Content uploaded by Alexander K. Hill
Author content
All content in this area was uploaded by Alexander K. Hill on Nov 02, 2017
Content may be subject to copyright.
Original Article
Quantifying the strength and form of sexual selection on men's traits☆
,
☆☆
Alexander K. Hill
a
, John Hunt
b
, Lisa L.M. Welling
a,c
, Rodrigo A. Cárdenas
d
, Michelle A. Rotella
a
,
John R. Wheatley
a
, Khytam Dawood
d
, Mark D. Shriver
a
, David A. Puts
a,e,
⁎
a
Department of Anthropology, Pennsylvania State University, University Park, PA 16802
b
Centre for Ecology and Conservation, School of Biosciences, University of Exeter, Penryn, UK
c
Department of Psychology, Oakland University, Rochester, MI 48309
d
Department of Psychology, Pennsylvania State University, University Park, PA 16802
e
Center for Brain, Behavior, and Cognition, Pennsylvania State University, University Park, PA 16802
abstractarticle info
Article history:
Initial receipt 13 March 2013
Final revision received 30 May 2013
Keywords:
Attractiveness
Dominance
Face
Masculinity
Sociosexuality
Voice
Although recent research has increasingly focused on human sexual selection, fundamental questions remain
concerning the relative influence of individual traits on success in competition for mates and the mechanisms,
form, and direction of these sexual selective pressures. Here, we explore sexual selection on men's traits by
ascertaining men's dominance and attractiveness from male and female acquaintances. On a large American
university campus, 63 men from two social fraternities provided anthropometric measurements, facial
photographs, voice recordings, and reported mating success (number of sexual partners). These men also
assessed each other's dominance, and 72 women from two socially affiliated sororities assessed the men's
attractiveness. We measured facial masculinity from inter-landmark distances and vocal masculinity from
acoustic parameters. We additionally obtained facial and vocal attractiveness and dominance ratings from
unfamiliar observers. Results indicate that dominance and the traits associated with it predict men's mating
success, but attractiveness and the traits associated with it do not. These findings point to the salience of
contest competition on men's mating success in this population.
© 2013 The Authors. Published by Elsevier Inc. All rights reserved.
1. Introduction
A rapidly growing literature suggests that sexual selection has
shaped men's phenotypic traits (Puts, Jones, & DeBruine, 2012).
Men's bodies (Pawlowski & Jasienska, 2005), faces (Penton-Voak &
Perrett, 2000; Rhodes, Chan, Zebrowitz, & Simmons, 2003), and
voices (Fitch & Giedd, 1999; Puts, Apicella, & Cárdenas, 2012)
exhibit features that are highly sexually differentiated and develop
at sexual maturity. These traits also appear to aid in competition for
mates. In men, more masculine, muscular bodies (Dixson, Dixson,
Bishop, & Parish, 2010; Frederick & Haselton, 2007; Hughes,
Dispenza, & Gallup, 2004) and tall stature (Nettle, 2002; Pawlowski,
Dunbar, & Lipowicz, 2000; Pawlowski & Jasienska, 2005) predict
reported number of sexual partners and perceptions of attractive-
ness and dominance. Masculine faces also convey dominance
(DeBruine et al., 2006; Perrett et al., 1998; Watkins, Jones, &
DeBruine, 2010), and some studies (DeBruine et al., 2006; Johnston,
Hagel, Franklin, Fink, & Grammer, 2001), but not others (DeBruine,
Jones, Smith, & Little, 2010; Perrett et al., 1998; Rhodes, Hickford, &
Jeffery, 2000), have found that women prefer masculine male faces,
particularly during the fertile phase of the ovulatory cycle (Johnston
et al., 2001; Penton-Voak & Perrett, 2000; Penton-Voak et al., 1999;
Welling et al., 2007). Similarly, masculine voices have been found to
predict men's number of reported sex partners (Hodges-Simeon,
Gaulin, & Puts, 2011; Puts, 2005) and reproductive success (Apicella,
Feinberg, & Marlowe, 2007). In addition, masculine acoustic features,
such as deep timbre and low pitch influence perceptions of both
dominance (Jones, Feinberg, DeBruine, Little, & Vukovic, 2010; Puts,
Hodges, Cárdenas, & Gaulin, 2007; Watkins, Fraccaro, et al., 2010;
Wolff & Puts, 2010) and attractiveness (Feinberg, Jones, Little, Burt,
& Perrett, 2005; Hodges-Simeon, Gaulin, & Puts, 2010), the latter
particularly during the fertile phase of the ovulatory cycle (Feinberg
et al., 2006; Puts, 2005, 2006).
Such research has helped illuminate whether and how sexual
selection has shaped men's phenotypes, yet a number of funda-
mental questions remain. First, prior studies have typically focused
on either female choice or male contests without attempting to
quantify the relative contributions of these mechanisms to the total
sexual selective pressure on a particular trait (Hunt, Breuker,
Sadowski, & Moore, 2009). Second, to our knowledge, no study
reporting relationships between a male trait and mating success has
investigated whether these relationships were mediated by
Evolution and Human Behavior 34 (2013) 334–341
☆This is an open-access article distributed under the terms of the Creative Commons
Attribution License, which permits unrestricted use, distribution, and reproduction in any
medium, provided the original author and source are credited.
☆☆ JH was funded by NERC and a University Royal Society Fellowship. LW was funded
by an AIB grant awarded to DAP.
⁎Corresponding author. Department of Anthropology, Pennsylvania State Univer-
sity, University Park, PA 16802. Tel.: +1 814 867 0453.
E-mail address: dap27@psu.edu (D.A. Puts).
1090-5138/$ –see front matter © 2013 The Authors. Published by Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.evolhumbehav.2013.05.004
Contents lists available at SciVerse ScienceDirect
Evolution and Human Behavior
journal homepage: www.ehbonline.org
attractiveness or dominance. Third, most studies of sexual selection
in men have measured success under female choice or male
contests from limited information, such as body size, strength, or
ratings of faces or voices made by strangers in the laboratory (but
see, e.g., Pillsworth, 2008; von Rueden, Gurven, & Kaplan, 2011).
Attractiveness and dominance have thus frequently been assessed
devoid of relevant information, such as personality and intelligence,
and in isolation from the complex webs of social relationships in
which we live. Fourth, facial, vocal, and bodily characteristics may
be developmentally correlated (Feinberg, 2008; Fink, Neave, &
Seydel, 2007), so when exploring relationships between each and
success in mating competition, it is necessary to measure and
statistically control for the others, as well as to explore their
interactions (i.e., correlational selection). Fifth, nearly all prior
research on these traits has assumed linear (directional) selection
without exploring quadratic selection gradients (i.e., stabilizing or
disruptive selection). Finally, previous studies have not explored
whether female choice or male contests contribute more strongly to
men's mating success, despite the centrality of this question in
understanding human sexual selection (Puts, 2010). In sum, we still
do not know the overall sexual selective pressures on individual
traits, the relative strengths of sexual selection on different traits,
the forms and mechanisms of this sexual selection, or even the
relative importance of female choice and male contests in men's
competition for mates.
In the present paper, we therefore investigated sexual selection on
some of the strongest candidates for sexually selected traits in men:
stature, body build, facial masculinity, and deep voices (Fig. 1). We
measured these traits in a sample of men and obtained assessments of
the men's success under intra- and intersexual selection from logically
the most authoritative source: familiar male and female peers
(Pillsworth, 2008; von Rueden et al., 2011). We also obtained
assessments of the men's facial and vocal attractiveness and
dominance from unfamiliar raters, as well as the men's self-reported
mating success. We then 1) investigated the contributions of these
traits to different mechanisms of sexual selection (mate choice and
contests) and to mating success via their linear, quadratic, and
correlational selection gradients, and 2) compared mechanisms of
sexual selection to each other and to mating success using a
sequential model-building approach (Draper & John, 1988) to identify
whether the strength, direction, and form of sexual selection on male
traits differ across these episodes, and if so, which traits contribute to
these overall differences.
Although we are interested in how past selection produced
present sexual dimorphisms, we take a behavioral ecological
approach, which emphasizes contemporary selection. We take this
approach because we expect that, in general, current function will
provide insight into past function. However, attractiveness, domi-
nance, and even mating success have likely been at least partly
decoupled from reproductive success by features of modern
industrial environments such as effective contraception and socially
imposed monogamy (Perusse, 1993). We therefore examined
components of reproductive success that are “upstream”of fertility.
Assuming that these components affected ancestral men's repro-
ductive output, and in parallel with nonhuman literature, which also
frequently measures only proximate components of fitness, we refer
to “selective surfaces”,“directional selection”, and the like.
2. Methods
2.1. Participants
We recruited members of two social fraternities (N = 63, mean
age ± SE = 19.9 ± 0.15) and two sororities (N = 72, mean age ±
SE = 19.4 ± 0.11) from a large university in the northeastern United
States. Each fraternity was socially affiliated with one of the sororities,
regularly attending joint social functions. Participating fraternity
members (male participants) were paid US$15, and participating
sorority members (female participants) were paid US$10. We also
recruited male (N = 36, mean age ± SE = 20.3 ± 0.42) and female
(N = 43, mean age ± SE = 19.0 ± 0.14) raters from the university's
psychology department subject pool. Raters were unfamiliar with the
male participants. All methods were approved by the university's
Institutional Review Board.
Fig. 1. Linear (β) and quadratic (γ) relationships (statistics shown for statistically significant relationships) between men's traits, success under female choice and male contests, and
mating success. Linear, quadratic and interaction terms for variables in each level to the left (e.g., men's traits) were entered into multiple regression models to predict variables at
higher levels to the right (e.g., success under female choice). Biceps, chest, and shoulder circumference, and weight were standardized and summed to produce the composite
variable, Girth. *pb.05, **pb.001, ***p b.0001.
335A.K. Hill et al. / Evolution and Human Behavior 34 (2013) 334–341
2.2. Participant procedures
We collected data at male participants' residence (fraternity)
houses using a series of stations. The first station acquired informed
consent; other stations collected voice recordings, facial photographs,
anthropometric measurements, and answers to online question-
naires. Male participants' voices were recorded in a quiet room using
a Shure SM58 vocal cardioid microphone placed approximately
9.5 cm from the participant's mouth. Voices were recorded in mono
with a sampling frequency of 44,100 Hz as male participants spoke
the first six sentences of the Rainbow Passage (Fairbanks, 1960). For
facial photographs, participants were instructed to remove all
earrings, glasses, and facial jewelry, and to use a headband if hair
covered any part of their faces. We asked participants to sit upright in
a chair and maintain a relaxed, neutral facial expression with the
mouth lightly closed. Participants posed approximately 2 m from the
camera, and the flash was always used. Height and chest, shoulder,
and flexed biceps circumference were taken with a measuring tape,
and weight was obtained using an electronic scale (Table E1). For
online questionnaires, participants were seated privately. Male
participants answered demographic questions; the revised Sociosex-
ual Orientation Inventory (SOI-R) (Penke & Asendorpf, 2008), which
targets attitudes, desires, and behavior regarding uncommitted sex;
and with how many women they had sexual intercourse in the past
year (Table E1).
One week after initial data collection, we returned to the
residences, at which time male participants were seated privately
at computers and sequentially shown facial photographs of all other
male participants from their fraternity. Participants were asked
what percentage of men each pictured man could beat in a physical
fight (0% to 100% in increments of 10%), and to estimate with how
many women the pictured man had had sexual intercourse in the
past year.
In order to control for any differences across photographs in
ambient light, we chose a photograph with optimal lighting and
matched all other photographs to its brightness using the Match Color
function in Adobe Photoshop CS6. Interpupillary distance was also
standardized across photographs. Order of rating tasks and presen-
tation order of stimuli were randomized.
Female participants were brought individually into the lab,
sequentially shown facial photographs on a computer of all male
participants from their affiliated fraternity, and asked to rate on a
10-point Likert scale how attractive each was for a “short-term,
purely sexual relationship, such as a one-night stand”,and
demographic questions. Rating task order and stimulus presentation
order were randomized as above. We utilized ratings of short-term,
sexual attractiveness, rather than attractiveness for a long-term,
committed relationship, because masculine traits more strongly
affect short-term, sexual attractiveness (e.g., Frederick & Haselton,
2007; Puts, 2005), and because we expected sexual attractiveness to
more strongly predict number of sexual partners. In other words,
short-term, sexual attractiveness measures the type of intersexual
selection that would more strongly link masculine traits with men's
mating success.
2.3. Independent male and female rater procedures
Independent male and female raters viewed photographs of the
male participants' faces and listened on Sennheiser HD 280 Pro
headphones to recordings of male participants reading the first
sentence of the Rainbow Passage. We standardized voice amplitude
(mean ± SD = 71.5 ± 2.4 dB) using Praat version 5.3, and photo-
graphs were standardized as above. Face and voice stimuli were split
randomly into two sets so that each independent rater rated half of
the stimuli, but all stimuli were evaluated an equal number of times.
Male raters evaluated faces and voices for fighting ability, while
female raters evaluated faces and voices for attractiveness in a short-
term, sexual relationship. The same questions and scales used for the
male and female participants were used with the independent raters.
Rating task order and stimulus presentation order were randomized
as above.
2.4. Data treatment
We calculated a facial masculinity index (Table E1) by placing a
series of 30 landmarks on each male participant's digital facial
photograph, then calculating seven sexually differentiated inter-
landmark distances and angles and standardizing and summing these
measures. Eye width, lower face height/total face height, cheekbone
prominence, and face width/lower face height were measured
following Penton-Voak and Perrett (2001), and eye height, jaw
angle, and nose width were measured following Burriss, Roberts,
Welling, Puts, and Little (2011).
Vocal masculinity was computed by standardizing and summing
two highly sexually differentiated acoustic parameters, fundamental
frequency (F
0
) and formant position (P
f
), obtained from male
participants' voice recordings measured in Praat, version 5.3 (Table
E1). Lower F
0
and P
f
correspond with deeper, more masculine pitch
and timbre, respectively. Pitch floor and ceiling were 75 Hz and
300 Hz; otherwise, default settings were used. Formant frequencies F
1
through F
4
were measured at each glottal pulse and averaged across
measurements, as in Puts et al. (2012). Formant measurements
obtained by this method correlate highly (rvaried between 0.93 and
0.98) with measurements obtained by measuring and averaging
across individual vowels (Puts et al., 2012). We then computed P
f
,
defined as the average standardized formant value for the first four
formants, as in (Puts et al., 2012). Means and SDs used to standardize
formants were: F
1
= 417.2 ± 31.7 Hz, F
2
= 1476.9 ± 47.0 Hz, F
3
=
2457.9 ± 68.9 Hz, F
4
= 3426.8 ± 91.2 Hz.
Anthropometric measurements were entered into a principal
components analysis with varimax rotation. Biceps, chest, and
shoulder circumference, and weight loaded onto the first component,
and height loaded onto the second component. We consequently
standardized and summed biceps, chest, and shoulder circumference,
and weight to produce the composite variable “girth”.
Each male participant's facial and vocal attractiveness and
dominance were measured as the mean short-term attractiveness
and fighting ability ratings, respectively, that he received from the
independent raters who assessed his facial photo and voice
recording (mean inter-class correlation coefficient (ICC) = 0.76,
range: 0.60–0.91, Table E1). Each male participant's success under
female choice was measured as the mean short-term attractiveness
rating that he received from female participants (ICC = 0.90, 0.95
for the two fraternities, Table E1). Each male participant's success in
male contests was measured as the mean rating of fighting ability
that he received from the male participants who rated him (ICC =
0.91, 0.95 for the two fraternities, Table E1). To measure
predisposition toward uncommitted sex, we summed the attitudinal
and desire components of the SOI-R (Penke & Asendorpf, 2008).
Lastly, we used self-reported number of sex partners in the past
year (Faurie, Pontier, & Raymond, 2004; Hodges-Simeon et al.,
2011) to measure mating success.
2.5. Statistical analysis
We first investigated the contributions of masculine traits to
different mechanisms of sexual selection and to mating success via
their linear, quadratic, and correlational selection gradients. After
standardizing trait values (to zero mean and unit variance) and
converting absolute fitness for each individual (mating success or
success under each mechanism of sexual selection) to relative fitness
(ω) by dividing by the mean fitness of the sample, we used multiple
336 A.K. Hill et al. / Evolution and Human Behavior 34 (2013) 334–341
regression to estimate the standardized linear selection gradients (β)
(Phillips & Arnold, 1989). We then estimated nonlinear forms of
selection by running a separate regression that includes quadratic and
cross-product terms to estimate the matrix of standardized nonlinear
selection gradients (γ)(Phillips & Arnold, 1989).
Interpreting the size and strength of individual elements in γcan
underestimate the true strength of nonlinear selection (Blows &
Brooks, 2003). We therefore performed a canonical analysis of the γ
matrix to find the major axes of the response surface (Phillips &
Arnold, 1989). This results in an Mmatrix consisting of ieigenvectors
(m
i
), each of which describes a major axis of the response surface
(where iis the original number of traits being examined). The
strength of linear selection along each eigenvector is given by θ
i
, and
the strength of nonlinear selection is given by its eigenvalue (λ
i
). θ
i
and λ
i
were estimated using the double linear regression method of
Bisgaard and Ankenman (1996). As our response variables were not
normally distributed, we tested the significance of our standardized
selection gradients and linear and nonlinear selection operating on
the eigenvectors of γusing randomization tests, as recommended by
Mitchell-Olds and Shaw (1987). These procedures, including ran-
domization tests, are provided in detail in Lewis, Wedell, and
Hunt (2011).
We used thin-plate splines (Green & Silverman, 1994) to visualize
the major axes of the response surface extracted from the canonical
analysis of γ. We used the Tps function in fields package of R (version
2.13.0; freely available at http://www.r-project.org)tofit a spline
surface using the lambda value that minimized the generalized cross-
validation (GCV) score. We then plotted the surface using perspective
view in R.
We compared mechanisms of sexual selection to each other and to
mating success using a sequential model-building approach (Draper &
John, 1988), a hierarchical model that first compares linear sexual
selection, then quadratic and correlational sexual selection to identify
whether the direction and form of sexual selection on male traits
differ across these episodes, and if so, which individual trait
contributes to the overall differences. Univariate interaction terms
from the complete models were used to determine which individual
traits contributed to any overall significant difference (see [Lewis et
al., 2011] for a full description of this approach).
3. Results
3.1. Female choice
Female choice exerted directional (linear) selection favoring
height (Fig. 1,Table 1A). There was also negative correlational
selection between girth and facial and vocal masculinity (Table 1A): as
girth increased, men with lower facial and vocal masculinity became
more attractive. Female choice did not exert significant quadratic
selection on male traits (Table 1A).
Canonical analysis of γrevealed two eigenvectors with significant
nonlinear sexual selection (m
1
and m
4
,Table 2A). The dominant
eigenvector of nonlinear selection (m
1
) had a positive eigenvalue,
indicative of disruptive selection, and was heavily weighted by a
positive loading from girth and negative loadings from facial and
vocal masculinity (Table 2A). This result parallels results of the
regression analysis in that it signifies negative correlational selection
between girth and facial and vocal masculinity. The second
significant eigenvector of nonlinear selection (m
4
) had a negative
eigenvalue, indicative of stabilizing selection, and was heavily
weighted by a positive loading from girth (Table 2A). This
combination of significant positive and negative eigenvalues sug-
gests that the fitness surface for female choice is best described as a
multivariate saddle (Fig. 2A). There was also significant linear
selection on m
2
,which was heavily weighted by a positive loading
from height (Table 2A).
3.2. Male contests
Male contests showed directional selection favoring increased
girth and vocal masculinity (Fig. 1,Table 1B). There was also
significant disruptive (positive quadratic) selection on facial mascu-
linity (Fig. 1,Table 1B), as well as negative correlational selection
between girth and facial masculinity (Table 1B).
Canonical analysis of γrevealed two eigenvectors with signifi-
cant nonlinear sexual selection (m
1
and m
4
,Table 2A). The
dominant eigenvector of nonlinear selection (m
1
) had a positive
eigenvalue and was heavily weighted by a positive loading from
girth and a negative loading from facial masculinity (Table 2A). This
eigenvector was also subject to significant positive linear selection
(Table 2A). The second eigenvector of nonlinear selection (m
4
) had
a negative eigenvector and was heavily weighted by positive
loadings from girth and facial masculinity (Table 2A). This
eigenvector also experienced significant positive linear selection
favoring an increase in girth and facial masculinity. As shown for
female choice, the combination of significant positive and negative
eigenvalues suggests that the fitness surface for male contests is best
described as a multivariate saddle (Fig. 2A). There was also
significant negative linear selection on m
3
, which favors increased
vocal masculinity (due to the negative contribution of this trait to
this eigenvector).
Table 1
The vector of standardized linear selection gradients (β) and the matrix of standardized
quadratic and correlational selection gradients (γ) for facial masculinity (FaMasc),
vocal masculinity (VoMasc), height and Girth operating through female choice, male
contests and mating success.
βγ
FaMasc VoMasc Height Girth
a. Fem. choice FaMasc 0.08 0.20
VoMasc −0.02 0.22 −0.01
Height 0.23* −0.15 0.06 0.27
Girth −0.09 −0.39** −0.34* 0.07 −0.07
b. Male cont. FaMasc −0.03 0.11*
VoMasc 0.07* 0.05 −0.04
Height −0.01 −0.04 0.04 0.01
Girth 0.18*** −0.19** −0.02 −0.05 0.05
c. Mat. succ. FaMasc −0.13 0.16
VoMasc −0.07 0.39* 0.02
Height −0.18 −0.22 0.01 0.11
Girth 0.59*** 0.05 −0.16 −0.32* 0.13
Randomization tests: ***pb0.0001, **pb0.001, *pb0.05.
Table 2
The Mmatrix of eigenvectors from the canonical analysis of γin Table 1 for female
choice, male contests and mating success.
MSelection
FaMasc VoMasc Height Girth θ
i
λ
i
a. Fem. choice m
1
−0.68 −0.47 0.07 0.56 −0.07 0.35**
m
2
−0.19 0.21 0.94 −0.17 0.21* 0.16
m
3
0.60 −0.74 0.30 0.07 0.12 −0.09
m
4
0.37 0.44 0.12 0.81 −0.02 −0.22**
b. Male cont. m
1
−0.75 −0.18 −0.04 0.64 0.13** 0.14**
m
2
0.27 −0.34 −0.89 0.16 0.01 0.02
m
3
−0.15 −0.89 0.21 −0.41 −0.14** −0.03
m
4
0.59 0.30 0.41 0.63 0.07* −0.07*
c. Mat. succ. m
1
0.69 0.40 −0.53 0.30 0.16 0.29**
m
2
−0.31 −0.53 −0.44 0.65 0.54** 0.22
m
3
0.15 0.16 0.71 0.67 0.24* −0.14
m
4
0.64 −0.73 0.19 −0.16 −0.17 −0.18*
The linear (θ
i
) and quadratic (λ
i
) gradients of selection along each eigenvector are
given in the last two columns. The quadratic selection gradients (λ
i
)ofeach
eigenvector (m
i
) are equivalent to the eigenvalue. Randomization tests: ** pb.001,
*pb.05.
337A.K. Hill et al. / Evolution and Human Behavior 34 (2013) 334–341
3.3. Female choice vs. contests
The strength and form of linear sexual selection acting on the four
male traits differed significantly between female choice and male
contests (F
4,106
= 3.44, p= 0.011). This was due to selection for
greater height through female choice (F
1,106
= 3.92, p= 0.050) and
greater girth through male contests (F
1,106
= 5.15, p= 0.025). There
was no difference in quadratic sexual selection (F
4,98
= 1.92, p=
0.113). Correlational selection differed between these mechanisms of
sexual selection (F
6,86
= 2.31, p= 0.041) due to selection for
negative covariance between girth and vocal masculinity through
female choice (F
1,86
= 3.99, p= 0.049).
3.4. Mating success
Mating success was measured by participants' self-reported
numbers of sex partners. These numbers were highly correlated
with average estimates made by their male peers (r
52
= 0.47,
pb0.0005), suggesting reliability in assessing mating success. Mating
success exerted directional selection favoring girth, negative correla-
tional selection on girth and height, and positive correlational
selection on facial and vocal masculinity (Table 1C).
Canonical analysis of γrevealed two eigenvectors with significant
nonlinear sexual selection (m
1
and m
4
,Table 2C). The dominant
eigenvector of nonlinear selection (m
1
) had a positive eigenvalue and
was heavily weighted by a positive loading from facial masculinity
and a negative loading from height. The second eigenvector of
nonlinear selection (m
4
) had a negative eigenvector and was heavily
weighted by a positive loading from facial masculinity and a negative
loading from vocal masculinity. As shown for female choice and male
contests, the combination of significant positive and negative
eigenvalues suggests that the fitness surface for mating success is
best described as a multivariate saddle (Fig. 2C). There was also
significant positive linear selection on m
2
and m
3
, which favors
increased girth and decreased vocal masculinity (m
2
) and increased
height and girth (m
3
).
Despite evidence of selection on men's traits through female
choice, linear sexual selection acting on the four male traits differed
significantly between female choice and mating success (F
4,111
=
3.94, p= 0.005). This was because female choice selected for greater
height (F
1,111
= 3.95, p= 0.049), and mating success selected for
greater girth (F
1,111
= 12.30, p= 0.001). Neither quadratic (F
4,103
=
1.75, p= 0.145) nor correlational (F
6,91
= .82, p= 0.561) sexual
selection differed significantly between female choice and mating
success. In contrast, the strength and form of linear (F
4,111
= 2.040,
p= 0.093), quadratic (F
4,103
= .962, p= 0.431) and correlational
sexual selection (F
6,91
= 1.192, p= 0.318) imposed on male traits
through male contests did not differ significantly from that imposed
by mating success.
When mating success was used as the fitness measure and success
under female choice (attractiveness) and male contests (dominance)
were treated as traits, there was directional selection for dominance,
but not attractiveness (Fig. 1,Table 3). Canonical analysis revealed
that dominance loaded positively and attractiveness loaded nega-
tively onto the first eigenvector, which exhibited positive linear
Fig. 2. Correlational selection on eigenvectors m
1
and m
4
(see Table 2) under (A) female
choice, (B) male contests, and (C) mating success.
Table 3
The vector of standardized linear selection gradients (β) and the matrix of standardized
quadratic and correlational s election gradients (γ) for the influence of men's
attractiveness and dominance on mating success.
βγ
Dominance Attractiveness
Dominance .37** .11
Attractiveness .07 −.11 .06
Randomization tests: ** pb.001.
338 A.K. Hill et al. / Evolution and Human Behavior 34 (2013) 334–341
selection. Both dominance and attractiveness loaded positively
onto the second eigenvector, which exhibited positive linear
selection (Table 4).
3.5. Additional analyses
3.5.1. Subjective ratings of faces and voices
Because our measurements of facial and vocal masculinity could
not capture all information available for perceptions of facial and vocal
attractiveness and dominance, we performed a second set of analyses,
similar to those above, substituting for objective masculinity
measurements the subjective assessments of attractiveness and
dominance made by independent raters.
Although facial and vocal attractiveness (Table E2a) and related
eigenvectors (Table E3a: m
1
,m
2
) positively linearly predicted success
under female choice, they did not predict mating success (Tables E2b,
E3b). Again, linear, but not quadratic or correlational, sexual selection
on male traits acting through female choice differed from that acting
through mating success (see ESM).
When ratings of facial and vocal dominance were substituted for
facial and vocal masculinity measurements, girth (Table E4a) and
related eigenvectors (Tables E5a: m
1
,m
3
,m
4
; E5b: m
1
) again linearly
predicted dominance and mating success. As above, there was no
significant difference in linear or quadratic sexual selection acting on
male traits under contests vs. mating success, although there was a
significant difference in correlational selection acting on the covari-
ance between height and vocal dominance and between height and
girth (see ESM).
3.5.2. Sociosexuality
To control for the effects of sociosexual psychology on sexual
behavior, we reran analyses involving mating success including the
summed attitudes and desires components of the SOI-R. Girth,
sociosexual psychology (Table E6), and related eigenvectors (Table
E7: m
1
,m
2
,m
4
) positively linearly predicted mating success.
When mating success was used as the fitness measure and
attractiveness, dominance, and sociosexual psychology were treated
as traits, there was directional selection for dominance, sociosexuality
(Table E8), and an eigenvector onto which dominance and socio-
sexuality loaded heavily (Table E9: m
1
), but not attractiveness (Table
E8). Dominance and sociosexuality also positively interacted in
predicting mating success (Table E8).
4. Discussion
Female choice exerted positive directional selection on height and
stabilizing selection on an eigenvector that was heavily weighted by
girth. These results corroborate previous research finding that women
prefer taller males particularly for short-term mating (Pawlowski &
Jasienska, 2005), and that they prefer men of intermediate brawniness
(Frederick & Haselton, 2007). Moreover, both multiple regression
analysis and canonical analysis indicated selection under female
choice for negative covariance between girth and facial and vocal
masculinity, suggesting that the brawnier a man is, the more
important it is for him to have a feminine face and voice, and vice
versa. Female choice favored more attractive, but not more masculine,
faces and voices, and facial attractiveness became more important as
height increased. These results indicate that beyond height, masculine
features tend not to make independent positive contributions to
success under female choice, suggesting that other factors may have
operated in the selection of masculine traits in men.
In contrast, male contests exerted positive directional selection
on girth and vocal masculinity. This was evident in the results of
both our multiple regression and canonical analyses. These results
support the notion that men's approximately 60% greater muscular-
ity (Lassek & Gaulin, 2009) and voices approximately five standard
deviations lower in pitch and timbre (Puts et al., 2012) evolved in
the context of male contests (Puts, 2010). There was also disruptive
selection for facial masculinity, and selection for negative covariance
between facial masculinity and girth, indicating that men high in
girth were more dominant if their faces were less masculine.
Canonical analysis indicated disruptive selection on an eigenvector
related to high girth and low facial masculinity, and stabilizing
selection on an eigenvector related to high girth and high facial
masculinity. Given little evidence that men generally deferred to, or
that women preferred, men with masculine faces in the present
study, perhaps facial masculinity evolved in men not so much as a
dominance signal or sexual ornament but because robust facial
skeletal structure was protective against facial fractures incurred in
physical fights (Puts, 2010).
The selective surface under female choice differed from those
under both male contests and mating success due to positive
directional selection for height under female choice and for girth
under male contests and mating success. The selective surfaces under
male contests and mating success were largely similar. Overall success
under male contests (male acquaintance-rated dominance) predicted
mating success, but success under female choice (female acquain-
tance-rated attractiveness) did not. Moreover, there was positive
correlational selection on dominance and sociosexual psychology,
such that positive sociosexual attitudes and desires contributed more
to mating success as men's dominance increased. This would be
expected if dominant men can more readily satisfy their interests in
uncommitted sex. Generally, dominance and the traits favored by
male contests predicted mating success, but attractiveness and the
traits favored by female choice did not.
These results suggest stronger sexual selection through male
contests than female choice in the population studied. Much research
in evolutionary psychology states or implies the contrary: stronger
sexual selection in men through female choice (reviewed in Puts,
2010). Yet, male contests tend to evolve in terrestrial species,
especially where females are social, as in humans (D. J. Emlen,
2008;S.T.Emlen & Oring, 1977; Puts, 2010), and frequent or intense
male contests characterize all extant great apes (Plavcan & van Schaik,
1992). Large human sex differences in muscle mass and same-sex
aggression also suggest the importance of male contests in shaping
men's traits (Archer, 2009; Puts, 2010). Thus, the present findings are
predicted from theory, as well as phylogenetic and functional analyses
of men's traits.
At the same time, these results appear incompatible with the
apparent autonomy with which Western women choose their mates.
One possibility is that female choice determines men's mating
success, but women choose dominant men (i.e., men's attractiveness
and dominance are functionally equivalent). However, women
preferred different traits from those favored under male contests,
and dominance rather than attractiveness predicted men's mating
success. Another possibility is that women choose from among
dominant men—that is, men's attractiveness and dominance posi-
tively interact, so that the influence of attractiveness on mating
success increases with increasing dominance. However, in predicting
mating success, we observed no statistically significant selection for
Table 4
The Mmatrix of eigenvectors from the canonical analysis of γin Table 3.
MSelection
Dominance Attractiveness θ
i
λ
i
m
1
.78 -.62 .25* .10
m
2
.62 .78 .28* -.02
The linear (θ
i
) and quadratic (λ
i
) gradients of selection along each eigenvector are
given in the last two columns. The quadratic selection gradients (λ
i
) of each eigenvector
(m
i
) are equivalent to the eigenvalue. Randomization tests: *pb.05.
339A.K. Hill et al. / Evolution and Human Behavior 34 (2013) 334–341
positive covariance between attractiveness and dominance: in fact, if
anything, the correlational selection gradient was negative in sign.
Nevertheless, perhaps women rate men's sexual attractiveness
differently from how they ultimately choose (but see Burriss, Welling,
& Puts, 2011 for correspondence between men's traits and their long-
term mates' preferences). For example, attractiveness ratings may not
adequately capture women's differential resistance to men's seduc-
tion attempts (Gangestad & Eaton, 2013; Kokko, Brooks, Jennions, &
Morley, 2003). Finally, men's dominance may limit female choice in
subtle ways. For example, in the bars, clubs, parties, and other venues
in which sexual affairs are initiated, a dominant man may have little
compunction against interfering with the mating attempts of a less
dominant man, whereas the reverse would be less likely. These
intriguing possibilities deserve future research, but certainly the
present results provide strong evidence that dominance remains
salient in men's competition for mates.
Despite the coherence of these results, we note several limita-
tions. First, although we measured what we believe are some of the
strongest candidates for sexually selected traits in men, traits that
exhibit large sex differences that emerge at sexual maturity and have
been implicated in men's mating competition, we did not assess all
possible traits. Among those that we might have included are
psychological traits, such as aggression (Archer, 2009) and humor
(Miller, 2000). Second, the population that we sampled may differ in
important ways from those in which men's sexually selected traits
were shaped over human evolution. Yet, we note similarities
between some traditional societies and American university social
fraternities: both small groups of allied males with a high degree of
social conformity and an ethos of hegemonic masculinity (Anderson,
2008; Chagnon, 1992) who interact regularly with the same females.
It was also critical that we sample from a population in which
participants view questions on sexual attitudes, desires, and behavior
as minimally invasive or stressful (Yeater, Miller, Rinehart, & Nason,
2012). Third, the use of hormonal contraception may have affected
some female participants' and raters' mate preferences (Roberts,
Gosling, Carter, & Petrie, 2008) and decoupled male participants'
copulatory patterns from their reproductive success. However,
copulatory patterns can predict the reproductive success that
would be realized in the absence of effective contraception (Perusse,
1993). Fourth, our data on mating success were based on self-report,
which may be unreliable. However, we found a highly significantly
correlation between self-reported numbers of sex partners and male
peers' assessments of men's numbers of sex partners. Fifth, although
we measured success under female choice and male contests, sexual
selection in men likely involves other mechanisms, such as sperm
competition and sexual coercion (Goetz & Shackelford, 2006). Finally,
we measured men's mating success by their number of sex partners,
but additional variables are clearly relevant to mating success, such
as the quality of men's mates, the number of copulations with each,
and mates' fecundability at the time. Nevertheless, the number of
women with whom a man has copulated likely strongly reflects his
ability to obtain mating opportunities (Faurie et al., 2004; Hodges-
Simeon et al., 2011).
The present study begins to fill significant gaps regarding the
mechanisms and forms of sexual selection in men and the relative
salience of men's traits to different mechanisms of sexual selection.
We do not, however, consider these questions resolved. Future
research should explore additional traits and other measures of
mating success in different populations, especially in traditional
societies. We hope that the present study can help guide these
future explorations.
Supplementary Materials
Supplementary data to this article can be found online at http://dx.
doi.org/10.1016/j.evolhumbehav.2013.05.004.
Acknowledgments
We thank Roy Baker, Scott Boren, Jay Bundy, Robert Burriss,
Valentina Ceccarelli, Joanna Colgan, Chelsey Culbert, Nate Davis, Helen
Geleskie, Beck Graefe, Jeffrey Krystek, Luke Lolla, Ashley Matz, Coralie
McEachron, Leela McKinnon, Jessica Pavliska, Diana Rosa-Leyra, Kevin
Singh, and Andrew Vrabel for their generous contributions to this
research, and Daniel Nettle and two anonymous reviewers for their
helpful comments.
References
Anderson, E. (2008). Inclusive masculinity in a fraternal setting. Men and Masculinities,
10(5), 604–620.
Apicella, C. L., Feinberg, D. R., & Marlowe, F. W. (2007). Voice pitch predicts
reproductive success in male hunter-gatherers. Biology Letters,3(6), 682–684.
Archer, J. (2009). Does sexual selection explain human sex differences in aggression?
The Behavioral and Brain Sciences,32(3–4), 249–266.
Bisgaard, S., & Ankenman, B. (1996). Standard errors for the eigenvalues in second-
order response surface models. Technometrics,38, 238–246.
Blows, M. W., & Brooks, R. (2003). Measuring nonlinear selection. The American
Naturalist,162, 815–820.
Burriss, R. P., Roberts, S. C., Welling, L. L., Puts, D. A., & Little, A. C. (2011a). Heterosexual
romantic couples mate assortatively for facial symmetry, but not masculinity.
Personality and Social Psychology Bulletin,37(5), 601–613.
Burriss, R. P., Welling, L. L., & Puts, D. A. (2011b). Mate-preference drives mate-choice:
Men's self-rated masculinity predicts their female partner's preference for male
facial masculinity. Personality and Individual Differences,51(8), 1023–1027.
Chagnon, N. A. (1992). Yanomamo. (4th ed.). Fort Worth: Harcourt Brace.
DeBruine, L. M., Jones, B. C., Little, A. C., Boothroyd, L. G., Perrett, D. I., Penton-Voak, I. S.,
et al. (2006). Correlated preferences for facial masculinity and ideal or actual
partner's masculinity. Proceedi ngs of the Royal Society B: Biological Sciences,
273(1592), 1355–1360.
DeBruine, L. M., Jones, B. C., Smith, F. G., & Little, A. C. (2010). Are attractive men's faces
masculine or feminine? The importance of controlling confounds in face stimuli.
Journal of Experimental Psychology. Human Perception and Performance,36(3),
751–758.
Dixson, B. J., Dixson, A. F., Bishop, P. J., & Parish, A. (2010). Human physique and sexual
attractiveness in men and women: A New Zealand–U.S. comparative study.
Archives of Sexual Behavior,39(3), 798–806.
Draper, N. R., & John, J. A. (1988). Response-surface designs for quantitative and
qualitative variables. Technometrics,30(4), 423–428.
Emlen, D. J. (2008). The evolution of animal weapons. Annual Review of Ecology,
Evolution, and Systematics,39, 387–413.
Emlen, S. T., & Oring, L. W. (1977). Ecology, sexual selection, and the evolution of mating
systems. Science,197(4300), 215–223.
Fairbanks, G. (1960). Voice andarticulation drillbook. (2nd ed.). New York: Harper & Row.
Faurie, C., Pontier, D., & Raymond, M. (2004). Student athletes claim to have more
sexual partners than other students. Evolution and Human Behavior,25(1), 1–8.
Feinberg, D. R. (2008). Are human faces and voices ornaments signaling common
underlying cues to mate value? Evolutionary Anthropology,17, 112–118.
Feinberg, D. R., Jones, B. C., Law Smith, M. J., Moore, F. R., DeBruine, L. M., Cornwell, R. E.,
et al. (2006). Menstrual cycle, trait estrogen level, and masculinity preferences in
the human voice. Hormones and Behavior,49(2), 215–222.
Feinberg, D. R., Jones, B. C., Little, A. C., Burt, D. M., & Perrett, D. I. (2005). Manipulations
of fundamental and formant frequencies affect the attractiveness of human male
voices. Animal Behaviour,69, 561–568.
Fink, B., Neave, N., & Seydel, H. (2007). Male facial appearance signals physical strength
to women. American Journal of Human Biology,19(1), 82–87.
Fitch, W. T., & Giedd, J. (1999). Morphology and development of the human vocal tract:
A study using magnetic resonance imaging. The Journal of the Acoustical Society of
America,106(3 Pt 1), 1511–1522.
Frederick, D. A., & Haselton, M. G. (2007). Why is muscularity sexy? Tests of the
fitness indicator hypothesis. Personality and Social Psychology Bulletin,33(8),
1167–1183.
Gangestad, S. W., & Eaton, M. A. (2013). Toward an integrative perspective on sexual
selection and men's masculinity. Behavioral Ecology,24(3), 594–595.
Goetz, A. T., & Shackelford, T. K. (2006). Sexual coercion and forced in-pair copulatio n as
sperm competition tactics in humans. Human Nature,17(3), 265–282.
Green, P. J., & Silverman, B. W. (1994). Nonparametric regression and generalized linear
models. London: Chapman & Hall.
Hodges-Simeon, C. R., Gaulin, S. J., & Puts, D. A. (2010). Different vocal parameters
predict perceptions of dominance and attractiveness. Human Nature,21(4),
406–427.
Hodges-Simeon, C. R., Gaulin, S. J., & Puts, D. A. (2011). Voice correlates of mating
success in men: Examining "contests" versus "mate choice" modes of sexual
selection. Archives of Sexual Behavior,40(3), 551–557.
Hughes, S. M., Dispenza, F., & Gallup, G. G. (2004). Ratings of voice attractiveness predict
sexual behavior and body configuration. Evolution and Human Behavior,25,
295–304.
Hunt, J., Breuker, C. J., Sadowski, J. A., & Moore, A. J. (2009). Male-male competition,
female mate choice and their interaction: Determining total sexual selection.
Journal of Evolutionary Biology,22(1), 13–26.
340 A.K. Hill et al. / Evolution and Human Behavior 34 (2013) 334–341
Johnston, V. S., Hagel, R., Franklin, M., Fink, B., & Grammer, K. (2001). Male facial
attractiveness: Evidence of hormone-mediated adaptive design. Evolution and
Human Behavior,22, 251–267.
Jones, B. C., Feinberg, D. R., DeBruine, L. M., Little, A. C., & Vukovic, J. (2010). A domain-
specific opposite-sex bias in human preferences for manipulated voice pitch.
Animal Behaviour,79,57–62.
Kokko, H., Brooks, R., Jennions, M. D., & Morley, J. (2003). The evolution of mate choice
and mating biases. Proceedings Biological Sciences,270(1515), 653–664.
Lassek, W. D., & Gaulin, S. J. C. (2009). Costs and benefits of fat-free muscle mass in men:
Relationship to mating success, dietary requirements, and natural immunity.
Evolution and Human Behavior,30, 322–328.
Lewis, Z., Wedell, N., & Hunt, J. (2011). Evidence for strong intralocus sexual conflict in
the indian meal moth, Plodia interpunctella. Evolution,65, 2085–2097.
Miller, G. F. (2000). The mating mind: How sexual choice shaped the evolution of human
nature. New York: Doubleday.
Mitchell-Olds, T., & Shaw, R. G. (1987). Regression analysis of natural selection:
Statistical inference and biological interpretation. Evolution,41, 1149–1161.
Nettle, D. (2002). Height and reproductive success in a cohort of British men. Human
Nature,13(4), 473–491.
Pawlowski, B., Dunbar, R. I., & Lipowicz, A. (2000). Tall men have more reproductive
success. Nature,403(6766), 156.
Pawlowski, B., & Jasienska, G. (2005). Women's preferences for sexual dimorphism in
height depend on menstrual cycle phase and expected duration of relationship.
Biological Psychology,70(1), 38–43.
Penke, L., & Asendorpf, J. B. (2008). Beyond global sociosexual orientations: A more
differentiated look at sociosexuality and its effects on courtship and romantic
relationships. Journal of Personality and Social Psychology,95, 1113–1135.
Penton-Voak, I. S., & Perrett, D. I. (2000). Female preference for male faces changes
cyclically: Further evidence. Evolution and Human Behavior,21,39–48.
Penton-Voak, I. S., & Perrett, D. I. (2001). Male facial attractiveness: Perceived
personality and shifting female preferences for male traits across the menstrual
cycle. Advances in the Study of Behavior,30, 219–259.
Penton-Voak, I. S., Perrett, D. I., Castles, D. L., Kobayashi, T., Burt, D. M., Murray, L. K.,
et al. (1999). Menstrual cycle alters face preference. Nature,399(6738), 741–742.
Perrett, D. I.,Lee, K. J., Penton-Voak, I.,Rowland, D., Yoshikawa, S., Burt,D. M., et al. (1998).
Effects of sexual dimorphism on facial attractiveness. Nature,394(6696), 884–887.
Perusse, D. (1993). Cultural and reproductive success in industrial societies: Testing the
relationship at proximate and ultimate levels. The Behavioral and Brain Sciences,16,
267–283.
Phillips, P. C., & Arnold, S. J. (1989). Visualizing multivariate selection. Evolution,43,
1209–1222.
Pillsworth, E. G. (2008). Mate preferences among the Shuar of Ecuador: Trait rankings
and peer evaluations. Evolution and Human Behavior,29, 256–267.
Plavcan, J. M., & van Schaik, C. P. (1992). Intrasexual competition and canine
dimorphism in anthropoid primates. American Journal of Physical Anthropology,
87(4), 461–477.
Puts, D. A. (2005). Mating context and menstrual phase affect women's preferences for
male voice pitch. Evolution and Human Behavior,26, 388–397.
Puts, D. A. (2006). Cyclic variation in women's preferences for masculine traits:
Potential hormonal causes. Human Nature,17(1), 114–127.
Puts, D. A. (2010). Beauty and the beast: Mechanisms of sexual selection in humans.
Evolution and Human Behavior,31, 157–175.
Puts, D. A., Apicella, C. L., & Cárdenas, R. A. (2012a). Masculine voices signal men's threat
potential in forager and industrial societies. Proceedings of the Royal Society B:
Biological Sciences,279(1728), 601–609.
Puts, D. A., Hodges, C., Cárdenas, R. A., & Gaulin, S. J. C. (2007). Men's voices as
dominance signals: Vocal fundamental and formant frequencies influence
dominance attributions among men. Evolution and Human Behavior,28, 340–344.
Puts, D. A., Jones, B. C., & DeBruine, L. M. (2012b). Sexual selection on human faces and
voices. Journal of Sex Research,49(2–3), 227–243.
Rhodes, G., Chan, J., Zebrowitz, L. A., & Simmons, L. W. (2003). Does sexual dimorphism
in human faces signal health? Proceedings of the Royal Society B: Biological Sciences,
270(Suppl 1), S93–S95.
Rhodes, G., Hickford, C., & Jeffery, L. (2000). Sex-typicality and attractiveness: Are
supermale and superfemale faces super-attractive? British Journal of Psychology,
91(Pt 1), 125–140.
Roberts, S. C., Gosling, L. M., Carter, V., & Petrie, M. (2008). MHC-correlated odour
preferences in humans and the use of oral contraceptives. Proceedings Biological
Sciences,275(1652), 2715–2722.
von Rueden, C., Gurven, M., & Kaplan, H. (2011). Why do men seek status? Fitness
payoffs to dominance and prestige. Proceedings of the Royal Society B: Biological
Sciences,278, 2223–2232.
Watkins, C. D., Fraccaro, P. J., Smith, F. G., Vukovic, J., Feinberg, D. R., DeBruine, L. M.,
et al. (2010a). Taller men are less sensitive to cues of dominance in other men.
Behavioral Ecology,21(5), 943–947.
Watkins, C. D., Jones, B. C., & DeBruine, L. M. (2010b). Individual differences in
dominance perception: Dominant men are less sensitive to facial cues of male
dominance. Personality and Individual Differences,49(8), 967–971.
Welling, L. L., Jones, B. C., DeBruine, L. M., Conway, C. A., Law Smith, M. J., Little, A. C.,
et al. (2007). Raised salivary testosterone in women is associated with increased
attraction to masculine faces. Hormones and Behavior,52(2), 156–161.
Wolff, S. E., & Puts, D. A. (2010). Vocal masculinity is a robust dominance signal in men.
Behavioral Ecology and Sociobiology,64, 1673–1683.
Yeater, E., Miller, G., Rinehart, J., & Nason, E. (2012). Trauma and sex surveys meet
minimal risk standards: Implications for institutional review boards. Psychological
Science,23(7), 780–787.
341A.K. Hill et al. / Evolution and Human Behavior 34 (2013) 334–341
