The medial bed nucleus of the stria terminalis (BSTm) is

Valence-sensitive neurons exhibit divergent functional profiles in gregarious and asocial species James L. Goodson* and Yiwei Wang Department of Psychology, University of California at San Diego, La Jolla, CA 92093-0109 Edited by Bruce S. McEwen, The Rockefeller University, New York, NY, and approved September 21, 2006 (received for review July 24, 2006) The medial bed nucleus of the stria terminalis (BSTm) influences both social approach and social aversion, suggesting that this structure may play an important role in generating motivational and behavioral differences between gregarious and asocial species. However, no specific neurons have been identified within the BSTm that influence species-typical levels of sociality or that mediate approach and avoidance. Using five songbird species that differ selectively in their species-typical group sizes, we now demonstrate that vasotocin-immunoreactive (VT-ir) neurons of the BSTm exhibit very different immediate early gene responses to same-sex stimuli in gregarious and asocial species. Exposure to a same-sex conspecific increases VT-Fos colocalization in gregarious species while decreasing colocalization in relatively asocial species. We additionally demonstrate that these neurons are selectively activated by social stimuli that normally elicit affiliation (positively valenced social stimuli) but not by stimuli that elicit aversion (negatively valenced social stimuli). Constitutive Fos activity of the VT-ir neurons is also significantly greater in the gregarious species, and the two most social species express significantly more VT-ir neurons. These findings demonstrate that the properties of valence-sensitive neurons evolve in relation to sociality and indicate that gregarious species accentuate positive stimulus properties during social interactions. amygdala sociality vasopressin vasotocin The medial bed nucleus of the stria terminalis (BSTm) is anatomically and functionally conserved in the land vertebrate classes (1 3), and in virtually all species, a subpopulation of BSTm neurons produce vasotocin (VT) or its mammalian homologue vasopressin (VP) (4, 5) (see Fig. 1). These neurons project to multiple areas of the basal forebrain [e.g., ventral pallidum and lateral septum (5, 6)], where VT VP influences a variety of social behaviors, sometimes in a potent and highly species-specific manner (4, 7, 8). For instance, differential expression of V 1a receptors in the ventral pallidum is both necessary and sufficient to account for species differences in monogamous pair-bonding (9), and V 1a receptors in the lateral septum are both necessary and sufficient for the display of social recognition in mice (10). However, despite the vast literature on VT VP functions, receptor distributions, sexual dimorphisms, and steroid sensitivities (4, 5, 7, 8), almost nothing is known about BSTm VT VP neuronal activity as it relates to stimulus parameters, social context, and species-specific aspects of social structure. Indeed, to our knowledge, the only relevant data come from monogamous male voles, in which VP mrna increases in the BSTm after cohabitation with a female (11). The present experiments were therefore conducted to test the hypotheses that (i) VT-immunoreactive (VT-ir) neurons in the BSTm are responsive to stimuli that elicit affiliation, and (ii) responses of these neurons evolve in relation to sociality. These hypotheses were tested by immunofluorescently double-labeling for VT and the immediate early gene protein Fos (an inducible marker of neural activity) after a variety of social interactions in males and females of five estrildid finch species that differ selectively in their species-typical group sizes (12, 13). These include two species that independently evolved a relatively asocial, territorial social structure (the melba finch, Pytilia melba, and the violet-eared waxbill, Uraeginthus granatina); two species that independently evolved a highly colonial social structure (the spice finch, Lonchura punctulata, and the zebra finch, Taeniopygia guttata); and an intermediately gregarious species (the Angolan blue waxbill, Uraeginthus angolensis, which is a congener of the territorial violet-eared waxbill). The mean group size of the territorial species is two (i.e., male female pair). Group sizes for the Angolan blue waxbill range from 8 40 and are 100 for the two highly gregarious, colonial species. These species are closely matched in other aspects of behavior and ecology: All are monogamous (typically pair-bonding for life), exhibit biparental care, breed semiopportunistically dependent on rainfall, and inhabit semiarid and or grassland scrub habitat. The species differences in sociality are stable across seasons and contexts (for evolutionary considerations, see ref. 13; also see refs. 14 16). In the present experiments, subjects were separated from stimulus animals by a wire barrier (this was modified in the final experiment), and tests were conducted outside of the normal housing environment. This paradigm successfully limits the performance of overt social behaviors, even in territorial subjects (12). Thus, Fos induction that is observed in this paradigm should primarily reflect motivational and or perceptual processes, not motoric responses to the stimulus. Results In our first experiment, males and females of the highly gregarious zebra finch were exposed to a same-sex conspecific through the wire barrier or to a control condition (wire barrier only). This simple manipulation produced a large increase in the percentage of VT-ir neurons that coexpressed Fos-ir nuclei [Fig. 2; two-way ANOVA, sex condition: F(6,1,1) 23.216; P 0.002]. No sex differences were observed [F(6,1,1) 1.701; P 0.234]. We subsequently repeated this experiment in the other four other species. Although no female melba finches could be committed to this study (collections in Africa and subsequent breeding have yielded few females), two-way ANOVAs (condition sex) revealed no main effect of sex for the violet-eared waxbill [F(6,1,1) 0.317; P 0.594], Angolan blue waxbill [F(10,1,1) 0.292; P 0.601], or spice finch [F(10,1,1) 0.070; P 0.797]. A three-way ANOVA including all three of these species likewise showed a clear lack of sex differences [F(25,1,1,2) 0.795; P 0.381]. We therefore pooled data from males and females for a two-way ANOVA (species condition) that included the melba finch (Fig. 3A). These analyses revealed a significant main effect of species [F(36,1,3) 4.356; P 0.010] and a significant Author contributions: J.L.G. designed research; J.L.G. and Y.W. performed research; J.L.G. and Y.W. analyzed data; and J.L.G. wrote the paper. The authors declare no conflict of interest. This article is a PNAS direct submission. Abbreviations: BSTm, medial bed nucleus of the stria terminalis; VP, vasopressin; VT, vasotocin; VT-ir, VT-immunoreactive. *To whom correspondence should be addressed. E-mail: jgoodson@ucsd.edu. 2006 by The National Academy of Sciences of the USA www.pnas.org cgi doi 10.1073 pnas.0606278103 PNAS November 7, 2006 vol. 103 no. 45 17013 17017

Fig. 2. Responses of VT neurons to a same-sex stimulus in the highly gregarious zebra finch. Percentage of VT-ir neurons in the BSTm that express Fos-ir nuclei (means SEM) after exposure to a same-sex conspecific or a control condition (males, solid bars; females, open bars). Two-way ANOVA (sex condition) shows a main effect of condition (P 0.002). Total n 10. Fig. 1. Photomicrographs of VT-ir neurons in the BSTm of a male zebra finch at the level of the anterior commissure (A) and immediately caudal to the anterior commissure (B) (color inverted fluorescence). Asterisks indicate the location of Insets showing neurons double-labeled for VT (green) and Fos (red). (Scale bars: A, 100 m; B, 200 m; Insets, 50 m.) AC, anterior commissure; BSTl, lateral bed nucleus of the stria terminalis; LS, lateral septum; MS, medial septum; PVN, paraventricular nucleus of the hypothalamus; v, ventricle. species condition interaction [F(36,1,3) 3.445; P 0.026], both of which appear to reflect differences between the two territorial species versus the two gregarious species. To directly test this idea, we combined subjects of the two territorial species and subjects of the two gregarious species for an additional two-way ANOVA (sociality condition; Fig. 3B). Consistent with our hypothesis, the territorial subjects exhibited significantly lower levels of VT-Fos colocalization than did the gregarious subjects [main effect of sociality: F(40,1,1) 8.555; P 0.005], and a significant interaction was observed for sociality and condition [F(40,1,1) 6.504; P 0.014]. This interaction reflects the fact that VT-Fos colocalization tended to decrease in the territorial subjects after exposure to a same-sex conspecific, while increasing in the gregarious subjects. Given these species differences in Fos response, a main effect of condition was not observed [F(36,1,3) 0.998; P 0.325]. Notably, the gregarious Angolan blue waxbill exhibits a pattern of colocalization that is similar to the colonial spice finch and strongly dissimilar to its closely related congener, the territorial violet-eared waxbill. A separate analysis of the two waxbill congeners confirms the divergence between them, yielding a significant species condition interaction [F(20,1,1) 6.620; P 0.018]. Overall, the constitutive levels of Fos expression were much higher in this second experiment than in the previous experiment, which may have produced a ceiling effect in the two gregarious species. This difference likely reflects the fact that the subjects used in this second experiment were taken from breeding cages (2 days before testing), whereas zebra finches in the first experiment had been in same-sex housing. Data presented below from breeding-condition zebra finches are consistent with this hypothesis, because these birds likewise show high constitutive coexpression of VT and Fos. Based on the finding that VT-Fos colocalization in the BSTm increases in response to same-sex stimuli in gregarious species but not in relatively asocial species, we hypothesized that these neurons may respond selectively to social stimuli that promote affiliation (positively valenced social stimuli) but not to stimuli that produce aversion or avoidance (negatively valenced social stimuli). To provide additional support for this hypothesis, we Fig. 3. Responses of VT-ir neurons to a conspecific stimulus differ between asocial and gregarious species. (A) Percentage of VT-ir neurons in the BSTm that express Fos-ir nuclei (means SEM) after exposure to a same-sex conspecific (filled bars) or a control condition (open bars) in two relatively asocial, territorial species [melba finch (MF) and violet-eared waxbill (VEW)], a moderately gregarious species [Angolan blue waxbill (ABW)], and a highly gregarious, colonial species [spice finch (SF)]. No sex differences were observed, and sexes are shown pooled. Two-way ANOVA (species condition) yields a main effect of species (P 0.01) and a species condition interaction (P 0.02). Different letters above the bars indicate significant species differences (P 0.05, Fisher s protected least squares difference). Two-way ANOVA (sociality condition) after pooling of the relatively asocial subjects (MF and VEW) and the gregarious subjects (ABW and SF) shows a main effect of sociality (P 0.005) and a sociality condition interaction (P 0.01). Total n 44. (B) Representative double-labeling for Fos (red) and VT (green) in an experimental male spice finch. DAPI nuclear stain is shown in blue. (Scale bar: 50 m.) 17014 www.pnas.org cgi doi 10.1073 pnas.0606278103 Goodson and Wang

Fig. 5. Highly colonial species express more VT-ir neurons than do less gregarious species. Numbers of VT-ir neurons per 40- m section (means SEM) in the BSTm of two relatively asocial, territorial species [melba finch (MF) and violet-eared waxbill (VEW)], a moderately gregarious species [Angolan blue waxbill (ABW)], and two highly gregarious, colonial species [spice finch (SF) and zebra finch (ZF)]. No sex differences were observed, and sexes are shown pooled. Different letters above the bars indicate significant species differences (P 0.05, Fisher s protected least squares difference) after significant ANOVA (P 0.0001). Total n 96. Fig. 4. Fos expression in VT-ir neurons increases in response to positively valenced social stimuli but not negatively valenced stimuli. (A) Percentage of VT-ir neurons in the BSTm that express Fos-ir nuclei (means SEM) in the asocial violet-eared waxbill after exposure to a control condition (open bars), a same-sex conspecific (black bars), or the subject s pairbond partner (gray bars). Different letters above the bars indicate significant group differences (Mann Whitney tied P 0.05, after significant Kruskal Wallis tied P 0.003). Total n 16. (B) VT-Fos colocalization in control (open bars), subjugated (black bars), and nonsubjugated (gray bars) zebra finches exposed to mate competition (Kruskal Wallis tied P 0.03). No sex differences were observed, and sexes are shown pooled. Total n 15. conducted two experiments in which subjects were exposed to positively valenced and negatively valenced social stimuli. The first of these experiments was conducted in the relatively asocial violet-eared waxbill. Violet-eared waxbills do not affiliate with same-sex individuals but maintain close contact with their pairbond partners. Thus, we exposed male and female waxbills to the control condition (wire barrier), a same-sex conspecific (negative valence), or their pairbond partner (positive valence). The species differences obtained above were anticipated based on pilot experiments; thus, subjects in the third group (pairbond partner) were run alongside the control and same-sex subjects from the previous experiment. As shown in Fig. 4A, the results provide strong support for our hypothesis. Relative to control subjects, subjects exposed to the same-sex stimulus showed a significant decrease in VT-Fos colocalization, whereas subjects exposed to their pairbond partners exhibited a very robust increase (Mann Whitney tied P 0.05 after significant Kruskal Wallis tied P 0.003). No sex differences were observed [sex condition ANOVA, main effect of sex: F(10,2,1) 0.031; P 0.969]. In a concurrent experiment, we exposed male and female zebra finches to a control condition or a mate-competition paradigm in which the subject and a same-sex conspecific competed for courtship access to an opposite sex bird. We constructed this paradigm such that subjects were intensely subjugated (negative social valence) or were dominant and allowed to court (positive social valence). As expected, the nonsubjugated subjects showed a significant increase in VT-Fos colocalization, whereas the subjugated birds did not (Mann Whitney tied P 0.05 after significant Kruskal Wallis tied P 0.03; Fig. 4B). This experiment additionally demonstrates that social arousal alone does not induce Fos activity in the VT-ir neurons, because the subjugated animals failed to exhibit an increase in VT-Fos colocalization, despite being aggressively displaced 71 205 times during the 10-min test. Two other observations further strengthen the association between these VT-ir neurons and positive affiliation. First, constitutive VT-Fos colocalization is significantly higher in gregarious versus territorial subjects, as quantified in breedingcondition control subjects of all five species (Figs. 3 and 4; 53.7 5.5% versus 30.7 6.0%, respectively; t 2.268, P 0.031, unpaired t test). A similar difference is observed when comparing the number of VT-ir neurons in gregarious and territorial subjects (gregarious territorial; t 3.751, P 0.0003, unpaired t test), but in this case the moderately gregarious Angolan blue waxbill is most similar to the two territorial species. Thus, the two highly colonial species (spice finch and zebra finch) exhibit significantly more VT-ir neurons than do the other species [one-way ANOVA on species: F(91,4) 45.201; P 0.0001; Fig. 5). These species differences in VT-ir cell numbers cannot be attributed to variation in immunocytochemical runs because brains from all species and conditions were run in parallel. Discussion Valence Sensitivity of BSTm VT Neurons. The combined data from the present experiments demonstrate that Fos expression in BSTm VT-ir neurons is increased after exposure to positively valenced social stimuli (stimuli that normally elicit affiliative behaviors) but not negatively valenced social stimuli (stimuli that elicit aversion or territorial aggression). Thus, in both gregarious and relatively asocial species, significant increases in VT-Fos colocalization are observed after exposure to an opposite-sex stimulus such as a potential mate or the subject s pairbond partner. Increases in VT-Fos colocalization are also observed after exposure to a same-sex conspecific in the gregarious species, which flock year-round, but decreases are observed in the relatively asocial species, which are territorial and do not associate with same-sex conspecifics. These species differences in VT neuronal response cannot be attributed to species differences in social arousal, because intensely arousing subjugation (71 205 aggressive displacements in a 10-min period; Fig. 4B) failed to influence VT-Fos colocalization. We additionally found that constitutive VT-Fos colocalization is higher in the three gregarious species than in the relatively asocial species, and the two colonial species exhibit more VT-ir neurons in the BSTm. These findings suggest that gregarious birds are tonically primed to respond to social stimuli in a positive, affiliative manner. Finally, sensitivity to positive social valence is not a property Goodson and Wang PNAS November 7, 2006 vol. 103 no. 45 17015

of the BSTm as a whole; in fact, BSTm Fos responses to same-sex stimuli are significantly greater in territorial birds than in gregarious birds (12). VT neurons therefore comprise a functionally distinct population within the BSTm. Whether or not these neurons are sensitive to nonsocial stimuli remains to be determined, although available evidence suggests that the avian BSTm responds preferentially to social stimuli (17). Valence Sensitivity in the BSTm and Amygdala: A General Organizational Scheme? The BSTm and medial amygdala are strongly interconnected and share many functional properties (1, 2). It is therefore noteworthy that various regions of the amygdala contain valence-sensitive neurons in mammals. Activity in the human amygdala is associated with stimulus valence (18) and the encoding of positive and negative items (19). Neurophysiological studies in monkeys further demonstrate that positive and negative sensitivity is exhibited by separate, but intercalated, neuronal populations (20, 21), and recent studies in rats and mice have also identified spatially segregated neuronal populations within the central nucleus that are responsive to appetitive or aversive stimuli (22). The present experiments now identify the neurochemical phenotype of valence-sensitive neurons that are activated during social interactions, which should greatly facilitate the analysis of neural circuits that are involved in social approach and avoidance. The importance of the medial amygdala and BSTm in social approach and avoidance is well established (1, 23), and intercalated populations of medial amygdala neurons exhibit Fos responses to reproductive and predator odors (24). Further analyses may show that these intercalated neurons are more broadly sensitive to positive and negative valence. Importantly, although we have thus far identified only BSTm neurons that are sensitive to positive social valence, it seems likely that an intercalated population of BSTm neurons is also sensitive to negative social valence. This hypothesis is suggested by the fact that same-sex stimuli induce greater BSTm Fos responses in the territorial violet-eared waxbill than in three gregarious species (12), a pattern reversed from that of the VT-ir neurons. VT VP Functions and the Evolution of Sociality. The Angolan blue waxbill and violet-eared waxbill are congeners that have evolved divergently from their common ancestor, becoming more social and less social, respectively (13). These species also diverge significantly in their patterns of Fos induction after exposure to a same-sex conspecific. Furthermore, in diverging from each other, their patterns of VT-Fos colocalization converge with other species that have independently evolved either highly social or highly asocial behavior (see ref. 13). Our finding that BSTm neurons are sensitive to positive social valence in finches provides a strong interpretational framework for other recent data on these same five species, which demonstrate that linear 125 I-VP V 1a antagonist binding in the lateral septum is positively related to sociality (13). Septal VT [presumably of BSTm origin (5, 6)] also influences aggression in these species, inhibiting territorial resident intruder aggression while facilitating aggression that is specific to the context of appetitive courtship behavior (2). Both of these effects are consistent with an increase in affiliative motivation [note that distinctly different influences are exerted by anterior hypothalamic VP neurons, which are known to promote offensive aggression in rodents (25)]. Septal V 1a receptors also mediate social recognition (10, 26, 27) and promote a variety of affiliative behaviors in rodents, including active social interaction in rats (26) and pairbonding in monogamous male voles (28). Somewhat counterintuitively, the various influences VT VP on affiliation are coupled with influences on anxiety and stress processes (10, 17, 29), although the functional significance of this connection remains to be elucidated. Influence of Sex and Reproductive Condition. The number of VT VP neurons in the BSTm is sexually dimorphic (males females) and steroid-dependent in most vertebrate species thus far examined (4, 5, 30), including songbirds (31, 32). In contrast, zebra finches have not been consistently found to exhibit sexually dimorphic or testosterone-dependent immunostaining (33, 34), and we here found no evidence for sex differences in any of the estrildid species examined, either in their numbers of VT-ir neurons or Fos responses to social stimuli. We did however observe a substantial difference in constitutive VT-Fos colocalization between zebra finches that had been housed in same-sex cages ( 12% of VT-ir neurons expressed Fos in control subjects) and those that had been housed in mixed-sex breeding cages until 2 days before testing ( 60%; Figs. 1 and 4B, respectively). Although there were methodological differences in those two experiments, socially induced levels of VT-Fos colocalization were high in both, suggesting that the constitutive differences are not methodological artifacts but rather reflect a strong social and or hormonal influence on constitutive Fos activity. Conclusions. The present findings substantially expand on a variety of findings in humans and other animals, which demonstrate that the amygdala contains neuronal populations that are sensitive to stimulus valence (18, 20, 21). To our knowledge, the present findings provide the first evidence that valence sensitivity is exhibited during social interactions and also represent the first case in which the neurochemical phenotype of the relevant neurons has been identified. Neurochemical identification of valence-sensitive neurons is key, because it essential for systemslevel analysis of valence-related behavioral regulation (e.g., approach avoidance gating). Finally, the general relevance of the present findings should be broad, because the VT VP system arising in the BSTm is conserved across the tetrapod classes (2, 4, 5). Even in cases of independent social evolution, the predictive validity of our results should be strong, given that we have elucidated functional properties of VT neurons that show reliable correspondence to behavior during both evolutionary divergence and convergence in species-typical group size. Methods Animals. Melba finches, violet-eared waxbills, and Angolan blue waxbills were wild-caught by mist net in South Africa and maintained in captivity for 3 years or were captive-bred from wild stock. Wild-caught spice finches were provided by a commercial supplier. Zebra finches were of domestic stock [note that extensive ethological examinations have failed to identify any behavioral differences between wild-caught and domestic zebra finches (35)]. Collections and procedures were conducted legally and humanely under applicable federal and state permits and in compliance with institutional and federal guidelines. Behavioral Testing. The protocol for exposure to conspecifics through a wire barrier has been fully described (12) and was used for experiments presented in Figs. 2, 3, and 4A. Two days before testing, subjects were removed from housing cages and placed into testing cages for acclimation. These cages were then placed into a quiet room the day before testing. In the experiment presented in Fig. 4B, we sought to create a negative social stimulus for the zebra finch by experimentally inducing social subjugation. However, intense subjugation is difficult to experimentally manipulate in this species, given their gregarious nature. Aggression is most reliably elicited by using a mate competition paradigm (36): After several days in same-sex housing, two same-sex individuals are exposed to an opposite sex individual (separated from the other two by a wire barrier). This induces aggressive competition, although it is typically mild. To increase the level of aggression, we used well known bullies from our colony as dominant partners. These were one female 17016 www.pnas.org cgi doi 10.1073 pnas.0606278103 Goodson and Wang

and one male that each had a history of excessive aggression in their housing cages. Two daily conditioning trials (10 min) were conducted before final testing and perfusion. To be retained for the experiment, we required that (i) aggressive subjugation be initiated within 1 min of the start of the trial, thereby minimizing the potential for positive interactions with the opposite-sex stimulus; (ii) a minimum of 40 displacements be exhibited in each of the conditioning trials; and (iii) the number of displacements on test day be equal to, or greater than, the average of the conditioning trials. Final tests were 10 min. Immunocytochemistry and Quantification of Labeling. Tissue was cut into three series of 40- m free-floating sections and one series was immunofluorescently double-labeled for Fos protein and VT by using standard methods (12, 17, 37). For the experiment presented in Fig. 2, we used a custom mouse anti-fos antibody (1:100) (12). A rabbit anti-fos antibody (1:1,000) was used for all other experiments (Santa Cruz Biotechnology, Santa Cruz, CA). Both antibodies are directed against portions of the chicken Fos sequence. Anti-VP raised in guinea pig (1:1,000; Bachem, Torrance, CA) was used for all experiments. The VP antibody was labeled with anti-guinea pig biotin (Vector Laboratories, Burlingame, CA) and visualized with streptavidin conjugated to Alexa Fluor 488 (Molecular Probes, Eugene, OR). The Fos antibodies were visualized with secondary antibodies conjugated to Alexa Fluor 594. Double-labeling was quantified bilaterally in sections containing VT-ir neurons in the commissural BSTm (dorsal to the anterior commissure) and postcommissural BSTm, where a band of VT-ir neurons drops ventrally behind the caudal margin of the anterior commissure (Fig. 1). Subjects were excluded from analyses of VT-Fos colocalization if fewer than five VT-ir neurons were observed. Slides were coded for observer-blind analysis. All counts were conducted on a Zeiss Axioscop with a 62 objective and 10 eyepieces. Statistics. Data were analyzed with StatView 5.0 software for the Macintosh, using ANOVA followed by Fisher s protected least squares difference if variances could be normalized (determined by F tests). Data shown in Figs. 2, 3, and 5 were analyzed after log normal transformations to normalize variances. Data shown in Fig. 4 could not be normalized and were analyzed by using Kruskal Wallis tests followed by Mann Whitney U tests. All tests were 1. Newman SW (1999) Ann NY Acad Sci 877:242 257. 2. Goodson JL (2005) Horm Behav 48:11 22. 3. Yamamoto K, Sun Z, Wang HB, Reiner A (2005) Brain Res Bull 66:341 347. 4. Goodson JL, Bass AH (2001) Brain Res Rev 35:246 265. 5. De Vries GJ, Panzica GC (2006) Neuroscience 138:947 955. 6. De Vries GJ, Buijs RM (1983) Brain Res 273:307 317. 7. Keverne EB, Curley JP (2004) Curr Opin Neurobiol 14:777 783. 8. Young LJ, Wang Z (2004) Nat Neurosci 7:1048 1054. 9. Lim MM, Wang Z, Olazabal DE, Ren X, Terwilliger EF, Young LJ (2004) Nature 429:754 757. 10. Bielsky IF, Hu SB, Ren X, Terwilliger EF, Young LJ (2005) Neuron 47:503 513. 11. Wang Z, Smith W, Major DE, De Vries GJ (1994) Brain Res 650:212 218. 12. Goodson JL, Evans AK, Lindberg L, Allen CD (2005) Proc R Soc London Ser B 272:227 235. 13. Goodson JL, Evans AK, Wang Y (2006) Horm Behav 50:223 236. 14. Skead DM (1975) Ostrich Suppl 11:1 55. 15. Goodwin D (1982) Estrildid Finches of the World (Cornell Univ Press, Ithaca, NY). 16. Zann RA (1996) The Zebra Finch: A Synthesis of Field and Laboratory Studies (Oxford Univ Press, Oxford). 17. Goodson JL, Evans AK (2004) Horm Behav 46:371 381. 18. Anders S, Lotze M, Erb M, Grodd W, Birbaumer N (2004) Hum Brain Mapp 23:200 209. 19. Kensinger EA, Schacter DL (2006) J Neurosci 26:2564 2570. two-tailed with an alpha of 0.05. Sex could not be included as a factor in the analyses presented in Fig. 3 and Fig. 5 because no female melba finches were available for this study. We therefore conducted ANOVAs with melba finches removed and further analyzed each species separately; these analyses showed no trends toward sex differences (see Results). Although the data shown in Fig. 3A were analyzed by Kruskal Wallis, two-way ANOVA (sex condition) shows a clear lack of sex differences [F(10,2,1) 0.031, P 0.969]. Due to a small n in Fig. 3B, sex differences were not tested. The two colonial species exhibited large numbers of VT-ir neurons, and none of the 13 spice finches or 25 zebra finches were excluded from analyses of VT-Fos colocalization. In contrast, the other three species exhibited relatively few VT-ir neurons, requiring that 1 of 8 melba finches, 16 of 32 violet-eared waxbills, and 14 of 28 Angolan blue waxbills be excluded (fewer than five VT-ir neurons observed). Final subject n values were as follows. In Fig. 2, zebra finches: two control males, three control females, three experimental males, and two experimental females. In Fig. 3, melba finches: three control males and four experimental males; violet-eared waxbills: three control males, two control females, three experimental males, and two experimental females; Angolan blue waxbills: five control males, three control females, four experimental males, and two experimental females; spice finches: four control males, two control females, five experimental males, and two experimental females. In Fig. 4A, violet-eared waxbills: three control males, two control females, three males with same-sex stimulus, two females with same-sex stimulus, three males with pairbond partner, and three experimental females with pairbond partner. In Fig. 4B, zebra finches: three control males, three control females, two subjugated males, one subjugated females, three nonsubjugated males, and three nonsubjugated females. In Fig. 5, eight male melba finches, 15 male and 17 female violet-eared waxbills, 14 male and 14 female Angolan blue waxbills, nine male and four female spice finches, and six male and nine female zebra finches. Note that Fig. 5 includes data from all subjects, including those that were excluded from analyses of VT-Fos colocalization. We thank Marcy Kingsbury, Ben Mines, and Renier van Heerden for assistance with bird collections in Africa. This work was supported by the National Institute of Mental Health. 20. Nishijo H, Ono T, Nishino H (1988) J Neurosci 8:3570 3583. 21. Paton JJ, Belova MA, Morrison SE, Salzman CD (2006) Nature 439:865 870. 22. Knapska E, Walasek G, Nikolaev E, Neuhausser-Wespy F, Lipp HP, Kaczmarek L, Werka T (2006) Learn Mem 13:192 200. 23. Sheehan TP, Paul M, Amaral E, Numan MJ, Numan M (2001) Neuroscience 106:341 356. 24. Choi GB, Dong HW, Murphy AJ, Valenzuela DM, Yancopoulos GD, Swanson LW, Anderson DJ (2005) Neuron 46:647 660. 25. Delville Y, De Vries GJ, Ferris CF (2000) Brain Behav Evol 55:53 76. 26. Landgraf R, Gerstberger R, Montkowski A, Probst JC, Wotjak CT, Holsboer F, Engelmann M (1995) J Neurosci 15:4250 4258. 27. Landgraf R, Frank E, Aldag JM, Neumann ID, Sharer CA, Ren X, Terwilliger EF, Niwa M, Wigger A, Young LJ (2003) Eur J Neurosci 18:403 411. 28. Liu Y, Curtis JT, Wang Z (2001) Behav Neurosci 115:910 919. 29. Pitkow LJ, Sharer CA, Ren X, Insel TR, Terwilliger EF, Young LJ (2001) J Neurosci 21:7392 7396. 30. Panzica GC, Aste N, Castagna C, Viglietti-Panzica C, Balthazart J (2001) Brain Res Rev 37:178 200. 31. Voorhuis TAM, Kiss JZ, De Kloet ER, De Wied D (1988) Brain Res 442:139 146. 32. Plumari L, Plateroti S, Deviche P, Panzica GC (2004) Brain Res 999:1 8. 33. Voorhuis TAM, De Kloet ER (1992) Develop Brain Res 69:1 10. 34. Kimura T, Okanoya K, Wada M (1999) Life Sci 65:1663 1670. 35. Morris D (1958) Proc Zool Sci London 131:389 439. 36. Adkins-Regan E, Robinson TM (1993) J Comp Psychol 107:223 229. 37. Goodson JL, Evans AK, Lindberg L (2004) J Comp Neurol 473:293 314. Goodson and Wang PNAS November 7, 2006 vol. 103 no. 45 17017