Sexual Skin Color Contains Information About the Timing of the Fertile Phase in Free-ranging Macaca mulatta
- First Online:
- Cite this article as:
- Dubuc, C., Brent, L.J.N., Accamando, A.K. et al. Int J Primatol (2009) 30: 777. doi:10.1007/s10764-009-9369-7
Females of several primate species undergo cyclical changes of their sexual skin, i.e., the development of a swelling or a change in color. The relationship between intracycle probability of fertility and the size of sexual swellings is well established, but in the only study to combine an objective measure of color with endocrinological data, researchers found no evidence that swelling color contains such information. To evaluate the role of female skin color in the context of sexual signaling further, we investigated whether changes in sexual skin color contain information about the timing of the fertile phase in rhesus macaques (Macaca mulatta), a species in which adult females do not develop sexual swellings, but do express visually detectable changes in the skin color of the face and hindquarters. Using an objective and quantitative measure of color, along with detailed data on fecal progestogen and estrogen metabolite levels collected from 8 females of the Cayo Santiago colony, we show that the ratio of red to green (R/G) for facial and hindquarter skin significantly varies throughout the ovarian cycle. In addition, facial skin R/G is significantly higher during the 5-d fertile phase versus the 5-d periods immediately before or after this time, but no such pattern occurs in hindquarter R/G. This suggests that skin color change in female rhesus macaques may potentially signal information about the intracycle probability of fertility to male receivers, but that only facial skin color may signal reliable information about its timing.
Keywordscolorfecal steroidsfertile phaseMacaca mulattasexual skin
In several primate species, there is variation over the ovarian cycle in the sexual skin of females, i.e., an increase in size (sexual swelling) or change in color (Dixson 1983; Nunn 1999; Zinner et al.2004). Among cercopithecines, the ancestral state appears to include changes in both the size and color of the skin of the anogenital region and its surrounding areas, and a lack of these traits is most likely due to secondary loss (Dixson 1983, 1998; Nunn 1999). Members of this subfamily that express changes in skin color, but do not have sexual swellings, may therefore represent an evolutionary stage intermediate to a complete loss of sexual skin changes (Dixson 1998; Sillén-Tullberg and Møller 1993). All such species express skin color changes in the anogenital region, e.g., vervets (Cercopithecus aethiops) and patas monkeys (Erythrocebus patas: Dixson 1983, 1998), while some also express color change in ventral anatomical areas such as the chest and abdomen, e.g., gelada baboons (Theropithecus gelada: Alvarez 1973; Matthews 1956), as well as the face, e.g., rhesus macaques (Macaca mulatta: Baulu 1976; Zuckerman et al.1938) and Japanese macaques (M. fuscata: Fujita et al.2004; Garcia et al.2009).
Given that changes in the sexual skin of cercopithecines occur over the ovarian cycle, they may contain information regarding the intracycle likelihood of conception. Indeed, a relationship between the size of sexual swellings and the timing of the fertile phase is well established (e.g., Macaca nigra: Thomson et al.1992; M. tonkeana: Aujard et al.1998; Pan troglodytes: Deschner et al.2003; M. fascicularis: Engelhardt et al.2005; M. sylvanus: Möhle et al.2005; Hylobates lar: Barelli et al.2007; Papio cynocephalus: Gesquiere et al.2007; P. anubis: Higham et al.2008). In contrast, few studies have investigated the potential link between female fertility and red skin coloration (Bradley and Mundy 2008), even though catarrhine species are trichromats and thus perceive red (Surridge et al.2003; Waitt and Buchanan-Smith 2006), and that researchers have proposed coloration to be an important sociosexual signal in these species (Changizi et al.2006; Fernandez and Morris 2007). Of the studies conducted, most lacked an objective means to quantify color (Czaja et al.1977; Gauthier 1999; Fujita et al.2004), endocrine information to determine the timing of the fertile phase (Setchell et al.2006), or both of these elements (Alvarez 1973; Baulu 1976; Matthews 1956). In the only study to date to combine an objective measure of color with endocrinological data, Higham et al. (2008) found no evidence that anogenital skin color contains precise information regarding the timing of the fertile phase in olive baboons (Papio anubis), a species with a prominent sexual swelling. To our knowledge, to date there is no comparable analysis on a species lacking a prominent swelling. To evaluate the role of female skin color in the context of sexual signaling more fully, data for more species based on objective measures of color and detailed hormone profiles are therefore required, with special attention given to species that do not have sexual swellings.
Adult female rhesus macaques do not exhibit sexual swellings, but do express changes in the skin color of the face and hindquarters, i.e., the anogenital areas, legs, thighs, and base of the tail, which are very pronounced, ranging from pale pink to deep red, and clearly visible to human observers (Baulu 1976; Bernstein 1963; Carpenter 1942; Cleveland et al.1943; Czaja et al.1977 Zuckerman et al.1938). There is some evidence to suggest skin color change in rhesus macaques may contain information about when females are most likely to be fertile during the ovarian cycle. For example, estrogen, in addition to its reproductive role, regulates variation in blood flow directly under the skin in this species, which in turn changes the red color of the skin (Dixson 1998; Rhodes et al.1997). Moreover, male rhesus macaques have a greater attraction to redder skin as shown by experiments that recorded gaze durations toward images of adult females (Gerald et al.2009; Waitt et al.2006). However, the most direct evidence is from studies by Baulu (1976) and Czaja et al. (1977). Using subjective observer ratings of color intensity, and reproductive assessment via reproductive hormones (Czaja et al.1977) or visual inspection of menstruation (Baulu 1976), these authors found that the hindquarters of captive females are reddest at midcycle, i.e., when females are more likely to be fertile. Though the results are promising, in order to understand the relationship between skin color and the timing of the fertile phase in rhesus macaques, studies in which both reproductive status and color are quantified objectively are required.
We combined an objective measure of color with detailed hormonal data to examine whether facial and hindquarter skin color of free-ranging adult female rhesus macaques varies over the course of the ovarian cycle in such a way as to reveal information about the timing of the fertile phase. This constitutes the first study of its type in nonhuman primates to investigate the potential role of skin color as a sexual signal in a species without a sexual swelling.
Study Site and Subjects
We studied free-ranging rhesus macaques on Cayo Santiago (Caribbean Primate Research Center, Puerto Rico). We collected data during the peak of the mating season from April 22 to July 12, 2007. At the time of study, our focal group (group V) comprised 22 adult females (≥7 yr old), 9 nulliparous females (3–5 yr old), and 15–20 sexually active males (≥4 yr old). Here we present data from 8 parous adult females (average age: 11.6, range: 7–18) for which both sufficient fecal samples for assessment of the ovarian cycle and skin color data were available. We analyzed 10 ovarian cycles for the face and 8 for the hindquarters. Of the 10 cycles used, 5 were conceptive (as indicated by maintenance of elevated pregnanediol glucuronide (PdG) levels for >4 wk or occurrence of birth).
Assessment of Skin Coloration
We used digital images of subjects’ faces and hindquarters to quantify color. We captured images in RAW format using a Canon EOS Digital Rebel XTi camera with a 10.1 megapixel CMOS censor and an EF28–135 mm f/3.5–5.6 IS USM lens and converted them to 16-bit TIFF files for analysis. We collected images approximately 1–3 meters from subjects with the flash disabled and with the shutter speed and aperture size determined automatically by the camera. We attempted to capture all images straight on, i.e. directly facing the camera, and avoided taking images of subjects in locations that were unevenly or heavily shaded, as well as those in full sunlight. We conducted image collection between 0730 h and 1030 h, a period leading up to and following the distribution of commercial food by CPRC employees and characterized by feeding behavior and general activity, e.g., traveling, vigilance.
Immediately after the capture of an image, we took a second photograph of a color rendition chart (GretagMacbeth ColorChecker [color chart]). We placed the color charts in the same location as the subjects, and photographed them under the same lighting conditions using the same shutter speed and aperture size as for the subject image (Bergman and Beehner 2008; Higham 2006; Higham et al.2008). The color chart consists of 24 colored squares of known and varying reflectance. Following Bergman and Beehner (2008), we adjusted subject images according to the known values of their corresponding color chart using the inCamera plug-in (Pictocolor Corporation, v. 4.0.1) for Adobe Photoshop (CS2, 9.0.1). The technique allows comparisons of color data between images captured under different lighting conditions and with different camera settings.
To verify whether our method measures color accurately, we tested our outputs for a linear relationship to light intensity and determined whether the reflectance values of the 3 color channels were equal (Stevens et al.2007; Stevens et al. this issue). To achieve this, we measured the red, green, and blue (R, G, and B) reflectance values for each of the 6 gray colored squares of 10 adjusted color chart images. Linear regressions of the measured R, G, and B values and the known reflectance values of the gray squares yielded an R2 of 1.0 for all 3 color channels. The absolute difference between measured reflectance values of R, G, and B was in the range of 0–2 (out of a maximum possible difference of 255) with a mean (± SD) difference of 0.52 ± 0.64. Based on these findings, we concluded our method measures color accurately.
Assessment of the Ovarian Cycle and Definition of the Fertile Phase
We collected a median of 31 fecal samples per focal female during the period of image collection (range: 14–42), with samples collected on average every 2.7 days (range: 1.9–5.8). We collected samples directly after defecation and discarded those that were contaminated with urine. We homogenized fecal boluses and placed 0.5–2 g in individual polypropylene tubes. We kept the samples on ice until we returned to the field station at the end of the observation day, where we were stored them at –20°C. We shipped the samples on dry ice to the German Primate Center for hormone analysis.
We used fecal progestogen metabolite profiles to determine the dates when ovulation most likely occurred (the ovulation window). We considered ovulation to have occurred when PdG concentrations rose above a threshold of the mean plus 2 standard deviations of 3–5 preceding baseline values, and maintained at this level for at least 3 consecutive samples (Jeffcoate 1983; Heistermann et al.2001). On the basis of a time lag of 24–56 h in the excretion of reproductive hormone metabolites in the feces of macaques (Shideler et al.1993) and to account for life span of the oocyte (Deschner et al.2003; France 1981; Higham et al.2008), we defined the most likely 2 d of ovulation as days –2/–3 relative to the defined PdG rise (Fig. 2; Brauch et al.2007; Heistermann et al.2001; Engelhardt et al.2004). Following Bosu et al. (1973), we set the length of the ovarian cycle at 28 d (Fig. 2). We defined the fertile phase as a 5-d period including the 2-d ovulation window and the 3 d preceding it to account for sperm life span in the female tract (Behboodi et al.1991; Wilcox et al.1995). We referred to the 5 d before and the 5 d after the fertile phase as pre- and postfertile phases (Fig. 2).
Data Analysis and Statistics
We included in the analysis only cycles for which the frequency of fecal sample collection during the periovulatory period allowed us to estimate the fertile phase with reasonable reliability, i.e., a maximum of a 3-d gap between the day of the PdG rise and the previous sample; median: 2, range: 0–3) and for which ≥1 picture was available per phase. We used a total of 10 cycles for the face (2 cycles for 2 females, 1 cycle for 6 females) and 8 for the hindquarters (1 cycle per female). A median of 12 images were available per 28-d ovarian cycle for facial skin (range: 10–15) and 11 for hindquarter skin (range: 7–14).
We performed general linear mixed models (GLMMs) to examine whether R/G varies in such a way as to reveal information about the timing of the fertile phase. GLMM is an extension of the general linear model that accounts for repeated measurements of the same subject and for unbalanced sample size by including random factors in the model. We analyzed the effect of a continuous fixed variable, day to estimated fertile phase, on R/G values and included female identity as a random factor, with cycle number included as a nested random factor in facial color analyses. In this analysis, we numbered each of the 5 d of the fertile phase 0; we labeled the day directly preceding the fertile phase day –1, the day directly after it day 1, and so on (Higham et al.2008, 2009). First, we tested whether R/G throughout the 28-d ovarian cycle follows a quadratic curve, i.e., highest values reached at midcycle when the fertile phase occurs. Because GLMMs test for linear relationships, we squared the numbers of the scale day to estimated fertile phase (Higham et al.2008, 2009). Next, to verify whether R/G values were higher during the fertile phase versus each of the other 2 phases, we tested the effect of the day to estimated fertile phase on R/G values during the 10-d periods spanning 2 of the 3 defined phases (fertile vs. prefertile and fertile vs. post-fertile). Given potential problems associated with GLMMs and sample sizes as small as ours, we also performed nonparametric statistics (Friedman tests with post hoc Wilcoxon signed-rank tests) using 1 cycle per female (the cycles for which R/G data were available for both facial and hindquarter skin) to confirm our results. The 2 types of analysis gave results with similar levels of statistical significance and we therefore present only the results of the GLMMs here. We used SPSS 17.0 for statistical analyses. All statistical analysis were 2-tailed and we set significance levels at p < 0.05.
Using an objective and quantitative measure of color, along with estimates of ovulation date based on measurements of fecal progestogen and estrogen metabolite levels, we show that red skin coloration (R/G) for 2 regions of free-ranging female rhesus macaque sexual skin significantly varies throughout the ovarian cycle in such a way that R/G values increase as the probability of fertility rises. Facial R/G values are significantly higher during the fertile phase versus the 5-d periods immediately before and after it, but such a pattern does not occur for hindquarter R/G. Therefore, although sexual skin color appears to contain general information about the probability of fertility during the ovarian cycle in rhesus macaques, only facial skin color seems to contain reliable information about its timing. Skin color in this species, which lacks a prominent swelling, therefore appears to contain similar graded information (Nunn 1999) about the timing of the fertile phase to swelling size in other catarrhine primates (Brauch et al.2007; Deschner et al.2004; Higham et al.2008).
Previous descriptive studies using subjective color measures have also reported that hindquarter color varies throughout the ovarian cycle and is most intense during midcycle —the presumed time of the fertile phase— in single-caged rhesus macaque females (Baulu 1976; Czaja et al.1977). Our results suggest that the period during which the highest R/G values are reached includes, but might exceed, the fertile phase and thus contains only partial information regarding its timing. In the only other study to combine detailed hormonal data with an objective measure of hindquarter sexual skin color, Higham et al. (2008) showed that, in olive baboons, the color of the sexual swelling does not contain information regarding the timing of the fertile phase. Although more studies are needed, it appears that in cercopithecines color changes in the skin of the anogenital region and its surrounding areas contain some information regarding the probability of fertility, but perhaps only in the absence of sexual swellings.
In addition to color in the hindquarters, Baulu (1976) also examined facial coloration in rhesus macaques and found that it did not show cycle-related changes. In contrast, our results suggest that facial color change contains reliable information about the timing of the fertile phase in this species. The discrepancy between these studies may be attributed to the accuracy of measurements. Baulu (1976) measured color based on weekly observer ratings and estimated the timing of ovulation from the menstruation date, both of which might not produce reliable data. Our results for facial coloration are in accord with studies in other primate species: facial skin was reddest during the periovulatory period of Japanese macaques (Fujita et al.2004) and mandrills (Mandrillus sphinx; Setchell et al.2006), although it should be noted that these studies used either an objective measure of color (Setchell et al.2006) or reproductive status (Fujita et al.2004), but not a combination of the 2. More studies using objective measures to investigate the role of color change as a sexual signal in areas outside the anogenital region are clearly needed.
To establish whether skin color change in rhesus macaques acts as a signal of the timing of the fertile phase, it is crucial to determine whether males can perceive and interpret the information contained therein (Maynard-Smith and Harper 2005; Snowdon 2004). An effective way to achieve this is with an experimental approach that examines the impact of skin color variation on male behavior in isolation from other potential signals and cues of female reproductive status, e.g., female behaviors (Engelhardt et al.2005). In pioneering experiments, Waitt et al. (2006) showed that single-caged rhesus macaque males gaze longer at red than at nonred images of female hindquarters, but showed no difference for female faces, whereas Deaner et al. (2005) found no effect of skin redness in the motivation of rhesus macaque males, as measured by juice sacrifice, to view female faces or hindquarters. As detailed reproductive hormone data were unavailable in these 2 experiments, it is unknown exactly what stage of the ovarian cycle the images used represented, which could have influenced results. Male rhesus macaques do pay selective attention to red color associated with pregnancy in images of female faces (Gerald et al.2009); thus red facial coloration is able to attract male attention. However, female skin color may contain information other than the timing of the fertile phase, such as age (swelling color: Setchell and Wickings 2004; Strum and Western 1982; facial color: Setchell et al.2006), competitive ability (facial color: Waitt et al.2006; cf. Setchell et al.2006), or parity (swelling color: Gauthier 1999; Higham et al.2008; cf. Setchell and Wickings 2004; facial color: Setchell et al.2006), which may or may not be perceived by males (Setchell and Wickings 2004). To investigate whether skin color change acts as a signal of the timing of the fertile phase in this species further, more experiments are required using stimuli based on detailed reproductive hormone data. Future experiments should also ideally be designed in a manner that takes into account the specifics of the rhesus visual system (Stevens et al. this issue).
If skin color does act as a visual signal of the timing of the fertile phase, it remains unclear why rhesus macaques have secondarily lost sexual swellings in their evolution only to express the information about the timing of the fertile phase with a different signal. Perhaps the costs of color change are less than those associated with swellings (e.g., increased body weight, parasite loads, risks of injuries, and water retention; Nunn 1999). As skin color change in the perineal area may be less conspicuous than swelling size, it may be more visible if it covers a larger skin surface: legs, thighs, tails, and face. Researchers have explained the sexual skin on the chest and abdomen of gelada baboons in terms of the large amount of time this species spends sitting on the ground feeding, which hides the anogenital area (Dixson 1983, 1998). This explanation could also apply to rhesus macaques because they may be one of the most terrestrial macaque species (Napier and Napier 1967); wild rhesus macaques spend a significant amount of time feeding on herbs and grass in some populations (Goldstein and Richard 1989), though there is variation in diet and habitat use between sites (Lindburg 1977; Seth et al.2001). Although we know little about the ecological conditions under which rhesus macaques evolved, it is likely that an ecological force would be at play in the evolution of a sexual signal in the skin in the upper body. If color change is more visible to potential male receivers in the face than in the hindquarters in rhesus macaques, it may be that facial skin color change is more likely to have been selected as a reliable signal of the timing of the fertile phase, as suggested by our results.
It is important to note that changes in color may occur more quickly and less predictably than changes to the size of a swelling because stress, emotion, and social interactions may affect blood flow, and thus skin redness, in a short-term manner independent of reproductive hormones (Bradley and Mundy 2008; Changizi et al.2006). If the information contained in color change can be interpreted by males, this information could perhaps be used more effectively by males who can monitor females on a regular basis, e.g., during a long consortship (Higham et al.2009). Moreover, as baseline and maximal colors vary between females in this species (Brent et al., unpub. data), previous experience with a given female may be crucial to the interpretation of the signal. As proposed by Higham et al. (2009), information regarding the timing of the fertile phase may therefore be unevenly distributed among males, which may potentially allow females to alter costs and benefits of male monopolization and bias paternity toward preferred males (Nunn 1999). A combination of behavioral and genetic data, along with objective measurements of hormones and color, may shed light on the function of sexual skin color in rhesus macaques, and lead to a greater general understanding of the evolution of sexual signaling in primates.
We thank the Caribbean Primate Research Center for permission to undertake research on Cayo Santiago; Juliet Alla, Julie Cascio, Camille Guillier, Edith Hovington, and Charlie McIntyre for assistance in data collection; Andrea Heistermann and Jutta Hagedorn for technical assistance in hormonal extraction and analysis; and the CPRC employees. We thank James Higham for inviting us to contribute to this special edition of the International Journal of Primatology on primate coloration. We thank Christoff Neumann for fruitful discussions; Roger Mundry for statistical advice; and Bernard Chapais, James Higham, and 2 anonymous reviewers for insightful comments on an earlier version of the manuscript. This project was funded by PhD fellowships awarded to C. Dubuc (SSHRC and Université de Montréal) and L.J.N. Brent (NSERC and Roehampton University). The investigation was approved by the IACUC of the University of Puerto Rico, Medical Sciences Campus. This publication was made possible by Grant No. CM-20-P40RR003640 from the National Center for Research Resources (NCRR), a component of the National Institutes of Health (NIH). Its contents are solely the responsibility of the authors and do not necessarily represent the official views of NCRR or NIH.