Endocrine and behavioral responses to aggression and social dominance in the green anole lizard, Anolis carolinensis

Abstract

Adult males of the small arboreal iguanid lizard, Anolis carolinensis, will fight and form social dominance hierarchies when placed in habitats with limited resources. The relationships between time since initial aggressive interaction, relative social dominance, reproductive activity, and corticosterone and androgen levels were determined for 34 pairs of lizards. A discriminant analysis established a “dominance index” which indicated that over 90% of the difference between individuals who had won or lost aggressive interactions (putative social dominants and subordinates) was attributable to a single discriminant function reflecting altered body color, perch site selection, and circulating androgen. Animals that had darker body color also selected lower perch sites and had depressed rates of courtship relative to winners of fights and were thus designated as social subordinates. These animals also had levels of circulating androgen significantly lower than that of dominants, but circulating corticosterone was not significantly affected. Winners of fights showed a dramatic surge in circulating androgen at 1 hr but returned to near control values by 1 week; losers, however, showed depressed circulating androgen levels at 1 week.

Full text

GENERAL AND COMPARATIVE ENDOCRINOLOGY 77,2462.55 (1990)
Endocrine and Behavioral Responses to Aggression and Social
Dominance in the Green Anole Lizard, Anolis carolinensis
NEIL GREENBERG* AND DAVID CREWS?
*Department of Zoology and Life Sciences Graduate Program in Ethology, University of Tennessee,
Knoxville, Tennessee 37996, and tlnstitute of Reproductive Biology, Department of Zoology, University of
Texas, Austin, Texas 78712
Accepted April 18, 1989
Adult males of the small arboreal iguanid lizard,
Anolis carolinensis,
will fight and form
social dominance hierarchies when placed in habitats with limited resources. The relation-
ships between time since initial aggressive interaction, relative social dominance, reproduc-
tive activity, and corticosterone and androgen levels were determined for 34 pairs of lizards.
A discriminant analysis established a “dominance index” which indicated that over 90% of
the difference between individuals who had won or lost aggressive interactions (putative
social dominants and subordinates) was attributable to a single discriminant function re-
flecting altered body color, perch site selection, and circulating androgen. Animals that had
darker body color also selected lower perch sites and had depressed rates of courtship
relative to winners of tights and were thus designated as social subordinates. These animals
also had levels of circulating androgen significantly lower than that of dominants, but cir-
culating corticosterone was not significantly affected. Winners of tights showed a dramatic
surge in circulating androgen at 1 hr but returned to near control values by 1 week; losers,
however, showed depressed circulating androgen levels at 1 week. o 1990 Academic PW, IK.
Aggressive behavior, social dominance,
and reproductive activity have been associ-
ated with adrenal and other physiological
responses to environmental stressors in
many vertebrates, including several reptiles
(Greenberg and Wingfield, 1987). The inter-
actions between these responsive systems
and behavior indicate that their underlying
physiological substrate involves hormones
that have multiple effects. For example,
hormones associated with activation of the
adrenal axis can also enhance acquisition
and delay extinction of active avoidance
behavior (Bohus and de Wied, 1980) and
hormones associated with gonadal activa-
tion can directly affect sensory-perceptual
functions (Beach, 1974; Gandelman, 1983)
as well as affecting attention either directly
(Andrew, 1972) or by means of interactions
with adrenocortical hormones (Oades,
1979). One of the primary ambitions of ex-
perimental behavioral endocrinology is to
determine the manner in which such multi-
ple effects are integrated and coordinated
with each other and the environment.
The green anole lizard,
Anolis carolinen-
sis,
is uniquely suited to the investigation of
the manner in which environmental stimuli
and physiological variables are integrated
in the control of agonistic behavior. The
dermal chromatophores of this species
have no direct sympathetic innervation
(Kleinholz, 1938), but are responsive to cir-
culating hormones associated with physio-
logical stress such as melanotropin (Green-
berg et al., 1986) and epinphrine, thus per-
mitting body color to be utilized as a
relatively direct indication of an endocrine
response to environmental stressors
(Greenberg and Crews, 1983). Studies on
A. carolinensis
have documented body
color changes in both agonistic interactions
and long-term relationships (Greenberg et
al., 1984). For example, the males that ini-
246
0016~6480&O $1.50
Copyright 0 1990 by Academic Press, Inc.
AU rights of reproduction in any form reserved.

RESPONSES TO AGGRESSION AND DOMINANCE
247
tiate almost all sexual activity are charac-
teristically completely green while subordi-
nate, sexually inactive males are typically
brown.
This study was undertaken to determine
the association between the outcome of an
agonistic interaction, subsequent social
dominance, reproductive behavior, and an-
drogen and corticosterone levels in the liz-
ard,
Anolis carolinensis,
at different points
in time after an initial interaction.
METHODS
Adult
A. carolinensis
were commercially supplied
by the Snake Farm, LaPlace, Louisiana. When re-
ceived, animals were placed in habitats under condi-
tions known to initiate and maintain gonadal recrudes-
cence throughout the year (Crews and Garrick, 1970;
Crews et
al.,
1974; Licht, 1967, 1971). The stimulatory
regimen had a photic cycle of 14 hr light:10 hr dark,
and provided corresponding daily temperature fluctu-
ations of 32:22”. Temperature is known to be a signif-
icant variable in the responses of lizard adrenal to
ACTH (Licht and Bradshaw, 1969), on spermatogen-
esis (Licht, 1971), and in the production and action of
androgens (Pearson er al., 1976). Humidity varied in-
versely with temperature ranging from 70 (day) to 90%
(night) RH.
Behavioral techniques.
Prior to behavioral experi-
mentation, 68 adult males (minimum size, 61 mm
snout-to-vent) were individually housed for a mini-
mum of 3 weeks in glass vivaria (25 x 25 x 30 cm)
provided with 20-cm diagonal wooden perches and a
substrate of sphagnum moss. Food consisted of com-
mercially supplied l-cm crickets which maintained
alertness in the lizards when provided on alternate
days and allowed to hide in the moss. Handling was
strictly minimized to eliminate this potential stressor
(Meier
et al.,
1973).
Each lizard was briefly exposed to individual recep-
tive females and intruding males to determine disposi-
tion to court or to manifest territorial aggression; tests
were halted in 5 min or if copulation or a tight seemed
imminent. Each was also characterized along two di-
mensions associated with physiological stress or social
status: body color and perch-site selection (Greenberg
et
al., 1984). The range of variability manifest for each
of these dimensions was divided into several clearly
delineated units and was represented on an ordinal
scale based upon their sequence of expression in an
agonistic situation (Table 1). Because the expression
of body color is also sensitive to ambient light reflec-
tance and temperature, habitats were provided with
uniform illumination levels, background color, and
temperature in order to minimize direct, nonsocial
sources of variability in the responses of chromato-
phores to circulating hormones.
Pairs of individuals randomly selected to be subjects
were introduced to each other by removing the opaque
divider separating two connecting vivaria. The distinc-
tive displays exchanged between subjects (Greenberg,
1977), including body color changes indicative of adre-
nal activation (Greenberg, 1983), were documented;
the pairs were left together in the enclosure and body
color and perch-site selection were recorded twice
daily for up to 1 week.
The terms “winner” or “loser” designate subjects
that (1) engaged in an agonistic encounter and mani-
fested a postorbital “eyespot,” indicating adrenal epi-
nephrine release (Greenberg, 1983), and (2) performed
displays that included sagittal expansion of the body
profile, a posture unique to conspecific aggression
(Greenberg, 1977). Losers were identified by (1) their
cessation of display and (2) exclusion from higher
perch sites, the adoption of a head-down or cryptic
posture, or attempts at flight. The terms “dominant”
and “subordinate” are used to designate subjects dur-
ing the period of cohabitation subsequent to the initial
agonistic encounter. Relative social dominance was
determined by daily observations of staged feeding or
female introduction episodes in which subjects mani-
fested priority of access to food or females in at least
75% of observations. Further, submissiveness was in-
dicated by observations of supplantation from perch
site, or withdrawal from an arena of activity when the
other lizard displays or approaches. Subordinate be-
havior was manifest in all instances of exposure to a
display and approximately 25% of the observed in-
stances of the other lizard’s approach. Dominant indi-
viduals, unlike subordinates, will not withdraw or be
supplanted from a perch site when approached. In the
conditions provided, winners of tights invariably be-
come social dominants.
Daily observations of body color and perch-site se-
lection were recorded on a hand-held Hewlett-Packard
51B computer. Staged agonistic and courtship epi-
sodes were documented with an electromechanical
event recorder (Esterline Angus 20-channel recorder);
observations of feeding and social dominance interac-
tions were recorded by pen. Data were then trans-
ferred or transcribed onto an IBM 5 150 computer and
subjected to analysis using SAS. Sexual behavior after
the introduction of a receptive female was character-
ized by rapid nodding (Table 1), a key element of
courtship behavior (Crews, 1975; Greenberg, 1977).
Observer effects on subjects were minimized by
means of closed-circuit color television surveillance.
Hormone assays.
Blood samples were taken by
rapid decapitation at 1 hr, 1 day, or 1 week following
the agonistic interaction. The procedure was per-
formed between 1400 and 1600, times at which corti-

248
GREENBERG AND CREWS
TABLE 1
SCORING PROTOCOL FOR BODY COLOR, SITE SELECTION, AND SOCIAL DISPLAY BEHAVIOR IN GREEN ANOLE
LIZARDS, ANOLIS CAROLINENSM
Color index: An ordinal scale of body colors progressively observed during agonistic encounters or
manifested in experimentally stressed lizards.
1. (G) all visible pigmented body surfaces a light to moderate green.
2. (GB) part green/part brown.
3. (B) all visible pigmented body surfaces a light to moderate brown.
4. (DB) more than 50% of the body surface is dark brown.
5. (DB/ES) dark brown with an eyespot (ES), a black spot near the posterior margin of the orbit.
6. (G/ES) all visible pigmented body surfaces moderate to dark green with eyespot.
7. (G/B/ES) body color is blotchy, irregularly pigmented dark green and dark brown, with eyespot.
Sire
selection:
An ordinal scale of perching sites, ranging from optimal predatory surveillance sites
through degree of displacement in response to aggression or social dominance behavior by another male.
1. top of perch, favored predatory/social surveillance site.
2. upper half of perch.
3. lower half of perch.
4. on substrate; sustained presence, not foraging for prey.
5. covert perching; hiding; squirreled around perch, in crevice, or out of line of sight of other lizard.
6. burrowed into substrate; frequently with head only above surface.
Display behavior:
Units of male behavior observed in contexts of nonspecific disturbance, intrusion or
activity of a conspecific males, or intrusion or activity of a female.
AD pushup/dewlap only (vertical movements of anterior body with brief extension of red dewlap;
constitutes the 2- to 4-set “assertion” display, response to a broad spectrum of perturbations)
ET extended throat (unique element of “threat” display, sustained expansion of the throat profile,
response to possible conspecific intrusion)
SE sagittal expansion (unique element of “challenge” display, sustained posture incorporating ET and
displaying expanded lateral profile, response to active conspecific intruder)
RN rapid nod of head (unique feature of “courtship” response to female; typically followed by approach
and attempt at copulation)
D Instantaneous scan or 30-set indent sampling of three 4-hr time bins during photoperiod (first and last hours
not counted); dyad tests conducted for a minimum of 5 min.
costerone levels are known to be relatively high
(Green and Greenberg, unpublished data). Total time
betwen the initial disturbance and decapitation was
less than 20 sec. Total androgen (ANDR) and cortico-
sterone (CS) were chromatographically separated and
their circulating levels were then determined by radio-
immunoassay using established procedures (Moore
et
al., 1985; Whittier et al., 1987). The intraassay coefft-
cient of variation (CV) and the interassay CV were 6.0
and 7.0%, respectively, for total androgens, and 3.3
and 13.0% for corticosterone. Accuracy values aver-
aged 95.9% for androgens and 86.2% for corticoste-
rone.
Statistical procedures.
Analyses of the contribution
of the variables documented to the differences be-
tween groups (“winners/dominants” and “losers/
subordinates”) employed multiple discriminant analy-
sis (Pimental and Frey, 1978). Discriminant scores
were used as a continuous measure of dominance, the
rationale being that those characters most effective in
distinguishing between or characterizing behaviors of
animals in two conditions will be most heavily
weighted by the multiple analysis. This procedure ef-
fectively maximizes the distances between two popu-
lations (winner/loser or dominant/subordinate) for the
variables documented. A dominance index (DI) was
derived by multiplying each variable documented by
its respective canonical coefftcient (color-dif, + 0.17;
site-dif, + 1.04; log CS, -0.09; log ANDR, -0.58).
The behavioral variables, body color and perch-site
selection, were represented as the difference between
the average pre- and postencounter values to minimize
the variability between individuals. The signitlcances
of hormonal differences attributable to outcomes of
encounters and social dominance were inferred by
contrasting the sampled hormone levels manifest in
each group with that of a control group randomly se-
lected from the same pool of subjects prior to experi-

RESPONSES TO AGGRESSION AND DOMINANCE
249
mentation. The endocrine variables, nanograms per
milliliter of plasma ANDR and CS, were log trans-
formed for statistical analysis. The significance of dif-
ferences between color and site changes and the hor-
mone levels of the winners and losers at the three sam-
pling intervals was analyzed by Welch’s modified t
test. The significance of hormone changes following
the initial interaction was determined by MANOVA;
courtship data was analyzed by Fisher’s exact proba-
bility test.
RESULTS
Body color and perch-site selection were
immediately and obviously affected by
fighting; circulating levels of ANDR, but
not of CS, were also affected as revealed by
the hormone assays.
Dominance index.
A multiple discrimi-
nant analysis indicated that 91% of the dif-
ference between groups could be accounted
for by a single discriminant function (ca-
nonical variable). This represents an index
of relative social dominance (DI) and re-
flects the confidence with which the con-
stituent variables of the index could be used
to distinguish a dominant from a subordi-
nate individual. The significance of the in-
dex (Z’ < 0.0001, Roy’s greatest root) is at-
tributable primarily to the contributions of
three of the four variables: perch height
change, body color change, and circulating
ANDR levels; CS levels did not contribute
significantly to the index.
Color differences.
Lizards that won
fights showed little or no change in color
from that seen before the encounter. Los-
ers, on the other hand, showed some vari-
ability in the first hour following a fight, but
generally were darker in body color than in
their preencounter period (Fig. 1). Because
of their initial variability, the body color
scores of losers were significantly lower
(indicating darker) than their prefight
scores only at 1 day and 1 week. The dif-
ference in body color between winners and
losers, however, was significant at all times
(1 hr,
F
= 6.21, prob >
F = 0.022;
1 day,
F
= 10.21, prob >
F = 0.007;
1 week,
F =
10.61, prob >
F
= 0.0035). The correlation
h-l
0 Dominant
0.2 - Subordinate LIGHTER
0.0
-0.2
-0 4
-0 6 DARKER
I.I
ONEHOUR ONE DAY ONE WEEK
FIG.
1. Changes in body color scores (see Table 1)
following aggressive interaction in cohabiting green
anole lizards
A. cnrolinensis
sampled at 1 hr (n = 12
pairs), 1 day (n = 10 pairs), and 1 week (n = 15 pairs)
following an aggressive interaction. Positive changes
indicate lighter color, negative changes indicate
darker.
between the extent of color change and the
dominance index (the first discriminant
function) was 0.62.
Site differences.
Following fights,
changes in site-selection scores (site-dif)
were evident as winners adopted higher
perch sites and losers adopted lower ones
(Fig. 2). The difference in perch-site scores
between winners and losers was significant
at all sampling intervals (1 hr,
F =
20.18,
prob >
F
= 0.0002; 1 day,
F
= 8.30, prob
>
F
= 0.0129; 1 week,
F
= 35.07, prob >
F
= 0.0001); however, the change from
prefight levels was significant for winners
only at 1 hr and 1 day. The postfight perch-
.” L- I
0.5 HffiHER
0.0
-0 5
-1 .o
-1.5 LOWER
-2.0 ! I
ONE HOUR ONE DAY ONEWEEK
FIG.
2. Changes in perch-site selection scores (see
Table 1) following aggressive interaction in cohabiting
green anole lizards
A. carolinensis
sampled at 1 hr (n
= 12 pairs), 1 day (n = 10 pairs), and 1 week (n = 15
pairs) following an aggressive interaction. Positive
changes indicate higher perch selected, negative
changes indicate lower.

250
GREENBERG AND CREWS
site scores of losers indicated that the adop-
tion of perch sites was significantly lower
than their prefight sites at 1 day and 1 week.
The correlation between the extent of site
change and the dominance index was 0.93,
indicating a large contribution of this vari-
able to the distinction between losers and
winners following a fight.
Reproductive activity.
Both winners and
losers recovered some courtship activity af-
ter fighting as indicated by the number of
displays incorporating rapid nods, but win-
ners had higher levels than losers at all
times tested, attaining sexual activity levels
comparable to their prefight levels by 1
week (Fig. 3). This was significantly higher
than the activity of losers (Fisher’s exact
test,
P = 0.026).
At 1 hr postfight, winners displayed 66%
of their prefight level of courtship activity
while none of the losers courted. One day
following a fight, winners and losers
courted at 77 and 15% of that group’s pre-
tight levels, respectively. One week follow-
ing a fight, winners manifested 100% of
their pretight levels of courtship head nods,
while only 12 of the losers performed this
display at prefight levels in response to in-
troduced females.
Androgen levels.
The mean level of cir-
culating total ANDR in winners was higher
than that found in control subjects at all
sampling times, but the elevation was sig-
nificant only 1 hr
(P
= 0.003 Roy’s greatest
100 SUBORDINATE
ii *O
2 60
B
g 40
20
ONE WR ONE DAY ONEWEEK
FIG.
3. Percentage of control values of courtship
behavior shown by subjects in each group during the
hour (n = 9 pairs), day (n = 10 pairs), or week (n =
15 pairs) following an aggressive interaction.
root). The mean levels of circulating andro-
gen in losers, while slightly depressed,
were not significantly lower at any sam-
pling time from that of control subjects.
However, because of this slight depression
in losers, the difference in mean hormone
levels between winners and losers was sig-
nificantly different at 1 hr and 1 week after
combat, but not at 1 day (Table 2). The
overall correlation between ANDR levels
and the dominance index was -0.66.
Corticosterone levels.
The mean levels of
circulating CS in either winning or losing
subjects were not significantly different
from those of control subjects at any of the
sampling times. Further, there were no sig-
nificant differences between winners and
losers at different times (Table 2). The
overall correlation between CS and the
dominance index was -0.12.
DISCUSSION
Following an initial aggressive interac-
tion, winning and losing green anole lizards
manifested significant differences in body
color, perch-site selection, and circulating
levels of androgen, but not in the circulat-
ing concentrations of CS. These differences
were generally sustained through the sub-
sequent period of cohabitation and were as-
sociated with the relative social dominance
of individual lizards.
Behavior.
The agonistic interactions be-
tween adult male green anole lizards in-
volve characteristic exchanges of mutual
stalking, display, and body color change
(Greenberg, 1977). The changes in color are
relatively rapid and indicate fluctuating lev-
els or availability of several chroma-
totrophic hormones, all of which are also
associated with effects on behavior as well
as physiological response to environmental
stressors (Greenberg and Crews, 1983).
Lizards that lost agonistic interactions
manifested body color darkening, fre-
quently several seconds before any obvious
action indicated the outcome (Sigmund,

RESPONSES TO AGGRESSION AND DOMINANCE
251
TABLE 2
RELATIVE HORMONE LEVELS OF CORTICOSTERONE (CS) AND ANDROGEN (ANDR) IN WINNERS AND LOSERS
OF AGONISTIC ENCOUNTERS
Variable
Winners/dominants Losers/subordinates
Levels (n) Levels (n) F” F’rob >
F
Corticosterone (% of control level)*
1 hr
1 day
1 week
Androgen (% of control levels)’
1 hr
1 day
1 week
124% (11) 149% (11)
0.02 0.89
172% (8) 131% (8)
0.33 0.57
107% (13) 109% (12)
0.49 0.49
470% (12) 81% (10) 8.78 0.008
127% (8)
94% (9) 0.13 0.73
104% (14)
60% (12) 8.46 0.008
’ rig/ml log transformed before
F
statistic calculated.
b Mean control level of CS = 6.82 r&ml (SEM = 1.45) (n = 9).
c Mean control level of ANDR = 1.64 r&ml (SEM = 0.55) (n = 9).
1978). Compared to their prelight behavior,
losers were also found at lower perch sites,
an indication of reduced social dominance
(Greenberg
et
al., 1984).
The darker body color manifest by sub-
ordinate
A. carolinensis
was associated in
earlier studies (Greenberg
et
al., 1984) with
a physiological stress response. This dark-
ening is probably attributable to increased
availability of the pituitary peptide, melan-
otropin (MSH), closely related to ACTH
and known to parallel its release in re-
sponse to CRF (Proulx-Ferland
et
al.,
1982). Recently, the level of circulating
MSH in
A. carolinensis
has been found to
be sensitive to social and environmental
stressors (Greenberg and Chen, 1987). It is
reduced in response to the acute stress as-
sociated with aggression, possibly because
of catecholamine suppression of its release
from the pituitary, but significantly ele-
vated in long-term social subordination
(Greenberg
et al.,
1986).
Courtship behavior was also markedly
reduced in losers at all sampling intervals
following an interaction; winners showed a
slight decrease following a fight but at-
tained prefight levels of activity within 1
week. Females were receptive, as indicated
by their expression of slow head nods
(Greenberg, 1977), even in response to sub-
ordinates. There was no indication that fe-
males discriminated males on the basis of
body color or were differentially responsive
to dominants and subordinates, but they
did appear more responsive to more active
males, typically the dominant.
Androgen levels.
The most dramatic dif-
ference between groups was the high circu-
lating androgen levels in winners sampled
at 1 hr. The levels of circulating androgen
were also significantly greater in winners
than in losers at 1 week postencounter.
This finding recalls the short-term fluctua-
tions in the ANDR levels associated with
changes in behavior following the aggres-
sive encounters of male songbirds (Wing-
field and Ramenofsky, 1987; Wingfield and
Moore, 1987), which are most apparent
during the period in which social relation-
ships are being established (Wingfield
et
al.,
1987). In male mountain spiny lizards,
Sceloporus jarrovi,
on the other hand, al-
though behavior following agonistic inter-
actions is altered, Moore (1988) found no
detectable changes in circulating levels of
testosterone. The studies of
S. jarrovi,
however, were different from those of
A.
carolinensis
in several significant details
apart from the species difference. For ex-
ample, in the study of
S. jarrovi,
territorial
aggression was elicited from free-living

252
GREENBERG AND CREWS
males in the field by tethered intruders that
were removed immediately after the inter-
action.
It appears that at least in some contexts,
physiological stress and a depressed hypo-
thalamic-pituitary-gonadal axis are not in-
variably linked. In song sparrows, CS treat-
ment does not affect plasma levels of LH,
and gonads are maintained in a near-
functional state, enabling a rapid resump-
tion of reproductive activity as soon as the
adrenal stress response abates (Wingfield
and Silverin, 1986). Such a response might
be of considerable utility to social subordi-
nates in a community where mortality of
dominants is high, possibly due to their
more conspicuous body color and perch
site selection.
Interestingly, there is an apparent corre-
spondence between androgen and perch
heights at which lizards are found. At the
time androgen is highest, winners are found
at significantly higher perches. Subordi-
nates, on the other hand, show the lowest
perch height selection scores at 1 week, the
time at which androgen is most depressed.
This observation is consistent with others
on this species: isolated castrated males, fe-
males, and juveniles will spontaneously
perch at lower heights than isolated intact
males (Greenberg and Hake, 1990; Green-
berg, unpublished data) as will castrated
males when they win aggressive interac-
tions (Greenberg
et
al., 1984). Such obser-
vations indicate that endocrine factors may
be as significant a variable in microhabitat
selection as is social experience.
Previous workers found that increased
testis weight and elevated plasma testoster-
one levels were associated with dominance
behavior in three of six green anoles (Pear-
son
et
al., 1976). In long-term (4 month)
undisturbed hierarchies, however, the an-
drogen levels of dominants were not higher
than those of subordinates (cited in Pearson
et
al., 1976). In the present study, ANDR
levels of dominants at 1 week were similar
to levels seen in the control group, but at
this time the subordinates showed a level of
circulating ANDR well below that of con-
trols. The control value of ANDR (mean,
1.64; SEM, 0.55) is lower than that seen in
another study of captive
Anolis
(Crews
et
al., 1974), but in the earlier study, the males
were long-term experimental animals sub-
jected to daily tests of courtship and aggres-
sion. While the results indicate that social
experience affects levels of circulating
ANDR, we cannot discount the possibility
that ANDR was highly elevated in future
winners of fights. The control values of
ANDR, however, were determined in indi-
viduals randomly selected from the same
pool of subjects as those that interacted so-
cially.
Subordinate green anoles were also re-
productively inactive, perhaps correspond-
ing to reduced ANDR levels, although
there is no direct evidence on how low an-
drogen levels must be before reproductive
dysfunctions are manifest. An elevated cir-
culating androgen level is often positively
correlated with reproductive activity
(Crews and Silver, 1980) and its depression
is associated with physological stress
(Greenberg and Wingfield, 1987). This
could be due to a cross-system feedback
inhibition of gonadotropin attributable to
adrenal androgen (Kime
et
al., 1980; and
see Christian, 1980) secondary to increased
adrenal activity, a classic indicator of the
physiological stress response. Alterna-
tively, there could be an altered sensitivity
of the testes to gonadotropin.
Corticosterone levels.
There was a slight
but comparable elevation in CS levels in
both winners and losers. This is in contrast
to an earlier study (Greenberg
et al.,
1986)
in which CS levels were significantly ele-
vated in subordinate males cohabiting with
a dominant following an initial aggressive
encounter. There are several possible rea-
sons for the present finding of relative
“stability” of CS levels within and between
the groups at varying times and in previous
research: First, in the present study, no

RESPONSES TO AGGRESSION AND DOMINANCE
253
more than 1 week had passed before blood
sampling, while the subordinates in the ear-
lier study were sampled after at least 3
weeks. In a study of rats, the response of
CS to chronic environmental stressors was
constant for the first 2 weeks but increased
markedly in subsequent weeks (Vogel and
Jensch, 1988). Second, short-term CS fluc-
tuations might be masked by already mildly
elevated levels in both groups as a function
of captivity (Grassman and Crews, 1987).
In baboons, comparable elevations in the
levels of glucocorticoids are shown by both
dominants and subordinates in response to
stress, although dominants have lower
basal concentrations (Sapolsky, 1986).
Third, in the present study, lizards were
subject to the perturbation of courtship
tests during the period of cohabitation; in
the earlier study in which a CS increase in
subordinates was observed, individuals
were undisturbed for the entire period of
cohabitation.
An inhibitory effect of CS on male ag-
gressive behavior and testicular function
was observed by Tokarz (1987) in a conge-
ner,
A. sagrei.
The effective subordination
of
A. carolinensis
in the absence of a dis-
tinct CS change in the present study, how-
ever, indicates that the CS-related inhibi-
tion of aggression seen by Tokarz is part of
a more complex ensemble of adaptive re-
sponses to a specific form of acute stress.
The CS response of
A. carolinensis,
while slight, might have been sufficient to
activate a mechanism similar to that de-
scribed in squirrel monkeys. When mon-
keys were allowed to establish social rela-
tions, plasma cortisol levels rose in all sub-
jects, but at 24 hr (but not 3 hr or 1 week),
only dominant males showed increases in
testosterone (Coe et al., 1982). In the first
hour following the acute stress of rapid cap-
ture and immobilization, high-ranking male
baboons showed an increase in ANDR lev-
els while subordinate males responded by
declines (Sapolsky, 1986). In both cases,
luteinizing hormone concentrations were
similarly suppressed, indicating a periph-
eral mechanism for the effect of acute
stress on dominants. This was attributed in
part to reduced sensitivity of tastes of tes-
tes of high-ranking males to glucocorticoid
suppression. However, Sapolsky’s (1986)
findings of attenuated transient ANDR rise
in dominants following stress-induced cate-
cholamine blockade indicate that sympa-
thetic epinephrine and norepinephrine are
acting peripherally to increase ANDR con-
centrations, possibly by means of altered
blood circulation through the testes and/or
direct stimulation of ANDR release. In
pigs, exogenous ACTH can stimulate a sig-
nificant and relatively rapid but transient
elevation in testicular testosterone secre-
tion by means of a mechanism that does not
involve pituitary gonadotropins (Fenske et
al., 1981).
ACKNOWLEDGMENTS
We gratefully acknowledge the expert technical help
of Yuki Morris with RIA procedures, the advice on
statistics of Dr. Ralph O’Brien, and the laboratory as-
sistance of Donna Layne. This work was made possi-
ble by NSF Grant BNS-8406028 and a University of
Tennessee Faculty Research Award to N. Greenberg
and MH41770 and RSA
MH00135
to D. Crews.
REFERENCES
Andrew, R. J. (1972). Recognition processes and be-
havior with special reference to effects of testos-
terone of persistence. In “Advances in the Study
of Behavior” (D. S. Lehrman, R. A. Hinde, and
E. Shaw, Eds.), Vol. 4, pp. 175-208. Academic
Press, New York.
Beach, F. A. (1974). Behavioral endocrinology and
the study of reproduction.
Biol. Reprod. 10, 2-18.
Bohus, B., and de Wied (1980). Pituitary-adrenal sys-
tem hormones and adaptive behavior.
In
“Gen-
eral, Comparative, and Clinical Endocrinology of
the Adrenal Cortex” (I. Chester Jones and I. W.
Henderson, Eds.), Vol. 3, pp. 256-347. Academic
Press, New York.
Christian, J. J. (1980). Endocrine factors in population
regulation. In “Biosocial Mechanisms of Popula-
tion Regulation” (M. N. Cohen, R. S. Malpass,
and H. G. Klein, Eds.), pp. 55-115. Yale Univ.
Press, New Haven.

254
GREENBERG AND CREWS
Coe, C. L., Frankhn, D., Smith, E. R., and Levine, S.
(1982). Hormonal responses accompanying fear
and agitation in the squirrel monkey. Physiol. Be-
hav. 29, 1051-1057.
Coe, C. L. , Mendoza, S. P., and Levine S. (1979).
Social status constrains the stress response in the
squirrel monkey.
Physiol. Behav. 23, 633-638.
Crews, D. (1975). Inter- and intraindividual variation
in display patterns in the lizard,
Anolis carolinen-
sis. Herpetologica 31, 37-47.
Crews, D., and Garrick, L. D. (1970). Methods of in-
ducing reproduction in captive reptiles. In “Re-
productive Biology and Diseases of Captive
Reptiles” (J. B. Murphy and J. T. Collins, Eds.),
pp. 4%70. Sot. Stud. Amphib. Rept., Miami Uni-
versity, Oxford, Ohio.
Crews, D., Rosenblatt, D. S., and Lehrman, D. S.
(1974). Effects of unseasonal environmental regi-
men, group presence, group composition, and
male’s physiological state on ovarian recrudes-
cence in the lizard,
Anolis carolinesis. Endocri-
nology 94,
541-547.
Crews, D., and Silver, R. (1980). Reproductive phys-
iology-behavior interactions in nonmammalian
vertebrates. In “Handbook for Behavioral
Neurobiology” (R. W. Coy and D. W. Pfaff,
Eds.). Plenum, New York.
Fenske, M., Holtz, W., Pitzel, L., and Konig, A.
(1981). Effects of ACTH-induced testosterone re-
lease on the secretion of pituitary gonadotrophin
and prolactin release in male pigs.
In
“Gonadal
Steroids and Brain Function, Exp. Brain Res.
Suppl. 3” (W. Wuttke and R. Harowski, Eds.),
pp. 342-343. Springer-Verlag, New York.
Gandelman, R. (1983). Gonadal hormones and sensory
function.
Neurosci. Biobehav. Rev. 7,
l-17.
Grassman, M., and Crews, D. (1987). Dominance and
reproduction in an all-female lizard species. Be-
hav. Ecol. Sociobiol.
21, 141-147.
Greenberg, N. (1977). A neuroethological study of dis-
play behavior in the lizard
Anolis carolinensis
(Reptilian, Lacertilia, Iquanidae).
Amer. Zool.
17,
191-201.
Greenberg, N. (1983). Central and autonomic aspects
of aggression and dominance in reptiles. In “Ad-
vances in Vertebrate Neuroethology” (J.-P.
Ewert, R. R. Capranica, and D. J. Ingle, Eds.),
pp. 1135-l 144. Plenum, New York.
Greenberg, N., and Chen, T. (1987). Aggression and
social submissiveness alter melanotropin (MSH)
in the lizard.
Amer. Zool.
27(4), 49A. [Abstract]
Greenberg, N., and Crews, D. (1983). Physiological
ethology of aggression in amphibians and rep-
tiles.
In
“Hormones and Aggressive Behavior”
(B. Svare, Ed.), pp. 46%506. Plenum, New York.
Greenberg, N., and Hake, L.
(1990).
Hatching and
neonatal behavior of the lizard,
Anolis carolinen-
sis. J. Herpetol. in
press.
Greenberg, N., and Wingfield, J. C. (1987). Stress and
reproduction: Reciprocal relationships.
In
“Hor-
mones and Reproduction in Fishes, Amphibians,
and Reptiles” (D. 0. Norris and R. E. Jones,
Eds.), pp. 461-503. Plenum, New York.
Greenberg, N., Chen, T., and Crews, D. (1984). Social
status, gonadal state, and the adrenal stress re-
sponse in the lizard,
Anolis carolinensis. Horm.
Behav. 18,
l-l 1.
Greenberg, N., Chen, T., and Vaughan, G. (1986).
Melanotropin is altered by acute and chronic so-
cial stress in lizards.
Sot. Neurosci. Abstracts 2,
834. [Abstract]
Hadley, M. E., and Goldman, J. M. (1969). Physiolog-
ical color changes in reptiles.
Amer. Zool. 9(2),
489-504.
Kime, D. E., Vinson, G. P., Major, P. W., and Kil-
patrick, R. (1980). Adrenal-gonadal relationships.
In “General, Comparative, and Clinical Endocri-
nology of the Adrenal Cortex” (I. Chester Jones
and I. W. Henderson, Eds.), Vol. 3, pp. 133-264.
Academic Press, New York.
Kleinholz, L. H. (1938). Studies in reptilian color
change. III. Control of light phase and behavior of
isolated skin.
J. Exp. Zool. 15, 492-499.
Leshner, A. I., and Politch, J. A. (1979). Hormonal
control of submissiveness in mice: Irrelevance of
the androgens and relevance of the pituitary adre-
nal hormones.
Physiol. Behav. 22,
531-534.
Licht, P. (1967). Environmental control of annual tes-
ticular cycles in the lizard
Anolis carolinensis.
In-
teraction of light and temperature in the initiation
of testicular recrudescence.
J. Exp. Zool. 165,
505616.
Licht, P. (1971). Regulation of the annual testis cycle
of photoperiod and temperature in the lizards,
An-
olis carolinensis. Ecology 52, 240-252.
Meier, A. H., Trobec, T. N., Haymaker, R., Mac-
Gregar, R., III, and Russo, A. C. (1973). Daily
variations in the effects of handling on fat storage
and testicular weight in several vertebrates.
J.
Exp. Zool.
184, 281-288.
Moore, M. C., Whittier, J. M., and Crews, D. (1985).
Sex steroid hormones during the ovarian cycle of
an all-female, parthenogenetic lizard and their
correlation with pseudosexual behavior. Gen.
Comp. Endocrinol. 60,
144-153.
Nie, N. H., Hull, C. H., Jenkins, J. G., Steinbrenner,
K., and Bent, D. H. (1975). “Statistical Package
for the Social Sciences,” 2nd ed. McGraw-Hill,
New York.
Oades, R. D. (1979). Search and attention: Interac-
tions of the hippocampal-septal axis, adrenocorti-
cal and gonadal hormones.
Neurosci. Biobehav.
Rev. 3,
31-48.

RESPONSES TO AGGRESSION AND DOMINANCE
255
Pearson, A. K., Tsui, H. W., and Licht, P. (1976).
Effect of temperatures on spermatogenesis, on
the production and action of androgens and on the
ultrastructure of gonadotropic cells in the lizard
Anolis carolinensis. J. Exp. Zool.
195, 291-304.
Pimentel, R. A., and Frey, D. F. (1978). Multivariate
analysis of variance and discriminant analysis. In
“Quantitative Ethology” (P. W. Colgan, Ed.),
pp. 247-274. Wiley, New York.
Proulx-Ferland, L., Labrie, F., Dumont, D., and
Cote, J. (1982). Corticotropin-releasing factor
stimulates secretion of melanocyte-stimulating
hormone from the rat pituitary.
Science
217, 62-
64.
Sandman, C. A., Miller, L. H., Kastin, A. J., Shally,
A. V., and Kendall, J. W. (1973). Neuroendo-
crine responses to physical and psychological
stress.
.I. Comp. Physiol. Psychol. 34, 386-390.
Sapolsky, R. M. (1986). Stress-induced elevation of
testosterone concentrations in high ranking ba-
boons: Role of catecholamines.
Endocrinology
118(4), 1630-1635.
Sigmund, W. R. (1978). The analysis of visual displays
in the lizard,
Anolis carolinensis.
Unpublished
Ph.D. dissertation, Indiana University, Blooming-
ton. 102 pp.
Tokarz, R. R. (1987). Effects of corticosterone treat-
ment on male aggressive behavior in a lizard
(An-
olis sagrez]. Horm. Behav.
21, 358-370.
Vogel, W. H., and Jensh, R. (1988). Chronic stress
and plasma catecholamine and corticosterone lev-
els in male rats.
Neurosci. Lett. 87,
183-188.
Whittier, J. M., Mason, R. T., and Crews, D. (1987).
Plasma steroid hormone levels of female red-sided
garter snakes,
Thamnophis sirtalis parietalis:
Re-
lationship to mating and gestation. Gen.
Comp.
Endocrinol. 67, 3343.
Wingfield, J. C., and Silverin, B. (1986). Effects of
corticosterone on territorial behavior of free-
living male song sparrows,
Melospiza melodul.
Horm. Behav. 20,
405417.
Wingtield, J. C., Ball, G. F., Dufty, A. M., Jr., Heg-
ner, R. E., and Ramenofsky, M. (1987). Testos-
terone and aggression in birds.
Amer. Sci. 75,
602608.