Body temperature stability achieved by the large body mass of sea turtles

Large reptiles, including giant dinosaurs, might have had relatively constant body temperatures (Colbert et al., 1946; Bakker, 1972; McNab and Auffenberg, 1976; Barrick and Showers, 1994). Theoretical simulation has raised the interesting idea that large reptiles might maintain a high body temperature as a result of large size alone (Spotila et al., 1973; Stevenson, 1985). Seebacher et al. (Seebacher et al., 1999) have previously used field data to demonstrate that high and stable body temperatures of land-living crocodiles are driven primarily by physical relationships between body temperature and environmental temperature. Water, in comparison with air, places much tighter constraints on thermoregulation in aquatic animals, owing to its high heat capacity and high thermal conductivity, which leads to a rapid transfer of heat from a warm animal to cold water. Thus, water strongly limits the warming effect of metabolism in aquatic living animals.

Sea turtles spend almost all their time under water, and their range of both vertical and horizontal movements are large. To substantiate their thermal characteristics under natural conditions, long and continuous measurements in the sea were needed. Some aspects of body temperature of free-ranging turtles during the internesting period have been revealed previously using animal-borne recorders (Sakamoto et al., 1990; Sato et al., 1994; Sato et al., 1995; Southwood et al., 2005). Core body temperatures of loggerhead turtles (Caretta caretta Linnaeus 1758) are higher than the water temperatures throughout their internesting periods (Sakamoto et al., 1990; Sato et al., 1994), and the mean temperature differences between core body and water varies from 0.7 to 1.7°C (data from 15 turtles), with larger animals having a significantly higher mean difference (Sato et al., 1998). Despite this, body temperature followed long-term (>24 hours) fluctuations in water temperature with a lag of 2–3 hours, although body temperatures do not respond to sudden changes in ambient water temperature when they dive in the ocean (Sato et al., 1994). The development of a dynamic heat transfer model will be informative in order to understand the unsteady correlation between body and water temperatures.

Here, I investigate the non-equilibrium thermodynamics of body temperature in free-ranging adult loggerhead turtles (56–118 kg) by applying the continuous temperature measurements of body and water to a dynamic heat transfer model. In addition, I make allometric comparisons of the temperature difference between body and water with data from other species of sea turtle. Finally, the ecological implications of body temperature on the life history of the turtles are discussed.

Fig. 1 shows a long-term relationship between the measured and calculated body temperature of a turtle (ID 9305) over an internesting period of 21.0 days. A radius of a spherical model for the turtle was calculated to be 0.25 m, and the thermal diffusivity and heat production rate were taken to be 4.5×10⁻⁷ m² s⁻¹ and 1.5×10² J s⁻¹ m⁻³, respectively. The calculated body temperature coincided well with measured body temperature (a coefficient of determination=0.94) and had a low mean residual (0.13°C). As shown in Table 1, the coefficients of determination between the measured and calculated body temperatures were high (0.79–0.98) and the mean residuals were low (0.12–0.33°C) in all turtles.

Vertical movements of turtles led to them being exposed to sudden changes in ambient water temperature; however, the measured body temperature did not follow such short-term (<90 min, maximum dive duration) fluctuations. One example of a turtle (ID 9305, 69 kg) is shown in Fig. 2. The calculated core body temperature for a spherical model (69 kg, r_R=0.25 m) fitted with the measured body temperature, which remained constant despite rapid changes in water temperature. Assuming smaller spheres, 10 kg (r_R=0.13 m) and 1 kg (r_R=0.06 m), with the same thermal diffusivities and heat production rates, the stability of their core body temperature was not

maintained, and the calculated body temperature followed the rapid changes in water temperature (Fig. 2).

The measured body temperatures of three turtles rose 2.6–5.1°C in 107–152 min during nesting behaviours on land (Fig. 3). The rate of change of body temperature was higher during digging the chamber and covering the nest than it was during landing and egg laying (Fig. 3A). The mean heat production rate that caused the rapid rise in body temperature was estimated for each turtle, assuming the same value of thermal diffusivity in the sea. The estimated heat production rates on land were 7.4–10.5 times the level of those for the same turtle in the sea (Table 1).

The body temperature stability of large reptiles, including dinosaurs, has already been described by several researchers as gigantothermy, thermal inertia or inertial homoiothermy (Frair et al., 1972; Spotila et al., 1973; Neill and Stevens, 1974; McNab and Auffenberg, 1976; Paladino et al., 1990). In the present study, the unsteady thermodynamic analysis was applied to measured data obtained from adult loggerhead turtles under natural conditions, and body temperature stability was not attained when assuming small body masses such as 10 and 1 kg (Fig. 2). This result leads me to conclude that adults could achieve their body temperature stabilities through their large body masses.

The observed responses of body temperature to long-term fluctuations in ambient water temperature were mostly explained by a hypothesis of a constant thermal diffusivity and a constant heat production rate (Fig. 1). The relationships between measured and calculated body temperatures had high coefficients of determination (0.79–0.98 in Table 1). Some individuals (ID 9101, 9102, 9103, 9201) had relatively lower coefficients of determination, which might be due to low sampling rate (Table 1). The others had high coefficients of determination (0.88–0.98), indicating that more than 88% of the variance in the body temperature was explained as dynamic heat transfer under constant thermal diffusivity and heat production.

During the internesting period, turtles repeat dives. The dives of loggerhead turtles have been classified into several types based on the time–depth profile, and some dominant types included three phases in each dive: (1) first descent, (2) gradual ascent, and (3) final ascent (Minamikawa et al., 1997). The turtles swim during the descent and ascent phases but stay at a certain depth without swimming during the gradual ascent phase. Turtles seem to be neutrally buoyant and rest for a large proportion of their dives in the middle of the water column (Minamikawa et al., 1997; Minamikawa et al., 2000). These inactive behavioural patterns of loggerhead turtles during the internesting period support the constant thermal diffusivity and heat production findings in the results of the present study. Thus, physiological thermoregulation does not appear to play an important role in determining the body temperature of adult female loggerhead turtles during their internesting periods. This conclusion is conspicuous in comparison with other aquatic animals.

Holland et al. (Holland et al., 1992) show that swimming Bigeye tunas actively regulate body temperature through a combination of physiological and behavioural means; therefore, these fish can raise the whole-body thermal conductivity by two orders of magnitude to allow rapid warming when they ascend from cold water into warmer surface waters, the reverse taking place when they return to the depths. Handrich et al. (Handrich et al., 1997) have also reported that diving king penguins decrease their abdominal temperature by more than 10°C, which would lead to a metabolic depression and might help to explain the long dive duration of these endotherms. The decline in body temperature in king penguin may be the result of an increase in heat loss and/or a local metabolic depression. In both the cases of the fish and the bird, distinct changes are needed in the physiological condition to explain the observed changes in body temperature. Although it has been reported that turtles can change their physiological condition according to environmental and behavioural characteristics (Weathers and White, 1971; Heath and McGinnis, 1980; Butler et al., 1984; Smith et al., 1986), a substantial change in thermal diffusivity or heat production rate is unnecessary to explain the body temperature response of loggerhead turtles to fluctuations in water temperature (Figs 1, 2).

The body temperature of the three turtles rose 2.6–5.1°C during a series of nesting behaviours on land (Fig. 3). In 1980, Mrosovsky discussed the body temperature of nesting turtles using the available data at that time and predicted that metabolic heat production may warm up nesting turtles less than 2°C (Mrosovsky, 1980). My data support the prediction qualitatively, but the scope of body temperature change was higher than the prediction. According to the numerical simulation, higher heat production rates are needed to explain the rapid rises in body temperature during nesting behaviour on land. Actual heat production rates on land are likely to be even higher than the estimations because I assumed that there was no evaporative water loss and that surface body temperatures were equal to air temperatures. Despite this, an increase in heat production of at least 7.4–10.5 fold above that in water would be needed to explain the rapid rises in the core temperatures of turtles nesting on land. The elevated heat production rates are probably due to the hard exercise during nesting behaviours, especially digging the chamber and covering the nest. The heat production rate is likely to be greatly affected by the behaviour of turtles, and it appears that loggerhead turtles are similar to green turtles in that they are able to raise their metabolism by a factor of 10 when they are active (Prange and Jackson, 1976; Jackson and Prange, 1979).

Southwood et al. (Southwood et al., 2005) have measured the subcarapace temperature of three leatherback turtles during internesting periods and reported that their body temperatures were maintained higher than ambient water temperature. The range of differences between the body and water temperatures of leatherback turtles (1.2–4.3°C) was similar to the prediction using the model (Table 2). Bradshaw et al. (Bradshaw et al., 2007) have previously estimated the diving metabolic rate from dive duration and depth data that was collected for nine free-ranging leatherback turtles over long periods (181–431 days). The behaviourally derived diving metabolic rates were close to the predicted field metabolic rate for a reptile of equivalent size and were nearly an order magnitude lower than the field metabolic rate predicted for a mammal of equivalent size (Bradshaw et al., 2007). Considering these previous studies and the present study, the relatively higher temperature gradient of leatherback turtles during internesting periods can be attributed mainly to their larger body sizes. A recent published paper reported interesting records from leatherback turtles foraging in the Northwest Atlantic Ocean (Casey et al., 2014). The mean body temperature of leatherback turtles (391–589 kg) ranged from 25.4 to 27.2°C, and the mean temperature gradient between body and water ranged from 10.0 to 12.2°C (Table 2). These values are 3.6–4.6 times larger than the values calculated from their body masses using Eqn 2. Thus, leatherback turtles swimming in high-latitude cold water (13.6–15.9°C) seemed to have different physiological conditions from turtles around nesting grounds.

As discussed in a previous paper (Schofield et al., 2009), shorter internesting intervals might have some advantages for female adult loggerhead turtles – they can begin nesting earlier in the year and generate more clutches to be incubated when sand conditions are optimal during the summer. Loggerhead turtles breeding at their northern margin (Greek island of Zakynthos: 37.7°N, 20.9°E) seem to search for small patches of warm water in order to make raise their body temperature (Schofield et al., 2009). My study sites (33.8°N, 134.7°E; 33.8°N, 135.3°E) are situated in the middle of Japanese nesting grounds of this species, and the available nesting season may not constrain their reproductive output. However, shorter internesting intervals might have another advantage because reducing the total time required to lay all clutches per season will contribute to turtles minimizing the time that they spend away from their foraging areas. Although adult loggerhead turtles are known to have some endothermic capacities during internesting periods (Sato et al., 1995), there is no evidence that they have higher rates of metabolism than that for reptiles of similar mass. Indeed, it would be less advantageous to have a higher metabolic heat production rate in order to elevate body temperature because these animals apparently do not feed actively during internesting periods (Tanaka et al., 1995), and energy reserves might constrain their reproductive output. Thus, adult loggerhead turtles during internesting periods seem to benefit from a passive thermoregulatory strategy, which depends primarily on the physical attributes of their large body masses, rather than physiological mechanisms.

Loggerhead turtles make several serial nests on the same beach at 13–25 day intervals on the Japanese nesting grounds (Sato et al., 1998). This regular pattern of reproduction is ideal for the deployment and retrieval of data loggers on turtles. Field studies for this work were conducted at nesting beaches in the Japanese archipelago in 1989 and from 1991 to 1994. We attached data loggers onto the carapaces of turtles to record water temperature, depth and light intensity during their internesting periods, and induced animals to swallow units to record core body temperature (Sato et al., 1995; Sato et al., 1998). Body and water temperatures were simultaneously recorded from a total of 16 turtles, in three of these body temperature was also measured during nesting behaviour on land (Table 1). The sampling interval was 1 min, except for four individuals where intervals were 5 or 10 min. Data were recorded for between 1.9 and 21.0 days (Table 1). The body mass of turtles was measured for each individual on the beach using a hanging scale and a net, and these data were used in the mathematical analysis in this paper.

The recapture ratio for turtles with data loggers was 0.69 (N=35), which is almost same as the ratio of tagged turtles without data loggers, 0.68 (N=66). It is unlikely that the attachment of data loggers led turtles to avoid the nesting beach or cease the next nesting. All experimental procedures were approved by a board of education in Minabe town, Wakayama Prefecture, Japan.

The first purpose of the analysis was to quantify the effect of the large body mass of the animal on the core body temperature. To do this, the model animal was regarded to be a sphere, this being the simplest three-dimensional shape. The diagrammatic representation of the model is shown in Fig. 5A. The radius (r_R) of the considered spherical model was calculated from the known body mass of each animal with a density taken to be ρ=1046.5 kg m⁻³. The density was measured in an aquarium using two green turtles, weighing 42.7 kg and 71.5 kg, and loggerhead turtles were assumed to have equivalent densities.

Solar radiation is an important parameter for analyzing the body temperature of reptiles on land, but adult loggerhead turtles in water do not benefit from solar radiation as an external heat source (Sato et al., 1995). Thus, only metabolic heat production is considered as a heat source in the model (the second term of the right hand side of Eqn 4).

Eqn 4 was substituted into the differential equation for each layer with Δr=1 cm and Δt=20 s being used as finite difference. At the beginning of a calculation (t=0), a uniform temperature=T_b(0,0), initial core body temperature, was assigned throughout the body (r=0 ~r_{R – Δr}). The measured water temperature was assigned as surface body temperature T_b(r_R, t) because surface body temperatures were nearly identical to water temperature in an experiment using captive turtles (K.S., unpublished data). Initially (t=Δt), T_b(r_R–Δr, Δt) was calculated from the differential equation, then, T_b(r_{R – 2Δr}, Δt) was calculated with decreasing value of r until reaching core body temperature T_b(0, Δt). The simulation was then run to the next time period (t=2Δt). After computation over extended periods, the calculated core body temperature was compared with the measured body temperature of turtles.

Fig. 5B,C shows the general relationship between fluctuating surface temperature T_b(r_R, t) and calculated core body temperature T_b(0, t) of a febrifacient sphere. The core body temperature is higher than surface temperature and lags behind the surface temperature fluctuation. In addition, the range of fluctuation of core temperature is less than that of the surface temperature. These phenomena qualitatively coincide with that described previously in free-ranging loggerhead turtles (Sato et al., 1994). Where core body temperatures were calculated with high value for thermal diffusivity, the lag between surface temperature and core body temperature became smaller (Fig. 5B).

The model simulation was run several times using several values for thermal diffusivity, and the calculated core body temperatures were compared with the measured body temperature of a turtle. The thermal diffusivity, under which the smallest coefficient of determination between measured and calculated core body temperatures was calculated, was assumed to be the most appropriate value for the turtle.

Thereafter body temperatures were calculated using an appropriate thermal diffusivity and several heat production rates. Differences in the heat production rate made the core body temperature lower or higher but did not affect the time lag between core and surface body temperatures (Fig. 5C). The heat production rate was assumed to be appropriate for the turtle when the mean residual between the measured and calculated temperatures was smallest. Both values, χ and Q, were determined for each turtle using the same procedure (Table 1).

Although my simple model enabled me to answer some biological questions, it is important to appreciate that the thermal diffusivities assumed to be appropriate for each turtle contain an effect of transformation of actual body shape to a sphere in the model. The thermal diffusivity χ values used for the turtles varied between 3.2×10⁻⁷ and 7.7×10⁻⁷ m² s⁻¹ (Table 1), which can be transformed to the thermal conductivities K of 1.2–2.9 J s⁻¹ m⁻¹ °C⁻¹ using Eqn 5. These values are greater than the expected level for turtle tissue by one order of magnitude, e.g. human muscle 0.46 J s⁻¹ m⁻¹ °C⁻¹ and adipose tissue 0.21 J s⁻¹ m⁻¹ °C⁻¹ (Schmidt-Nielsen, 1990). The thermal diffusivity χ and thermal conductivity K used in this paper represent the degree of the thermal insulation, including the effect of the actual body shape of animals, which has a larger surface area in comparison with a sphere.

The author wishes to thank Yasuhiko Naito and Wataru Sakamoto for their helpful advice to complete this study. The author acknowledges Tateki Fujiwara for providing significant feedback on data analysis using the thermodynamic model. The field study was conducted in cooperation with Hideji Tanaka, Yoshimasa Matsuzawa, Shingo Minamikawa, Kiyoshi Goto and many volunteers. Graeme Hays and an anonymous reviewer gave valuable comments on the manuscript.