Relationships between Hormones and Aggressive Behavior in Green Anole Lizards: An Analysis Using Structural Equation Modeling

Hormones and Behavior 42, 192–205 (2002) doi:10.1006/hbeh.2002.1811

Relationships between Hormones and Aggressive Behavior in Green Anole Lizards: An Analysis Using Structural Equation Modeling Eun-Jin Yang and Walter Wilczynski Department of Psychology and Institute for Neuroscience, University of Texas at Austin, Austin, Texas 78712 Received September 17, 2001; revised February 1, 2002; accepted February 4, 2002

We investigated the relationship between aggressive behavior and circulating androgens in the context of agonistic social interaction and examined the effect of this interaction on the androgen–aggression relationship in response to a subsequent social challenge in male Anolis carolinensis lizards. Individuals comprising an aggressive encounter group were exposed to an aggressive conspecific male for 10 min per day during a 5-day encounter period, while controls were exposed to a neutral stimulus for the same period. On the sixth day, their responses to an intruder test were observed. At intervals, individuals were sacrificed to monitor plasma androgen levels. Structural equation modeling (SEM) was used to test three a priori interaction models of the relationship between social stimulus, aggressive behavior, and androgen. Model 1 posits that exposure to a social stimulus influences androgen and aggressive behavior independently. In Model 2, a social stimulus triggers aggressive behavior, which in turn increases circulating levels of androgen. In Model 3, exposure to a social stimulus influences circulating androgen levels, which in turn triggers aggressive behavior. During the 5 days of the encounter period, circulating testosterone (T) levels of the aggressive encounter group followed the same pattern as their aggressive behavioral responses, while the control group did not show significant changes in their aggressive behavior or T level. Our SEM results supported Model 2. A means analysis showed that during the intruder test, animals with 5 days of aggressive encounters showed more aggressive responses than did control animals, while their circulating androgen levels did not differ. This further supports Model 2, suggesting that an animal’s own aggressive behavior may trigger increases in levels of plasma androgen. © 2002 Elsevier Science (USA)

Key Words: aggression; social experience; androgens; structural equation modeling; lizards.

192

The relationship between aggressive behavior and circulating androgens has been investigated in a variety of territorial species. The fluctuation in plasma level of androgen between breeding and nonbreeding seasons coincides with fluctuation in aggression between male conspecifics in many species (in lizards, Crews, 1974, 1975; Klukowski and Nelson, 1998; in birds, Harding, 1981; Wingfield, Ball, Dufty, Hegner, and Ramenofsky, 1987). This suggests two possibilities: changes in behavior cause changes in circulating androgen concentrations or changes in androgens cause changes in behavior. Although extensive studies have been conducted to elucidate the relationship between androgens and aggression, little attempt has been made to test specific models of the interrelationships among social stimulus, hormone, and aggression in intact animals (for an example, see Burmeister and Wilczynski, 2000). Several studies show that behavior influences plasma androgen concentrations. A brief aggressive encounter increases circulating testosterone (T) (Harding, 1981; Wingfield, 1984) and gonadotropin levels (Wingfield, 1985) in birds early in the breeding season. Winning a staged fight results in higher levels of plasma androgen than losing in Anolis lizards (Greenberg and Crews, 1990). Furthermore, changes from nonterritorial to territorial status in cichlid fish, with accompanying increases in territorial aggression, coincide with increases in gonad size, gonadotropinreleasing hormone-immunoreactive (GnRH-ir) cell size, and levels of GnRH messenger RNA in the preoptic area (White and Fernald, 1997). In other fishes, such as bluehead wrasse and anemonefish that adopt alternative male reproductive tactics, the territorial

0018-506X/02 $35.00 © 2002 Elsevier Science (USA) All rights reserved.

Hormonal Responses to Aggressive Behavior

males show more GnRH-ir neurons than nonterritorial males (for a review, see Foran and Bass, 1999). Other studies show that changes in androgen levels in response to a social stimulus can trigger aggressive behaviors in a territorial owner. Hormonal manipulation studies report increases in territorial aggression when animals received exogenous testosterone (in lizards, Adkins and Schlesinger, 1979; Crews, 1974; DeNardo and Licht, 1993; DeNardo and Sinervo, 1994; Moore, 1987a, 1988; in birds, Wingfield, 1985) or decreases in aggression after castration (in lizards, Adkins and Schlesinger, 1979; Crews, Traina, Wetzel, and Muller, 1978; DeNardo and Licht, 1993). In addition, androgen receptor (AR) antagonists successfully blocked the effect of androgens on aggression in a variety of vertebrates (in birds, Beletsky, Orians, and Wingfield, 1990; Hegner and Wingfield, 1986; Schwabl and Kriner, 1991; Soma, Sullivan, and Wingfield, 1999; in lizards, Tokarz, 1987, 1995; in mice, Clark and Nowell, 1980; in rats, Taylor, Haller, Rupich, and Weiss, 1984). For example, Beletsky et al. (1990) demonstrated that red-winged blackbird territory owners given T engaged in more aggressive behaviors than controls, while territory owners given an AR antagonist (Flutamide) and ATD, an aromatase blocker, showed reduced aggressive behavior and were thereby unable to maintain territories. In addition, Anolis lizards treated with anti-androgens were less aggressive toward conspecifics (Tokarz, 1987) and less likely to acquire and defend high-quality territory than sham-treated lizards (Tokarz, 1995). This androgen-dependency of the expression of aggressive behavior in various species appears to be under the strong influence of social context. For example, the positive correlation between aggression and plasma T levels is more pronounced in socially unstable situations like the territorial intrusion of a conspecific male or the disruption of a previously stable hierarchy (Rose, Bernstein, Gordon, and Catlin, 1975; Sapolsky, 1983b; Wingfield, 1985). The widely accepted “challenge hypothesis,” which asserts that a strong positive relationship between aggression and T prevails only when there are frequent aggressive interactions among males (e.g., during the establishment of territory or a dominance hierarchy), captures the importance of social context on the emergence of the hormone– behavior relationship (Wingfield et al., 1987). In Anolis lizards, the detrimental effect of gonadectomy on aggression was manifested only in a novel environment (Crews, 1980). Furthermore, the outcome of a single agonistic encounter (Greenberg,

193

Chen, and Crews, 1984a; Greenberg and Crews, 1990; Knapp and Moore, 1995) and length of interaction with the same individual after the agonistic encounter (Summers and Greenberg, 1994, 1995) may influence hormone responses and neuronal mechanisms associated with aggression. Provided there are dynamic interactions between androgen and aggressive behavior when a territorial owner faces an aggressive intruder, the coinciding increases in both aggressive behavior and plasma androgen can be explained if either behavioral changes drive hormonal changes or hormonal changes drive behavioral changes. This interrelationship between aggressive behavior and androgen can be summarized into three possible interaction models (Fig. 1). First, exposure to an appropriate social stimulus influences androgen and aggressive behavior independently. Second, a social stimulus triggers aggressive behavior, and in turn this behavior causes an increase in circulating levels of T. Third, exposure to a social stimulus influences circulating androgen levels, which in turn triggers aggressive behavior. Such models have been proposed previously in the context of reproductive behavior (Burmeister and Wilczynski, 2000). In this study, we indirectly tested the relationship between social stimulus, aggressive behavior, and androgens over the course of several days of encounters in the temperate lizard, Anolis carolinensis. Anoles have distinct aggressive display behaviors (Crews, 1974; DeCourcy and Jenssen, 1994; Greenberg and Noble, 1944; Greenberg, 1977; Jenssen, Lovern, and Congdon, 2001), whose physiological and neuronal correlates are well documented (Baxter, 2001; Baxter, Ackermann, Clark, and Baxter, 2001a; Baxter, Clark, Ackermann, Lacan, and Baxter, 2001b; Korzan, Summers, Ronan, Renner, and Summers, 2001; Lovern, McNabb, and Jenssen, 2001; Summers, 2001). We used three interaction models, described above, as our working hypotheses to examine a dynamic, reciprocal hormone– behavior relationship in aggressive interactions. We performed structural equation modeling analyses to find an interaction model that best describes the pattern in which social stimulus, hormone, and aggressive behavior interact with one another in dynamic male–male encounters. Structural equation modeling (SEM) is widely used in social sciences to map out complex causal interactions among multiple factors influencing the studied phenomenon (Aspinwall and Taylor, 1992; Crowley and Fan, 1997). Some of these factors may influence the outcome directly, while other factors influence it indirectly, or both. This

194

Yang and Wilczynski

FIG. 1. Three proposed a priori structural equation models with their corresponding paths (path diagrams) are illustrated. Boxes represent observed or measured variables. An oval represents a latent variable: the aggressive behavior factor, derived from measured variables, PU (push-up) and LC (lateral compression). E indicates residuals in a measured variable, and D indicates residuals in a latent variable. The solid arrows indicate directional influence from the origin of the arrow, while dotted arrows indicate residuals in corresponding direct paths. Model 1 shows the simultaneous and independent influence of social stimulus on both aggressive behavior and plasma T. Model 2 shows a direct influence of social stimulus on aggressive behavior and an indirect influence of social stimulus on plasma T via aggressive behavior. Model 3 shows a direct influence of social stimulus on plasma T and an indirect influence of social stimulus on aggressive behavior via plasma T.

method is tailored toward model testing in which a specified model of the relationships can be tested against empirical relationships in gathered data. Unlike other multivariate procedures that are in essence descriptive and can incorporate either only observed

or latent variables but not both, SEM provides simultaneous statistical analysis of the entire array of observed variables and latent factors. Although it has rarely been used in studies in biological sciences (e.g., by McIntosh and Gonzalez-Lima, 1992), SEM offers a

195

Hormonal Responses to Aggressive Behavior

powerful technique for characterizing the relationships among factors controlling any social behavior, including aggression. This study consists of two parts: model testing with SEM, and comparisons of aggressive behaviors and levels of plasma androgen between groups, which had different types of encounters over 5 days, in response to a subsequent social challenge. We predicted that the pattern of changes in the level of circulating androgens would reflect that of behavioral changes as a function of aggressive encounters. If high androgen triggered by the stimulus is required to produce aggressive behavior, Model 3 would yield the best fit to the data, while if aggressive behavior is independent and rather drives androgen increase, Model 2 would yield the best fit among three a priori models.

METHODS Animals and Materials Adult male and female A. carolinensis were obtained from Charles Sullivan, Inc. (Nashville, TN). Each male was housed with a conspecific female in a clear glass cage measuring 17.8 cm (length) ⫻ 30.5 cm (width) ⫻ 45.7 cm (height) for 2 weeks prior to the experiment to allow recovery from shipping and handling stress. Each cage contained a wooden perch, and its bottom was covered with sphagnum moss. Water was supplied in gravel-filled water dishes, and crickets were provided three times a week. The light was on a 14-h light/10-h dark cycle with Vitalites and a 60-W incandescent light, with room temperature varying between 22°C (lowest) and 32°C (highest). Procedure Fifty-one males were randomly assigned to one of two different groups: the aggressive encounter group and the control group. Snout-to-vent lengths (SVL) did not differ between the two groups (F 1,49 ⫽ 0.37, ៮ ⫽ 61.3 mm, SD ⫽ 3.28). Animals in the P ⬎ 0.54, X aggressive encounter group interacted with a conspecific male in an adjacent cage for 10 min, while animals in the control group were exposed to a video clip depicting two moving green colored balls (for details, see Yang, Phelps, Crews, and Wilczynski, 2001). Briefly, the control video was constructed by recording two green balls (dimensions 4189 mm 3) bouncing on a wooden perch with a Panasonic AG-450 color

video camera and Ampex S-VHS 120 videotape. The image of the balls on the video monitor was comparable to the size of an adult male anole. Separate subsets of animals in each group (N ⫽ 5–6 per subset) were exposed to either an aggressive conspecific male or a neutral video once daily for either 1, 3, or 5 consecutive days. A fourth set (N ⫽ 12) of aggressive encounter and control group animals interacted with stimuli for 5 consecutive days and on the following day they were tested for their aggressive responses by having a conspecific male intruder placed inside their home cage (Test). The subject animals in the aggressive encounter group were exposed to a different stimulus animal each day. Stimulus conspecifics were housed individually with a female and were used every 2 days and twice per day with at least an hour interval between encounters. They were not used as experimental subjects. Anoles were not able to bite or chase off the stimulus conspecifics, because during the 5 days of the encounter period, subject and stimulus animals remained in their home cages, which were brought together to face through a glass wall for an encounter session. In each encounter session, two experimenters recorded the subject animals’ display behaviors. Immediately following the end of their encounter sessions (on day 1, 3, 5, or Test), animals were sacrificed by rapid decapitation and trunk blood was drawn. All blood samples were spun in a refrigerated centrifuge and plasma was stored at ⫺20°C until processing. Aggressive Display Measures We measured three aggressive display behaviors in A. carolinensis: dewlap (DL) extension, push-up (PU) display, and lateral compression (LC). Dewlap extension is a brief extension of the red throat, and push-up display is a characteristic up and down motion of the front part of the body. Both of these displays occur in response to a broad range of perturbations including aggression (Greenberg and Crews, 1983, 1990). Lateral compression is an enlargement of the lateral profile of the body accompanied by expansion of the throat. It is considered to be a “challenging or threatening” display toward an intruding conspecific because it exclusively occurs in aggressive encounters and usually precedes attacking (Greenberg and Crews, 1990), and it is suggested to be strongly associated with plasma androgen (Tokarz, 1986). The DL and PU responses are seen during aggressive encounters in green anoles, suggesting significant overlap of display behaviors in

196

different contexts, and both behaviors can be considered as aggressive displays with less intensity than challenge displays (see DeCourcy and Jenssen, 1994; Greenberg, 1977). We counted the number of DL and PU that occurred during each trial. We treated LC as a categorical variable, and the score for LC within a given trial would be either 1 (expression of LC) or 0 (absence of LC). Hormone Assay Plasma steroid levels of T and dihydrotestosterone (DHT) were measured with radioimmunoassay (RIA) methods using chromatographic columns as previously published (Rhen, Ross, and Crews, 1999; Tousignant and Crews, 1995). Briefly, before extraction of steroids, approximately 500 cpm each of tritiated DHT and T was added to the plasma samples. After reaching equilibrium, steroids were extracted by addition of 3 ml of anesthesia-grade ether and separation of the ether phase from the aqueous phase. Then, separated steroids were placed in a dry bath under a gentle stream of nitrogen. Steroids were purified by running through water-celite:glycol-celite mixture packed columns (for details, see Tousignant and Crews, 1995). We added 4 ml of pure iso-octane to rid the sample of other hormone elutes and then added 4.5 ml of 10% ethyl acetate in iso-octane for DHT, followed by 4.5 ml of 20% ethyl acetate-iso-octane solution for T. After each steroid fraction was dried in the dry bath with a stream of nitrogen at 35°C and resuspended in 330 ␮l phosphate gelatin buffer, RIA for each plasma steroid was conducted by using antibodies DT3-351 for DHT and T3-125 for T (Endocrine Sciences, Calabasas Hills, CA). Approximately 2000 cpm of corresponding tritiated steroid and antibody was added to duplicate 100-␮l aliquots of each steroid, while an additional 100-␮l aliquot of each steroid was used for determining individual recoveries. After a 24-h incubation, bound steroids were separated from free steroids with dextran-coated charcoal–phosphate gelatin buffer. Average recoveries for DHT and T were 50 and 84%, respectively. Assay sensitivities were 59 pg DHT/ml plasma and 67 pg T/ml plasma. Intra-assay coefficients of variance were 2.9 and 12% for DHT and T, respectively. All the plasma samples were run in a single assay. Structural Equation Modeling Unlike Path analysis, which employs only measured variables, SEM is an excellent technique not only to

Yang and Wilczynski

assess complex interrelated dependence relationships but also to incorporate both measured variables and constructs (e.g., latent factors) with corresponding error terms for model testing (Byrne, 1994). By using a variance– covariance matrix of observed data as inputs, the weights, or structural coefficients, are derived through iterative estimations. The structural coefficients represent how reliable the independent variables are in explaining a dependent variable. We minimized the number of structural coefficients to be estimated because of the small sample size in our study (N ⫽ 31; for a review, see Loehlin, 1998) and therefore included only variables that are absolutely required for the model construction. Behavioral and hormonal data from six subsets in the 5-day encounter period in both aggression and control groups were used for this analysis (i.e., Days 1, 3, and 5 in aggression and control groups). We did not include data from the intruder test animals in order to avoid confounding the model with different social contexts. We included social stimulus (whether presentation of a conspecific male or an inanimate object), PU, LC, and plasma T levels during the 5-day encounter period as independent or dependent variables in this analysis. DL and plasma DHT level variables were dropped from the model because both showed very high correlations to other variables (DL to PU, and DHT to T). Incorporating highly correlated variables results in multicolinearity, which makes SEM unsolvable. Before testing structural models, we conducted a confirmatory factor analysis to construct a latent variable, “aggressive behavior,” from PU and LC. Three a priori models of the relationships among the variables including the aggressive behavior factor were tested using the software program EQS (Bentler, 1989). Figure 1 illustrates path diagrams for three models representing possible interactions among social stimulus, hormone, and aggression. The first model hypothesizes direct effects of social stimulus on both plasma T levels and aggression. The second model hypothesizes an indirect effect of social stimulus on plasma T levels mediated by aggression, which receives a direct effect from social stimulus. The third model hypothesizes an indirect effect of social stimulus on aggression mediated by plasma T levels, which receives a direct effect from social stimulus. A ␹ 2 goodness-of-fit test, testing the discrepancy between observed (data) and estimated (model) covariance matrices, was used to evaluate the fit of each model. In addition, we inspected Bentler’s Comparative Fit Index (CFI), a descriptive fit measure, and the size of standardized residuals to

197

Hormonal Responses to Aggressive Behavior

evaluate the fit of each model because the ␹ 2 goodness-of-fit test is sensitive to the sample size (Loehlin, 1998). The convergence criterion for finding optimal parameter estimates via iterations in maximum likelihood was to reproduce the estimated model covariance matrix that is closest to the sample covariance matrix. In addition, we examined the CFI, a measure of relative fit of a model on a continuum ranging from 0 to 1, with 0 indicating a null model (zero intercorrelations among variables) and 1 indicating an unrestricted model (significant correlations among all variables). A CFI value of 1 indicates a perfect fit and a CFI value close to 1 indicates a reasonable fit. Means Analysis Because DL and PU are highly correlated with each other, multivariate analysis of variance (MANOVA) was performed for DL and PU displays on Days 1, 3, and 5. Given significant differences between groups in MANOVA for both variables, separate ANOVAs for DL and PU followed. Fisher’s exact tests were performed for LC on Days 1, 3, and 5, and Test. Before analyses, we log-transformed hormone data because of heterogeneity of variances. To examine the effects of stimuli, days, and their interaction with plasma steroid levels, we performed a MANOVA with DHT and T as dependent variables followed by subsequent ANOVAs. To test differences in plasma levels of T and DHT between animals that showed LC and those that did not show LC during the 5-day encounter period or during the test, we divided all the subject animals into two groups based on their LC response, i.e., whether they showed LC or not. Then, ANOVAs were performed to test differences in T and DHT between LC and No-LC groups.

RESULTS Structural Equation Modeling: Relationship between Social Stimulus, Aggressive Behavior, and Androgen The first step in structural equation modeling was to construct a factor which represented aggressive behavior among aggressive displays measured. Since PU and LC have a hierarchical relationship between them, and yet they both occur during the aggressive encounters, we restricted the factor model to have only one factor, which reflects the latent factor aggressive be-

havior in a statistically reliable manner. Examination of a Scree plot indicated the presence of a factor with an eigenvalue greater than 1. The factor explains 71.5% of total variance, and factor loadings for both PU and LC are 0.846. Three a priori models of interaction relations were tested individually including a latent aggressive behavior factor structure. The factor loadings on PU and LC were fixed to 0.846. Structural coefficients for direct paths from social stimulus (Model 1), for an indirect path from social stimulus via aggressive behavior (Model 2), and for an indirect path from social stimulus via plasma T (Model 3) were free to be estimated by maximum-likelihood estimation methods. In Model 1, a ␹ 2 goodness-of-fit test was statistically significant, indicating a significant difference between the model and the observed data and suggesting that we reject Model 1 (␹ 32 ⫽ 14.6, P ⬍ 0.01). The CFI value for model 1 was 0.489, indicating a poor fit. The inspection of residuals (i.e., observed minus estimated) for Model 1 also suggests that Model 1 was not a good fit to the data: the largest standardized residual was 0.525 and the average off-diagonal absolute standardized residual was 0.15. For Model 2, a ␹ 2 test was not statistically significant, indicating a good fit of Model 2 (␹ 32 ⫽ 5.43, P ⬎ 0.14). The CFI value for Model 2 was 0.893, suggesting a moderate fit. The residuals for Model 2 were moderate: The largest standardized residual for Model 2 was ⫺0.188, and the average offdiagonal absolute standardized residual was 0.09. The ␹ 2 test for Model 3 was statistically significant, indicating a poor fit of the model to the data (␹ 32 ⫽ 9.53, P ⬍ 0.03), and we rejected Model 3. The CFI value for Model 3 was 0.713, indicating that the fit is poor. Residuals for Model 3 were also fairly large: The largest standardized residual for model 3 was 0.347 and the average off-diagonal absolute standardized residual was 0.14. In sum, goodness-of-fit statistics and standardized residual data suggest that among the three a priori models, only Model 2, hypothesizing the indirect effect of social stimulus on plasma T via aggressive behavior, shows a good fit to the data. Table 1 shows standardized parameter estimates of structural Model 2 with R 2 values. Behavior During the 5 days of encounter conditions, animals in the aggressive encounter group increased aggressive responses up to Day 3 and decreased during later days of encounter conditions (MANOVA Wilk’s ␭:

198

Yang and Wilczynski

TABLE 1 Standardized Parameter Estimates of Structural Model 2 Outcome variable

Predictor Social stimulus Aggressive behavior (latent factor) Error variance R2

Aggressive Lateral behavior Plasma Push-up compression (latent factor) T

cant. Immediately after both group animals were tested with a conspecific intruder in their home cage, circulating androgens did not differ between groups. In addition, the levels of androgens in both groups during the test were equivalent to those during Day 1 (Fig. 3, Table 2). Overall levels of plasma T were lower than those observed in the field (Jenssen et al., 2001). In

0.458* 0.681a 0.732 0.464

0.733 a 0.680 0.538

0.889 0.210

0.655** 0.756 0.429

Note. Significance levels are determined by a critical ratio of the unstandardized parameter estimate divided by its standard error. * P ⬍ 0.05. ** P ⬍ 0.01. a Indicates a fixed and constrained estimate in the model.

Group: F 2,24 ⫽ 5.27, P ⬍ 0.02; Day: F 4,48 ⫽ 4.08, P ⬍ 0.01; Group ⫻ Day: F 4,48 ⫽ 3.55, P ⬍ 0.02). Anoles that encountered aggressive conspecifics showed more DL and PU than anoles in the control group on Day 1 (ANOVA: DL: F 1,51 ⫽ 8.15, P ⬍ 0.01, and PU: F 1,51 ⫽ 7.01, P ⬍ 0.01) and Day 3 (DL: F 1,41 ⫽ 7.95, P ⬍ 0.01, and PU: F 1,41 ⫽ 8.42, P ⬍ 0.01), but not at Day 5 as the animals habituated (see Yang et al., 2001). In terms of LC, anoles that interacted with another conspecific male showed more LC on Day 1 [Fisher’s Exact test (two-tailed): P ⬍ 0.01] and Day 3 (P ⬍ 0.001), but not on Day 5. However, anoles with 5 days of aggressive encounters showed more LC in response to an intruder than did anoles with the neutral experience (P ⬍ 0.02, Fig. 2). Although the patterns of DL and PU from both aggressive encounter and control groups followed that of LC during the 5-day period, during the test with a conspecific intruder, DL and PU were not different between groups (Fig. 2). Hormones T levels on Day 1 did not differ between groups. In the aggressive encounter group, there was a significant effect of Day on circulating T levels (F 4,40 ⫽ 2.90, P ⬍ 0.05). T levels within the aggressive encounter group followed the same pattern as their behavioral responses with the peak level at Day 3 (F 2,12 ⫽ 3.90, P ⬍ 0.05). The control group showed no significant change over days. DHT showed a pattern similar to that seen in T levels in both groups although the changes were less robust and not statistically signifi-

FIG. 2. Behavioral responses of the aggressive encounter group (black) and neutral video group (gray) at different time points are shown. Different animals were used on each day. Each bar (N ⫽ 5–6/bar) represents the mean number of dewlap (A) and push-up (B) displays ⫹ SEM, and the percentage of animals displaying lateral compression (C). An asterisk indicates a statistically significant difference in an individual comparison between groups at ␣ ⬍ 0.05.

199

Hormonal Responses to Aggressive Behavior

Finally, when animals that showed LC were compared with those that did not show LC, there were significant differences in plasma levels of T and DHT during the 5-day encounter period (T: F 1,29 ⫽ 17.38, P ⬍ 0.001; and DHT: F 1,29 ⫽ 5.99, P ⬍ 0.03), but no differences were found immediately following the aggressive intrusion on the test day with an intruder (Fig. 4).

DISCUSSION

FIG. 3. Mean levels of plasma T (A) and DHT (B) concentrations ⫹ corresponding SEM in each subset (N ⫽ 5–6) of aggressive encounter (black) and neutral video (gray) groups at different time points. An asterisk indicates a statistically significant difference at ␣ ⬍ 0.05.

order to detect any delayed changes in plasma androgen that might have occurred as a function of this experimental procedure, we used a new set of animals for each group (N ⫽ 5–6 per group), which were subjected to the same experimental procedure, including 5-day encounters and the intruder test, and were sacrificed 1 day after the test. There were no differences between the groups in plasma androgen, and the levels of androgens in both groups were equivalent to those during Day 1 and the test (Fig. 3, Table 2).

Our results support the hypothesis that animals that experience 5 days of aggressive encounters respond more aggressively to a conspecific intruder placed into their home cage. Results in this study confirm those found earlier using video playbacks (Yang et al., 2001). Our results also demonstrated that, during the 5 days of the encounter period, circulating levels of T followed a pattern similar to that of aggressive behavior: Both T and aggression peaked on Day 3 and decreased thereafter. However, although animals with 5 days of aggressive encounters showed more aggressive responses than did control animals when they encountered an aggressive intruder in their home cage 1 day after the 5-day period, their T levels were no higher. In addition, although during the 5 days of encounters animals that showed LC had higher levels of plasma androgens than animals that did not, this difference was not observed in the intruder test. Relationships among Social Stimulus, Aggressive Behavior, and Androgen Although previous studies have demonstrated both the androgenic control of aggressive behavior and the behavioral influence on androgen levels in various species, our SEM analysis results provide a unique

TABLE 2 Androgen Levels [Mean ng/ml Plasma (SEM, N)] of Each Group

T (ng/ml) Aggressive encounter group Control DHT (ng/ml) Aggressive encounter group Control

Day 1

Day 3

Day 5

Test

1 day after the test

0.93 (0.40, N ⫽ 5) 0.74 (0.15, N ⫽ 5)

4.89 (2.34, N ⫽ 5) 2.60 (1.08, N ⫽ 5)

0.80 (0.21, N ⫽ 6) 1.53 (1.08, N ⫽ 6)

0.96 (0.18, N ⫽ 6) 0.86 (0.34, N ⫽ 6)

0.67 (0.20, N ⫽ 5) 0.62 (0.17, N ⫽ 6)

0.42 (0.06, N ⫽ 5) 0.42 (0.08, N ⫽ 5)

0.64 (0.18, N ⫽ 5) 0.47 (0.16, N ⫽ 5)

0.26 (0.01, N ⫽ 5) 0.40 (0.12, N ⫽ 6)

0.41 (0.08, N ⫽ 6) 0.34 (0.04, N ⫽ 6)

0.21 (0.02, N ⫽ 5) 0.29 (0.03, N ⫽ 5)

200

Yang and Wilczynski

FIG. 4. Mean levels (⫹SEM) of plasma T (A and B) and DHT (C and D) concentrations in anoles who showed (LC; stipled bar) or did not show (No-LC; white bar) lateral compression are presented. Left panels (A and C) show statistically significant differences between groups during 5 days of encounters (indicated by an asterisk at ␣ ⬍ 0.05), while right panels (B and D) show very low and not different levels of plasma androgen between groups on the test day in an encounter with an intruder in the subjects’ home cages.

perspective on the dynamic interrelationship between androgens and aggressive behavior in this experimental context. To our knowledge this study is the first to test different a priori models hypothesizing different interrelations among social stimulus, behavior, and androgen concentrations. Among the three a priori models, our SEM results supported Model 2, which posits that aggressive behavior (a latent factor accounted for by PU and LC displays) receives a direct positive effect from social stimulus, and plasma T levels receive an indirect positive effect from social stimulus via the animal’s own aggressive behavior. In other words, our results suggest that, in a dynamic male–male encounter, the presentation of another male conspecific elicits aggressive behavior from a territorial owner, which in turn leads to elevation of the owner’s plasma androgen. Furthermore, our SEM results showed that the simultaneous influences of social stimulus on both aggressive behavior and

plasma T (Model 1), or an indirect influence of social stimulus on aggressive behavior via plasma T (Model 3), are unlikely. In order to find a model that fits best to our data, we tested a fourth model, which combined all the paths from Models 1 and 2 (i.e., direct effects of social stimulus on both behavior and plasma T and an indirect effect of social stimulus on plasma T via aggressive behavior). The fit was worse than that of Model 2. Adding a direct path from social stimulus to T appeared to be responsible for this result, indicated by a statistically not significant addition of a path to the model (Wald test: ␹ 12 ⫽ 0.76, P ⬍ 0.39) while increasing the number of parameters to be estimated. Taken together, our results showed that Model 2, rather than a more exhaustive and complicated model, fits our data better, suggesting that during the 5-day period, presentation of a male conspecific to a subject animal alone may not directly increase the subject animal’s plasma androgen, but rather that in-

201

Hormonal Responses to Aggressive Behavior

creased levels of androgen appear to be a direct result of aggression displayed by the subject animals. Our results are consistent with previous reports that aggression, particularly successful aggression, results in an increase in plasma T (Sapolsky, 1986). For example, an aggressive encounter early in the breeding season increased circulating testosterone (Harding, 1981) and gonadotropin levels in song sparrows (Wingfield, 1985). Early in the breeding season, male birds have high levels of circulating T when they establish new territories through aggressive encounters (Wingfield, 1985). Territorial sparrows showed elevations of circulating T concentrations in response to a simulated intrusion (Wingfield, 1984), and flycatchers attacking the decoy used for the intrusion had high T levels (Silverin, 1993). Furthermore, dominant Anolis lizards had plasma T levels fivefold greater than subordinates after winning a male–male aggressive encounter (Greenberg and Crews, 1990). The fact that Models 1 and 3 were not supported does not invalidate the previously established androgen-dependency of challenge behavior in various species. The previous studies that demonstrated a causal link between plasma androgen and challenge behavior consist of either comparisons of the association between androgen and aggression during the breeding season with that during nonbreeding seasons or experimental effects of the presence/absence of androgen on aggression. For example, aggressive behavior in seasonal breeders covaries with seasonal fluctuation of plasma androgen level (for a review, see Nelson, 1995). Experimental studies with castration or castration plus treatment with exogenous T showed that castration decreases aggressive behavior, but does not eliminate it, and T replacement elevated aggressive display behaviors in Anolis lizards (Adkins and Schlesinger, 1979; Crews, 1974, 1978; Crews et al., 1978; Mason and Adkins, 1976). These data address the causal relationship between androgen and aggressive behavior to some extent, but they do not deal with dynamic relationships between social stimulus, androgen, and aggressive behavior in terms of mechanisms of behavioral expression as agonistic encounters proceed. We suggest that, in intact Anolis lizards, a baseline level of plasma T is necessary to show aggressive behavior. Once T levels pass this threshold (e.g., in an intact breeding animal), aggressive behavior is elicited primarily by social cues (e.g., the appearance of a strange conspecific male in its territory), and, subsequently, an animal’s own aggressive behavior causes plasma T levels to increase.

Comparisons between Group Means Our behavioral results were consistent with previous findings with animals that were exposed to a video featuring an aggressive conspecific male (Yang et al., 2001). Animals that interacted with a video showing an aggressive male showed aggressive behaviors in a pattern similar to that shown by animals that encountered live aggressive conspecifics in this experiment, with the peak response levels on Day 3. An interesting observation in the current study was that more LC, but not more DL or PU, toward the intruder in their home cage was displayed by the animals with 5 days of aggressive encounters as compared to control animals. This might be related to the hierarchical organization between DL, PU, and LC (Crews, 1975; Greenberg and Crews, 1983, 1990). In general, social displays are shown in a ritualized and predictable sequence in this species. Although all three displays are observed in male–male interactions, DL and PU occur across various contexts including patrolling territory and engaging in agonistic encounters (Jenssen, Greenberg, and Hovde, 1995), whereas LC occurs exclusively during agonistic encounters with other conspecifics (Crews, 1975; Greenberg and Crews, 1983). LC display consists of expanded throat, formation of a crest on the back and neck, and extreme compression of the body laterally, which requires a great deal of energy and reflects high levels of aggression, and LC is associated with escalation to physical fighting (Greenberg and Crews, 1983; Tokarz, 1985). Therefore, our results that animals with 5 days of aggressive encounters showed more LC than control animals suggest that this experience increased animals’ propensity to engage in highly costly behavior. Our results show that plasma levels of androgen follow the pattern of aggressive responses during the 5-day aggressive encounters; androgens peaked on Day 3 and decreased thereafter. However, increased aggression to an intruder at Test was not accompanied by high androgen levels. This suggests that the relationship between androgen levels and aggressive behavior is not invariable and that 5 days of aggressive encounters resulted in an increase in aggression in response to a subsequent social challenge, but not via increases in androgen levels. This result further supports our SEM results in that increases in plasma T levels do not necessarily trigger aggressive behavior. This is consistent with findings in a series of field studies demonstrating that mountain spiny lizards (Sceloporus jarrovi) showed T-independent aspects of

202

aggressive behavior, which were more pronounced during the nonbreeding season (Moore, 1988; Moore and Marler, 1987). Taken together, our results from both means analysis and SEM analysis suggest that aggressive behavior can occur independent of high levels of plasma T and can trigger increases in plasma T levels. Unlike in other androgen-dependent behaviors, the context in which an aggressive encounter occurs is known to play an important role in the relationship between androgens and aggressive displays. Interacting with other males in an adjacent cage during the 5-day period in our study is a different circumstance from defending its “territory” when we placed a stimulus male into the subject’s home cage. The behavioral patterns in these two situations are different: animals that were confined in the same cage displayed and quickly established a social hierarchy (Summers and Greenberg, 1994; Tokarz, 1985), whereas animals that were in the adjacent cages would display but not necessarily adopt the dominance/subordination relationship (personal observation). The significance of context or environmental condition in which an aggressive encounter occurs has been reported previously in various species, including Anolis lizards (e.g., Crews, 1980). Similarly, castrated rats that remained in their home cages displayed high levels of aggressive behaviors for more than 2 weeks (Schuurman, 1980). In addition, in some species of birds and lizards, T is associated with territorial aggression during the breeding season (DeNardo and Licht, 1993; Wingfield et al., 1987), but this positive association between T and territorial aggression diminished during the nonbreeding season (Moore, 1987b; Moore and Marler, 1987; Wingfield, 1994). Furthermore, social environment, often characterized by the degree of stability of the social group dynamics, has been demonstrated to be important. Birds (Wingfield et al., 1987) and monkeys (Brockman, Whitten, Richard, and Benander, 2001; Rose et al., 1975; Sapolsky, 1983a) in unstable social relationships or in the early stages of the establishment of social hierarchies showed closer association between aggression and T levels than those in well-established social relationships (Dufty and Wingfield, 1986; Hegner and Wingfield, 1986; Rohwer and Rohwer, 1978). The 10 min of aggressive encounters after which we measured plasma T appears sufficient to reveal a dynamic relationship between plasma androgens and aggressive behavior in our study. Transient increases in plasma T in response to ongoing aggressive inter-

Yang and Wilczynski

actions lasted for 10 –19 min in the field in blackbirds and song sparrows (Harding and Follett, 1979; Wingfield, 1994). In addition, Ramenofsky (1985) has reported that elevated circulating androgens in winners are evident up to 32 min after the onset of an aggressive encounter in Japanese quail. The time course of changes in plasma androgens in response to an agonistic encounter remains unknown in this species. In this study, we did not measure minute-by-minute changes in plasma T in response to exposure to another conspecific, preceding the subject animal’s aggressive behavior, and this may limit the interpretation of our results. It is possible that changes in plasma androgen that occur rapidly differ from the level of androgen at 10 min after the onset of the aggressive encounter. To resolve this issue, it would be necessary to do further systematic experiments manipulating and measuring one component at a time at various short-term time points in minutes, as an agonistic encounter between animals unfolds. Although it still remains unclear what aspect of, and by what mechanism, an animal’s aggressive behavior influences changes in levels in its own plasma androgen, illuminating precedents exist in the context of reproductive behavior in birds (Brockway, 1967; Cheng, 1986) and frogs (Marler and Ryan, 1996; Mendonc¸ a, Licht, Ryan, and Barnes, 1985). For example, in birds and frogs where vocalizing during the breeding season is an important courtship behavior, experimentally devocalized male budgerigars showed reduced testis size (Brockway, 1967) and male frogs producing mating calls in the field showed higher androgen levels than those producing no calls (Marler and Ryan, 1996). In addition, a female’s own courtship vocalization (nest-coo) stimulates her follicle growth in the ring dove (Cheng, 1986) by acting on the direct connections between auditory nuclei and hypothalamic nuclei related to the production and release of GnRH (Cheng, Peng, and Johnson, 1998). A parallel mechanism is possible for the influence of aggressive behavior on plasma androgen in Anolis lizards. The brain areas reported to be involved in aggressive behavior, such as the preoptic area, the basal hypothalamus (Farragher and Crews, 1979; Wheeler and Crews, 1978), and the paleostriatum (in part the mammalian basal ganglia homologue, Greenberg, Scott, and Crews, 1984b), are also known to interact with androgens (Morrell, Crews, Ballin, Morgentalin, and Pfaff, 1979; Rosen, O’Bryant, Matthews, Zacharewski, and Wade, 2002). In summary, using a SEM analysis technique, our

Hormonal Responses to Aggressive Behavior

study successfully tested models of the array of interacting influences that occur during aggressive encounters. When animals were interacting with other males in an adjacent cage, dynamic interactions among social cues, behavior, and hormones were best explained by a model hypothesizing that the social stimulus triggers an animal’s aggressive behavior, which in turn elevates levels of plasma androgen.

ACKNOWLEDGMENTS We thank Jessica Berndt for assistance in collecting data and Dr. David Crews for help with hormone assays. We also thank Jon Sakata for helpful comments on the manuscript and JongHan Kim for help with SEM analyses. This study was supported by Research Grants-in-Aid from the Society for Integrative and Comparative Biology to E.-J.Y. and NSF IBN 0090739 to W.W. The experimental protocol was approved on July 3, 1998 by the University of Texas at Austin (Animal Research Protocol 98070301).

REFERENCES Adkins, E., and Schlesinger, L. (1979). Androgens and the social behavior of male and female lizards (Anolis carolinensis). Horm. Behav. 13, 139 –152. Aspinwall, L. G., and Taylor, S. E. (1992). Modeling cognitive adaptation: A longitudianl investigation of the impact of individual differences and coping on college adjustment and performance. J. Personal. Soc. Psychol. 63, 989 –1003. Baxter, L. R. (2001). Brain mediation of anolis social dominance displays. III. Differential forebrain 3H-sumatriptan binding in dominant and submissive males. Brain Behav. Evol. 57, 202–213. Baxter, L. R., Ackermann, R. F., Clark, E. C., and Baxter, J. E. G. (2001a). Brain mediation of Anolis social dominance displays. I. Differential basal ganglia activation. Brain Behav. Evol. 57, 169 –183. Baxter, L. R., Clark, E. C., Ackermann, R. F., Lacan, G., and Baxter, J. E. G. (2001b). Brain mediation of Anolis social dominance displays. II. Differential forebrain serotonin turnover and effects of specific 5-HT receptor agonists. Brain Behav. Evol. 57, 184 –202. Beletsky, L. D., Orians, G. H., and Wingfield, J. C. (1990). Effects of exogeneous androgen and antiandrogen on territorial and nonterritorial red-winged blackbirds (Aves: Icterinae). Ethology 85, 58 –72. Bentler, P. M. (1989). EQS Structural Equations Program Manual. BMDP Statistical Software, Los Angeles, CA. Brockman, D. K., Whitten, P. L., Richard, A. F., and Benander, B. (2001). Birth season testosterone levels in male Verreaux’s sifaka, Propithecus verreauxi: Insights into socio-demographic factors mediating seasonal testicular function. Behav. Ecol. Sociobiol. 49, 117–127. Brockway, B. F. (1967). The influence of vocal behavior on the performer’s testicular activity in budgerigars (Melapsittacus undulatus). Wilson Bull. 79, 328 –334. Burmeister, S., and Wilczynski, W. (2000). Social signals influence hormones independently of calling behavior in the treefrog (Hyla cinerea). Horm. Behav. 38, 201–209.

203

Byrne, B. M. (1994). Structural Equation Modeling with EQS and EQS/Windows. Sage, Thousand Oaks, CA. Cheng, M.-F. (1986). Female cooing promotes ovarian development in ring doves. Physiol. Behav. 37, 371–374. Cheng, M.-F., Peng, J. P., and Johnson, P. (1998). Hypothalamic neurons pereferentially respond to female nest coo stimulation: Demonstration of direct acoustic stimulation of luteinizing hormone release. J. Neurosci. 18(14), 5477–5489. Clark, C. R., and Nowell, N. W. (1980). The effect of the nonsteroidal antiandrogen flutamide on neural receptor binding of testosterone and intermale aggressive behavior in mice. Psychoneuroendocrinology 5, 39 – 45. Crews, D. (1974). Castration and androgen replacement on male facilitation of ovarian activity in the lizard, Anolis carolinensis. J. Comp. Physiol. Psychol. 87, 963–969. Crews, D. (1975). Inter- and intraindividual variation in display patterns in the lizard, Anolis carolinensis. Herpetology 31, 37– 47. Crews, D. (1978). Endocrine control of reptilian reproductive behavior. In C. Beyer (Ed.), Endocrine Control of Sexual Behavior, pp. 167–222. Raven Press, New York. Crews, D. (1980). Interrelationships among ecological, behavioral, and neuroendocrine processes in the reproductive cycle of Anolis carolinensis and other reptiles. In J. S. Rosenblatt, R. A. Hinde, C. G. Beer, and M.-C. Busnel (Eds.), Advances in the Study of Behavior, Vol. 11, pp. 1–73. Academic Press, New York. Crews, D., Traina, V., Wetzel, F. T., and Muller, C. (1978). Hormonal control of male reproductive behavior in the lizard, Anolis carolinensis: Role of testosterone, dihydrotestosterone, and estradiol. Endocrinology 103, 1814 –1821. Crowley, S. L., and Fan, X. (1997). Structural equation modeling: Basic concepts and applications in personality assessment research. J. Personal. Assess. 68, 508 –531. DeCourcy, K. R., and Jenssen, T. A. (1994). Structure and use of male territorial headbob signals by the lizard Anolis carolinensis. Anim. Behav. 47, 251–262. DeNardo, D. F., and Licht, P. (1993). Effects of corticosterone on social behavior of male lizards. Horm. Behav. 27, 184 –199. DeNardo, D. F., and Sinervo, B. (1994). Effects of steroid hormone interaction on activity and home-range size of male lizards. Horm. Behav. 28, 273–287. Dufty, A. M., and Wingfield, J. C. (1986). Endocrine changes in breeding brown-headed cowbirds and their implications for the evolution of brood parasitism. In Y. Que´ innec and N. Delvolve´ (Eds.), Behavioural Rhythms, pp. 93–108. Universite´ Paul Sabatier, Toulouse. Farragher, K., and Crews, D. (1979). Role of basal hypothalamus in the regulation of male reproductive behavior in the lizard, Anolis carolinensis: Lesion studies. Horm. Behav. 13, 185–206. Foran, C. M., and Bass, A. H. (1999). Preoptic GnRH and AVT: Axes for sexual plasticity in teleost fish. Gen. Comp. Endocrinol. 116, 141–152. Greenberg, B., and Noble, G. K. (1944). Social behavior of the American chameleon (Anolis carolinensis Voigt). Physiol. Zool. 17, 392– 439. Greenberg, N. (1977). A neuroethological study of display behavior in the lizard Anolis carolinensis (Reptilia, Lacertilia, Iguanidae). Am. Zool. 17, 191–201. Greenberg, N., Chen, T., and Crews (1984a). Social status, gonadal state, and the adrenal stress response in the lizard, Anolis carolinensis. Horm. Behav. 18, 1–11. Greenberg, N., and Crews, D. (1983). Physiological ethology of

204

aggression in amphibians and reptiles. In S. Bruce (Ed.), Hormones and Aggressive Behavior, pp. 469 –506. Plenum, New York. Greenberg, N., and Crews, D. (1990). Endocrine and behavioral responses to aggression and social dominance in the green anole lizards, Anolis carolinensis. Gen. Comp. Endocrinol. 77, 246 –255. Greenberg, N., Scott, M., and Crews, D. (1984b). Role of the amygdala in the reproductive and aggressive behavior of the lizard, Anolis carolinensis. Physiol. Behav. 32, 147–151. Harding, C. F. (1981). Social modulation of circulating hormone levels in the male. Am. Zool. 21, 223–231. Harding, C. F., and Follett, B. K. (1979). Hormone changes triggered by aggression in a natural population of blackbirds. Science 203, 918 –920. Hegner, R. E., and Wingfield, J. C. (1986). Behavioral and endocrine correlates of multiple brooding in the semi-colonial house sparrow, Passer domesticus. Horm. Behav. 20, 294 –312. Jenssen, T. A., Greenberg, N., and Hovde, K. A. (1995). Behavioural profile of free-ranging male lizards, Anolis carolinensis, across breeding and post-breeding seasons. Herp. Monogr. 9, 41– 62. Jenssen, T. A., Lovern, M. B., and Congdon, J. D. (2001). Field-testing the protandry-based mating system for the lizard, Anolis carolinensis: Does the model organism have the right model? Behav. Ecol. Sociobiol. 50, 162–172. Klukowski, M., and Nelson, C. E. (1998). The challenge hypothesis and seasonal changes in aggression and steroids in male northern fence lizards (Sceloporus undulatus hyacinthinus). Horm. Behav. 33, 197–204. Knapp, R., and Moore, M. C. (1995). Hormonal responses to aggression vary in different types of agonistic encounters in male tree lizards, Urosaurus ornatus. Horm. Behav. 29, 85–105. Korzan, W. J., Summers, T. R., Ronan, P. J., Renner, K. J., and Summers, C. H. (2001). The role of monoaminergic nuclei during aggression and sympathetic social signaling. Brain Behav. Evol. 57, 317–327. Loehlin, J. C. (1998). Latent Variables Models: An Introduction to Factor, Path, and Structural Analysis, 3rd ed. Erlbaum, Hillsdale, NJ. Lovern, M. B., McNabb, F. M. A., and Jenssen, T. A. (2001). Developmental effects of testosterone on behavior in male and female green anoles (Anolis carolinensis). Horm. Behav. 39, 131–143. Marler, C. A., and Ryan, M. J. (1996). Energetic constraints and steroid hormone correlates of male calling behaviour in the tu´ ngara frog. J. Zool. 240, 397– 409. Mason, P., and Adkins, E. K. (1976). Hormones and social behavior in the lizard, Anolis carolinensis. Horm. Behav. 7, 75– 86. McIntosh, A. R., and Gonzalez-Lima, F. (1992). The application of structural modeling to metabolic mapping of functional neural systems. In F. Gonzalez-Lima (Ed.), Advances in Metabolic Mapping Techniques for Brain Imaging of Behavioral and Learning Functions, pp. 219 –255. Kluwer Academic, Dordrecht. Mendonc¸ a, M. T., Licht, P., Ryan, M. J., and Barnes, R. (1985). Changes in hormone levels in relation to breeding behavior in male bullfrogs (Rana catesbeiana) at the individual and population levels. Gen. Comp. Endocrinol. 58, 270 –279. Moore, M. C. (1987a). Castration affects territorial and sexual behaviour of free-living male lizards, Sceloporus jarrovi. Anim. Behav. 35, 1193–1199. Moore, M. C. (1987b). Circulating steroid hormones during rapid aggressive responses of territorial male mountain spiny lizards, Sceloporus jarrovi. Horm. Behav. 21, 511–521. Moore, M. C. (1988). Testosterone control of territorial behavior: Tonic-release implants fully restore seasonal and short-term ag-

Yang and Wilczynski

gressive responses in free-living castrated lizards. Gen. Comp. Endocrinol. 70, 450 – 459. Moore, M. C., and Marler, C. A. (1987). Effects of testosterone manipulations on nonbreeding season territorial aggression in free-living male lizards, Sceloporus jarrovi. Gen. Comp. Endocrinol. 65, 225–232. Morrell, J. L., Crews, D., Ballin, A., Morgentalin, A., and Pfaff, D. W. (1979). 3H-Estradiol, 3H-testosterone, and 3H-dihydrotestosterone localization in the brain of the lizard, Anolis carolinensis. J. Comp. Neurol. 188, 201–224. Nelson, R. J. (1995). In An Introduction to Behavioral Endocrinology. Sinauer, Sunderland, MA. Ramenofsky, M. (1985). Acute changes in plasma steroids and agonistic behavior in male Japanese quail. Gen. Comp. Endocrinol. 60, 116 –128. Rhen, T., Ross, J., and Crews, D. (1999). Effects of testosterone on sexual behavior and morphology in adult female leopard geckos, Eublepharis macularius. Horm. Behav. 36, 119 –128. Rohwer, S., and Rohwer, F. C. (1978). Status signalling in Harris’s sparrows: Experimental deceptions achieved. Anim. Behav. 26, 1012–1022. Rose, R. M., Bernstein, I. S., Gordon, T. P., and Catlin, S. F. (1975). Consequences of social conflict on plasma testosterone levels in rhesus monkeys. Psychosomatic Med. 37, 50 – 61. Rosen, G., O’Bryant, E., Matthews, J., Zacharewski, T., and Wade, J. (2002). Distribution of androgen receptor mRNA expression and immunoreactivity in the brain of the green anole lizard. J. Neuroendocrinol. 14, 19 –28. Sapolsky, R. (1983a). Endocrine aspects of social instability in the olive baboon (Papio anubis). Am. J. Primatol. 5, 365. Sapolsky, R. (1983b). Individual differences in cortisol secretory patterns in the wild baboon: Role of negative feedback sensitivity. Endocrinology 113, 2263. Sapolsky, R. M. (1986). Stress-induced elevation in testosterone concentrations in high ranking baboons: Role of catecholamines. Endocrinology 118, 1630 –1635. Schuurman, T. (1980). Hormonal correlates of agonistic behavior in adult male rats. Prog. Brain Res. 53, 415– 420. Schwabl, H., and Kriner, E. (1991). Territorial aggression and song of male European robins (Erithacus rubecula) in autumn and spring: Effects of antiandrogen treatment. Horm. Behav. 25, 180 – 194. Silverin, B. (1993). Territorial aggressiveness and its relation to the endocrine system in the pied flycatcher. Gen. Comp. Endocrinol. 89, 206 –213. Soma, K. K., Sullivan, K., and Wingfield, J. (1999). Combined aromatase inhibitor and antiandrogen treatment decreases territorial aggression in a wild songbird during the nonbreeding season. Gen. Comp. Endocrinol. 115, 442– 453. Summers, C. H. (2001). Mechanisms for quick and variable responses. Brain Behav. Evol. 57, 283–292. Summers, C. H., and Greenberg, N. (1994). Somatic correlates of adrenergic activity during aggression in the lizard, Anolis carolinensis. Horm. Behav. 28, 29 – 40. Summers, C. H., and Greenberg, N. (1995). Activation of central biogenic amines following aggressive interaction in male lizards, Anolis carolinensis. Brain Behav. Evol. 45, 339 –349. Taylor, G. T., Haller, J., Rupich, R., and Weiss, J. (1984). Testicular hormones and intermale aggressive behaviour in the presence of a female rat. J. Endocrinol. 100, 315–321.

Hormonal Responses to Aggressive Behavior

Tokarz, R. R. (1985). Body size as a fctor determining dominance in staged agonistic encounters between male brown anoles (Anolis sagrei). Anim. Behav. 33, 746 –753. Tokarz, R. R. (1986). Hormonal regulation of male reproductive behavior in the lizard Anolis sagrei: A test of the aromatization hypothesis. Horm. Behav. 20, 364 –377. Tokarz, R. R. (1987). Effects of the antiandrogens cyproterone acetate and flutamide on male reproductive behavior in a lizard (Anolis sagrei). Horm. Behav. 21, 1–16. Tokarz, R. R. (1995). Importance of androgens in male terriorial acquisition in the lizard Anolis sagrei: An experimental test. Anim. Behav. 49, 661– 669. Tousignant, A., and Crews, D. (1995). Incubation temperature and gonadal sex affect growth and physiology in the leopard gecko (Eublepharis macularius), a lizard with temperature-dependent sex determination. J. Morphol. 224, 159 –170. Wheeler, J. M., and Crews, D. (1978). The role of the anterior hypothalamus–preoptic area in the regulation of male reproductive behavior in the lizard, Anolis carolinensis: Lesion studies. Horm. Behav. 11, 42– 60.

205

White, S. A., and Fernald, R. D. (1997). Changing through doing: Behavioral influences on the brain. Rec. Prog. Horm. Res. 52, 455– 474. Wingfield, J. C. (1984). Environmental and endocrine control of reproduction in the song sparrow, Melospiza melodia. II. Agonistic interactions as environmental information stimulating secretion of testosterone. Gen. Comp. Endocrinol. 56, 417– 424. Wingfield, J. C. (1985). Short-term changes in plasma levels of hormones during establishment and defense of a breeding territory in male song sparrows, Melospiza melodia. Horm. Behav. 19, 174 –187. Wingfield, J. C. (1994). Regulation of territorial behavior in the sedentary song sparrow, Melospiza melodia morphna. Horm. Behav. 28, 1–15. Wingfield, J. C., Ball, G. F., Dufty, A. M. J. Hegner, R. E., and Ramenofsky, M. (1987). Testosterone and aggression in birds. Am. Sci. 75, 602– 608. Yang, E.-J., Phelps, S. M., Crews, D., and Wilczynski, W. (2001). The effects of social experience on aggressive behavior in the green anole lizard (Anolis carolinensis). Ethology 107, 777–793.