The Science of Nature

, 104:81 | Cite as

Skin lipids of the striped plateau lizard (Sceloporus virgatus) correlate with female receptivity and reproductive quality alongside visual ornaments

  • Jay K. Goldberg
  • Alisa K. Wallace
  • Stacey L. Weiss
Original Paper


Sex pheromones can perform a variety of functions ranging from revealing the location of suitable mates to being honest signals of mate quality, and they are used in the mate selection process by many species of reptile. In this study, we determined whether the skin lipids of female striped plateau lizards (Sceloporus virgatus) can predict the reproductive quality of females, thereby having the potential to serve as pheromones. Using gas chromatography/mass spectrometry, we identified 17 compounds present in skin lipids of female lizards. Using principal component analysis to compare the skin lipid profile of receptive and non-receptive females, we determined that an uncharacterized compound may allow for chemical identification of receptive mates. We also compared extracted principal components to measures of female fitness and reproductive qualities and found that the level of two 18 carbon fatty acids present in a female’s skin lipids may indicate her clutch size. Finally, we compared the information content of the skin lipids to that of female-specific color ornaments to assess whether chemical and visual cues transmit different information or not. We found that the chroma of a female’s orange throat patch is also related to her clutch size, suggesting that chemical signals may reinforce the information communicated by visual ornamentation in this species which would support the “backup signals” hypothesis for multiple signals.


Chemical cues Lizards Multimodal communication Pheromones Reptilian Skin lipids 


Chemical signature mixtures and pheromones come in various forms and can transmit myriad messages between conspecific organisms. For example, they have been shown to reveal the location of an individual to conspecifics, mark territory boundaries, and transmit information regarding the quality of potential mates (Johanssen and Jones 2007; Wyatt 2014). Signature mixtures are blends of compounds that transmit information to conspecifics without an evolutionarily derived signaling function whereas pheromones are chemical compounds with a signaling function that is the result of selection acting upon both signalers and receivers (Wyatt 2014). Both forms of chemical cues are understudied in the sexual selection literature (Penn and Potts 1998), as research has emphasized the role of striking visual and auditory displays (Moore et al. 2016).

Reptiles, like many other animals, mediate social interactions with chemical cues (Houck 2009). Inter-species and inter-population variation in chemical profiles is known to drive species identification, reproductive isolation, and speciation in both snakes (LeMaster and Mason 2002) and lizards (Barbosa et al. 2006). Inter-individual variation also plays an important role in mate choice by honestly signaling phenotypic information (LeMaster and Mason 2002). Reptilian pheromones can originate from many physiological sources (Weldon et al. 2008), yet prior research with lizards has focused on the secretions of the femoral glands (Alberts 1990; Mártin et al. 2013). The femoral glands are more productive during the breeding season, suggesting they play a role in mating (Martins et al. 2006). In many species, femoral glands are only active in males and play a role in rival assessment (Carazo et al. 2007), species identification (Gabriot et al. 2010), and mate assessment (Mártin et al. 2007). Less is known about the role of female chemical signals in the reproductive behavior of lizards.

The striped plateau lizard (Sceloporus virgatus) is a medium-sized lizard native to the Chiricahua Mountains of Arizona, USA. Males of the genus Sceloporus usually have bright colored ornamentation, yet S. virgatus has recently lost this ancestral trait (Wiens 1999). Instead, females possess orange throat patches which signal reproductive quality to conspecifics and influence male behavior (Weiss 2002; Weiss 2006; Weiss et al. 2009). The loss of male visual ornamentation in S. virgatus has been linked to higher rates of chemosensory behaviors (Hews et al. 2011), suggesting that chemical communication may play a greater role in this species than in related species. Consistent with this idea, S. virgatus males frequently tongue flick the flanks of females during courtship (Weiss, personal observation), and behavioral evidence has been found suggesting that information regarding female body size is transmitted via chemical cues (Fritzche and Weiss 2012). Despite this, little progress has been made in elucidating the composition of chemical cues in S. virgatus.

Females of this species lack active femoral glands, thus making skin lipids a likely source for female pheromones although others certainly exist. The primary function of skin lipids is believed to be preventing water loss (Roberts and Lillywhite 1980), yet they are also known to function as pheromones in some species (Weldon et al. 2008). Components of the epidermal lipids in female red-sided garter snakes (Thamnophis sirtalis parietalis) produced only during the mating season (Uhrig et al. 2012) serve as an attractiveness pheromone communicating phenotypic information to conspecifics (Mason et al. 1989) whereas differences between the skin lipids of male and female leopard geckos mediate sex recognition (Mason and Gutzke 1990).

The use of both visual and chemical cues by S. virgatus females presents the unique opportunity to study the information content of female cues in separate sensory modalities. Theoretical models have shown that when a species utilizes multiple displays, the displays can signal different information (multiple messages) or strengthen transmission of another signal (backup signals; Johnstone 1996). Thus, by relating measures of female color with phenotypic measures, we set out to determine if communication via skin lipids could serve as a backup to visual ornamentation, or signal a separate message.

In this study, we address three questions regarding the epidermal lipids of S. virgatus females: (1) what is the chemical makeup of their skin lipids? (2) does the profile of the skin lipids indicate a female’s receptivity? and (3) can a female’s skin lipid profile predict her phenotype/reproductive quality? We also examined the information content of a female’s color ornament following Weiss (2006) and compared this to the information content of her skin lipid profile.

Materials and methods

Lizard collection and maintenance

Sceloporus virgatus females (hereafter “lizards”) were captured by noose from the vicinity of the Southwestern Research Station (SWRS, Portal, AZ) on May 26 (N = 8), June 7 (N = 6), and June 20 (N = 7) 2010. Lizards (total N = 21) were housed individually in terraria (22.9 × 15.2 × 16.5 cm) on a screened porch, allowing for natural light and temperature fluctuations for 7 days. Additional heat was provided by a 40 W incandescent bulb in a metal reflector on a 12:12 light:dark cycle. Lizards were fed crickets (Acheta sp., Fluker Farms) and provided access to water twice during this period; feedings occurred in separate 10-gal glass terraria to control the duration of exposure to prey-borne chemicals.

Phenotypic measurements

Upon capture, lizards were weighed to determine body mass, measured with a transparent ruler to determine body size (snout to vent length, SVL), and had the number of trombiculid and pterygosomatid mites on their body counted. The residuals of a regression of SVL and body mass were used as the measure of body condition (Weiss 2006). Following euthanization and extraction of skin lipids (detailed below), females were dissected in order to classify the reproductive state of each female as non-reproductive (NR, N = 7), vitellogenic (V, N = 5), or gravid (G, N = 9). NR females were those that had not yet reached sexual maturity. Reproductive females are vitellogenic before ovulation occurs whereas after ovulation, females become gravid and are no longer sexually receptive. The ovaries of V females were weighed and the number of enlarged follicles was counted. For G females, the egg-containing oviducts were removed, weighed for mass, and the number of eggs was counted.

Color measurement

On the 8th day following capture, the right throat patch of each female was photographed along with a small ruler using an Olympus C-5050 ZOOM 5-megapixel digital camera set to macro mode with a Super Bright zoom F1.8 lens under standardized indoor lighting. Ornament area was measured by selecting orange pixels from the photograph using the “color range” command of Adobe Photoshop 9.0 (Weiss 2006) and determining the area of the selected pixels (in mm2) in National Institutes of Health ImageJ 1.60. The value (relative lightness or darkness) and chroma (degree of saturation) of the ornament were determined by matching the orange color of females to Munsell color chips; SLW performed all color matches under standardized light conditions following Weiss (2006). Each Munsell color chip is classified by separate measures of hue, value, and chroma. All reproductive females were matched to Munsell hue 10R but varied in both value (range 4.0–6.0) and chroma (range 10.0–16.0). Higher numbers for value and for chroma indicate lighter and more saturated color, respectively. Spectrometry shows that the female orange color is non-reflective in the UV spectrum (Weiss et al. 2012) and is similar in shape to that of the matching Munsell color chip (Weiss, unpublished data).

Lipid collection

Immediately following color measurements, lizards were euthanized with 0.1 mL of 50 mg/mL pentobarbital solution. Euthanized lizards were then placed in glass jars, covered to the neck in hexane, and soaked for 12 h. Hexane was decanted into 32-mL storage vials and left to dry overnight in a fume hood. Vials were then weighed, wrapped in aluminum foil, and transported to the University of Puget Sound, Tacoma, WA, where they were stored at − 70 °C until analysis began in February 2013. Lipids collected from lizards captured on June 20 were not dried or weighed at SWRS due to time constraints. These samples were transported and stored with the others, then dried under reduced pressure via rotary evaporation at 45 °C in March 2013. Control samples (N = 6) of hexane subjected to the above procedure were prepared (on 29 June 2010), transported, and stored along with the lipid samples.


Preliminary analysis revealed that the skin lipids were primarily composed of triacylglycerides, the precise structures of which are hard to obtain from direct analysis; thus, samples were derivatized to allow for the identification of fatty acids within lipid samples using gas chromatography/mass spectrometry. An aliquot of lipid (too small to be accurately weighed) was scraped from storage vials with a metal spatula and dissolved in 0.2 mL toluene. 1.5 mL anhydrous methanol and 0.3 mL of 8% HCl in methanol were added (Ichihara and Fukubayashi 2010). The mixture was then heated to 120 °C for 10 min. This reaction breaks triacylglycerides into fatty acid methyl esters (FAMEs) and glycerol. FAMEs were then separated by adding 10 mL distilled water followed by 1 mL hexane and shaking in a separatory funnel. The hexane layer (FAME containing) was dried over sodium sulfate and concentrated to 200 μL under a stream of dry nitrogen before 50 μL MSTFA (N-methyl-N-trimethylsilyl-trifluoroacetamide) was added. This solution was then heated to 110 °C for 1 h. The final product was reduced to 100 μL under a stream of nitrogen before being analyzed immediately using GC/MS. Residue from control vials was subjected to this procedure as well. Samples from multiple individuals were never pooled. All chemical reagents and standards used were purchased from Sigma-Aldrich (St. Louis, MO) and used without further refinement or derivatization except when stated below. All reactions were heated in pressure vials using a microwave reactor (CEM Corporation, Matthews, NC).

Gas chromatography and mass spectrometry

Derivatized samples were analyzed using an Agilent Technologies (Santa Clara, CA) 6850 Network gas chromatograph system connected to a 5973 Network mass selective detector. Separations were achieved using a 15 m × 0.32 mm ZB-5HT capillary column (Phenomenex, Torrance, CA) coated with 0.1 μm of 5% diphenyl/95% dimethylpolysiloxane. Five microliter sample injections were performed in splitless mode with helium as the carrier gas at a constant flow rate of 1.2 mL/min. Injector and detector temperatures were set to 300 and 290 °C, respectively. Oven protocol was as follows: 50 °C isothermal for 5 min, ramp to 150 °C at 10 °C/min, and isothermal for 10 min, then ramp to 300 °C at 3 °C/min and isothermal for 10 min, then ramp to 350 °C at 10 °C/min and isothermal for 10 min. Electron impact-mass spectrometer (EI-MS) conditions were source temperature of 230 °C, ionizing energy at 70 eV, and scan range from 45 to 800 m/z. Peaks were identified and their areas calculated as percent of the total ion current (TIC) using the Agilent G1701DA ChemStation software. Peaks representing > 0.1% of the TIC and present in all samples of at least one reproductive state were selected for identification and analysis. Contaminants identified in control vials were excluded. Initial identification of the 17 remaining peaks (Table 1) was made via comparison of spectra with the National Institute of Standards and Technology (NIST) mass spectral library. Identifications were confirmed via comparison of retention times and spectra with those obtained from authentic standards, when available. Sterol standards were obtained in the unsilylated form and were derivatized using the procedure described above. All fatty acids (FA) were identified as the corresponding FAME and all sterols were identified as the corresponding trimethylsilylether. The double bond position and geometry of unsaturated FA could not be determined using this method.
Table 1

Mean (± standard deviation) for the % TIC of each compound identified in the skin lipids from females of each reproductive state. Sample sizes for mean % are 7 (NR), 5 (V), and 9 (G) except when a peak was not found in all individuals of a given reproductive state; for these exceptions, n is shown in parentheses. For unidentified compounds, m/z of notable peaks are shown in parentheses

Peak #








Tetradecanoic acid

0.88 ± 0.18

0.52 ± 0.16

0.61 ± 0.16



Hexadecanoic acid

32.96 ± 4.68

24.41 ± 6.41

28.97 ± 4.88



Heptadecanoic acid

0.63 ± 0.07

0.54 ± 0.16

0.67 ± 0.15



Putative octadecadienoic acida

0.83 ± 0.53

0.48 ± 0.24

0.63 ± 0.18



Octadecenoic acid

2.84 ± 0.92

2.10 ± 0.77

2.67 ± 0.34



Octadecanoic acid

15.90 ± 3.31

13.24 ± 3.92

17.69 ± 2.76



Eicosanoic acid

2.58 ± 0.57

2.13 ± 0.54

2.95 ± 0.39



Docosanoic acid

2.12 ± 0.94

1.21 ± 0.36

2.27 ± 0.79



Tetracosanoic acid

0.66 ± 0.20

0.69 ± 0.22

0.61 ± 0.13



Hexacosanoic acid

0.35 ± 0.05 (4)

0.37 ± 0.08 (3)

0.49 ± 0.10




13.02 ± 3.96

14.22 ± 6.56

14.24 ± 2.11



Octacosanoic acid

0.51 ± 0.23 (6)

0.50 ± 0.33

0.94 ± 0.19



Unidentified steroid (343, 367, 382, 395, 413, 457, 472)

0.93 ± 0.58

0.78 ± 0.67

1.34 ± 0.35




0.66 ± 0.42

1.05 ± 0.68

1.19 ± 0.28



Unidentified steroid (215, 305, 383, 398, 473, 488)

0.45 ± N/A (1)

0.65 ± 0.32 (4)

0.57 ± 0.14



Putative triacontanoic acida (367, 423, 466)

0.32 ± 0.08 (4)

0.24 ± 0.05 (2)

0.44 ± 0.13



Unidentified compound (393, 451, 494)

0.25 ± 0.11 (2)

0.40 ± 0.17

aDenotes that the analyte spectrum matched that of the authentic standard; however, their retention times did not

Skin lipids as indicators of receptivity

All statistical analyses were performed using RStudio (R Core Team 2012). To determine whether skin lipid profiles can indicate a female’s receptivity, we looked for differences in the skin lipid profile between females of different reproductive states. The areas of 12 peaks identified in all females were restandardized to 100% and transformed using Aitchison’s formula as this is the most accepted way of analyzing composition data of this nature (Dietemann et al. 2003; Gabriot et al. 2010). Homogeneity of variance in the transformed areas (across the three reproductive states) was tested using Levene’s test and no variables violated this assumption. The transformed areas were then used as variables in a principal component analysis. The three principal components (PCs) with eigenvalue > 1 extracted together accounted for 75.1% of the variation (Table 2) and were used as response variables in a multivariate analysis of variance (MANOVA) with reproductive state as the explanatory factor (Whittaker et al. 2010). One-way analysis of variance (ANOVA) and Tukey’s HSD were used for post hoc analysis.
Table 2

Eigenvalue, percent of variance explained, one-way ANOVA across reproductive state results, and variable loadings for the three PCs extracted. PCA includes compounds found in all three reproductive states









% of variance explained




F statistic (ANOVA)




P value




Tetradecanoic acid

− 0.339

− 0.167


Hexadecanoic acid

− 0.351



Heptadecanoic acid

− 0.122



Octadecadienoic acid

− 0.341

− 0.350


Octadecenoic acid

− 0.314

− 0.148


Octadecanoic acid

− 0.161



Eicosanoic acid

− 0.092


− 0.211

Docosanoic acid

− 0.346

− 0.042

− 0.344

Tetracosanoic acid


− 0.144






Unidentified steroid



− 0.552



− 0.058


P value in italics is significant at the 0.05 level. Loadings in italics indicate that the compound was strongly associated with that PC

Skin lipids as predictors of reproductive quality

To determine whether skin lipid profiles can predict the reproductive quality of reproductively active (V and G) females, we looked for correlation between components of the skin lipid profile and various phenotypic traits. The areas of the 13 peaks identified in all reproductive females were restandardized to 100% and transformed using Aitchison’s formula (Dietemann et al. 2003; Gabriot et al. 2010). Homogeneity of variance of the transformed areas (across the two reproductive states) was tested using Levene’s test and no variables violated this assumption. The transformed areas were then used as variables in a principal component analysis. The five PCs with eigenvalue > 1 extracted together accounted for 87.9% of the variation (Table 3) and were used as explanatory variables in multivariate regression models. Each phenotype variable was used as the response variable in separate models. Regression models may also include additional explanatory variables; these control variables were used to examine the effect of collection date, reproductive state, body size, and body condition in cases when that variable (1) was suspected to influence the relationship of interest, (2) significantly correlated to the phenotype variable, and (3) had a significant effect on the regression model (Weiss 2006). No control variables had an effect on any measure of female color, mite load, body size, or body condition, and thus, regression models for these phenotype variables did not include any additional explanatory variables. Body condition correlated with a female’s clutch size (Pearson’s R2 = 0.419, F(1,12) = 8.662, P = 0.012); thus, it was controlled for in that model. Both collection date and reproductive state had an effect on a female’s average egg mass (date F(1,12) = 46.79, P < 0.001; reproductive state F(2,11) = 41.19, P < 0.001), yet collection date was not included in that model due to redundancy issues. Orange patch area was natural log transformed prior to being used in all analyses.
Table 3

Eigenvalues, percent of variance explained, and variable loadings for the five PCs used in multiple regression models. PCA includes only compounds found in V and G females













% of variance explained






Tetradecanoic acid

− 0.303



− 0.015


Hexadecanoic acid

− 0.413



− 0.219


Heptadecanoic acid

− 0.329




− 0.243

Octadecadienoic acid


− 0.216


− 0.243

− 0.170

Octadecenoic acid



− 0.240

− 0.701


Octadecanoic acid

− 0.436


− 0.251

− 0.620


Eicosanoic acid

− 0.428

− 0.043

− 0.265


− 0.151

Docosanoic acid

− 0.220

− 0.039

− 0.139

− 0.244

− 0.172

Tetracosanoic acid



− 0.066






− 0.175

− 0.172

− 0.275

Octacosanoic acid


− 0.353

− 0.070



Peak #13




− 0.081

− 0.132



− 0.307

− 0.178


− 0.564

Loadings in italics indicate that the compound was strongly associated with that PC

Orange coloration as predictors of reproductive quality

To determine if orange coloration can predict the reproductive quality of reproductively mature (V and G) females, we used multivariate regression to determine if the measurements of female color (area, value, and chroma) are correlated with any phenotypic variables (Weiss 2006). NR females were excluded from this analysis as they do not possess color ornaments. Control variables identified above were included when they met the criteria described above. Variance inflation factors were calculated to determine that multicolinearity was not an issue with any model.


Skin lipid profile

We identified 17 compounds in the derivatized lipid samples (see Table 1). The three most prevalent compounds were hexadecanoic acid, octadecanoic acid, and cholesterol. The other compounds were identified as long-chain and very long-chain fatty acids (FA) ranging from 14 to 28 carbons in size. Two 18 carbon unsaturated FA were identified as octadecenoic acid and octadecadienoic acid. The sterols cholesterol and beta-sitosterol were identified along with two putative sterols of unknown structures. The structure of one compound remains unknown altogether.

Skin lipids as indicators of receptivity

The skin lipid principal components showed a statistically significant relationship with female reproductive state (MANOVA, F(6,34) = 4.488, P = 0.026), yet only PC3 differed significantly between reproductive states (Table 2). Pairwise comparisons showed that only V and G females were significantly different (Tukey’s HSD, P = 0.035) and when plotted against PC1, V females cluster separately from G and NR females (Fig. 1). The PC3 scores of V females were not statistically different from that of NR females (Tukey’s HSD, P = 0.550) nor were the PC3 scores of G females (Tukey’s HSD, P = 0.201). Peak #13 (Table 1) was heavily and negatively factored (loading < − 0.5) into PC3 (Table 2). Since V individuals had a higher average PC3 score (Fig. 1), this indicates that they have lower proportions of peak #13 in their skin lipids.
Fig. 1

PC1 and PC3 scores derived from the relative proportions of the 12 compounds found in all 21 females. See Table 2 for PC loadings. Vitellogenic females form a cluster indicating that they have a similar skin lipid profile that is distinct from that of gravid and non-reproductive females

Skin lipids as predictors of reproductive quality

PC loadings for each skin lipid component present in V and G females only are shown in Table 3. All PCs failed to predict female coloration, SVL, body condition, mite load, and average egg mass (Table 4). Clutch size was predicted by the skin lipid profiles, yet only PC4 was significant (Table 4, Fig. 2). Only octadecanoic acid and octadecenoic acid were heavily and negatively factored into PC4 (Table 3); this means that females with larger clutches have less of these fatty acids on their skin.
Table 4

Statistics for multiple linear regression models of skin lipid profile versus phenotype. Phenotype variables were predicted by multiple predictors simultaneously. Full model statistics (F) are shown alongside β coefficients for specific predictors. P values associated with each model statistic are shown in parentheses. Values in italics are significant at P < 0.05 level

Phenotype variable

Full model statistic






Reproductive state

Body condition

Orange area

0.35 (0.87)

0.02 (0.90)

− 0.12 (0.38)

0.04 (0.81)

0.11 (0.58)

0.16 (0.51)


1.29 (0.39)

0.17 (0.55)

0.67 (0.09)

− 0.14 (0.74)

0.29 (0.51)

− 0.19 (0.82)


0.92 (0.52)

− 0.001 (0.99)

0.002 (0.99)

0.06 (0.69)

− 0.13 (0.44)

− 0.51 (0.13)

Body size

0.38 (0.85)

− 0.06 (0.86)

− 0.47 (0.86)

0.21 (0.72)

0.32 (0.60)

− 0.21 (0.78)

Body condition

1.23 (0.38)

< 0.01 (0.71)

− 0.15 (0.22)

− 0.25 (0.16)

0.13 (0.47)

− 0.23 (0.30)

Mite load

0.53 (0.75)

− 0.62 (0.58)

− 1.40 (0.29)

1.55 (0.39)

− 0.45 (0.81)

− 0.74 (0.74)

Average egg mass

10.91 (< 0.01)

− 0.01 (0.15)

− 0.01 (0.62)

− 0.001 (0.92)

− 0.02 (0.07)

− 0.01 (0.65)

− 0.13 (0.06)

Clutch size

4.17 (0.04)

0.22 (0.14)

0.22 (0.22)

− 0.08 (0.73)

0.56 (0.04)

− 0.01 (0.98)

1.14 (0.04)

Fig. 2

Partial regression plot showing the significant positive (P = 0.041) relationship between PC4 and clutch size when other explanatory variables in the multivariate model are accounted for. Table 3 shows loading values for PC4. This relationship shows that compounds factored heavily into PC4 may be capable of signaling a female’s clutch size to conspecifics

Orange coloration as a predictor of reproductive quality

Orange coloration was unable to predict female SVL, body condition, and mite load (Table 5). Chroma was found to be positively correlated with clutch size (Fig. 3), whereas value was not correlated with any phenotypic variables (Table 5). This indicates that both visual and chemical cues are capable of signaling similar information regarding female reproductive quality, despite not being directly correlated with each other (Table 4).
Table 5

Statistics of multiple regression models comparing color ornaments to phenotype. Model statistics are shown in the same fashion as Table 4

Phenotype variable

Full model statistic

Orange area



Reproductive state

Body condition

Body size

0.63 (0.62)

0.12 (0.91)

1.47 (0.20)

0.04 (0.92)

Body condition

1.02 (0.43)

− 0.09 (0.80)

0.13 (0.72)

− 0.19 (0.18)

Mite load

1.36 (0.32)

5.17 (0.12)

3.73 (0.27)

− 1.06 (0.40)

Average egg mass

8.30 (< 0.01)

− 0.01 (0.56)

0.01 (0.75)

− 0.01 (0.61)

− 0.15 (< 0.01)

Clutch size

5.26 (0.02)

− 0.92 (0.07)

− 0.37 (0.44)

0.46 (0.04)

1.61 (< 0.01)

Fig. 3

Partial regression plot showing the significant positive (P = 0.043) relationship between a female’s clutch size and the chroma of her orange throat patches. This relationship indicates that chroma is an indicator trait capable of signaling a female’s clutch size


All compounds identified in the skin lipids of S. virgatus females were previously identified in the skin lipids of female Iberian wall lizards Podarcis hispanica (Font et al. 2012). Although the identities of the compounds present in the skin lipids of these two species are similar, the relative proportions of these compounds differ substantially between them. For example, in P. hispanica, the most prevalent compound was octadecanoic acid (Font et al. 2012), whereas hexadecanoic acid was the primary component of S. virgatus skin lipids. One of the compounds identified in our samples (beta-sitosterol) is a phytosterol (Piironen et al. 2000). The presence of phytosterols in the skin of predatory reptiles has been documented before and thus our observation is not entirely unexpected (Ahern and Downing 1974; Weldon et al. 2008). Diet has been demonstrated to influence components of human skin lipids (Meinke et al. 2013) and this relationship may also exist for reptiles, but has yet to be directly investigated.

Some notable compounds were absent in the skin lipids of S. virgatus. For example, long-chain methyl ketones, which function as the attractiveness pheromone of red-sided garter snakes (Mason et al. 1989), were absent. Methyl ketones have also been identified in the skin lipids of the brown tree snake (Boiga irregularis; Jones et al. 1991), leopard gecko (Eublepharis macularius, Mason and Gutzke 1990), and Florida indigo snake (Drymarchon coarais; Ahern and Downing 1974). We also identified no squalenes or hydrocarbons in the skin lipids of S. virgatus. Hydrocarbons are a common component of reptilian skin lipids (Weldon et al. 2008) and the triterpene hydrocarbon, squalene, mediates sex recognition in the red-sided garter snake (Mason et al. 1989); this finding may be an artifact of our methodology, which was not equipped to quantify volatile compounds.

We found evidence that receptive V females possess less of an unidentified steroid in their skin lipids relative to non-receptive NR and G females. Given that a phytosterol (beta-sitosterol) was identified, it is possible that the unknown compound is also a plant sterol obtained via diet. Reproductive state is known to affect dietary behavior in captive S. virgatus, with V females eating more than G females (Weiss 2001), yet diet composition in the wild has not been examined. Given this, it is possible that reproduction-associated dietary shifts are the source of skin lipid variation. Furthermore, phytosterols have important physiological functions (Piironen et al. 2000) and diverting them to the skin to serve as pheromones may impose a cost, yet receptive (V) females do not bear this cost since they store lower levels of the compound in their skin lipids. Low levels of campesterol are known to be transported into the yolk of developing chicken eggs (Elkin 2007). A similar mechanism may be present here, whereby V females divert dietary sterols to their developing follicles, thus maintaining low skin-borne levels as an indicator of receptivity via physiological constraints. Future studies may wish to examine if NR and G females benefit from reduced male harassment via expression of chemical profiles indicative of their non-receptive state. Specific blends of triacylglycerides are known to function as suppressive sex pheromones in a species of desert-dwelling fruit fly (Chin et al. 2014). It is possible that triacylglycerides containing the fatty acids we identified may be performing a similar role; however, the precise arrangement of these fatty acids within triacylglycerides is likely important to their information content and thus must be determined before behavioral assays can confirm their signaling function.

We also found evidence linking octadecanoic acid and octadecenoic acid, a monounsaturated fatty acid, to female clutch size, a reliable metric for female phenotypic quality (Weiss 2006). Previous work has shown that scent marks (femoral pore secretions) deposited by male Carpetane wall lizards (Ibolacerta cyreni) are most attractive when they contain high levels of oleic acid ((Z)-9-octadecenoic acid; Mártin and López 2010). Contrary to the results of Mártin and López (2010), our results show that lower—not higher—levels of octadecanoic acid and octadecenoic acid correlate with higher quality phenotype. This is contradictory to theory which states that higher quality signals must be costly to produce, and the primary cost of chemical signals is diversion of compounds from metabolic functions (Mártin and López 2008). Future studies will wish to examine if in fact male S. virgatus will exhibit stronger chemosensory responses to skin lipids with lower levels of octadecenoic acid and octadecanoic acid. Should this be the case, it will also be necessary to investigate the costs of storing low levels of these compounds in skin lipids in order to understand how the honesty of this potential signal is maintained.

We found that both the levels of octadecenoic and octadecanoic acid in a female’s skin lipids and the chroma of her orange ornament correlate with her clutch size. This suggests that both chemical and visual cues in this species can potentially communicate the same information. These results fit with the “backup signals” hypothesis which states that multiple signals can remain stable when they enhance the accuracy by which the receiver assesses a single signal’s quality (Johnstone 1996). Our support for the backup signals hypothesis contradicts the notion that there is a trade-off between visual and chemical signaling in iguanid lizards (Hews et al. 2011) as we have shown that information transfer through a visual signal does not preclude the redundant use of chemical signals; however, our results do not conclusively rule out the possibility of a trade-off. Skin lipids and orange coloration do not correlate with each other despite both independently correlating with clutch size; this could be due to the correlations being driven by “visually signaling” and “chemically signaling” subsets of females, but may also be due to our small sample size.

It is also notable that our results differ from those of Weiss (2006) which found that both orange patch area and chroma correlated with a number of the phenotypic variables we analyzed here and did not correlate with clutch size. This may be the result of true annual variation in the information content of indicator traits (Chaine and Lyon 2008). Therefore, future studies may benefit from comparing female fitness correlates and male preference for female signals across years.

This study has laid the groundwork for a wide range of future studies. Most importantly, behavioral studies should be used to determine how variation in the skin lipid profile—both natural and experimental—influences the mate choice of males. Prior work typically focused on chemosensory responses to standards of compounds found in chemical ornaments (Mártin and López 2010). This is an effective method for determining which compounds in a mixture elicit chemosensory behaviors, yet studies linking response to natural variation in skin lipid composition are required to fully understand the information content and ecological roles of such complex chemical signals. Studies such as these will allow us to determine if female skin lipids are sexually selected chemical signals or simply signature mixtures that males learn to recognize (Wyatt 2014). A preliminary study on S. virgatus suggests that increasing the lipids on a female’s skin may reduce the interest of males (Gill, Goldberg, and Weiss, unpublished data), and future studies will further disentangle the relationship between chemical signal structure, female condition, and male behavior.



We thank Eric Scharrer and John Hanson for their help with chemical analysis, James Bernhard for statistical help, and Mark Martin for helpful comments on an early draft of the manuscript. We also thank all the staff and volunteers at the Southwestern Research Station.

Compliance with Ethical Standards

Ethical approval

This work was conducted under Arizona Game and Fish scientific collecting permit SP603461 and with University of Puget Sound IACUC approval (F0708-01). All applicable international, national, and/or institutional guidelines for the care and use of animals were followed. All procedures performed in studies involving animals were in accordance with the ethical standards of the institution or practice at which the studies were conducted.


  1. Ahern DG, Downing DT (1974) Skin lipids of the Florida indigo snake. Lipids 9:8–14. CrossRefPubMedGoogle Scholar
  2. Alberts AC (1990) Chemical properties of femoral gland secretions in the desert iguana Dipsosaurus dorsalis. J Chem Ecol 16:13–25.
  3. Barbosa D, Font E, Desfilis E, Carretero MA (2006) Chemically mediated species recognition in closely related Podarcis wall lizards. J Chem Ecol 32:1587–1598. CrossRefPubMedGoogle Scholar
  4. Carazo P, Font E, Desfilis E (2007) Chemosensory assessment of rival competitive ability and scent-mark function in a lizard, Podarcis hispanica. Anim Behav 74:895–902.
  5. Chaine AS, Lyon BE (2008) Adaptive plasticity in female mate choice dampens sexual selection on male ornaments in the lark bunting. Science 319:459–462. CrossRefPubMedGoogle Scholar
  6. Chin JSR, Ellis SR, Pham HT, Blanksby SJ, Mori K, Koh QL, Etges WJ, Yew JY (2014) Sex-specific triacylglycerides are widely conserved in Drosophila and mediate mating behavior. ELife 3:e01751.
  7. Dietemann V, Peeters C, Liebig J, Thivet V, Hölldobler B (2003) Cuticular hydrocarbons mediate discrimination of reproductives and nonreproductives in the ant Myrmecia gulosa. PNAS 100:10341–10346.
  8. Elkin RG (2007) Reducing shell egg cholesterol content. II. Review of approaches utilizing non-nutritive dietary factors or pharmacological agents and an examination of emerging strategies. Worlds Poult Sci J 63:5–32. CrossRefGoogle Scholar
  9. Font E, Barbosa D, Sampedro C, Carazo P (2012) Social behavior, chemical communication, and adult neurogenesis: studies of scent mark function in Podarcis wall lizards. Gen Comp Endocr 177:9–17. CrossRefPubMedGoogle Scholar
  10. Fritzche AK, Weiss SL (2012) Effect of signaler body size on the response of male striped plateau lizards (Sceloporus virgatus) to conspecific chemical cues. J Herp 46(1):79–84.
  11. Gabriot M, Castilla AM, López P, Mártin J (2010) Differences in chemical signals may explain species recognition between an island lizard, Podarcis atrata, and related mainland lizards, P. hispanica. Biochem Syst Ecol 38:521–528.
  12. Hews DK, Date P, Hara E, Castellano MJ (2011) Field presentation of male secretions alters social display in Sceloporus virgatus but not S. undulatus lizards. Behav Ecol Sociobiol 65:1403–1410.
  13. Houck LD (2009) Pheromone communication in amphibians and reptiles. Annu Rev Physiol 75:161–176. CrossRefGoogle Scholar
  14. Ichihara K, Fukubayashi Y (2010) Preparation of fatty acid methyl esters for gas-liquid chromatography. J Lipid Res 51:635–640. CrossRefPubMedPubMedCentralGoogle Scholar
  15. Johanssen BG, Jones TM (2007) The role of chemical communication in mate choice. Biol Rev 82:265–289CrossRefGoogle Scholar
  16. Johnstone RA (1996) Multiple displays in animal communication: ‘backup signals’ and ‘multiple messages’. Philos Trans R Soc Lond Ser B Biol Sci 351:329–338. CrossRefGoogle Scholar
  17. Jones T, Fales H, Mulata Y, Yeh H, Pannell L, Mason RT (1991) New ketodienes from the integumental lipids of the guam brown tree snake, Boiga irregularis. J Nat Prod 54:233–240.
  18. LeMaster MP, Mason RT (2002) Variation in a female sexual attractiveness pheromone controls male mate choice in garter snakes. J Chem Ecol 28:1269–1285. CrossRefPubMedGoogle Scholar
  19. Mártin J, Civantos E, Amo L, López P (2007) Chemical ornaments of male lizards Psammodromus algirus may reveal their parasite load and health state to females. Behav Ecol Sociobiol 62:173–179.
  20. Mártin J, López P (2008) Female sensory bias may allow honest chemical signaling by male Iberian rock lizards. Behav Ecol Sociobiol 62:1927–1934. CrossRefGoogle Scholar
  21. Mártin J, López P (2010) Condition-dependent pheromone signaling by male rock lizards: more oily scents are more attractive. Chem Senses 35(4):253–262. CrossRefPubMedGoogle Scholar
  22. Mártin J, Ortega J, López P (2013) Lipophilic compounds in femoral secretions of male collared lizards, Crotaphytus bicinctores (Iguania, Crotaphytidae). Biochem Syst Ecol 47:5–10.
  23. Martins EP, Ord TJ, Slaven J, Wright JL, Housworth EA (2006) Individual, sexual, seasonal, and temporal variation in the amount of sagebrush lizard scent marks. J Chem Ecol 32:881–893CrossRefPubMedGoogle Scholar
  24. Mason RT, Gutzke WHN (1990) Sex recognition in the leopard gecko Eublepharis macularius (Sauria: Gekkonidae) possible mediation by skin-derivied semiochemicals. J Chem Ecol 16:27–36Google Scholar
  25. Mason RT, Fales HM, Jones TH, Pannell LK, Chinn JC, Crews D (1989) Sex pheromones in snakes. Science 245:290–292CrossRefPubMedGoogle Scholar
  26. Meinke MC, Friedrich A, Tscherch K, Haag SF, Darvin ME, Vollert H, Groth N, LadeMann J, Rohn S (2013) Influence of dietary carotenoids on radical scavenging capacity of the skin and skin lipids. Eur J Pharm Biopharm 84:365–373CrossRefPubMedGoogle Scholar
  27. Moore FR, Shuker DM, Dougherty L (2016) Stress and sexual signalling: a systematic review and meta analysis. Behav Ecol 27:363–371. CrossRefGoogle Scholar
  28. Penn D, Potts WK (1998) Chemical signals and parasite-mediated sexual selection. TREE 13:391–396. PubMedGoogle Scholar
  29. Piironen V, Lindsay DG, Miettinen TA, Toivo J, Lampi A (2000) Plant sterols:biosynthesis, biological function and their importance to human nutrition. J Sci Food Agr 80:939–996CrossRefGoogle Scholar
  30. R Core Team (2012) R: language and environment for statistical computing. R Foundation for Statistical Computing, ViennaGoogle Scholar
  31. Roberts JB, Lillywhite HB (1980) Lipid barrier to water exchange in reptile epidermis. Science 207:1077–1079. CrossRefPubMedGoogle Scholar
  32. Uhrig EJ, Lutterschmidt DI, Mason RT, LeMaster MP (2012) Pheromonal mediation of intraseasonal declines in the attractivity of female red-sided garter snakes, Thamnophis sirtalis parietalis. J Chem Ecol 38:71–80.
  33. Weiss SL (2001) The effect of reproduction on food intake of a sit-and-wait foraging lizards, Sceloporus virgatus. Herpetologica 57:138–146Google Scholar
  34. Weiss SL (2002) Reproductive signals of female lizards: pattern of trait expression and male response. Ethology 108: 793–813Google Scholar
  35. Weiss SL (2006) Female-specific color is a signal of quality in the striped plateau lizard (Sceloporus virgatus). Behav Ecol 17:726–732.
  36. Weiss SL, Foerster K, Hudon J (2012) Pteridine, not carotenoid, pigments underlie the female-specific orange ornament of striped plateau lizards (Sceloporus virgatus). Comp Biochem Phys B 161:117–123. CrossRefGoogle Scholar
  37. Weiss SL, Kennedy EA, Bernhard JA (2009) Female-specific ornamentation predicts offspring quality in the striped plateau lizard, Sceloporus virgatus. Behav Ecol 20:1063–1071.
  38. Weldon PJ, Flachsbarth B, Schulz S (2008) Natural products from the integument of nonavian reptiles. Nat Prod Rep 25:738–756. CrossRefPubMedGoogle Scholar
  39. Whittaker DJ, Soini HA, Atwell JW, Hollars C, Novotny MV, Ketterson ED (2010) Songbird chemosignals: volatile compounds in preen gland secretions vary among individuals, sexes, and populations. Behav Ecol 21:608–614. CrossRefPubMedPubMedCentralGoogle Scholar
  40. Wiens JJ (1999) Phylogenetic evidence for multiple losses of a sexually selected character in phrynosomatid lizards. Proc R Soc Lond B Biol Sci 266:1529–1535. CrossRefGoogle Scholar
  41. Wyatt TD (2014) Pheromones and Animal Behavior. 2nd edition. Cambridge University PressGoogle Scholar

Copyright information

© Springer-Verlag GmbH Germany 2017

Authors and Affiliations

  • Jay K. Goldberg
    • 1
    • 2
  • Alisa K. Wallace
    • 1
  • Stacey L. Weiss
    • 1
  1. 1.Department of BiologyUniversity of Puget SoundTacomaUSA
  2. 2.Department of BiologyIndiana UniversityBloomingtonUSA

Personalised recommendations