Despite the importance of the thermal environment to physiological processes and, ultimately, energy acquisition and expenditure in ectothermic vertebrates (Huey 1982; Congdon 1989; Peterson et al. 1993), an understanding of body temperature variation of free-ranging freshwater turtles has only recently emerged (Edwards and Blouin-Demers 2007; Rowe and Dalgarn 2009). Studies of body temperature (T_b) variation in freshwater turtles focused on individuals while aquatic (Edwards and Blouin-Demers 2007). However, an understanding of T_b variation during terrestrial nesting forays in north-temperate freshwater turtles would seem important because timing and activities associated with nesting may be determined by the thermal environment (Congdon and Gatten 1989). Midland painted turtles (Chrysemys picta marginata), for instance, initiate terrestrial nesting movements in late morning or afternoon and nest in open areas at distances of up to > 400 m from the aquatic home range (Christens and Bider 1987; Congdon and Gatten 1989; Baldwin et al. 2004; Rowe et al. 2005). Because nesting forays can last a few hours to days in northern latitudes (Rowe et al. 2005), C. p. marginata could face extreme fluctuations in environmental temperatures (Blouin-Demers and Weatherhead 2001; Isaac and Gregory 2004). Although a minimum T_b value would be required to sustain nesting movements (Congdon and Gatten 1989), individuals must also avoid high and potentially lethal T_b values (Burke et al. 1994). C. p. marginata presumably maintain T_b within an acceptable range through the use of sunny and shaded areas (Rowe et al. 2005). In fact, individuals may seek terrestrial refugia during the afternoon before nesting, presumably to avoid overheating (Congdon and Gatten 1989; Rowe et al. 2005).

Environmental temperature constraints on T_b, the risk of predation by mammalian predators, and the location of nesting sites with suitable thermal characteristics (Weisrock and Janzen 1999) probably interact to determine timing and behavior of C. p. marginata during terrestrial nesting forays (Congdon and Gatten 1989). Nesting relatively far from an aquatic home range may reduce nest predation by increasing search times for predators (Spencer 2002; Kolbe and Janzen 2002; Spencer and Thompson 2003). However, long periods of time on land may increase exposure of females to predators (Spencer and Thompson 2003) and so behaviors that minimize predation rates (e.g., use of terrestrial refugia and efficient navigation) would be expected to evolve (Rowe et al. 2005). Up to 75% of C. p. marginata spent the night in terrestrial refugia after nesting (Christens and Bider 1987; Congdon and Gatten 1989; Rowe et al. 2005), possibly because low evening T_b impedes mobility (Isaac and Gregory 2004; Rowe et al. 2005) or individuals were unable to navigate to water under low light intensity (Bowne and White 2004; Rowe et al. 2005) or both.

We studied T_b variation in nesting C. p. marginata at a small north-temperate marsh in northern Michigan. We used radiotelemetry and surgically implanted thermally sensitive radio transmitters to record T_b of nesting individuals before, during, and after nesting forays. First, we expected that mean T_b of mobile individuals during nesting forays would be similar to aquatic activity T_b at that time of year (1 week before and after the nesting foray) and would exceed ambient air temperature (T_a) as the result of heating by the sun or substrate. Second, turtles that sought terrestrial refugia before nesting later that day would have T_b lower than T_b of, or T_a associated with, mobile and terrestrial turtles. Third, mean T_b of nesting individuals would be similar to aquatic activity T_b but would decline over time because most individuals at Miller's Marsh finish nesting after dark (Rowe et al. 2005). Finally, if low T_b is responsible for inducing individuals to seek terrestrial refugia after nesting (Congdon and Gatten 1989), then T_b values should be relatively low, 10°–20°C, a temperature range that is well below the preferred T_b range (21.3°–25.0°C) determined for C. p. marginata (Edwards and Blouin-Demers 2007) and within which diminished terrestrial locomotor capacity occurs (Elnitsky and Claussen 2006).

METHODS

We collected turtles by hand, basking traps, or hoop nets. Turtles were sexed, and the reproductive status (ovigerous or nonovigerous) of females was determined by palpation. Temperature-sensitive, calibrated (1°–45°C) radio transmitters (2.8 g; Advanced Telemetry Systems) with unique radio frequencies were implanted in the peritoneal cavities (Rowe and Dalgarn 2009) of 4 females between 28 May and 7 June 2003, and 5 females between 2 and 26 May 2004. Of the original 9 radio tagged turtles, T_b of 8 individuals (mean CL  =  140.9 ± 13.25 mm, 50–170 mm; mean mass  =  433.125 ± 29.17 g, 360–620 g) was recorded during nesting forays.

Handheld radio receivers and antennae were used to radio locate each turtle from at least 3 directions while attempting to maintain a distance of several meters from the turtle. Before an individual's nesting activity, radio locations of aquatic individuals were made approximately every 2 hours each day. Upon emergence from the marsh, individuals were radio located every 0.5 to 1 hour. Locations of nest and refugia sites were marked with flags and later recorded by using a Trimble global positioning system unit. At the initiation of the nesting process, pulse intervals (time between radio pulses in milliseconds) were measured by using a pulse interval timer (AVM Instruments) or by tallying radio pulses per minute at 15–30-minute intervals. For individuals that remained in terrestrial refugia after nesting, T_b data were recorded at 2- to 3-hour intervals between 0000 and 0600 hours and then hourly thereafter until the turtle returned to the marsh. To measure aquatic T_b, a Telonics TR5 radio receiver was programmed to record pulse intervals for each radio transmitter frequency at 15-minute increments for a maximum of 96 daily observations per individual. Pulse intervals or pulses per minute were converted to T_b values by using regression equations (Rowe and Dalgarn 2009). As a relative measure of environmental temperatures, shaded air temperature (T_a) was measured by using an Optic Stowaway data logger (Onset Computer Corporation) at a location that was 15 m from the marsh edge and at a height of 1 m.

We analyzed differences in T_b among nesting foray activities (prenesting mobile, prenesting refugium, nesting, postnesting refugium, and postnesting mobile for the day of nesting and the day after nesting) and while individuals were aquatic 1 week before and 1 week after an individual's nesting foray. Mean T_b values per individual for each activity were derived from mean hourly T_b values determined during each activity or while aquatic to minimize biases that would result from differences in sampling frequency at various times during the day. We analyzed T_b by analysis of variance (ANOVA) with activity (mean T_b 7 days pre- and postaquatic and nesting foray activity) included as a main effect and individual turtle identification number included as a random variable to account for autocorrelation. We also used ANOVA to assess environmental temperature (T_a) variation during aquatic and terrestrial nesting foray activities of each turtle. Statistics were considered to be significant at the 95% level.

RESULTS

Turtles began nesting forays during late afternoon, and most remained on land overnight after nesting. All turtles began nesting forays between 1500 and 1900 hours, initiated the nesting process between 1800 and 2015 hours (mode  =  1915 hours, n  =  7 turtles), and terminated nesting between 1900 and 2315 hours (mode  =  2015 hours, n  =  8). Nesting duration averaged 105.0 ± 27.58 minutes, 45–255 minutes (n  =  7). Three turtles traveled through woodlands, and 5 passed through dense sedge-cattail and grassy areas on their way to nesting sites. Three individuals emerged from the marsh, nested, and returned to the marsh between 2110 and 2345 hours the same day. Three individuals emerged from the marsh, nested, and then spent the night under terrestrial leaf litter to return to the marsh between 1130 and 1330 hours the following day. After nesting, 1 individual remained under leaf litter at 2 separate locations for 2 consecutive nights but was killed by a predator on the second night. Finally, 1 individual spent 2 nights under leaf litter before nesting and then spent another night in leaf litter before returning to the marsh at 1415 hours on the fourth calendar day of its nesting foray. All turtles nested in locations that were in full exposure to the early evening sun. The nest-to-water distance averaged just over 170 m and pre- or postnesting leaf litter refugia sites were relatively close to the nesting location when compared with their proximities to water (Table 1).

Table 1 Distances (m) among nests, refugium sites, and the marsh edge for Chrysemys picta marginata, 2003–2004.

View Fullscreen

For individuals on nesting forays, T_b largely paralleled T_a throughout the day. Based on 268 individual measurements, T_b was highly variable and ranged between 9.9°C and 30.2°C. Both T_b and T_a were roughly equivalent during the early morning and afternoon hours, but mean T_b exceeded mean T_a during the late afternoon and early evening hours (Fig. 1).

View Figure

• Download as Powerpoint
• Download Figure

Body temperatures of active turtles on nesting forays and while aquatic 1 week before, or after, nesting forays were higher than in terrestrial turtles beneath leaf litter. ANOVA (F_6,34  =  6.9, p < 0.0001, r²  =  0.75) and post hoc comparisons indicated that the mean T_b of aquatic individuals was similar to mean T_b while active and terrestrial (mobile on their way to, or from, a nesting site, or while nesting; p > 0.05 in all comparisons; Fig. 2). Mean T_b of individuals in terrestrial refugia (before and after nesting) was significantly lower than during other terrestrial activities (p < 0.05 in all comparisons; Fig. 2).

View Figure

• Download as Powerpoint
• Download Figure

Body temperature tended to decline during the nesting process. Mean T_b of individuals for which we recorded T_b at both the initiation and termination of the nesting process (n  =  7 individuals) was 23.5° ± 4.55°C (16.8°–28.4°C) and 20.0° ± 1.43°C (13.4°–23.9°C), respectively. Mean T_b between the initiation and termination of the nesting process (3.5° ± 0.98°C, −1.6°C to 6.8°C) was significantly greater than zero (Wilcoxon signed-ranked matched pairs T_{df  =  6}  =  13, p_one-tailed < 0.05), which indicated that T_b declined during the nesting process. Similarly, the mean T_a at the initiation of nesting was 17.9° ± 1.37°C, 14.1°–22.9°C and at the termination of nesting was 16.4° ± 1.28°C (11.7°–21.0°C), a significant decline (mean difference  =  1.5 ± 0.31°C, 0.3°–2.4°C; Wilcoxon signed-rank T  =  14, p_1-tailed < 0.01).

Mean T_b of individuals at the initiation of terrestrial use of refugia after nesting was 19.9° ± 1.66°C (19.8°–23.6°C, n  =  5 individuals), mostly within the range of aquatic activity temperatures. The mean T_b of 4 individuals that were measured within 15 minutes of emergence from terrestrial refugia between 1100–1200 hours the day after nesting was 16.9° ± 2.53°C (12°–22°C).

DISCUSSION

Body temperatures of C. p. marginata on nesting forays at Miller's Marsh varied with variation in both environmental temperatures and with the phase of the nesting foray activity. Our turtles emerged during late afternoon and traveled through highly vegetated areas (forests or grass-sedge-cattail) and remained on land for several hours before nesting. The relatively large proportion of turtles that remained on land after nesting (75%; Rowe et al. 2005; 62.5% in this study) relative to that found in other studies (< 25%; Congdon and Gatten 1989; Christens and Bider 1987) probably occurred because of relatively long water-nest distances (Rowe et al. 2005). As a consequence, terrestrial individuals faced a broad range of diel environmental temperatures that presumably influenced T_b values as evidenced by our observed coincidental oscillations of T_b and T_a values over time.

Active (mobile or nesting) C. p. marginata during terrestrial nesting forays attained a mean T_b that was similar to a mean T_b of individuals while aquatic and well below the critical thermal maximum reported for C. picta (41.5°–42.3°C, Brattstrom 1965; Ernst 1972). Because the mean T_b of active terrestrial turtles was significantly greater than T_a, T_b was probably elevated by thermal conduction or by basking in sunny spots. Yet, T_b rarely exceeded 30°C, even during the late afternoon when operative temperatures would have been very high (Hertz et al. 1993; Edwards and Blouin-Demers 2007). Our observations, therefore, are consistent with the notion that T_b of mobile turtles is maintained within acceptable limits by use of warm or cool thermal patches during nesting migrations (Rowe et al. 2005). Because females may select nest sites that have favorable thermal conditions for the developing embryos (Morjan and Valenzuela 2001), T_b could also be influenced by the environmental temperatures associated with nest sites or potential nest sites. Unlike C. p. marginata previously studied at Miller's Marsh (Rowe et al. 2005) and in southeastern Michigan (Congdon and Gatten 1989), we did not find that turtles used terrestrial refugia during the afternoon before nesting later that day. Therefore, our prediction that afternoon T_b would be temporarily reduced before nesting later in the day could not be evaluated. Turtles that used terrestrial refugia before nesting did so relatively late in the day, apparently because they did not locate a suitable nesting site on the day of emergence from the marsh. As a result, mean T_b of turtles in “prenesting” and “postnesting” refugia were similar in value because both were measured mainly after dark.

We found that C. p. marginata did not enter terrestrial refugia after nesting in the evening because T_b was too low for locomotion or because light levels were too low for navigation at that point in time (Rowe et al. 2005). As expected, we found that both T_a and T_b declined during the nesting process. In contrast, Congdon and Gatten (1989) found that cloacal temperatures at the termination of nesting were relatively high (mean  =  30°C, 21°–38°C) when compared with our mean value of 20°C. Because turtles at Miller's Marsh nest later in the day and farther from water (Rowe et al. 2005) than in the population of Congdon and Gatten (1989), T_b of nesting turtles at Miller's Marsh would be expected to be relatively low at the termination of nesting. Body temperatures after nesting and when turtles entered terrestrial refugia overlapped with T_b of mobile terrestrial turtles and was well within the range of possible aquatic activity T_b values (Rowe and Dalgarn 2009). The T_b of diurnal turtles as they emerged from terrestrial refugia the day after nesting indicated that turtles were capable of terrestrial locomotion at T_b values as low as 12°C. Therefore, turtles did not enter refugia after nesting because T_b (mean ca. 20°C) was too low for skeletal muscle activity per se (Rowe et al. 2005). Turtles that entered refugia after nesting did so before the onset of dark, and 3 turtles returned to the marsh after the onset of dark. Therefore, low light levels at the time of entry into terrestrial refugia were not the direct cause of individuals remaining terrestrial overnight. Rather, turtles most likely responded to impending low environmental temperatures and low light levels that would occur during their relatively lengthy return to the marsh at a time when they are vulnerable to nocturnal mammalian predators.

Terrestrial activity during nesting forays clearly has its risk of predation (Tucker et al. 1999), even if turtles are concealed in vegetation (Rowe et al. 2005 and this study). Apparently, the risk of predation on slow-moving C. p. marginata with low T_b while attempting to navigate at night outweighs the predation risks associated with remaining in terrestrial refugia overnight (Rowe et al. 2005). We predict that, in populations in which nesting sites are relatively close to the aquatic home range, turtles will attempt to return to water even after dark and at relatively low T_b. Such a prediction would be consistent with our observations that marsh-nest distance of turtles that returned to the marsh on the day of nesting (mean  =  105.3 m, 31–249 m, n  =  3) was about one-half that of turtles that used terrestrial refugia after nesting (mean  =  200 m, 70–381 m). Other costs of remaining terrestrial after nesting would include a lack of foraging opportunities (Burke et al. 1994) and low T_b values that retard digestive processes and, therefore, energy acquisition. However, relatively low T_b of nesting turtles and the low energetic demands of terrestrial individuals relative to aquatic individuals (Stockard and Gatten 1983) should result in a low energy expenditure relative to that required for the construction of a clutch of eggs (Congdon and Gatten 1989). Because C. p. marginata at Miller's Marsh are active from at least mid-April (Rowe, pers. obs.) to late November at Miller's Marsh (Rowe and Dalgarn 2009), we suggest that energetic cost of not eating for 1–3 days would be minimal.