Shell Disease in Painted Turtles Causes Changes in Stripe Color

1. Introduction

Emergent diseases are exacerbating challenges among wild turtle populations (Frederickson et al. 2024, Fey et al., 2015; Adamovicz et al., 2018; Zhang et al., 2019; Waltzek et al., 2022). Turtles already face declines in wild population numbers because of Illegal pet trade, habitat loss & destruction, harvesting as a human food source (Nardoni and Mancianti, 2023). Turtles have been on earth for hundreds of millions of years and are among the oldest extant reptile (Nardoni and Mancianti, 2023). Despite this, many species are currently faced with extinction (King et. al, 2026; Nardoni and Mancianti, 2023).

Painted Turtles (Chrysemys picta picta) are model organisms for the study of physiology, behavior, and evolution (Valenzuela, 2009). Painted Turtles are common North American turtles whose skin possess a variety of integumentary stripes and spots that are red, yellow and orange. Turtles are prized by humans world-wide because of these colorful stripes and spots.

Painted Turtle stripe and spot colors are generated by pterins and carotenoids bound in dermal xanthophores of the integument (Bagnara, 1983; Bagnara and Matsumoto, 2006), and the colors are dependent on access to carotenoids. Painted turtle forelimb stripes are orange-red in color to the unaided human eye, chin stripes are orange-red and post-orbital spots are yellow to the human eye. These colors are in part due to access to xanthophyll carotenoids (Steffen et al., 2015, 2019).

Carotenoids are lipid-soluble photosynthetic pigments in plants that animals must acquire from the diet (Goodwin, 1984). In contrast, pterins are nitrogen-rich, ultraviolet- (UV) fluorescent compounds that animals synthesize from basic purine (e.g., guanine) precursors (McGraw, 2006). Carotenoids play diverse and important roles in photoprotection, free-radical scavenging, immunomodulation, visual tuning, and animal coloration (Vershinin, 1999), and are purported to function as honesty-reinforcing mechanisms that underlie costly yet beneficial colorful ornamental traits (Lozano, 1994; Hill, 2002; Ibanez et al., 2014).

The Hamilton-Zuk hypothesis posits that animals use secondary sex characters, like spots and stripes, because they are costly to the individual who possesses them and reflect the possessor’s health state or condition (1982). This cost may be related to maintaining the immune function (Wedekind, 1992; Wedekind and Folstad, 1994). Carotenoids are used as an integumentary colorant but because they are also known to be important to modulating the immune system (reviewed in Svensson and Wong, 2011), they mediate a trade-off in colorful animals that consume them, so that animals either use specific carotenoid molecules as an integumentary colorant or a health enhancement, but not both (Koch and Hill, 2018; Hill, 2019; Hill et al., 2023). This is known as the Resource trade-off hypothesis.

Shell Disease is a general term for lesions that occur on turtle carapaces and skin, especially in captivity (Hatt, 2010; Zwart et al., 1985). Here I present experimental research that opportunistically provides support for the Resource trade-off hypothesis of carotenoids. I was performing a behavioral lab experiment in the lab when an accidental outbreak of shell disease (actual biological cause was unknown) in a lab population, spread to several (but not all) individuals in the lab. I decided to opportunistically test the hypothesis that if shell lesions grow on the carapace of captive turtles, then half of the turtles in the lab which have been supplemented with xanthophylls, the other half which have not, then turtles will use xanthophylls for either an integumentary colorant or as an immunological agent (but not both), and I predicted that there will be an interaction between turtles who lack dietary access to carotenoids and the carotenoids use as an integumentary colorant.

I show that turtles who have been supplemented xanthophylls have lower forelimb stripe brightness than turtles who have not been xanthophyll supplemented, but that there is an interaction: In turtles with xanthophyll supplementation but who are free of shell disease forelimb (whose stripes are orange-red) yellow chroma is similar to yellow chroma in turtles who have been supplemented xanthophylls but experience shell disease. But turtles who experience shell disease but lack xanthophyll supplementation have significantly lower forelimb stripe yellow chroma compared to those that have been supplemented xanthophylls. This indicates that carotenoids are used by Painted Turtles as either an integumentary colorant or for some unidentified immunological function.

2. Methods and Materials.

Turtle collection and care. Turtles were captured with 4-foot diameter hoop nets (baited with sardines & dry pellet dog chow) in a variety of ponds throughout Jefferson County, West Virginia. Turtles were then transported back to a lab in Shepherd University upon capture, and cared for according to Shepherd University IACUC protocol # 2018-001. Turtles were assigned unique identities by carving notches along the marginal scutes with a bastard file. Turtles were then transferred into 36.8 cm×22.2cm×24.8cm (20.379L3) plastic Critter Keepers™ tanks (LLL Reptile Large Exo Terra Faunarium Critter Keeper). The Critter Keeper tanks were filled with tap water and placed on shelving units with full spectrum fluorescent tubes (Vitalite T8, 32 watt) suspended 30 cm above each tank top. The room was maintained on a14:10 hour light: dark photoperiod, at an air temperature of 28.0°C, with relative humidity ranging between 40 and 60%. The Critter Keeper water was changed every other day and scrubbed with an abrasive brush.

Initial experimental protocol: Maintenance & husbandry: Upon initial entry into the lab, turtles were fed Tetra ReptoMin™ floating food sticks ad libitum, 3-4 times per week and housed for variable lengths of time before the experiment (range May 5-June 9^th, 2019). On June 10, 2019 a sufficient number of turtles were in the lab to start the experiment (N = 46), and turtles were randomly assigned to one of two diet groups: 1) fed nutritional gel with no xanthophyll supplementation (i.e., the negative control level of the independent variable, or IV hereafter) or 2) fed a nutritional gel enriched with xanthophyll supplements (i.e., the experimental level of the IV).

Once the experiment started (i.e., when a sufficient number of turtles were placed in the lab, N = 46), turtles were maintained by feeding individuals prepared and pre-cut squares of Mazuri™ Aquatic Turtle Gel (formula code 5M7N, Purina Mills, LLC, St. Louis, Missouri) ranging in mass between 3.8 and 6.8 grams. The base Mazuri Turtle Gel 5M7N (i.e., with no xanthophyll supplementation) contained 38 ppm beta-carotene and consisted of 55% protein, 15% fat, and 5% fiber. The Xanthophyll supplementation level of the experimental level of the IV consisted of the base Mazuri Turtle Gel 5M7N enriched with 40 ppm lutein in addition to the beta-carotene already present in the base diet. The experimental feeding regiment was performed for 9 weeks and 2 days (or 65 calendar days, June 10^th – August 13^th, 2019).

Spectrometry.—I measured spectral variation (300-700 nm) of the chin, post-orbital spot, and right forelimb stripes of Painted Turtles using an Ocean Optics (UV-Vis) S2000 spectrometer (Dunedin, Florida) coupled to a tungsten–deuterium light source. I used a bifurcated fiber-optic cable (400 µm inner diameter) mounted in a black plastic sleeve that was fastened to the probe tip and that was placed at an angle of 90° to, and 2 mm away from, the measured skin surface. I dried the turtles’ skin surfaces before the reflectance measurement, and I took three unique, adjacent, and non-overlapping measurements in the colorful stripe or spot of each body region. Color data were gathered as percent reflectance at 1 nm wavelength increments from 300 to 700 nm (integration time = 2 ms, scans to average = 22) and this output was smoothed and processed using CLR software v 1.0 (Montgomerie, 2006). From the smoothed spectral reflectance data, I constructed spectral reflectance curves for chin, forelimb and post-orbital color patches. The following wavelength ranges delimit hue for three wavebands: 300–399 nm = UV, 550–599 nm = yellow, and 650–700 nm = red. UV chroma (i.e., UVC) describes spectral purity in the UV waveband (300–399 nm), yellow chroma (i.e., YC, provided as S1Y in the Montgomerie CLR output) describes spectral purity in the yellow waveband and red chroma (RC, provided as S1R in the Montgomerie CLR output) describes spectral purity in the red waveband. Total reflectance from 300-700 nm describes overall brightness, B1 (Montgomerie, 2006). Initial spectrometric measurements were made on dates ranging from June 10 to 13, and we made a final series of spectrometric measurements on August 13^th as a post-experimental measure of stripe and spot color.

Shell lesions were discovered on captive turtles on July 11, 2019. I present results of forelimb stripes only, to simplify presentation of the results, and because they were the only stripe or spot region that showed a significant effect.

Treatment of Shell Disease Before Turtles were Released Back to the Wild

Once shell disease was detected on the carapace of a turtle, the turtle was handled by me or my lab students with latex gloves. The lesion was washed with mild liquid soap with a toothbrush. The toothbrush was rinsed to ensure that the lesion causing agent was not spread to another turtle in the future. Then the lesion was dried off with a paper towel, and covered with Betadine iodine. The turtle was placed in a dry terrarium for 15 minutes. Then the turtle was placed back into its Critter Keeper tank that was scrubbed and re-filled with fresh tap water. The turtle was subjected to this treatment every day until the lesion disappeared.

Statistical analysis.—I studied the effects of shell disease lesions (IV₁) on post-experimental skin stripe and spot color (the dependent variables, or DVs) when xanthophyll supplementation is a second indpendent variable, (or IV₂) and pre-experimental skin stripe and spot color was treated as a covariate. By using this covariate approach, I could determine if the IV had significant effects on the DV after the covariate influence was removed.

Before this can be done, however, data need to meet all assumptions of ANOVA and ANCOVA, multivariate significance needs to be investigated, and covariates need to be assessed for reliability. If covariates are reliable, then one usually wants to determine the ability of specific covariates to adjust specific DV’s. Only after these steps have been performed can the researcher reliably evaluate the dependent variables, and infer statistical significance of the effects of factors on DV’s (Tabachnik and Fidell, 2001).

I used SPSS v. 20 to evaluate normality of distributions of spectral variation. Several variables did not meet assumptions of normality; when yellow, red or UV chroma or overall brightness was non-normally distributed, the values were Box-Cox- transformed. The transformations were retained and used for the analyses only if they improved the normality of the variable (Shapiro-Wilks W test, available in SPSS). Otherwise, the original, untransformed variables were used. After appropriate transformations were retained or rejected, the data met all assumptions of homogeneity of variance-covariance matrices.

The following analyses were all performed in SPSS in syntax mode (with syntax provided by Tabachnik and Fidell, 2001). SPSS’s MANOVA command was used to perform MANOVA and determine multivariate significance. MANOVA was also used to assess the reliability of covariates. In this assessment, Wilks’ lambda was used to test for significant relationships between the linear combinations of DV’s (representing post-experimental skin stripe and spot color) and the covariates (i.e., pre-experimental skin stripe and spot color).

I also used SPSS MANOVA to investigate the ability of covariates to adjust specific wavebands of post-experimental spectral variation (i.e., the DV’s) by performing multiple regressions for each DV in turn, with the covariates acting as IV’s in the model. Regression coefficients (beta coefficients) and statistical tests of covariate significance were provided as output in SPSS (Table 1). All pre-experimental wavebands were able to adjust post-experimental wavebands (i.e., all post-experimental spectral variation were influenced by pre-experimental spectral variation).

I used MANCOVA to analyze the effects of carotenoid supplementation on post treatment skin color (post-yellow chroma, post-red chroma, post-UV chroma, post-carotenoid chroma, or post-brightness) in Painted Turtles. Pre-experimental measures of color (pre-carotenoid chroma, pre-yellow chroma, pre-UV chroma or pre-brightness) were used as covariates. Data were Box-Cox transformed when their distributions did not meet assumptions of normality or homogeneity of variance. I measured the colorful stripes along the chin (yellow to the human eye) and fore-leg (orange-red to the human eye) and post-orbital eyespots (yellow to the human eye).

Once the contributions of covariates to adjust the DV were evaluated, I wanted to see if the DV’s (post-experimental stripe and spot color) were affected by the IV’s (shell disease lesions and xanthophyll supplementation) once the effect of covariates (pre-experimental color) were removed. Because many of the DV’s were significantly correlated to each other (i.e., they showed a correlation coefficient > 0.30, see Tabachnick and Fidell, 2001), we used SPSS MANOVA to perform Roy-Bargmann step-down analysis. In Roy-Bargmann step-down analysis, DV’s are assigned priority based on theoretical or practical considerations, and the highest priority DV is analyzed as a univariate ANOVA where the covariates are entered as IV’s. The second priority DV is then entered into ANCOVA as a DV, while the highest priority DV is entered as a covariate (along with covariates from previous ANOVA). Following a first ANCOVA, a second ANCOVA is performed, with the higher priority DV from the previous ANCOVA being entered as a covariate along with previous covariates. This succession of ANCOVA’s is performed until the lowest priority DV is entered as a DV, and all previous DV’s are entered into ANCOVA as covariates (along with initial covariates, so that adjustment is made for the previous DV’s as well as the pre-experimental spectral variation (i.e the covariates). I assigned yellow chroma as the highest priority in red-striped patches, and red chroma the second highest priority (because xanthophylls likely influence portions of the red spectral curve, i.e., yellow and red make orange). UV chroma was assigned third highest priority, H4c (carotenoid chroma, as designated by Montgomerie, 2006) and overall brightness was assigned the fourth and lowest priority.

One assumption of MANOVA is that the regression between different covariates and each DV have similar slopes, and use of the average regression to investigate the effect of the IV on the relationship between covariates and the DV’s are therefore justified (Tabachnik & Fidell, 2001). If there is heterogeneity of regression for any step (i.e., P > 0.01) within the Stepdown analysis , then one cannot reliably interpret any of the steps following the heterogeneous step. I tested the Homogeneity of regression assumption of MANOVA according to Tabachnik & Fidell (2001)) and described in detail in Steffen et al. (2019). The assumption of homogeneity of regression was satisfactory for all steps and combinations of covariates and DV’s, and the DV’s were judged to be sufficiently reliable to act as covariates Table 2.

After performing Roy-Bargmann stepdown analysis and testing for homogeneity of regression I used SPSS MANOVA to perform univariate ANCOVA on each DV, with xanthophyll supplementation as IV, and pre-experimental measures of yellow, red, and UV chroma as well as overall brightness as covariates. To test for significant effects of xanthophyll supplementation on the linear combination of DV’s that represent post-experimental skin stripe and spot spectral variation, one needs to compare significant differences in spectral variation between univariate ANCOVA and stepdown F tests. According to Tabachnik and Fiddell (2001), DV’s that show both significant univariate and stepdown F statistics are statistically pure tests of significance because they demonstrate that there is unique variability shared by the treatments after adjustments for differences created by the covariates. I combined F statistics, and effect and error degrees of freedom from the univariate and stepdown analyses and present them side by side in Table 3 to facilitate comparison and evaluate the significance of each analysis type. When results show significant univariate but not significant stepdown F statistics, there is not unique treatment variability after covariate adjustment. Thus, significance is not reliable. If there is a significant stepdown F but not a significant univariate F, significant DV’s take on importance in the presence of higher order DV’s, and the results may be related to the order in which DV’s were entered into the stepdown analysis. Thus, significance is not reliable in this case, either. I controlled for increased likelihood of Type I inflation error due to multiple comparisons by dividing a typical P value indicating an experiment-wise error rate of 0.05 by 5 (i.e., P < 0.01) to accommodate the increased probability of making type I errors among 5 separate tests (i.e., one separate test for each DV).

3. Results

Roy-Bargmann Step down Analysis showed the following results. Pre-experimental yellow chroma (preS1y) was a significant covariate of post-experimental yellow chroma in painted turtle forelimb stripes (Table 3).

There was a marginal main effect of xanthophyll supplementation on yellow chroma (Table 3, Figure 1). Xanthophyll supplementation created a marginal decrease in yellow chroma (S1y) of forelimb stripes. Forelimb stripes of xanthophyll deprived turtles (C-) had an average yellow chroma of 0.199 (S.E. = 0.004) while forelimb stripes of xanthophyll supplemented turtles (C+) had an average yellow chroma of 0.188 (S.E.=0.005). Univariate F = 2.077, Stepdown F = 2.077, P = 0.073.

There was also a significant effect of shell disease state on forelimb stripe brightness (Table 3, Figure 2). Forelimb stripes of disease-free turtles showed lower average overall brightness of 46.71 (S.E. = 7.02), and forelimb stripes of turtles with shell disease showed an overall brightness of 65.67 (S.E.= 6.9). Univariate F = 3.145, Stepdown F = 5.024, P = 0.031.

There was a marginally significant interaction between xanthophyll supplementation and shell disease on forelimb stripe yellow chroma. In turtles free of shell disease, forelimb yellow chroma is similar between turtles who are supplemented with xanthophylls and in turtles who are supplemented with xanthophylls, but experienced shell disease. But turtles who experience shell disease but lack xanthophyll supplementation have significantly lower forelimb stripe yellow chroma compared to those that were supplemented with xanthophylls (Figure 3).

Turtles that were disease free, but who were xanthophyll supplemented (i.e. C+) had an average yellow chroma of 0.217 (S.E. = 0.01), and disease free turtles who lacked xanthophyll supplementation (i.e. C-) had an average yellow chroma of 0.220 (S.E.=0.009). Among turtles with shell disease, those who were xanthophyll supplemented (i.e., C+) had an average yellow chroma of 0.199 (S.E.= 0.01) and turtles who lacked xanthophyll supplementation (i.e., C-) had an average yellow chroma of 0.221 (S.E. = 0.009). Univariate F = 3.516, Step-down F = 3.886, P = 0.056

4. Discussion

The research presented here adds to published literature demonstrating that turtles, archosaur reptiles that have been on the earth for 200 million years (Ernst and Lovich, 2009), must ingest carotenoids to maintain their colorful spots & stripes (Steffen et.al 2019). Other reptiles that have been shown to maintain color with carotenoids, which also respond to some aspect of homeostatic physiology include Frillneck lizards (Chlamydosaurus kingii, McLean et al., 2019), Common lizards (Lacerta vivipara, Fitze et al., 2009; Cote et al., 2010), Iberian Green lizards (Lacerta schreiberi, Kopena et al., 2014) and even a 50% probability of tissue-based expression in Archelosauria, Archosauria, and Aves (i.e., a clade joining turtles and archosaurs—birds and crocodilians—and their closest relatives, Davis and Clarke, 2022).

In Painted Turtles, carotenoids appear to be used for maintaining an (undetermined) aspect of health in turtles that have shell lesions, because their orange-red forelimb stripes lose chroma from the yellow waveband (i.e., spectral purity in 550-559 nm, Steffen et. al., 2019). This finding supports the resource trade-off hypothesis (Koch and Hill, 2018; Koch et al., 2019) of carotenoids as colorant of stripes and spots in turtles. The disease that provides this support is a common illness experienced by turtles in captivity (Rothstein and Steffen, 2020). Shell disease increases the overall brightness (or overall reflectance) of Painted Turtle forelimb stripes.

Painted Turtle color generating mechanisms have not been comprehensively studied, but Red-eared sliders (Trachemys scripta) and River Cooters (Pseudemmys concinna) have been studied with respect to cellular (i.e., pigmentary), ultra- and nano-structure color generating elements (Brejcha et al., 2019). How pigments (e.g. xanthophyll) interact with other mechanisms can have important outcomes for reflectance patterns of spots or stripes in turtles. The yellow forelimb stripes of both P. concinna and T. scripta contain xanthophores and iridophores, but the post-orbital regions of the two species differ in cell-type distribution. Turtles also employ novel color-generating mechanisms (e.g., nano-organization of dermal collagen).

When carotenoids are diverted from the dermal chromatophore unit to some other location, and for some other purpose (e.g. health enhancement) overall brightness increases because carotenoid saturation decreases (Shawkey et al., 2005, 2017). Finally, this research showed a marginal interaction between shell disease state, xanthophyll supplementation and yellow chroma of the orange-red forelimb stripes. Turtles who were supplemented with xanthophylls had similar forelimb stripe yellow chroma regardless of whether the turtles had shell disease or not. Presumably xanthophyll supplementation provided a sufficient amount of xanthophylls to allow turtles to use carotenoids for integumentary color maintenance and immunological purposes. But among turtles who lacked xanthophyll supplementation, forelimb stripes from turtles who had shell disease had lower yellow chroma than turtles who were free of shell disease. When carotenoids are diverted from the dermal chromatophore unit to some other location, and for some other purpose (e.g. health enhancement) overall brightness increases because carotenoid saturation decreases (Shawkey et al., 2005, 2017).

Several pieces of evidence indicate that these disease-based changes in color are perceptible to other Painted Turtles (Steffen et. al., 2019). Red-eared slider turtles (and all Emydid turtles) have a well-developed tetra-chromatic visual system (Loew and Govardovskii, 2001). Emydid turtle retina are capable of complex visual processing (Loew and Govardovskii, 2001; Zana et al., 2001). Turtles possess UVS, SWS, MWS and LWS retinal photoreceptor cones (Loew and Govardovskii, 2001). These retinal cones contain keto-carotenoid pigmented oil droplets, which aid in wavelength discrimination by reducing scattered light and improving contrast, as well as signal-to-noise ratio (Loew and Govardovskii, 2001; Neumeyer and Jäger, 1985, 2001; Zana et al., 2001; Lopes et al., 2016; Twyman et al., 2016).

Fresh water lakes and ponds that Painted Turtles inhabit transmit light maximally at longer wavelengths (Lythgoe, 1984). When viewed against low-reflecting brown colors of the carapace or skin, the increases in yellow and red spots and stripes are likely to be highly visible to conspecific individuals. Finally, Painted Turtles are a quasi-social species of reptile, and they show these stripes and spots in close contact social encounters such as basking or in water during courtship (Bury, 1979; Lovich, 1988).These results provide support for the Resource Trade-off hypothesis of carotenoids. Turtles suffering shell disease presumably lost yellow chroma in their forelimb stripes because the xanthophyll supplementation was diverted to an undetermined health use instead of the integument as a colorant (Koch and Hill, 2018).

I did not identify the cause of the shell lesions that appeared in the lab and I noted shell disease by observing lesions on the keratin-covered scutes of the carapace only. Lesions could have been caused by fungi, bacteria, and/or ecto parasites. In previous studies, wild-caught River Cooters (Pseudemmys concinna) and Yellow-Bellied Turtles (Trachemys scripta) lesions correlated with bacterial and trematode infections (Lovich et.al 1996; Garner et al., 1997). In map turtles (Graptemys species) maintained under natural conditions but brought to University of Georgia’s Veterinary Teaching Hospital, six biopsies of lesions revealed yielded aerobic isolates (Chryseobacterium indologenes, Aeromonas hydrophila, Ralstonia pickettii, and Morganella morganii), whereas eleven shell samples grew various clostridial anerobes (Hernandez-Divers et al., 2009). Emydomyces testavorans is not likely to be the causative agent of the lesions because it was found at low prevalence when investigated as causes to lesions in aquatic Painted Turtles (Frederickson et al., 2024, White et. al., 2025).

5. Conclusions

The cause of shell lesions in the present study is presumably due to poor turtle husbandry and omission of aquatic animal water conditioner in the lab (Hatt, 2010, Rothstein and Steffen, 2020, Nardoni and Manchianti, 2023). No hygienic practice was used in the present study that might harm turtles (e.g., turtle housing was not washed with bleach to sterilize tanls), and turtles were treated with standard veterinary care procedures to alleviate disease and suffering. Nevertheless, Painted Turtle color, one of the most aesthetically pleasing aspects of Painted Turtle appearance which leads to Painted Turtles to be sold through the pet trade and renders Painted Turtles as common pets, is affected by shell disease that is common to anthropogenic landscapes (i.e., captivity). In other words, shell disease, a common disease of human captivity reduces the vibrancy of Painted Turtle stripe and spot color.

Ulcerative shell disease has been found on Geomydid fossil turtles from the early Eocene in southwestern Wyoming (Zonneveld and Bartels, 2023). Thus, turtles have been co-evolving with shell disease for millions of years. Nonetheless, turtles are facing extinction at a very high rate from so many causes, that shell disease due to human captivity or other indirect anthropogenic causes needs to be better understood and treated. Chelonian conservation efforts should continue to include turtle health and disease treatment.