Embryonic Temperature and Gonadal Sex Organize
Male-Typical Sexual and Aggressive Behavior in a Lizard
with Temperature-Dependent Sex Determination*
TURK RHEN AND DAVID CREWS
Section of Integrative Biology, School of Biological Sciences, University of Texas, Austin, Texas 78712
ABSTRACT
Temperature during embryonic development determines gonadal
sex in the leopard gecko, Eublepharis macularius. Moreover, both
embryonic temperature and gonadal sex influence adult behavior. Yet
it remains unclear whether the effects of embryonic temperature and
gonadal sex on behavior are irreversibly organized during develop-
ment. To address this question, we gonadectomized adult females and
males generated from a temperature that produces mostly females (30
C) and a temperature that produces mostly males (32.5 C). Females
and males from both temperatures were then treated with equivalent
levels of various sex steroids. We found that both embryonic temper-
ature and gonadal sex had persistent effects on the expression of
male-typical sexual and aggressive behaviors. For example, adult
females do not scent mark and display very little courtship and
mounting behavior even when treated with levels of hormones (pri-
marily androgens) that activate these behaviors in males. In contrast,
species-typical aggressive displays were less sex specific and were
activated by both dihydrotestosterone and testosterone (T) in males
and by T in females. Nevertheless, the average duration of aggressive
displays was significantly shorter in T-treated females than that in
T-treated males. With regard to submissive behavior, androgens de-
creased flight behavior in males, but had no effect in females. Em-
bryonic temperature had enduring effects on certain behaviors in
males. For instance, males from a male-biased embryonic tempera-
ture scent-marked more than males from a female-biased embryonic
temperature when treated with dihydrotestosterone or T. Conversely,
and across hormone treatments, males from a female-biased embry-
onic temperature mounted more than males from a male-biased em-
bryonic temperature. Finally, treatment with 17b-estradiol decreased
submissive behavior in males from a male-biased embryonic temper-
ature compared with that in males from a female-biased embryonic
temperature. Courtship and aggressive behavior were not influenced
by temperature. These results strongly suggest that male-typical
behaviors in the adult leopard gecko are permanently organized by
both embryonic temperature and gonadal sex during development.
(Endocrinology 140: 4501–4508, 1999)
WHEREAS SEX chromosomes determine gonadal sexin mammals and birds (1), embryonic temperature
determines sex in some lizards, many turtles, and all croc-
odilians (2–4). Despite this dramatic difference in the signal
that initiates testicular or ovarian development, it appears
that much of the molecular machinery for gonadogenesis is
evolutionarily conserved. Indeed, genes clearly involved in
mammalian sex determination [e.g. anti-Mu¨llerian hormone,
steroidogenic factor-1, Wilms’ Tumor (WT-1) gene, and Sry-
like, High Mobility Group Box-like (SOX)9 gene] (5–7) have
also been identified and implicated in avian sex determina-
tion (5, 8–10) and temperature-dependent sex determination
(TSD) in reptiles (5, 11–13). Moreover, the gonadal anlagen
is initially bipotential and consists of a cortical region that
gives rise to the ovary and a medullary region that gives rise
to the testis in all amniotic vertebrates (14). Considering such
similarities in gonadal differentiation, a fundamental ques-
tion is whether other aspects of sexual differentiation are also
alike in mammals, birds, and reptiles with TSD.
In this respect, the sexual differentiation of reproductive
and aggressive behavior is very well studied in mammals
and birds and depends upon the sexually dimorphic pro-
duction of steroids by the differentiated gonads. Our under-
standing of sex differences in these behaviors is based on the
organization-activation paradigm formulated by Pheonix et
al. (15). Classically, behavioral activation is the process by
which circulating sex steroids affect specific neural sub-
strates to induce sexual or aggressive behavior in adults that
are exposed to the appropriate external stimuli (i.e. individ-
uals of the opposite or same sex, respectively). For example,
sex differences in the display of male-typical sex behavior
occur because of sex differences in plasma testosterone (T)
levels in intact rats. In fact, exogenous T can activate male-
typical mounting behavior in gonadectomized male and fe-
male rats (reviewed in Refs. 16–19). In contrast, castrated
male rats do not display female-typical sex behavior when
treated with a sex steroid regimen [i.e. 17b-estradiol (E2)
followed by progesterone] that activates lordosis in ovari-
ectomized female rats. A perinatal T surge in male, but not
female, rats causes this dimorphic response to hormonal
activation of female sex behavior in adulthood (reviewed in
Refs. 16 and 19). Such permanent developmental effects of
sex steroids on subsequent behavior are called organiza-
tional effects.
Although details vary, the basic paradigm of organiza-
tional vs. activational effects of sex steroids has been sup-
ported in a variety of mammals and birds. For instance,
female quail are demasculinized (i.e. organized) by circu-
lating estrogens during the perinatal period and lose the
ability to display male-typical mounting behavior when
Received March 22, 1999.
Address all correspondence and requests for reprints to: Dr. Turk
Rhen, Section of Integrative Biology, School of Biological Sciences, Uni-
versity of Texas, Austin, Texas 78712. E-mail: turkrhen.uta@mail.
utexas.edu.
* This work was supported by Individual National Research Service
Award MH-11369 from the NIMH (to T.R.), NSF Dissertation Improve-
ment Grant IBN-9623546 (to T.R.), and NIMH Grant MH-57874 (to D.C.).
0013-7227/99/$03.00/0 Vol. 140, No. 10
Endocrinology Printed in U.S.A.
Copyright © 1999 by The Endocrine Society
4501
treated with T as adults (20, 21). Gonadectomized male
and female quail, when treated with E2 as adults, can
display female-typical receptive behavior in response to
male sexual overtures. Copulatory behavior in the zebra
finch also fits this general pattern, even though its song
system is paradoxical in that exogenous E2 organizes the
male phenotype (22). Thus, the sexual differentiation of
reproductive behaviors in birds can be classified as acti-
vational and/or organizational in nature. In contrast, very
little is known about the sexual differentiation of behavior
in reptiles with TSD.
In the leopard gecko, Eublepharis macularius, an embryonic
temperature of 30 C produces a female-biased sex ratio (ap-
proximately one male to three females), whereas 32.5 C pro-
duces a male-biased sex ratio (approximately three males to
one female) (23). Furthermore, both embryonic temperature
and gonadal sex influence reproductive and aggressive be-
havior in intact adult leopard geckos (24, 25; reviewed in Ref.
26). However, it is unclear whether these effects are organi-
zational or activational, because embryonic temperature and
gonadal sex also influence adult sex steroid physiology. For
example, female leopard geckos have lower circulating levels
of T and 5a-dihydrotestosterone (DHT) than males and nor-
mally do not exhibit male-typical sex behaviors (24, 25, 27,
28), yet females treated with male-typical levels of T can
display male-typical courtship behavior (27). However, in
this latter study, there was no quantitative comparison be-
tween levels of courtship behavior in males and females
given identical hormone treatments and tested in exactly the
same manner. Consequently, it is not clear whether sex dif-
ferences in courtship behavior are purely activational in
nature.
Similarly, males from the male-biased incubation tem-
perature (i.e. 32.5 C) are more aggressive but less sexually
active toward females than are males from the female-
biased incubation temperature (i.e. 30 C) (25). Males from
the male-biased incubation temperature also have lower
E2 levels than males from the female-biased incubation
temperature, whereas their T levels are similar (28, 29).
Overall, the combined data clearly show that sexual dif-
ferentiation of behavior in the leopard gecko depends
upon both gonadal sex and embryonic temperature (re-
viewed in Ref. 26). Nevertheless, it is uncertain whether
such effects are activated or organized because there has
been no systematic examination of temperature and sex
effects on reproductive and aggressive behavior while
controlling for circulating hormone levels.
A definitive answer to this question would provide fun-
damental information concerning sexual differentiation of
behavior in a reptile with TSD. The following study of
male-typical behaviors was designed to determine
whether the sexes are behaviorally organized in the way
that the mammalian and avian sexes are organized. An-
other goal was to determine whether embryonic temper-
ature has permanent (i.e. organizational) effects on behav-
ior within each sex. Overall, this experiment illuminates
how embryonic temperature and gonadal sex during de-
velopment and sex steroids in adulthood act and interact
to influence sexual and aggressive behaviors in the leop-
ard gecko.
Materials and Methods
Animals
Animals were treated according to a research protocol approved by
the university’s institutional animal care and use committee. Leopard
gecko eggs from our captive breeding colony at the University of Texas
were collected within 24 h of oviposition and candled for fertility. Fertile
eggs were placed in individual cups filled with moist vermiculite (1 part
water/1 part vermiculite) and split between two constant incubation
temperatures (30 and 32.5 6 0.1 C). An incubation temperature of 30 C
produces a female-biased sex ratio, whereas 32.5 C produces a male-
biased sex ratio (23). Geckos hatched from these eggs were raised in
isolation for 49–52 weeks as previously described (25). Leopard geckos
reach sexual maturity at roughly 45 weeks of age (28).
Surgical and hormonal manipulation
Approximately equal numbers of adult males and females from each
incubation temperature were gonadectomized under cold anesthesia. At
the same time these animals were implanted sc with SILASTIC brand
tubing (Dow Corning Corp., Midland, MI) containing cholesterol (C), E2,
DHT, or T for a fully factorial experimental design, with embryonic
temperature, gonadal sex (before gonad removal), and adult hormone
treatment as main effects. Although sample sizes ranged from 8–15 for
each group, all except 1 group had 10 or more individuals (see Table 1).
Implant length was 10 mm for C, E2, and DHT and 20 mm for T.
Otherwise, implants were identical in dimension (id, 1.47 mm; od, 1.95
mm), were packed in the same manner, and were all soaked in reptilian
Ringers solution for 24 h before surgery. Animals were allowed 4 weeks
to recover after surgery/implantation, and then behavior was tested.
One day after behavior testing was completed, a blood sample was taken
via cardiac puncture for RIA to confirm hormone delivery. Animals were
then killed, dissected, and examined for residual gonadal tissue. Gon-
adectomies were complete in all cases.
Behavior testing
We used a behavior testing procedure similar to that described by
Flores et al. (25). Briefly, animals were tested three times for 5 min each
time in a neutral cage with one of two types of stimulus animals to assess
levels of male-typical and female-typical sexual and aggressive behavior
(six tests total per animal). In this paper we report the results of behavior
tests in which experimental animals were exposed to intact vitellogenic
females (i.e. sexually receptive females) on 3 consecutive days; each
experimental animal interacted with a given female only once. This set
of tests allowed us to examine the factors controlling the display of
male-typical behaviors toward female stimulus animals. Experimental
animals were first placed in a neutral cage (43 3 22 3 20 cm) with a clean
paper towel as a liner. Stimulus females were then placed, facing the
experimental subject, in the same cage. Subject animal and stimulus
female behavior was recorded using a keypad timer (Witt/Timer Pro-
gram courtesy of Diane Witt, NIH, Bethesda, MD). Tests ended after 5
min or if an attack or attempted copulation occurred. In contrast to the
testing protocol used previously (25), experimental animals in the cur-
rent study were tested on 6 consecutive days (vs. over a 5-week period)
and were first tested with female stimulus animals for 3 days and then
with male stimulus animals for 3 days (vs. a randomized sequence). The
latter change was made because aggressive behavior of stimulus males
toward experimental animals could alter subsequent behavior and thus
would have confounded our measures of male-typical and female-
typical sexual behaviors.
We measured scent marking, courtship (i.e. tail vibrations), and
mounting (i.e. body grips) as male-typical sex behaviors in our exper-
imental animals. We also recorded aggressive (i.e. high posture display
and attacks) and submissive (i.e. flight) behaviors. In a sexual encounter,
a male slowly approaches a female, first licking the substrate or the air
with his tongue and then licking the female. An attractivity pheromone
in the skin of females (30) elicits a male-typical tail vibration that creates
an audible buzz and a tactile vibration of the substrate. During these
encounters males may also drag their preanal pores on the substrate,
presumably to deposit pheromones in a scent-marking behavior. Males
then body grip the female’s skin with their jaws during courtship and
mounting. Body grips are a major component of mounting behavior, as
4502 SEXUAL DIFFERENTIATION OF BEHAVIOR IN A LIZARD Endo • 1999
Vol 140 • No 10
they position the male for copulation and nearly always accompany
intromission. We measured the cumulative duration (in seconds) of
scent marking, tail vibration, and mounting (i.e. body grip) behaviors.
Overall, these behaviors are a fairly complete index of male-typical sex
behavior. We also measured high posture duration (an aggressive dis-
play) and the frequency of tests in which an attack occurred as an index
of aggressive behavior. Conversely, submissive behavior was recorded
as the cumulative duration (in seconds) of flight from the stimulus
female.
RIA
On the day after the last behavior test (with a male), a blood sample
was drawn from each experimental animal by cardiocentesis using a
heparinized 1-cc syringe with a 25-gauge needle. Blood was centrifuged
at 3000 rpm for 10 min at 4 C, and plasma was stored in plastic microfuge
tubes at 280 C until assayed for levels of DHT, E2, and T. The antibodies
used for RIA were DT3–351 for DHT, E26–47 for E2, and T3–125 for T
(Endocrine Sciences, Inc., Calabasas Hills, CA). Column chromatogra-
phy and RIAs were performed as previously described (28). Recoveries
averaged 57%, 56%, and 70% for DHT, E2, and T, respectively. Assay
sensitivity was 71 pg DHT/ml plasma, 92 pg E2/ml plasma, and 86 pg
T/ml plasma. For a pooled plasma sample, intraassay coefficients of
variation were 16%, 18%, and 17% for DHT, E2, and T, respectively.
Interassay coefficients of variation for the same sample were 18%, 17%,
and 13% for DHT, E2, and T, respectively. We also ran quality control
standards of known concentration in the low, medium, and high ranges
of the standard curve for each steroid. For DHT, intraassay coefficients
of variation were 12%, 6%, and 6% in the low, medium, and high parts
of the curve, respectively. Interassay coefficients of variation for DHT
were 18%, 9%, and 11% in the low, medium, and high parts of the curve,
respectively. For E2, intraassay coefficients of variation were 11%, 4%,
and 6% in the low, medium, and high parts of the curve, respectively.
Interassay coefficients of variation for E2 were 10%, 8%, and 9% in the
low, medium, and high parts of the curve, respectively. For T, intraassay
coefficients of variation were 9%, 4%, and 5% in the low, medium, and
high parts of the curve, respectively. Interassay coefficients of variation
for T were 14%, 9%, and 10% in the low, medium, and high parts of the
curve, respectively.
Statistical analyses
All data were analyzed using embryonic temperature, gonadal sex
(before gonadectomy), adult hormone treatment, and day of testing as
main effects in a four-way repeated measures design. All dependent
variables, scent marking, tail vibration, body grip, high posture, and
flight durations, were analyzed with univariate ANOVA. Independent
variables were considered nonsignificant when P . 0.05. Dependent
variables are presented as least squares mean 6 one se. Post-hoc com-
parisons were made using the Dunn-Sida´k method to provide a signif-
icance level of a9 5 1 2 (1 2 0.05)1/k, where k is the number of individual
comparisons for an experimentwise a 5 0.05 (31). Hormone concentra-
tions were first log transformed and then compared using Tukey’s
honestly significant difference test. All statistics were performed using
version 3.1 of JMP (32) for Macintosh (Apple Computer, Inc., Cupertino,
CA).
Results
Hormone levels
As expected, treatment with SILASTIC capsules contain-
ing E2, DHT, and T elevated plasma levels of these hormones
above the levels observed in geckos treated with C (see Table
1). Importantly, treatment with a given steroid resulted in
equivalent levels of hormones in gonadectomized female
and male leopard geckos from each embryonic temperature.
Consequently, our experimental manipulations achieved the
desired goal, which was to separate the normally confound-
ing effects of embryonic temperature and gonadal sex on sex
steroid physiology and behavior. The steroid levels pro-
duced by these implants are in the normal physiological
ranges for intact males and/or females of this species
(24–28).
Scent marking behavior
Scent marking behavior was organized by embryonic tem-
perature [F(1,458) 5 12.1; P 5 0.0005], gonadal sex [F(1,458) 5
24.0; P , 0.0001], and a significant interaction between em-
bryonic temperature and gonadal sex during development
[F(1,458) 5 12.1; P 5 0.0005]. Specifically, females never scent
marked regardless of their embryonic temperature (results
not shown), whereas, overall, males from the male-biased
temperature marked significantly more than did males from
the female-biased temperature (see Fig. 1). Scent marking
was activated by adult hormone treatment [F(3,458) 5 5.8;
TABLE 1. Circulating concentrations of DHT, E2, and T (nanograms per ml plasma) in female and male leopard geckos from two
incubation temperatures (30 or 32.5 C) after receiving SILASTIC implants filled with C, DHT, E2, or T
Treatment groups Circulating steroid levels
Sample size
Sex Temperature (C) Hormone E2 DHT T
Female 30 C 1.4 6 0.3a 0.3 6 0.1a 0.3 6 0.05a 13
E2 7.2 6 2.4
b 0.3 6 0.1a 0.4 6 0.1a 10
DHT 2.9 6 0.4a,c 42.9 6 5.3b 1.6 6 0.2b 14
T 1.7 6 0.3a 17.4 6 1.6b 172 6 15c 15
32.5 C 0.8 6 0.1a 0.2 6 0.05a 0.3 6 0.07a 11
E2 6.7 6 1.3
b,c 0.2 6 0.04a 0.2 6 0.05a 10
DHT 1.2 6 0.2a 39.6 6 8.3b 1.4 6 0.3b 10
T 1.6 6 0.2a 16.7 6 3.8b 190 6 23c 10
Male 30 C 1.0 6 0.5a 0.4 6 0.3a 0.3 6 0.05a 10
E2 8.1 6 1.7
b 0.3 6 0.2a 0.4 6 0.1a 10
DHT 1.1 6 0.3a 35.0 6 4.3b 1.3 6 0.2b 10
T 2.4 6 0.3a,c 17.9 6 4.4b 227 6 15c 10
32.5 C 1.4 6 0.3a 0.3 6 0.2a 0.4 6 0.1a 8
E2 7.8 6 1.1
b 0.4 6 0.2a 0.3 6 0.1a 10
DHT 2.0 6 0.4a,c 44.2 6 6.3b 1.5 6 0.3b 10
T 2.0 6 0.2a 14.7 6 2.6b 187 6 25c 10
Mean hormone levels are shown in nanograms per ml plasma 6 1 SE. Numbers of samples assayed are shown. For each hormone, groups
with different superscripted letters are significantly different from each other using Tukey’s post-hoc comparisons.
SEXUAL DIFFERENTIATION OF BEHAVIOR IN A LIZARD 4503
P 5 0.0006], but there was also a signif