Introduction

An animal’s pigmentation patterns influence the effectiveness of communication with conspecifics, degree of crypsis, as well as body heating and cooling rates (Cooper and Greenberg, 1992). Stripes and spots, as well as color reflectance properties that includes ultraviolet reflectance (mainly 320-400 nm), can serve as important conspecific signals to potential mates or rivals of the same sex (Endler, 1992; Andersson, 1994). How closely an individual matches its background has consequences for detection by both predators and prey (Endler, 1990). Crypticity is not only influenced by integument color, such as dark skin that facilitates matching with a dark background, but also by spotting or striping patterns that help break up the organism’s outline (Cooper and Greenberg, 1992; Cuthill et al., 2005). Pigmentation may also facilitate radiative warming in individuals with dark integuments or limit heat absorption in relatively light-colored individuals (Trullas et al., 2007).

In freshwater turtle species, pigmentation patterns seem to be related to conspecific signaling, to crypsis, and to their thermal biology. Sexual dichromatism that is presumably related to conspecific signaling between sexes is uncommon and restricted to seasonal dichromatism in the heads of some male Asian species (Moll et al., 1981), or to eye or chin color in some North American emydid species (Ernst and Lovich, 2009; Rowe et al., 2013a). That conspicuous color traits are typically restricted to males suggests a sexually selected trait whereby females select suitable males based on sufficient expression of those traits (Endler, 1990, 1992). However, why males of some species increasingly become melanistic with age while females remain relatively light in color is unclear (Schueler, 1983; Lovich et al., 1990; Stuart, 1998; Bailey et al., 2005). Among populations of some freshwater turtle species that live in environments with variable substrate colors, individuals that reside in habitats with dark substrates are relatively dark in color while those residing in habitats with light-colored substrates are light in color (Rowe at al., 2006a; McGaugh, 2008). Apparently, environmental signals elicit a phenotypically plastic process (Schlichting and Smith, 2002) whereby melanin deposition in the integument is induced by dark or light substrate colors (Rowe et al., 2006b; Rowe et al., 2013b). Color change in Painted Turtles (Chrysemys picta) apparently involves variable degrees of melanosome content of melanophores, the transfer of pigment to surrounding keratinocytes from melanocytes, or both (Albardi, 2013; Rowe et al., 2013b). Although the typically dark to nearly black dorsal pigmentation of shells and skin of the head or limbs is presumably related to crypsis, radiative warming by heliothermic species, or both, the functions of spots, blotches, or stripes are less clear. The blotching and mottling of the ventral portions of the plastron and marginal and bridge scutes of various North American pond turtles may serve a cryptic function. The yellow or red stripes of the head and limbs could conceivably facilitate crypsis against aquatic vegetation (Ross and Lovich, 1992; Cuthill et al., 2005) or could be involved with conspecific recognition in shallow water where red light is prominent (Rowe et al., 2013a). Some turtle species may have pigmentation that facilitates mimicry of vegetation (Ross and Lovich, 1992) or even noxious species (Janzen, 1980).

The North American Painted Turtle (Chrysemys picta) is an interesting species for the study of pigmentation patterns in a freshwater Emydid species. Four subspecies of C. picta are recognized, based in part, on differences in pigmentation (Ernst and Lovich, 2009). The head has a brown to black ground color with yellow spots, stripes, or both, that are posterior to the eye and mouth depending on the subspecies while the limbs are dark brown or black in ground color with a series of longitudinal red stripes (Ernst and Lovich, 2009). The ground color of the carapace is green to black and may have light cream or orange lateral or longitudinal stripes, and the ventral regions of the marginal scutes have alternating black and red blotches. Because courtship of C. picta occurs “face-to-face” (Ernst, 1971), the accentuation of colored features at the cranial end of a turtle by UV reflectance such that they are more conspicuous to conspecifics (Cooper and Greenberg, 1992) might be expected. Pond turtles possess UV sensitive retinal cones (Loew and Govardovskii, 2001) and UV reflectance of sexually selected traits in males of some reptiles, such as lizards, is common (LeBas and Marshall, 2000). While sexual dimorphisms in size and morphometry are obvious in C. picta (Jolicoeur and Mossimann, 1960), the only known intersexual difference in pigmentation (dichromatism) in Midland Painted Turtles occurs in some relatively old males that develop reticulate melanization (Smith et al., 1969; Ultsch, 1999; Gronke et al., 2006). Substrate color-induced melanization occurs in juvenile Midland Painted Turtles (C. p. marginata) and may explain why turtles that reside in dark-bottomed habitats have dark integuments while those that reside in light-bottomed habitats have relatively light integuments thus promoting crypsis in their respective environments (Rowe et al., 2006a, b, 2009). While substrate color-induced melanization is reversible in juvenile C. picta marginata (Rowe et al., 2009), the degree to which substrate color-induced melanism is manifested in adults, that are likely to travel among habitats with different substrate types and colors, is unknown (Rowe et al., 2009).

We studied spectral reflectance and substrate color-induced melanization in adult C. picta marginata from a population in central Michigan. We measured spectral reflectance in adult C. picta marginata that were collected from a natural population and then placed on a black or a white substrate for 150 days. In terms of reflectance of wild caught individuals, we expected that turtles would show peak reflectance in the spectral regions that are specific for the brighter colors (red-orange: 400-510 nm and yellow: 570-590 nm) with little reflectance and intensity (overall brightness) in the darker regions (dark green, brown or black) of the head, limbs, and carapace. Since the ground color of the dorsal head skin and carapace show little to no UV reflectance in hatchling and lab-reared juveniles (Rowe et al., 2006b, 2009), we did not expect substantial UV reflectance in these regions of wild caught turtles. However, accentuation of the integument by UV reflectance in the red, yellow, and even black markings of the turtle’s cranial end could occur (Rowe et al., 2013a). Hatchling Chrysemys picta marginata maintained on a white substrate for extended periods tend to lighten over time while those maintained on a dark substrate show substantial darkening (Rowe et al., 2006b, 2009). If color change is a phenotypically plastic trait in older individuals as it is in juveniles (Rowe et al., 2006b, 2009), then ground color (darker background color) of the carapace, head, and limbs of turtles reared on a white substrate would be lighter than in turtles reared on a black substrate. Relative to turtles reared on a white substrate, ground colors of the soft skin, carapace, and relatively dark markings of the shell (e.g. blotches of the marginal scutes) of turtles reared on black substrates would be expected to be dark as the result of melanocyte and melanophore activity (Alibardi, 2013; Rowe et al., 2013b). The red, orange, or yellow regions of the skin are determined by carotenoids or pterins in xanthophores and lipophores (Alibardi, 2013). Since red or yellow regions contain some dermal melanophores but few melanocytes (Alibardi, 2013), we did not expect that turtles reared on a black substrate would be substantially darker than in individuals reared on a white substrate.

Materials and methods

Experimental design and turtle husbandry

We randomly assigned turtles to either black ($n=33$) or white ($n=35$) substrate color treatments and of these, 30 individuals survived through the experiment in the black substrate treatment (F: $n=17$ and M: $n=13$) and 32 survived in the white substrate treatment (F: $n=19$ and M: $n=13$). Turtles were maintained in opaque plastic bins (5 l × 34 × 30 cm) with the external bottoms painted to 25 cm on the sides using either semi-gloss black or white paint. Each bin contained two to three turtles and a total of 25 bins were arranged in two rows of 12 or 13 bins per row. Approximately 10 l of water were added to each bin and water was conditioned using Prime^® (Seachem Laboratories Inc., Madison, GA, USA) and changed as needed. Each bin was equipped with two water filters (Duetto 100, Marineland United Pet Group, Blacksburg, VA, USA), and water heater (Visi-Therm Deluxe Submersible Heater, 150 W, Marineland United Pet Group, Blacksburg, VA, USA) set at 28°C. We illuminated bins with 34 watt Philips© Alto full spectrum fluorescent lights (Eindhoven, Netherlands) set at 12:12 L:D light cycles and that we positioned 20 cm above each container and each bin was equipped with black or white basking bricks (19 × 9 × 3.5 cm). Algae was scrubbed from the inside surfaces of the containers when observed. Turtles were fed to satiation five days per week using an alternating diet of ground beef heart and Reptomin^® (Spectrum Brands, Inc., Cincinnati, OH, USA).

Reflectance spectra of soft integument areas included dorsal head skin, the lateral head stripe located just posterior to the eye, the dark ground color of the forelimb, and the red stripe of the forelimb. For the shell, we obtained reflectance spectra of the third vertebral scute, the black and red blotches of the ventral surfaces of the marginal scutes, and the lateral, light-colored regions of the abdominal scutes of the plastron. Reflectance spectra were obtained on days 0, 30, 60, 90, 120, and 150. We used a reflectance probe (Ocean Optics R-400) connected to a deuterium-halogen lamp (Ocean Optics) of an Ocean Optics USB 2000 portable spectrometer, and a notebook computer running the SpectraSuite software program (Ocean Optics, Inc.). Subjects were placed at a fixed distance of 0.5 cm from the probe and reflectance of the integument surface was sampled at 30° from the perpendicular. A white standard (Labsphere Spectral WS-1) was scanned and dark current removed from the signal immediately before a reading was taken. Three measurements were taken per location per turtle and values averaged per individual per body region. If mortality precluded intensity measurements beyond day 0, the individual was entirely eliminated from the data set. Our data included wavelengths that included the visible and ultraviolet spectra (320-700 nm).

We analyzed intersexual differences in color by use of intensity and hue as determined by CLR analysis (Montgomerie, R., 2008; CLR, Color Analysis Programs, version 1.05. Queen’s University, Kingston, Canada; available at http://post.queensu.ca/~mont/color/analyze.html). Using CLR, spectral data were extracted from raw SpectraSuite files in 1 nm increments over the range of 320-700 nm. These extracted spectral files were then de-spiked in the range of 650-665 nm to remove the deuterium spike. To produce spectral reflectance curves and additional analyses, de-spiked spectral files were averaged over the three samples per body region measured per individual. For estimating hue, the de-spiked spectral files were processed further with CLR Var program to extract hue measurements and then averaged per body region per individual. Intensity, a dimensionless measure of brightness (Cooper and Greenberg, 1992), was measured as the total area under the spectral curve between 320-700 nm where values were extracted at 20 nm increments (Macedonia et al., 2003; Rowe et al., 2006b). For descriptive purposes, hue (spectral location) was determined as the wavelength at peak reflectance ($\lambda R_{\max }$) as it correlates well with the color perception of humans (Andersson et al., 1998).

View Full Size

Reflectance spectra of the head, limbs, and shell of wild caught C. p. marginata ($n=68$).

Citation: Amphibia-Reptilia 35, 2 (2014) ; 10.1163/15685381-00002934

• Download Figure
• Download figure as PowerPoint slide

Statistical analyses

Potential intersexual differences in intensity of the eight measured body regions were assessed by MANOVA. Because hue was not normally distributed, we conducted Wilcoxon rank sum tests for each of the eight measured body regions with Bonferroni adjustment of alpha to 0.006. Changes in intensities of turtles that were maintained on black or white substrates for 150 days were assessed using a series of ANOVAs on the intensities of the four regions of the carapace and of the four regions of the head and neck. Each model included substrate color (black or white substrate), day-of-measurement (days 0-150), sex, and all interaction terms. Turtle identification number was included in the models as a random variable to account for autocorrelation that would occur from multiple measurements on the same individual. Least squares means multiple t-tests adjusted for all main and interaction effects were used for post hoc comparisons of means.

View Full Size

Mean (±1SE) peak reflectance (hue) for various body regions of female ($n=36$) and male ($n=26$) C. picta marginata.

Citation: Amphibia-Reptilia 35, 2 (2014) ; 10.1163/15685381-00002934

• Download Figure
• Download figure as PowerPoint slide

Results

Spectral reflectance of wild-caught turtles

The ground colors of the soft integument regions and of the shell were nearly black or dark brown-green and provided a contrasting background for stripes and blotches and particularly for the yellow head stripe. The ground color of the soft skin (dorsum of the head and the legs) showed little reflectance across the UV and visible spectra indicating a nearly black coloration (fig. 1). The yellow head stripe, with peak reflectance in the 600-650 nm range (fig. 2), was the brightest region of the skin that we measured followed by the red stripe of the leg and both regions showed a small degree of reflectance in the UV (fig. 1). The ground color of the carapace (third vertebral scute) showed little reflectance across the UV and visible spectra with slightly elevated reflectance in the 500-700 nm region indicating somewhat more color (brown-green) than in the head and legs. For the marginal scutes, the black blotches showed low and relatively even reflectance across the measured spectrum while the red blotches showed little reflectance except in the 550-700 nm range (fig. 1) where peak reflectance occurred (fig. 2). The plastron, with peak reflectance at about 615 nm (fig. 2), had a small degree of UV reflectance and relatively uniform but moderately high reflectance across the measured spectra indicating a somewhat gray color (fig. 1).

There were little to no intersexual differences in brightness or color quality of any body region in wild caught turtles. The F statistic in the MANOVA of intensity that included all eight dependent variables (shell and soft skin regions) and with sex as an independent variable was not significant ($F=1.3$, df = 7, 54, $P=0.2486$). Following adjustment of alpha to 0.006 for eight comparisons, Wilcoxon rank sum tests of hue with sex as independent variable were not significant for most body regions ($P>0.006$) although the red stripe of the forelimb was significantly different between females and males ($X^2=10.1$, df = 1, $P=0.002$). The wavelength at peak reflectance was higher in males (mean = 675.8 ± 3.1 nm, $n=26$) than in females (mean = 661.5 ± 2.6 nm, $n=36$) indicating that the leg stripes of males were slightly redder than in those of females.

Table 1.

Analyses of variance of reflectance intensities of four head and limb locations in Chrysemys picta that were reared on either black or white substrates and measured at 30 day intervals over six months.

View Table

Table 2.

Analyses of variance of reflectance intensities of four shell locations in Chrysemys picta that were reared on either black or white substrates and measured at 30 day intervals over six months.

View Table

Effects of prolonged exposure to black or white substrates

The head and limbs tended to lighten over time but generally, more lightening occurred in turtles that were reared on a white substrate when compared to those reared on a black substrate. Analysis of variance of dorsal head skin ground color intensity (table 1) revealed significant effects of substrate color, day-of-measurement, and substrate color × day-of-measurement. Overall, mean intensity of the head ground color was greater in turtles reared on a white substrate (mean = 12.4 ± 0.4, $n=32$) than in those reared on a black substrate (mean = 7.8 ± 0.4, $n=30$). The significant day-of-measurement term and LS mean multiple t-tests post hoc comparisons indicated that, overall, head skin ground color of turtles maintained on both black and white substrates increased in intensity (became lighter) between day 0 and 30 and between days 90 and 120 ($P<0.05$ in both tests; table 1 and fig. 3). The significant substrate color × day-of-measurement term (table 1) and post hoc comparisons indicated that turtles reared on white substrates followed the overall pattern for all turtles combined but that turtles on a black substrate increased in mean intensity only between days 30 and 60. Overall, the ground color of the forelimb in turtles that were maintained on a black substrate was significantly darker (mean intensity = 11.2 ± 1.0) than in those maintained on a white substrate (mean intensity = 20.7 ± 1.0; table 1). Over time, mean intensity of leg ground color of all turtles changed in a similar pattern to that of the head ground color (table 1 and fig. 3). Ground color of the leg of turtles that were maintained on a white substrate increased in intensity between each 30 d interval except between days 60 and 90 while those reared on a black substrate did not change significantly between days 0 and 150. In the ANOVA of yellow head stripe intensity, the day-of-measurement term was significant (table 1) and post hoc comparisons indicated a significant increase in intensity between days 90 and 120. Changes in mean intensity of the red leg stripe over time paralleled changes that occurred for the yellow head stripe (table 1 and fig. 3).

View Full Size

Mean (±1SE) intensities of soft integument regions of C. p. marginata reared on either black ($n=30$) or white ($n=32$) substrates for 150 days.

Citation: Amphibia-Reptilia 35, 2 (2014) ; 10.1163/15685381-00002934

• Download Figure
• Download figure as PowerPoint slide

The ground color of shells and black blotches of the marginal scutes of turtles that were maintained on a black substrate were darker than in those maintained on a white substrate but other shell tissue regions generally darkened over time. In the ANOVA of intensity of the third vertebral scute (TVS), significant effects of substrate color, day-of-measurement, and substrate color × day-of-measurement terms were observed (table 2). Turtles maintained on a black substrate tended to have a slightly darker carapace (mean intensity = 7.8 ± 0.3) than did turtles maintained on a white substrate (mean intensity = 9.3 ± 0.3). Overall, mean intensity of the TVS was lower on day 30 relative to other days of measurement but this was due to an initial darkening of individuals in the black substrate treatment (table 2 and fig. 4). On day 60, mean intensity of the TVS of turtles reared on a black substrate remained constant although turtles maintained on a white substrate slowly increased in intensity such that the mean value was significantly greater on days 120 and 150 than on days 0-90 (fig. 4). For the black blotch on the marginal scute, substrate color, day-of-measurement, and sex terms were significant (table 2). Turtles maintained on a black substrate (mean = 11.3 ± 0.5) had significantly darker black blotches than did turtles that were maintained on a white substrate (mean = 13.7 ± 0.5), and overall, mean intensity was higher on days 120 and 150 than on days 0-90 ($P<0.05$ in all comparisons). Males had slightly, but significantly, darker black blotches of the marginal scutes (mean = 11.5 ± 0.5) than did females (mean = 13.6 ± 0.5; table 2). Intensity of the red blotch of the marginal scutes was not affected by substrate color on which the turtles were maintained (table 2) but generally decreased between days 0 and 90 and then increased between days 90 and 120. Intensity of the plastron was not affected by substrate color or by sex but significantly decreased between days 30 and 60 and then increased again between days 90 and 120 (table 2, fig. 4, $P<0.05$ in all post hoc comparisons). The increases in intensity that occurred between days 90 and 120, particularly in the marginal scutes, was associated with an increase in the degree of opacity that we believe marked the initiation of the scute shedding process rather than in changes in chromatophore activity.

View Full Size

Mean (±1SE) intensities of shell integument regions of C. p. marginata reared on either black ($n=30$) or white ($n=32$) substrates for 150 d.

Citation: Amphibia-Reptilia 35, 2 (2014) ; 10.1163/15685381-00002934

• Download Figure
• Download figure as PowerPoint slide

Discussion

In general, we found few intersexual differences in color regions of the head, extremities, or shell in C. picta marginata. We did not find that the ground color of the carapace or plastron, or the black or red blotches of the ventral marginal scutes, varied between sexes. It seems likely then, that the ground colors of the carapace and plastron facilitate background matching through countershading (Rowland et al., 2008). The dark carapace would converge in color with the dark bottom substrate when viewed by a predator from above while the lighter plastron would converge in color with the sunlit sky when viewed from below. The black and red blotches of the marginal scutes might function to break up the outline of the shell when viewed from below, particularly against the irregular shapes of aquatic vegetation. Initially, we reasoned that the stripes of the head and forelimbs in males could function during courtship as males court females from an anterior direction. However, we found no, or extremely minor, intersexual differences in head or forelimb and visible or UV reflectance and so we suggest that the pigmentation patterns of the head and forelimbs could be involved with cryptic coloration. Indeed, the black, red, and yellow pigmentation of the hind limbs, which are similar to the pigmentation of the anterior regions of the turtle, would not be expected to be involved with courtship in a species where males face females. Similarly, except for dichromatic chins, no intersexual differences in ground color or spots were found in Spotted Turtles (Clemmys guttata) suggesting that spots of the head might function in crypsis rather than in communication (Ross and Lovich, 1992; Rowe et al., 2013b). Bulté et al. (2013) found that the post-orbital spots of adult male Northern Map Turtles (Graptemys geographica) were brighter than in adult females and argued that the post-orbital spot may be a sexually selected trait. However, because adult male and juvenile female G. geographica had similar bright post-orbital spots that were not correlated with testosterone levels (Bulté et al., 2013), it may be that adult female spots simply become duller as they age and grow.

The yellow and red color markings in the shells, head, and extremities of C. picta marginata could function in recognition of conspecifics. The retina of the closely related T. scripta is highly sensitive to red wavelengths of light (Granda, 1979; Loew and Govardovskii, 2001) that are prominent in shallow water habitats and it is possible that C. picta marginata has similar retinal sensitivities. Still yet, the quality of carotenoid-based conspicuous head markings in freshwater turtles, including T. scripta elegans, may convey information to conspecifics regarding mate quality because color expression and immune function are positively correlated (Polo-Cavia et al., 2012; Ibáñez et al., 2013). It seems increasingly likely that pigmentation patterns of freshwater turtle integuments serve multiple functions that include crypsis, thermoregulation, and conspecific signaling that is involved with species recognition and mate choice.

We found that maintaining adult C. picta marginata on different substrate colors significantly affected ground color as well as stripes and blotches of the shell, head, and extremities. The ability for a turtle to either darken or lighten to more closely match its background would presumably reduce predation rates, particularly in juveniles that would encounter such predators as wading birds and fish. Following emergence from a terrestrial nest, hatchling turtles could encounter a number of aquatic habitats with bottom substrates that range from light-colored sand or algae to black muck-bottoms (Rowe et al., 2006a, 2009). As adults, color convergence of the carapace with a background might reduce predation as individuals move among aquatic habitats with variable substrate colors throughout their lives (Rowe et al., 2009) in response to drought (Gibbons et al., 1983), or during the course of seasonal migrations (McAuliffe, 1978; Bodie and Semlitsch, 2000). However, throughout most of its range, aquatic predators that would be large enough to consume adult C. picta are likely to be rare and so the adaptive value of substrate color convergence may be rather low for adults. Substrate color-induced melanization in adults may be a simple hormonal reflex that involves MSH or epinephrine (Alibardi, 2013) that may not have survival consequences for adults.

While change in the brightness of the ground color of the carapace and soft skin regions would presumably facilitate background color convergence, we found that the brightness of the stripes and blotches was also influenced by exposure to different substrate colors over time. Since epidermal melanocytes are scarce or absent in yellow or red stripe regions (Alibardi, 2013), color change in those areas would occur through changes in the content and distribution of melanin within the dermal melanophores (Woolley, 1957). It remains unclear, however, if darkening or lightening of yellow or red stripes or blotches would enhance the degree of background matching. It is possible that color changes in light-colored skin regions occur coincidentally with the changes in the ground color regions of C. picta. Alibardi (2013) concluded that pigmentation of dark spots and stripes of C. picta occur mainly through transfer of melanosomes from melanocytes to keratinocytes early in the process of new scute production prior to the shedding of an older scute. The mechanisms that underlie darkening of the skin involve melanosome transfer to keratinocytes by melanocytes (morphological change), increased melanin content within dermal melanophores (physiological change), or both processes (Wasmeier et al., 2008; Albardi, 2013; Rowe et al., 2013b). However, since color change does not require scute shedding in juvenile turtles (Rowe et al., 2009) or in adults (this study), there must be some mechanism whereby melanosomes within superficial keratinocytes are broken down during lightening of the integument (Rowe et al., 2013b). Our results here indicate that the significance of melanism as it changes with age, varies within and between sexes, in turtles such as C. picta and T. scripta (Lovich and Garstka, 1990) should be interpreted in the context of environmental influences.