eds), pp. 43–77. Taylor & Francis, Miami.
Mason, R. & Parker, M.R. 2010. Social behavior and pheromo-
nal communication in reptiles. J. Comp. Physiol. A 196: 729–
749.
Mayr, E. 1963. Animal Species and Evolution. Harvard University
Press, Massachusetts.
McCarthy, E.M. 2006. Handbook of Avian Hybrids of the World.
Oxford University Press, Oxford.
Michaelides, S., While, G.M., Bell, C. & Uller, T. 2013. Human
introductions create opportunities for intra-specific
hybridization in an alien lizard. Biol. Invasions 15: 1101–
1112.
Michaelides, S., While, G.M., Zajac, N. & Uller, T. 2015. Wide-
spread primary, but geographically restricted secondary,
human introductions of wall lizards, Podarcis muralis.Mol.
Ecol. 24: 2702–2714.
Mu~
noz-Fuentes, V., Vil#
a, C., Green, A.J., Negro, J.J. & Soren-
son, M.D. 2007. Hybridization between white-headed ducks
and introduced ruddy ducks in Spain. Mol. Ecol. 16: 629–638.
Nagel, L. & Schluter, D. 1998. Body size, natural selection, and
speciation in sticklebacks. Evolution 52: 209–218.
€
Ockinger, E., Schweiger, O., Crist, T.O., Debinski, D.M.,
Krauss, J., Kuussaari, M. et al. 2010. Life-history traits
predict species responses to habitat area and isolation: a
cross-continental synthesis. Ecol. Lett. 13: 969–979.
Olsson, M. 1993. Male preference for large females and assor-
tative mating for body size in the sand lizard (Lacerta agilis).
Behav. Ecol. Sociobiol. 32: 337–341.
Olsson, M. 1994. Nuptial coloration in the sand lizard, Lacerta
agilis: an intra-sexually selected cue to fighting ability. Anim.
Behav. 48: 607–613.
Olsson, M. & Madsen, T. 1995a. Female choice on male quan-
titative traits in lizards: why is it so rare? Behav. Ecol.
Sociobiol. 36: 179–184.
Olsson, M. & Madsen, T. 1998. Sexual selection and sperm
competition in reptiles. In: Sperm Competition and Sexual Selec-
tion (T.R. Birkhead & A.P. Møller, eds), pp. 503–577.
Academic Press, London.
Panhuis, T.M., Butlin, R., Zuk, M. & Tregenza, T. 2001. Sexual
selection and speciation. Trends Ecol. Evol. 16: 364–371.
Pearson, S.F. 2000. Behavioral asymmetries in a moving
hybrid zone. Behav. Ecol. 11: 84–92.
Pellitteri-Rosa, D., Sacchi, R., Pupin, F., Bellati, A., Cocca, W.,
Gentilli, A. et al. 2012. Testing the ability to store sperm: an
experimental manipulation of mating opportunities in the
common wall lizard, Podarcis muralis.Acta Herpetol. 7: 111–118.
Peterson, M.A., Honchak, B.M., Locke, S.E., Beeman, T.E.,
Mendoza, J., Green, J. et al. 2005. Relative abundance and
the species specific reinforcement of male mating preference
in the Chrysochus (Coleoptera: Chrysomelidae) hybrid zone.
Evolution 59: 2639–2655.
Powell, R.A. 2000. Animal home ranges and territories and
home range estimators. In: Research Techniques in Animal Ecol-
ogy: Controversies and Consequences (L. Boitani & T.K. Fuller,
eds), Columbia University Press, New York.
Price, T. 2008. Speciation in Birds. Roberts and Company,
Greenwood Village.
R Core Team 2013. R: A Language and Environment for Statistical
Computing. R Foundation for Statistical Computing, Vienna,
Austria. http://www.R-project.org/.
Randler, C. 2002. Avian hybridization, mixed pairing and
female choice. Anim. Behav. 63: 103–119.
Rhymer, J.M. & Simberloff, D. 1996. Extinction by hybridiza-
tion and introgression. Annu. Rev. Ecol. Syst. 27: 83–109.
Sax, D.F., Stachowicz, J.J., Brown, J.H., Bruno, J.F., Dawson,
M.N., Gaines, S.D. et al. 2007. Ecological and evolution-
ary insights from species invasions. Trends Ecol. Evol. 22:
465–471.
Schulte, U., Hochkirch, A., L€
otters, S., R€
odder, D., Schweiger,
S., Weimann, T. et al. 2012a. Cryptic niche conservatism
among evolutionary lineages of an invasive lizard. Glob. Ecol.
Biogeogr. 21: 198–211.
Schulte, U., Veith, M. & Hochkirch, A. 2012b. Rapid genetic
assimilation of native wall lizard populations (Podarcis mura-
lis) through extensive hybridization with introduced lin-
eages. Mol. Ecol. 21: 4313–4326.
ª2016 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY. J. EVOL. BIOL. doi: 10.1111/jeb.12840
JOURNAL OF EVOLUTIONARY BIOLOGY ª2016 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY
12 R. J. P. HEATHCOTE ET AL.

Senn, H.V. & Pemberton, J.M. 2009. Variable extent of
hybridization between invasive sika (Cervus nippon) and
native red deer (C. elaphus) in a small geographical area. Mol.
Ecol. 18: 862–876.
Servedio, M.R. 2007. Male versus female mate choice: sexual
selection and the evolution of species recognition via rein-
forcement. Evolution 61: 2772–2789.
Sperling, F.A.H. & Spence, J.R. 1991. Structure of an asym-
metric hybrid zone between two water strider species (Hemi-
ptera: Gerridae: Limnoporus). Evolution 45: 1370–1383.
Stein, A.C. & Uy, J.A.C. 2006. Unidirectional introgression of a
sexually selected trait across an avian hybrid zone: a role for
female choice. Evolution 60: 1476–1485.
Svensson, E.I., Karlsson, K., Friberg, M. & Eroukhmanoff, F.
2007. Gender differences in species recognition and the evo-
lution of asymetric sexual isolation. Curr. Biol. 17: 1843–
1947.
Tokarz, R.R. 1985. Body size as a factor determining domi-
nance in staged agonistic encounters between male brown
anoles (Anolis sagrei). Anim. Behav. 33: 746–753.
Tokarz, R.R. 1995. Mate choice in lizards: a review. Herpetol.
Monogr. 9: 17–40.
Tynkkynen, K., Rantala, M. & Suhonen, J. 2004. Interspecific
aggression and character displacement in the damselfly
Calopteryx splendens.J. Evol. Biol. 17: 759–767.
Uller, T., Stuart-Fox, D. & Olsson, M. 2010. Evolution of
primary sexual characters in Reptiles. In: The Evolution of Pri-
mary Sexual Characters in Animals (J.L. Leonard & A. Cordoba-
Aguilar, eds), pp. 425–452. Oxford University Press, Oxford.
Valdar, W., Solberg, L.C., Gauguier, D., Cookson, W.O., Rawl-
ins, J.N.P., Mott, R. et al. 2006. Genetic and environmental
effects on complex traits in mice. Genetics 174: 275–281.
Verzijden, M.N., ten Cate, C., Sevedio, M.R., Kozak, G.M.,
Boughman, J.W. & Svensson, E.I. 2012. The impact of learn-
ing on sexual selection and speciation. Trends Ecol. Evol. 27:
511–519.
West-Eberhard, M.J. 1983. Sexual selection, social competi-
tion, and speciation. Q. Rev. Biol. 58: 155–183.
Westneat, D.F. 1987. Extra-pair fertilizations in a predomi-
nantly monogamous bird: genetic evidence. Anim. Behav. 35:
877–886.
While, G.M., Michaelides, S., Heathcote, R.J.P., MacGregor,
H.E.A., Zajac, N., Beninde, J., Carazo, P., Perez I de Lanuza,
G., Sacchi, R., Zuffi, M., Horvathova, T., Fresnillo, B.,
Schulte, U., Veith, M., Hochkirch, A. & Uller, T. 2015a. Sex-
ual selection drives asymmetric introgression in wall lizards.
Ecol. Letts. 18: 1366–1375.
While, G.M., Williamson, J., Prescott, G., Horv!
athov!
a, T., Fres-
nillo, B., Beeton, N.J. et al. 2015b. Adaptive responses to
cool climate promotes persistence of a non-native lizard.
Proc. R. Soc. Lond. B 282: 20142638.
Whitehead, H. 2009. SOCPROG programs: analysing animal
social structures. Behav. Ecol. Sociobiol. 63: 765–778.
Wirtz, P. 1999. Mother species–father species: unidirectional
hybridization in animals with female choice. Anim. Behav.
58:1–12.
Supporting information
Additional Supporting Information may be found in the
online version of this article:
Table S1 UK population origins of animals for each
experiment [further details of these populations are
found in Michaelides et al. (2013, 2015)].
Table S2 Details on morphometrics that comprised the
principal components used in different parts of the
study.
Table S3 Identified compounds from femoral pore
secretions of male lizards, with their retention time and
characteristic ion signature.
Table S4 Eigenvector loadings of the different chemical
compounds for the seven main principal components.
Table S5 Summary of Protected ANOVAs conducted on
individual principal components from the femoral pore
extractions.
Table S6 Summary statistics of GLMMs for variables
that predict the number of tongue flicks that male and
female lizards direct towards scented-swabs.
Table S7 Separate contrasts from binomial GLMM for
male-female lineage combinations on likelihood of cop-
ulation during the staged mating trials.
Table S8 Separate contrasts from LMM for male-female
lineage combinations on duration of copulation during
the staged mating trials.
Table S9 Summary statistics of mixed models for fac-
tors that predict home range size (in m
2
) and the num-
ber of individuals overlapping in home range in the
enclosure experiment for male and female lizards.
Figure S1 Number of tongue flicks from male lizards in
response to scent from female lizards based on the
interaction between lineage combination and female
body size in a Poisson GLMM.
Figure S2 Core home range size (in m
2
) for male
French and Italian lizards in clumped and dispersed
basking treatments.
Figure S3 Relationship between female body size and
the number of courtships received from males in the
replicated contact zone experiment.
Received 15 October 2015; revised 18 January 2016; accepted 1
February 2016
ª2016 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY. J. EVOL. BIOL. doi: 10.1111/jeb.12840
JOURNAL OF EVOLUTIONARY BIOLOGY ª2016 EUROPE AN SOCIETY FOR EVOLUTIONARY BIO LOGY
Males drive assortative mating in lizards 13