The capacity for the integuments of organisms to match their background, which facilitates the avoidance of detection by predators and prey, can be attained through various cellular mechanisms (Endler 1990; Cooper and Greenberg 1992). In reptiles, dermal chromatophores form a range of colors and mottling of the integument (Cooper and Greenberg 1992). In physiological color change, the distribution of pigments within chromatophores results in temporal variation in color that includes hue (color quality or the peak wavelength reflected), saturation (degree of peak reflectance of the dominant wavelength), and brightness (lightness or darkness) of the integument (Cooper and Greenberg 1992). Dermal melanophores are largely responsible for the brightness of an individual through the distribution of melanosomes within cells. Melanosomes are large vesicles that contain the tyrosine derivative melanin and are easily viewed by light microscopy (Wasmeier et al. 2008). Many lizards show physiological color change within minutes by concentrating melanosomes near the nucleus of the melanophore, thereby lightening the integument, or by transporting melanosomes into dendritic processes of the cell, thereby darkening the integument (Taylor and Hadley 1970). In other reptile groups, such as chelonians, color change occurs relatively slowly over several days, weeks, or years (Woolley 1957; Moll et al. 1981; Banks 1986; Lovich et al. 1990; Rowe et al. 2009). Such slow color change is often morphological where darkening involves the concentration of intracellular melanosomes in dermal melanophores (Lovich et al. 1990) or dermal melanophores deposit melanosomes in the more superficial epidermal cells (Moll et al. 1981).

In turtles, the degree of melanization may be established early in life, gradually increase with age, or change depending on reproductive cycles or environmental conditions (Lovich et al. 1990; Hays and McBee 2009). Over several months of observations of Chelodina longicollis in the laboratory, Woolley (1957) concluded that melanization involved a slow physiological color change through the distribution of melanophores in cytoplasmic cell extensions. In constrast, melanistic and nonmelanistic Trachemys scripta (Lovich et al. 1990) appear to differ in their melanophore distribution in the dermis and epidermis. In the Asian river turtle (Callagur borneoensis), Moll et al. (1981) found that, when compared with light-phase individuals, dark-phase individuals had relatively high concentrations of melanosomes in the strata germinativum, spinosum, and corneum. Similarly, reticulate melanism in aging male Chrysemys involves deposition of melanosomes in the epidermal scutes (Smith et al. 1969). Therefore, the melanophores of some freshwater turtles may behave like mammalian melanocytes in that they are involved with intercellular deposition of melanosomes (Wasmeier et al. 2008).

In some freshwater turtles, the degree of melanization may facilitate background matching that could minimize predation rates. Freshwater turtles that reside in dark-bottomed or in light-bottomed habitats may show dark or light colored integuments, respectively (Rowe et al. 2006a; McGaugh 2008). Substrate color-induced melanization has been shown in distantly related Cryptodire and Pleurodire species (Woolley 1957; Banks 1986; Rowe et al. 2006b). In laboratory studies, where individual Midland painted turtles (Chrysemys picta marginata) and red-eared sliders (Trachemys scripta elegans) were reared on dark or light substrates, significant substrate color-induced melanization occurred within 30 d (Rowe et al. 2006b, 2009). Melanization is reversible within several weeks in laboratory-reared C. picta marginata and T. scripta elegans and does not involve the shedding of epidermal scutes, the tissue structures that could contain many of the epidermal melanosomes. Reversal of melanization may simply involve retraction of melanosomes from dendritic processes to a perinuclear location (Woolley 1957) or by a decrease in melanophore number or melanin content as occurs in some teleost fish (Sugimoto 2002). However, species such as Chrysemys picta and T. scripta apparently do not have melanophores with extensive dendritic processes (Lovich et al. 1990; Alibardi 2002); therefore, color change likely involves changes in intracellular melanosome concentrations within melanophores and epidermal keratinocytes (Hoekstra 2006; Rowe et al. 2009). For melanization to be reversed in C. picta and T. scripta, melanosomes would have to be degraded, presumably in the living cell regions such as the stratum germinativum or the stratum spinosum.

We studied the histological basis of color change in C. picta marginata under laboratory conditions. Hatchlings (individuals that have recently emerged from the egg shell) were reared on either black or white substrates for 120 d (Rowe et al. 2006b) or on black or white substrates for 160 d (controls), or with a reversal in substrates (black to white or white to black) at 80 d and then maintained on the new substrates for another 80 d (Rowe et al. 2009). From histological preparations, we quantified relative melanin levels in the dermal melanophore layer and in the transitional stratum spinosum and superficial-most stratum corneum of the epidermis. If color change occurred by variation in melanosome content or dispersion within melanophores, then pixel counts of the melanophore layer in dark substrate-reared individuals should exceed those in individuals reared on a white substrate. However, mean pixel counts of the epidermal layers should not vary between groups that were reared on black or on white substrates. If color change involved variation in melanosome content of the epidermis, then mean pixel counts of the epidermal layers should be higher in turtles that were reared on a black substrate than in those that were reared on a white substrate. Given that melanization is reversible, however, we expected that most epidermal color change would be in the living stratum spinosum where melanosomes could be actively degraded.

METHODS

Tissue Collection and Histological Techniques

The source of hatchlings, maintenance, rearing conditions, and experimental treatments of the C. picta marginata used in the present study has been previously described (Rowe et al. 2006b, 2009). We obtained histological tail-tip preparations from hatchlings and in older turtles at the termination of each study and measured the degree of tissue darkness by pixel counts from digital photos. Tail-tip sections were used rather than shell or other skin regions to facilitate ease of tissue collection and preparation. We obtained tissue samples from individuals that were randomly selected from the various treatment groups (120-d experiment: n  =  10 hatchlings and n  =  6 for each of the white substrate and black substrate groups; 160-d experiment: n  =  5 for each of the white substrate control, black substrate control, white-to-black substrate reversal, and black-to-white substrate reversal groups). Each individual selected was derived from a different clutch of eggs to avoid maternal effects. We sampled tail tips from hatchlings to evaluate the initial pigmentation levels of each tissue layer. Some histological preparations could not be used because of incomplete decalcification of caudal vertebrae that led to fragmentation of the sample during sectioning. Therefore, sample sizes varied across treatments within experiments.

The animals were anesthetized on ice, and 2–3 mm of the tips of their tails were removed with a razor blade. Excised tail tips were fixed in buffered neutral formalin and decalcified with EDTA. After dehydration in graded concentrations of ethanol and clearing in toluene, the tissues were embedded in paraplast. Serial sections were cut at 6 µm and mounted on poly-L-lysine–coated glass slides. Following removal of the paraplast and rehydration of the tissues, some sections were stained with hematoxylin and eosin to clearly delineate cell layers, whereas other sections were not stained for the quantification of melanin. Digital images of stained and unstained sections were obtained under standardized lighting conditions with a Moticam 1000 digital camera attached to a Leica DLB microscope (Wetzler, Germany) using ×200 magnification.

Quantification of Melanin

Each image was white-balanced and converted to a binary image using ImageJ (National Institutes of Health, Bethesda, MD). In cross-section, dorsal, ventral, and lateral scale regions of the tail tips were visible. We sampled pixel counts of the lateral scales, which were continuous with, and similar to, the relatively dark ground colors of other skin and shell regions. For each lateral scale region, we determined pixel counts within 6, 50 × 50 pixel squares (2500 square pixel regions) across the melanophore region, the stratum spinosum, and the stratum corneum layers of the epidermis (Fig. 1A). We used ImageJ software to count the number of dark pixels in each square.

View Figure

• Download as Powerpoint
• Download Figure

Statistical Analyses

Pixel counts were approximately normally distributed; therefore, we used analyses of variance (ANOVAs) with mixed effects to evaluate pixel counts among treatment groups. In the ANOVA for which turtles were reared for 120 d, our main effects were substrate color exposure (hatchlings [no substrate exposure], white substrate, or black substrate) and tissue layer (melanophore layer, stratum spinosum, and stratum corneum). For the substrate color reversal experiment, our ANOVA included substrate color exposure (black control, white control, black-to-white reversal, or white-to-black reversal), tissue layer, and their interaction. Therefore, each individual was represented in only a single substrate color treatment, but each individual had 3 values in the data set that corresponded with each of the 3 tissue layers. In both ANOVAs, we included turtle identification number as a random variable to account for autocorrelation. Post hoc comparisons were made using least squares (LS) means multiple t-tests. Test statistics were considered significant at the 95% significance level, and mean values are reported with ± 1 SE.

RESULTS

The melanophore layer was readily identifiable although melanophores were scattered throughout the dermis (Fig. 1). We did not observe melanophores within the epidermis, but some portions of the melanophores appeared to intermingle with the deepest epidermal keratinocytes. Melanosomes were apparent within the keratinocytes of the stratum germinativum, stratum spinosum, and stratum corneum (Fig. 1).

Hatchlings and Individuals Reared on Black or White Substrates

For hatchlings and older turtles that were reared on black or white substrates for 120 d, pixel counts varied among substrate color exposure treatments, among the cell layers, and among cell layers in different substrate color exposure treatments. Overall, older turtles reared on a black substrate had higher mean pixel counts than did hatchlings and older turtles that were reared on a white substrate (Table 1). LS mean multiple t-tests indicated that mean pixel counts of older turtles reared on a black substrate (LS mean  =  657.1 ± 77.7 pixels) were significantly greater than in hatchlings (LS mean  =  414.4 ± 57.9 pixels, p < 0.05) and in older turtles that were reared on a white substrate (LS mean  =  270.4 ± 70.9 pixels; p < 0.05 in both comparison). For individuals combined across all treatments, mean pixel counts varied significantly among all cell layers (Table 1) where the melanophore layer had a higher mean pixel count (LS mean  =  968.2 ± 52.2 pixels) than either the stratum corneum (LS mean 264.5 ± 52.2 pixels, p < 0.05) or the stratum spinosum (LS mean  =  109.6 ± 52.2 pixels, p < 0.05). The mean pixel count of the stratum corneum was significantly greater than in the stratum spinosum (p < 0.05) indicating that the stratum corneum was intermediate in darkness between the melanophore layer and stratum spinosum. The substrate color exposure × cell layer term was significant (Table 1) indicating that there was significant variation in pixel counts within and among substrate colors and cell layers. LS mean pixel counts of the melanophore cell layer of hatchlings, and of older turtles that were reared on a black substrate were significantly greater than in turtles that were reared on a white substrate (p < 0.05 in both comparisons; Figs. 1A–D and 2A). LS mean pixel count of the stratum corneum of turtles that were reared on a black substrate were significantly greater than in the stratum corneum layers of both hatchlings (p < 0.05) and of older turtles that were reared on a white substrate (p < 0.05; Fig. 2A). There were no significant differences in mean pixel count of the stratum corneum among hatchlings and older turtles that were reared on black or white substrates (p > 0.05 in all 3 post hoc comparisons; Fig. 2A).

View Figure

• Download as Powerpoint
• Download Figure

Table 1. Analyses of variance for pixel counts where main effects were substrate color exposure (hatchling [no exposure], turtles reared on a black or white substrate for 120 d) and cell layer (melanophore layer, stratum spinosum, and stratum corneum) and for the “substrate reversal” experiment, main effects were substrate color exposure (rearing on a single substrate color for 160 d [black or white controls] or on a white or black substrate for 80 d with subsequent reversal to black or white substrates for an additional 80 d) and cell layer.

View Fullscreen

Substrate Color Reversal Experiment

Variation in color of turtles that were reared on black or white substrates for 160 d and in turtles that were initially reared on a black or white substrate for 80 d and then placed on a white or black substrate, respectively, for 80 d, involved variation in melanin content of the melanophore, stratum spinosum, and stratum corneum layers. All terms in the ANOVA were significant (Table 1). Both the black substrate control and white-to-black substrate reversal groups had higher mean pixel count (LS mean  =  511.2 ± 68.6 pixels and LS mean  =  563.8 ± 68.6 pixels, respectively) than did the white substrate control and black-to-white substrate reversal groups (LS mean  =  278.1 ± 68.6 pixels and LS mean  =  235.5 ± 68.6 pixels, respectively; p < 0.05 in all 4 comparisons). Regardless of treatment, mean pixel count of the melanophore layer (LS mean  =  818.9 ± 39.7 pixels) was significantly greater than mean pixel count of the stratum corneum (LS mean  =  254.9 ± 39.7 pixels, p < 0.05) that was in turn significantly greater than mean pixel count of the stratum spinosum (LS mean  =  117.6 ± 39.7 pixels, p < 0.0001).

The substrate color exposure × cell layer term of the ANOVA was significant (Table 1) and post hoc comparisons revealed that reversal of substrate color at 80 d caused marked decreases or increases in melanin content of all 3 cell layers. LS mean pixel count of the melanophore layer of the black substrate control group was significantly greater than in the white substrate control and black-to-white substrate reversal groups (p < 0.05 in both comparisons; Fig. 2B). In terms of the stratum spinosum, mean pixel count of the white-to-black substrate reversal group was significantly greater than that in the white control and black-to-white substrate reversal groups (p < 0.05 in all 3 comparisons; Fig. 2B). Mean pixel count of the stratum corneum layer of the black control and white-to-black substrate reversal groups were significantly greater than mean pixel count of the stratum corneum of the white control and black-to white substrate reversal groups (p < 0.05 in all 4 comparisons; Fig 2B).

DISCUSSION

We determined that color change of the integuments of C. picta marginata occurred through intracellular melanosome concentrations within the living melanophore and stratum spinosum as well as within the nonliving stratum corneum. As hatchlings, tail-tip scale coloration was mainly attributable to pigments within the melanophore layer. In older individuals that were reared on different colored substrates for several months, the densities of melanosomes were relatively high in the melanophore layer, but the relative contributions of epidermal cell layers to overall darkness varied between and among substrate color treatments. In our study, the melanophore layer appeared to be a densely packed dermal layer of cells with cell extensions within both the dermis and epidermis. Our histological profile of C. picta was concordant with that of T. scripta (Lovich et al. 1990) in that the amount of pigmentation was in part related to the degree of melanin within the melanophore layer. In contrast to our study, Lovich et al. (1990) did not describe dermal melanophore cell extensions in the epidermis but did conclude that melanophores were located in the epidermis. In general, epidermal melanophores seem to be uncommon in turtle integuments and in reptile integuments in general (Cooper and Greenberg 1992) but are known in crocodilians (Alibardi and Thompson 2000).

Clearly, because melanophores produce the melanosomes that are eventually transferred to the more superficial keratinocytes (Wasmeier et al. 2008), some of the pigmentation difference between turtles reared on black or white substrates in our study was attributable to variation in intracellular melanin content of melanophores. Using our light transmission histological techniques, however, we could not discern individual melanophores nor quantify extensive networks of dendritic processes (Woolley 1957). Therefore, we cannot discount color variation that was attributable to differential dispersion of melanosomes within melanophores (Taylor and Hadley 1970; Wasmeier et al. 2008). Changes in the numbers of melanophores over time appear to be common in teleost fish (Sugimoto 2002) and, although apparently rare in reptiles, could have occurred in the turtles of our study.

Our prediction that color change within the epidermal layers of C. picta would involve mainly the deeper, living regions of the integument was not supported. We reasoned that the living keratinocytes would be capable of retaining or breaking down melanosomes. However, melanosomes were ubiquitous throughout the epidermis. In fact, changes in melanin content of the most superficial layer, the stratum corneum, had a major effect on color change. Although it is clear that much of the color change in C. picta involved epidermal melanosomes, the intercellular and intracellular mechanisms of color change remain to be determined. As turtles darkened, melanosomes could have been directly transferred from melanophores to the cells of the stratum germinativum (Wasmeier et al. 2008). Alternatively, extracellular channels that penetrate the epidermis could conduct melanosomes to more superficial epidermal layers (Wasmeier et al. 2008), a cellular arrangement and process that likely explain melanosome dispersion in Callagur borneoensis (Moll et al. 1981). Conversely, turtles reared on a white substrate would have low melanosome production levels. During the reversal of melanization, melanosome breakdown in living epidermal cells, coupled with reduced melanosome production, might occur as the turtle lightens. In the carapace of the shell, however, darkening and lightening occur without scute-shedding (Rowe et al. 2009), and we assume that a similar situation exists in the scales of the tail. Therefore, melanosomes would have to penetrate a previously established stratum corneum in a turtle that darkens and be broken down in the stratum corneum when a turtle lightens.

The relatively slow rate of substrate color-induced melanization in C. picta marginata suggests hormonal control that probably involves melanophore-stimulating hormone and visually mediated by light reflected from the substrate (Cooper and Greenberg 1992; Sugimoto 2002; Wasmeier et al. 2008). From a histological perspective, phenotypic plasticity in turtle coloration is interesting because the mechanism, or mechanisms, that underlie it apparently differs from other species of reptiles (Cooper and Greenberg 1992) and possibly among species of turtles (Woolley 1957; Lovich et al. 1990; this study). Detailed descriptions of the cellular mechanisms of color change in C. picta and more distantly related turtle species would be useful to discerning a more complete picture of similarities and differences among species.