“Breath holding” as a thermoregulation strategy in the scalloped hammerhead, a deep-diving tropical ectothermic shark
Data files
Apr 26, 2023 version files 1.20 GB
-
Depth_Temp_Accel_Duration_Dates_Sampling_Rates.xlsx
-
Fig_S12_Allometric_scaling_thermal_k.xlsx
-
HH10_accel.csv.zip
-
HH10_depth-temp.csv
-
HH11_accel.csv.zip
-
HH11_depth-temp.csv
-
HH6_accel.csv.zip
-
HH6_depth-temp.csv
-
HH7_170711-181310.txt
-
HH7_170712-172900.txt
-
HH7_170713-183510.txt
-
HH7_170714-203440.txt
-
HH7_170715-172630.txt
-
HH7_170716-174640.txt
-
HH7_170717-180740.txt
-
HH7_170718-175330.txt
-
HH7_accel.csv.zip
-
HH7_depth-temp.csv
-
HH8_170716-212610.txt
-
HH8_170717-204930.txt
-
HH8_170718-203120.txt
-
HH8_accel.csv.zip
-
HH8_depth-temp.csv
-
HH9_170805-201610.txt
-
HH9_accel.csv.zip
-
HH9_depth-temp.csv
-
README.md
-
SL1_bodytemp.txt
-
SL2_bodytemp.txt
-
Table_S2-4_hh_deepdive__telemetry.xlsx
-
Table_S5_hh_deepdive_thermal_k_analysis_results.xlsx
-
Thermal-k-modeling_script5_results_Archive.zip
Abstract
Fish moving between different thermal environments experience heat exchange via conduction through the body wall and convection from blood flow across the gills. Here we report a strategy of preventing convective heat loss at the gills during excursions into deep cold water by the tropical scalloped hammerhead shark (Sphryna lewini). Adult scalloped hammerhead sharks dive rapidly and repeatedly from warm (~26ºC) surface waters to depths exceeding 800 meters and temperatures as low as 5°C. Biologgers attached to adult sharks show that warm muscle temperatures were maintained throughout the deepest portion of each dive. Substantive cooling only occurred during the latter stages of the ascent phase, and once initiated, was rapid. Heat transfer coefficient modeling indicated that convective heat transfer was suspended, probably by suppressing gill function during deep dives. This previously unobserved strategy has broad similarities to marine mammal “breath hold” diving.
Methods
Note: (Refer to References section of publication for numbered citations)
Measuring swimming performance, environment, and body temperature
To measure scalloped hammerhead shark swimming performance, depth, ambient water and muscle temperatures we used an instrument package consisting of a tri-axial accelerometer tag and a depth and temperature archival tag housed in a syntactic foam float (2,000 m depth rating) equipped with a timed release mechanism and radio beacon transmitters to facilitate recovery. The tri-axial accelerometer tag was either a TDR10-XB-340 (56 x 38 x 24mm 69g; Wildlife Computers., Redmond, WA) or a TDR10-Daily Diary-278 (74 x 57 x 36mm, 117g; Wildlife Computers., Redmond, WA). Tri-axial acceleration was sampled at either 16Hz or 32Hz. Depth, water temperature and body temperature were sampled every 5 or 10 seconds using a MK9 archival tag with an 8cm thermistor stalk (Wildlife Computers, Redmond, WA). Each package also contained a SPOT5 or SPOT6 satellite-linked transmitter (80 x 20 x 11mm, 30g; Wildlife Computers., Redmond, WA) to indicate the package’s position when it floated to the surface following release from the tagged animal. Packages that contained a SPOT5 tag were also equipped with a VHF transmitter (MM130B; 16mm diameter, 60mm length, 20g; ATS, USA).
Shark capture and handling
All sharks were caught using baited hooks on demersal longlines inside Kāneʻohe Bay (N 21.45°, W 157.80°), Oʻahu (Hawaiʻi, USA). To ensure captured sharks were in good condition, lines were checked every 30 minutes. Captured sharks were brought along the side of a 17ft skiff, secured with a rope around the caudal peduncle and provided with constant gill ventilation during measurement and instrumentation via a hose inserted into the mouth and connected to an in-water bilge pump. The tag package was attached to each shark via a fusible stainless steel cable tie (360 mm, 8 g; Little Leonardo Co., Tokyo, Japan) passed through two holes drilled through the base of the dorsal fin and secured around the syntactic foam float. The sensor stalk of the MK9 was inserted approximately 8 cm below the skin into the dorsal musculature while the package was being secured to the dorsal fin. Each package contained a timed-release mechanism (RT-4, 16mm diameter x 19mm length 10g; RT-5, 20mm diameter x 38mm length, 20g; Little Leonardo Co., Tokyo, Japan) set to release after 23 days. At release time, an integrated capsule severed the stainless steel band, releasing the package to float to the surface. We used a small vessel and handheld receivers equipped with directional antennas to home in on and recover the floating package locator beacons. Contact information was written on the packages in case they were found by members of the general public. Individual deployment durations ranged from 23 to 29 days (Table S1).
Swimming behavior data processing
Archived accelerometer and depth data were downloaded from ten recovered tag packages. A total of 103 deep dives were recorded from 7 of the 10 sharks (Supplemental Table S1). Average maximum dive depth was 635.5m (± 91.8m; maximum 825m [HH10]). All 32Hz tri-axial acceleration data were resampled at 16Hz to facilitate analyses. Acceleration and depth data were analyzed using the ‘Ethographer’ package (46) in Igor Pro 8 (WaveMetrics Inc., Portland, OR, USA). A low pass filter of 0.3 Hz was used to estimate the static (gravitational) and dynamic (tail stroking) components of the acceleration signal for each axis. The static acceleration components from the x,y,z axes were used to calculate the roll and pitch angles of the shark, where x is the surge axis, y is the sway axis, and z is the heave axis (47):
Eq. 2
Roll = arctan(y/x2+z2)1/2) (180/π)
Eq. 3
Pitch = arctan(x/y2+z2)1/2) (180/π)
The tag attachment angle was corrected to zero degrees centered for roll and to horizontal for pitch by quantifying tag pitch during upright swimming at a constant depth and subtracting this value from the archived data (48) (49). The overall dynamic body acceleration (ODBA) was calculated by summing the absolute values of the dynamic acceleration from all three orthogonal axes (23) (50). We used continuous wavelet transformation to generate a spectrogram of the dynamic component of swaying acceleration, classified the dominant peak as tail beat cycles, and calculated tail beat frequency and amplitude of acceleration at every second.
Thermal conductivity coefficient analysis
Packages from three sharks (HH7, HH8, HH9) contained useful body temperature data for thermal coefficient analysis. HH11’s thermal probe was not inserted into the musculature and HH10 had unusable body temperature data, likely due to a malfunctioning thermal probe (see Table S1). HH1-3 and HH5 remained in shallow water and did not conduct deep dives following tagging. The package for HH4 was never recovered.
To test the sensor response times of the ‘stalk’ thermistor (the one that measured body temperature) and the thermistor embedded in the body of the tag (the one that measured ambient temperature) we empirically tested the thermistor response times of the MK9 tags used for our study and compared our values with previously published values for MK9 tag thermistor response times in (51). Our empirical tests were similar to those of (51) and consisted of moving the MK9 tags back and forth between two buckets of water differing in temperature by 20 ºC. Our tests results concur with those of (51) and show a fast response in the ‘stalk’ thermistor that was inserted into the scalloped hammerhead musculature, and a minor lag in the thermistor embedded in the epoxy casing of the MK9. The lag response of the embedded MK9 thermistor yielded different depth-temperature profiles between the fast descents and the comparatively slower ascents of each dive. To account for this disparity, the depth-temperature profiles of the ascents were used to adjust the depth-temperature profile of the fast descents.
To evaluate whether scalloped hammerhead sharks actively retain body heat or rely on simple thermal inertia during deep dives into cold water, predicted values of the whole-body heat transfer coefficient (k) and the rate of temperature change due to metabolic heat production (Τo) were modeled to match the observed rates of body warming and cooling (sensu (27) (28) (Eq. 1)). The modeling used in (27) (28) (30) assumed the following conditions for k;
(a) k = a constant value;
(b) k = {k1; Ta(t) < Tb (t) k2; Ta(t) ≥ Tb(t),
where k1 and k2 are two values for the whole-body heat transfer coefficient. Model (a) assumes the rate of heat exchange between the shark and the environment is constant and not altered by the shark. Model (b) assumes the rate of heat exchange is different when the shark is cooling (k1) compared to warming (k2). These conditions were not able to adequately model the body temperature of the sharks (all R2 fits >0.60). A post-hoc analysis of swimming activity (ODBA, tailbeat frequency and amplitude) revealed high activity levels during each deep dive (Fig. S5). Body temperature was observed to increase throughout the deepest portion of each dive and then followed by a sharp cooling rate during ascent to the surface and subsequent decrease in swimming activity (Fig. 2, S5-8). These high activity rates will affect metabolic heat generation (Τo) and possibly rates of blood flow through the core muscles or through the gills. These phenomena would ultimately affect the rate of change of body temperature (dTb(t)/dt) and whole-body thermal rate coefficient (k) (equation 3) (3) (31). We therefore modeled predicted values of the whole-body heat transfer coefficient (k) and the rate of temperature change due to metabolic heat production (Τo) based on different conditions associated with the periods of intense swimming activity during each dive. The following four conditions of k and Τo were assumed:
(1) k = a constant value, Τo = a constant value;
(2) k = a constant value, Τo = {Τo1; normal swimming Τo2; deep diving};
(3) k = {k1; normal swimming k2; deep diving}, Τo = a constant value;
(4) k = {k1; normal swimming k2; deep diving}, Τo = {Τo1; normal swimming Τo2; deep diving};
where k1 and k2 are two values for the whole-body heat transfer coefficient. Model (1) assumes the rate of heat exchange between the shark and the environment (k) and rate of temperature change due to metabolic heat production (Τo) is constant and not altered by the shark. Model (2) assumes rate of heat exchange between the shark and the environment (k) is constant while the rate of temperature change due to metabolic heat production (Τo) alters between (Τo1) during normal swimming and (Τo2) during the high activity phases of a deep dive. Model (3) assumes (k1) when the shark is swimming normally and (k2) during the high activity phases of a deep dive while the rate of temperature change due to metabolic heat production (Τo) is constant. Model (4) assumes (k1) and (Τo1) when the shark is swimming normally and (k2) and (Τo2) during the high activity phases of a deep dive. The determination of switching (models 3 & 4) between (k1) (normal swimming) and (k2) (deep diving) was based on established thresholds of vertical velocity between the descent phase and ascent phase of each dive. The same thresholds were used for the determination of switching (models 2 & 4) between (Τo1) (normal swimming) and (Τo2) (deep diving). The optimized parameters for each model were estimated using an iterative least-squares algorithm comparing the observed and predicted body temperature, and coefficient of determination (R-squared) computed to assess the fit across models. Models were run for each night that consisted of more than one deep dive for each shark. Start and stop times for each model duration were arbitrarily set to include all deep dives within a night duration and to include the durations where body-temperature reaches the ambient temperature after the conclusion of a series of deep dives.
While both constant-k models yielded poor fits, both variable-k models performed well, with model (4) giving higher R2 values but also requiring an additional fit parameter (Τo1 and Τo2 vs only Τo). Given the significantly enhanced tailbeat activity occurring during the rapid diving phase it is reasonable to assume higher metabolic heat generation during this period and that different Τo values should be used to reflect different metabolic heat generation during the low and high activity phases.
Thermal conductivity of dead sharks
To account for the rate of temperature change in the absence of convective heat transfer (blood flow across the gills), the body temperatures of two dead adult male scalloped hammerhead sharks with total lengths 240cm and 273cm were recorded by individually placing the sharks in an insulated tub and inducing abrupt alternations of water temperature from 23 ºC to 6 ºC and vice versa (sensu (2) (30)). Core muscle and ambient water temperatures were measured using the same TDR-Mk9 archival tag used in field experiments. Constant heat-transfer coefficients k were calculated using the heat budget model without metabolic heat production (Τo) = 0 (Table S6).
Allometric relationship of thermal coefficient rates
To compare the influence of body size on the rate of heat exchange between the scalloped hammerheads from this study and other large fish species, the heat transfer coefficients from scalloped hammerheads HH7, HH8, and HH9 from model (4) were plotted with heating and cooling thermal coefficients against body mass published from other fish species collated in (32). For the scalloped hammerheads, normal swimming heat transfer coefficients (k1) were incorporated into the warming regression and high activity deep dive phase (k2) heat transfer coefficients from model (4) were incorporated into the cooling regression. Body mass estimates for each scalloped hammerhead were calculated using length-weight relationships from (52).
Usage notes
Thermal coefficient modeling of all scalloped hammerhead shark deep-diving was run using Matlab code "Script5.m" with Matlab version 2021b. (included in this dataset).
Thermal coefficient modeling of the dead scalloped hammerhead sharks was run using Matlab code "Script3.m" (included in this dataset) with Matlab version R2020a.
Telemtry was analyzed using IgroPro 8 (Wavemetrics, Portland, OR).