Minutes after emerging from underground nests, hatchling green turtles (Chelonia mydas L.) enter the sea and begin a migration towards the open ocean. To test the hypothesis that migrating hatchlings use wave cues to maintain their seaward headings, we released turtles offshore during unusual weather conditions when waves moved in atypical directions. Hatchlings swam into approaching waves in all experiments, even when doing so resulted in orientation back towards land. These data suggest that green turtle hatchlings normally maintain seaward headings early in the offshore migration by using wave propagation direction as an orientation cue. Because waves and swells reliably move towards shore in shallow coastal areas, swimming into waves usually results in movement towards the open sea.
The physiological mechanisms that underlie wave detection by sea turtle hatchlings are not known. Calculations indicate that, at the depth at which hatchlings swim, accelerations produced beneath typical waves and swells along the Florida coast are sufficient to be detected by the vertebrate inner ear. We therefore hypothesize that hatchlings determine wave direction while under water by monitoring the sequence of horizontal and vertical accelerations that occur as waves pass above.
Sea turtle hatchlings emerge from underground nests on oceanic beaches, scramble into the sea, and swim towards the open ocean in a migration lasting several days. Hatchlings establish an offshore course a few minutes after entering the sea, then maintain the seaward heading even after swimming beyond sight of land (Frick, 1976; Ireland et al. 1978).
Recent experiments have demonstrated that loggerhead sea turtle hatchlings (Caretta caretta L.) can orient using the magnetic field of the earth (Lohmann, 1991). Initial experiments in the field, however, have suggested that neither loggerhead (Salmon and Lohmann, 1989) nor green turtle hatchlings (Lohmann et al. 1990) rely on magnetic cues during the initial phase of the offshore migration. Instead, hatchlings tethered offshore in floating orientation arenas consistently swam towards the direction of wave approach (Salmon and Lohmann, 1989; Lohmann et al. 1990). Wave tank experiments have confirmed that tethered hatchlings will readily orient to waves in the laboratory (Lohmann et al. 1990; Wyneken et al. 1990).
Although these experiments clearly demonstrated that hatchlings can use waves as an orientation cue, the restraint of the tethering system used in earlier field experiments (Salmon and Lohmann, 1989; Lohmann et al. 1990) might conceivably have altered the normal orientation behavior of the turtles or precluded use of other, as yet unidentified, cues. To determine whether green turtle hatchlings also use wave cues to maintain seaward orientation when unrestrained, we released hatchlings offshore during unusual weather conditions when waves moved away from shore or in other uncommon directions. Hatchlings swam into the approaching waves regardless of wave propagation direction. These results are consistent with the hypothesis that wave direction is indeed the primary orientation cue employed by hatchlings in the initial stages of the offshore migration (Salmon and Lohmann, 1989). Because waves move in reliable and predictable patterns over large areas of the ocean surface (Bascom, 1980), they could also represent an important open-sea orientation cue for sea turtles and other long-distance ocean migrants such as fish and cetaceans.
MATERIALS AND METHODS
Hatchling green turtles (Chelonia mydas L.) were obtained from nests deposited on beaches in the vicinity of Fort Pierce or Boca Raton, Florida, USA. Each morning during the nesting season, a beach patrol identified green turtle nests deposited the night before, marked each nest site, and noted the date of deposition. Nests were then checked daily as the time of expected emergence neared. When a depression formed in the sand above the eggs (indicating that the eggs had hatched and that emergence would probably occur that evening), we gently dug into the sand by hand and extracted 12–24 hatchlings. The turtles were placed into styrofoam coolers and transported within 1h to a laboratory at the Harbor Branch Oceanographic Institution (in Fort Pierce) or to Florida Atlantic University (in Boca Raton). All turtles were tested and released within 48h of capture.
Transport of hatchlings offshore
All experiments were conducted between 10:00 h and 17:00 h at distances of 7.4–11.1 km offshore near Fort Pierce or 4.6–5.6km offshore near Boca Raton. The sun and at least some blue sky were visible during all experiments. Hatchlings were placed in styrofoam coolers lined with damp paper towels and transported offshore by motorboat. The lids of the coolers were left ajar to ensure adequate ventilation, but the coolers were kept beneath the canopy of the boat where hatchlings could not see the sky.
Assessment of wave direction and weather conditions
Immediately before and after each experiment, detailed notes were made of wind, wave and weather conditions. At each offshore position, latitude, longitude and distance from shore were recorded using a Loran C recorder system accurate to within 6m. Wind speed was measured using a mechanical wind gauge (Davis Turbometer). Wind direction was determined by mounting a length of yarn on a pole and sighting along the yarn using a handheld digital compass (Autohelm). Wave direction was measured by sighting down the axis of wave propagation with the digital compass (Lohmann et al. 1990). When more than one wave type could be discerned, we calculated the mean wave direction using the procedures of Lohmann et al. (1990).
Treatment of hatchlings before experiments
In all but two releases (see below), hatchlings were placed into a plastic bucket filled with sea water at least 15 min prior to release. This step was taken because (1) hatchlings released into a floating orientation cage in previous experiments (Lohmann et al. 1990) usually dove or circled for their first several minutes in the water before establishing a course, and (2) preliminary experiments indicated that dry hatchlings placed directly into the ocean often dove and disappeared from view before their orientation could be assessed, whereas hatchlings that had been in water for several minutes before release usually swam within 0.5 m of the surface (as hatchlings normally do while migrating offshore).
In two releases, we compared the orientation of hatchlings placed in water beforehand with that of nestmates kept dry until release. In each case hatchlings were obtained from a single nest; half were placed into a bucket of sea water at least 15min before the trial (as before), while the other half remained in a styrofoam cooler lined with damp paper towels. Hatchlings from the two groups were then released into the ocean alternately (i.e. first a turtle from the wet group, then one from the dry group, then a wet hatchling, and so on).
Assessment of orientation
Hatchlings were dropped over the side of the boat so that they entered the water approximately 1.5 m away. Swimming sea turtle hatchlings reportedly ignore nearby boats and swimmers and do not change course to avoid them (Frick, 1976; Salmon and Wyneken, 1987; Salmon and Lohmann, 1989). Nevertheless, we released turtles alternately to the north and south of the boat to reduce possible directional biases that could result if hatchlings consistently sought to avoid the boat. Evidence of such biases was not found (see Results).
After each hatchling had been released, two observers began independently recording its directional heading. Each observer was equipped with an Autohelm handheld digital compass which recorded the bearing of the turtle each time a button was pressed. Readings were recorded approximately once per second until the turtle disappeared from view. The last bearing obtained by each observer was recorded; a vanishing bearing for each turtle was then calculated as the mean of these two measurements. Vanishing bearings for each group were analyzed using the Rayleigh test (Batschelet, 1981) to determine whether the turtles were significantly oriented as a group.
Calculations of water mass accelerations beneath waves
The calculations assumed typical wave parameters near the east coast of Florida during August, the first month of the green turtle hatching season; these include wave heights of 0.5–1 m and a wave period of 5 s (Hogben and Lumb, 1967). We also assumed an average hatchling swimming depth of 20cm (Frick, 1976), an average swimming speed of 43.6 cm s−1 (Frick, 1976) and a swimming course in the same direction that the waves moved (so that the calculated accelerations represent the minimum a turtle would encounter).
Swimming behavior after release
Upon release into the ocean, hatchlings usually submerged for 5–30 s. During this time most turtles (approximately 90% ) remained within 1 m of the surface and were easily visible from the boat. A few turtles (5–10%) dove to greater depths and were immediately lost from view; no data were obtained for these hatchlings.
After resurfacing from their initial dive, hatchlings usually paddled with their heads above water for about 5–20 s before beginning to swim rapidly just below the surface. Some hatchlings then immediately established courses into the waves and did not deviate from them during our observations. Others started on initial courses that did not lead into the waves (but were nearly always within ±70° to the direction of wave approach); in these cases, hatchlings typically changed course once or twice during the next minute before establishing a course that they maintained consistently until vanishing from sight. Adoption of the ‘final’ course was often associated with a marked increase in swimming speed. Between 45 and 150 s typically elapsed between the time a hatchling was released and the time it swam out of sight (at a distance of about 30–50 m from the boat).
In 10 of 11 sets of releases, there was no significant difference (Watson test, using P < 0.05 as criterion) between the orientation of hatchlings released north and south of the boat. These data suggest that the position of release with respect to the boat had little or no effect on the orientation of hatchlings under the experimental conditions. Data from turtles released to the north and south each day under otherwise identical conditions were therefore combined.
Results of three different releases conducted on days when waves moved in a typical onshore direction are shown in Fig. 1. In each experiment hatchlings oriented in a seaward direction and towards the approaching waves; mean angles of orientation for the three groups of hatchlings ranged from 65° to 92°, similar to the range of directions for wave approach (74–104°).
Results of four releases conducted when waves moved in directions other than directly towards land are shown in Fig. 2. In each case, the mean angle of orientation was within 21° of the direction of wave approach. In one experiment in which waves moved almost directly away from land and towards the open sea, hatchlings swam back towards land (Fig. 2C).
In two different releases, the orientation of hatchlings kept dry prior to release was compared to that of nestmates previously placed in sea water. In both cases, the ‘wet’ group was significantly oriented (Rayleigh test, P<0.001) and swam towards approaching waves (Fig. 3A,C), whereas the orientation of the ‘dry’ turtles was statistically indistinguishable from random (Fig. 3B,D). The vanishing bearings of the dry hatchlings were significantly more dispersed (P<0.001in the first experiment, P<0.025 in the second) than those of the wet hatchlings (Wallraff’s modified Wilcoxon-Mann-Whitney U-test; Batschelet, 1981).
The mean angles of orientation for all releases (excluding the hatchlings kept dry prior to release) are plotted as a function of the direction of wave approach in Fig. 4. Circle-circle correlation analysis (Batschelet, 1981) indicated that wave direction and hatchling orientation were significantly related (P<0.001).
Calculation of accelerations beneath waves
The vertical and horizontal accelerations that a migrating green turtle hatchling presumably encounters while swimming beneath typical Florida waves are plotted in Fig. 5. For waves 1 m in height, the peak accelerations of approximately 72 cm s−2 (Fig. 5) easily exceed the minimum linear acceleration of 5 cm s−2 that can be detected by the vertebrate inner ear (Lewis et al. 1985). Even smaller waves of 0.5 m would produce accelerations that exceed this threshold (Fig. 5).
Orientation to waves
The results provide strong evidence that unrestrained hatchlings orient by swimming into approaching waves when released offshore. The mean angle of orientation for turtles in each experiment was strongly correlated with wave direction (Figs 1, 2, 3A,C, 4). Hatchlings swam into waves even when doing so resulted in orientation back towards land instead of out to sea (Fig. 2C).
These results are consistent with data obtained using tethered green turtle (Lohmann et al. 1990), leatherback (Lohmann et al. 1990) and loggerhead (Salmon and Lohmann, 1989) hatchlings restrained inside a floating cage. The present results suggest that the behavior of tethered hatchlings accurately reflects the behavior of free-swimming turtles; hatchlings tested under both conditions oriented towards waves.
As waves enter shallow water and approach shore, the propagation direction is refracted until it is nearly perpendicular to the beach (Denny, 1988). Thus, for hatchlings entering the ocean under natural conditions, swimming into waves reliably results in offshore movement. Several marine crustaceans (Walton and Herrnkind, 1977; Nishimoto and Herrnkind, 1978) and molluscs (Hamilton and Russell, 1982; Gendron, 1977) are also known to use waves or wave surge (the horizontal oscillation of water near the sea floor) as an orientation cue in shallow water.
In deeper water far from shore, wave and swell direction often still provide a consistent directional cue for orientation. Because swells originate in the open ocean and move great distances before reaching shore, swell direction is established far out at sea and is largely independent of local weather patterns occurring near land. The seasonal constancy of swell direction has long been recognized; Polynesian navigators, for example, used swell propagation direction as a cue in long-distance voyages (Lewis, 1978). Swell direction could hypothetically be used in open-ocean orientation by hatchling and adult sea turtles, fish (Cook, 1984), cetaceans and perhaps even flying migrants (birds and insects) able to perceive swell patterns from the air.
Whether sea turtle hatchlings continue to use wave direction as a cue after entering deep water is not known. Recent efforts to track loggerhead hatchlings leaving the east coast of Florida suggest that turtles may normally swim into waves for only a relatively short time after entering the sea (Witherington, 1992). After distancing themselves from shore, hatchlings may subsequently maintain their directional bearings using the geomagnetic field (Lohmann, 1991) or other, as yet unidentified, cues.
Effects of handling and the role of the beach crawl in ocean orientation
When hatchlings emerge naturally from their underground nests, they must crawl across the beach before entering the sea. In a previous study of hatchling green turtle orientation in the ocean, hatchlings deprived of a beach crawl and released offshore dove repeatedly and swam in erratic circles, apparently unable to establish a seaward course (Frick, 1976). This result led to the hypothesis that a beach crawl is essential for subsequent orientation at sea (Frick, 1976; Mrosovsky, 1983).
In the present experiments, however, none of the hatchlings crawled across a beach before they were released offshore, but all groups placed in water prior to release were significantly oriented (Figs 1, 2, 3A,C). Moreover, other recent experiments have failed to provide evidence that beach crawl experience modifies the way in which hatchlings respond to waves early in the offshore migration (Salmon and Lohmann, 1989; Lohmann et al. 1990).
The difference between the present results and those of Frick (1976) may be attributable to differences in treatment of hatchlings immediately prior to release into the sea. In experiments comparing two different handling methods, some hatchlings were placed into a bucket of sea water at least 15 min before release, while others were kept dry (see Materials and methods). Turtles from the wet group reliably oriented towards waves (Fig. 3A,C), whereas dry hatchlings were not significantly oriented as a group (Fig. 3B,D).
Hatchlings may therefore require several minutes in water before they attain the behavioral or physiological state in which they reliably orient to waves. This time could represent a brief transition period during which (1) hatchlings are attentive to cues other than waves, (2) hatchlings become acclimated to the water and the change from a motor program for crawling to one for swimming, and/or (3) the motivational state of the hatchlings changes, perhaps as a consequence of the slightly lower temperature of water or sensory cues associated with submergence. We speculate that the hatchlings released offshore by Frick (1976) appeared disoriented because they were released while dry and thus disappeared from the view of observers before they began to orient to waves.
Physiological mechanisms underlying wave detection
The mechanism underlying wave detection by sea turtle hatchlings is not known. In principle, however, hatchlings could use one of at least three different mechanisms to determine wave propagation direction. First, turtles could detect wave direction visually by looking at approaching waves while at or just below the surface. Second, hatchlings could indirectly determine wave direction by detecting the frequency of waves encountered while swimming in different directions (a Doppler shift). Finally, a swimming hatchling might determine wave direction by detecting the sequence of horizontal and vertical accelerations caused by a wave passing above.
Wave tank experiments have indicated that green turtle hatchlings can orient to waves even in the absence of visible light (Lohmann et al. 1990). Thus, hatchlings do not need to see waves in order to determine wave propagation direction.
A Doppler shift could also be used to determine wave propagation direction. Relative to a stationary hatchling, a turtle swimming towards approaching waves will experience an increase in wave frequency, whereas a turtle swimming in the direction of wave movement will experience a decrease. The number of waves encountered per unit time could be determined by monitoring the frequency of accelerations due to the passing waves (see below). Whether hatchlings can detect differences in wave frequency with sufficient resolution to discriminate different directions is, however, not known. Moreover, tethered hatchlings oriented to waves while essentially stationary (Salmon and Lohmann, 1989; Wyneken et al. 1990; Lohmann et al. 1990), and unrestrained hatchlings often established courses towards waves almost immediately after release into the ocean (see Results). Because a hatchling would presumably need to contrast the Doppler shifts encountered while swimming in different directions to determine the direction of wave approach, this mechanism appears unlikely.
Cook (1984) hypothesized that salmon might orient using wave cues and might determine wave propagation direction by monitoring the phase relationship between the horizontal and vertical components of acceleration occurring in water beneath oceanic swells. Calculations revealed that the accelerative forces produced beneath typical swells are probably detectable by migrating fish down to depths of about 30 m.
We repeated the calculations of Cook (1984) using parameters appropriate for sea turtle hatchlings swimming beneath waves off the Florida coast. The results (Fig. 5) indicate that accelerations encountered by migrating sea turtle hatchlings exceed by an order of magnitude the minimal 5 cm s −2 threshold linear acceleration known to be detectable by a vertebrate inner ear (Lewis et al. 1985).
Given that the horizontal and vertical accelerations are theoretically detectable by a turtle, the sequence of accelerations could be used to determine orientation relative to wave direction. The phase relationship between horizontal and vertical accelerations can be seen quantitatively in Fig. 5 for a turtle swimming with waves and qualitatively in Fig. 6 for a turtle swimming either with or against waves. In qualitative terms, a hatchling swimming towards approaching waves would experience a sequence of upward, backward, downward and forward accelerations with each wave cycle (Fig. 6). In contrast, a hatchling swimming in the same direction as the waves moved would encounter a sequence of upward, forward, downward and backward accelerations (Fig. 6). The nervous system of the turtle would need only to distinguish between these sequences (perhaps by simply determining whether a backward or forward acceleration followed the initial upward movement) to differentiate orientation against and with wave propagation direction.
We thank Michael Salmon, Jeanette Wyneken and Andrew Smith for critical reading of the manuscript. We also thank Erik Martin and Robert Ernest for help in locating green turtle nests and the Harbor Branch Oceanographic Institution for providing boat moorage. The work was supported by NSF grant IBN-9120338 (to K.J.L.) and NSF grant BNS-87-07173 (to M. Salmon and J. Wyneken). Endangered species research was authorized under Florida DNR special permit TP 073.