Similarity of personalities speeds up reproduction in pairs of a
Marylin Rangassamy, Morgan Dalmas, Christophe F
eron, Patrick Gouat, Heiko G. R€
Laboratoire d'Ethologie Exp
erimentale et Compar
ee E.A. 4443 (LEEC), Universit
e Paris 13, Sorbonne Paris Cit
e, Villetaneuse, France
Received 19 October 2014
Initial acceptance 3 December 2014
Final acceptance 22 January 2015
MS. number: 14-00845R
onset of reproduction
Animal personality, i.e. stable individual differences in behaviour, is considered to be subject to evolu-
tionary processes, as it has been shown to be heritable and to entail ﬁtness consequences. Different
hypotheses have been developed to explain why personality variation is maintained within populations,
most likely via processes of balancing selection. One mechanism discussed is that combinations of
similar personalities within breeding pairs could increase ﬁtness. We investigated this purported
mechanism by studying the association between personality combinations within nulliparous pairs and
their onset of reproduction in a monogamous rodent, the mound-building mouse, Mus spicilegus.In
seasonal iteroparous breeders, reproductive timing is relevant to ﬁtness, as a delayed onset of breeding
potentially limits the number of breeding occasions and lowers the chance that offspring will start
reproducing during the same season. Repeated standardized tests carried out prior to pairing revealed
consistent individual differences in subjects' behavioural responses with respect to anxiety and
exploratory activity, indicating the existence of personality types. Within-pair similarity in anxiety levels
affected the chance of reproduction: pairs with similar anxiety scores had a higher probability of
breeding, and were quicker to start, during the observation period, independently of the scores of both
partners of the pair. We propose that this association between ‘personality matching’and the onset of
reproduction could have important life history consequences in this species and might be one of the
mechanisms leading to the maintenance of personality variation within populations.
©2015 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.
Animals frequently show consistent individual differences in
their behavioural responses across time and context; this phe-
nomenon has been termed temperament, behavioural syndrome,
coping style or animal personality (Bell, 2007; Koolhaas et al., 1999;
Wilson, Clark, Coleman, &Dearstyne, 1994). To date, a growing
body of evidence supports the existence of animal personality
across a wide range of taxa, including vertebrates and invertebrates
(Gosling &John, 1999; Kralj-Fi
ser &Schuett, 2014; Stamps &
Groothuis, 2010). Furthermore, there is convincing evidence that
personality traits are heritable (van Oers, de Jong, van Noordwijk,
Kempenaers &Drent, 2005; van Oers &Sinn, 2011) and they
have been shown to entail ﬁtness consequences, e.g. by affecting
survival, reproduction and longevity (Bergeron et al., 2013;
Dingemanse, Both, Drent, &Tinbergen, 2004; R
Coltman, Poissant, &Festa-Bianchet, 2009; R€
odel et al., 2015;
Smith &Blumstein, 2008; Wilson, Godin, &Ward, 2010). Thus,
animal personality can be assumed to be subject to natural selec-
tion (Dall, Houston, &McNamara, 2004; R
eale, Reader, Sol,
McDougall, &Dingemanse, 2007; Sih, Bell, &Johnson, 2004).
The evolutionary mechanisms governing the emergence of this
phenomenon, and in particular the processes leading to the
generally high variability of personality types within species and
populations, are still being debated (reviewed in: Bergmüller, 2010;
Kight, David, &Dall, 2013). Persistent personality variation within
populations may seem surprising at ﬁrst sight, given that certain
personalities, e.g. more aggressive and exploratory individuals, are
frequently reported to have ﬁtness advantages (cf. Smith &
Blumstein, 2008). Such types might then be expected to be fav-
oured by natural selection, thus spreading in their population and
reducing overall variability. Different processes of balancing selec-
tion are frequently proposed to be involved in maintaining varia-
tion in personalities. For example, it has been considered that
negative frequency-dependent selection processes, i.e. ﬁtness
beneﬁts of rare personalities, could favour the stable coexistence of
different types (Dall et al., 2004; Mathot et al., 2011; Wolf, van
Doorn, &Weissing, 2008). In addition, personality variation could
be maintained by differential ﬁtness advantages under
*Correspondence: H. G. R€
odel, Laboratoire d'Ethologie Exp
ee E.A. 4443 (LEEC), Universit
e Paris 13, Sorbonne Paris Cit
E-mail address: firstname.lastname@example.org (H. G. R€
Contents lists available at ScienceDirect
journal homepage: www.elsevier.com/locate/anbehav
0003-3472/©2015 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.
Animal Behaviour 103 (2015) 7e15
heterogeneous environmental conditions (Cote, Dreiss, &Clobert,
2008; Dingemanse &R
eale, 2005). Balancing selection could also
play a role when the direction of the ﬁtness consequences of per-
sonality changes at different stages of an animal's life history
(Nettle, 2006). For example, a study in North American red squir-
rels, Tamiasciurus hudsonicus, showed that the mother's personality
was associated with offspring survival prior to weaning and in
winter in a contrasting way (Boon, R
eale, &Boutin, 2007).
Similar patterns of selection might also arise from ﬁtness dif-
ferences between certain combinations of personality types within
breeding pairs (Schuett, Tregenza, &Dall, 2010). This is, for example,
supported by studies on great tits, Parus major, in which pairs with
similar personalities, either slow or fast explorers, produced
offspring in better body condition and recruited more offspring
than pairs with other personality combinations (Both, Dingemanse,
Drent, &Tinbergen, 2005; Dingemanse et al., 2004). Further sup-
port for potentially ﬁtness-increasing effects of similar (positive
assortative) personalities comes from a study on zebra ﬁnches,
Taeniopygia guttata. Breeding pairs that were both highly explor-
atory and aggressive fostered chicks in better body condition than
pairs of other behavioural combinations (Schuett, Dall, &Royle,
2011), while females actively seem to choose their partners on the
basis of personality (Schuett, Godin, &Dall, 2011). In addition, a
study in eastern bluebirds, Sialia sialis, reported that parental pairs
that behaved more similarly with respect to territorial aggression
reared the heaviest offspring (Harris &Siefferman, 2014). Apart
from these studies in birds, the existence of a link between assor-
tative personality and reproduction ﬁnds support by studies in ﬁsh
(Ariyomo &Watts, 2013; Budaev, Zworykin, &Mochek, 1999) and
squid (Sinn, Apiolaza, &Moltschaniwskyj, 2006).
However, to the best of our knowledge, no study has investi-
gated such associations between similarities in personality types
and parameters of reproduction in a mammal (Schuett et al., 2010).
Furthermore, most studies reporting advantages of such ‘person-
ality matching’within breeding pairs show associations with
offspring parameters (in birds, offspring growth and survival: Both
et al., 2005; Harris &Siefferman, 2014; Schuett, Dall, et al., 2011),
which might have been mainly the result of optimized biparental
care. However, the timing of reproduction, i.e. the probability of an
early start to breeding, can also contribute importantly to parental
ﬁtness. This is especially relevant in seasonal environments, in
which the number of annual breeding events and favourable con-
ditions for offspring development are restricted by the length of the
vegetation period, and thus a prompt onset of breeding can in-
crease ﬁtness (Eccard &Yl€
onen, 2001; R€
odel, von Holst, &Kraus,
2009). This also applies to the mound-building mouse, Mus spic-
ilegus, which has a life expectancy of less than 1 year under ﬁeld
conditions. Almost exclusively juveniles survive the winter period,
in nest chambers under specially constructed mounds (Garza et al.,
1997; Poteaux, Busquet, Gouat, Katona, &Baudoin, 2008; Szenczi
et al., 2011). These overwintering individuals then start to repro-
duce in spring after dispersal from the mound and give birth to a
few consecutive litters, with early born offspring having the po-
tential to mature and to reproduce within the same season (Gouat,
Katona, &Poteaux, 2003; Milishnikov, Raﬁev, &Muntianu, 1998).
To shed light on the adaptive signiﬁcance of assortative per-
sonality in breeding pairs of this monogamous rodent (Baudoin
et al., 2005), we studied the timing and probability of reproduc-
tion in pairing experiments of nulliparous mound-building mice
under laboratory conditions. Prior to pairing, we attempted to
quantify personality types with respect to the animals' consistent
behavioural responses in two standardized tests that are frequently
used in murine rodents (e.g. Lewejohann, Zipser, &Sachser, 2011;
odel &Meyer, 2011): open ﬁeld, usually interpreted to indicate
activity and exploration tendency, and elevated plus maze,
reﬂecting anxiety-related behaviour (Archer, 1973). In accordance
with the results obtained in several other taxa, we predicted that
pairs of this monogamous mammal with a similar (i.e. positive
assortative) personality would start to reproduce earlier or with a
higher probability than pairs with more dissimilar combinations.
Furthermore, we also considered the nonexclusive hypothesis that
male or female personality might exert distinct effects on repro-
duction of the pair. For example, such effects have been shown in
great tits, in which ﬂedgling size was directly associated with the
mother's exploratory behaviour (Both et al., 2005).
Study Animals and Housing Conditions
Our breeding stock were descendants (16th and 17th genera-
tion) of mound-building mice caught from the wild at different sites
in Hungary in 1999. In addition, we attempted to maintain genetic
variation of the breeding stock by regularly (every 2e4 years)
adding some new individuals, captured at the same Hungarian
collection sites. We kept the animals on a 14:10 h light/dark cycle
(white lights off at noon) in standard polycarbonate cages
(26 14 cm and 16 cm high, Iffa Credo, Lyon, France), containing
wood shavings as bedding. Animals always had ad libitum access to
rodent standard diet (Special Diets Services, Ext. M20, Witham,
Essex, U.K.) and water. Temperature in the housing rooms was
maintained at 21 ±0.5
C, and relative humidity at approximately
50%. All animals, including the focal animals of our study, were
transferred to clean cages around every second week but never
within 10 days before behavioural tests. We always provided several
cotton balls (diameter approximately 5 cm), which the animals
(including juveniles) use for nest building in a corner of their cage.
This study consisted of two trials (hereafter referred to as Cohort
1 and Cohort 2), repeating the same experimental procedure with
different animals in 2 years (Cohort 1 in 2012 and Cohort 2 in 2013).
All focal animals stemmed from unculled litters born to adult virgin
maleefemale breeding pairs (Cohort 1: 10 litters from 10 breeding
pairs; Cohort 2: 12 litters from 12 breeding pairs). On postnatal day
14, four pups were randomly chosen from each litter (preferably
two females and two males) as focal animals and individually
marked with different symbols on their back using a black, per-
manent nontoxic hair dye (Nyanzol-D, Greenville Colorants, Jersey
City, NJ, U.S.A.). For this, the animals were not removed from their
home cage, but were held tight at the base of their tail (front paws
remaining on the ground) by the experimenter with one hand,
while the hair dye was applied to the subject's back with a small
brush. This procedure took about 30 s per individual. At this time,
we also determined their sex by external genital inspection. The
remaining pups were not removed and thus all siblings remained
together until weaning (separation from the mother) on postnatal
day 28, when the four focal individuals per litter were transferred
together into a new cage. Subjects were re-marked shortly af ter the
ﬁrst behavioural test (T
, see below) in order to allow individual
recognition within their sibling groups. On postnatal day 55, before
reaching maturity (Busquet, Nizerolle, &F
eron, 2009), all animals
were transferred into separate cages.
The total number of focal animals used for the study was 82 (41
females and 41 males), with 36 (18 pairs) in Cohort 1 and 46 (23
pairs) in Cohort 2.
Animals were kept and treated according to the ethical and
animal care guidelines of France (where the project was carried
out) and the institutional guidelines of animal welfare.
M. Rangassamy et al. / Animal Behaviour 103 (2015) 7e158
Experimental procedures were approved by the local authority for
laboratory animal care and use (Comit
e d'Ethique en Exp
mentation Animale ‘Charles Darwin’; authorization codes: Ce5/
Subjects were kept singly during postnatal days 55e90. Being
solitary is not an unusual situation in this species, as solitary fe-
males and males have been frequently found under natural con-
ditions (Simeonovska-Nikolova, 2012). Furthermore, the growth
rates of the animals during the last 2 weeks before being separated
from their siblings (mean ±SE ¼0.69 ±0.19 g) did not differ
signiﬁcantly from those during the ﬁrst 2 weeks of single housing
(0.82 ±0.15 g; paired ttest: t¼0.72, P¼0.47), indicating that the
animals did not suffer from obvious effects of separation stress.
Furthermore, our recent experiments show that single housing of
females and males prior to pairing (most probably resembling the
natural situation of dispersal) increases their probability of suc-
cessful reproduction (F
eron &Gheusi, 2003).
After the experiments, and after the offspring of the experi-
mental pairs were weaned, adults were killed in accordance with
French animal law. For this, they were ﬁrst anaesthetized by putting
them into a closed transparent plastic box ﬁlled with isoﬂurane gas
at a concentration of 3% (IsoFlo, Axience, France), administered by
an automatic system (Univentor 400 Anaesthesia Unit, Univentor
Ltd, Zejtun, Malta). Then, anaesthetized individuals were killed by
putting them into a closed transparent plastic box ﬁlled with a high
dose of CO
gas (delivered by a compressed gas cylinder) for at least
5 min. They were observed until all muscle activity and other signs
of life had been absent for at least 30 s. After removal, we carried
out cervical dislocation in order to ensure death. This whole pro-
cedure was conducted by a qualiﬁed and experienced person. The
offspring of the experimental animals were used for further
behavioural experiments (not included in this paper).
Procedures and Experimental Settings
All procedures and experiments were carried out with two
separate cohorts (Cohort 1 and 2), conducted by two different ex-
perimenters. We used the same experimental setting in both co-
horts, except for slight differences in the apparatuses used for the
battery tests, as outlined below. The animals of the different co-
horts were kept in different housing rooms, but with the same
conditions of temperature, humidity and illumination (see above).
Repeated Behavioural Tests
To assess individual personality types, we carried out repeated
behavioural testing using an open ﬁeld test and an elevated plus
maze test. Two test batteries were done prior to the pairing with a
potential mating partner (see below). The ﬁrst test battery (T
carried out on postnatal days 40 (open ﬁeld) and 44 (elevated plus
maze), when individuals were kept in sibling groups of four, and
the second (T
) on days 70 (open ﬁeld) and 74 (elevated plus maze),
when individuals were housed singly.
Experiments were always conducted between 1400 and 1600
hours during the animals' activity period (red light phase). In-
dividuals were returned to their home cages immediately after each
test, and we cleaned the apparatuses with water and detergent
(Cleansinald, Johnson Diversey, Fontenay-sous-Bois, France) be-
Open ﬁeld test
The open ﬁelds used for both cohorts were made of white
polyethylene and consisted of a circular arena surrounded by walls.
However, they differed slightly in size. The one used for experi-
ments conducted with Cohort 1 had a diameter of 48 cm and the
surrounding walls were 50 cm high. The one used for Cohort 2 had
a diameter of 60 cm with surrounding walls of 69 cm.
Each animal was placed on a deﬁned position at the edge of the
arena, and its behaviour was recorded for 5 min by use of a digital
video camera mounted 120 cm above the centre of the arena. For
analysis, we deﬁned a (circular) central zone of the arena with a
diameter of 16 cm for Cohort 1 and 20 cm for Cohort 2. For both
cohorts, this represented one-third of the total diameter of the
open ﬁeld. We quantiﬁed the total distance covered by the animal
and the distance covered in the centre of the arena with Ethovision
XT7 (Noldus Information Technology, Wageningen, The
Elevated plus maze
The apparatuses used for Cohort 1 and Cohort 2 were both made
of rigid PVC consisting of four arms, 5 cm wide and 30 cm long,
arranged at a 90
angle, and mounted 70 cm above the ﬂoor by a
stable construction. Two opposite arms enclosed by 30 cm high
walls and two open arms without walls were connected by a
10 10 cm central platform. This platform was surrounded by clear
Plexiglas walls with round openings (diameter 5 cm) to all four
directions. Preliminary tests had revealed that such a box around
the platform helps to considerably reduce the chance that subjects
(which were descendants of wild animals) would jump off the
maze. During the experiments with Cohort 1, four of 18 individuals
jumped off the apparatus during the ﬁrst trial but none during the
second trial and none during the experiments with Cohort 2. We
did not exclude the data from animals that jumped off the maze for
later analysis. The apparatuses used for the two cohorts differed
slightly in the way the closed arms were constructed. For Cohort 1,
the terminal ends of the closed arms were left open. In contrast, the
closed arms of the elevated plus maze used for the experiments
with Cohort 2 were surrounded by walls on three sides.
Each animal was placed on the central platform facing an open
arm and also here its behaviour was recorded for 5 min by use of a
digital video camera mounted 120 cm above the centre of the maze.
The percentage time spent in the open area of the elevated plus
maze and the frequencies of exits from the closed arm were
quantiﬁed. These were deﬁned as crossing the respective thresh-
olds with more than 50% of the animal's body length. Both mea-
surements have been shown to reﬂect anxiety-related behaviour in
laboratory rodents (Pellow, Chopin, File, &Briley, 1985; Walf &Frye,
Behavioural tests using an open ﬁeld as well as an elevated plus
maze of similar size and construction have been previously and
successfully used in the mound-building mouse to describe its age-
related dynamics in risk-taking behaviour (Lafaille &F
Formation of Pairs and Survey of Reproduction
When subjects were about 95 days old (average age ±SE: fe-
males: 94.9 ±1.5 days; males: 94.9 ±1.3 days), we formed the pairs
(18 in Cohort 1; 23 in Cohort 2) and housed them in standard
polycarbonate laboratory mouse cages (26 14 cm and 16 cm
high). Note that this species usually matures prior to an age of 80
days (Busquet et al., 2009). Before pairing, animals were weighed to
the nearest 0.01 g. We checked the pedigree of each individual and
veriﬁed that partners stemmed from different direct kin lines for at
least two generations. Apart from this condition, partners were
Experimental pairing of an unfamiliar female and male in this
species is usually characterized by an initial increase in chasing
behaviour (Patris, Gouat, Jacquot, Christophe, &Baudoin, 2002).
However, these initial chasing events typically decrease in fre-
quency and appearance within the ﬁrst 30 min. During this time,
M. Rangassamy et al. / Animal Behaviour 103 (2015) 7e15 9
we always monitored the new pairs in order to terminate the
pairing if the aggression did not stop, but this never occurred
during our experiments.
Females were then weighed once per week in order to detect
pregnancies. In total, pairs were checked for reproduction during a
period of 90 days; this duration corresponds to more than three
times the gestation period of this species (around 20 days of
gestation in primiparous females; F
eron &Gouat, 2007). Starting at
20 days after pairing, we checked the cages for new litters on a daily
basis. In addition to the date of birth of each litter, we noted the litter
size. We also weighed the pups on postnatal day 12 (Cohort 1) or day
14 (Cohort 2) in order to obtain a measure of offspring development.
All statistical analyses were done using the program R, version
3.1.1 (R Core Team, 2014). Data obtained from experiments with
Cohort 1 and Cohort 2 were analysed separately, as they were
conducted by different experimenters and there were slight dif-
ferences in the setting (size of open ﬁeld arena, construction of
elevated plus maze, see above).
Calculation of personality scores
To capture the information of the different behavioural re-
sponses per test (open ﬁeld, elevated plus maze) we calculated
scores (exploration scores based on open ﬁeld responses; anxiety
scores based on elevated plus maze responses) using the scaled ﬁrst
component as provided by a principal component analysis, PCA (R
function: prcomp). This was done separately for T
and for T
for Cohorts 1 and 2. These new variables were used to test for
repeatability across time. For further analysis (see below), we then
averaged the scores over both time steps. In all cases, the ﬁrst
component captured more than 51% (up to 88%) of the variability of
the behavioural responses considered (see Appendix Table A1 with
the explained variations of all ﬁrst principal components).
Furthermore, in all cases the ﬁrst component had an eigenvalue >1,
indicating that this component accounted for more variance than
any of the original variables of the standardized data.
The scores of males and females were compared by linear
mixed-effects models LMM, using litter identity as a random factor.
This was done with the R package lme4 (Bates, Maechler, &Bolker,
2014). Pvalues were extracted by Wald chi-square tests (type III).
Repeatability of behavioural responses
Repeatability, i.e. the intraclass correlation calculated as the
proportion of phenotypic variation that can be attributed to
between-subject variation (Lessells &Boag, 1987), was calculated
for both scores (anxiety, exploration), separately for Cohorts 1 and
2. We used LMM-based calculations of repeatability with the R
package rptR (Nakagawa &Schielzeth, 2010), and we assessed 95%
conﬁdence intervals by 1000 bootstrap steps. Individual identity
was used as a random factor. Pvalues were calculated by 1000
Within-pair similarity and reproductive timing
We tested the effects of sets of different predictor variables on
the timing of reproduction of the different breeding pairs using
multiple Cox proportional hazards regression (R package: survival),
allowing for the analysis of censored data (Therneau &Grambsch,
2000). This kind of model tests for the combined timing and
probability of occurrence of an event. The latency until pairs had
their ﬁrst litter after pairing and the probability of reproduction,
censored after an observation period of 90 days, were used as
response variables. Predictor variables were based on the explo-
ration scores and anxiety scores as obtained by PCA (see above).
For each of the two variables, we used the absolute difference
between the scores of the male and the female within each pair as a
measure of similarity of personality types, calculated as:
Similarity index ¼
Low values indicate high similarity and high values indicate low
similarity within pairs. The distributions of values for similarity
indices calculated for both traits and for both experiments and
cohorts are given in Appendix Fig. A1. We also calculated models
separately based on the scores of males and of females in order to
test for distinct (‘direct’) effects of the personality type of each
partner (see Both et al., 2005; Schuett, Dall, et al., 2011). In addition,
we considered additive effects of female body mass (covariate)
measured on the day of pairing in combination with all other
predictor variables as mentioned above. This was done in order to
adjust for potential effects of female body condition on reproduc-
tive timing and performance (Millesi, Huber, Everts, &Dittami,
odel et al., 2005). We also considered models in which we
concurrently included the within-pair similarity of exploration and
of anxiety scores (see Appendix Table A2).
Information theory-based model selection
For model selection, we constructed sets of 16 candidate models
(including a ‘null’model, which did not include any slope para-
meter), separately for each of the two cohorts, where we consid-
ered the above-mentioned combinations of the predictor variables
(Burnham &Anderson, 2002). The full sets of models are given in
Appendix Table A2.
The aim of the AIC-based, information-theoretical approach that
we used is to identify the most parsimonious model, i.e. the model
of the set considered that represents the data adequately with the
smallest possible number of parameters. We used the second-order
AIC, the AICc, which includes a correction term for small sample
sizes. All models of the set were ordered from ‘best’(lowest AICc) to
‘worst’(highest AICc). We report AICc differences
) to compare the support that the
different models had for best approximating the data. Burnham and
Anderson (2002) suggested that models with
2 can be
considered to have substantial support, and models with
about 4e7 have considerably less support. We also calculated
normalized Akaike weights (w
) for each model, which can be
interpreted as a measure of the evidence in favour of model ias
being the best approximating model of the set (Burnham &
Anderson, 2002). In addition, Pvalues were calculated for the
best approximating model of the set using a likelihood ratio test.
Model diagnostics and graphical presentation
For model diagnostics of Cox proportional hazards regressions,
we veriﬁed the assumption of proportional hazards by visually
checking plots of Schoenfeld residuals versus the transformed time.
In addition, we ruled out the existence of potential nonlinear effects
by plotting the Martingale residuals of the models against the
different covariates (Fox, 2002). There was a certain degree of
collinearity between the different anxiety scores and exploration
scores (Cohort 1: F
¼0.054, P¼0.17; Cohort 2:
¼0.127, P¼0.015); however, the level of explained
) of these correlations was not particularly high. We
calculated variance inﬂation coefﬁcients (VIF) in order to check for
(multi)collinearities among predictor variables (Zuur, Ieno, &
Elphick, 2010) for all models of the set including more than one
predictor variable. VIF were always lower than 1.5, indicating no
notable interfering effects of multicollinearities.
For graphical presentation only (see Results), the covariate
(Table 1) was split into three levels, with the two
M. Rangassamy et al. / Animal Behaviour 103 (2015) 7e1510
quartiles [0e25% of the data] and [75e100% of the data] repre-
senting the groups with high and low similarity, and 25e75% of the
data representing the level with intermediate similarity. Note that
the effects of within-pair similarity in anxiety scores on repro-
duction were still statistically signiﬁcant after the transformation of
this predictor variable into three categories (see Pstatistics for
Cohorts 1 and 2 in the Results).
Repeatability across Time
Elevated plus maze
With respect to data obtained from Cohort 1 as well as from
Cohort 2, the PCA revealed similar results during both time steps.
Higher scores of the ﬁrst components were always positively
associated with more signs of anxiety, such as a lower percentage
time spent in the open arms and fewer exits from the closed arms.
These scores did not differ between females and males during any
of the time steps in Cohort 1 or 2 (LMM: all P>0.10). Anxiety scores
were repeatable across time with respect to Cohort 1 (R¼0.648,
95%CI ¼[0.389, 0.813], P¼0.001) and Cohort 2 (R¼0.519, 95%CI ¼
[0.286, 0.697], P¼0.001; Fig. 1a, c).
Open ﬁeld test
Animals with higher scores, representing the ﬁrst components
of PCAs for the two time steps of Cohorts 1 and 2, showed more
Sets of models (Cox proportional hazards regressions; models with
explaining the timing/probability of reproduction (R) by means of the birth of a ﬁrst
litter in pairs of nulliparous mound-building mice
Model terms K
Cohort 1 R[Anxiety
] 2 0 0.266
] 2 1.57 0.121
] 3 1.80 0.108
] 3 2.01 0.097
] 3 2.35 0.082
4 2.71 0.069
] 3 2.93 0.061
] 3 3.06 0.058
Cohort 2 R[Anxiety
] 2 0 0.426
] 3 2.47 0.124
R[Null model] 1 3.27 0.083
4 3.56 0.072
The models are ordered by
AICc. The onset of reproduction was monitored for 90
days after pairing (censored data). Predictor variables considered are the similarity
(calculated as the absolute score difference between males and females) in anxiety
or exploration scores within each pair. Furthermore, distinct effects of score values
of males and of females were considered. In addition, female body mass prior to
pairing was included in the construction of the models. The number of estimable
AICc and Akaike weights (w) is given. Analyses are based on data
from two cohorts (Cohort 1: N¼18 pairs; Cohort 2: N¼23 pairs). Full sets of models
are given in Appendix Table A2.
–3 –2 –1 0 1 2 3
–3 –2 –1 0 1 2 3
–3 –2 –1 0 1 2 3
–3 –2 –1 0 1 2 3
Anxiety score T2
Exploration score T2
Exploration score T1
Anxiety score T1
Figure 1. Consistencies across time in (a, c) anxiety scores (measured in an elevated plus maze) and (b, d) exploration scores (measured in an open ﬁeld test) of female (circles) and
male (triangles) mound-building mice at two times of testing (T
). Data from two cohorts are presented. (a, b) Cohort 1: N
¼36. (c, d) Cohort 2: N
scores indicate more pronounced anxiety-related or exploratory behaviours during the tests, respectively. All behaviours were signiﬁcantly repeatable based on the proportion of
phenotypic variation attributed to between-subject variation. There were no signiﬁcant sex-speciﬁc differences. See text for details on statistics.
M. Rangassamy et al. / Animal Behaviour 103 (2015) 7e15 11
signs of exploratory activity, covering a greater total distance and a
greater distance in the central zone of the arena. Also here, scores
did not differ between females and males during any of the time
steps in Cohort 1 or 2 (LMM: all P>0.10). Exploration scores were
also repeatable across time in Cohort 1 (R¼0.510, 95%CI ¼[0.225,
0.712], P¼0.001) and Cohort 2 (R¼0.387, 95%CI ¼[0.118, 0.606],
P¼0.004; Fig. 1b, d).
Personality Matching and Onset of Reproduction
Of 18 pairs, 14 (77.8%) started to reproduce in Cohort 1, and of 23
pairs 10 (43.5%) started to reproduce in Cohort 2, within 90 days
after pairing. The analysis of the probability of starting to repro-
duce, taking into account the timing of reproduction by a Cox
proportional hazards survival model, revealed highly similar results
in both cohorts, despite the variation in reproductively active pairs
The start of reproduction was best explained, by means of the
lowest AICc, by the similarity in anxiety scores within pairs
(Table 1). According to this model, pairs with more similar anxiety
levels (covariate) reproduced signiﬁcantly earlier/with a higher
probability than more dissimilar ones (Cohort 1: c
¼0.342, P¼0.007; Cohort 2: c
Results were also signiﬁcant when we split the covariate into three
¼7.05, P¼0.008; c
¼4.49, P¼0.034; Fig. 2a, b).
Models considering similarities in exploration score, or the in-
dividual scores of either the male or the female partner in anxiety
or exploration, did not ﬁnd notable support by the data, in either
Cohort 1 or 2 (Table 1). In both model sets, and in particular in
Cohort 1, the effects of the initial female body mass also found some
support by the data (
AICc <2; c
¼5.73, P¼0.016). This result
revealed that heavier females tended to reproduce earlier/with a
higher probability than lighter ones.
Personality Traits, Litter Size and Pup Body Mass
For pairs that started to reproduce, we tested for correlations
(Spearman rank due to moderate sample sizes) of parental
temperamental scores and the within-pair similarity of these
scores (anxiety and exploration, respectively) with litter size
(mean ±SD: 6.4 ±1.2) or with the averaged individual pup body
masses (Cohort 1: mean body mass on postnatal day 12 ±SD:
4.1 ±0.5 g; Cohort 2: mean body mass on postnatal day 14 ±SD:
5.1 ±0.7 g).
There were no signiﬁcant bivariate correlations between these
variables (all P>0.10). That is, based on our moderate sample sizes,
there was no evidence for an association between either similar-
ities in temperamental traits or parental temperamental traits and
litter size or offspring body mass.
Behavioural tests revealed consistent individual differences in
exploratory activity and anxiety-related responses across time,
indicating the existence of distinct personality types in the mound-
building mouse. Most importantly, our results suggest ﬁtness-
related beneﬁts in pairs with assortative personalities with
respect to anxiety, showing a higher probability of starting to
reproduce than more dissimilar pairs. These ﬁndings were based on
coherent results obtained in independent experiments with two
cohorts of animals.
To our knowledge, this is the ﬁrst study reporting effects of
personality matching on reproductive activity in a mammal. Our
report supports the results obtained in some other taxa. A study in
convict cichlids, Amatitlania nigrofasciata, after spawning revealed a
positive correlation in freezing behaviour between partners of
reproducing pairs but no such similarity in nonreproductive ones
(Budaev et al., 1999). Further evidence comes from a recent study in
guppies, Poecilia reticulata, in which females that mated with males
with a more similar boldness level had higher parturition success
than females with more dissimilar mating partners (Ariyomo &
Watts, 2013). In a study on dumpling squid, Euprymna tasmanica,
the eggs of bold and intermediate females were predominantly
fertilized by males having similar phenotypes; however, repro-
duction of shy females was not associated with male personality
(Sinn et al., 2006). Studies on birds, reporting a link between per-
sonality matching and offspring body condition or recruitment
rates, also showed that male and female behavioural types per se
had distinct effects on reproductive parameters such as chick body
mass (Schuett, Dall, et al., 2011), nest survival and the choice of
high-quality nestboxes (Both et al., 2005). Such independent effects
of components of parental personality were not supported by our
study. Furthermore, we did not ﬁnd signiﬁcant correlations be-
tween parental personality matching and parameters of offspring
30 60 90 0 6030 90
Proportion of reproducing pairs
Figure 2. Kaplan-Meier survival curves for (a) Cohort 1 and (b) Cohort 2 showing the evolution of the proportion of reproducing pairs of mound-building mice, for groups with
high, medium and low within-pair similarities in anxiety scores (see text for details of calculation). Note that for statistical analyses (Table 1) the original data on similarities in
anxiety scores (covariates) were used; however, analyses based on categorized data were also statistically signiﬁcant (see text). Day 0 indicates the day of pairing; the census of
reproductive activity of each pair was terminated at day 90.
M. Rangassamy et al. / Animal Behaviour 103 (2015) 7e1512
condition or development (i.e. pup body mass). However, conclu-
sions based on this nonsigniﬁcant result are difﬁcult to draw as the
sample size available for this analysis was rather low. Furthermore,
keeping animals under laboratory conditions, and thus protecting
them from environmental constraints such as food restriction,
might signiﬁcantly hamper the detection of potential maternal/
parental effects on offspring development (Lafaille, Gouat, &F
2015; Ung, F
eron, Gouat, Demouron, &Gouat, 2014).
Although the percentage of reproductively active pairs differed
strongly between the two cohorts (77.8% versus 43.5% of repro-
ducing pairs), we obtained similar results providing evidence for a
link between personality matching and the onset/probability of
reproduction. In addition, our study emphasizes that potential
differences in conditions such as handling (e.g. due to different
experimenters) or housing (e.g. different housing rooms) between
consecutive trials might strongly affect the performance of animals
kept under supposedly standardized laboratory conditions (see also
Richter et al. (2011) for variation in behavioural responses to
standardized tests between different laboratories).
Different mechanisms might underlie the observed association
between within-pair personality matching and the timing/proba-
bility of reproduction. In general, assortative mating with respect to
different phenotypic traits is not uncommon. A meta-analysis
across published data from 254 species showed an overall ten-
dency towards positive assortative mating within populations
(Jiang, Bolnick, &Kirkpatrick, 2013). It could be argued that the
results of our experiment may in fact reﬂect mate choice decisions
ser, Sanguino Mostajo, Preik, Pek
ar, &Schneider, 2013;
Schuett, Dall, et al., 2011). Individuals might adaptively choose
partners with a similar behavioural type, as it could improve the
coordination of behaviour between partners as compared to non-
assortative pairs (Schuett et al., 2010). This could be particularly
relevant to ﬁtness in species with biparental care, such as the
mound-building mouse, in which high levels of cooperation be-
tween both partners (Patris &Baudoin, 2000) could help optimize
the development and survival of the young.
On the proximate level, the observed association between per-
sonality matching and reproduction might be affected by the
quality of the pair relationship. Putting two potential mating
partners together does not inevitably lead to the formation of a pair
bond (von Holst, 1998). We propose that combinations of dissimilar
personalities could have resulted in more unstable and disharmo-
nious conditions. For example, disharmonious pairs of monoga-
mous tree shrews, Tupaia belangeri, as characterized by a
comparatively lower degree of positive social interactions and oc-
casionally increased within-pair aggression, show escalated stress
hormone concentrations and increased heart rates in comparison
to animals in harmonious pairings (von Holst, 1998). Generally,
chronically increased levels of stress might exert suppressive ef-
fects on reproductive function, such as decreases in fertility or
higher rates of fetal resorption during early stages of pregnancy, all
leading to a lower probability of reproduction (von Holst, 1998;
Sapolsky, Romero, &Munck, 2000). In the mound-building
mouse, pairwise encounters of unfamiliar adult males and fe-
males are frequently characterized by agonistic interactions, initi-
ated by the refusal and defensive behaviour of the female towards
approaches of the male (Patris et al., 2002). Enduring and frequent
aggressive encounters in unstable pairings could entail negative
stress effects, which may play a role in suppressing reproduction.
However, studies in this species point out that an initial increase in
agonistic interactions between partners of unfamiliar pairings is
essential in triggering the onset of reproduction (Busquet et al.,
2009). There is need for further studies exploring to what extent
personality differences are involved in mate choice decisions in this
Our ﬁndings highlight the mound-building mouse as another
example of a mammal providing indications for the existence of
personality by means of consistent individual differences in
behaviour across time. But most importantly, this study demon-
strates for the ﬁrst time in a monogamous mammal that personality
matching provides potential ﬁtness beneﬁts to the partners of a pair
by increasing the probability of reproduction. A rapid onset of
breeding is clearly an important component of individual ﬁtness in
short-lived species. In particular in seasonally breeding iteroparous
small mammals, delays in the onset of reproductive activity can limit
the number of litters produced per season (Eccard &Yl€
Furthermore, offspring born earlier during the vegetation period
potentially have a higher chance of starting to reproduce during the
same season, with positive implications for their parents' inclusive
ﬁtness. Assuming there is a genetic basis for anxiety/emotionality
(cf. Willis-Owen &Flint, 2006), we propose that the observed
advantage of positive assortative pairing with respect to this trait
could contribute to maintaining personality variation within pop-
ulations. Thus the results of this study could add to the under-
standing of the evolution of personality in monogamous mammals.
We thank Ludivine Jaravel, Simone Demouron, Samira Varela
and Sonia Varela for excellent animal care, and we are grateful to
Paul Devienne for technical support. This study was supported by a
research grant from the Universit
e Paris 13 to H.G.R., C.F. and P.G.
e Recherche 26.02.2013). M.R. was supported by a
PhD fellowship provided by the Ecole Doctorale Galil
Archer, J. (1973). Tests for emotionality in rats and mice: a review. Animal Behaviour,
Ariyomo, T. O., & Watts, P. J. (2013). Disassortative mating for boldness decreases
reproductive success in the guppy. Behavioral Ecology, 24, 1320e1326.
Bates, D., Maechler, M., & Bolker, B. (2014). lme4: Linear mixed-effects models using
Eigen and S4. R package version 1.1-7.http://CRAN.R-project.org/package¼lme4.
Baudoin, C., Busquet, N., Dobson, F. S., Gheusi, G., F
eron, C., Durand, J.-L., et al.
(2005). Maleefemale associations and female olfactory neurogenesis with pair
bonding in Mus spicilegus.Biological Journal of the Linnean Society, 84, 323e334.
Bell, A. M. (2007). Future directions in behavioural syndromes research. Proceedings
of the Royal Society B: Biological Sciences, 274, 755e761.
Bergeron, P., Montiglio, P. O., R
eale, D., Humphries, M. M., Gimenez, O., & Garant, D.
(2013). Disruptive viability selection on adult exploratory behaviour in eastern
chipmunks. Journal of Evolutionary Biology, 26,766e774.
Bergmüller, R. (2010). Animal personalities and behavioural syndromes. In
P. Kappeler (Ed.), Animal behaviour: Evolution and mechanisms (pp. 587e621).
Heidelberg, Germany: Springer.
Boon, A. K., R
eale, D., & Boutin, S. (2007). The interaction between personality,
offspring ﬁtness and food abundance in North American red squirrels. Ecology
Letters, 10, 1094e1104 .
Both, C., Dingemanse, N., Drent, P. J., & Tinbergen, J. M. (2005). Pairs of extreme
avian personalities have highest reproductive success. Journal of Animal Ecology,
Budaev, S. V., Zworykin, D. D., & Mochek, A. D. (1999). Individual differences in
parental care and behaviour proﬁles in the convict cichlid, a correlation study.
Animal Behaviour, 58,195e202.
Burnham, K. P., & Anderson, D. R. (2002). Model selection and multimodel inference: A
practical information-theoretic approach (2nd ed.). New York, NY: Springer.
Busquet, N., Nizerolle, C. L., & F
eron, C. (2009). What triggers reproductive life?
Effects of adolescent cohabitation, social novelty and aggression in a monoga-
mous mouse. Ethology, 115,87e95.
Cote, J., Dreiss, A., & Clobert, J. (2008). Social personality trait and ﬁtness. Pro-
ceedings of the Royal Society B: Biological Sciences, 275, 2851e2858.
Dall, S. R. X., Houston, A. I., & McNamara, J. M. (2004). The behavioural ecology of
personality: consistent individual differences from an adaptive perspective.
Ecology Letters, 7,734e739.
Dingemanse, N. J., Both, C., Drent, P. J., & Tinbergen, J. M. (2004). Fitness conse-
quences of avian personalities in a ﬂuctuating environment. Proceedings of the
Royal Society B: Biological Sciences, 271,847e852.
M. Rangassamy et al. / Animal Behaviour 103 (2015) 7e15 13
Dingemanse, N. J., & R
eale, D. (2005). Natural selection and animal personality.
Behaviour, 142,1159e118 4 .
Eccard, J. A., & Yl€
onen, H. (2001). Initiation of breeding after winter in bank voles:
effects of food and population density. Canadian Journal of Zoology, 79,
eron, C., & Gheusi, G. (2003). Social regulation of reproduction in the female
mound-builder mouse (Mus spicilegus). Physiology and Behavior, 78,717e722.
eron, C., & Gouat, P. (2007). Paternal care in the mound-building mouse reduces
inter-litter intervals. Reproduction, Fertility and Development, 19, 425e429.
Fox, J. (2002). Cox proportional-hazards regression for survival data. Appendix to: An R
and S-PLUS companion to applied regression (pp. 12e17). Comprehensive R
Archive Network. http://cran.r-project.org/doc/contrib/Fox-Companion/
Garza, J. C., Dallas, J., Duryadi, D., Gerasimov, S., Croset, H., & Boursot, P. (1997).
Social structure of the mound-building mouse Mus spicilegus revealed by ge-
netic analysis with microsatellites. Molecular Ecology, 6,1009e1017.
Gosling, S. D., & John, O. P. (1999). Personality dimensions in nonhuman animals: a
cross-species review. Current Directions in Psychological Science, 8,69e75.
Gouat, P., Katona, K., & Poteaux, C. (2003). Is the socio-spatial distribution of
mound-building mice, Mus spicilegus, compatible with a monogamous mating
system? Mammalia, 67,15e24.
Harris, M. R., & Siefferman, L. (2014). Interspeciﬁc competition inﬂuences ﬁtness
beneﬁts of assortative mating for territorial aggression in Eastern bluebirds
(Sialia sialis). PLoS One, 9, e88668.
von Holst, D. (1998). The concept of stress and its relevance for animal behavior.
Advances in the Study of Behavior, 27,1e131.
Jiang, X., Bolnick, D. I., & Kirkpatrick, M. (2013). Assortative mating in animals.
American Naturalist, 181,E125eE138.
Kight, C. R., David, M., & Dall, S. R. X. (2013). The evolution of animal personality
variation. In H. Kehrer-Sawatzki (Ed.), Encyclopedia of life sciences (eLS).http://
Koolhaas, J. M., Korte, S. M., De Boer, S. F., Van Der Vegt, B. J., Van Reenen, C. G.,
Hopster, H., et al. (1999). Coping styles in animals: current status in behavior
and stress-physiology. Neuroscience and Biobehavioral Reviews, 23, 925e935.
ser, S., Sanguino Mostajo, G. A., Preik, O., Pek
ar, S., & Schneider, J. M. (2013).
Assortative mating by aggressiveness type in orb weaving spiders. Behavioral
Ecology, 24, 824e831.
ser, S., & Schuett, W. (2014). Studying personality variation in invertebrates:
why bother? Animal Behaviour, 91,41e52.
Lafaille, M., & F
eron, C. (2014). U-shaped relationship between ageing and risk-
taking behaviour in a wild-type rodent. Animal Behaviour, 97,45e52.
Lafaille, M., Gouat, P., & F
eron, C. (2015). Efﬁciency of delayed reproduction in Mus
spicilegus.Reproduction, Fertility and Development.http://dx.doi.org/10.1071/
Lessells, C. M., & Boag, P. T.(1987). Unrepeatable repeatabilities: a common mistake.
Lewejohann, L., Zipser, B., & Sachser, N. (2011). “Personality”in laboratory mice used
for biomedical research: a way of understanding variability? Developmental
Psychobiology, 53, 624e630.
Mathot, K. J., van den Hout, P. J., Piersma, T., Kempenaers, B., Reale, D., &
Dingemanse, N. J. (2011). Disentangling the roles of frequency- vs. state-
dependence in generating individual differences in behavioural plasticity.
Ecology Letters, 14, 1254e1262.
Milishnikov, A. N., Raﬁev, A. N., & Muntianu, A. I. (1998). Genotypic variability in
populations of the mound-builder mouse Mus spicilegus Pet., 1882, at various
life cycle stages. Genetika, 34, 9947e9952.
Millesi, E., Huber, S., Everts, L. G., & Dittami, J. P. (1999). Reproductive decisions in
female European ground squirrels: factors affecting reproductive output and
maternal investment. Ethology, 105,163e175.
Nakagawa, K., & Schielzeth, H. (2010). Repeatability for Gaussian and non-Gaussian
data: a practical guide for biologists. Biological Reviews of the Cambridge Phil-
osophical Society, 85,935e956.
Nettle, D. (2006). The evolution of personality variation in humans and other ani-
mals. American Psychologist, 61, 622e631.
van Oers, K., de Jong, G., van Noordwijk, A. J., Kempenaers, B., & Drent, P. J. (2005).
Contribution of genetics to the study of animal personalities: a review of case
studies. Behaviour, 142,1191e1212.
van Oers, K., & Sinn, D. (2011). Towards a basis for the phenotypic gambit: advances
in the evolutionary genetics of animal personality. In A. Weiss, M. Inoue-Mur-
ayama, S. Kawamura, & E. Inoue (Eds.), From genes to behavior (pp. 165e183).
Tokyo, Japan: Springer.
Patris, B., & Baudoin, C. (2000). A comparative study of parental care between two
rodent species: implications for the mating system of the mound-building
mouse Mus spicilegus.Behavioural Processes, 51,35e43.
Patris, B., Gouat, P., Jacquot, C., Christophe, N., & Baudoin, C. (2002). Agonistic and
sociable behaviors in the mound-building mice, Mus spicilegus: a comparative
study with Mus musculus domesticus.Aggressive Behavior, 28,75e84.
Pellow, S., Chopin, P., File, S. E., & Briley, M. (1985). Validation of open:closed arm
entries in an elevated plus-maze as a measure of anxiety in the rat. Journal of
Neuroscience Methods, 14,149e167.
Poteaux, C., Busquet, N., Gouat, P., Katona, K., & Baudoin, C. (2008). Socio-genetic
structure of mound-building mice, Mus spicilegus, in autumn and early spring.
Biological Journal of the Linnean Society, 93, 689e699.
R Core Team. (2014). R: A language and environment for statistical computing. Vienna,
Austria: R Foundation for Statistical Computing. www.R-project.org.
eale, D., Martin, J., Coltman, D. W., Poissant, J., & Festa-Bianchet, M. (2009). Male
personality, life-history strategies and reproductive success in a promiscuous
mammal. Journal of Evolutionary Biology, 22, 1599e1607.
eale, D., Reader, S. M., Sol, D., McDougall, P. T., & Dingemanse, N. J. (2007). Inte-
grating animal temperament within ecology and evolution. Biological Reviews,
Richter, S. H., Garner, J. P., Zipser, B., Lewejohann, L., Sachser, N., Touma, C., et al.
(2011). Effect of population heterogenization on the reproducibility of mouse
behavior: a multi-laboratory study. PLoS One, 6, e16461.
odel, H. G., Bora, A., Kaetzke, P., Khaschei, M., Hutzelmeyer, H. D., Zapka, M., et al.
(2005). Timing of breeding and reproductive performance of female European
rabbits in response to winter temperature and body mass. Canadian Journal of
Zoology, 83, 935e942.
odel, H. G., von Holst, D., & Kraus, C. (20 09). Family legacies: short- and long-term
ﬁtness consequences of early-life conditions in female European rabbits. Journal
of Animal Ecology, 78, 789e797.
odel, H. G., & Meyer, S. (2011). Early development inﬂuences ontogeny of per-
sonality types in young laboratory rats. Developmental Psychobiology, 53,
odel, H. G., Zapka, M., Talke, S., Kornatz, T., Bruchner, B., & Hedler, C. (2015). Sur-
vival costs of fast exploration during juvenile life in a small mammal. Behavioral
Ecology and Sociobiology, 69, 205e2017.
Sapolsky, R. M., Romero, M., & Munck, A. U. (2000). How do glucocorticoids inﬂu-
ence stress responses? Integrating permissive, suppressive, stimulatory, and
preparative actions. Endocrine Reviews, 21,55e89.
Schuett, W., Dall, S. R. X., & Royle, N. J. (2011). Pairs of zebra ﬁnches with similar
‘personalities’make better parents. Animal Behaviour, 81, 609e618.
Schuett, W., Godin, J. G. J., & Dall, S. R. X. (2011). Do female zebra ﬁnches, Taenio-
pygia guttata, choose their mates based on their ‘personality’?Ethology, 10,
Schuett, W., Tregenza, T., & Dall, S. R. X. (2010). Sexual selection and animal
personality. Biological Reviews of the Cambridge Philosophical Society, 85,
Sih, A., Bell, A., & Johnson, J. C. (2004). Behavioral syndromes: an ecological and
evolutionary overview. Trends in Ecology and Evolution, 19,372e378.
Simeonovska-Nikolova, D. M. (2012). Neighbour relationships and spacing behav-
iour of mound-building mouse, Mus spicilegus (Mammalia: Rodentia) in sum-
mer. Acta Zoologica Bulgarica, 64,135e143.
Sinn, D. L., Apiolaza, L. A., & Moltschaniwskyj, N. A. (2006). Heritability and ﬁtness-
related consequences of squid personality traits. Journal of Experimental Biology,
Smith, B. R., & Blumstein, D. T. (2008). Fitness consequences of personality: a meta-
analysis. Behavioral Ecology, 19,448e455.
Stamps, J., & Groothuis, T. G. G. (2010). The development of animal personality:
relevance, concepts and perspectives. Biological Reviews, 85,301e325.
Szenczi, P., B
anszegi, O., Dúcs, A., Gedeon, C. I., Mark
o, G., N
emeth, I., et al. (2011).
Morphology and function of communal mounds of overwintering mound-
building mice (Mus spicilegus). Journal of Mammalogy, 92, 852e860.
Therneau, T. M., & Grambsch, P. M. (2000). Modeling survival data: Extending the Cox
model. New York, NY: Springer.
Ung, D., F
eron, C., Gouat, M., Demouron, S., & Gouat, P. (2014). Maternal energetic
investment in a monogamous mouse. Mammalian Biology, 79,221e224.
Walf, A. A., & Frye, C. A. (2007). The use of the elevated plus maze as an assay of
anxiety-related behavior in rodents. Nature Protocols, 2, 322e328.
Willis-Owen, S. A. G., & Flint, J. (2006). The genetic basis of emotional behaviour in
mice. European Journal of Human Genetics, 14,721e728.
Wilson, A. D. M., Godin, J. G. J., & Ward, A. J. W. (2010). Boldness and reproductive
ﬁtness correlates in the eastern mosquitoﬁsh, Gambusia holbrooki.Ethology, 116,
Wilson, D. S., Clark, A. B., Coleman, K., & Dearstyne, T. (1994). Shyness and boldness
in humans and other animals. Trends in Ecology and Evolution, 9,442e446.
Wolf, M., van Doorn, G. S., & Weissing, F. J. (2008). Evolutionary emergence of
responsive and unresponsive personalities. Proceedings of the National Academy
of Sciences of the United States of America, 105, 15825e15830.
Zuur, A. F., Ieno, E. N., & Elphick, C. S. (2010). A protocol for data exploration to avoid
common statistical problems. Methods in Ecology and Evolution, 1,3e14.
M. Rangassamy et al. / Animal Behaviour 103 (2015) 7e1514
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
Differences in anxiet
scores Differences in ex
Figure A1. Distributions of within-pair differences in (a, c) anxiety and (b, d) exploration scores calculated as the absolute score differences between male and female. Data were
obtained by experiments with two cohorts. (a, b) Cohor t 1: N¼18 pairs. (c, d) Cohort 2: N¼23 pairs. Values of 0 denote maximum similarity in scores between the partners of a pair.
Explained variation of the ﬁrst component of the PCA
Cohort 1 Cohort 2
Time 1 Time 2 Time 1 Time 2
Elevated plus maze 0.743 0.773 0.743 0.509
Open ﬁeld 0.877 0.838 0.817 0.857
PCA was based on measures taken during the elevated plus maze tests and open
Full sets of models (Cox proportional hazards regressions) explaining the timing/
probability of reproduction (R) by means of the birth of a ﬁrst litter in pairs of
nulliparous mound-building mice
Model terms K
Cohort 1 R[Anxiety
] 2 0 0.266
] 2 1.57 0.121
] 3 1.80 0.108
] 3 2.01 0.097
] 3 2.35 0.082
4 2.71 0.069
] 3 2.93 0.061
] 3 3.06 0.058
] 3 4.21 0.032
] 3 4.22 0.032
R[Null model] 1 4.89 0.023
] 2 6.01 0.013
] 2 6.24 0.012
] 2 6.44 0.011
Table A2 (continued )
Model terms K
] 2 6.99 0.008
] 2 7.30 0.007
Cohort 2 R[Anxiety
] 2 0 0.426
] 3 2.47 0.124
R[Null model] 1 3.27 0.083
4 3.56 0.072
] 2 4.36 0.048
] 3 4.36 0.048
] 2 4.85 0.038
] 2 5.25 0.031
] 2 5.44 0.028
] 2 5.58 0.026
] 2 5.65 0.025
] 3 6.79 0.014
] 3 7.26 0.011
] 3 7.50 0.010
] 3 8.10 0.007
] 3 8.11 0.007
The models are ordered by
AICc; all models of the sets are denoted. The onset of
reproduction was monitored during 90 days after pairing (censored data). Predictor
variables considered are the similarity (calculated as the absolute score difference
between males and females) in anxiety or exploration scores within each pair.
Furthermore, distinct effects of score values of males and of females were consid-
ered. In addition, female body mass prior to pairing was included in the construction
of the models. The number of estimable parameters (K),
AICc, and Akaike weights
(w) is given. Analyses are based on data from two cohorts (Cohort 1: N¼18 pairs;
Cohort 2: N¼23 pairs).