Ethics statement
All procedures for fieldwork in Pacuare Nature Reserve followed approved protocol under Monash University’s School of Biological Sciences Animal Ethics Committee (Protocol No. BSCI/2016/13), the University of Maryland Center for Environmental Sciences’ Institutional Animal Care and Use Committee (IACUC) (Research Protocol No. S-CBL-16–11), and the Costa Rican Ministerio Del Ambiente y Energia, Sistema Nacional de Áreas de Conservación (SINAC), Área de Conservación La Amistad Caribe (ACLAC) (RESOLUCIÓN SINAC-ACLAC-PIME-VS-R-022-2016; RESOLUCIÓN SINAC-ACLAC-PIME-VS-R-025-2016). The study was performed in accordance with the approved guidelines.
Hatchling tracking
To examine in-situ factors of turtle dispersal into the offshore environment, leatherback hatchlings were tagged with coded acoustic transmitters between 20 August and 3 September 2016 in Pacuare Nature Reserve, Limón Province, Costa Rica (Fig. 1). At Pacuare, hatchlings were obtained from hatchery, incubator-reared, and relocated nests. Hatchery nests consisted of eggs collected as they were laid, transported in plastic bags for less than 5 km, and reburied in two separate protected areas. In these protected, monitored enclosures, the eggs were safeguarded (e.g. from predation) and otherwise developed naturally. Relocated eggs were collected from nests laid the night prior, transported in plastic bags a short distance above the high tide line, reburied on the nesting beach and unmonitored thereafter, until such time as hatchlings emerged. Hatchlings from hatchery and relocated nests were collected as they naturally emerged from the buried nests. Incubator-reared eggs were collected as they were laid, transported up to 1 km in vacuum-sealed bags, and raised under 3 treatments: control, low-oxygen, and high-oxygen in accordance with the Williamson38 protocol. Eggs were incubated for the first 5 days of development in: hypoxia (1% O2) for the low-oxygen treatment and hyperoxia (42% O2) for the high-oxygen treatment. As they did not have to expend time and energy exiting a nest, incubator-hatched turtles were left to absorb their yolk for 2 days38. Turtles held post-emergence from their nest (hatchery and relocated hatchlings) or eggs (incubator-reared hatchlings) were kept in moistened, sand-lined incubators at approximately 30 °C to reduce energy expenditure prior to trial release and prevent potential decreases in swimming performance39. To minimise the influence of genetic relatedness, hatchlings were taken from all available nests (n = 9 in total from hatchery, relocated, and incubator nests) at the time of the study, resulting in parentage by nine females. Turtles were weighed and measured prior to trials using a scale and calliper. To prevent overheating on the boat, turtles were transported in a bucket covered by a wet towel with a moistened cloth inside.
Acoustic tracking was conducted using Vemco V5-180 kHz transmitter tags (0.38 g in-water weight; 0.65 g in-air weight) and tethered to the turtle via a line-float-transmitter assembly (6.85 g in-air weight) and Vetbond based on Gearheart et al.24 and Hoover et al.25 (Fig. S1). The monofilament line in the line-float-transmitter assembly was a total length of 2 m; the first float was suspended 1.5 m behind the hatchling, and the second float was an additional 0.5 m. The brightly coloured orange floats (4.4 cm by 1.9 cm) allowed for visual tracking in the water when acoustic signal was insufficient. Tracking began outside the surf zone, approximately 0.4 km from shore, where turtles were taken via a small 6 m, 150 hp motorboat. The release location was the approximate midpoint of the two hatcheries where hatchlings were collected. Between sunrise and sunset, each turtle was followed at a distance of 10–20 m in the boat using a Vemco VR100 acoustic receiver and VH180-D-10M directional hydrophone22. The V5 tag detection range was approximately 200 m. The VR100 receiver stored the detections, and the data were downloaded to reconstruct hatchling movement paths. The mobile acoustic receiver allowed tracking of the turtles’ movements for a longer period and over a broader area than visual tracking alone because turtles were found acoustically when visual contact was lost.
Hatchlings were tracked only during daylight hours over a 3-week period given hatchling and boat availability. Although hatchlings generally emerge during cooler, evening hours of the day in Costa Rica, no effect on the overall innate behaviour of hatchlings was anticipated18,40. The tracking data should still be indicative of the orientation and speed at which hatchlings are likely to swim. Nighttime tracking was logistically infeasible because of the hazards associated with the oceanic entry point. For a track to provide enough data for inclusion in the analysis, a minimum tracking time of 30 min was established. Turtles were tracked individually for approximately 90 min. Track duration was a trade-off between obtaining a large sample of tracks to account for individual variability, while providing robust speed and orientation information. At the end of each track, the turtle was recovered with a small net, the line-float-transmitter attachment was completely removed, and the turtle was released at the recovery location. The Velcro piece easily removed from the carapace, and there were no evident damages, marks, or lesions from this attachment method on the leatherback hatchlings. Handling was kept to a minimum to reduce any unnecessary stress on the turtles.
Surface current trajectories
Two drifters were used to obtain data on local sea surface currents to evaluate the effect of currents on hatchling movements. A Pacific Gyre Microstar drifter was deployed at the beginning of each turtle track (Fig. S2A). The drifter’s surface float was equipped with a GPS unit that used the iridium short burst data service to broadcast location coordinates every 5 min. A flag was attached to the surface float for increased visibility. Sea surface temperature was recorded by the drifter with a Pacific Gyre probe with 0.1 °C accuracy. The position and temperature data of each drifter release were retrieved from the Pacific Gyre website (https://www.pacificgyre.com). One drifter track was removed from analysis because it entered the surf zone and did not represent nearshore surface currents.
A secondary drifter was launched when equipment permitted at the approximate halfway point during tracking of a turtle. This better estimated the immediate currents the hatchling was experiencing and was used to estimate shifts in the nearshore currents as the turtles headed offshore. This second drifter was constructed using a Davis Instruments aluminium radar reflector with 80 cm of parachute cord attached to a 20.3 cm diameter Panther Plast trawl float (Fig. S2B). The centre of the drifter sat 1 m from the water’s surface, similar to the depth of the Microstar drifter. A piece of wood affixed to the top of the float had a Samsung Galaxy Core Prime mobile phone attached in a waterproof bag. A GPS application was started with each drifter release to provide locations at one minute intervals. Foam tubing was zip-tied around the middle of the trawl float to maintain the GPS unit in an upright position. The float had a flag attached for visibility on the water. Positions were stored on the mobile phone and downloaded upon retrieval of the drifter. Both drifters were recovered at the completion of each individual turtle track.
Environmental data
To understand the influence of tidal states and bathymetry on local currents experienced by hatchling leatherbacks, daily tidal currents were obtained at Limón, Costa Rica (10.00° N, 83.03° W; https://tides.mobilegeographics.com). Periods of peak tides (i.e. high and low) were defined as one hour before and after the measured minima or maxima, with ebb and flow tides between those periods of peak tides. Tidal states were categorised as: high, ebb, low, and flow (Fig. S3A). High-resolution, near-shore bathymetry data were obtained at a 0.0011° resolution from the Global Multi-Resolution Topography Synthesis dataset (https://www.gmrt.org). Missing values were filled with data from the General Bathymetric Chart of the Oceans dataset (GEBCO-2014; https://www.gebco.net; 0.0042° resolution). Bathymetric values for each hatchling track were extracted with a ‘bilinear’ interpolation in the R ‘raster’ package41 (Fig. S3B). All analyses were conducted in the R environment42.
Hatchling movement analysis
An acclimation period of five min was applied to each turtle track to provide time for the hatchling to orient and adjust to the water temperature. Intervals greater than five min between recorded hatchling positions were removed to prevent erroneous calculations (0.03% of recorded positions). These time lapses occurred when the boat was actively searching for a hatchling. Despite the combination of surface floats and the directional hydrophone, maintaining visual and acoustic contact with turtles was challenging, even in calm waters. Many hatchlings dove for short periods (< 60 s); when contact with a hatchling was lost, the visual observer would maintain watch on the last known bearing. The boat would ‘dead reckon’ the hatchling position, maintaining speed and heading until visual contact was re-established. Distances resulting in over-ground speeds greater than 0.75 m/s (0.02% of positions) were removed as spurious positions because they were extreme outliers and inconsistent with adjacent and published values8,27,43. Final positions were mapped using the ‘mapdata’ package in the R statistical software (R Version 3.5.2; https://www.R-project.org)44.
To correct for boat movement as it repositioned relative to the hatchling to maintain a 10–20 m distance, mean latitude and longitude values were calculated for every five-minute period. This provided a regularised track representative of hatchling movement throughout the study period from which distance and speed were calculated. Drifter distances were calculated using the GPS locations from the Microstar surface float GPS and the mobile phone GPS at five min intervals. After converting these distances to speed, values exceeding 1.0 m/s (n = 7) were removed as they were inconsistent with adjacent values, mainly occurring at the beginning and end of tracks. The ‘argosfilter’ package in the R statistical software was used in distance and bearing calculations45. Mean bearing and heading were calculated with circular statistics.
Over-ground speed of hatchlings was calculated based on the total distance travelled during every 5-min interval of each hatchling track, represented as observed distance (m) over time (s). This over-ground speed is the apparent speed of the turtle moving through the water, which includes the turtle’s movements and that of the surface water. The speed of the drifter was calculated in the same manner.
Variations in the over-ground swimming speed of hatchlings were examined using a linear mixed effects model framework. The response variable of over-ground speed was square-root transformed based on results of a Box-Cox transformation to meet model assumptions. The explanatory variables of the model were categorical time as five-minute intervals (Time 1 = 0–5 min, Time 2 = 5–10 min, etc.), treatment (control, low-oxygen, high-oxygen, and combined hatchery-reared and relocated nests), bathymetry, and tidal state, and interactions between these variables were tested. The time variable was restricted to the first 85 min of each track because only a few turtles were tracked for longer durations. Hatchery-reared and relocated nests were treated as one treatment due to the low sample size of relocated nests; hatchlings from those nests experienced a natural emergence onto the beach under similar rearing conditions, with no anticipated differences. Parentage and day of release were confounded with treatment and were not examined. The best fit model was selected based on the model with the lowest Akaike information criterion. Fitting the model runs with restricted maximum likelihood, the chosen random effects structure was intercepts varying amongst individuals, and the error structure was an autoregressive lag 2, suggesting a dependency 2 time lags apart. ANOVA F-tests were used to test for significance of model effects, and ANOVA summary statistics were calculated to describe effects.
Changes in the bearing of hatchlings were examined in the same manner as over-ground speed, with bearing as the response variable. Diagnostic checks were performed on model residuals, which indicated a linear mixed effects model was suitable after data transformation. Squared bearing was run as a result of the Box-Cox transformation with explanatory variables of categorical time, treatment, bathymetry, and tidal state. The best fit linear mixed effects model included random effects differing by individual and an autoregressive lag 2 error structure.
Effects of surface currents on swimming speed and bearing
To obtain a value for the true swimming speed component of hatchling leatherbacks, the surface water flow in which the turtles were swimming was removed from the measured speed of the turtle46. This in-water swimming speed was calculated as the difference of the hatchling over-ground velocity and the velocity of the surface currents, as estimated by the drifters in this study. Over-ground speed of hatchlings and drifters was broken into velocity components using equations in Bailey et al.47, which accounted for each turtle’s speed and bearing to obtain east–west (u) and north–south (v) components. The nearest five-minute intervals of each hatchling were matched with the corresponding drifter released. Some hatchlings (n = 11) did not have drifters deployed with them due to equipment issues and were not included in this portion of the analysis. For turtles with two drifters launched during the trial, one in the beginning and one in the middle, the second drifter’s data were used once recording started. Where a single drifter was deployed, those data were used for analysis throughout the track. This provided surface current values closer to those directly experienced by the hatchlings at each given time interval. The drifter’s u and v-velocity components were differenced from each hatchling’s corresponding over-ground speed components. The in-water speed of the hatchlings was then defined as the square root of the sum of the squared u and v-velocities. To determine true bearing for each location, in-water speed components were then used to correct hatchling tracks for currents from equations in Gaspar et al.46,48,49.
Mean current speeds were used to compare hatchling movements from tracks during the lowest, middle, and highest flow periods from the study. The relationship between over-ground and in-water swimming speed was examined using a linear regression on the mean values from each hatchling.
Source: Ecology - nature.com
