behavioral ecology 2016 bonnet beheco arw023

8/18/2019 Behavioral Ecology 2016 Bonnet Beheco Arw023

1/10

© The Author 2016. Published by Oxford University Press on behalf of

the International Society for Behavioral Ecology. All rights reserved. For

permissions, please e-mail: [email protected]

The official journal of the

ISBEInternational Society for Behavioral Ecology

BehavioralEcology

Original Article

A prison effect in a wild population: a scarcityof females induces homosexual behaviorsin males

Xavier Bonnet,a Ana Golubović,b Dragan Arsovski,a,b Sonja Ðorđević,b Jean-Marie Ballouard,c Bogoljub Sterijovski,d Rastko Ajtić,e Christophe Barbraud,a and Liljana Tomovićb

aCEBC, UMR-7372, CNRS, 79360 Villiers en Bois, France, bFaculty of Biology, Institute of Zoology,

University of Belgrade, 11000 Belgrade, Serbia, cCRCC, SOPTOM, 83590 Gonfaron, France,dMacedonian Ecological Society, Herpetology Group, 1000 Skopje, Former Yugoslav Republic of

Macedonia, and eInstitute for Nature Conservation of Serbia, 11070 Belgrade, Serbia

Received 9 October 2015; revised 29 January 2016; accepted 2 February 2016.

The high frequency of same-sex sexual behaviors (SSB) in free-ranging animals is an evolutionary puzzle because fitness benefits areoften unclear in an evolutionary context. Moreover, the physiological and genetic underpinnings of SSB remain unclear. We exploitedan extraordinary natural experiment to examine the impact of environmental factors (local sex ratio [SR]) and testosterone (T) levels onSSB in a dense population of Hermann’s tortoises monitored for 7 years. Under the combination of high density and extremely skewedSR (~50 females, >1000 males), males courted and mounted other males more frequently than females. They even exhibited extrava-gant sexual behaviors, attempting to copulate with dead conspecifics, empty shells, and stones. T levels remained within the species’normal range of variation. SSB was not observed in other populations where SR is not, or less skewed, and where density is lower.This study reports the first natural example of a “prison effect,” whereby a high population density combined with female deprivation triggered SSB as a mere outlet of sexual stimulation. More generally, it supports the hypothesis that SSB can be a nonadaptive conse-quence of unusual proximate factors rather than reflecting physiological disorders.

Key words : ecophysiology, reptiles, sex ratio, testosterone, tortoise.

INTRODUCTION

Same-sex sexual behaviors (SSB) have been recorded in numer-

ous free-ranging animal species from a wide array of lineages, and

in many cases are common ( Sommer and Vasey 2006; Bailey and

Zuk 2009; Poiani 2010; Scharf and Martin 2013 ). Because sexual

behaviors are tightly linked to procreation and often entail major

survival and energy costs, the high frequency of these apparentlyunproductive behaviors is a biological paradox. Both adaptive and

nonadaptive hypotheses have been proposed to explain the origin

and maintenance of SSB. Possibly due to their greater explanatory

power, adaptive hypotheses (e.g., social glue, intrasexual conflict)

have attracted most of the research ( Camperio-Ciani et al. 2004;

MacFarlane et al. 2007; Bailey and Zuk 2009; Bierbach et al. 2013;

VanderLaan et al. 2014 ). Moreover, SSB is considered not only as

the outcome of selection but also as a selective agent per se ( Bailey

and Zuk 2009 ). Technical difficulties may also explain why investi-

gations of SSB tend to focus on ultimate rather than on proximate

causations. Indeed, sexually active individuals often express a wide

spectrum of behaviors, and SSB can be intermingled with hetero-

sexual behaviors (HSB), posing obstacles to isolate physiological

processes that specifically trigger SSB. Consequently, physiologi-

cal mechanisms underlying adaptive SSB have rarely been experi-

mentally elucidated in the field ( Shine et al. 2000 ), and there are

only limited data on this topic for nonadaptive SSB ( Scharf and

Martin 2013 ).

In captive animals, genetic, anatomical, neurophysiological, or

hormonal investigations have failed to identify clear distinctions

between the mechanisms that promote SSB versus HSB ( Poiani

2010; Hoskins et al. 2015 ). Even in humans, one of the most inten-

sively studied vertebrate species, the abundant efforts to link sexual

orientation with anatomy, physiology, and genetic mechanisms

have revealed complex and equivocal patterns ( Banks and Gartrell

1995; Rice et al. 1999; Zitzmann and Nieschlag 2001; Mustanski

et al. 2002; Jannini et al. 2010 ). Although recent epidemiologic Address correspondence to X. Bonnet. E-mail: [email protected].

Behavioral Ecology (2016), 00(00), 1–10. doi:10.1093/beheco/arw023

Behavioral Ecology Advance Access published March 20, 2016

mailto:[email protected]?subject=mailto:[email protected]?subject=


2/10

Behavioral Ecology

studies revealed that genetic factors influence human sexuality

( Jannini et al. 2015 ), underlying genetic and physiological mecha-

nisms of sexual orientation are not yet understood such that adap-

tive hypotheses remain controversial ( Rahman 2005 ).

Nevertheless, there is no doubt that physiological factors (e.g.,

hormones) influence behaviors and thus SSB ( Nelson 2005;

Goldey and van Anders 2015 ). In vertebrates, investigations have

concentrated on mechanisms that involve sex steroids ( Poiani 2010; Goldey and van Anders 2015 ). Such studies have explored

several experimentally induced disorders (e.g., using knockout

mice; Bakker et al. 2002, 2006 ). Experimental masculinization

can induce SSB but also perturbs fertility and reproductive behav-

iors, with hyperaggressiveness or behavioral asexuality as possible

outcomes ( Balthazart, et al. 1997 ). Consequently, such disorders

represent reproductive impasses and are unlikely to be selected

in free-ranging animals. Moreover, empirical results suggest that

male vertebrates that engage in SSB do not exhibit behavioral

anomalies, are able to discriminate between the sexes, reproduce

normally, and do not display unusual circulating testosterone

levels (T) ( Poiani 2010 ). Overall, available results regarding hor-

mones and sexual behaviors provide little insights into the preva-

lence of SSB in natural populations. However, the difficulty of

identifying proximate physiological mechanisms does not mean

that sexual hormones are not key factors. Examining these issues

remains important to test adaptive versus nonadaptive causes, to

understand why individuals engage in SSB in natural populations,

and thus to frame physiological investigations in a natural non-

pathological context.

We benefited from a quasi-experimental natural situation to

investigate one of the main nonadaptive hypotheses proposed by

Bailey and Zuk (2009): The “prison effect” where “depriving indi-

viduals of members of the opposite sex causes them to engage in

sexual interactions with members of the same sex.” This notion

received support in a recent review on insects studied in captivity:

The paucity of sexual partner (i.e., one sex kept in isolation) com-bined with high population density and exposure to sexual chemi-

cal stimulation is assumed to trigger SSB ( Scharf and Martin 2013 ).

However, comparable data are lacking in vertebrates or in natural

populations. In the current study, a strong spatial and population

contrast allowed us to explore the effect of environmental (female

scarcity) and physiological (T levels) factors on the high SSB fre-

quency exhibited by free-ranging males.

MATERIALS AND METHODS

Studied species

The Hermann’s tortoise ( Testudo hermanni ) is a medium-sized

Mediterranean species that comprises 3 subspecies, one (e.g.,Testudo hermanni boettgeri ) is locally abundant in the eastern part of

its distribution range ( Bertolero et al. 2011 ). This species exhib-

its typical tortoise sexual behaviors: Males display a repertoire

of behaviors (e.g., head bobbing, biting) to court and (often forc-

ibly) mount females ( Hailey 1990; Sacchi et al. 2013 ). During

copulation, they produce characteristic high pitch vocaliza-

tions ( Galeotti et al. 2007 ). Females usually try to escape before

accepting copulation, sometimes flipping males onto their backs.

Male rivals often engage in vigorous combats characterized by

shell ramming, severe bites, and attempts to flip back opponents,

but without mounting or vocalization. Because adult chelonians

are sexually dimorphic ( Bonnet et al. 2010 ), including the studied

subspecies ( Djordjević et al. 2011, 2013 ), sex can be accurately

determined and thus SSB can be unambiguously distinguished

from HSB.

SSB were never observed in the thousands of free-ranging indi-

viduals of other Hermann’s tortoise populations ( T. h. boettgeri and

Testudo hermanni hermanni ) or in other Testudo species studied over

more than 20 years ( Testudo graeca ; Testudo horsfieldi ; Bonnet et al.

2001; Lagarde et al. 2003, 2008 ). Additionally, despite the very

abundant literature, we are not aware of published SSB in free-ranging tortoises.

Study sites and study population

A large population was monitored for 7 years ( 2008 –2014) on

Golem Grad Island (Former Yugoslav Republic of Macedonia, ~18

ha; N40°52′; E20°59′ ) by 2–12 people (6 on average) during 13

field sessions (total 137 searching days). The location and behavior

of each tortoise were first recorded in order to limit disturbance by

the observer; individuals were then captured, sexed, permanently

marked using a notche-code on the marginal-scutes, and rapidly

released ( Djordjević et al. 2011; Golubović et al. 2013 ). During

resightings, behavior was also recorded first; individuals were then

recaptured to check identity. Importantly, sexual activity, althoughinfluenced by climatic conditions (Tomović L, Bonnet X, Sterijovski

B, Ðorđević S, unpublished data), was observed during the entire

activity period, from April to September. We marked 1737 tor-

toises: 1208 adults (and 529 juveniles, discarded from analyses) and

collected abundant recaptures ( N = 7829).

On Golem Grad Island, a very particular landscape ( Figure 1 ) splits

the population into 2 almost-distinct subpopulations. Approximately

20% of the adults live on the shoreline (~1 ha), which is separated by

vertical cliffs (10- to 25-m high) from the plateau subpopulation (~17 ha;

71% of individuals). A few tortoises (9%, mostly males [79%]) have been

recorded to commute between subpopulations through 2 narrow steep

paths (Tomović L, Bonnet X, Sterijovski B, Ðorđević S, unpublished

data). For unknown reasons (perhaps reflecting temperature-dependent

sex determination coupled with a scarcity of potential nest sites that provide daughter-producing temperatures; Pieau 1971 ), the SR is highly

male biased on the shoreline subpopulation (23% adult females) and

even more strongly male biased on the plateau (5% adult females).

As in other Testudo populations, males patrol their habitat, search

for females, and harass them to copulate, whereas females never

chase conspecifics ( Hailey and Willemsen 2000 ). The very high pop-

ulation density (~67 adults per hectare) means that many substrate-

deposited pheromonal trails intersect in the field whereas body size

and coloration overlap between sexes, reducing male’s ability to

locate females using olfactory or visual clues. Consequently, for a

male tortoise locating a female is like looking for a needle in a hay-

stack of rivals, notably for males living on the plateau where adult

females are very scarce. In contrast, shoreline tortoises are confinedwithin a narrow 1-ha zone where female density is 10 times greater

than on the plateau. Thus, shoreline males can find females more

easily, but they cannot avoid many male–male encounters. We

observed a high frequency of male SSB in the 2 subpopulations of

Golem Grad (see Results for details).

A neighboring population situated on the mainland (Konjsko, 4

km away from the island, 407 individuals marked; 322 adults; ~20

ha) was monitored between 2010 and 2014 using similar method.

Although not the main focus of this study, it provided comparative

data regarding possible effects of SR and T levels on sexual behav-

iors. In this neighboring mainland population, the SR is close to 1:1

(53% adult females) and density (~20 individuals per hectare) lower

compared with Golem Grad. We never observed any male SSB in

Page 2 of 10


3/10

Bonnet et al. • Same-sex sexual behaviors in wild tortoises

this population, whereas classical HSB (e.g., courting, mating) were

frequent (e.g., N = 45).

Adapting behavioral data to a capture–mark–recapture framework

Capture–mark–recapture (CMR) analyses have been developed to

assess demographic parameters (e.g., population size, annual sur-

vival) of populations when the detectability of individuals is imper-

fect ( Jolly 1965; Seber 1965 ). The encounter history of each animal

is implemented as a basic input in the analyses, thus, estimates of

model parameters can be performed using numerical maximum

likelihood techniques ( Lebreton et al. 1992 ). Various factors and

their interactions can be modeled, individuals can be assigned into

groups and covariates can be used. The robustness and flexibility

of this approach explain the exponentially increasing popularity

of CMR analyses in the ecological literature ( Kendall and Nichols

2002; Choquet, Lebreton, et al. 2009 ). Because the probability to

observe a given behavior in the field is subjected to survival, sight-

ing, and observation biases, we adapted the CMR approach to our

behavioral observations. In other words, we evaluated the annual

probabilities of male tortoises to engage in SSB using the indi-

vidual CMR data from 2008 to 2014. Indeed, many individuals

were observed repeatedly during the study (resighted 5.5 ± 3.9 times

on average, range 1–27), enabling us to build individual capture/

behavioral histories ( Supplementary Annex 1 ) and to use matrix

analyses where transition probabilities among successive behaviors

within the histories can be estimated.

In the current study, one crucial aspect of the construction of

the CMR dataset was the assignment of each individual’s behav-ioral capture history to a biologically relevant group allowing us

to take into account the effects of sex and site: males on plateau,

females on plateau, males on shoreline, and females on shoreline

( Supplementary Annex 2 ). Site assignment was based on the place

of capture of each individual during sampling sessions. Data for

the few individuals captured within the 2 narrow paths, commuting

between the plateau and shoreline, were discarded. The assignment

of each individual to a site allowed us to include SR and density

into modeling. Although possible interaction between SR and den-

sity could not be disentangled, the principal difference between the

2 subpopulations was due to skewed SR (5.4 times more skewed on

the plateau than on the beach; this difference was only 2.1 for den-

sity). Consequently, our null hypothesis was a lack of difference in

the probability to exhibit SSB between plateau and shoreline males.

Conversely, a significant site effect would suggest an influence of

SR and density.

The dataset consisted of 1208 adult capture histories (we

retained only 1 observation per year and per individual). In case of

multiple resighting of individuals within a given year, we retained

the most informative behavior. For example, SSB was preferred

over heterosexuality because this study focuses on SSB. In practice,

this selection was applied on 13 occasions only, and possible asso-

ciated bias was negligible considering the large number of sexual

behaviors observed (removing these 13 cases from the analyses did

not change the results).

Figure 1

Golem Grad island is an 18-ha natural prison where tortoise population density is high, ~60 individuals per hectare. (a) Vertical cliffs separate a plateau

subpopulation (highly skewed SR: 5% of females) from a narrow shoreline subpopulation (skewed SR: 23% of females) of Hermann’s tortoises (photographed

by X.B.). (b) A male mounting (+ vocalization) another male, which himself is mounting a juvenile male (left) (photographed by A.G.). (c) A male mounting a

stone (center) (photographed by A.G.). (d) A male mounting a female (right) (photographed by X.B.).

Page 3 of 10

http://beheco.oxfordjournals.org/lookup/suppl/doi:10.1093/beheco/arw023/-/DC1http://beheco.oxfordjournals.org/lookup/suppl/doi:10.1093/beheco/arw023/-/DC1http://beheco.oxfordjournals.org/lookup/suppl/doi:10.1093/beheco/arw023/-/DC1http://beheco.oxfordjournals.org/lookup/suppl/doi:10.1093/beheco/arw023/-/DC1


4/10

Behavioral Ecology

Designing the multistate mark–recapture model

Over time, individuals exhibit different behaviors, shifting from sex-

ual activity to inactivity, etc. Therefore, we used a multistate mark–

recapture (MSMR) model to estimate transition probabilities among

behaviors at each sighting occasion (e.g., shifts from SSB to HSB).

MSMR can distinguish between the behavioral states within a CMR

probabilistic framework in order to take into account imperfect

detectability of tortoises during capture sessions in the field ( Burnham

and Anderson 2002 ). This CMR approach allowed us to analyze

the annual probability of detecting different behaviors, while taking

into account possible differences in detectability and survival between

years, sites, sexes, and states. It also allowed us to take into account

the previous behavioral state of individuals (e.g., successive behav-

iors might not be independent), which might be important to obtain

unbiased estimates of demographic parameters ( Lebreton and Pradel

2002; Choquet, Lebreton, et al. 2009; Choquet, Rouan, et al. 2009 ).

In practice, from an initial departure state of initial state, individ-

uals could be categorized in 4 distinct states (see Table 1 for details):

sexually inactive or sexually passive (SP, individual exhibiting no

sexual behavior or passive during sexual behavior), sexually active

(SA, individual displaying active HSB or involved in male-to-malecombat), SA with the same sex (SSB, individual observed in an

active homosexual behavior, essentially a male mounting another

male, thus excluding combats), and dead (†). Three state transitions

were modeled to account for survival probability between capture

occasions (matrix S; Supplementary Annex 2 ): S SP, S SA, and S SBB

were the respective survival probabilities of SP (line 1), SA (line 2),

and SSB (line 3) tortoises. Six state transitions were implemented

to model behavioral shifts between capture occasions (matrix Ψ;

Supplementary Annex 2 ): ψSP–SA and ψSP–SSB were the respective

probabilities that a SP tortoise would become SA or display SSB

(line 1); ψSA–SP and ψSA–SSB were, respectively, the probabilities that

a SA tortoise would become SP or SSB (line 2); ψSSB–SP and ψSSB-SA

were the respective probabilities that a tortoise displaying same sex

behavior would become SP or SA (line 3). Detection probability

was modeled as follow (matrix P; Supplementary Annex 2 ): P SP,

P SA, and P SSB were the probabilities that a tortoise was recaptured

at a capture occasion and observed as SP, SA, or displaying SSB,

respectively. Note that the state “dead” was not observable. This

is an absorbing state representing death (there was no emigration

from the island), used to ensure that model parameters would be

estimable ( Lebreton and Pradel 2002; Pradel 2005; Choquet,

Lebreton, et al. 2009; Choquet, Rouan, et al. 2009 ).

Building biological scenarios

Survival ( S ) was constrained to be constant over time and state but

was allowed to vary with sex and site. Transitions between sexual

behaviors ( ψ ) were held constant over time but they were sex, site,

and state dependent. Detection probability ( P ) was time, sex, site,

and state dependent to allow us to take into account variations in

detectability. Our initial model was thus S (sex × site) ψ(sex × site × state) P (sex

× site × state × t). We then built alternative models to test specific hypoth-

eses by using a backwards stepwise procedure while removing

effects from each parameter in order to obtain the best linear com-

bination at each step. Model selection was first performed on detec-

tion probability by testing time, sex, site, and state dependencies.

We then tested for sex and site dependencies for survival. Finally,we tested for site dependency on transition probabilities between

behavioral states, thus testing for an effect of SR and density.

We tested the goodness-of-fit of the time-dependent MSMR model

using U-CARE ( Lebreton and Pradel 2002; Choquet Lebreton, et al.

2009 ). Model selection was based on the quasi-likelihood counterpart

of Akaike’s information criterion (QAIC) and all models were run

under program E-SURGE 1.9.0 ( Lebreton and Pradel 2002; Choquet,

Rouan, et al. 2009 ). Resighting probability decreased during the study

period for all states ( Figure 2a ) due to decreasing search effort over

time and was influenced by the type of behavior exhibited ( Table 2,

Figure 2a ), but did not vary between subpopulations or sexes.

Testosterone and behaviors

To assign a precise behavior with T level, we rapidly took a blood

sample (less than 3 min on average, range 0.2–7 min) on a random

Table 1

Tortoise behaviors observed in the field

Behavior Definition Category N SSB (%)

Burrowed Partly burrowed, head and legs retracted Inactive 677In the shade Motionless in the shade Inactive 2513Basking Fully exposed to sunrays, legs deployed Active 1729Foraging Eating plants, fungi, or animals Active 195Hiding Moving toward a shelter Active 718Walking Walking alone Active 755Chased Trying to escapea from the sexual assaults of a male Sexual (P) 30 25 (83)

Chasing Male pursuing a conspecific Sexual (A) 31 25 (81)Courted Courted by a male Sexual (P) 22 10 (45)Courting Male courting a conspecific, mounting attempts Sexual (A) 22 10 (45)Fighting Male to male combat, violent shell shocks, and bites Sexual (A) 30 0 (0)Sex group Sexually active group, e.g., 2–5 males courting a female Sexual 101 36 (36)Mounting Male mounting any conspecific, vocalizing, ejaculation Sexual (A) 667 506 (76)Necrophilia Male mounting a dead tortoise, vocalizing Sexual (A) 5 5 (100)Skeletophilia Male mounting an empty shell (bones), vocalizing Sexual (A) 5 5 (100)Sex train Several males forming a line while pursuing a female Sexual 3 0 (0)Not classified Information missing, ambiguous behavior (e.g., partly shaded), dead individual 2075

Behaviors (resightings of 1737 individuals, 2008–2014) were classified into 3 main categories: inactive, active, or sexual; (A) and (P) mean, respectively, sexuallyactive versus sexually passive. Sample size ( N ) indicates the total number of observations; for example, 15 pairs of fighting males involve 30 individuals (someinformation was missing). SSB (%) indicates the actual number and proportion of SSBs observed; these involved only males (except for one female mountinganother female).aPossible confusion with a male breaking the fight. Two males courting stones were included in the not-classified category.

Page 4 of 10

http://beheco.oxfordjournals.org/lookup/suppl/doi:10.1093/beheco/arw023/-/DC1http://beheco.oxfordjournals.org/lookup/suppl/doi:10.1093/beheco/arw023/-/DC1http://beheco.oxfordjournals.org/lookup/suppl/doi:10.1093/beheco/arw023/-/DC1http://beheco.oxfordjournals.org/lookup/suppl/doi:10.1093/beheco/arw023/-/DC1http://beheco.oxfordjournals.org/lookup/suppl/doi:10.1093/beheco/arw023/-/DC1http://beheco.oxfordjournals.org/lookup/suppl/doi:10.1093/beheco/arw023/-/DC1


5/10


subsample of adults monitored for the CMR study on Golem

Grad. We notably ensured that the individuals did not detect the

observers (i.e., no sign of disturbance) at capture. This fast proce-

dure limited the possible rapid negative effect of stress on steroid

levels ( Michel and Bonnet 2014 ), thus enabling us to consider that

T levels and behaviors were recorded simultaneously. We also col-

lected blood samples in adult tortoises from the nearby mainland

Konjsko population using the same method. In total, we collected283 blood samples ( N = 216 on Golem Grad [31 females and 185

males] and N = 67 on Konjsko [46 females and 21 males]).

Blood samples (100–300 L, adjusted to tortoise body size) were

taken during activity time (morning and afternoon) from the jugu-

lar vein or cervical sinus using very small needles (27–30 G) and

slightly heparinized 1-mL syringes. Blood was centrifuged in the

field, plasma samples were immediately stored in liquid nitrogen,

and then at −25 °C in the laboratory until assays. Plasma T levels

were measured by radioimmunoassay at the CEBC ( Lagarde et al.

2003 ) on 50 L of plasma after diethyl ether extraction (efficiency

0.93 ± 0.1, standard deviation [SD]). Cross-reactivity of the spe-

cific antibody (Sigma Laboratory) with other steroids was low (B/

B0: 5α-dihydrotestosterone: 17.8%; 5β-androstene-3β, 17β-diol:

1.4%; 5α-androstene-3β, 17β-diol: 1.2%; androstenedione: 1.4%;

epitestosterone: 0.7%; progesterone: 0.07%). The assay’s sensitivity

was 7.8 pg by tube, that is, 156 pg/mL. Intra-assay and interassay

variations were 6% and 12%, respectively.

Ethical note

All procedures were performed in accordance with Macedonian,Serbian, and French guidelines and regulations. No individual was

mistreated or injured. Permits no. 03-246, 353-01-46/2012-03 and

no. 11-4093/5 to perform population monitoring were issued by Galičica

National Park authorities (FYROM) and the Ministry of Environment,

Mining and Spatial Planning (Serbia). Ethical procedures were approved

by the ethical committee COMETHEA (permit no. CE2013-6).

RESULTS

Sexual behaviors

Between 2008 and 2014, we collected 7503 behavioral observations

on Golem Grad ( Table 1 ). Adult females were more frequently

1.1

0.20

0.18

0.16

0.14

0.12

0.10

0.08

0.06

0.04

0.02

0.00

P r o b a b i l i t y t o s h i f t f r o m

s e x u a l l y i n a c t i v e t o S S B

f r o m t

t o t + 1

Plateau Shore

1.0

0.9

0.8

0.7

0.6

0.5

0.4

0.3 S i g h t i n g p r o b a b i l i t y

0.2

0.10.0

2009

Active SSB

(a) (b)

Active heterosexual

Sexually inactive

2010 2011 2012 2013 2014

Figure 2

Annual resighting probabilities (mean ± standard error [SE]) of adult male tortoises as a function of their behaviors. (a) Active heterosexual refers to

males courting females or involved in male-to-male combats; active SSB refers to males courting other males (e.g., mounting); and sexually inactive includes

individuals observed: 1) inactive (e.g., resting), 2) active but not involved in sexual activity (e.g., feeding), and 3) SP (e.g., mounted) (see Methods for details).

(b) Annual probabilities for a male in year t (sexually inactive) to engage in SSB in year t + 1 in each of the 2 subpopulations of Golem Grad (mean ± SE).

Table 2

MSMR analyses of male tortoise behaviors

Model np Deviance QAIC ∆QAIC

Modeling resighting probability S (sex × site) ψ(sex × site × state) P (sex × site × state × t) 1 90 10 220.4 10 400.4 43.2 S (sex × site) ψ(sex × site × state) P (sex × site × state) 2 34 11 092.4 11 160.4 803.2 S (sex × site) ψ(sex × site × state) P (sex × state × t) 3 58 10 257.3 10 373.3 16.1 S (sex × site) ψ(sex × site × state) P (state × t) 4 40 10 277.2 10 357.2 0 S (sex × site) ψ(sex × site × state) P (t) 5 28 10 391.5 10 447.5 90.3Modeling survival probability S (sex) ψ(sex × site × state) P (state × t) 6 38 10 218.3 10 357.3 0.1 S (.) ψ(sex × site × state) P (state × t) 7 37 10 305.2 10 379.2 22Modeling transition probability S (sex × site) ψ(sex × state) P (state × t) 8 34 10 290.1 10 358.1 0.9 S (sex × site) ψ(site × state) P (state × t) 9 37 10 284.9 10 358.9 1.7

Modeling probabilities of survival ( S ), resighting ( P ), and transition between behavioral states ( ψ ) of Hermann’s tortoises on Golem Grad, 2008–2014. Theselected model is indicated in bold. The initial model (model 1) fitted to the data ( χ105

2

= 110.2, P = 0.345). np, number of parameters; sex, sex dependence;site, site dependence; state, state dependence; t, time dependence; “×”, interaction; “.” constant parameter.

Page 5 of 10


6/10

Behavioral Ecology

inactive (i.e., partly sheltered and motionless) compared with adult

males ( χ2 = 30.6, P < 0.001, Table 1 ).

We observed many classical HSB and male-to-male combats,

but most of the sexual behaviors recorded on this small island were

SSB ( Table 1 ): For instance, on 295 occasions, a male was observed

mounting another male (with vocalizations and/or ejaculation).

SSB (i.e., mounting + courting and so on, Table 1 ) largely domi-

nated among plateau-males (74% SSB, N = 786), whereas HSBwere more frequent among shoreline-males (33% SSB, N = 102).

SSB was exhibited by different individuals ( N = 363) and thus was

attributable to males at large (not to few individuals).

CMR modeling results suggest that resighting probability was low-

est for SA individuals involved in HSB, highest for sexually inactive

individuals, and intermediate for individuals displaying SSB ( Table 2,

Figure 2a ). Transition probabilities between SSB versus HSB were

sex and site dependent ( Table 2, model 4). However, ∆QAIC between

models 4, 6, and 8 were low. We selected model 4 (lowest absolute

QAIC) for biological reasons ( Burnham and Anderson 2002 ). Indeed,

this model takes into account differential survival between sites, and

this conforms, for example, to the fact that females face greater envi-

ronmental difficulties on the plateau compared with the shoreline

(e.g., water unavailability during summer). Furthermore, by selecting

the second best model (6), the transition probabilities between sexual

behaviors were still site and sex dependent.

Individual males were not strictly homosexual or heterosexual,

shifting between SSB and HSB. However, the annual probabilities

for a sexually inactive male in year t to engage in SSB in year t +

1 were significantly higher on the plateau than on the shoreline

( Figure 2b ).

We estimated the annual number of heterosexually active indi-

viduals versus individuals exhibiting SSB using raw numbers of

individuals captured each year along with their capture probabili-

ties from the best-supported model ( Table 2 ). In order to estimate

confidence intervals, we used parametric bootstrapping. Each

year on average, the total estimated number of adult tortoises was

1080 ± 40 (standard error); 97 ± 27 displaying HSB versus 139 ± 34

exhibiting SSB.Following the same approach, we also estimated the total popu-

lation size and the numbers of females and males. Mean annual

population size was 1038 ± 228 adult tortoises, 56 ± 15 females, and

979 ± 228 males. The mean estimated SR was 17.5.

Testosterone

We first compared T levels between sexes and mainland/island

populations (using generalized linear model [GLM; Statistica 12.0,

StatSoft 2013, www.statsoft.fr ] [log Poisson model] with sex [male/

female] and population [mainland/island] as fixed effects and

testosterone level as the dependent variable). Mean T levels dif-

fered between sexes (Wald test = 215.82, P < 0.001) and between

mainland and island populations (Wald test = 24.81, P < 0.001).

Females displayed markedly lower T levels compared with males;

the difference between mainland and island populations appeared

weak ( Figure 3 ). However, removing the population factor from the

model decreased adjustment quality (AIC increasing from 1015.81

to 1036.47, δ AIC > 2), suggesting a significant effect. This popu-

lation effect was caused by a difference between mainland and

island males: T levels were slightly albeit significantly lower on

Golem Grad (4.6 ± 3.5 [SD] ng/ml), compared with Konjsko males

12

10

8

6 a

123

a

17

bc

26ab

13

c

3

31

KONJSKOGOLEM GRAD

bc abc

21

13 4

4

2

0

A L O N

E

A L O N

E

P A S S I V E S

S B

A C T I V E S

S B

H E T E

R O M A T I N

G

H E T E

R O M A T I N

G

C O M B A T

F E M A L E

F E M A L E

BEHAVIOUR AT BLOOD SAMPLING

T E S T O S T E R O N E P L A S M A

C O N C E N T R A T I O N ( n g / m l )

Figure 3

Mean plasma levels of testosterone (± standard error [SE]) and sexual behaviors in randomly sampled free-ranging adult tortoises from Golem Grad Island

(circles, left panel) and from a neighboring mainland population (Konjsko, diamonds, right panel). The vertical dashed line facilitates the distinction between

island and mainland populations. Numbers above symbols stand for sample sizes. Males found alone (ALONE) or mounted by another male (PASSIVE SSB)

were considered as SP (gray circles and gray diamond). Males mounting another male (ACTIVE SSB), mounting a female (HETERO MATING), or during

male-to-male combat (COMBAT) were considered as SA (black circles and black diamond). Females (open circle and open diamond) were not observed

SA. Note that SSB were observed only on Golem Grad. Strong sex differences and population effects were observed (see text for details). Letters denote

significance differences among groups of males.

Page 6 of 10

http://www.statsoft.fr/http://www.statsoft.fr/


7/10


(6.9 ± 3.3 ng/ml; Tukey post hoc test P < 0.01); no difference was

observed in females ( P = 0.916).

SSB and other SA behaviors (e.g., courting, mounting, fighting),

and high T levels were exclusively observed in males. Therefore, we

focused on this sex. Breaking up the males’ behaviors into catego-

ries was required to gain a more fine-scale view of how T related to

different behaviors. Using the main behaviors exhibited by males,

we distinguished 5 types of behaviors: males found alone, mountedby another male, mounting another male, mounting a female, or

during male-to-male combats ( Figure 3 ). A GLM revealed a sig-

nificant effect of behavior types on T levels (Wald test = 38.41,

P < 0.001) and a significant difference between mainland and

island populations (Wald test = 16.87, P < 0.001; as above AIC

was lower in the full model, δ AIC = 13.7). Further analyses (com-

paring all parameters estimated or using post hoc tests) suggested

that SP males (alone or mounted) exhibited lower T levels com-

pared with males involved in the most vigorous sexual behaviors

(combat) whereas other active males (mounting) displayed inter-

mediate T levels ( Figure 3 ). Because SSB was observed exclusively

in Golem Grad males, we restricted some analyses to these indi-

viduals: Results were unchanged (GLM analysis revealed a strong

effect of behavior type, Wald test = 37.37, P < 0.001). Thus, we

observed a trend of higher T levels in those males sampled dur-

ing most demanding sexual behaviors, without an influence of SSB.

Furthermore, the significant effect of mainland/island populations

and examination of Figure 3 suggest that this trend was specific to

Golem Grad.

Sample sizes were small in males from the continent and in

fighting island males, shedding uncertainty on some of the results

above. Therefore, to examine the main effect (i.e., higher T level

in the most active males) with sufficient statistical power, we used

broad behavioral categories: SA ( N = 46) versus sexually inactive

males ( N = 157). Mean T levels differed between these groups

(GLM, Wald test = 19.86, P < 0.001) with a mainland/island effect

(GLM, Wald test = 20.77, P < 0.001). Overall, our results suggest

a relationship between T level and sexual activity in males without

specific effect of SSB.

DISCUSSION

The expression of any phenotypic trait, including sexual behav-

iors, can be examined at proximate and ultimate levels. Laboratory

studies of SSB in vertebrates suggest that intrinsic drivers, prenatal

factors notably, influence sexual orientation in adults ( Balthazart

2011 ). In natural populations, mostly adaptive hypotheses have

been proposed for the high frequency of SSB observed in differ-

ent taxa ( Bailey and Zuk 2009 ). Yet, a recent review on arthropods

revealed that SSB are principally displayed by sexually stimulated

males and result essentially from nonadaptive mistaken identifica-

tions, especially under high density and strongly skewed SR ( Scharf

and Martin 2013 ). Our data collected in wild tortoises monitored

during 7 years support this option where SSB is triggered by spe-

cific environmental factors and where adaptive explanations play a

limited role.

In free-ranging male Hermann’s tortoises, high population den-

sities combined with strongly male-biased SRs possibly generated a

“prison effect” ( Bailey and Zuk 2009 ) and SSB became more com-

mon than HSB in this situation only ( Table 3 ). Hormonal investiga-

tions suggested that the high expression of SSB correlated with high

T levels. These high T values remained within the range of natural

Table 3

Population density (number of individuals per hectare [ D]), numbers of adults sampled ( N ), and SR (adult males/adult females)

among 16 populations/subpopulations of Testudo hermanni boettgeri monitored in Greece (data from the Tables 2 and 3 in Haileyand Willemsen 2000) and in Macedonia (Golem Grad, Konjsko) and in 3 other species of Testudo (Testudo hermanni hermanni ,Ballouard JM, unpublished data; Testudo horsfieldii , Bonnet et al. 2001; Lagarde et al. 2002; and Testudo graeca, Lagarde et al. 2008;Bonnet X, unpublished data)

Species Site D N SR SSB % HSB

T. h. hermanni Callas 1.2 258 0.93 0 27T. h. hermanni Saint Daumas 1.2 123 0.84 0 71T. h. hermanni Lambert 1.6 64 1.21 0 4T. h. boettgeri Deskati 3.7 201 4.02 0 — T. h. hermanni Redon 4.2 105 1.55 0 29T. h. boettgeri Mikra Volvi 4.6 338 3.69 0 — T. h. hermanni Riaux 5.2 78 1.69 0 27T. h. boettgeri Litochoron 5.3 102 2.09 0 — T. horsfieldii Bukhara 5.4 354 1.21 0 314T. h. boettgeri Langadia 5.9 71 0.55 0 —

T. graeca Marrakech 5.9 192 1.21 0 192T. h. boettgeri Kilkis 11.3 99 2.81 0 — T. h. boettgeri Parga 12 187 0.83 0 — T. h. boettgeri Olympia 15.4 1378 1.61 0 — T. h. boettgeri Kalamata 18.2 262 1.65 0 — T. h. boettgeri Agios Dimitrios 20 230 0.84 0 — T. h. boettgeri Konjsko 20 322 1 0 45T. h. boettgeri Meteora 20.2 4855 1.9 0 — T. h. boettgeri Kastoria 20.5 265 1.85 0 — T. h. boettgeri Sparta 20.6 814 2.31 0 — T. h. boettgeri Igoumenitsa 39.6 237 6.41 0 — T. h. boettgeri Golem Grad Plateau 59.7 1075 18.19 74 204T. h. boettgeri Golem Grad Beach 130 131 3.36 33 68

SSB were observed in males only (expressed as % relative to all sexual behaviors recorded). The last column indicates the number of HSB observed. Populationswere ranked according to increasing density.

Page 7 of 10


8/10

Behavioral Ecology

variations however (e.g., Golem Grad vs. Konjsko), thus SSB did

not require specific adaptive or pathological explanation. In both

the mainland and island populations, T levels were high in adult SA

males, in accordance with the paradigm that androgenic steroids

(e.g., T, 11-keto testosterone) are primary stimulators of sexual activ-

ity in male vertebrates. Moreover, the most SA males (i.e., mating

with females, engaged in SSB or combats, Figure 3 ) exhibited higher

T levels, as observed in other reptiles ( Aubret et al. 2002; King and Bowden 2013 ). Because an individual is not continuously SA, some

of the “inactive” individuals monitored during random-focal blood

sampling undoubtedly were SA animals that we observed between

bouts of reproductive behavior (especially in mainland tortoises

where lower density decreases encounter rate). Nonetheless, we

recorded higher T in males displaying the most vigorous behaviors:

fighting or mounting other males. Therefore, this suggests that the

combination of high density and skewed SR led to frequent SSB in

males, notably in individuals strongly stimulated by high T levels.

External sexual stimuli (e.g., visual/chemical signals from conspe-

cifics) can activate the brain structures involved in sexual arousal in

adults ( Arnow et al. 2002 ), high density promotes courtship behaviors

in male tortoises ( Hailey and Willemsen 2000 ), and T level correlates

with sexual reactiveness ( Amstislavskaya and Popova 2004; Goldey

and van Anders 2015 ). On Golem Grad, these factors were strongly

amplified in males. Under strong sexual stimulation caused my multi-

ple factors (high density + scarcity of females + high T levels), males

attempted to copulate with the first encountered “partner,” generally

females on the shoreline but more often with males on the plateau

(video in Supplementary Material ). A relatively similar behavioral

“libido syndrome” where the occurrence of heterosexual courtships

positively correlates with SSB has been described in captive cock-

roaches ( Logue et al. 2009 ). Interestingly, libido scores also correlated

with mating success, suggesting that selection for high heterosexual

courtship intensity entailed nonadaptive SSB ( Logue et al. 2009 ).

The studied population presents several exceptional characteris-

tics and thus suffers from a lack of replication. Experiments wheredensity and SR are manipulated are needed to gauge the validity

of our conclusions. However, our results are based on a large sam-

ple size and long-term monitoring. Furthermore, the quasi-experi-

mental shoreline/plateau situation of Golem Grad, along with the

comparison with other populations belonging to different species

( Table 3 ), offers support to the nonadaptive “prison effect” hypoth-

esis ( Eigenberg 2000; Bailey and Zuk 2009 ). Plateau-males and to

a lesser extent shoreline-males can rarely find adult females, but

instead encounter many male tortoises that morphologically resem-

ble females while they are under T stimulation. Hyperstimulation

and/or frustration created by a similar situation may promote fre-

quent SSB as documented in feral-domestic cats ( Yamane 2006 ).

Deprivation of member(s) of the opposite sex generates sexual

strain that can be relaxed by different behavioral outlets including

onanism, SSB, or interspecific sexual behaviors ( Exton et al. 2001;

Yamane 2006; Sakaguchi et al. 2007 ). A positive impact of absti-

nence (a proxy of frustration) on both T levels and sexual motivation

has been documented in humans ( Exton et al. 2001; Sakaguchi et al.

2007 ). Yet, nonadaptive SSB may simply reflect behavioral mistakes,

such as inaccurate sexual partner discrimination ( Scharf and Martin

2013 ). Although these explanations are not exclusive, the hypothesis

that males cannot discriminate sex has been rejected experimentally

( Galeotti et al. 2007, 2011 ). Many studies demonstrated the strong

ability of tortoises, and of nonavian reptiles in general, to use chem-

ical cues to distinguish sexual partner ( Cooper and Pèrez-Mellado

2002; Shine et al. 2003; Poschadel et al. 2006; Ibáñez et al. 2012 ).

Overall, SSB may emerge and persist in natural populations because

the costs for total SSB inhibition outweigh the costs of nonadaptive

SSB ( Thornhill and Alcock 1983; Burgevin et al. 2013; Scharf and

Martin 2013; Engel et al. 2015 ). Mechanisms that eliminate SSB

may negatively impact on HSB, for example, through a decrease of

general sexual motivation if common neuroendocrine pathways are

involved. Precisely, in vertebrates, T stimulates hypothalamic areas

that promote both heterosexual and homosexual activity ( Bakker et al. 2002, 2006 ). We speculate that the development, maintenance,

and functioning of specific neuroendocrine structures that selectively

filter out SSB might be too expensive (and useless) in male tortoises.

Other unusual sexual behaviors observed on Golem Grad Island,

especially on the plateau, reinforce the notion that the most SA adult

males deprived of females were highly stimulated ( Figure 1 ). Males

( N = 12) were observed mounting juveniles, dead males, and tortoise

skeletons (empty shells of males). Two were seen attempting to copu-

late with stones that vaguely resembled a tortoise ( Figure 1 ). In cap-

tivity, male SSB plus strange sex-toy behaviors have been reported in

tortoises (Internet search using keywords “tortoise and shoe”). These

cases of juvenile mounting, homosexual Davian behaviors, skeleto-

philia, or petrophilia illustrate that the strong sexual motivation of

males can overrule accurate discrimination of appropriate sexual part-

ners. In the nearby mainland Konjsko population, in other Hermann’s

tortoise populations, and in other Testudo species, uneven SR (generally

unbiased or male biased) and varying population densities have been

documented ( Table 3 ). However, the combination of extremely skewed

SR with very high population density and SSB was observed only on

Golem Grad. We thus suggest that frequent SSB are expressed above a

threshold that combines high density and skewed SR in physiologically

predisposed individuals (e.g., male tortoises with high T levels).

Considering other adaptive or nonadaptive explanations does

not change our main conclusions. For example, mate selection by

females would reinforce the scarcity of partners for the males; the

effect of practice (“Immature individuals learn … through SSB”;

Bailey and Zuk 2009 ) is unlikely in tortoises because sexual behav-iors were never observed in juveniles or in other contexts than Golem

Grad. The tortoises of Golem Grad provide the first documented

example of a “prison effect” in free-ranging animals and thus offer

a strong support to the hypothesis of nonadaptive causation for SSB.

This finding suggests that proximate factors underlying SSB are not

necessarily restricted to physiological disorders or require an adaptive

basis (although indirect links between nonadaptive SSB and mating

success exist; Logue et al. 2009 ). Under conditions with strong social

stimulation and a scarcity of partner, SSB may be common.

SUPPLEMENTARY MATERIAL

Supplementary material can be found at http://www.beheco.oxfordjournals.org/

FUNDING

The Ministry of Education, Science and Technological

Development of Republic of Serbia provided fund (grant #173043).

We thank the Galičica National Park authorities for issuing permits. Wethank Mitko Tasevski, colleagues, and students for support in the field.R. Shine offered valuable help to improve the English. Anonymous review-ers provided very useful comments.

Handling editor: John Fitzpatrick

Page 8 of 10

http://beheco.oxfordjournals.org/lookup/suppl/doi:10.1093/beheco/arw023/-/DC1http://beheco.oxfordjournals.org/lookup/suppl/doi:10.1093/beheco/arw023/-/DC1http://beheco.oxfordjournals.org/lookup/suppl/doi:10.1093/beheco/arw023/-/DC1http://beheco.oxfordjournals.org/lookup/suppl/doi:10.1093/beheco/arw023/-/DC1http://beheco.oxfordjournals.org/lookup/suppl/doi:10.1093/beheco/arw023/-/DC1http://beheco.oxfordjournals.org/lookup/suppl/doi:10.1093/beheco/arw023/-/DC1


9/10


REFERENCES

Amstislavskaya TG, Popova NK. 2004. Female-induced sexual arousal inmale mice and rats: behavioral and testosterone response. Horm Behav.46:544–550.

Arnow BA, Desmond JE, Banner LL, Glover GH, Solomon A, Polan ML,Lue TF, Atlas SW. 2002. Brain activation and sexual arousal in healthy,heterosexual males. Brain. 125:1014–1023.

Aubret F, Bonnet X, Shine R, Lourdais O. 2002. Fat is sexy for females

but not males: the influence of body reserves on reproduction in snakes( Vipera aspis ). Horm Behav. 42:135–147.

Bailey NW, Zuk M. 2009. Same-sex sexual behavior and evolution. TrendsEcol Evol. 24:439–446.

Bakker J, Honda S, Harada N, Balthazart J. 2002. Sexual partner prefer-ence requires a functional aromatase (cyp19) gene in male mice. HormBehav. 42:158–171.

Bakker J, De Mees C, Douhard Q, Balthazart J, Gabant P, Szpirer J, SzpirerC. 2006. Alpha-fetoprotein protects the developing female mouse brainfrom masculinization and defeminization by estrogens. Nat Neurosci.9:220–226.

Balthazart J. 2011. Minireview: hormones and human sexual orientation.Endocrinology. 152:2937–2947.

Balthazart J, Castagna C, Ball GF. 1997. Aromatase inhibition blocks theactivation and sexual differentiation of appetitive male sexual behavior in Japanese quail. Behav Neurosci. 111:381–397.

Banks A, Gartrell NK. 1995. Hormones and sexual orientation: a question-able link. J Homosex. 28:247–268.Bertolero A, Cheylan M, Hailey A, Livoreil B, Willemsen RE. 2011. Testudo

hermanni Gmelin, 1789, Hermann’s tortoise. In: Rhodin AGJ, PritchardCH, van Dijk PP, Saumure RA, Buhlmann KA, Iverson JB, MittermeierRA, editors. Biology of freshwater turtles and tortoises. Lunenburg (MA):Chelonian Research Monographs. p. 059.1–059.20.

Bierbach D, Jung CT, Hornung S, Streit B, Plath M. 2013. Homosexualbehaviour increases male attractiveness to females. Biol Lett. 9:20121038.

Bonnet X, Delmas V, El-Mouden H, Slimani T, Sterijovski B, Kuchling G.2010. Is sexual body shape dimorphism consistent in aquatic and terres-trial chelonians? Zoology (Jena). 113:213–220.

Bonnet X, Lagarde F, Henen BT, Corbin J, Nagy KA, Naulleau G, BalhoulK, Chastel O, Legrand A, Cambag R. 2001. Sexual dimorphism insteppe tortoises: influence of the environment and sexual selection onbody shape and mobility. Biol J Linn Soc. 72:357–372.

Burgevin L, Friberg U, Maklakov AA. 2013. Intersexual correlation for

same-sex sexual behaviour in an insect. Anim Behav. 85:759–762.Burnham KP, Anderson DR. 2002. Model selection and multimodel

inference: a practical information-theoretic approach. New York:Springer.

Camperio-Ciani A, Corna F, Capiluppi C. 2004. Evidence for maternallyinherited factors favouring male homosexuality and promoting femalefecundity. Proc Biol Sci. 271:2217–2221.

Choquet R, Lebreton JD, Gimenez O, Reboulet AM, Pradel R. 2009.U-CARE: Utilities for performing goodness of fit tests and manipulatingCApture-REcapture data. Ecography. 32:1071–1074.

Choquet R, Rouan L, Pradel R. 2009. Program E-SURGE: a softwareapplication for fitting multievent models. In: Thomson DL, Cooch EG,Conroy MJ, editors. Modeling demographic processes in marked popula-tions. New York: Springer. p. 845–865.

Cooper WE Jr, Pèrez-Mellado V. 2002. Pheromonal discriminations of sex,reproductive condition, and species by the lacertid lizard Podarcis hispanica .

J Exp Zool. 292:523–527.Djordjević S, Djurakić M, Golubović A, Ajtić R, Tomović L, Bonnet X.

2011. Sexual body size and body shape dimorphism of Testudo her manni incentral and eastern Serbia. Amphib Reptil. 32:445–458.

Djordjević S, Tomović L, Golubović A, Simović A, Sterijovski B, DjurakicM, Bonnet X. 2013. Geographic (in-)variability of gender-specific traitsin Hermann’s tortoise. Herpetol J. 23:67–74.

Eigenberg HM. 2000. Correctional officers and their perceptions of homo-sexuality, rape, and prostitution in male prisons. Prison J. 80:415–433.

Engel KC, Männer L, Ayasse M, Steiger S. 2015. Acceptance thresh-old theory can explain occurrence of homosexual behaviour. Biol Lett.11:20140603.

Exton MS, Krüger TH, Bursch N, Haake P, Knapp W, Schedlowski M,Hartmann U. 2001. Endocrine response to masturbation-induced orgasmin healthy men following a 3-week sexual abstinence. World J Urol.19:377–382.

Hailey A. 1990. Adult survival and recruitment and the explanation of anuneven sex ratio in a tortoise population. Can J Zool. 68:547–555.

Hailey A, Willemsen RE. 2000. Population density and adult sex ratio ofthe tortoise Testudo hermanni in Greece: evidence for intrinsic populationregulation. J Zool. 251:325–338.

Galeotti P, Sacchi R, Pellitteri-Rosa D, Fasola M. 2007. Olfactory discrim-ination of species, sex, and sexual maturity by the Hermann’s tortoiseTestudo hermanni . Copeia. 2007:980–985.

Galeotti P, Sacchi R, Pellitteri-Rosa D, Fasola M. 2011. The yellow cheek-

patches of the Hermann’s tortoise (Reptilia, Chelonia): sexual dimor-phism and relationship with body condition. Ital J Zool. 78:464–470.

Goldey KL, van Anders SM. 2015. Sexual modulation of testosterone:insights for humans from across species. Adapt Hum Behav Physiol.1:93–123.

Golubović A, Bonnet X, Djordjević S, Djurakić M, Tomović L. 2013.Variations in righting behaviour across Hermann’s tortoise populations. J Zool. 291:69–75.

Grimbos T, Dawood K, Burriss RP, Zucker KJ, Puts DA. 2010. Sexual ori-entation and the second to fourth finger length ratio: a meta-analysis inmen and women. Behav Neurosci. 124:278–287.

Hoskins JL, Ritchie MG, Bailey NW. 2015. A test of genetic models for theevolutionary maintenance of same-sex sexual behaviour. Proc Biol Sci.282. doi: 10.1098/rspb.2015.0429.

Ibáñez A, López P, Martín J. 2012. Discrimination of conspecifics’ chemi-cals may allow Spanish terrapins to find better partners and avoid com-

petitors. Anim Behav. 83:1107–1113. Jannini EA, Blanchard R, Camperio-Ciani A, Bancroft J. 2010. Male

homosexuality: nature or culture? J Sex Med. 7:3245–3253. Jannini EA, Burri A, Jern P, Novelli G. 2015. Genetics of human sexual

behavior: where we are, where we are going. Sex Med Rev. 3:65–77. Jolly GM. 1965. Explicit estimates from capture-recapture data with both

death and immigration-stochastic model. Biometrika. 52:225–247.Kendall WL, Nichols JD. 2002. Estimating state-transition probabilities for

unobservable states using capture-recapture/resighting data. Ecology.83:3276–3284.

King RB, Bowden RM. 2013. Seasonal, condition-dependent, and individ-ual variation in testosterone in a natricine snake. J Herpetol. 47:179–183.

Lagarde F, Bonnet X, Corbin J, Henen B, Nagy K. 2002. A short springbefore a long jump: the ecological challenge to the steppe tortoise ( Testudohorsfieldi ). Can J Zool. 80:493–502.

Lagarde F, Bonnet X, Henen B, Nagy K, Corbin J, Lacroix A, Trouvé C.2003. Plasma steroid and nutrient levels during the active season in wildTestudo horsfieldi . Gen Comp Endocrinol. 134:139–146.

Lagarde F, Guillon M, Dubroca L, Bonnet X, Ben Kaddour K, SlimaniT, El Mouden HE. 2008. Slowness and acceleration: a new method toquantify the activity budget of chelonians. Anim Behav. 75:319–329.

Lebreton JD, Burnham KP, Clobert J, Anderson DR. 1992. Modeling survival and testing biological hypotheses using marked animals: a unifiedapproach with case studies. Ecol Monogr. 62:67–118.

Lebreton JD, Pradel R. 2002. Multistate recapture models: modellingincomplete individual histories. J Appl Stat. 29:353–369.

Logue DM, Mishra S, McCaffrey D, Ball D, Cade WH. 2009. A behav-ioral syndrome linking courtship behavior toward males and females pre-dicts reproductive success from a single mating in the hissing cockroach,Gromphadorhina portentosa . Behav Ecol. 20:781–788.

MacFarlane GR, Blomberg SP, Kaplan G, Rogers LJ. 2007. Same-sex sex-ual behavior in birds: expression is related to social mating system andstate of development at hatching. Behav Ecol. 18:21–33.

Michel CL, Bonnet X. 2014. Effect of a brief stress on progesterone plasmalevels in pregnant and non-pregnant guinea pigs. Anim Biol. 64:19–29.

Mustanski BS, Chivers ML, Bailey JM. 2002. A critical review of recentbiological research on human sexual orientation. Annu Rev Sex Res.13:89–140.

Nelson RJ. 2005. An introduction to behavioral endocrinology. 3rd edn.Sunderland (MA): Sinauer Associates, Inc.

Pieau C. 1971. Sur la proportion sexuelle chez les embryons de deuxChéloniens ( Testudo graeca L. et Emys orbicularis L.) issus d’œufs incubésartificiellement. C R Acad Sci Paris. 272:3071–3074.

Poiani A. 2010. Animal homosexuality. Cambridge (UK): CambridgeUniversity Press.

Poschadel JR, Meyer-Lucht Y, Plath M. 2006. Response to chemical cuesfrom conspecifics reflects male mating preference for large females andavoidance of large competitors in the European pond turtle, Emys orbicu-laris . Behaviour. 143:569–587.

Page 9 of 10


10/10

Behavioral Ecology

Pradel R. 2005. Multievent: an extension of multistate capture-recapturemodels to uncertain states. Biometrics. 61:442–447.

Rahman Q. 2005. The neurodevelopment of human sexual orientation.Neurosci Biobehav Rev. 29:1057–1066.

Rice G, Anderson C, Risch N, Ebers G. 1999. Male homosexuality: absenceof linkage to microsatellite markers at Xq28. Science. 284:665–667.

Sacchi R, Pellitteri-Rosa D, Marchesi M, Galeotti P, Fasola M. 2013. A comparison among sexual signals in courtship of European tortoises. J Herpetol. 47:215–221.

Sakaguchi K, Oki M, Honma S, Uehara H, Hasegawa T. 2007. The lowersalivary testosterone levels among unmarried and married sexually activemen. J Ethol. 25:223–229.

Scharf I, Martin OY. 2013. Same-sex sexual behavior in insects andarachnids: prevalence, causes, and consequences. Behav Ecol Sociobiol.67:1719–1730.

Seber GA. 1965. A note on the multiple-recapture census. Biometrika.52:249–259.

Shine R, Harlow P, LeMaster MP, Moore IT, Mason RT. 2000. The trans- vestite serpent: why do male garter snakes court (some) other males? Anim Behav. 59:349–359.

Shine R, Phillips B, Waye H, LeMaster M, Mason RT. 2003.Chemosensory cues allow courting male garter snakes to assess bodylength and body condition of potential mates. Behav Ecol Sociobiol.54:162–166.

Sommer V, Vasey PL, editors. 2006. Homosexual behaviour in animals: anevolutionary perspective. Cambridge (UK): Cambridge University Press.

Thornhill R, Alcock J. 1983. The evolution of insect mating systems.Cambridge (MA): Harvard University Press.

VanderLaan DP, Garfield ZH, Garfield MJ, Leca JB, Vasey PL, Hames

RB. 2014. The “female fertility–social stratification–hypergyny”hypothesis of male homosexual preference: factual, conceptual andmethodological errors in Barthes et al [Commentary]. Evol HumBehav. 35:445–447.

Yamane A. 2006. Frustrated felines: male-male mounting in feral cats. In:Sommer V, Vasey PL, editors. Homosexual behaviour in animals: an evo-lutionary perspective. Cambridge (UK): Cambridge University Press. p.172–189.

Zitzmann M, Nieschlag E. 2001. Testosterone levels in healthy men and therelation to behavioural and physical characteristics: facts and constructs.Eur J Endocrinol. 144:183–197.

Page 10 of 10

behavioral ecology 2016 bonnet beheco arw023

Documents