Sex differences in the brains of capuchin monkeys (Sapajus [Cebus] apella)

Abstract

This study reports an analysis of 20 T1-weighted magnetic resonance imaging scans from tufted capuchin monkeys (5 male, 15 female). We carried out a data-driven, whole-brain volumetric analysis on regional gray matter anatomy using voxel-based morphometry. This revealed that males showed statistically significant expansion of a region of the hypothalamus, while females showed significant expansion in a distributed set of regions, including the cerebellum, early visual cortex, and higher-order visual regions spanning occipital and temporal cortex. In order to elucidate the network connectivity of these regions, we employed probabilistic tractography on diffusion tensor imaging data. This showed that the female-enlarged regions connect with distributed association networks across the brain. Notably, this contrasts with rodent studies, where sex differences are focused in deep, ancestral limbic regions involved in the control of reproductive behavior. Additionally, in our data set, for several regions, male and female volumetric measures were completely nonoverlapping. This contrasts with human studies, where sex differences in cortical regions have been reported but are characterized by overlapping rather than divergent male and female values. We suggest that these results can be understood in the context of the different lifetime experiences of males and females, which may produce increased experience-dependent cortical plasticity in capuchins compared to rodents, and in humans compared to capuchins.

1 |. INTRODUCTION

Sex differences in neuroanatomy are an important topic of study because they may help explain sex differences both in typical healthy behavior and in responses to disease and treatment. Currently, the bulk of our knowledge about sex differences in brain structure and function come from rodent studies, which have been noted for decades. These are reported mostly in the deep, ancestral hypothalamic, and limbic circuits that control instinctual behaviors like reproduction and territory defense (for a review, see de Vries & Södersten, 2009). However, the brains of our own species are substantially larger and more complex than those of rodents. Human studies on sex differences in brain organization have had variable and sometimes conflicting results, but in general, have reported sex differences not only in ancestral limbic circuitry like the amygdala and hippocampal formation but also in higher-order association regions that are involved in complex social behavior and social cognition and show evidence of recent adaptive change, such as the temporal pole, orbitofrontal cortex, lateral prefrontal cortex, parietal operculum, and frontal pole (for a meta-analysis of human neuroimaging studies, see (Ruigrok et al., 2014)).

When might these additional sex differences in brain anatomy have evolved, and what sort of selection pressures might have led to their emergence? Clearly, humans have substantial behavioral and socioecological differences from rodents, which makes it difficult to link specific sex differences in the brain to specific sex differences in behavior. Moreover, rodents and humans are separated by over 96 MY of divergent evolution (Nei & Glazko, 2002), and our brains have a multitude of dissimilarities that are unrelated to sex. Furthermore, sex-linked variation ranges on a continuum from sexual dimorphism, where a trait is present in one sex and absent in the other, to sex differences, where a trait exists on a continuum with average differences between males and females, to sex convergence and divergence, where a trait appears similar in males and females but has different neural underpinnings, or where a sex difference emerges only after a particular exposure or experience (Mccarthy, Arnold, Ball, Blaustein, & De Vries, 2012). Relative to rodents, humans may be skewed toward the latter half of this spectrum (Joel et al., 2015). This raises the possibility that the selection pressures that produce sex differences in the brain may have different effects in large, complex brains than in smaller, simpler brains. These questions can be at least partially addressed via male/female comparisons in species that are more closely related to humans, that is, nonhuman primates. Taking a comparative approach to studying sexual dimorphisms in the brain using nonhuman primates as models provides a better estimate of when such dimorphisms might have emerged and allows us to more directly compare these differences to homologues in the human brain.

Despite the benefits of such an approach, a relatively small number of studies have investigated sex differences in the brains of nonhuman primates, possibly due to the challenges of doing so. One study reported that in rhesus macaques, the caudate, putamen, and hippocampus are larger in females, whereas the corpus callosum is larger in males (Knickmeyer et al., 2010). Another study in a very small sample size of rhesus macaques tentatively identified hypothalamic regions homologous to the human first, second, and third interstitial nuclei of the anterior hypothalamus, which is sexually dimorphic in humans; the putative rhesus homologues also showed enlargement in males (Byne, 1998). Although the specific genetic underpinnings of these results are at present unknown, it is likely that many of these sex differences in the brain are the result of different selection pressures experienced by males and females (i.e., are not a secondary effect of experience) and that these will differ across primate species. For example, a comparison of sex differences in the expression of 14,621 HUGO annotated genes in the cortex of humans and cynomolgus macaques (both relatively sexual dimorphic Old World primates) and marmosets (a relatively sexually monomorphic New World species) found that humans and macaques showed sex differences in the expression of hundreds of genes, whereas in marmosets, there were fewer than 10, potentially suggesting that the more closely related rhesus are a better model for humans (Reinius et al., 2008). On the other hand, despite these convergences in neuroanatomical sex differences between macaques and humans, other, further-diverged primate species show important convergences in behavioral ecology with humans, warranting further investigation. Thus, it is important to expand our inquiry to those species.

In the current study, we extended this area of research to a New World monkey species, the tufted capuchin monkey (Sapajus [Cebus] apella). While humans are more distantly related to New World monkeys than they are to Old World monkeys and apes, capuchins show a number of convergences with humans that make them an important study group for understanding sex differences in brain structure. First, capuchins, like humans, show extraordinarily high brain-to-body ratios; capuchins exhibit the largest brain to body ratio of any nonhuman primate, second only to humans in the primate taxa (Dicke & Roth, 2016; Jerison, 1975). In addition, they show socially sophisticated behaviors not seen in all primates. Captive experiments in S. apella show evidence of cooperation (De Waal & Berger, 2000) as well as evidence that they understand the contingencies of cooperation, such as attention to their partner (Mendres & De Waal, 2000) and sensitivity to inequity between partners (Brosnan & De Waal, 2003; Brosnan, Freeman, & De Waal, 2006; De Waal & Davis, 2003). Although there are fewer field studies of S. apella, species in the closely related Cebus genus cooperate in the wild (Perry & Rose, 1994; Rose, 1997), show triadic awareness in their choice of alliance partners (Perry, Barrett, & Manson, 2004), and even show culturally transmitted social behaviors that are argued to be used to show trust and solidify social bonds (Perry, 2011; Perry et al., 2003).

Like humans, capuchins exhibit sexual dimorphism both in body morphology and in behavior, which might be associated with parallel sexual dimorphisms in the brain (Fragaszy, Visalberghi, & Fedigan, 2004). Capuchins exhibit moderate sexual dimorphism. Although infants and juveniles are nearly impossible to distinguish visually by sex, adult males grow on average 30% larger in body size than females (“robust” S. apella: Fragaszy et al., 2004; “gracile” white-faced capuchins: Rose, 1994), develop bigger canines (Fragaszy et al., 2004; Rose, 1994), and have larger mastication muscles at maturity than females (Masterson & Hartwig, 1998). Capuchin males also develop distinctive secondary sex characteristics such as larger brow ridges, wider jaws, and broadened shoulders, which contribute to their overall larger appearance compared to females (Jack et al., 2014). Males and females develop at a similar rate, but males continue to develop longer than females and continue to build body mass during this time (Fragaszy et al., 2016). In addition, male capuchins exhibit a “secondary puberty” upon transition into an alpha position. In the presence of novel females, males show drastic increases in testosterone levels, body size, and heightened aggression (Benítez, Sosnowki, Reilly, & Brosnan, n.d.; Schoof, Jack, & Carnegie, 2011), and these changes persist in males that obtain the alpha position (Figure 1).

FIGURE 1.

Sexual dimorphism in male and female capuchin face morphology and magnetic resonance imaging (MRI) scans. Top: A capuchin male as an adolescent (left, age 7 years) and as a fully mature adult male, after he had become alpha male (right, age 13 years), with MRI scan collected after the alpha-male transition. Bottom: A capuchin female at age 4 years (left) and age 11 years (right) and MRI scan.

Source: Photographs taken by Catherine Talbot and Meg Sosnowski, reproduced with permission of the Language Research Center, Georgia State University

Sapajus genus also exhibit several forms of sex differences in behavior that may point toward underlying sex differences in brain organization. They live primarily in uni-male, multi-female social groups with strong bonds among the primarily philopatric females, although multi-male groups have been documented. Males are generally the sex that disperses at puberty (but see Tokuda, Martins, & Izar, 2018). Males’ mating access is based upon rank, which they obtain through fighting to either take over a group or to be the dominant male within a group. Adult males are also the primary defenders of their group from out-group capuchins and will form male–male coalitions to defend their territory from these rival groups (Scarry, 2013). Females will also participate in intergroup aggression, but not to the extent that males do (Scarry, 2013). Female capuchins form strong bonds with other females in their social groups. Behavioral studies show that females are more likely to reciprocally share food (De Waal, 1997) and groom (Di Bitetti, 1997) with specific social partners, whereas males are more likely to share indiscriminately with all group members, perhaps because most males are alphas, who maintain their tenure for years, so that other group members are their current or future mates or their offspring. The exception to this is when a new alpha male enters a group, in which case the existing infants are not his offspring. This is a dangerous situation for infants that are not yet weaned. Infanticide by males is common in robust capuchin monkeys (Lynch, & Rímoli, 2000; Izawa, 1992; Janson, Baldovino, & Di Bitetti, 2012; Ramírez-Llorens, Di Bitetti, Baldovino, & Janson, 2008) and is the most important source of mortality of non-weaned infants (Janson et al., 2012). Aside from supporting the sex difference in aggression, this also suggests selective pressure on females for tactics to avoid infanticide. Thus, we might expect to see sex differences in brain circuits related to aggression and/or social recognition, communication, and vigilance.

In general, there is little work on capuchin brain morphology, but one consistent finding is that throughout capuchin development (Phillips & Sherwood, 2008), females exhibit a larger corpus callosum: brain ratio than males. Other literature has shown that male capuchins that were right-handed had a smaller anterior midbody of the corpus callosum than female capuchins that were right-handed (Phillips, Sherwood, & Lilak, 2007), as well as a smaller corpus callosum:brain volume ratio than females overall. These females were also found to have a larger rostral body, posterior midbody, splenium, and isthmus than adult males. These studies suggest that sexual dimorphisms in the capuchin brain might be related to extractive foraging behavior, in which monkeys manipulate and destroy substrate to find invertebrate prey. Capuchins spend nearly half of their day manipulating substrate in search of food (Janson & Boinski, 1992). De Andrade and de Sousa (2018) showed that male capuchins (S. apella) engaged in extractive foraging significantly more often than females (but see Fedigan, 1990, for gracile capuchins) and exhibited sex differences in handedness in tool tasks, with females favoring their right hand and males favoring their left (de Andrade & de Sousa, 2018). This finding is also supported by a prey capture study in which males favored their left hand and were more successful at catching fish than females. The more efficient fish hunters—males—exhibited smaller corpus callosum: brain volume ratios (Hellner-Burris, Sobieski, Gilbert, & Phillips, 2010). It has been suggested that the functional significance of these morphological differences could stem from sex differences related to motor processing and spatial ability that arise through differences in foraging behaviors (Hellner-Burris et al., 2010).

Given that the corpus callosum is the main white matter tract linking the left and right cerebral cortex, these sex differences in corpus callosum structure suggest that sex differences might also exist in cortical anatomy. However, subcortical comparisons are also warranted, given that sex differences are commonly observed in these regions across mammals (and specifically, in the hypothalamus, caudate, putamen, and hippocampus in macaques (Byne, 1998; Knickmeyer et al., 2010). Accordingly, here we report the results of a whole-brain, male–female comparison of structural neuroimaging scans in tufted capuchin monkeys. We extend previous work by employing voxel-based morphometry (VBM) to examine differences in gray matter volume, a method that does not rely on a priori delineation of regions of interest and is thus sensitive to anatomical effects of any size or location. Furthermore, we use probabilistic tractography to elucidate the white-matter network connectivity of sex-differentiated gray matter regions.

2 |. METHODS

2.1 |. Subjects

This study involved 20 adult tufted capuchin monkeys housed at Georgia State University’s Language Research Center (5 males, 15 females). Subjects ranged in age from 6.5 to 21.0 years (mean = 14.1 years; standard deviation = 4.8 years), so none of the monkeys were considered to be in “old age” for this species. Four out of the five males were considered to be the alpha male of their group, and one was a beta male (lower ranked, but still an adult). A Mann–Whitney U test indicated no significant difference in age between males and females (U = 35.0, p = .827). Table 1 lists subject details. Capuchins were socially housed in mixed-sex stable, long-term groups of 4–10 individuals, each group with their own large indoor/outdoor enclosure, and were provided with a variety of enrichment items, including physical enrichment such as climbing structures and food enrichment. All subjects received a daily diet of fruits, vegetables, primate biscuits, nuts, and other enrichment foods. The Language Research Center is fully accredited by the Association for Assessment and Accreditation of Laboratory Animal Care. Animal involvement in research was in accordance with the Georgia State University guidelines. All research procedures were reviewed and approved by the Institutional Animal Care and Use Committee of Georgia State University.

TABLE 1.

Subject details

Name	Sex	Age at scan (years)
Albert	Male	6.6
Bailey	Female	17.8
Beeker	Female	11.5
Gambit	Female	21.0
Gonzo	Female	10.7
Gretel	Female	13.7
Griffina	Male	20.0
Ira	Female	6.5
Irene	Female	15.9
Ivory	Female	19.1
Lexi	Female	9.0
Liama	Male	13.9
Lily	Female	20.1
Logana	Male	12.1
Lychee	Female	18.3
Masona	Male	19.3
Nala	Female	15.4
Paddy	Female	7.7
Widget	Female	9.3
Wren	Female	15.1

^aMales who are the alpha of their social group and have undergone “secondary puberty” upon gaining alpha male status.

2.2 |. Scan acquisition and preprocessing

Neuroimaging scans were acquired at the Yerkes National Primate Research Center at Emory University, approximately 10 miles away. Monkeys were transported by van under veterinary supervision. Scans were acquired using a 3T Siemens Trio under anesthesia (3–4 mg/kg telazol). T1-weighted magnetic resonance imaging (MRI) images were acquired with a matrix size of 176 × 254 × 254 and a voxel size of 0.5 mm × 0.5 mm × 0.5 mm. Four T1 averages were acquired for each subject. Diffusion-weighted images were acquired with a matrix size of 94 × 110 × 52 and a voxel size of 1.0 mm × 1.0 mm × 1.0 mm. For each subject, two diffusion-weighted data sets were acquired with phase encoding in the left-to-right direction, and two with phase encoding in the right-to-left direction. Each data set included 11 images with no diffusion weighting and 128 diffusion-weighted directions.

2.3 |. Image preprocessing

FSL, an open-source neuroimaging software toolkit, was used for image preprocessing (Jenkinson, Beckmann, Behrens, Woolrich, & Smith, 2012; Smith et al., 2004; Woolrich et al., 2009). Preprocessing steps for averaged T1-weighted images included bias correction, accomplished using fast (Zhang, Brady, & Smith, 2001), and brain extraction, accomplished using bet (Smith, 2002). Diffusion-weighted images were preprocessed as follows. EPI distortion was corrected using topup (Andersson, Skare, & Ashburner, 2003), and eddy current correction was accomplished using eddy (Andersson & Sotiropoulos, 2016). Diffusion tensors were calculated using dtifit, and a probabilistic distribution of fiber orientations was modeled at each voxel using bedpostx, both part of FSL’s FDT toolkit (Behrens et al., 2003; Behrens, Berg, Jbabdi, Rushworth, & Woolrich, 2007).

2.4 |. Image analysis

In order to produce a common image space for analysis, we created a template brain which represents the unbiased anatomical average of all the subjects in this study. This was accomplished using the ANTS open-source software platform (Avants, Tustison, Song, & Gee, 2009). Because no prior template existed, an initial template was first constructed using rigid-body registration to the geometric mid-space of all images. Next, nonlinear registrations were computed using the ANTS buildtemplateparallel.sh script with the following parameters: 30 × 50 × 20 iterations per registration, Greedy-SyN transformation model, and cross-correlation similarity metric.

VBM is a type of neuroimaging analysis that assesses if and where regional brain volume differs in relation to a variable of interest (Ashburner & Friston, 2000). In order to perform VBM on our data set, the following steps were carried out. First, images were segmented into gray matter, white matter, and CSF using fast (Zhang et al., 2001). Jacobian determinant images were then computed for the native-space-to-template-space warps using ANTS. These represent an index of where and how much an image has to deform in order to match the template—in other words, they map each individual brain’s deviation from the group average. Jacobian determinant images were then masked with the gray matter segmentations, so that only gray matter voxels were compared across subjects. These masked images were then fed into a Monte Carlo permutation analysis using FSL’s randomise tool (Winkler, Ridgway, Webster, Smith, & Nichols, 2014). The statistical threshold was set at p > .05, and threshold-free cluster enhancement (Smith & Nichols, 2009; Spisak et al., 2019) was used to correct for multiple comparisons. The analysis controlled for differences in total brain size.

The sample size of this study is small and unbalanced. Our analysis approach addresses this issue in the following ways. First, permutation testing is nonparametric and is appropriate for data sets with unevenly matched groups. It involves comparing the actual male vs. female data distributions to bootstrapped distributions generated by randomizing the independent variable across subjects (sex). Second, our analyses were carried out in a whole-brain, data-driven manner. This is statistically more conservative than a region-of-interest-based approach. Third, we report individual data points so that readers can evaluate for themselves the extent of the observed effects.

In order to elucidate the network connectivity of regions showing sex differences, we carried out probabilistic tractography, a technique that traces the structural connectivity of seeded regions. This was accomplished using FSL’s probtrackx tool, which repeatedly samples the probability distribution of fiber orientations from bedpostx (Behrens et al., 2003; Behrens et al., 2007). Tractography was seeded using the voxels that showed significant female>male or male>female sex differences in the VBM analysis. The parameter settings for probtrackx were as follows: loopchecks enabled, curvature threshold 0.2; 2000 steps per sample, step length 0.5 mm, fiber threshold 0.1, and 5,000 samples per seed voxel. Tract output was thresholded at 0.05% of the waytotal. Tractography was run in subjects’ native diffusion space, with forward and inverse warp fields to the template included in order to visualize the results in a common group space. Groupwise connectivity maps were created by binarizing and summing subjects’ thresholded tractography output and then dividing by the total number of subjects. Thus, the intensity of each voxel in the groupwise map represents the percentage of subjects who showed above-threshold connectivity at a given location. We thresholded these group images to show connectivity common by at least 10% of subjects (i.e., at least two monkeys).

3 |. RESULTS

VBM analyses revealed a distributed cluster of regions that were larger in female capuchin brains, including parts of left posterior occipitotemporal cortex and bilateral cerebellum. Additionally, one cluster of voxels showed enlargement in male capuchin brains, located in the posterior hypothalamus. Table 2 lists the voxel coordinates and statistics for each significant cluster. Figure 2, red, shows clusters that were significantly larger in females. Figure 2, blue, depicts clusters that were significantly larger in males. Although age did not differ significantly between males and females, to be thorough, we also reran the age analysis using sex as a covariate. There was no continuous covariate interaction between age and sex anywhere in the brain, meaning the effect of age on gray matter volume did not differ significantly between males and females.

TABLE 2.

Statistics for all clusters showing significant sex differences

Cluster location	Sex difference	Voxels	Volume (mm3)	p value	Max voxel coordinates			Center of gravity voxel coordinates
					x	y	z	x	y	z
Left lateral cerebellar hemisphere	F > M	964	120.500	.005	67	82	79	65.3	80.8	81.9
Left occipitotemporal cortex: Superior temporal sulcus, inferior temporal gyrus, and ectocalcarine sulcus	F > M	715	89.375	.027	44	100	125	45	89.6	118
Right lateral cerebellar hemisphere	F > M	194	24.250	.025	111	76	82	113	78.8	83.1
Left occipital cortex: Inferior occipital sulcus	F > M	8	1.000	.048	61	67	107	61.3	67.2	108
Posterior hypothalamus	M > F	79	9.875	.019	93	124	92	92.8	124	93.1

Notes: p values are corrected for multiple comparisons using threshold-free cluster enhancement (Smith & Nichols, 2009; Spisak et al., 2019). Max voxel coordinates correspond to the voxel with the lowest (most significant) p value.

FIGURE 2.

Regions showing enlargement in female capuchin brains (red) and male capuchin brains (blue)

While the identity of these regions cannot be pinpointed precisely without histological and/or functional data, putative identifications can be made via comparison with past work. The female-enlarged cerebellar regions are located in cerebellar cortex in the lateral aspects of the posterior lobe, likely in Crus II. The smaller, more posterior female-enlarged cerebral cortical region is located in the inferior occipital sulcus, occupying parts of both the dorsal and ventral banks. This is congruent with the reported location of V2 and/or V3 (Nascimento-Silva, Gattass, Fiorani, & Sousa, 2003; Nascimento-Silva, Pinõn, Soares, & Gattass, 2014; Pinon, Gattass, & Sousa, 1998; Rosa, Piñon, Gattass, & Sousa, 2000; Sousa, Piñon, Gattass, & Rosa, 1991). The larger, more anterior female-enlarged cerebral cortical region spans a number of locations: the dorsal and ventral banks of the inferior occipital sulcus, potentially including V2 and/or V3 (Nascimento-Silva et al., 2003; Nascimento-Silva et al., 2014; Pinon et al., 1998; Rosa et al., 2000; Sousa et al., 1991); the lateral convexity of the inferior temporal gyrus, potentially including V4 and/or TEO (Pinon et al., 1998; Rosa et al., 2000); and the dorsal and ventral banks of the posterior superior temporal sulcus, just anterior to its junction with the lateral sulcus, potentially including V4 and/or MT (Diogo, Soares, Koulakov, Albright, & Gattass, 2003; Fiorani, Gattass, Rosa, & Sousa, 1989; Pinon et al., 1998; Rosa et al., 2000). The male-enlarged region is located in the posterior hypothalamus around the level of the fornix. MRI resolution is insufficient for positive identification of hypothalamic nuclei, but candidate nuclei near this location might include the mammillary bodies, paraventricular nucleus, and/or lateral hypothalamus (Eidelberg & Saldias, 1960).

In using group-level neuroimaging analyses to assess sex differences, it is crucial to plot individual data points, in order to confirm that apparent sex differences are not being driven by a small number of outlier values from one sex and to visualize the actual degree of separation vs. overlap between male and female values (Joel et al., 2015). To this end, we extracted the mean Jacobian determinant value for each individual within each region that showed a significant sex difference. Values greater than one represent enlargement relative to the group mean and values less than one represent reduction. Notably, for the regions that showed expansion in females, there is no overlap between male and female values. Figure 3 shows all individual subjects’ Jacobian determinant values for the regions that showed significant sex differences.

FIGURE 3.

Gray matter expansion (Jacobian determinant) values for individual subjects within the regions that showed significant sex differences. Blue: Males; red: Females. Error bars: ±SEM

In order to elucidate the potential function of the regions that showed sex differences, we examined their network connectivity using probabilistic tractography. Figure 4 shows these results. The tractography seeds, that is, the gray matter voxels that showed significant sex differences—are also plotted and are color-coded to correspond with Figures 2 and 3. Tractography is presented in order to inform anatomical understanding of the sex-differentiated gray matter regions; no statistical analysis of sex differences in white matter connectivity was performed here and is a separate question beyond the scope of the current research.

FIGURE 4.

Connectivity of regions showing sex differences in gray matter volume. Red: Tractography seeded in regions showing expansion in females. Blue: Tractography seeded in regions showing expansion in males. Color bars indicate percent of subjects showing above-threshold connectivity at each voxel. Only voxels with connectivity common to at least 10% of the group (two monkeys) are shown. CS, calcarine sulcus; IOS, inferior occipital sulcus; IPS, intraparietal sulcus; LS, lateral sulcus; pSTS, posterior superior temporal sulcus. In locations where sulci are obscured by tractography, the sulcus in the contralateral hemisphere is indicated

Tractography from the female-enlarged cerebellar clusters appears to pass through the superior cerebellar peduncle, which would be consistent with the location of the clusters in Crus II. Connections reach the inferior colliculus, the superior colliculus, and the posterior thalamus in the vicinity of the medial geniculate nucleus, lateral geniculate nucleus, and thalamus.

Tractography from the female-enlarged occipital and temporal cortex clusters reaches large portions of the inferotemporal cortex, likely via the inferior longitudinal fasciculus (Borges, Nishijo, Aversi-Ferreira, Ferreira, & Caixeta, 2015). Connections reach cortex surrounding the calcarine, inferior occipital, and posterior superior temporal sulci, where multiple regions in the ventral visual stream are located from V1 to TEO (Pinon et al., 1998; Rosa et al., 2000). Additionally, the more anterior female-enlarged occipitotemporal cluster reaches the fundus of the posterior superior temporal sulcus and the intraparietal sulcus, the site of the dorsal visual stream including regions MT, MST, VIP, LIP, and AIP (Boussaoud, Ungerleider, & Desimone, 1990; Fiorani et al., 1989). Tracts additionally reach the claustrum, amygdala, basal forebrain, and orbitofrontal cortex.

Tractography from the male-enlarged hypothalamus cluster reaches lateral prefrontal cortex, anterior caudate, claustrum, amygdala, basal forebrain, and brainstem areas including the periaqueductal gray.

4 |. DISCUSSION

The current study compared regional gray matter morphometry—that is, increases or decreases in gray matter volume—in male and female capuchin monkeys. We found that males showed expansion of a region of the hypothalamus, while females showed expansion in a distributed network of regions, including the cerebellum and occipital and temporal visual regions likely spanning from V2 to TEO and MT. The diffusion tensor imaging (DTI) connectivity of these areas indicates that they are part of a distributed network of regions associated with social processing, including the dorsal and ventral visual streams, higher-order cortical association regions like the orbitofrontal cortex and temporal pole, and subcortical regions such as the amygdala, basal forebrain, and claustrum. The female-enlarged visual and temporal cortical regions we observed may be a coincidence with the enlargement of the caudal midbody of the corpus callosum reported by Phillips and Sherwood (2012), given that posterior sectors of the corpus callosum provide the interhemispheric connections for these gray matter regions, as we observed in our tractography analyses (see Figure 3).

What might drive these differences in morphology? An initially appealing but ultimately overly simplistic explanation is that each sex is somehow doing “more” processing in its expanded regions and that the anatomical differences we report here are therefore indicative of innate, sex-specific adaptations for behavior, perception, or cognition. However, the link between variation in structure and variation in function is not always so straightforward (Striedter, 2005). Furthermore, some sex differences in neuroanatomy are thought to be compensatory in nature in that they exist in order to counteract or normalize the effects of differential hormone exposure (for a review, see De Vries, 2004). We propose two potential explanations for different aspects of our results.

First, sex differences in the brain and behavior can be driven by sex differences in hormone exposure. For example, evidence in support of gonadal hormones influencing brain organization has been reported in rhesus macaques (Clark & Goldman-Rakic, 1989). Females that were treated with testosterone propionate (TP) and that had lesions in the orbital prefrontal cortex performed the same as males that had lesions in the same region compared to unmanipulated females on an object reversal task, and unlesioned TP-treated females performed the same as normal males, suggesting an influence of hormones on cortical development.

Thus, one possible explanation for our results is that the hypothalamic expansion found in capuchin males may be related to biological differences in testosterone exposure in males compared to females. The male-enlarged hypothalamic region we observed in capuchins may be homologous to the male-enlarged hypothalamic regions in macaques and humans (Allen, Hines, Shryne, & Gorski, 1989; Byne, 1998; Swaab & Fliers, 1985), suggesting that this is widespread in primates. Notably, like capuchins, macaques also experience secondary growth upon transitioning into alpha. Thus, it could be that the elevated testosterone exposure that adult males experience upon becoming the alpha male of a group (Benítez et al., n.d.; Schoof et al., 2011) is related to this hypothalamic expansion. Four of our five males were alpha males in this study, which makes it difficult to discern differences due to sex androgen exposure (innate biological differences) vs. those due to social status (individual social experience). These results open the door to future studies examining this link between circulating levels of hormones, specifically testosterone, and expansion of these brain regions. Future research could address this possibility via comparison of hypothalamic gray matter expansion values with measurements of testosterone and social behavior.

A second, not mutually exclusive, possibility is that some sex differences are not immediately driven by innate sex differences in hormone exposure but rather are emergent, experience-dependent secondary effects of the differential life histories, experiences, and social niches. This second scenario seems like a more probable explanation for the expansion of early visual cortex and higher-order occipitotemporal regions in female capuchins. Notably, our DTI results indicate that these regions show connectivity with distributed association networks across the brain. Data on trajectories for brain development in capuchins are lacking, but in both humans and macaques, higher-order association regions are among the last to develop, both in terms of gray matter morphology and white matter connectivity (Buckner & Krienen, 2013; Flechsig, 1920; Hill et al., 2010; Miller et al., 2012; Yakolev & Lecours, 1967). Additionally, early visual cortex is well-known to show experience-dependent plasticity both in early development and adulthood (for reviews, see Espinosa & Stryker, 2012; Karmarkar & Dan, 2006). Thus, expansion of these regions in female capuchins might support increased social-perceptual processing, vigilance, and individual recognition. These differences may be due to sex differences in social and ecological niches resulting in sex-specific behaviors. For female capuchins, reproductive success is contingent on acquiring access to high-quality resources and forming strong social bonds often with maternal kin (Fragaszy et al., 2004), not to mention navigating infanticide, as discussed earlier. As such, social-perceptual processing (e.g., navigating a social group), vigilance (e.g., keeping track of dominant individuals when foraging or males’ whereabouts), and individual recognition (e.g., necessary for forming and maintaining preferred social bonds) would all be important to a female’s overall fitness. For a male capuchin, reproduction is contingent on his social status; he typically leaves his social group and competes against males for a dominant position in the hierarchy. In the population of capuchins studied here, males were members of a stable social group and were often the only male, as is true in most Sapajus groups; thus there was little reproductive competition for access to females, while females compete daily for resources and male attention. The expansion of early visual cortex and higher-order occipitotemporal regions in female capuchins could reflect this difference in how male and female capuchins move through social spaces. A promising avenue of research would be to examine sex-difference in these regions of the brain in nonhuman primate species, like chimpanzees, where males remain in their social groups, form strong stable social bonds with kin, and routinely compete with each other and against other social groups. Doing so would help elucidate how sex-based social experiences may alter brain morphology.

Differences in social experience are also hypothesized to play a role in the emergence of neural sex differences in humans (Eliot, 2011). Additionally, outside the realm of sex differences, there is substantial research indicating that differential life experiences can produce differences in brain anatomy that are detectable with the type of methods we used here (e.g., Bremner, Elzinga, Schmahl, & Vermetten, 2007; Li, Legault, & Litcofsky, 2014; Maguire et al., 2000; for a review on potential mechanisms, see Zatorre, Fields, & Johansen-Berg, 2012). Capuchins offer a valuable point of comparison for understanding human sex differences in brain anatomy because they are far more closely related to us than rodents, in which most research on sex differences has taken place. Notably, rodent studies reliably find anatomical differences in brain regions involved in the generation and regulation of instinctive behaviors such as mating and territory defense; these results are highly reproducible within a species (for a review, see de Vries & Södersten, 2009). In contrast, research on human sex differences has been marked by variable and sometimes contradictory findings, particularly in higher-order cortical regions, and human research generally has not produced the sort of clear-cut differentiation in anatomy visible in rodents (for a meta-analysis, see Ruigrok et al., 2014). The results of the current study are in some ways intermediate to reports from the rodent and human literature. As in rodent studies, we found clear differences in brain anatomy in regions involved in the regulation of instinctive, sex-differentiated behavior (i.e., the hypothalamus). Furthermore, as in rodent studies, we also observed total sex differentiation in the anatomy of several regions, with completely non-overlapping male and female values. However, in contrast to rodent studies, and similar to human studies, we also observed sex differences in cortical regions that are likely involved in attention, perception, and more complex aspects of social behavior. Future work could build upon the results reported here by confirming which specific behaviors are controlled by the sexually dimorphic areas that we identified, as we were unable to determine a causal relationship between the differences in brain structure and the functional significance of those differences in this study.

McCarthy et al. (2012) distinguish between sexual dimorphism, where male and female traits are largely nonoverlapping, vs. sex differences, where male and female traits partially overlap on a continuum. The capuchin results we report here are more dimorphic than what has been reported in human brains (Joel et al., 2015). What might explain this disparity, and what implications could this have for our understanding of our own species? One important point is that while capuchins are moderately sexually dimorphic in, for example, canine size and total body mass, humans are even less so; the selection pressures that led to this state in humans may also have resulted in reduced sexual dimorphism in the brain. Another important point is that while capuchins are highly socially complex and do show a great deal of social learning, humans far outpace all other primate species in these capacities. Human behavior is more flexible and more dependent on social and cultural learning than any other species on the planet. Likewise, human brains show evidence of adaptations to increase plastic responses to the social and cultural environment (reviewed in Buckner & Krienen, 2013). Association networks in the human neocortex are characterized by volumetric enlargement relative to other species (Genovesio, Wise, & Passingham, 2014; Passingham & Smaers, 2014; Preuss, 2011; Sherwood, Bauernfeind, Bianchi, Raghanti, & Hof, 2012), reduced genetic constraint (Gomez-Robles, Hopkins, Schapiro, & Sherwood, 2015), prolonged developmental plasticity (Buckner & Krienen, 2013; Flechsig, 1920; Yakolev & Lecours, 1967), rapid growth during development (Hill et al., 2010), and elaborated white matter connectivity (Buckner & Krienen, 2013). Interestingly, comparisons with macaques suggest that this pattern of developmental expansion is mirrored by the pattern of evolutionary expansion, perhaps because it is adaptive for recently evolved regions to mature more slowly, to increase the influence of experience on those regions (Hill et al., 2010). Humans also show increased expression of genes related to synaptic plasticity and remodeling (Bauernfeind et al., 2015; Caceres, Suwyn, Maddox, Thomas, & Preuss, 2006; Fu et al., 2011; Konopka et al., 2012; Preuss, Cáceres, Oldham, & Geschwind, 2004) and “transcriptional neoteny” in the developmental timing of gene expression (Somel et al., 2009). Viewed in this light, humans’ increased capacity for neural plasticity means that our brains are especially sensitive to the effects of environmental experience, including culturally driven social experiences. Thus, perhaps more in our own species than in any other, our own behavior has the possibility to either further separate or equalize any innate sex differences in brain anatomy. Future work could address this possibility by investigating sex differences in additional primate species, in order to more accurately identify neural systems where sex differences are impacted by prolonged developmental plasticity. Furthermore, future work could build upon the results reported here by confirming which specific behaviors are controlled by the sexually divergent areas we identified.

The primary limitation of the present study is its sample size, which is small and unbalanced. We employed statistical approaches designed to deal with this issue as best as possible: we used a whole-brain, data-driven approach, which is more conservative than a region-of-interest-based approach, we used nonparametric statistics including permutation testing, and we reported individual subjects’ gray matter morphometry measurements, revealing a clear separation between male and female values. The present study notably did not identify sex differences in the morphology of the medial preoptic area, commonly observed to be sexually dimorphic. It is possible that future research with a larger number of subjects might uncover this. Furthermore, sex differences in capuchin brain anatomy could be better understood with future research addressing links with behavior, hormones, and genes.

DATA AVAILABILITY STATEMENT

Research data are not shared.