Are Painted Turtles (Chrysemys picta) Resilient to the Potential Impact of Climate Change on Vitamin D via Overgrown Floating Vegetation?

Abstract

Floating aquatic vegetation and algal blooms are increasing with global warming, potentially reducing UVB exposure and, consequently, vitamin D (vit-D) synthesis in freshwater turtles. Vit-D mediates calcium metabolism and overall health, yet the effects of floating aquatic vegetation on vit-D levels remain unclear, as is whether turtles actively avoid habitats with abundant floating vegetation. Here, we address these questions by quantifying vit-D₃ levels in the blood of adult female painted turtles (Chrysemys picta) exposed to high-vegetation (darker/colder) or clear-water (lighter/warmer) treatments for one month outdoors and one month indoors at a single temperature during late summer and early fall. The observed circulating vit-D₃ levels resembled those reported for other freshwater turtles, declined over time in both treatments, and were marginally lower under high vegetation after 60 days compared to clear water. However, this difference disappeared after correcting for lymph contamination and multiple comparisons, suggesting that perhaps adult females are robust to the effect of floating vegetation, but whether they were buffered by vit-D₃ stores in lipids is unclear. Additionally, in subsequent years, females were exposed to habitat choice experiments and exhibited a strong preference for high floating vegetation over clear water, both as a group (outdoors) and individually (outdoors, and indoors at 21 °C and 26 °C), consistent with known benefits conferred by floating vegetation (food, predator avoidance). While no ill effects of high vegetation nor behavioral avoidance were detected here, longer experiments at different seasons on both sexes and varying ages are warranted before concluding whether painted turtles are truly resilient in their vit-D levels or if, instead, a tradeoff exists between the known benefits of floating vegetation and potential [yet unidentified] detrimental effects (lower dissolved oxygen or vit-D) when vegetation is overgrown for extended periods.

1. Introduction

Turtles inhabit numerous aquatic and terrestrial environments across the globe. North American freshwater turtles exhibit preferences for various water bodies, such as open water (softshell turtles), backwaters of rivers (red-eared slider turtles), or ponds with vegetation and ample basking sites (painted turtles) [1]. These areas often contain floating aquatic vegetation (macrophytes) and algae, whose proliferation can lead to complete coverage of the water surface [2]. Indeed, warmer springs induce higher macrophyte growth later in the turtles’ reproductive season [3], as elevated temperatures generally increase plant growth rates [4]. Thus, because global temperatures are projected to rise due to anthropogenic climate change by 2.5 °C [5], up to 6 °C [6], macrophyte growth is expected to increase over the next several decades [7,8]. Freshwater turtles benefit from areas of floating vegetation, which allow them to forage for plants and invertebrates while avoiding predators [9,10], thus increasing their fitness. Consistently, painted turtles’ abundance is higher in areas with more floating vegetation [11], but whether any ill effects on turtle biology might be anticipated from such habitat modifications remains unknown, even though this information is crucial for effective management [12,13]. One important turtle behavior that might be altered is basking, which can be aquatic or atmospheric and is affected by many factors, such as solar radiation, submerged vegetation, water turbidity, and flow speed. Basking site preferences may be species-specific [14,15,16,17] and can also impact turtles’ nesting [18].

During atmospheric basking, turtles rest on land or on exposed objects out of the water, primarily fallen logs and debris near the shore with sun exposure [19]. In contrast, during aquatic basking, the turtle’s carapace, head, and limbs remain just under the water’s surface, exposed to the sun [20,21], often with limbs outstretched to maximize surface area exposure [16]. Turtles benefit from using atmospheric basking as a thermoregulatory strategy to elevate their body temperature above that of the water [17], as this enhances gonadal recrudescence, somatic cell growth, locomotion, and potentially foraging success, digestion, and sexual maturation (reviewed in [22]). Other benefits include the removal of harmful ectoparasites by drying the body thoroughly at the appropriate basking duration and air temperature [23], the removal of excess algae (which improves swimming dynamics) [16], and the avoidance of aquatic predators [24,25,26], which explains infrequent nocturnal basking [24,27,28,29]. But atmospheric basking also imposes costs by reducing foraging time, exposing turtles to terrestrial predators, and augmenting energy expenditure by raising their metabolism.

Turtles stop basking atmospherically once the water temperature reaches their preferred body temperature [20], around 21–25 °C for painted turtles [30,31], as occurs in late summer, leading to a drastic decrease in atmospheric basking and a transition to floating, often at the surface of the water exposed to sunlight, which also occurs more frequently in habitats lacking basking structures [15]. While much less is known about aquatic basking, it also has thermoregulatory benefits, with turtles’ body temperatures increasing anywhere from 0 to 3 °C above the water temperature [16,32], which is less than the 2–10 °C warming attained with atmospheric basking [33,34]. This more modest effect on body temperature suggests that turtles might bask aquatically for reasons other than purely for thermoregulation. Indeed, basking may also serve to regulate UV exposure, which is essential for vitamin D synthesis. For example, eastern fence lizards (Sceloporus undulatus) have been shown to prioritize UV exposure over thermoregulation during basking [35], although whether a similar behavior occurs in turtles remains unknown.

Turtles require sunlight to synthesize vitamin D (vit-D) in their skin [36,37,38,39,40,41], as do other reptiles such as snakes and lizards [39,42,43,44], a process that is impacted positively by body temperature [45]. Vit-D is crucial for many functions in turtles, such as calcium metabolism and immune response, helping them to recover from diseases such as fibropapillomatosis in green sea turtles (Chelonia mydas) [46]. Indeed, vit-D actively transports dietary calcium from the intestines to the bloodstream [47,48], which strengthens eggshells, as well as the turtles’ shell and skeletal bones, which help buffer against lactic acid accumulation during brumation. In the absence of sufficient sunlight exposure, the only other vit-D source for turtles is dietary. Consistently, carnivorous turtles such as Chelydrids (snapping turtles), Kinosternids (mud and musk turtles), and several others obtain sufficient vit-D from their food sources [49] and bask less often than omnivorous turtles [50,51,52,53] such as Emydids and Batagurids (freshwater pond turtles). Specifically, Emydid and Batagurid turtles regularly bask and spend large amounts of time in the sunlight, suggesting that they mainly synthesize vit-D from UVB exposure, as dietary sources do not appear to be sufficient for them [49]. Similarly, snakes are obligate carnivores that presumably obtain most of their vit-D from their diet [43] while retaining the ability to synthesize it from UVB radiation when needed [42,43].

Notably, several factors, such as cloud cover, water turbidity, and shading by aquatic and terrestrial vegetation, influence how much UVB radiation reaches the water and, thus, reaches turtles that bask aquatically. Over the next century of global warming, algal blooms are expected to increase in frequency and persistence, potentially blocking light similarly to floating aquatic vegetation [54]. These habitat features certainly reduce the penetration of sunlight into the water column, but whether this could negatively affect the ability of wild turtles to synthesize sufficient vit-D remains untested. Answering this question is urgent, because insufficient vit-D could have adverse cascading effects, such as metabolic bone disease, when dietary calcium cannot be absorbed due to vit-D₃ deficiency [55], as observed in captive reptiles [48,56]. Furthermore, reproductive females are predicted to be affected more strongly in their vitellogenic cycle and calcium allocation to the yolk and eggshells under reduced UVB radiation [57], potentially affecting their offspring’s fitness detrimentally. Thus, to understand the full spectrum of the challenges posed to turtle populations by global warming, it is critical to understand the associations among aquatic vegetation, UVB exposure, and circulating vit-D levels in reproductive females.

Here, we tested the hypothesis that increased floating aquatic vegetation decreases circulating vit-D₃ levels in adult female painted turtles (Chrysemys picta, Emydidae) by reducing UVB exposure through the attenuation of solar radiation by dense floating vegetation during aquatic basking (Figure 1). This hypothesis was motivated by previous lab experiments, where vit-D₃ levels decreased without UVB light but increased when a UVB light source was introduced immediately after aestivation [36,40,58]. Alternatively, turtles may be buffered from the adverse effects of blocked sunlight by other means (perhaps through their diet, or by limited exposure of their head above water). While UVB can penetrate the water to some degree, the dense vegetation cover used in our study likely diminished exposure significantly for turtles floating at the surface. We also tested whether turtles actively avoid areas that are heavily covered by floating vegetation that might potentially reduce vit-D production. This project used a combined experimental approach in the field and the laboratory, quantifying circulating vit-D₃ in adult females exposed to one of two treatments: a pond/tank with full vegetation cover (‘vegetation’) or no vegetation cover (‘clear water’), as described below. In complementary experiments, we presented females with a choice between clear water or high vegetation coverage within the same water body to study their habitat preferences.

2. Materials and Methods

2.1. 2022 Vitamin D Experiment

2.1.1. Animal Procurement and Experimental Design

Forty-six adult female painted turtles 337.1–935.6 g in mass (similar to or greater in mass than the smallest 340 g gravid female found among the study population) were captured during the spring and summer of 2022 under an Iowa DNR permit and held in an outdoor artificial pond at the Iowa State University (ISU) Horticulture Research Station. The turtles were individually marked on their marginal scutes using Cagel’s [59] notching system and with an additional numbered metal tag. On 2 August 2022, the females were retrieved, randomly assigned to one of two treatments (n = 23 females per treatment), and released in two experimental ponds ~49 × 10 m in surface area, with an average depth of 2.5 m: one pond with clear water, and one with extensive floating vegetation cover (Figure 2A,B). Both the ‘clear water’ and ‘vegetation’ treatment ponds were surrounded by 1 m high landscape fencing buried in the soil to prevent the turtles from escaping. A shaded and dry area was created to allow the turtles to rest out of the water while preventing them from basking atmospherically and, thus, being exposed to UVB light outside of their treatments. Three Onset Hobo Pendant mxTemp/Light loggers were deployed in the center of each pond, at 5 cm, 15 cm, and 30 cm below the water’s surface. The loggers recorded temperature and light continuously at one-minute intervals throughout the experiment, and all data were used for statistical tests of differences between treatments and depths, whereas the daily mean temperature and light for each depth and treatment were calculated for visualization. The turtles were provisioned with Mazuri^® Aquatic Turtle Diet pellets daily, which contain a vit-D₃ concentration of 1531 IU/kg, a lower concentration than that of the diet used in previous experiments on turtles (1730 IU/kg), where daily feeding allowed for the detection of an effect of UVB deprivation on circulating vit-D₃ [36]. The amount of food provisioned daily represents ~1.5% of their body weight, or ~14.29 IU of vit-D₃ (~22.95 IU/kg). We should note that turtles in the outdoor ponds could potentially supplement their diet with naturally occurring organisms.

A blood sample (1 cc) was taken from each individual before the start of the experiment (day zero, 0 d) and one month later (30 d) at the end of the outdoor experiment. Blood was drawn from the subcarapacial sinus [60], centrifuged at 3400 rpm for 2.5 min to separate the plasma from the blood cells, flash-frozen in dry ice, and stored at −80 °C until processing. Of the 46 females, 34 were retrieved one month after the onset of the outdoor experiment (as some individuals escaped), brought into the lab, and kept in the same treatment group, but this time in 1.2 × 1.2 × 0.3 m plastic tanks, either covered in duckweed brought in from the outdoor pond (n = 16, ‘vegetation’) or without vegetation (n = 18, ‘clear water’) (Figure 2C,D). The tanks were filled to a depth of 15 cm with water and had a UVB light placed 30 cm above the water in a temperature-controlled room set at 23 °C. The turtles were fed the same daily amount of Mazuri^® Aquatic Turtle Diet pellets (PMI Nutrition International, LLC, Arden Hills, MN, USA) as during the outdoor experiment. One Onset HOBO Pendant mxTemp/Light logger (HOBO, Bourne, MA, USA) was placed at the bottom of each tank, and measurements were taken every 5 min. Blood was collected in the same manner as described above after one month indoors, i.e., 60 days after the onset of the outdoor experiment (60 d). The turtles were released at their location of capture at the ISU Horticulture Research Station at the end of the experiment. All animal procedures were approved by the ISU IACUC Animal Care Committee.

2.1.2. Vitamin D Sample Quantification

All plasma samples were submitted to Heartland Assays, LLC (Ames, IA, USA), for the quantification of 25-hydroxyvitaminD₂ (25OHD₂) and 25-hydroxyvitaminD₃ (25OHD₃) using mass spectrometry in an AGILENT 1290 infinity HPLC (High-Performance Liquid Chromatography) coupled with an AGILENT 6460 MS/MS (Tandem Mass Spectrometry) with an ESI (Electrospray Ionization) source. A standard curve was made in house with 25(OH)D2 (Sigma-Aldrich, St. Louis, MO, USA), 25(OH)D3 (Sigma-Aldrich), (6,19,19)-d3-25(OH)D2—25(OH)D2-[d3] (IsoSciences), and (6,19,19)-d3-25(OH)D3—25(OH)D3-[d3] (IsoSciences). As a control, National Institute of Standards and Technology NIST 972a vit-D metabolites in frozen human serum—four-level standard (Gaithersburg, MD, USA) was used. Plasma samples, along with a standard curve and controls [61], underwent protein precipitation with 0.2 M zinc sulfate solution, followed by vortexing and subsequent addition of methanol. D3-25(OH)D2/D3-25(OH)D3 internal standards were then added to appropriate samples and controls, followed by vortexing. Hexanes were added to all samples and controls, capped, and vortexed, followed by centrifugation. The organic layer was then transferred and dried. All standards, controls, and samples were reconstituted using LCMS-grade methanol and water, both containing 0.1% buffer, before being loaded onto the auto-sampler for analysis. The assay accuracy was determined to be >94%, based on assessment against the NIST-certified standards for 25OHD. Samples below the limit of detection (LOD), i.e., <1.5 ng/mL, were assigned a constant value of 0.00001 in the data matrix for statistical analysis. Circulating levels of vit-D₂ were undetectable for all samples (<1.5 ng/mL), such that all of the analyses described below correspond to vit-D₃.

2.1.3. Vitamin D Statistical Analysis

Because different amounts of lymph are often drawn when sampling blood from the subcarapacial sinus [62], which would dilute the plasma, we weighed the vials containing the plasma and blood cells (BC) fractions separately, and we calculated a dilution factor as the ratio of plasma-to-BC weight. The raw values of vit-D₃ were multiplied by this dilution factor before statistical tests, and the results obtained using these ‘relative’ values were compared to those using the uncorrected raw values.

The values for 25OHD₃ were analyzed using a linear model in R version 4.4.2 [63], using the RRPP package v 2.0.0 [64], where vit-D₃ levels were modeled as a function of treatment (T), day (d), and their interaction (T × d), including individual female ID as a random effect and female size as a covariate. This ANCOVA was performed on (a) uncorrected raw vit-D₃ levels, (b) relative values (i.e., raw values multiplied by their dilution factor), (c) delta values (change in raw values compared to the baseline values at day 0), and (d) relative delta values (change in relative values from day 0), with Bonferroni correction for multiple comparisons. The values for all variables were Box–Cox-transformed to improve the normality of the data, and the results were compared to those using untransformed data.

Because no significant treatment-by-day interaction was detected, a reduced model was run after removing the interaction term, fitting all other effects, followed by three pairwise comparisons to assess the differences between treatments at each time point (0 d, 30 d, and 60 d). A further reduced model was then run excluding day, and pairwise tests were conducted to test exclusively for differences in vit-D₃ over time.

2.2. 2023–2024 Habitat Choice Experiments

2.2.1. 2023 Outdoor Group Choice Experiment

Reproductive females (n = 39 in 2023 and n = 38 in 2024) were collected, marked as described above, and housed in an outdoor artificial pond for 10 days prior to the experiments. A 16 m × 12.5 m artificial pond coated with a black pond liner (1 m average depth) was divided into four equal quadrants by deploying a cotton fabric (30 cm wide) across the pond from north to south and east to west, positioned 15 cm above and 15 cm below the water surface, which allowed the turtles to move freely between quadrants underwater or on land (Figure 3A,B). Duckweed and watermeal from nearby wetlands were added to the northwest (NW) and southeast (SE) quadrants, while the northeast (NE) and southwest (SW) quadrants contained clear water. Hereafter, the quadrants will be referred to as veg-NW, veg-SE, clear-NE, and clear-SW. Field cameras (Cabela’s Outfitter™ Gen 3 30MP IR Game Camera) were positioned 3.5 m high on posts placed ~5 m from each pond corner, and photographs were taken every ten minutes (6:00 AM–8:00 PM) from 21 August to 1 September 2023. One Onset Hobo™ Pendant mxTemp/Light logger was deployed 5 cm below the water’s surface in each quadrant, to simulate a turtle basking aquatically. We should note that the temperature measurements were taken at a fixed depth, while the turtles could occupy a range of depths when underwater. Therefore, the temperatures reported here may not fully capture the thermal environment experienced by all individuals, especially those deeper in the water column, where temperatures are more stable. Each morning, duckweed that may have drifted into the clear quadrants was removed with a leaf blower. Adult females were randomly assigned to one of the four quadrants at the onset of the experiment and allowed to move ad libitum thereafter. The turtles were collectively provisioned daily with 2.5 oz of the Mazuri Aquatic Turtle Diet pellets per quadrant. The total number of turtles observable per quadrant in each of the photos was counted, noting whether they were basking in the water, on the black pond liner devoid of vegetation (shore), or in the grass.

2.2.2. 2024 Individual-Choice Experiments

To account for the possibility of social facilitation and environmental effects [65,66] during the 2023 group experiment—where females may have followed one another instead of choosing quadrants on their own, and perhaps may have been affected also by the environmental temperature—three individual-choice experiments were conducted in the late summer/early fall of 2024. For these, six 52 × 76 cm, 102.21 L (27 gallon) Rubbermaid tubs were partitioned with plastic panels to separate the vegetation cover (duckweed) and clear-water sections while allowing the turtles to swim underneath (Figure 3C,D). Each female was tested thrice under different conditions: (1) outdoors in the late morning, (2) indoors at 31 °C, and (3) indoors at 21 °C, in a random order for each set of trials. First, for the outdoor experiment, turtles were collected from the holding pond each morning by net or pitfall trap (n = 38, 36, 36 per trial, since not all 38 females could be re-captured for re-testing). The turtles were positioned on a wooden board above the water at the center of the tub (Figure 3E) and allowed to choose their starting compartment (vegetation or clear water). Then, to test whether environmental temperature might affect habitat preference, eight partitioned tubs were placed in a temperature-controlled Percival chamber (Model CTH-810), and the females (n = 29 that could be re-captured from the outdoor experiment) completed three trials at 31 °C on three separate days, along with three more trials at 21 °C. The animals were held at 26 °C in a second chamber for 24 h before the first indoor trial and between trials. Individuals rested for at least 16 h between trials and were tested in a random order each day. For all individual-choice experiments, photographs were taken every five minutes for 100 min using cameras mounted above the tubs. At the conclusion of the study, the turtles were released at the ISU Horticulture Research Station. Data per individual female were collected as presence in the clear side of the tub (coded as 1), in the water covered with vegetation (coded as 0), or in the middle (coded as 0.5). Coded observations were multiplied by 5 and summed to calculate the total minutes spent in the partition with clear water or vegetation per trial (100 min minus the time spent in clear water), with the average calculated across the three trials per female (thus accounting for non-independence of the repeated measures), and these averages were then summed across all females within each experimental condition (outdoors, indoors at 31 °C, and indoors at 21 °C). After the animals were released at the end of the experiment, some errors were discovered, i.e., some of the animals were not tested for one of their triplicate trials by mistake, the camera did not take pictures during a trial, or data from a trial were deleted unintendedly. Thus, the sample sizes differed between trials (n = 29, 29, 28 for the experiments at 31 °C, and n = 29, 29, 22 for the experiments at 21 °C). If a turtle was missing data for a trial, we calculated the average of the available trials.

2.2.3. Habitat Choice Statistical Analysis

2023 Temperature and Sunlight Data: To assess the effects of vegetation cover and quadrant on temperature and sunlight (lux) levels, a full-factorial two-way ANOVA was performed for each of the response variables (temperature and sunlight), with treatment (clear, vegetation) and quadrant (NE, SW, NW, SE) as fixed factors, and with their interaction included. Follow-up pairwise comparisons were conducted to assess differences in temperature and sunlight between the treatment-by-quadrant groups. Analyses were performed using a linear model in R with the RRPP package (v2.0.0) [64].

2023 and 2024 Group and Individual-Choice Data: The total counts of females observed basking per pond quadrant for the 2023 outdoor group study, or the time (minutes) spent by all females in the vegetated tank partition for the 2024 individual-choice experiments, were analyzed using a chi-squared test [chisq.test() function in base R] to determine whether the distributions deviated from random expectations (i.e., from 25% null expectation in each pond quadrant, and 50% null expectation in vegetation versus clear water). All statistical tests were run in R version 4.4.2 [67].

3. Results

3.1. Temperature and Sunlight

Drastically less sunlight was detected, and the temperature was consistently colder, at 5 cm, 15 cm, and 30 cm (by 2 °C) in the outdoor pond covered by floating vegetation compared to the clear-water pond in 2022 (Figure 4). The same was true for sunlight in the 2023 outdoor experiment, whereas no significant thermal differences were found between quadrants that year (Supplementary Table S1, Figure S1), likely because the 2023 pond was shallower than the 2022 ponds (1 m versus 2.5 m average depth, respectively). Additionally, the indoor tubs used in the 2022 vitamin D experiment differed significantly in light (8.43 lux and 13.78 lux for the vegetation and clear tubs, respectively; p < 3.75 × 10⁻²⁰). On the other hand, the temperature was almost identical between tubs (21.34 °C vs. 21.40 °C for the vegetation and clear tubs, respectively), although the difference was statistically significant due to the large number of measurements taken (every 5 min).

3.2. Circulating Vit-D Levels

First, vit-D₂ was undetectable in our samples, which is not surprising, since vit-D₂ is significantly less abundant in vertebrates than vit-D₃ [68] and raises total circulating vit-D much less than dietary vit-D₃ sources do [39]. In contrast, vit-D₃ was detectable (>1.5 ng/mL) in all but five samples and declined over time in both treatments (

Table 1. Average and range of raw values of circulating vit-D₃ quantified in reproductive Chrysemys picta females at 0, 30, and 60 days of exposure to the experimental treatments.

Treatment	Day	n	Vitamin D3 (ng/mL)
			Mean	Min	Max
Clear	0	21	7.84	5.30	12.70
Vegetation	0	21	8.61	3.20	14.90
Clear	30	18	6.86	4.30	11.10
Vegetation	30	15	7.19	3.40	11.30
Clear	60	18	3.26	1.06	5.70
Vegetation	60	15	2.48	1.06	7.40

, Figure 5).

An initial ANOVA on untransformed data revealed no significant interaction between treatment and sampling date in uncorrected vit-D₃ values (p > 0.189) but found significant effects of treatment (p = 0.016) and female ID (p = 0.012) (Supplementary Table S1). A follow-up reduced ANOVA including only the main effects detected consistent effects of treatment (p = 0.017), day (p = 0.001), and female ID (p = 0.009) on vit-D₃ levels (Supplementary Table S1), with pairwise comparisons indicating marginally lower vit-D₃ in the vegetation treatment at 60 d (p = 0.055) (

Table 2. Uncorrected p-values from pairwise comparisons of vit-D₃ levels (A) between vegetation and clear-water treatments at days 0, 30, and 60, and (B) for both treatments combined (overall) after one or two months relative to day 0, evaluated before and after Box–Cox transformation of the data, with and without female size as a covariate. Bold italic font denotes p < 0.1, and bold underlined font denotes p < 0.05. Shaded cells denote significant Bonferroni-corrected p-values at alpha = 0.004.

	(A) Between Treatments at Each Day			(B) Overall Compared to Day 0
Variable	Day 0	Day 30	Day 60	Day 30	Day 60
Untransformed values
Raw Vit-D3	0.15	0.627	0.055	0.325	0.001
Relative Vit-D3	0.291	0.533	0.985	0.011	0.005
Delta Vit-D3	1	0.813	0.368	0.302	0.001
Delta Relative Vit-D3	1	0.102	0.913	0.012	0.007
Box–Cox-transformed values
BC Vit-D3	0.274	0.795	0.008	0.426	0.001
BC relative Vit-D3	0.134	0.451	0.272	0.084	0.001
BC delta Vit-D3	1	0.815	0.458	0.318	0.001
BC delta relative Vit-D3	1	0.101	0.944	0.009	0.008
Untransformed values, female size covariate
Size-covariate Vit-D3	0.183	0.403	0.963	0.224	0.001
Size-covariate relative Vit-D3	0.377	0.629	0.863	0.032	0.002
Size-covariate delta Vit-D3	0.921	0.845	0.437	0.318	0.001
Size-covariate delta relative Vit-D3	0.773	0.21	0.579	0.011	0.016
Box-Cox-transformed values, female size covariate
Size-cov BC Vit-D3	0.147	0.276	0.979	0.264	0.001
Size-cov BC relative Vit-D3	0.129	0.57	0.779	0.356	0.001
Size-cov BC delta Vit-D3	0.923	0.859	0.483	0.333	0.001
Size-cov BC delta relative Vit-D3	0.777	0.212	0.578	0.009	0.017

A). After Box–Cox transformation, only day and the treatment-by-day interaction remained significant (Supplementary Table S2), but pairwise comparisons again showed significantly lower vit-D₃ in the vegetation treatment at 60 d (p = 0.008) (Table 2A) compared to 0 d and 30 d. The delta values (transformed and untransformed) exhibited the same pattern, indicating that vit-D₃ tended to changed across all days, but decreased more substantially between 30 d and 60 d (p = 0.001) (Table 2B).

However, the effects of treatment or treatment-by-day interactions disappeared when the raw vit-D₃ values were corrected by their dilution factor (relative values), such that only the effect of day remained significant (p = 0.001) (Supplementary Table S1). The same was true for the adjusted relative delta values, and pairwise comparisons on the reduced model showed that, consistent with absolute levels, the reduction in relative vit-D₃ from baseline was greater by 60 d than by 30 d (p = 0.005) (Table 2B). After Box–Cox transformation, the effect of treatment at 60 d was no longer significant (Supplementary Table S2), while the results for all other comparisons between treatments and across days remained consistent with the results using the untransformed data (Table 2).

The analyses described above also revealed a significant effect of female ID, which might be attributable to differences in body size. To account for this possibility, we analyzed the data after adding female size to the model as a covariate, and the results for both the untransformed and Box–Cox-transformed data were generally robust to this inclusion (Supplementary Tables S3 and S4). Specifically, most trends remained consistent, with one exception: the trend toward lower vit-D₃ in the vegetation treatment at 60 d disappeared after accounting for female body size, such that, while the full model revealed significant effects of treatment and day (Supplementary Table S4), the pairwise comparisons showed no significant effect of treatment (Table 2A). Following the Box–Cox transformation, pairwise comparisons indicated that the overall vit-D₃ levels were significantly lower at 60 d compared to 0 d or 30 d for all variables except relative delta, for which the vit-D₃ levels differed among all sampling points (Table 2B).

3.3. Habitat Choice Experiments

For the 2023 group behavioral dataset, the chi-squared analysis revealed significant deviations from the null expectation that the turtles would utilize all quadrants equally for aquatic basking (p < 2.2 × 10⁻¹⁶ for all comparisons,

Table 3. Female habitat choice recorded in this study: (A) Total counts of turtles basking in the water per quadrant and adjacent areas every 10 min summed through the 2023 experiment. (B) Time spent by individual females in the vegetation or clear-water partition per condition in the 2024 experiment, averaged across their three trials and summed across all females. p-Values from the chi-squared tests of differences from random expectations among quadrants (2023) or between partitions (2024) are also presented.

(A) Total Female Count 2023 Experiment					(B) Time Spent (min) 2024 Experiments
		Zone				Habitat
Habitat	Quadrant	Water	Shore	Grass	Experiment	Vegetation	Clear	Chi-sq p
Clear	NE	540	291	112	Outdoor	3266.66	530	<2.2 × 10−16
Clear	SW	526	561	202	Indoor Cold	2090	860	<2.2 × 10−16
Vegetation	NW	4444	1978	331	Indoor Warm	1900	1033.33	<2.2 × 10−16
Vegetation	SE	2485	331	42
	Chi-sq p	<2.2 × 10−16	<2.2 × 10−16	<2.2 × 10−16

). This analysis was repeated on the total count of turtles basking in the vegetation or clear-water habitats (veg-NW + veg-SE counts versus clear-NE + clear-SW counts), which confirmed the significant preference for habitats with vegetation (p < 2.2 × 10⁻¹⁶) (Table 3). Lastly, the chi-squared results for the zones utilized by the females when out of the water revealed their preference to bask on the shore next to vegetated quadrants and in the grass next to the veg-NW quadrant.

The results from the 2024 individual-choice experiments were consistent with the outdoor group results. Specifically, the chi-squared analysis per experimental condition (outdoor, indoor cold, and indoor warm) revealed significant deviations from the null expectation that the turtles would use the vegetation and clear-water partitions equally (p < 2.2 × 10⁻¹⁶, Table 3). Instead, the turtles spent significantly more time on the vegetation side compared to the clear-water side, and this preference was more marked in the outdoor experiment than in the indoor experiments (Figure 6B).

4. Discussion

Individual and population survival is affected by interacting abiotic and biotic factors in their habitat. Thus, understanding the effects of environmental alterations on individual physiology and behavior, which ultimately leads to their reproductive success (or failure), is crucial to ensuring species persistence. Previous work on the effect of environmental incubation temperature on sex ratios has documented the well-known threat of extreme population feminization or masculinization faced by taxa exhibiting temperature-dependent sex determination [18,69,70,71,72]. But while rising global temperatures will increase nest temperatures, other aspects of the environment are also altered by warmer temperatures and represent indirect challenges imposed by climate change on turtle biology. This is the first study, to the best of our knowledge, to investigate a potential indirect impact of global warming on the circulating vit-D levels of reproductive adult female reptiles via the influence of aquatic vegetation abundance in their habitats.

In general, the female painted turtles in our study had comparable levels of vit-D₃ to those reported for wild freshwater turtles (painted and red-eared slider, Trachemys scripta) ranging from yearlings [36] to various age classes [73], but higher than those reported for captive red-eared sliders [38]. The values were also comparable to those reported for sea turtles (Chelonia mydas) [46,49,58,74]. In contrast, several Testudo tortoises (T. hermanni, T. graeca, and T. marginata [75,76]) and an individual box turtle (Terrapene carolina) [77] exhibited substantially higher vit-D₃ levels, likely due to their more terrestrial habits. Little consensus exists in the literature about what constitutes an appropriate daily intake of vit-D₃ or a healthy plasma vit-D₃ concentration in turtles, despite the importance of information on reference intervals for veterinary care and wildlife monitoring [78]. Reference intervals of 56.9 nmol/L (43.2–57.9) for Hermann’s tortoises and 15.6 nmol/L (11.4–19.0) for red-eared sliders have been reported [38], underscoring the variability across chelonians.

At first sight, the results of our two-month study might suggest that turtles’ vit-D levels are resilient to the presence of abundant floating aquatic vegetation that obstructs solar radiation (Figure 5), and thus UV light, from reaching the animals compared to clear-water habitats. Specifically, any differences in vit-D₃ levels between treatments observed in the raw data disappeared after Bonferroni correction (at adjusted alpha = 0.004) once the data were Box–Cox-transformed for normality and the effect of female body size was accounted for. Even when accounting for the dilution factor of our samples due to lymph contamination (relative values), no significant effects were detected between treatments (Table 2A). This lack of differences between treatments is somewhat surprising, because temperature affects the rate of vit-D synthesis in the body [45], and the duckweed-covered pond was significantly cooler than the clear-water pond, such that females basking aquatically under vegetation experienced a colder habitat that was expected to dampen vit-D production [45] and exacerbate any effect of reduced UVB exposure in this treatment. However, this was not the case. Because turtles are capable of behavioral thermoregulation but we did not measure their body temperatures or metabolic rates directly, we cannot rule out the possibility that the turtles may have mitigated temperature differences by adjusting their depth in the water column during the outdoor experiment (all of the animals experienced identical thermal conditions during the subsequent laboratory phase).

The one effect that we observed consistently was the overall reduction in vit-D₃ at day 60 in both treatments (Table 2B), perhaps due to the decreasing photoperiod in the fall [79]. Specifically, the turtles received 14.2 h of daylight outdoors at 0 d (6:13 AM–8:26 PM), which decreased gradually to ~13 h by 30 d (6:45 AM–7:39 PM), and instantaneously to 12 h/day of UBV thereafter when they were moved indoors until 60 d. Any seasonal vit-D₃ reduction could have been exacerbated if the UVB bulbs failed to replicate the full spectrum and intensity of natural sunlight. Specifically, vit-D synthesis is most efficient with exposure to UVB-1 (particularly between 290 and 295 nm), which is blocked by the ozone layer in nature [80]. However, lower vit-D₃ levels are reported consistently for captive turtles [58,76] and iguanas [81] under commercial UV lamps for reptiles that emit UVB-1 but have lower overall irradiance [80], compared to individuals exposed to natural sunlight (which has high intensity and a broad UVB-2 range of 305–315 nm), suggesting that the irradiance and spectral distribution of solar radiation matter more than the presence of UVB-1 alone. This notion is consistent with the steady vit-D levels observed during the outdoor experiment when the photoperiod shortened gradually.

While floating vegetation certainly reduces light penetration in the water column (Figure 2), there are factors that may have contributed to the lack of substantial differences in vit-D₃ levels between treatments, whereas others may be ruled out. For instance, the possibility that UV could pass through instead of being blocked by the duckweed cover is unlikely, since the biochemical properties of the duckweed’s waxy cuticle protect against UV radiation [82]. We should note that because we quantified full solar radiation instead of UVB radiation directly, our ability to infer UVB exposure is somewhat limited. On the other hand, duckweed provides shelter for small fish and aquatic invertebrates containing vit-D₃ that the turtles might have predated on [83,84], supplementing the dietary vit-D₃ sources during the outdoor experiment.

Additionally, our experiment may have been too short to elicit a significant change in vit-D₃ levels between treatments. For instance, monitor lizards deprived of both dietary and UVB-synthesized vit-D₃ experienced a gradual decline in plasma vit-D₃ over four weeks [85]. Given that our turtles consumed some dietary vit-D₃ (some provisioned and perhaps some naturally), any reduction in vit-D₃ would have been slow, potentially preventing the detection of significant treatment effects within two months, particularly if the female vit-D₃ levels were high at the onset of the experiment [85]. Consistently, the vit-D₃ levels detected in our study were higher than those reported for other freshwater turtles [38]. Furthermore, whether the females tapped into any vit-D3 stored in their fat tissue to maintain relatively steady circulating vit-D levels remains unclear.

Another question addressed here was whether turtles respond behaviorally to increasing macrophyte cover in their habitat by avoiding areas that are heavily covered by floating vegetation. That is, perhaps the females in the 2022 experiment basked preferentially in gaps in the pond cover where the floating vegetation had dispersed, as occurred daily by the wind, thus receiving more sunlight than anticipated. Counter to this hypothesis, our habitat choice results showed that female turtles strongly preferred being under the water covered by vegetation both during the 2023 outdoor group choice experiment and during the 2024 individual-choice experiments, regardless of condition (outdoor, indoor cold, or indoor warm) (Figure 5). These results indicate that the decision to opt for vegetation cover is made by each female, as the individual-choice experiments removed the potential for social facilitation [65,66]. Our findings align with experimental reports in other freshwater turtles, such as box turtles, which prefer more naturalistic habitats when given a choice between enriched and unenriched treatments indoors [86,87], and with the preference displayed by yellow mud turtles (Kinosternon flavescens) for submerged vegetation in divided experimental pools indoors [88]. Likewise, observational studies documented that freshwater turtles prefer vegetated habitats in the wild near floating algae and plants or by submerged and emergent vegetation compared to clearer water, including Blanding’s (Emydoidea blandingii), painted (Chrysemys picta), map (Graptemys geographica), snapping (Chelydra serpentina), and musk (Sternotherus odoratus) turtles [89,90], native turtles in Florida (Florida softshell—Apalone ferox, red-bellied cooter—Pseudemys nelsoni, Peninsula cooter—Pseudemys peninsularis, striped mud turtle—Kinosternon baurii, and snapping turtle—Chelydra serpentina [91]), Chinese pond turtles (Mauremys reevesii) [92], and two African Pelusios species [93]. To the best of our knowledge, this is the first study to experimentally evaluate turtles’ preference for habitats with floating vegetation. These results counter the notion that painted turtles prefer clear water to maximize their sunlight exposure for vit-D synthesis, as they consistently chose the vegetated habitats across the experiments, where their exposure to sunlight was significantly lower. This behavioral pattern may be a response to benefits afforded by vegetated habitats, such as thermoregulation, food acquisition, and predator avoidance.

Notably, while there is currently no substantial evidence that reduced UVB exposure under overgrown floating vegetation significantly impacts turtles’ health, it is premature to conclude that no ill effects exist, and future research is warranted to answer the many questions that remain outstanding. Specifically, it remains essential to investigate habitat features or climate phenomena (at various times of the year) that may have complex effects in ecological communities that freshwater turtles clearly prefer and benefit from, such as future rising water temperatures, more frequent and intense storms, prolonged flooding, and high water turbidity from nutrient overload. This includes early spring, when light levels are lower but UVB exposure may be particularly important to breeding females, whose vit-D needs are higher during vitellogenesis. While floating macrophytes are expected to increase under warmer conditions, the composition of aquatic communities may shift, partially because submerged plants and phytoplankton rarely coexist at high biomass levels, due to light competition [94,95,96,97,98], leading to the dominance of one or the other. Floating macrophytes may coexist with phytoplankton, particularly in eutrophic systems; however, excessive phytoplankton growth can reduce water clarity and hinder the early-season establishment of rooted aquatic plants [96]. For instance, when warming occurs earlier in the spring and nutrient levels are high, phytoplankton may outcompete larger aquatic plants, preventing the growth of macrophytes [99]. Under typical eutrophication conditions, phytoplankton tends to dominate, often suppressing macrophyte recovery through increased turbidity and limiting sunlight penetration; these effects vary widely by season [94,95,98]. Increased nutrient loading further favors algal growth, which reduces plant cover [94,97]. Macrophytes serve as a direct food source for turtles and provide an essential habitat for invertebrates and small aquatic vertebrates, such as fish and amphibians, which turtles also consume. Consequently, as macrophytes decline due to phytoplankton overgrowth, both primary and secondary food sources for turtles might be reduced. Additionally, a reduction in submerged vegetation lowers dissolved oxygen levels contributed by its photosynthesis, and excessive plant growth also decreases dissolved oxygen due to respiration and decomposition, impacting aquatic organisms, further decreasing the food available for turtles, and decreasing the oxygen that turtles use for respiration while submerged [100,101].

5. Conclusions

The results from our study revealed no effect of overgrown floating vegetation cover on vit-D levels in reproductive female painted turtles over two months (one month outdoors and one month indoors). While this preliminary study suggests that turtles might be resilient to changing environmental conditions, several factors other than the primary variable (clear water versus vegetation cover) may have contributed to the lack of differences between treatments in our study, and so it is premature to make conclusive statements about this potential indirect effect of climate change based on the scant existing data. For instance, an outstanding question is whether the females may have tapped into their vit-D stores during our short study. Therefore, experiments of longer duration, in different seasons, that study males and females of various ages and monitor vit-D levels in fat tissues, are needed to test whether ill effects may exist for growing juveniles or during the vitellogenic period of adult females, when vit-D requirements for calcium metabolism might be highest.

Our results indicate that turtles’ use of floating vegetation may be associated with benefits related to thermoregulation, predator avoidance, and/or possibly food acquisition, rather than a strong selection for habitats that maximize UVB exposure. Although no conclusive evidence exists for a tradeoff between reduced UVB exposure, leading to decreased circulating vit-D levels in reproductive females, and the potential advantages associated with floating vegetation, further research is warranted to investigate the physiological costs (if any) of this preference, to better understand the benefits of aquatic basking, which remain understudied, by uncovering the complex relationships between the ever-changing habitat and turtles’ physiology and behavior. This information will be useful for conservation efforts to determine management criteria that benefit the taxa residing in these aquatic habitats.