Spatial and Thermal Ecology of Snapping Turtles (Chelydra serpentina) in a Small, Dystrophic Lake in Central Michigan
Abstract
Understanding the interface between spatial and thermal ecology is integral to understanding energy acquisition and the life histories of ectotherms. Snapping turtles (Chelydra serpentina) occupy a wide range of habitats that vary greatly in their thermal properties. We studied activity, movements, and thermoregulation of C. serpentina in a small, land-locked lake in central Michigan, USA, using radiotelemetry. Consistent with our a priori predictions, turtles were active within core areas along the lake's edge, showed both diurnal and nocturnal activity, and did not make extensive interwetland movements. Turtles left the lake only for nesting or to hibernate in Sphagnum peat or in the banks of a nearby stream. Home range and core area size estimates of C. serpentina were small compared with other previously studied populations, perhaps in part because of the small dimensions of our lake. Contrary to our prediction of a broad Tset (thermoregulatory set-point) range for a large-bodied, habitat generalist turtle, we found a comparatively low and narrow laboratory-determined Tset range (22°C–26°C). Turtle body temperatures (Tb) cycled between May and August and attained maximal values during the evening hours, a pattern that likely results from thermal inertia, the selection of aquatic thermal patches, or both, as Te (operative temperatures) declined. Turtles most effectively maintained Tb within Tset during July and August when thermal conditions were most favorable. Throughout most of the active season the highly aquatic habits of C. serpentina apparently negated the effects of variations in daily weather conditions and their effects on incoming levels of solar radiation on variation in Tb. However, turtles maintained higher average Tb on sunny days compared with overcast days in April and May, the coolest months of the study. Diel Tb cycling ceased during September and October and average Tb declined despite favorable thermal conditions, at least during September, a pattern that could reflect a downward shift in the Tset range. Comparatively, C. serpentina is less effective at thermoregulating than is a small-bodied species at a similar latitude (Chrysemys picta marginata). Apparently, thermal inertia and lack of atmospheric basking proclivities influenced the thermoregulatory precision in our turtles.
Spatial and thermal ecology of ectothermic reptiles are inextricably linked because they must exploit thermal patches that allow sufficient body temperatures (Tb) for daily activities from within heterogeneous thermal environments (Huey 1982, 1991). Environmental temperatures permitting, ectotherms may regulate Tb within some optimal limits, largely through behavioral means and ultimately facilitating growth and reproduction (Carrière et al. 2008).
Behavioral thermoregulation in freshwater turtles includes atmospheric basking and the use of aquatic thermal patches (Dubois et al. 2009), although some thermoconforming is not uncommon (Manning and Grigg 1997). High thermal conductance of water may prevent freshwater turtles from attaining their physiological thermoregulatory set-point (Tset) range (Hertz et al. 1993; Christian and Weavers 1996), particularly when atmospheric basking opportunities are limited (Bulté and Blouin-Demers 2010). Conversely, Tb values may be maintained within the Tset range very precisely with little energetic expenditure if permitted by prevailing operative temperatures (i.e., the equilibrium Tb values that could be attained by a nonthermoregulating individual; Bakken and Gates 1975). The degree to which individuals thermoregulate likely varies among turtle species based on species-specific behavioral and other physical attributes. Relatively small-bodied species may warm and cool rapidly (Tilkens et al. 2007), resulting in large Tb fluctuations throughout the day and season (Rowe and Dalgarn 2009, 2010; Rowe et al. 2014, 2017). In large-bodied species, thermal inertia may slow warming and cooling rates, thereby reducing the range of variation in daily Tb (Sajwaj and Lang 2000; Seebacher et al. 2003; Millar et al. 2012) and even limiting the ability to thermoregulate (Bulté and Blouin-Demers 2010). Thermoregulatory set-point ranges themselves may vary in species that live in fine-grained (relatively narrow set-point ranges) vs. those in course-grained environments (relatively broad set-point ranges; Ben-Ezra et al. 2008; Rowe et al. 2017).
The broadly distributed snapping turtle (Chelydra serpentina) is a large freshwater species in North America that occupies a wide range of aquatic habitat types. Long distance movements can exceed several kilometers (Hammer 1969; Steen et al. 2010). Home range size and location in C. serpentina may be approximately stable between years and the core areas within them do not seem to be exclusively defended (Galbraith et al. 1987). Chelydra serpentina is highly aquatic but may elevate Tb by atmospheric basking (Obbard and Brooks 1979; Brown et al. 1990). Laboratory-determined Tset range in hatchling or juvenile C. serpentina approaches 30°C (Schuett and Gatten 1980; Williamson et al. 1989; Knight et al. 1990; Bury et al. 2000), although the Tb of free-ranging adults is relatively low (22.7°C; Brown et al. 1990). Comparative ecological studies of populations that reside in habitats with different spatial and thermal characteristics could reveal local adaptive, or phenotypically plastic, thermoregulatory responses and ultimately lend insights into local and geographic variation in life-history characteristics in C. serpentina (Brown et al. 1994; Iverson et al. 1997).
We studied spatial and thermal ecology of C. serpentina in a small dystrophic lake and its associated northern wetlands in central Michigan using radiotelemetry. We expected that turtles would establish stable home ranges and core areas among years, a pattern that seems to be common in freshwater turtles (Rowe and Dalgarn 2010; Rowe and Horton 2017). Within and among years, home ranges would be expected to overlap among individuals (Galbraith et al. 1987). We predicted that turtles would maintain at least some nocturnal movements (Obbard and Brooks 1981). Our relatively deep lake was unlikely to dry entirely; therefore, we did not expect interwetland movements that would result from drought periods (Steen et al. 2010), but tested for seasonal declines in daily distances moved that may occur in north-temperate freshwater turtles prior to hibernation (Rowe 2003).
Relative to a north-temperate and small-bodied species such as C. picta marginata (Rowe and Dalgarn 2009; Rowe et al. 2014, 2017), we anticipated that our relatively large-bodied C. serpentina would show less variable diel Tb cycling that is less susceptible to variation in weather patterns (and therefore variation in incoming solar radiation levels). We predicted relatively precise thermoregulation within a broad Tset range for C. serpentina, a species that typically occupies course-grained aquatic habitats (Bulté and Blouin-Demers 2010), and more effective thermoregulation during summer than during spring and autumn (Rowe et al. 2017).
MATERIALS AND METHODS
Study Site
Davis Lake is located in Montcalm County, Michigan (43°23′29″N, 84°53′37″W), on the property of the Alma College Ecological Station. Davis Lake (0.86-ha surface area) is located in a basin of glacial origin and is surrounded by a northern wetland Sphagnum sp. mat and forest that is surrounded by a shallow, “moat-like” wetland that is adjacent to a terrestrial mixed deciduous–conifer forest. With a maximum depth of 11–12 m at its center, Davis Lake has a littoral shelf of variable width (< 5 m) that is vegetated with bullhead lily (Nuphar variegata), watershield (Brasenia schreberi), and white water lily (Nymphaea odorata), and that grades down a steep slope into deep water. Immediately adjacent to the open water is a broad Sphagnum mat with leatherleaf (Chamaedaphne calyculata), pitcher plants (Sarracenia purpurea), various ferns, and marsh cinquefoil (Comarum palustre) scattered throughout with a band of pickerelweed (Pontederia cordata) at the Sphagnum mat–open water interface. There is a small marsh 370 m to the east of Davis Lake, a small spring-fed creek 100 m to the south, an artificial lake 1230 m to the east, and a roadside ditch system 445 m to the south.
Radiotransmitter Implantation
Beginning in June 2013, we collected turtles in baited hoop nets, uniquely notched the marginal scutes for individual recognition (Cagle 1939), and measured maximum carapace length (CL; ± 1 mm) using calipers (50 cm, Haglöf aluminum tree caliper; Långsele, Sweden) and mass (± 10 g) with a spring scale (Viking Model 9920; Hanson Scale Company, Subuta, MS). Between 2013 and 2015, we collected 2 females (CL = 250 and 290 mm; mass = 4.2 and 5.0 kg) and 4 males (meanCL = 273 ± 3.8 mm, 265–283 mm; meanmass = 4.8 ± 740 kg, 3.8–7.0 kg). After 24 hrs in captivity to allow gut clearance, we anesthetized turtles with intraperitoneal injections of MS222 (450 mg/kg turtle wet mass) and inserted cylindrical radiotransmitters (Model M1230T, 8 × 2 cm, 25 g; Advanced Telemetry System, Isanti, MN) into their coeloms through a 2-cm incision in the left femoral pocket, anterior to the hind limb. We closed the incisions with resorbable sutures and maintained turtles in a dry plastic bin for 24 hrs to allow the healing process to initiate.
Determination of Tset
We determined Tset ranges for 6 individual C. serpentina in a wet–dry thermal gradient constructed from a 2.4 × 0.8 × 0.6-m galvanized tank that was modified from Rowe et al. (2014, 2017). A single individual was placed at the center of the thermal gradient and allowed 6 hrs of acclimation before we began recording Tb data at 15-min intervals for 48 hrs using a Telonics TR5 radioreceiver and YAGI antenna (Telonics Inc., Mesa, AZ). We considered the middle 50% of the Tb observations, bounded by Tset-min and Tset-max, to be the Tset range (Hertz et al. 1993; Christian and Weavers 1996). We used Tb values recorded between 0600 and 1800 hrs for Tset range calculation because the time frame represented the daylight hours when turtles exploit solar radiation in the field (Tset range = 22°–26°C).
Determination of Operative Temperature
Operative temperatures can be measured by biophysical models that are similar to study subjects in size, shape, and reflectance (Bakken and Gates 1975). We constructed aluminum biophysical models using a C. serpentina shell that we filled with a liquid foam to form a smooth contoured 3-dimensional model and placed it dorsumdown in a wooden frame that was filled with casting sand (a clay–sand mixture) buried up to the level of the plastron and then removed to form an impression of the shell (260 × 225 × 75 mm) that we filled with molten aluminum. We bored a cylindrical cavity (32 mm wide × 20 mm deep) in the central plastron region that was at a right angle to the long-axis of the shell and that terminated approximately one-half way from the top to the bottom of the model. We placed HOBO TidbiT v2 Water Temperature Data Logger-UTBI-001 (Onset Computer Corporation, Bourne, MA) with the top appressed to the terminal surface of the bored cavity, which was lined with silicone heat sink gel (Techspray, 1977-DP; Kennesaw, GA) to facilitate heat transfer. We painted the model with satin black spray paint to approximate the color of our C. serpentina.
We used aluminum as a model medium to reduce heating lag times and thermal stratifications within the models (O'Connor et al. 2000). We tested the validity of our models using a euthanized (lethal dose of MS222) snapping turtle (263 mm CL) that we obtained from a commercial source and that was of the same shell dimensions of our models. Core temperature of the turtle carcass was measured at 10-min intervals by inserting a thermal probe linked to a HOBO U12 Outdoor/Industrial 4-channel unit (Onset Computer Corporation, Bourne, MA) into the cloaca and model temperatures were obtained using Tidbit data loggers. We placed the model and turtle on 0.64-cm-thick plywood that itself was placed on at 40 × 20 brick in a circular wading pool (110 × 20 cm), with water filled to the bottom of the turtle and model to simulate our field placement of our models. We allowed water, model, and turtle temperatures to equilibrate to room temperature (20°C ± 1°C) for ≥ 12 hrs and then allowed the model and turtle to equilibrate beneath 250-watt heat lamps that we placed at 5 fixed distances above the carapaces between 35 and 75 cm at 25-cm increments. Equilibrium temperatures of the model and turtle ranged from 23.5°C to 39.8°C and from 23.1°C to 33.2°C, respectively, diverging most markedly at the 35-cm light–carapace distance. To correct for inflated estimates of field-collected Te values from models, we fit third-order polynomial regression lines (light–carapace distance vs. core temperature) to each data set and, based on the difference in values between lines, adjusted model temperatures downward to carcass temperature at 1°C increments.
In the field, we mounted models on 30 × 30-cm plywood platforms that were supported by 2.5-m × 2.5-cm polyvinyl chloride (PVC) poles and that we inserted into the soft substrate at the Sphagnum–open water interface. The bottom of the model was in contact with water to simulate an individual turtle basking on the moist Sphagnum mat. Models were deployed at north, south, east, and west locations around the lake, and at each model location, we measured water temperature using HOBO TidbiT v2 Water Temperature Data Loggers-UTBI-001 that were tethered to PVC just above (0.5–1.0 m) the substrate of the littoral shelf and also at 1.5–2 m on the slope where turtles sometimes ventured. Model and water temperatures were recorded at 10-min intervals, but for data analyses, we determined hourly temperature values for models and water at each location. We averaged mean hourly littoral and slope zone water temperatures because they did not differ and, assuming equal access to atmospheric basking sites and water on the littoral shelf, we averaged mean model and mean water temperature values to determine an average hourly Te value (Christian and Weavers 1996; Edwards and Blouin-Demers 2007).
Radiotelemetry and Tb of Free-ranging Turtles
To plot turtle locations in the field, we placed a linear array of numbered, wire-stemmed (90-cm) flags (10 × 10 cm vinyl) at 5-m increments at the margins of the Sphagnum mat–open-water interface and mapped each flag using a Trimble GeoXT GPS unit. We used hardcopy maps to record turtle locations (0900 hrs ± 30 min, 1200 hrs ± 30 min, and 1500 hrs ± 30 min) in the field from watercraft between June and August using handheld Advanced Telemetry Systems R410 radio receivers 4–6 d/wk. On some days, we obtained only a morning location for each turtle. On 10 and 31 July and on 5 August 2014, we radiolocated each turtle 9 times daily at 3-hr intervals (± 30 min) between 0000 and 0000 hrs of the following day to evaluate the degree of nocturnal activity. We avoided shining lights directly on turtles because that could induce movements (Obbard and Brooks 1981). We plotted daily turtle locations on an electronic version of the map in ArcView 3.2 and measured distances between successive locations for each turtle on each day and summed them to determine a total daily distance moved (TDDM). To determine annual home range size, we plotted only a single (early morning) location in ArcView 3.2 for each individual to reduce autocorrelation effects that would occur from multiple locations within a day (De Solla et al. 1999). We used the Animal Movement extension in ArcView 3.2 to determine minimum convex polygon (MCP) size and 50% and 90% kernel sizes using a smoothing value that was determined by least-squares cross-validation for data collected between June and August. We considered the “active season” as April to October, which is when at least some turtles showed discernable daily movements. During the spring and autumn months, turtles were located once per day at weekly or biweekly intervals.
We remotely collected Tb values of free-ranging turtles using a R4500SD Standard Receiver-Data Logger linked to a 3-m omnidirectional antenna (Advanced Telemetry Systems Inc.) that we placed at the southeast end of the lake. The receiver recorded Tb values for each turtle at 10-min intervals (maximum observations/individual/d = 144). To assess daily and seasonal variation in Tb, we determined mean hourly Tb for each turtle for each day and then averaged those values across days, within months (April–October). We obtained daily weather conditions from the National Centers for Environmental Information (National Oceanic and Atmospheric Administration).
Indices of Thermoregulation
For each individual turtle, we determined mean hourly thermal accuracy (db) values as a measure of how well Tb values corresponded to the Tset range (db = |Tb – Tset-min| when Tb < Tset-min, db = |Tb – Tset-max| when Tb > Tset-max, and db = 0 when Tset-min < Tb < Tset-max; Hertz et al. 1993). Thermal accuracy is high when db values are close to zero and low when db values are relatively far from zero. Similarly, we calculated mean hourly thermal quality (Hertz et al. 1993) as de = |Te – Tset-min| when Te < Tset-min, |Te – Tset-max| when Te > Tset-max, and de = 0 when Tset-min < Te < Tset-max where thermal quality is high when de values are relatively small. Thermoregulatory effectiveness (de – db) measures the relative investment in thermoregulation where positive values suggest relatively high investment, negative values indicate avoidance of favorable thermal conditions, and values near zero indicate thermoconformity (Blouin-Demers and Weatherhead 2002; Edwards and Blouin-Demers 2007). To evaluate seasonal variation in thermal quality, accuracy, and investment in thermoregulation, we averaged de values and hourly db and de – db values for each turtle across days and within months. The index of exploitation (Ex; Christian and Weavers 1996) is a measure of thermoregulation that is calculated as the percentage of time that Tb occurs within the Tset range when permitted by operative temperatures (i.e., Te values are within the Tset range). We determined Ex values for each turtle and for each month (0600–1800 hrs). Across seasons, we tested for deviations between Tb and Te distributions that could suggest thermoregulation (Hertz et al. 1993).
Statistical Analyses
We analyzed seasonal variations in TDDM and mean daily Tb that were each averaged per individual over 10-d periods (June and August for TDDM and April–October for Tb) using a General Linear Mixed Model (GLMM), with weather, 10-d time period, and their interaction included as main effects with year and individual turtle identification number included as random effects. Hourly variation in Tb was assessed by a GLMM with hour-of-day, month, and their interaction included as fixed effects and turtle identification number included as a random effect to account for the multiple representations of each individual with the data set. We used least-square means multiple t-tests for post hoc comparisons of means (JMP; SAS Institute Inc., Cary, NC). We tested for differences in the distributions of Tb and Te using Kolmogorov-Smirnov (KS) 2-sample tests. All means are followed by ± SE and significance was determined at alpha = 0.05.
RESULTS
Activity and Movements
Of the 6 C. serpentina that we radiotagged in June 2014, we monitored the remaining 4 through early June or August 2015. One female died approximately 2 wks after release, presumably from a surgical complication. The other radiotagged female survived through the autumn of 2014, but failed to emerge from her hibernaculum during the spring of 2015.
During the summer months of 2014, all radiotagged individuals remained in Davis Lake except for a female that left the lake for a period of 4 d in mid-June, presumably for nesting. Turtles moved along the littoral shelf adjacent to the Sphagnum mat, occasionally moving beneath the mat's edge or into small channels in the mat (Fig. 1). Turtles that ventured away from the mat's edge mostly occupied the shallow (< 2 m), northwest end of the lake.
Citation: Chelonian Conservation and Biology: Celebrating 25 Years as the World's Turtle and Tortoise Journal 19, 1; 10.2744/CCB-1358.1
Home range size, as estimated by MCPs or 95% fixed-kernel (K95%), averaged between 0.6 and 0.8 ha and were relatively large in 2015 when we monitored turtles throughout most of the active season (Table 1). In most cases, MCPs encompassed the entire area of the lake while K95% areas were represented as continuous, or nearly continuous, “U-shaped” home ranges that often projected onto the adjacent terrestrial habitats (Fig. 1). Core areas (50% kernels) existed as 1–3 different areas/individual, averaged approximately 0.1 ha, and overlapped within individuals across years and among some individuals within years (Fig. 2).
Citation: Chelonian Conservation and Biology: Celebrating 25 Years as the World's Turtle and Tortoise Journal 19, 1; 10.2744/CCB-1358.1
During summer (June–August), turtles maintained activity throughout all hours of the day. Based on 3 radiolocations/d, TDDM averaged 53.5 ± 12.82 m (19.6–89.5 m, n = 441 observation days in 5 individual turtles). Total daily distance moved was not affected by diel weather conditions overall (Table 2), although some relatively long-distance movements made by multiple individuals during core area shifts in early July 2015 (Fig. 3) may have obscured the effect. Over 9 consecutive 10-d increments, TDDM significantly declined between early June and late August (Table 2; Fig. 3), but no interaction between weather period and 10-d increment was observed. We observed declines in TDDM over the summer months with most statistically significant differences occurring during early June and mid-July (increments 1 vs. 5) and mid-September (increment 8 vs. all other increments; t = –1.93–3.77, p < 0.0001–0.030; Fig. 3). On a diel basis, turtles moved throughout the entire day, traveling an average of 240.6 ± 15.12 m (201–276 m, n = 3 d in 5 turtles). There were no differences in distances moved among the 8 3-hr time periods (F7,112 = 0.80, p = 0.5862), indicating that turtles maintained both daylight and nocturnal activity.
Citation: Chelonian Conservation and Biology: Celebrating 25 Years as the World's Turtle and Tortoise Journal 19, 1; 10.2744/CCB-1358.1
Turtles showed hibernation-site fidelity and most individuals remained active only through early October of each year. Turtles entered hibernacula between 18 September and 15 October in 2013 and between 28 September and 11 October in 2014. During the 2013–2014 winter, 2 individuals overwintered beneath Sphagnum adjacent to the lake (9 and 47 m from the water's edge), 1 individual overwintered in shallow water in a narrow moat adjacent to the lake (27 m), and 2 others overwintered under the bank of a small stream at distances of 222 and 224 m away from the lake and within 10 m of one another. Interannual interhibernacula distance averaged 8.3 m (0–42 m). Four of our 5 turtles overwintered in identical hibernaculum sites during successive winters, but a single individual changed hibernaculum sites between years, with each at approximately 25 m from the lake's edge and 42 m apart. Emergence from hibernacula occurred between 22 March and 16 May in 2014 and between 15 March and 4 May in 2015. Prior to re-entering Davis Lake during spring, turtles resided in the moat for as long as 15 d.
Operative Temperatures and Thermal Quality
During the activity season, mean hourly Te cycled with a midafternoon peak (Fig. 4), mostly as a function of temperature cycling in our atmospheric model. Owing to relatively high atmospheric temperatures during the afternoon, hourly thermal quality was low at that time (de values were relatively high) but improved (de values declined) during midday in the spring and autumn. Thermal quality was relatively high throughout the day during the summer months (Fig. 4).
Citation: Chelonian Conservation and Biology: Celebrating 25 Years as the World's Turtle and Tortoise Journal 19, 1; 10.2744/CCB-1358.1
Seasonal and Daily Variation in Tb and Thermal Accuracy
Body temperature varied with season and hour-of-day and coincidentally with changes in temporal variations in environmental temperatures. Seasonally, mean daily Tb values averaged for each individual (0600–1800 hrs) across 10-d intervals differed significantly across the first 6 10-d intervals during April and May (mean difference between successive means = 3.6°C, range = 1.0°C–5.2°C; t = 3.8–5.3, p < 0.0001 in 8 comparisons; Table 2; Fig. 5). There were no significant differences in mean daily Tb among 9 successive, 10-d interval means from late May to late August (p > 0.05 in all comparisons), but we observed significant declines in mean daily Tb values, which declined in mean values between late August and late October (mean difference between successive means = 2.2°C, range = 0.6°C–3.2°C; t = 3.7–7.1, p < 0.05 in 5 comparisons; Fig. 5). There was no significant effect of daily weather conditions overall (Table 2), although we found that average Tb values were higher on sunny days compared with overcast days during most of the 10-d intervals that were recorded between late April and late May (mean difference between means = 2.8°C, range = 2.4°C–3.5°C; t = 2.4–3.4, p = 0.001–0.01 in 4 comparisons; Table 2, Fig. 5).
Citation: Chelonian Conservation and Biology: Celebrating 25 Years as the World's Turtle and Tortoise Journal 19, 1; 10.2744/CCB-1358.1
On a daily basis, body temperatures of individuals cycled with occasional, brief spikes and with peak values occurring between 1800 and 2000 hrs during summer (Fig. 6). We occasionally observed turtles at the water's surface, but we never observed atmospheric basking. The amplitude of hourly Tb fluctuations was relatively high during May and June, but declined during July and August, at which time average hourly Tb remained within the Tset range throughout most of the day (Fig. 7). During autumn, the amplitude of mean hourly Tb fluctuations was small despite relatively high Te values, at least during September. In the GLMM of hourly Tb across months of the activity seasons, all effects were significant (month: F6,351 = 637.09, p < 0.0001; hour-of-day: F12,351 = 14.73, p < 0.0001; month × hour-of-day interaction: F72,351 = 1.71, p = 0.0009). Variation in mean hourly Tb among months (Fig. 7) mirrored the results from our analysis of mean daily Tb variation measured across 10-d increments (see Fig. 5). During May and June, mean hourly Tb at 1200 hrs was significantly greater than mean hourly Tb during the morning hours (0600–1000 hrs; t = 1.67–2.21, p = 0.017–0.047) and gradually increased hourly until 1500 hrs (t = 2.23–3.23, p = 0.0007–0.013), with peak and statistically similar Tb values occurring through 1800 hrs (t = 0.53–1.31, p = 0.089–0.19). Significant differences among the mean hourly Tb values were consistent with overall diel cycling during May and June (Fig. 7), although the only significant differences were between mean Tb at 1800 hrs and means that occurred between 0600 and 1200 hrs in July (t = 2.21–3.07, p = 0.0024–0.0044) and between 0600 and 1200 hrs in August (t = 1.86–2.26, p = 0.012–0.031; Fig. 7). There were no significant differences in mean hourly Tb throughout the day during April, September, or October.
Citation: Chelonian Conservation and Biology: Celebrating 25 Years as the World's Turtle and Tortoise Journal 19, 1; 10.2744/CCB-1358.1
Citation: Chelonian Conservation and Biology: Celebrating 25 Years as the World's Turtle and Tortoise Journal 19, 1; 10.2744/CCB-1358.1
Thermal accuracy (db) was relatively low during April (mean hourly db values were relatively high), increased between May and June (mean hourly db values were low), peaked during July and August, but then declined during September and October (Fig. 8). The magnitude of diel cycling in db was highest during May and June when thermal quality was relatively low and relatively low during July and August when thermal quality was high. During September, we observed very little diel cycling in db. The index of thermoregulatory effectiveness (de – db) indicated that turtles avoided favorable midday thermal conditions during April, probably as a result of their residence in partially frozen conditions of the moat under canopy (Fig. 9). Between May and August, turtles largely sought favorable conditions during the mid- to late-day, but during September and October, turtles tended to avoid favorable, midday thermal conditions (Fig. 9), as indicated by negative de – db values.
Citation: Chelonian Conservation and Biology: Celebrating 25 Years as the World's Turtle and Tortoise Journal 19, 1; 10.2744/CCB-1358.1
Citation: Chelonian Conservation and Biology: Celebrating 25 Years as the World's Turtle and Tortoise Journal 19, 1; 10.2744/CCB-1358.1
Relative Distributions of Values of Tb vs. Te and Ex
Seasonally, comparisons of hourly Tb and Te distributions (0600–1800 hrs) indicated that Tb values lagged behind the upward shift in Te during spring, were maintained below average hourly Te distributions during summer, with both Tb and Te distributions showing a discernable downward shift during autumn (Fig. 10). After Bonferonni adjustment of alpha to 0.007 for 7 comparisons, KS 2-sample tests revealed significant differences between the distributions of Tb and Te during each month (KS D = 0.40–0.60, p < 0.0001–0.0007), except during October (KS D = 0.38, p = 0.02). The distributions of Tb in the low temperature range likely reflect the influences of the low water temperature during April and May. Average Te values were largely shifted above Tb distributions in the warmer months of the year. Atmospheric basking was never observed in our turtles; therefore, we tested for differences in the distributions of Tb and water column temperature (Tw), but the distributions did not differ from one another in any month (KS D = 0.20–0.43, p = 0.04–0.92).
Citation: Chelonian Conservation and Biology: Celebrating 25 Years as the World's Turtle and Tortoise Journal 19, 1; 10.2744/CCB-1358.1
The indices of thermal exploitation (Ex) attained maximal values during July and August (Table 3). The relatively low mean Ex values that we observed during May and June were likely limited by the relatively small number of recorded Te values that were within the Tset range during those months. Apparently, low average Te values during April and October prevented turtles from frequently attaining Tb values within Tset, but in September, mean Ex value was relatively low despite the large number of Te values that would have permitted Tb to be within Tset (Table 3).
DISCUSSION
Turtles were active at Davis Lake between mid-April and mid-October, although elevated body temperatures and most movement activity occurred between May and September, a period that would likely best represent the duration of energy acquisition in the population. Similar phenology and activity season durations have been reported for other northern populations of C. serpentina (Obbard and Brooks 1981; Strain et al. 2012), but annual activity seasons are comparatively long in C. serpentina at southern latitudes (Ernst and Lovich 2009), with hibernaculum entry and exit dates driven largely by environmental temperature variation (Strain et al. 2012).
We found that 50% kernels best described space use in our C. serpentina at Davis Lake and that individual turtles repetitively used specific core areas and that they shared space within the lake. Neither MCPs nor K95% values adequately described space use by C. serpentina because each measure included central areas of the lake, or adjacent bog mat and forest, that were not typically used by turtles. Core areas (50% kernels) adequately reflected the linear movements of turtles along the littoral shelf and largely overlapped within individuals among years, possibly indicating that familiarity of microhabitats over time is important. Overall, home ranges and core areas partially overlapped among individuals such that we cannot conclude that interspecific aggression entirely dictated the movements of our individuals (Galbraith et al. 1987). Given the relatively small dimensions of Davis Lake, we are not surprised that our mean MCP value was relatively small compared with values reported for C. serpentina in various larger aquatic systems throughout its range (1.5–35 ha; Obbard and Brooks 1981; Galbraith et al. 1987; Brown et al. 1994; Pettit et al. 1995; Paisley et al. 2009). We cannot discount variation in resource distribution and availability among the various populations as a cause of differences in home range size among populations. Brown et al. (1994), however, found no evidence that variation in productivity caused differences in home range size in C. serpentina.
Daily movements by our C. serpentina along the littoral shelf were consistent with the use of shallow water and structure for thermoregulation and for foraging and we found that C. serpentina is capable of extensive nocturnal activity. The relatively short movement distances of C. serpentina at Davis Lake were likely influenced by the small dimensions of Davis Lake. For instance, Brown and Brooks (1993) reported that between-day, straight-line movement distances ranged between 100 and 300 m in a relatively large lake in Ontario, Canada. However, our 24-hr sampling regimen that accounted for a greater degree of circuitous movements by turtles revealed mean daily distances moved that approached the values reported by Brown and Brooks (1993). Our data corroborate previously observed nocturnal activity in C. serpentina (Obbard and Brooks 1981). Our seasonal trend of declining TDDM values is consistent with a pattern of diminishing resource levels or energy conservation measures prior to hibernation (Rowe 2003).
The C. serpentina of our study showed hibernation biology that seems typical for north temperate populations. In Ontario, Canada, C. serpentina overwintered at lakeside sites, under stream banks, and in benthic sites adjacent to lakes (Brown and Brooks 1994). Hibernation site fidelity could be a common phenomenon in C. serpentina where preferred hibernation sites reduce the potentially lethal effects of anoxia or acidosis (Brown and Brooks 1994; Reese et al. 2002) but sometimes require travel to distant hibernation sites (Brown and Brooks 1994).
Chelydra serpentina appeared to thermoregulate throughout most of the active months, except during April. As we expected, C. serpentina showed relatively smooth, diurnal oscillations during the warm months but with occasional brief “basking spikes” in Tb that presumably resulted from dynamic body positioning between the benthos of the littoral shelf and the water's surface (Grayson and Dorcas 2004; Rowe and Dalgarn 2009). The air–water interface can be of relatively high thermal quality (Millar et al. 2012; Rowe et al. 2014, 2017) that can be exploited by turtles. Although we did not observe atmospheric basking during our summer radio-tracking, we cannot discount that atmospheric basking may have occurred in the spring under prolonged, low environmental temperatures (Obbard and Brooks 1979, 1981). That we found significantly greater mean Tb values on sunny vs. overcast days could imply that our turtles used radiant energy to elevate body temperature relatively early during the active season. Distributions of mean hourly Te values were shifted into the warmer temperature ranges, mostly above Tb values that remained approximately stable between June and August. Relative to peak Te values and thermal quality, turtles maintained the lowest db values (Tb was close to, or within, the Tset range) and attained peak Tb late in the day. We did not anticipate the occurrence of peak Tb late in the day, but the lag time between body and environmental temperatures could be attributed to the large thermal inertia that occurs in our large-bodied turtles (Hertz et al. 1993; Seebacher et al. 2003). Even so, Tb generally continued to increase as average Te values declined, indicating that turtles were likely employing behavioral and physiological mechanisms to increase and maintain relatively high body temperature late in the day (Weathers and White 1971; Spray and May 1972). During the autumn months, noncycling diel Tb patterns, declining mean Tb values, and downward shifts in Tb distributions relative to Te distributions, particularly during September, would be consistent with a downward shift in the Tset range (Rowe et al. 2014). Voluntary hypothermia in C. serpentina would depress oxygen consumption and baseline metabolic rates (Gatten 1978), thereby promoting energy conservation prior to hibernation (Seebacher and Grigg 1997; Seebacher et al. 2003; Rowe et al. 2014).
The thermal biology of our C. serpentina contrasts with that of the small-bodied C. picta marginata at similar north-temperate latitudes in habitats with similar thermal quality. In our study, thermal accuracy (db) and indices of thermal exploitation (Ex values) were maximal when thermal quality was maximal (July and August), whereas thermal accuracy and exploitation in a central Michigan population of C. picta marginata were relatively high throughout the summer months (Rowe et al. 2017). In contrast to our initial prediction that the C. serpentina Tset range would be broad, spanning low to high temperatures, it was in fact relatively narrow and low in C. picta marginata (22°C–26°C vs. 25°C–31°C; Rowe et al. 2014, 2017). The relatively small body size of C. picta marginata would presumably allow both rapid body temperature changes and agile shuttling among thermal patches that would facilitate more precise thermoregulation compared with the larger C. serpentina (Bulté and Blouin-Demers 2010). It is also not surprising, then, that our C. serpentina were not affected by daily weather conditions as has been observed in C. picta marginata at similar latitudes (Rowe and Dalgarn 2009; Rowe et al. 2014, 2017).
In conclusion, our C. serpentina were less effective thermoregulators than are small-bodied species such as C. picta marginata that have been described as “moderate thermoregulators” (Edwards and Blouin-Demers 2007). Future investigations could focus on the spatial and thermal ecology of C. serpentina in different habitats at various latitudes. Of particular interest is how local adaptation, or phenotypic plasticity, in Tset ranges evolves in response to environments with different thermal regimens.
Tribute to Peter C.H. Pritchard
As an initially aimless undergraduate in 1980, my coursework brought me into the company of freshwater turtles in Michigan. My interest in turtle biology quickly turned to the more high-profile and endangered sea turtles, which is where I first encountered the name “P.C.H. Pritchard”. Dr Pritchard's Encyclopedia of Turtles and The Turtles of Venezuela were my earliest herpetological acquisitions that introduced me to formal or technical treatments of turtle biology. Although I did not know Dr Pritchard personally, he was an inspirational figure who most certainly influenced the course of my education and career in biology.
Daily locations, and home range estimations as derived from minimum convex polygon (MCP) and 95% and 50% fixed kernel (K95% and K50%) analysis for an individual Chelydra serpentina (2014).
Core area use by 5 individual Chelydra serpentina at Davis Lake monitored over as many as 3 consecutive summers (2013: dashed; 2014: dot–dash; 2015: dotted).
Total daily distances moved (TDDM) overall and on overcast and sunny days in 5 Chelydra serpentina at Davis Lake, compiled between 2013 and 2015. Symbols are means and vertical lines are standard errors (SEs.)
(Top) Mean (± SE) hourly operative temperature (Te) model, and water temperature measured between June and August (n = 5317 hourly observations; dashed horizontal lines indicate the laboratory-determined Tset range). (Middle) Mean (± SE) hourly Te (n = 13,006 hourly observations overall). (Bottom) Mean (± SD) thermal quality, de (n = 11,359 hourly observations) during the active months at Davis Lake.
Mean daily body temperature (Tb) averaged per individual Chelydra serpentina per day (0600–1800 hrs) and then across all individuals (above; n = 453 observation days in 5 individuals) and mean (± SE) daily Tb averaged per individual at 10-d increments and then across all individuals during the active months (below; n = 30 observations based on 441 observation days in 5 individuals).
Body temperatures (Tb) logged every 10 min in 2 representative individual radiotagged Chelydra serpentina recorded at 10-min intervals over 3 consecutive days (24–26 June 2014) at Davis Lake. The dashed horizontal lines delimit the laboratory-determined Tset range.
Mean (± SE) hourly body temperature (Tb) averaged per individual per month and then across Chelydra serpentina during each month of the active season at Davis Lake (n = 27,337 hourly observations in 5 individuals). The dashed horizontal lines delimit the laboratory-determined Tset range.
Mean (± SE) thermal accuracy (db = |Te – Tb|; n = 27,337 hourly observations in 5 individuals) for values averaged across individual Chelydra serpentina during the active season at Davis Lake.
Mean (± SE) indices of thermoregulatory effectiveness (de – db) for values averaged across individual Chelydra serpentina (n = 5) throughout the active season at Davis Lake.
Distributions of Tb (n = 12,291 hourly observations in 5 individuals) and Te (n = 6180 hourly observations) measured between 0000 and 1800 hrs during the active season in Chelydra serpentina at Davis Lake. The dashed vertical lines delimit the laboratory-determined Tset range.
Contributor Notes
Handling Editor: Peter V. Lindeman
