Content uploaded by Long Yu
Author content
All content in this area was uploaded by Long Yu on Jul 23, 2021
Content may be subject to copyright.
Aggressive males are more attractive to females and more likely to
win contests in jumping spiders
Bernetta Zi Wei Kwek
a
,
y
, Min Tan
a
,
y
, Long Yu
a
,
b
, Wei Zhou
a
,
Chia-chen Chang
a
,
*
, Daiqin Li
a
,
*
a
Department of Biological Sciences, National University of Singapore, Singapore
b
State Key Laboratory of Biocatalysis and Enzyme Engineering &Centre for Behavioural Ecology and Evolution, School of Life Sciences, Hubei University,
Wuhan, Hubei, China
article info
Article history:
Received 22 December 2020
Initial acceptance 10 March 2021
Final acceptance 21 May 2021
MS. number: 20-00911R
Keywords:
aggression
aggression predictability
personality
Salticidae
sexual selection
Siler semiglaucus
Consistent interindividual differences in behaviour (i.e. personality) and intraindividual variability in
behaviour (higher intraindividual variability means lower behavioural predictability) are common across
animal taxa. However, how personality and behavioural predictability of males and females influence
female mate choice and maleemale competition remains poorly understood. Here, we investigated this
in the jade jumping spider, Siler semiglaucus. After assessing the level of aggression (an individual's
average aggression) and aggression predictability (the variability around average aggression within an
individual) of both S. semiglaucus males and females, we performed female mate choice trials to test
whether aggression and aggression predictability in females, males or both would affect female mate
choice. We also conducted male contest trials to test whether male aggression or aggression predict-
ability would influence the outcomes of male contests. We found that both females and males showed
consistent interindividual differences in aggression, and aggressive spiders were more predictable than
less aggressive ones. Despite a positive correlation between aggression and predictability, male
aggression predicted female mate choice better than aggression predictability. Females showed a
directional preference for aggressive males over docile males regardless of female aggression or male
aggression predictability. Predictable aggressive males were also more likely to win contests. Our results
suggest that both female mate choice and maleemale competition favour males with high aggression,
and thus total sexual selection that acts on male aggression may be reinforcing. These findings also
highlight that male S. semiglaucus with a higher level of aggression may have better reproductive
performance.
©2021 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.
It is well established that animals show consistent interindi-
vidual variation in behaviour across time or contexts (i.e. person-
ality). For example, some individuals tend to be more aggressive,
bolder or more exploratory than other individuals. The repeat-
ability of a behavioural trait in a population can be estimated as the
variance between individuals divided by the total variance (sum of
between-individual variance and within-individual variance).
Higher repeatability indicates more consistent differences across
individuals. Repeatability has been used as the theoretical upper
bound of heritability in a population (Boake, 1989;Dohm, 2002).
However, high repeatability does not mean that individuals always
show the same behavioural response under the same situations. An
unexplained variation within an individual across time is the
intraindividual variability (IIV) of behaviour (i.e. behavioural pre-
dictability; Stamps, Briffa, &Biro, 2012;Biro &Adriaenssens, 2013;
Cleasby, Nakagawa, &Schielzeth, 2015). A lower IIV value indicates
higher predictability. Within a population, some individuals are
more predictable than others. Selection pressure on a behavioural
type (e.g. being aggressive or being docile) and behavioural pre-
dictability (e.g. being predictable or unpredictable in aggression)
has been found to have fitness consequences via natural selection
(Bengston et al., 2018;Briffa, 2013;Dingemanse, Kazem, R
eale, &
Wright, 2010;Dingemanse &R
eale, 2005;Moiron, Laskowski, &
Niemel€
a, 2020;R
eale, Reader, Sol, McDougall, &Dingemanse,
2007;Sih, Bell, &Johnson, 2004;Smith &Blumstein, 2008;
Stamps et al., 2012). Fitness consequences of personality and pre-
dictability can also come from sexual selection (Munson, Jones,
*Corresponding authors.
E-mail addresses: dbschcc@nus.edu.sg (C. Chang), dbslidq@nus.edu.sg (D. Li).
y
These authors contributed equally to this work.
Contents lists available at ScienceDirect
Animal Behaviour
journal homepage: www.elsevier.com/locate/anbehav
https://doi.org/10.1016/j.anbehav.2021.06.030
0003-3472/©2021 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.
Animal Behaviour 179 (2021) 51e63
Schraft, &Sih, 2020;Schuett, Tregenza, &Dall, 2010). However,
empirical studies have primarily focused on natural selection on
personality; the limited empirical research hampers our under-
standing of the role of sexual selection, such as via female mate
choice and maleemale competition, in the evolution of personality
and behavioural predictability.
In the context of female mate choice, personality type and
behavioural predictability of an individual may influence whether a
male would be preferred by a female in a population (Munson et al.,
2020;Schuett et al., 2010). On the one hand, personality- or
predictability-based mate choices can be directional (i.e. the
directional mate preference hypothesis; Scherer, Kuhnhardt, &
Schuett, 2018;Scherer &Schuett, 2018). This hypothesis posits
that the preferred behavioural type of a male is consistent across
females regardless of female behavioural type. For example, field
cricket, Gryllus assimilis, females preferred more aggressive males
that were generally winners in agonistic maleemale interactions
(Loranger &Bertram, 2016). Most Siamese fighting fish, Betta
splendens, females preferred aggressive males, and aggressive
males also tended to have a higher chance of winning against a nest
intruder and better defend their offspring (Jaroensutasinee &
Jaroensutasinee, 2003). Therefore, females may use a male's
behavioural type or behavioural predictability to predict its quality
or potential future parental care (Jaroensutasinee &
Jaroensutasinee, 2003;Royle, Schuett, &Dall, 2010).
On the other hand, mating preference based on personality- or
predictability can also be (dis)assortative (i.e. the mate compati-
bility hypothesis, Kralj-Fi
ser, Sanguino Mostajo, Preik, Pek
ar, &
Schneider, 2013;Laubu, Schweitzer, Motreuil, Lou^
apre, &
Dechaume-Moncharmont, 2017). The mate compatibility hypoth-
esis posits that mate preference depends on the personality type
(or behavioural predictability) of both males and females. For
instance, in the zebra finch, Taeniopygia guttata, intermediate and
highly exploratory females showed an assortative preference for
highly exploratory males (assortative preference; Schuett, Godin, &
Dall, 2011). In rainbow kribs, Pelvicachromis pulcher, bold females
preferred docile males, showing a disassortative mate preference,
but females preferred males with a similar level of boldness pre-
dictability to themselves, indicating an assortative mate preference
(Scherer, Kuhnhardt, &Schuett, 2017).
In addition to mate choice, maleemale competition is another
important aspect of sexual selection. In maleemale competition,
individuals can exhibit various forms of agonistic interactions to
signal their resource-holding potential (RHP). RHP is a measure of
an individual's fighting ability as well as the fighting effort exerted
during a contest (Arnott &Elwood, 2009;Lane &Briffa, 2020). Thus,
it plays a crucial role in determining the outcome of a male contest
in which individuals can assess their own and/or their opponent's
RHP (Taylor &Elwood, 2003). Various traits that are used to assess
RHP include, among others, body size (Songvorawit, Butcher, &
Chaisuekul, 2018), weaponry (Jennings, Gammell, Carlin, &
Hayden, 2004;Yasuda &Koga, 2016) and body condition
(Bjornson &Anderson, 2018;Moretz, 2003). An individual's level of
aggression can also signal RHP (Briffa, 2014). For example, more
aggressive males (with a higher RHP) are more likely to win male
contests (Briffa, Sneddon, &Wilson, 2015;Moretz, 2005). Male
aggression predictability can also influence male contests (Briffa
et al., 2015), yet experimental tests of this hypothesis are scarce.
More importantly, female mate choice and maleemale compe-
tition rarely operate in isolation (e.g. Berglund, Bisazza, &Pilastro,
1996;Hunt, Breuker, Sadowski, &Moore, 2009;Wong &
Candolin, 2005). Previous studies have investigated how sexual
selection acts on personality traits either via female mate choice or
via maleemale competition (Munson et al., 2020;Schuett et al.,
2010), yet few have tested how selection imposed by both
mechanisms acts on personality traits. As a result, whether both
mechanisms favour personality/behavioural predictability in the
same direction remains largely unknown. For example, high
aggression levels in males may be favoured in both female mate
choice and maleemale competition (reinforcing); alternatively,
while males with high aggression levels may benefit in maleemale
competition, docile males can be favoured in female mate choice
(opposing; Hunt et al., 2009). The total selection pressure from two
mechanisms would influence the strength and form of sexual
pressure on animal personality and behavioural predictability.
Thus, it is important to explore how both female mate choice and
maleemale competition act on personality and/or behavioural
predictability in the same species.
Another research gap is that prior studies on personality and
behavioural predictability have primarily focused on vertebrates
(Kalb, Lindstr€
om, Sprenger, Anthes, &Heubel, 2016;Munson et al.,
2020;Neumann, Agil, Widdig, &Engelhardt, 2013;Schuett et al.,
2010). Fewer studies have been conducted to investigate the ef-
fects of personality and behavioural predictability on female mate
choice and maleemale competition in invertebrates, including
spiders (Kralj-Fi
ser &Schuett, 2014). In the bridge orb-weaving
spider, Larinioides sclopetarius, aggressive males preferred mating
with aggressive females, showing an assortative mating preference
(Kralj-Fi
ser et al., 2013). Although sexual selection has been well
studied in spiders, how both the personality and behavioural pre-
dictability of males, females or both may influence sexual selection
remains an open question.
In this study, we investigated whether the aggression and
aggression predictability in either or both sexes influence female
mate choice and maleemale competition in the jade jumping spi-
der, Siler semiglaucus (Araneae: Salticidae). Jumping spiders, the
largest spider family (World Spider Catalog, 2021), are highly
diverse and have a unique visual system coupled with an elaborate
vision-mediated behaviour (Harland, Li, &Jackson, 2012;Land &
Nilsson, 2012). Their amenability to experimentation makes them
a highly suitable model system for personality studies (Royaut
e,
Buddle, &Vincent, 2014,2015;Chang, Connahs, Tan, Norma-
Rashid, Mrinalini, Li, &Chew, 2020). Siler semiglaucus can be
found in tropical and subtropical Asia (Grob, 2015;World Spider
Catalog, 2021). They are often found on shrubs in open habitats,
with a clumped distribution (D. Li, personal observations). This
suggests that, in this species, female mate choice and male contests
are more likely to occur simultaneously. Like most salticids,
S. semiglaucus is a visual specialist, with an acute vision that allows
for elaborate visual displays during courtship and agonistic in-
teractions (Zeng, 2019). However, whether there is consistent inter-
and intraindividual variation in aggression and, if so, how sexual
selection acts on this variation in S. semiglaucus are unknown.
We first quantified the level of aggression in S. semiglaucus
spiders using a mirror test with five trials for each spider. The
(average) level of aggression (hereafter aggression), as well as the
predictability of aggression expression (high predictability in-
dicates less variability in aggression within an individual across
the five trials; hereafter aggression predictability), was estimated.
Then, we investigated the influence of the aggression and
aggression predictability of both males and females on female
mate choice. Finally, we examined the effects of male aggression
and aggression predictability on the outcome of maleemale
competition. We hypothesized that aggression and aggression
predictability in S. semiglaucus females, males or both would in-
fluence female mate choice. Females would show a directional
preference for males with high or low aggression and/or aggres-
sion predictability. Alternatively, females may prefer males that
were similar (i.e. assortative) or dissimilar (i.e. disassortative) in
terms of aggression and/or aggression predictability. For male
B. Z. W. Kwek et al. / Animal Behaviour 179 (2021) 51e6352
contests, we hypothesized that the aggression or aggression pre-
dictability of S. semiglaucus males would influence the outcomes
of the contests.
METHODS
Spider Collection and Maintenance
We collected juvenile and subadult S. semiglaucus in two pop-
ulations: China (Hainan Province) in August 2019 (N¼19) and
Singapore from October 2019 to August 2020 (N¼72). We kept all
spiders individually in cylindrical containers (diameter: 6.5 cm;
height: 8 cm) and housed them in the laboratory under controlled
environmental conditions (80 ±5% relative humidity, at approxi-
mately 25 ±1
C, 12:12 h light:dark cycle, lights on at 0800)
following the protocols used in the maintenance of typical salticids
(Lim, Land, &Li, 2007;Zeng, Wee, Painting, Zhang, &Li, 2019).
Spiders were exposed to ultraviolet (UV) light provided by full-
spectrum fluorescence tubes (Arcadia Natural Sunlight Lamp,
Croydon, U.K.) for 4 h every day (1000e1200 during the day;
160 0e1800 in the evening) to mimic the light conditions in the
wild. We fed each spider five to seven laboratory-cultured fruit
flies, Drosophila melanogaster, twice a week. We monitored the
juvenile and subadult spiders until they reached adulthood and
recorded the date of their final moult to determine their post-
maturation age. In total, 42 adult females and 49 adult males were
used in this study.
Ethical Note
These experiments were performed in accordance with all
relevant Singapore laws and the ASAB/ABS Guidelines for the use of
animals in research, and under the Institutional Biosafety Com-
mittee (IBC) of the National University of Singapore (Ref. No: 2018-
00115).
Body Size Measurements
Once the spiders reached maturation, we measured their body
size before the experiments. We first anaesthetized them with
carbon dioxide for 1 min for easy positioning and then took dorsal
photographs of them using a Samsung Galaxy A7 (2018) camera
and a digital SLR camera (Nikon D800) with flash (Sigma
EM140DG) and lens (Micro Nikkor 105 mm 1:2.8G). We then
analysed the photographs using ImageJ (https://imagej.nih.gov)to
measure the total body length (from the anterior tip of the
cephalothorax to the posterior tip of the abdomen excluding
spinnerets) and carapace width to the nearest 0.01 mm. We also
measured their body mass to the nearest 0.01 mg using an elec-
tronic balance (KERN ABT 120e5DM, Germany) before the
behavioural trials.
General Trial Conditions
We conducted all behavioural trials (aggression/aggression
predictability assessment, female mate choice and male contest) in
a dark room under full spectral light provided by six full spectral
light tubes (Hitachi BL/B, 20W, Tokyo, Japan). UV light was able to
pass through the full-spectrum glass covering the set-ups. This is
because both S. semiglaucus males and females have UV reflectance
and use it for mate choice and male contests (Zeng, 2019).
Assessment of Aggression and Aggression Predictability
We assessed the aggression and aggression predictability of
both S. semiglaucus males (N¼49) and females (N¼42) using a
mirror-image method modified from the procedure described in
Chang et al. (2020). Jumping spiders often participate in compli-
cated escalating contests when encountering conspecifics and in-
crease their level of escalation by displaying more intense agonistic
behaviour when approaching one another (Elias, Kasumovic,
Punzalan, Andrade, &Mason, 2008;McGinley, Prenter, &Taylor,
2015). Both male and female jumping spiders are known to
display same-sex aggressive behaviours (Clark, Jackson, &Waas,
1999;Elias et al., 2008), although femaleefemale aggressive
behaviour tends to be less intense than maleemale aggressive
behaviour (Jackson, Walker, Pollard, &Cross, 2006). Thus, the dis-
tance to the mirror image can be used as a good proxy for the level
of aggression and using the mirror image as an ‘opponent’can
minimize any potential factors associated with the differences in
body size or aggression of live conspecifics which could affect the
quantification of aggression (Royaut
e et al., 2014). Before the mirror
image stimulation test, we placed the spider into a container
(10 5 cm and 4 cm high) made of full-spectrum-transmitting
glass with a mirror (5 5 cm) attached to the set-up (Fig. 1a).
The spider was allowed to acclimatize for 5 min, during which an
opaque barrier covered the mirror. Upon removing the barrier, the
spider was allowed to interact with its mirror image until it stopped
displaying/approaching the mirror or for a maximum of 20 min,
whichever came first. The shortest distance (shown in Fig. 1a)
10 cm
5 cm
Mirror
5 cm
5 cm
Female
Transparent wall
Distance
5 cm
18 cm
(a)
(b)
(c)
5 cm
4 cm
5 cm
Male Male
5 cm
Image Female
Male Male
Opaque wall
Figure 1. Diagrams of the set-ups used in (a) aggression assessment of S. semiglaucus,
(b) female mate choice trials and (c) male contest trials. Diagrams not to scale.
B. Z. W. Kwek et al. / Animal Behaviour 179 (2021) 51e63 53
between the spider and its mirror image when it stopped dis-
playing or approaching the mirror was recorded: a shorter distance
indicates a higher level of aggression. To estimate repeatability and
predictability in aggression, each individual was tested five times
with an interval of 2e3 days in between trials. All tests were con-
ducted at least 24 h after spiders had been fed to control for the
effect of feeding on aggression. For the estimation of aggression
predictability, we excluded five males and three females that al-
ways showed the minimum (0 cm) or maximum (10 cm) distance
from the mirror across the trials. They were excluded because these
spiders showed zero variation within an individual (i.e. behaving
completely the same across the five trials), possibly due to con-
straints in the experimental set-up (e.g. aggression measurements
are bounded by the length of the set-up). We performed additional
analyses using the full data set containing these ‘extreme’eight
spiders and obtained similar results in aggression behaviour, fe-
male mate preference and male contests (for more details, see
Appendix Tables A1eA3).
Female Mate Choice
To investigate whether the aggression and aggression predict-
ability of both S. semiglaucus males and females would affect a
female's mate preference, we conducted a series of mate choice
trials by giving a female the choice between a relatively docile
male and an aggressive male. We matched the two males for each
trial on carapace width (with a maximum difference of 0.1 mm),
colour pattern and postmaturation age to minimize the effects of
these on female mate preference. The mate choice apparatus
consisted of three chambers: two male chambers (5 5 cm and
2.5 cm high) separated by an opaque wall and one female chamber
(10 5 cm and 2.5 cm high) separated by a transparent wall and
an opaque barrier (Fig. 1b). We covered the set-up with a full-
spectrum-transmitting glass and put in an opaque barrier to
separate the spiders during the 5 min acclimatization period. We
started a trial upon the removal of the opaque barrier when both
males were located at a similar distance to the female. Each trial
lasted for 10 min. The behaviour of the spiders was observed and
video recorded with a phone camera (Samsung Galaxy A7 [2018];
5.0 MP; 1080p@30fps). We played back the videos and recorded
the total female attention time (i.e. the time (s) a female spent
watching a particular male performing courtship display), which
acted as a proxy for female mate preference (Elias, Hebets, &Hoy,
2006;Hebets &Maddison, 2005). We defined the preferred male
as the one the female spent a longer time watching (i.e. longer
female attention). We also recorded the number of courtship
displays (e.g. body shakes, leg waves, body movements; Byers,
Hebets, &Podos, 2010) and the duration (s) of total courtship
display for each male. We conducted a total of 36 mate choice
trials (two trials using the ‘extreme’spiders were removed during
data analysis, see above). No male pairs or individual females were
reused.
Male Contests
To determine how male aggression and aggression predictability
affect theoutcome of male contests, we conducted a series of contests
between two S. semiglaucus males in a contest arena (Fig. 1c). This
arena was made from a rectangular plastic container (18 5cmand
4.5 cm high) and was divided half by an opaque divider and covered
with a sheet of full-spectrum-transmitting glass. Before the trial, we
colour-coded the males by applying acrylic paint to the top of their
carapaceto allow for individual identification.Males were not marked
on the surface of the dorsal abdomen, as the colour patterns
observed might have been used as signals to conspecifics during the
experimental trials(Hsiung et al., 2017;Lim &Li, 2007). We paired the
two males chosen for each trial by matching them on body size (i.e.
carapace width with a maximum difference of 0.1 mm), colour
pattern and postmaturation age to control for these confounders on
the male contests. We then introduced the two males to each half of
the arena and allowed them to acclimatize for 5 min. We began the
trial upon removal of the opaque divider and ended it when a male
retreated or 5 min elapsed, whichever came first. We performed a
total of 46 male contests (two trials using the ‘extreme’spiders were
removed during data analysis,see above) using 41male spiders. Some
spiders participated in more than one trial (i.e. of 41 males, 11 were
used once, nine were used twice, 21 were used three time s). However,
no pair was used in more than one trial.The behaviours of the males
were video recorded. We played back the videos and recorded the
winner and loser for each contest (i.e. based on whichever individual
‘decamped’first; Lim &Li, 2004). The prior experience with fighting
gained by the males that were reused (i.e. the contest outcome of an
individual's previous contest; Kasumovic, Elias, Punzalan, Mason, &
Andrade, 2009) was recorded to control for the effect of experience
in the subsequent statistical analyses.
Data Analysis
Aggression and aggression predictability
We used a double hierarchical generalized linear modelling
approach (DHGLMs, Cleasby et al., 2015) from the brms package
(Bürkner, 2017) to model the between-individual variance (i.e. mean
aggression) and the residual variance (i.e. predictability) in aggres-
sion simultaneously (Mitchell, Fanson, Beckmann, &Biro, 2016). To
model aggression, we included the trial number, sex and carapace
width as fixed effects and spider identity as a random effect.
Collection site (China or Singapore) was not included as a random
effect because no variance was captured by collection site (lower
bound of 95% CI close to zero; Appendix Table A4). We also included
a residual variance model (sigma) which was partitioned for each
individual (spider identity included as a random effect). We ran
5000 iterations for each of the four Markov chains with a thinning
rate of 2. We set inits as ‘random’to generate initial values for the
parameters randomly.We discarded the initial 200 0 iterations as the
model converged during the burn-in period. We then extracted each
individual's mean posterior distribution in intercept and added to
the population level of intercept and coefficient of sex (i.e. individual
average aggression). We also extracted each individual's mean
posterior distribution in residual variance (sigma intercept) and
added to the population level of residual variance (i.e. individual IIV
of aggression). The aggression and IIV values were subsequently
used for analyses on female mate choice and male contests.
We calculated the repeatability of aggression by dividing the
between-individuals variance by the total variance (sum of the
between-individual, between-population and residuals [i.e. within-
individual] variances) after accounting for the fixed effects
(Dingemanse &Dochtermann, 2013). We also extracted the 95%
confidence interval (CI) of the repeatability using the highest pos-
terior density intervals from the coda package (Plummer, Best,
Cowles, &Vines, 2006) to assess the level of repeatability of the
aggression (Wolak, Fairbairn, &Paulsen, 2012). Repeatability was
considered as ‘significant’when the 95% CI did not cross zero.
We also conducted two separate pairwise Spearman rank cor-
relations to determine whether there were correlations between
carapace width, body mass, aggression and IIV for females and
males, respectively.
Female mate choice
We used generalized linear models with quasibinomial error
structure due to the overdispersion to investigate the effects of both
B. Z. W. Kwek et al. / Animal Behaviour 179 (2021) 51e6354
male and female aggression and aggression predictability on fe-
male mate choice. We performed the models with female choice for
the focal male (preferred ¼1; not preferred ¼0) as the response
variable. We used the binary variable (1/0) because eventually only
one of the males would mate with the female. We used an
information-theoretic approach (Grueber, Nakagawa, Laws, &
Jamieson, 2011) by proposing a series of hypothesis-driven
models. We included female aggression, female aggression pre-
dictability, female carapace width, the difference in male aggres-
sion, the difference in male aggression predictability and the
difference in male carapace width as continuous predictors. We
proposed a total of 18 models (Appendix Table A5), which
comprised (1) each behavioural type of either female or paired-
male difference alone (i.e. female aggression, female aggression
predictability, difference in male aggression and difference in male
aggression predictability; four models), (2) the combination of
aggression and aggression predictability of both sexes (eight
models; with and without interactions) to test whether considering
both the male and female behavioural types predicted better than
each behavioural type alone, (3) carapace width of each and both
sexes, and its combination with the opposite sex's aggression (five
models) and (4) a null model (one model). The same-sex aggression
and aggression predictability were not included in the same model
to avoid the problem of multicollinearity because of their high
correlation (Fig. 2). We then compared models using the quasi-
likelihood selection QAIC in the package MuMIn (Barto
n, 2009),
Carapace width
0
2
4
6
1 1.4 1.8
6
10
14
0246
0.13
Aggression
–0.29
–0.77
***
Aggression IIV
–0.4 0 0.4 0.8
81216
1
1.4
1.8
–0.07
0.09
–0.4
0.2
0.8
0.03
Mass
(b)
Carapace width
0
2
4
6
1.1 1.3 1.5
4
8
12
0246
–0.27
Aggression
0.26
–0.41
**
Aggression IIV
-0.4 0 0.4
46810
1.1
1.3
–0.02
0.14
–0.4
0
0.014
Mass
(a)
1.5
0.4
Figure 2. Spearman rank correlation matrix plot showing the correlations between aggression, intraindividual variation in aggression (IIV), carapace width and body mass in (a)
male and (b) female S. semiglaucus.**P<0.01; ***P<0.001. Males: N¼44; females: N¼39.
B. Z. W. Kwek et al. / Animal Behaviour 179 (2021) 51e63 55
and the model with the smallest QAIC was selected as the best-
fitting model. The QAIC difference for each model (delta QAIC)
was computed with the best fitting model as the reference for all
candidate models. Top models with delta QAIC <2 were included
for model averaging using the model.avg function in the package
MuMIn.
We also carried out additional analysis using linear regressions
with female attention time (continuous variable) as the response
variable following the same procedures; the results are shown in
Appendix Table A6. The list of 18 models and model comparison is
presented in Appendix Table A7.
Male contest
To determine the effects of male aggression and aggression
predictability on the outcomes of male contests, we ran generalized
linear models with a quasibinomial error structure due to over-
dispersion. We also used the information-theoretic approach
(Grueber et al., 2011) to predict the outcome of male contests with
the focal male's outcome (winner ¼1, loser ¼0) as the response
variable. We included prior experience with fights (naïve, win,
lose), and differences in male aggression, difference in male
aggression predictability and difference in male carapace width
between two males as the predictors. Prior experience with fights
and carapace width were included as predictors to account for the
winnereloser effect (Harrison, Jennions, &Head, 2018) and po-
tential RHP of the males (Songvorawit et al., 2018), respectively. We
proposed a total of 15 models comprising the carapace width,
experience, behavioural predictors alone (four models), all pairwise
combinations except for aggression difference and aggression
predictability difference (10 models, with and without interaction
terms) and the null model (one model). The difference in male
aggression and difference in male aggression predictability were
not included in the same model to avoid the issue of multi-
collinearity because of the high correlation between aggression and
aggression predictability (Fig. 2). The list of 15 models and com-
parison is shown in Appendix Table A8. Similar to mate choice
analysis, we conducted model comparisons and selections, and
model averaging if delta QAIC <2.
We performed all data analyses using R version 4.0.0 (R Core
Team, 2020).
RESULTS
Aggression and Aggression Predictability
Siler semiglaucus showed consistent interindividual variation in
aggression when both males and females were combined (N¼83,
repeatability ¼0.624, 95% CI ¼[0.539, 0.718]). Males (N¼44,
repeatability ¼0.636, 95% CI ¼[0.514, 0.761]) showed a similar
level of repeatability as females (N¼39, repeatability ¼0.611, 95%
CI ¼[0.471, 0.742]).
Males were more aggressive than females, and spiders became
slightly more aggressive with increasing trial number (Table 1).
However, the carapace width of spiders was not associated with
aggression (Table 1). There was a clear individual variation in IIV of
aggression (SD of residual variance between spider ID ¼0.43, 95%
CI ¼[0.30, 0.56]; Table 1,Fig. 2).
Aggression was negatively correlated with aggression IIV for
both male and female S. semiglaucus (Fig. 2). This indicates that
aggressive spiders were more predictable than docile spiders.
Carapace width and body mass had no significant correlation
with aggression or aggression IIV in either males or females
(Fig. 2).
Effects on Female Mate Choice
The model including male aggression difference alone was the
best model (weight ¼0.368), followed by the model including male
aggression difference and female aggression, and the model
including male aggression difference and female carapace width
(Appendix Table A5, model averaging summary in Table 2). Male
aggression difference was the only statistically significant predictor
associated with female mate choice after model averaging (Table 2).
This indicates that females preferred more aggressive males (Fig. 3,
Table 2).
Table 1
Double hierarchical generalized linear mixed-effect model (DHGLM) showing the effects of the number of trials, sex and carapace width on the aggression of S. semiglaucus
Estimate SE Lower 95% CI Upper 95% CI
Group level effects
~ID sd(Intercept) 1.72 0.16 1.42 2.06
sd(sigma_Intercept) 0.43 0.07 0.30 0.56
Population level effects Intercept 3.41 2.35 1.13 8.07
sigma_Intercept 0.17 0.07 0.04 0.30
Trial number 0.10 0.04 0.02 0.18
Sex (Male) 1.02 0.46 0.09 1.91
Carapace width 2.46 1.68 0.89 5.69
Males: N¼44; females: N¼39. CI: confidence interval.
Table 2
Fully and conditionally averaged models from the top models to predict female mate choice
Model Predictor Estimate SE Z P
Full average Intercept 1.43 2.08
Male aggression difference 1.29 0.38 3.25 0.001
Female aggression 0.04 0.20 0.19 0.846
Female IIV 0.08 0.62 0.12 0.904
Female carapace width 0.12 1.28 0.09 0.930
Conditional average Intercept 1.43 2.09
Male aggression difference 1.29 0.38 3.25 0.001
Female aggression 0.23 0.43 0.52 0.604
Female IIV 0.48 1.48 0.32 0.753
Female carapace width 0.74 3.17 0.23 0.821
Number of trials: 34 (two trials using ‘extreme’spiders were excluded). IIV: intraindividual variation (predictability).
B. Z. W. Kwek et al. / Animal Behaviour 179 (2021) 51e6356
Effects on Male Contests
The model with male aggression difference alone was the best
model (weight ¼0.371), followed by the model with male aggres-
sion difference and male carapace width difference and the model
with male IIV aggression difference alone (Appendix Table A8,
averaged model summary in Table 3). More aggressive or more
predictable (low IIV) males were more likely to win the contests
(Fig. 3,Table 3).
DISCUSSION
In this study, we found that, in S. semiglaucus, both males and
females showed consistent individual difference in aggression level
and aggression predictability, and more aggressive spiders were
more predictable. This was consistent with several studies in birds
and fishes showing that more proactive animals (more aggressive,
more risk-taking and more exploratory) were more predictable and
less plastic/flexible (Mitchell et al., 2016;Scherer et al., 2018;Thys
et al., 2017). Importantly, we found that male aggression is an
important predictor of the outcomes of both female mate choice
and maleemale competition. Specifically, females preferred more
aggressive males, and this preference was not dependent on female
aggression level or aggression predictability, thus supporting the
hypothesis that females showed a directional but not (dis)assor-
tative preference for more aggressive males. In addition, highly
predictable aggressive males were more likely to win contests than
less predictable docile males. Taken together, our results show that,
in the jade jumping spider S. semiglaucus, sexual selection imposed
by both female mate choice and maleemale competition acts on
–4 –2 0 2 4 6 8
0
0.2
0.4
0.6
0.8
1
Female preference
(a)
–6–4–202468
Difference in male a
gg
ression
Male contest outcomes
(c)
0
0.2
0.4
0.6
0.8
1
–0.5 0 0.5 1
(b)
0
0.2
0.4
0.6
0.8
1
–0.5 0 0.5
Difference in male a
gg
ression IIV
(d)
0
0.2
0.4
0.6
0.8
1
Figure 3. The relationships between (a) difference in aggression between two males and probability of the males being preferred by females, (b) difference in intraindividual
variation in aggression (IIV) between two males and probability of the males being preferred by females, (c) difference in aggression between two males in a contest and probability
of the males winning and (d) difference in IIV between two males in a contest and probability of the males winning. Number of female mate choice trials: 34 (two trials using
‘extreme’spiders were excluded). Number of male contest trials: 44.
Table 3
Fully and conditionally averaged models from the top models to predict the outcomes of male contests
Model Predictor Estimate SE Z P
Full average Intercept 0.003 0.37
Male aggression difference 0.43 0.26 1.67 0.095
Male carapace width difference 1.56 4.33 0.35 0.725
Male IIV difference 0.46 1.15 0.40 0.688
Conditional average Intercept 0.003 0.37
Male aggression difference 0.52 0.19 2.68 0.007
Male carapace width difference 5.79 6.74 0.84 0.404
Male IIV difference 2.88 1.12 2.50 0.013
Number of trials: 44 (two trials using ‘extreme’spiders were excluded). IIV: intraindividual variation (predictability).
B. Z. W. Kwek et al. / Animal Behaviour 179 (2021) 51e63 57
male aggression in the same direction. Hence, this selection ap-
pears to be reinforcing a male's level of aggression.
Our results demonstrate that females showed a general prefer-
ence for more aggressive males, and this preference was not
influenced by female aggression level or aggression predictability.
Consistently, previous studies have also found that aggressive
males tend to have higher fecundity and reproductive success
(reviewed in Schuett et al., 2010;Munson et al., 2020). The
aggression level of males may signal their quality (Schuett et al.,
2010) and the benefits that males may provide (Munson et al.,
2020;Royle et al., 2010). For instance, more aggressive males are
better at acquiring limited resources under competitive conditions,
as aggression may increase their chances of accessing rare re-
sources in resource-deprived environments (Manenti, Pennati, &
Ficetola, 2015;Wilson, Grimmer, &Rosenthal, 2013).
Indeed, we found that more aggressive S. semiglaucus males
were more likely to win a contest, suggesting that aggressive males
are better fighters. This is also consistent with previous studies in
which aggressive males often win maleemale contests (Brown,
Smith, Moskalik, &Gabriel, 2006;Garnham, Porth
en, Child,
Forslind, &Løvlie, 2019). For example, in crickets, G. assimilis,
winners of male contests were more aggressive than losers with
similar body size (Bertram, Rook, Fitzsimmons, &Fitzsimmons,
2011). These empirical results suggest that aggression can be a
distinctive trait or measure of RHP in males (Briffa et al., 2015;
Hurd, 2006). Although body condition is a common indicator of
RHP, we found that the residuals from a regression of body mass on
carapace width (a commonly used proxy of body condition) of male
S. semiglaucus had no significant correlation with male aggression
or aggression predictability (Appendix Table A9). Therefore, apart
from body condition, aggression may be a measure of an in-
dividual's internal motivation to initiate and win a contest during
its assessment of RHP (Camerlink, Turner, Farish, &Arnott, 2015;
Favati, Løvlie, &Leimar, 2017).
However, how female S. semiglaucus evaluate the aggression
level of a male without observing maleemale interactions remains
unknown. One possibility is that they do so through the male's
courtship behaviour. Previous studies have shown the positive
relationship between aggression and courtship (Cornuau,
Schmeller, Courtois, Jolly, &Loyau, 2015), and we also detected a
positive, but weak, correlation between courtship duration and
aggression level of males (see Appendix and Appendix Table A10).
Future studies should use a finer scale of the courtship behaviour. It
is also possible that females use other male traits that are correlated
with male aggression (Kern, Robinson, Gass, Godwin, &
Langerhans, 2016;Tognetti, Ganem, Raymond, &Faurie, 2018),
such as colour patterns (Pauers, Kapfer, Doehler, Lee, &Berg, 2012),
so that they indirectly select for aggressive males. These potential
mechanisms require further studies.
In our study, male aggression was positively correlated with
male aggression predictability; therefore, it is difficult to tease apart
whether being aggressive or being predictable is more likely to win
a contest and be preferred by females. There may be a positive
feedback loop where aggressive males tend to have more resources
(e.g. via food intake) or rewards (e.g. positive previous experience),
which reinforces them to behave aggressively in the future, leading
to high predictability in aggression. For example, in a one-time
encounter between two males, the relatively more aggressive
male may have slightly more energy or higher motivation and then
win the contest. The experience gained from winning the contest
may hence encourage it to continue behaving aggressively to win
future contests. This reinforcement suggests that the opponent can
accurately assess the aggression (and the predictability in aggres-
sion) from just one observation (e.g. in a 5 min contest). Future
studies could replicate this experiment in other study systems to
better understand the correlation between male aggression and
predictability in aggression.
The strength of this study lies in providing a comprehensive pic-
ture of sexual selection on aggressive behaviour, combining two
components of behavioural variation, interindividual (personality)
and intraindividual (behavioural predictability) variation, and two
sexual selection mechanisms, female mate choice and maleemale
competition. Our results show that, in S. semiglaucus, both female
mate choice and maleemale competition may occur simultaneously;
thus, the selection imposed by these two mechanisms on male
aggression may be directionally reinforced (Hunt et al., 2009).
Conclusions
This study emphasizes the importance of aggression and
aggression predictability on female mate choice and/or maleemale
competition in the jumping spider S. semiglaucus. Our findings
demonstrate male aggression may bring fitness consequences via
directional female mate choice and maleemale competition. As
behavioural studies focusing on personality and behavioural pre-
dictability are still in their early stages, future research should
investigate how personality and behavioural predictability are
shaped by total sexual selection imposed by both mechanisms in
other animal species.
Author Contributions
D.L., B.Z.W.K., M.T., L.Y. and C.C.C. conceived and designed the
study. B.Z.W.K. conducted field and lab work. D.L., B.Z.W.K., L.Y. and
C.C.C. analysed the data. All authors contributed to the writing and
editing of the manuscript. All authors gave final approval for
publication.
Data Availability
Behavioural data (in .csv format) and R code for data analyses
and figures associated with this article are available in Mendeley
Data (doi: 10.17632/92d7yc7sv3.1).
Acknowledgments
This work was supported by Singapore Ministry of Education
AcRF Tier 1 grants (R-154-000-B18-112 &R-154-000-B72-114) and
by a grant from the National Natural Science Foundation of China
(NSFC-31872229). We thank Boon Hui Wong and Yueying Lee for
rendering assistance in acquiring materials needed for the study,
and Lu Wee Tan for overseeing the maintenance of the insectary.
References
Arnott, G., & Elwood, R. W. (2009). Assessment of fighting ability in animal contests.
Animal Behaviour, 77(5), 991e1004.https://doi.org/10.1016/j.anbehav.2009.02.010
Barto
n, K. (2009). MuMIn: Multi-model inference.http://r-forge.r-project.org/
projects/mumin/.
Bengston, S. E., Dahan, R. A., Donaldson, Z., Phelps, S. M., van Oers, K., Sih, A., et al.
(2018). Genomic tools for behavioural ecologists to understand repeatable in-
dividual differences in behaviour. Nature Ecology &Evolution, 2(6), 944e955.
https://doi.org/10.1038/s41559-017-0411-4
Berglund, A., Bisazza, A., & Pilastro, A. (1996). Armaments and ornaments: An
evolutionary explanation of traits of dual utility. Biological Journal of the Linnean
Society, 58(4), 385e399. https://doi.org/10.1006/bijl.1996.0043
Bertram, S. M., Rook, V. L. M., Fitzsimmons, J. M., & Fitzsimmons, L. P. (2011). Fine-
and broad-scale approaches to understanding the evolution of aggression in
crickets. Ethology, 117(12), 1067e1080. https://doi.org/10.1111/j.1439-
0310.2011.01970.x
Biro, P. A., & Adriaenssens, B. (2013). Predictability as a personality trait: Consistent
differences in intraindividual behavioral variation. The American Naturalist,
182 (5), 621e629. https://doi.org/10.1086/673213
B. Z. W. Kwek et al. / Animal Behaviour 179 (2021) 51e6358
Bjornson, F., & Anderson, G. W. (2018). Body condition, rather than size, predicts
risk-taking and resource holding potential in hatchery reared juvenile lake
sturgeon Acipenser fulvescens.Journal of Fish Biology, 93(6), 1188e1196. https://
doi.org/10.1111/jfb.13840
Boake, C. R. B. (1989). Repeatability: Its role in evolutionary studies of mating
behavior. Evolutionary Ecology, 3(2), 173e182. https://doi.org/10.1007/
BF02270919
Briffa, M. (2013). Plastic proteans: Reduced predictability in the face of predation
risk in hermit crabs. Biology Letters, 9(5). https://doi.org/10.1098/rsbl.2013.0592
Briffa, M. (2014). What determines the duration of war? Insights from assessment
strategies in animal contests. PLos One, 9(9), Article e108491. https://doi.org/
10.1371/journal.pone.0108491
Briffa, M., Sneddon, L. U., & Wilson, A. J. (2015). Animal personality as a cause and
consequence of contest behaviour. Biology Letters, 11(3), 20141007. https://
doi.org/10.1098/rsbl.2014.1007
Brown, W. D., Smith, A. T., Moskalik, B., & Gabriel, J. (2006). Aggressive contests in
house crickets: Size, motivation and the information content of aggressive
songs. Animal Behaviour, 72(1), 225e233. https://doi.org/10.1016/
j.anbehav.2006.01.012
Bürkner, P.-C. (2017). brms: An R package for Bayesian multilevel models using Stan.
Journal of Statistical Software, 80(1), 1e28. https://doi.org/10.18637/jss.v080.i01
Byers, J., Hebets, E., & Podos, J. (2010). Female mate choice based upon male motor
performance. Animal Behaviour, 79(4), 771e778. https://doi.org/10.1016/
j.anbehav.2010.01.009
Camerlink, I., Turner, S. P., Farish, M., & Arnott, G. (2015). Aggressiveness as a
component of fighting ability in pigs using game-theoretical framework. Animal
Behaviour, 108,183e191. https://doi.org/10.1016/j.anbehav.2015.07.032
Chang, C., Connahs, H., Tan, E. C. Y., Norma-Rashid, Y., Mrinalini, Li, D., & Chew, F. T.
(2020). Female spider aggression is associated with genetic underpinnings of
the nervous system and immune response to pathogens. Molecular Ecology,
29(14), 2626e2638. https://doi.org/10.1111/mec.15502
Clark, R. J., Jackson, R. R., & Waas, J. R. (1999). Draglines and assessment of fighting
ability in cannibalistic jumping spiders. Journal of Insect Behavior, 12(6),
753e766. https://doi.org/10.1023/A:1020900925863
Cleasby, I. R., Nakagawa, S., & Schielzeth, H. (2015). Quantifying the predictability of
behaviour: Statistical approaches for the study of between-individual variation
in the within-individual variance. Methods in Ecology and Evolution, 6,27e37.
https://doi.org/10.1111/2041-210X.12281
Cornuau, J. H., Schmeller, D. S., Courtois, E. A., Jolly, T., & Loyau, A. (2015). It takes
two to tango: Relative influence of male and female identity and morphology
on complex courtship display in a newt species. Ethology, 121(3), 218e226.
https://doi.org/10.1111/eth.12333
Dingemanse, N. J., & Dochtermann, N. A. (2013). Quantifying individual variation in
behaviour: Mixed-effect modelling approaches. Journal of Animal Ecology, 82(1),
39e54. https://doi.org/10.1111/1365-2656.12013
Dingemanse, N. J., Kazem, A. J. N., R
eale, D., & Wright, J. (2010). Behavioural reaction
norms: Animal personality meets individual plasticity. Trends in Ecology &
Evolution, 25(2), 81e89. https://doi.org/10.1016/j.tree.2009.07.013
Dingemanse, N. J., & R
eale, D. (2005). Natural selection and animal personality.
Behaviour, 142(9/10), 1159e1184. https://doi.org/10.1163/156853905774539445
Dohm, M. R. (2002). Repeatability estimates do not always set an upper limit to
heritability. Functional Ecology, 16(2), 273e280. https://doi.org/10.1046/j.1365-
2435.2002.00621.x
Elias, D. O., Hebets, E. A., & Hoy, R. R. (2006). Female preference for complex/novel
signals in a spider. Behavioral Ecology, 17(5), 765e771. https://doi.org/10.1093/
beheco/arl005
Elias, D. O., Kasumovic, M. M., Punzalan, D., Andrade, M. C. B., & Mason, A. C. (2008).
Assessment during aggressive contests between male jumping spiders. Animal
Behaviour, 76(3), 901e910. https://doi.org/10.1016/j.anbehav.2008.01.032
Favati, A., Løvlie, H., & Leimar, O. (2017). Individual aggression, but not
winnereloser effects, predicts social rank in male domestic fowl. Behavioral
Ecology, 28(3), 874e882. https://doi.org/10.1093/beheco/arx053
Garnham, L. C., Porth
en, S. A., Child, S., Forslind, S., & Løvlie, H. (2019). The role of
personality, cognition, and affective state in same-sex contests in the red jun-
glefowl. Behavioral Ecology and Sociobiology, 73(11), 149. https://doi.org/
10.1007/s00265-019-2762-0
Grob, P. (2015). Notes on the jumping spider Siler semiglaucus (Simon, 1901) in
Thailand (Araneae: Salticidae: Heliophaninae). Peckhamia, 126(1), 1e5.
Grueber, C. E., Nakagawa, S., Laws, R. J., & Jamieson, I. G. (2011). Multimodel infer-
ence in ecology and evolution: Challenges and solutions. Journal of Evolutionary
Biology, 24(4), 699e711. https://doi.org/10.1111/j.1420-9101.2010.02210.x
Harland, D., Li, D., & Jackson, R. (2012). How jumping spiders see the world. In
O. F. Lazareva, T. Shimizu, & E. A. Wasserman (Eds.), How animals see the world:
comparative behavior, biology, and evolution of vision (pp. 133e163). New York,
NY: Oxford University Press.
Harrison, L. M., Jennions, M. D., & Head, M. L. (2018). Does the winnereloser effect
determine male mating success? Biology Letters, 14(5), 20180195. https://
doi.org/10.1098/rsbl.2018.0195
Hebets, E. A., & Maddison, W. P. (2005). Xenophilic mating preferences among
populations of the jumping spider Habronattus pugillis Griswold. Behavioral
Ecology, 16(6), 981e988. https://doi.org/10.1093/beheco/ari079
Hsiung, B.-K., Siddique, R . H., Stavenga, D. G., Otto, J. C., Alle n, M. C., Liu, Y.,
et al. (2017). Rainbow peacock s piders inspi re miniature super-iridescen t
optics. Nature Commun ications, 8(1), 2278. h ttp://doi. org/10.1038/s41467-
017-02451-x.
Hunt, J., Breuker, C. J., Sadowski, J. A., & Moore, A. J. (2009). Maleemale competition,
female mate choice and their interaction: Determining total sexual selection.
Journal of Evolutionary Biology, 22(1), 13e26. https://doi.org/10.1111/j.1420-
9101.2008.01633.x
Hurd, P. L. (2006). Resource holding potential, subjective resource value, and game
theoretical models of aggressiveness signalling. Journal of Theoretical Biology,
241(3), 639e648. https://doi.org/10.1016/j.jtbi.2006.01.001
Jackson, R. R., Walker, M. W., Pollard, S. D., & Cross, F. R. (2006). Influence of seeing a
female on the maleemale interactions of a jumping spider, Hypoblemum
albovittatum.Journal of Ethology, 24(3), 231e238. https://doi.org/10.1007/
s10164-005-0185-z
Jaroensutasinee, M., & Jaroensutasinee, K. (2003). Type of intruder and reproductive
phase influence male territorial defence in wild-caught Siamese fighting fish.
Behavioural Processes, 64(1), 23e29. https://doi.org/10.1016/S0376-6357(03)
00106-2
Jennings, D. J., Gammell, M. P., Carlin, C. M., & Hayden, T. J. (2004). Effect of body
weight, antler length, resource value and experience on fight duration and
intensity in fallow deer. Animal Behaviour, 68(1), 213e221. https://doi.org/
10.1016/j.anbehav.2003.11.005
Kalb, N., Lindstr€
om, K., Sprenger, D., Anthes, N., & Heubel, K. U. (2016). Male per-
sonality and female spawning consistency in a goby with exclusive male care.
Behavioral Ecology and Sociobiology, 70(5), 683e693. https://doi.org/10.1007/
s00265-016-2089-z
Kasumovic, M. M., Elias, D. O., Punzalan, D., Mason, A. C., & Andrade, M. C. B. (2009).
Experience affects the outcome of agonistic contests without affecting the se-
lective advantage of size. Animal Behaviour, 77(6), 1533e1538. https://doi.org/
10.1016/j.anbehav.2009.02.026
Kern, E. M. A., Robinson, D., Gass, E., Godwin, J., & Langerhans, R. B. (2016). Corre-
lated evolution of personality, morphology and performance. Animal Behaviour,
117,79e86. https://doi.org/10.1016/j.anbehav.2016.04.007
Kralj-Fi
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(4), 824e831. https://doi.org/10.1093/beheco/art030
Kralj-Fi
ser, S., & Schuett, W. (2014). Studying personality variation in invertebrates:
Why bother? Animal Behaviour, 91,41e52. https://doi.org/10.1016/
j.anbehav.2014.02.016
Land, M. F., & Nilsson, D. E. (2012). Animal eyes (2nd ed.). Oxford, U.K.: Oxford
University Press.
Lane, S. M., & Briffa, M. (2020). Perceived and actual fighting ability: Determinants
of success by decision, knockout or submission in human combat sports. Biology
Letters, 16(10), 20200443. https://doi.org/10.1098/rsbl.2020.0443
Laubu, C., Schweitzer, C., Motreuil, S., Lou^
apre, P., & Dechaume-Moncharmont, F. X.
(2017). Mate choice based on behavioural type: Do convict cichlids prefer
similar partners? Animal Behaviour, 126, 281e291. https://doi.org/10.1016/
j.anbehav.2017.02.020
Lim, M. L. M., Land, M. F., & Li, D. (2007). Sex-specific UV and fluorescence signals in
jumping spiders. Science, 315(5811). https://doi.org/10.1126/science.1134254,
481e481.
Lim, M. L. M., & Li, D. (2004). Courtship and male-male agonistic behaviour of
Cosmophasis umbratica Simon, an ornate jumping spider (Araneae: Salticidae)
from Singapore. Raffles Bulletin of Zoology, 52, 435e448. http://archive.org/
details/raffles-bulletin-zoology-52-435-448.
Lim, M. L. M., & Li, D. (2007). Effects of age and feeding history on structure-based
UV ornaments of a jumping spider (Araneae: Salticidae). Proceedings of the
Royal Society B: Biological Sciences, 274(1609), 569e575. https://doi.org/10.1098/
rspb.2006.0020
Loranger, M. J., & Bertram, S. M. (2016). The effect of male dominance on female
choice in a field cricket (Gryllus assimilis). Animal Behaviour, 114,45
e52. https://
doi.org/10.1016/j.anbehav.2016.01.020
Manenti, R., Pennati, R., & Ficetola, G. F. (2015). Role of density and resource
competition in determining aggressive behaviour in salamanders. Journal of
Zoology, 296(4), 270e277. https://doi.org/10.1111/jzo.12241
McGinley,R.H.,Prenter,J.,&Taylor,P.W.(2015).Assessmentstrategies
and decision making in maleemale contests of Servaea incana jumping
spiders. Animal Behaviour, 101,89e95. https://doi.org/10.1016/
j.anbehav.2014.12.014
Mitchell, D. J., Fanson, B. G., Beckmann, C., & Biro, P. A. (2016). Towards powerful
experimental and statistical approaches to study intraindividual variability in
labile traits. Royal Society Open Science, 3(10), 160352. https://doi.org/10.1098/
rsos.160352
Moiron, M., Laskowski, K. L., & Niemel€
a, P. T. (2020). Individual differences in
behaviour explain variation in survival: A meta-analysis. Ecology Letters, 23(2),
399e408. https://doi.org/10.1111/ele.13438
Moretz, J. A. (2003). Aggression and RHP in the northern swordtail fish, Xiphophorus
cortezi: The relationship between size and contest dynamics in maleemale
competition. Ethology, 109(12), 995e1008. https://doi.org/10.1046/j.0179-
1613.2003.00938.x
Moretz, J. A. (2005). Aggression and fighting ability are correlated in the
swordtail fish Xiphophorus cortezi: The advantage of being barless. Behavioral
Ecology and Sociobiology, 59(1), 51e57. https://doi.org/10.1007/s00265-005-
0008-9
Munson, A. A., Jones, C., Schraft, H., & Sih, A. (2020). You’re just my type: Mate
choice and behavioral yypes. Trends in Ecology &Evolution, 35, 823e833.
https://doi.org/10.1016/j.tree.2020.04.010
B. Z. W. Kwek et al. / Animal Behaviour 179 (2021) 51e63 59
Neumann, C., Agil, M., Widdig, A., & Engelhardt, A. (2013). Personality of wild male
crested macaques (Macaca nigra). PLos One, 8(8), Article e69383. https://doi.org/
10.1371/journal.pone.0069383
Pauers, M. J., Kapfer, J. M., Doehler, K., Lee, J. T., & Berg, C. S. (2012). Gross colour
pattern is used to distinguish between opponents during aggressive encounters
in a Lake Malawi cichlid. Ecology of Freshwater Fish, 21(1), 34e41. https://
doi.org/10.1111/j.1600-0633.2011.00520.x
Plummer, M., Best, N., Cowles, K., & Vines, K. (2006). CODA: Convergence diagnosis
and output analysis for MCMC. R News, 6(1), 7e11.
R Core Team. (2020). R: a language and environment for statistical computing. Vienna,
Austria: R Foundation for Statistical Computing. https://www.R-project.org/.
R
eale, D., Reader, S. M., Sol, D., McDougall, P. T., & Dingemanse, N. J. (2007). Inte-
grating animal temperament within ecology and evolution. Biological Reviews,
82(2), 291e318. https://doi.org/10.1111/j.1469-185X.2007.00010.x
Royaut
e, R., Buddle, C. M., & Vincent, C. (2014). Interpopulation variations in
behavioural syndromes of a jumping spider from insecticide-treated and
insecticide-free orchards. Ethology, 120(2), 127e139. https://doi.org/10.1111/
eth.12185
Royaut
e, R., Buddle, C. M., & Vincent, C. (2015). Under the influence: Sublethal
exposure to an insecticide affects personality expression in a jumping spider.
Functional Ecology, 29(7), 962e970. https://doi.org/10.1111/1365-2435.12413
Royle, N. J., Schuett, W., & Dall, S. R. X. (2010). Behavioral consistency and the
resolution of sexual conflict over parental investment. Behavioral Ecology, 21(6),
112 5 e1130. https://doi.org/10.1093/beheco/arq156
Scherer, U., Kuhnhardt, M., & Schuett, W. (2017). Different or alike? Female rainbow
kribs choose males of similar consistency and dissimilar level of boldness.
Animal Behaviour, 128,117e124. https://doi.org/10.1016/j.anbehav.2017.04.007
Scherer, U., Kuhnhardt, M., & Schuett, W. (2018). Predictability is attractive: Female
preference for behaviourally consistent males but no preference for the level of
male aggression in a bi-parental cichlid. PLos One, 13(4), Article e0195766.
https://doi.org/10.1371/journal.pone.0195766
Scherer, U., & Schuett, W. (2018). No male mate choice for female boldness in a bi-
parental West African cichlid, the rainbow krib (Pelvicachromis pulcher). e5373
PeerJ, 6,1e19. http://doi.org/10.7717/peerj.5373.
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, 117(10),
908e917. https://doi.org/10.1111/j.1439-0310.2011.01945.x
Schuett, W., Tregenza, T., & Dall, S. R. X. (2010). Sexual selection and animal per-
sonality. Biological Reviews, 85(2), 217e246. https://doi.org/10.1111/j.1469-
185X.2009.00101.x
Sih, A., Bell, A., & Johnson, J. C. (2004). Behavioral syndromes: An ecological and
evolutionary overview. Trends in Ecology &Evolution, 19(7), 372e378. https://
doi.org/10.1016/j.tree.2004.04.009
Smith, B. R., & Blumstein, D. T.( 2008). Fitness consequences of personality: A meta-analysis.
Behavioral Ecology, 19(2), 448e455. https://doi.org/10.1093/beheco/arm144
Songvorawit, N., Butcher, B. A., & Chaisuekul, C. (2018). Resource holding potential
and the outcome of aggressive interactions between paired male Aegus chelifer
chelifer (Coleoptera: Lucanidae) stag beetles. Journal of Insect Behavior, 31(4),
347e360. https://doi.org/10.1007/s10905-018-9683-z
Stamps, J. A., Briffa, M., & Biro, P. A. (2012). Unpredictable animals: Individual dif-
ferences in intraindividual variability (IIV). Animal Behaviour, 83(6), 1325e1334.
https://doi.org/10.1016/j.anbehav.2012.02.017
Taylor, P. W., & Elwood, R. W. (2003). The mismeasure of animal contests. Animal
Behaviour, 65(6), 1195e1202. https://doi.org/10.1006/anbe.2003.2169
Thys, B., Pinxten, R., Raap, T., Link to external site, this link will open in a new
window, De Meester, G., Rivera-Gutierrez, H. F., & Eens, M. (2017). The female
perspective of personality in a wild songbird: Repeatable aggressiveness relates
to exploration behaviour. Scientific Reports, 7,1e10. https://doi.org/10.1038/
s41598-017-08001-1
Tognetti, A., Ganem, G., Raymond, M., & Faurie, C. (2018). Female mound-building
mice prefer males that invest more in building behavior, even when this
behavior is not observed. Behavioral Ecology and Sociobiology, 72(9), 155. https://
doi.org/10.1007/s00265-018-2569-4
Wilson, A. J., Grimmer, A., & Rosenthal, G. G. (2013). Causes and consequences of
contest outcome: Aggressiveness, dominance and growth in the sheepshead
swordtail, Xiphophorus birchmanni.Behavioral Ecology and Sociobiology, 67(7),
1151 e1161. https://doi.org/10.1007/s00265-013-1540-7
Wolak, M. E., Fairbairn, D. J., & Paulsen, Y. R. (2012). Guidelines for estimating
repeatability. Methods in Ecology and Evolution, 3(1), 129e137. https://doi.org/
10.1111/j.2041-210X.2011.00125.x
Wong, B. B. M., & Candolin, U. (2005). How is female mate choice affected by male
competition? Biological Reviews, 80(4), 559e571. https://doi.org/10.1017/
S1464793105006809
World Spider Catalog. (2021). World spider catalog. Version 22.0. Bern, Switzerland:
Natural History Museum http://wsc.nmbe.ch. (Accessed 2 April 2021).
Yasuda, C. I., & Koga, T. (2016). Importance of weapon size in all stages of
maleemale contests in the hermit crab Pagurus minutus. Behavioral Ecology and
Sociobiology, 70(12), 2175e2183. https://doi.org/10.1007/s00265-016-2221-0
Zeng, H. (2019). Colour-based sexual selection in a genus of jumping spider Siler
(Araneae: Salticidae) (Doctoral dissertation). Singapore: National University of
Singapore.
Zeng, H., Wee, S. S. E., Painting, C. J., Zhang, S., & Li, D. (2019). Equivalent effect of UV
coloration and vibratory signal on mating success in a jumping spider. Behav-
ioral Ecology, 30(2), 313e321. https://doi.org/10.1093/beheco/ary167
Appendix
To determine whether the male courtship duration or court-
ship intensity (i.e. courtship duration divided by the number of
courtship displays) is correlated with male aggression or male
aggression predictability, we ran linear mixed-effect models from
the package lme4 (Bates et al., 2015) with male identity as a
random effect due to reuse of males in the female mate choice
experiments. The afex package (Singmann et al., 2021) was used
to obtain the Pvalues. We ran four models with male courtship
duration and courtship intensity as response variables, male
aggression and aggression predictability (separately) as fixed
effects and male identity as a random effect. We also ran two
generalized mixed-effect models to determine the relationship
between the number of courtship displays and male aggression
or male aggression predictability. We coded the number of
courtship displays as response variable, male aggression and
aggression predictability (separately) as fixed effects and male
identity as a random effect. The results are presented in Ap-
pendix Tables A10 and A11.
Table A1
Double hierarchical generalized linear mixed-effects model (DHGLM) showing the
effects of the number of trials, sex and carapace width on aggression of S. semi-
glaucus using the full data set (including eight ‘extreme’spiders)
Estimate SE Lower 95% CI Upper
95% CI
Group
level effects
~ID sd(Intercept) 1.73 0.16 1.44 2.07
sd(sigma_Intercept) 0.59 0.08 0.44 0.75
Population
level effects
Intercept 3.13 2.21 1.12 7.46
sigma_Intercept 0.05 0.08 0.11 0.20
Trial number 0.04 0.03 0.02 0.11
Sex (Male) 1.10 0.46 0.20 1.98
Carapace width 2.94 1.56 0.12 5.88
Males: N¼49, females: N¼42. CI: confidence interval.
B. Z. W. Kwek et al. / Animal Behaviour 179 (2021) 51e6360
Table A2
Fully and conditionally averaged models from the top models (delta QAIC <2) showing the effects of explanatory variables on female mate preference using the full data set
(including eight ‘extreme’spiders)
Model Predictor Estimate SE Z P
Full average Intercept 1.17 2.03
Male aggression difference 1.15 0.34 3.24 0.001
Female IIV 0.08 0.39 0.19 0.849
Female carapace width 0.21 1.32 0.16 0.877
Female aggression 0.01 0.14 0.09 0.927
Conditional average Intercept 1.17 2.03
Male aggression difference 1.15 0.34 3.24 0.001
Female IIV 0.45 0.85 0.51 0.610
Female carapace width 1.29 3.04 0.41 0.683
Female aggression 0.08 0.34 0.24 0.814
Number of trials: 36 (two trials using ‘extreme’spiders were included). IIV: intraindividual variation (predictability).
Table A3
Fully and conditionally averaged models from the top models (delta QAIC <2) showing the effects of explanatory variables on the outcomes of male contests using the full data
set (including eight ‘extreme’spiders)
Model Predictor Estimate SE Z P
Full average Intercept 0.04 0.35
Male aggression difference 0.42 0.16 2.55 0.011
Male carapace width difference 2.06 4.82 0.42 0.675
Conditional average Intercept 0.04 0.35
Male aggression difference 0.42 0.16 2.55 0.011
Male carapace width difference 6.11 6.65 0.89 0.372
Number of trials: 46 (two trials using ‘extreme’spiders were included).
Table A4
Double hierarchical generalized linear mixed-effects model (DHGLM) showing the effects of the number of trials, sex and carapace width on aggressionofS. semiglaucus
Estimate SE Lower 95% CI Upper 95% CI
Group level effects
~Population sd(Intercept) 1.08 1.19 0.03 4.27
sd(sigma_Intercept) 0.82 1.04 0.02 3.66
~Population : ID sd(Intercept) 1.73 0.16 1.43 2.06
sd(sigma_Intercept) 0.58 0.08 0.44 0.75
Population level effects Intercept 3.19 2.32 1.38 7.78
sigma_Intercept 0.06 0.69 1.57 1.52
Trial number 0.04 0.03 0.02 0.11
Sex (Male) 1.07 0.45 0.18 1.92
Carapace width 2.90 1.52 0.12 5.87
Group level population is the collection site (China and Singapore). Males: N¼44; females: N¼39. ID: spider ID. CI: confidence interval.
Table A5
Comparisons of 18 generalized linear models to predict female mate choice
Predictor df LogLik QAIC Delta Weight
Male aggression difference 2 12.1 22.9 0.00 0.368
Female aggression þMale aggression difference 3 11.8 24.5 1.52 0.172
Female carapace width þMale aggression difference 3 12.1 24.9 1.95 0.139
Female aggression IIV þMale aggression difference 3 12.1 24.9 2.00 0.136
Female aggression*Male aggression difference 4 11.7 26.4 3.48 0.065
Female aggression IIV*Male aggression difference 4 12.1 26.9 4.00 0.050
Male aggression IIV difference 2 15.9 28.2 5.30 0.026
Female aggression IIV þMale aggression IIV difference 3 15.2 29.3 6.33 0.016
Female aggression IIV*Male aggression IIV difference 4 14.0 29.7 6.71 0.013
Female aggression þMale aggression IIV difference 3 15.9 30.2 7.30 0.010
Female aggression*Male aggression IIV difference 4 15.3 31.4 8.47 0.005
Null model ~1 1 23.6 37.0 14.06 0.000
Male carapace width difference 2 22.3 37.2 14.27 0.000
Female aggression 2 23.1 38.3 15.39 0.000
Female carapace width þMale carapace width difference 3 21.9 38.6 15.66 0.000
Female carapace width 2 23.4 38.7 15.77 0.000
Female aggression IIV 2 23.4 38.8 15.82 0.000
Female aggression þMale carapace width difference 3 22.1 38.9 15.96 0.000
IIV: intraindividual variation (predictability).
B. Z. W. Kwek et al. / Animal Behaviour 179 (2021) 51e63 61
Table A6
Summary of the best model (delta QAIC <2) showing the effects of explanatory variables on female attention time
Model Predictor Estimate SE Z P
Full average Intercept 1.97 21.53
Male aggression difference 11.96 10.22 1.16 0.248
Male IIV difference 31.67 49.86 0.63 0.529
Conditional average Intercept 1.97 21.53
Male aggression difference 18.11 6.85 2.55 0.011
Male IIV difference 93.23 39.75 2.26 0.024
Number of trials: 34 (two trials using ‘extreme’spiders were excluded). IIV: intraindividual variation (predictability).
Table A7
Comparisons of 18 linear regression models to predict the female mate choice using female attention time
Predictor df LogLik QAIC Delta Weight
Male aggression difference 3 205.4 417.5 0.00 0.305
Male IIV difference 3 206.0 418.9 1.33 0.157
Female IIV þMale aggression difference 4 205.4 420.1 2.56 0.085
Female aggression þMale aggression difference 4 205.4 420.1 2.57 0.084
Female carapace width þMale aggression difference 4 205.4 420.1 2.58 0.084
Female aggression þMale IIV difference 4 205.9 421.3 3.72 0.048
Female IIV þMale IIV difference 4 206.0 421.4 3.85 0.045
Null model ~1 2 208.7 421.8 4.31 0.035
Female aggression*Male aggression difference 5 205.1 422.4 4.87 0.027
Male carapace width difference 3 207.9 422.5 4.99 0.025
Female IIV*Male aggression difference 5 205.3 422.8 5.25 0.022
Female IIV*Male IIV difference 5 205.4 422.9 5.41 0.020
Female aggression*Male IIV difference 5 205.6 423.4 5.83 0.017
Female aggression 3 208.7 424.1 6.56 0.011
Female carapace width 3 208.7 424.2 6.70 0.011
Female IIV 3 208.7 424.2 6.71 0.011
Female carapace width þMale carapace width difference 4 207.8 425.0 7.42 0.007
Female aggression þMale carapace width difference 4 207.9 425.1 7.55 0.007
Number of trials: 34 (two trials using ‘extreme’spiders were excluded). IIV: intraindividual variation (predictability).
Table A8
Comparisons of 15 generalized linear models to predict the outcomes of male contests
Predictor df LogLik QAIC Delta Weight
Male aggression difference 2 24.1 40.0 0.00 0.371
Male aggression difference þMale carapace width difference 3 23.7 41.5 1.43 0.181
Male aggression IIV difference 2 25.4 41.8 1.78 0.152
Male aggression difference*Male carapace width difference 4 23.1 42.5 2.49 0.107
Male aggression IIV difference þMale carapace width difference 3 24.8 43.0 2.94 0.085
Male aggression IIV difference*Male carapace width difference 4 24.5 44.6 4.57 0.038
Male aggression difference þExperience 5 23.9 45.7 5.71 0.021
Male aggression IIV difference þExperience 5 24.8 46.9 6.90 0.012
Null model ~1 1 30.5 47.0 6.97 0.011
Male aggression IIV difference*Experience 8 20.8 47.3 7.28 0.010
Male carapace width difference 2 30.2 48.6 8.57 0.005
Male aggression difference*Experience 8 22.0 49.0 8.95 0.004
Experience 4 28.7 50.4 10.37 0.002
Male carapace width difference þExperience 5 28.4 52.1 12.05 0.001
Male carapace width difference*Experience 8 27.4 56.7 16.66 0.000
IIV: intraindividual variation (predictability).
Table A9
Spearman correlations between aggression, aggression IIV and the residuals of
carapace width and body mass in male S. semiglaucus
Male Aggression Aggression IIV Residual
Aggression e0.412 0.119
Aggression IIV 0.006 e0.233
Residual 0.442 0.129 e
The Pvalues are shown below and correlation coefficients above the diagonal. IIV:
intraindividual variation (predictability).
Table A10
Linear mixed-effect models testing for the effects of aggression and aggression IIV in
males on courtship duration and courtship intensity
Response variable Predictor Estimate SE tP
Courtship duration Intercept 115.77 29.12
Aggression 12.26 6.96 1.76 0.087
Intercept 8.54 0.69
Aggression IIV 1.29 1.79 0.72 0.480
Courtship intensity Intercept 15.53 8.78
Aggression 2.64 2.10 1.26 0.217
Intercept 25.76 3.85
Aggression IIV 3.89 9.99 0.39 0.699
N¼44 (two trials using ‘extreme’spiders were excluded). Male ID was included as a
random effect to account for the reuse of males. IIV: intraindividual variation
(predictability).
B. Z. W. Kwek et al. / Animal Behaviour 179 (2021) 51e6362
Table A11
Generalized linear mixed-effect model for the effects of aggression and aggression IIV in males on number of courtship displays
Response variable Predictor Estimate SE Z P
Number of courtship displays Intercept 2.32 0.17
Aggression 0.06 0.04 1.57 0.117
Number of courtship displays Intercept 2.07 0.08
Aggression IIV 0.13 0.20 0.67 0.503
N¼44 (two trials using ‘extreme’spiders were excluded). IIV: intraindividual variation (predictability).
B. Z. W. Kwek et al. / Animal Behaviour 179 (2021) 51e63 63
