The Development of Early Diving Behavior by Juvenile Flatback Sea Turtles (Natator depressus)
Abstract
The flatback turtle is the only species of marine turtle that lacks an oceanic phase of development in its early life history. Instead, the turtles grow to maturity in shallow turbid shelf waters of tropical to subtropical Australia. We studied the development of diving behavior in neonate flatbacks to determine whether diving under those ecological conditions resulted in differences from leatherbacks (Dermochelys coriacea) and green turtles (Chelonia mydas) at the same age when diving in clear, deep oceanic waters. Data were obtained from flatbacks that varied in both age (1–7 weeks) and mass (38–100 g). Each turtle towed a miniature time–depth tag during a single 30-minute trial in shallow (≤ 12 m) turbid shelf waters near Townsville, Queensland, Australia. In total, 192 dives were recorded from 22 turtles from 4 nests. Most dives were short (< 100 seconds) and shallow (< 4 m), but even young turtles could dive to the bottom. The most common flatback dives had V- or W-profiles, whereas, in leatherbacks, most dives were V-profiles, and, in green turtles, the dives were either V- or U-profiles. Routine flatback dives were accomplished by swimming slowly (like leatherbacks), but, when sufficiently motivated, flatbacks could swim faster (> 1 m/s) than green turtles. They could also make repeated deep dives after surfacing only briefly to replenish their oxygen supply. Changes in performance (longer, shallower dives) were correlated with increases in mass but not age. We hypothesize that, as neonates, flatback dives enable the turtles to 1) search efficiently for prey throughout the water column under conditions of limited visibility, 2) minimize surface time so that even in murky water the turtles can return to previously attractive locations, and 3) swim rapidly to evade their predators.
Marine turtles are diving specialists that as larger juveniles and adults can submerge to depths and for durations that rival many marine mammals. Those diving capabilities are the result of physiological (Kooyman 1989; Lutcavage and Lutz 1997), morphological (Wyneken 1997), and behavioral (Hays 2008) adaptations that together enable rapid gas exchange during typically brief surface breathing episodes, enhanced mechanisms for oxygen storage and release, a streamlined body shape, and efficient propulsion (predominantly lift-based, “aquatic flight”; Wyneken 1997; Renous et al. 2008) during the dive. Previous studies focused particularly on the dive profiles of adult turtles, especially during long-distance migration between foraging sites and internesting behavior near nesting beaches (Hays et al. 2006; James et al. 2006; Rice and Balazs 2008) and by juvenile turtles foraging within circumscribed home ranges (van Dam and Diez 1998; Seminoff et al. 2002; Makowski et al. 2006; Seminoff and Jones 2006), where activity is typically concentrated around important core areas (the best feeding and resting sites).
In contrast to the relative wealth of data available for large juvenile and adult marine turtles, little is known about diving frequencies, functions of dives, or types of dive profiles shown by young juveniles during the oceanic phase of their development. This deficiency occurs because small turtles disperse over vast and largely unknown oceanic areas and, therefore, are unavailable for study (Musick and Limpus 1997). Western Atlantic loggerheads (Caretta caretta) are the exception because for the first few weeks after they enter the sea, they can be found at “downwelling” sites adjacent to coastlines (Carr 1987). Weeks to months later, they are transported eastward by oceanic surface currents to “nursery” sites near the Azores, Canary Islands, and Madiera (Bolten 2003). But as small turtles, even surface-dwelling loggerheads are difficult to locate and challenging to observe while they reside in weed lines because they are camouflaged and rarely move. This behavior probably promotes their survival as “float and wait” predators (Witherington 2002) and also reduces their vulnerability to predators. At this stage of development, diving in open water occurs infrequently.
These gaps in our knowledge are unfortunate because the survival strategies and associated adaptations of young turtles in open water habitats are certainly important and probably differ from those shown by larger juveniles and adults in coastal waters that are less vulnerable to a smaller subset of predators. The behavior of young turtles is also of interest from a developmental perspective. With an increase in size, young turtles are likely to become more competent as swimmers and divers, which enables them to change how (and where) they search for food, what kind of prey they select, and how they defend themselves against predators. Unfortunately, we know virtually nothing about any of these aspects of their behavior.
How can this situation be remedied? One possibility is to carry out “staged” experiments, that is, to introduce young turtles of particular size or age classes into an open-water environment and observe what they do. Studies of this kind must be done with care because there is always the possibility that environmental conditions will be inappropriate and result in behavioral artifacts. However, small turtles seem amazingly impervious even to highly simplified and artificial laboratory conditions. In fact, they behave in ways that seem consistent with what little we know about their ecology in the open ocean (e.g., orientation relative to locations in gyre currents, Lohmann et al. 1997; migratory activity, Wyneken and Salmon 1992; Witherington 1995).
In one such study, Salmon et al. (2004) compared the development of diving and feeding behavior between young leatherbacks and green turtles. The turtles were reared in the laboratory and, at 2-week intervals, several were released offshore in deep water for a single, brief trial before they were released. Observers recorded behavior associated with each dive while dive profile data were stored on miniature time–depth recorders (TDR) that the turtles towed. These studies showed that the 2 species differed in the type and frequency of dive profiles, fed on different prey, and as they developed, older leatherbacks made deeper dives whereas older green turtles made longer dives. These changes were apparent even over a relatively short (8–10 week) period of growth.
Here, we use similar methods to describe the diving behavior of young flatback turtles (Natator depressus) during the first 7 weeks of development. This species is of particular interest because, unlike other marine turtles, the hatchlings do not disperse into open oceanic waters; instead, they remain within the relatively shallow Australian continental shelf waters (Walker and Parmenter 1990). Diving profiles of the adults were recently described (Sperling 2008), but no information is available on the diving performance of neonates (defined by Witherington [2002] as posthatchlings that have matured beyond the period of frenzied swimming).
We hypothesized that flatback dives, unlike those of developing leatherbacks and green turtles in the open ocean, would (1) show few changes in profile, because the turtles have little opportunity to vertically expand their niche in shallow (coastal) waters. Flatbacks consume a variety of prey found at the surface, in the water column, and on the bottom in turbid waters (Zangerl et al. 1988). We predicted that they would use (2) V- or W-shaped dives to search for prey and detect predators over a wide swath in the vertical plane. Finally, the morphology (streamlined body shape, large flippers) and vigorous activity shown by flatbacks after entering the sea (Salmon et al. 2009), coupled with a high probability of encountering predators (Walker 1991a) should select for a turtle capable of (3) rapid and powerful swimming movements. A similar prediction was made earlier by Walker and Parmenter (1990), who speculated that neonate flatbacks would be more powerful swimmers than the neonates of other marine turtle species.
METHODS
Turtles
We reared 33 hatchlings (6–11 per nest from 4 nests) obtained from a rookery in Mackay, Queensland, Australia (lat 21°08′S, long 149°11′E). Turtles were captured during January and February, either as they emerged naturally or from the sand column above the hatched eggs during the late afternoon just before their evening emergence. Hatchlings were immediately placed into covered buckets that contained a shallow layer of moist sand and were transported within a few hours to screened outdoor pools located in the Marine and Aquaculture Research Facilities Unit on the campus of James Cook University, Townsville, Queenland, Australia (lat 19°15′S, long 146°45′E). Each turtle was isolated in a rectangular plastic container (34 cm × 24 cm × 18 cm deep) that was continuously supplied with filtered (recirculating) seawater. The turtles began feeding within a day after capture. They were provided with an in-house manufactured diet composed of ground fish, gelatin, human infant formula, freshwater turtle food pellets, and reptile vitamins. The food was cut into small cubes that were dropped into the plastic containers; turtles dove to the bottom to consume the food (10%–15% of their body weight daily). At capture and every 7–10 days thereafter, we measured growth (straight-line carapace length [SCL]) with vernier calipers (nearest 0.1 mm) and mass with an electronic scale (nearest 0.1 g).
Diving Trials
Diving trials generally were carried out during the morning and early afternoon to avoid late afternoon thunderstorms. A subset (n = 22) of the 33 turtles we reared was used in these trials. Turtles were not fed on the day of their trial.
Using a small boat, turtles were transported up to 12 km offshore to 1 of 6 sites located between Townsville Port and Magnetic Island, or on the opposite (northwestern) side of Magnetic Island (Fig. 1). The turtles were kept in shaded buckets during transport. No more than 4 turtles from 2 clutches were tested on any one day. Each turtle was released after its trial. Water temperature at the surface varied between 29°C and 30°C and declined by no more than one degree with depth. Depth at our trial sites ranged between 9.0 and 12.5 m. Our intent was to test the turtles at regular (2-week) intervals. However, storms that lasted for several days made it impossible to follow an exact schedule. We instead opted to test the turtles as weather permitted.
Citation: Chelonian Conservation and Biology 9, 1; 10.2744/CCB-0803.1
During trials, each turtle towed a Lotek miniature TDR (LTD 1100, 11 × 32 mm, sampling interval = 14 seconds; LAT 1500, 11 × 35 mm, sampling interval = 10 seconds), which we previously calibrated by submergence at 2-m increments to a maximum depth of 12 m. TDRs were made slightly positive in buoyancy by encasing them inside a thin foam covering shaped to minimize drag (Fig. 2). Slight positive buoyancy allowed the TDR to reach the surface (and mark the end of each dive) near the time when the turtle surfaced to breathe but also enabled the device to be easily submerged by the turtle during each dive.
Citation: Chelonian Conservation and Biology 9, 1; 10.2744/CCB-0803.1
Each TDR was attached to the turtle by an approximately 1-m length of light (2-kg test) monofilament line threaded through the keratinous part of a supracaudal scute with a fine needle. If a predator took the turtle, then this line broke to release the TDR. The TDR was attached by stronger (9-kg test monofilament) line to a plastic “handline” spool held by an observer at the front of the boat. This line was released as the turtle dove and retrieved while it was on the surface so that there was always sufficient slack for the turtles to swim, dive, and surface without restriction or excessive drag.
Each trial began by gently releasing the turtle in the water in front of the boat and by paying out sufficient slack to permit an initial, relatively long and deep “escape” dive. During an initial 5-minute acclimation period, the turtle distanced itself from the boat approximately 50 m. At this distance, the turtle gave no indication that it reacted to the presence of the boat. The driver adjusted the boat's forward progress to match that of the turtle so that this distance was maintained during the entire 30-minute trial after acclimation. Unfortunately, underwater visibility was severely restricted (< 1 m), which made it impossible to directly observe how the turtles behaved during their dives.
Data Analysis and Statistics
Dive profiles (shape, depth [m], and duration [seconds]) were measured from the TDR records and downloaded to a computer. A dive was defined as any submergence to a depth ≥ 0.5 m. Profiles were classified by using the criteria of Hochscheid et al. (1999).
Individual turtles varied considerably in growth rates. We divided the turtles into 2 groups to determine whether diving behavior changed most as a function of age or mass. These were “younger” (≤ 3 weeks) and “older” (4–7 weeks after emergence) turtles, and “small” and “large” turtles. Small turtles were those of any age whose mass at the time of their dive trial fell within the range (38.0–62.4 g) shown by the turtles at the end of 1 week in captivity. Large turtles were individuals that, during their dive trial, weighed between 63.0 and 100.2 g. (The latter was the heaviest turtle we used in our field trials.) Comparisons between the groups focused on possible changes in dive-profile frequency, as well as changes in dive depth and duration. Comparisons were made by using the Fisher's exact and χ2 tests (the latter corrected for continuity; Zar 1999). Swimming speeds (cm/s) were estimated from the time it took the turtles on the surface to reach depth while performing the most common (V-shaped) dives. Comparisons between the swimming speeds of the different groups of turtles were made by using nonparametric (Kruskal-Wallis and Mann-Whitney; Zar 1999) tests. Time intervals between consecutive dives were estimated by the number of TDR readings at the surface multiplied by the recording interval (either once every 10 or 14 seconds) of the TDR.
In all statistical tests, significance was set at p ≤ 0.05.
RESULTS
Growth
At capture, the turtles were (mean ± SD) 45.94 ± 6.77 g mean ± SD in mass and 6.66 ± 0.60 cm mean ± SD in SCL (n = 33 turtles from 4 clutches). By the end of their first week in captivity, the turtles averaged 50.90 ± 4.41 g mean ± SD in mass (range, 38–62.4 g) and 7.09 ± 0.34 cm mean ± SD in SCL (Fig. 3). By 7 weeks after emergence, they averaged 91.86 ± 8.38 g mean ± SD in mass and 9.43 ± 0.34 cm mean ± SD in SCL (Fig. 3).
Citation: Chelonian Conservation and Biology 9, 1; 10.2744/CCB-0803.1
Dive Profiles
Records were obtained for 192 dives by 22 turtles; one turtle swam vigorously but did not dive (Table 1). Dives fell into 1 of 3 profiles: U, W, and V (Fig. 4). Profiles differed in their frequencies and in their spatio-temporal characteristics (Fig. 5). U-dives (n = 11) were least commonly performed (7 of the 21 diving turtles; Table 1). They were of short duration (80–125 seconds) and, with one exception, shallow (≤ 4 m). W-dives (n = 51) were more common (18 of 21 turtles) and spanned a broad range of durations (25–350 seconds) at typically shallow (≤ 4 m) depths. V-dives (n = 130) were most commonly performed (21 turtles), typically of short duration (≤ 150 seconds), and spanned a broad depth range (0.5 to > 11 m).
Citation: Chelonian Conservation and Biology 9, 1; 10.2744/CCB-0803.1
Citation: Chelonian Conservation and Biology 9, 1; 10.2744/CCB-0803.1
Two of the turtles in our study (turtles 18 and 19, Table 1), both from the same clutch and tested at the same location on the same day, performed more V-dives (n = 20, n = 27, respectively) than any of their siblings and all of the other older turtles combined. Their V-dives were also deeper and shorter in duration than those of the other turtles. For those reasons their V-dives are analyzed separately from the 83 V-dives of the other turtles.
Changes in Dive Frequency and Profile During Development
We found no statistical differences between small and large turtles, or between younger and older turtles, in diving frequency on a per turtle basis (77 dives by 12 smaller vs. 68 dives by 9 larger turtles, χ2 = 0.03, not significant [n.s.]; 78 dives by 11 younger vs. 67 dives by 9 older turtles; χ2 = 0.01, n.s.). Neither the small vs. large nor the younger vs. older turtles differed statistically in the proportion of W-dives (23 dives by the 12 small and 28 by 9 large turtles; χ2 = 0.62, n.s.; 23 dives by the 11 younger and 30 by the 11 older turtles; χ2 = 0.24, n.s.), V-dives (45 by the small and 38 by the larger turtles; χ2 = 0.04, n.s.; 40 by the younger and 43 by the older turtles; χ2 = 0.09, n.s.), or U-dives (8 small and 3 larger turtles; 8 younger and 3 older turtles) that they performed.
Changes in Dive Duration and Depth During Development
The turtles of small mass (Fig. 6, upper graphs) made proportionally more deep (≥ 4 m; 14 of 77) dives than did the turtles of large mass (2 of 68; p < 0.003 by the Fisher's exact test). However, the larger turtles made proportionally more long (≥ 200 seconds; 13 of 68) dives than the smaller turtles (1 of 77; p < 0.0003 by the Fisher's exact test). Comparisons between the age subgroups when using the same criteria (Fig. 6, lower graphs) revealed no statistical differences in either dive depth (χ2 = 0.37, n.s.) or dive duration (χ2 = 0.05, n.s.).
Citation: Chelonian Conservation and Biology 9, 1; 10.2744/CCB-0803.1
Descent Swimming Speeds During V-Dives
The descent speeds (Table 2) of the turtles of smaller mass (mean and median, 8.93 and 8.00 cm/s, respectively) were significantly faster than those shown by the larger turtles (3.58 and 2.86 cm/s, respectively; Mann-Whitney Z = 5.9, p < 0.0001). The mean and median descent speeds shown by turtles 18 and 19 (Fig. 7) were significantly faster (mean and median 53.5 and 54.0 cm/s, respectively) than those of the 2 mass subgroups (Kruskal-Wallis H = 95.11, p < 0.0001).
Citation: Chelonian Conservation and Biology 9, 1; 10.2744/CCB-0803.1
DISCUSSION
Characteristics of Neonate Flatback Diving
Flatbacks released in open water are active swimmers capable of diving for time periods in excess of 5 minutes and to depths > 8 m, even within the first 3 weeks of age. However, the majority of their dives are shorter (≤ 2 minutes) and shallower (< 4 m) in depth, regardless of profile (Figs. 5 and 6). Even young flatbacks can dive in excess of 11 m (Fig. 6) or close to depths (12–13 m) where somewhat larger juvenile flatbacks (11–20 cm SCL) were most frequently captured as bycatch in trawls (Walker 1991a) before the introduction of turtle excluder devices (TED).
Although our knowledge of flatback migration at an early age is meager, current thinking suggests it consists of 2 stages. After hatchlings from eastern Queensland swim offshore and consume their yolk supply, they assume “a surface water dwelling, planktonic life … over the continental shelf … inside the Great Barrier Reef lagoon” (Limpus 2007, p. 20). How long this neritic stage lasts is unknown. It terminates when the turtles as larger juveniles move to subtidal soft-bottomed habitats at shallower depths, without otherwise changing their geographic distribution as coastal organisms (Limpus 2007). These slightly larger turtles are preyed upon by white-bellied sea eagles (Haliaeetus leucogaster). Curved carapace lengths (CCL) of flatback skeletons at sea eagle middens range between 12.2 and 20.3 cm at a site in the Gulf of Carpentaria and between 11.3 and 20.5 cm at a site in the Great Barrier Reef (Cullen Island and Arch Rock; Walker 1991b). Although we lack a conversion of CCL to SCL for this species, CCL is generally about 7%–10% longer than the SCL in the sister species, C. mydas (J. Wyneken, unpubl. data). The smallest turtles taken by sea eagles were not much larger than the average size of our turtles at 7 weeks of age (9.4 cm SCL). These comparisons suggest that in nature (where growth rates can be slower than in captivity), the neritic phase is likely short in duration (months rather than years). The dive depths achieved by our turtles also suggest that even small, neritic phase turtles are capable of searching for food on the bottom at the shallower habitats they later occupy as juvenile turtles.
Flatback turtles forage in habitats characterized as turbid (Limpus et al. 1983; Zangerl et al. 1988). Walker (1991a) provides some general data on underwater visibility. For neritic turtles foraging inside the Great Barrier Reef lagoon, visibility (as determined by secchi disk measurements) ranges between 2.2 and 8.8 m. By comparison, in the outer shelf (ocean side of the Barrier Reef), it ranges between 15 and 20 m and is probably comparable with the visibility experienced by other species of marine turtles during their oceanic phase. For flatbacks that have returned as juveniles to inshore waters, visibility probably varies with the season. During the “wet” (rainy season between December and February), outflow from rivers contributes an increased volume of suspended material to shallow bays, estuaries, and mangroves where flatbacks are often found (Zangerl et al. 1988). We completed our study between February and April. During that time period, we were unable to see turtles that dove > 1.0 m below the surface. Under those conditions of restricted visibility, most flatback dives were (as predicted) V- and W-shaped in profile. Most dives were also shallow (4 m or less), with only occasional dives to deeper (up to 10 m) depths (Fig. 5). This pattern suggests that during the neritic phase the turtles most often scan the environment for objects of interest in the upper portion of water column (prey such as macroplankton; predators such as sharks; Limpus 2007). An alternative hypothesis is that this distribution of dive depths most efficiently enables the turtles to search for their prey under reduced visibility (see next section, below).
We postulated that because neonate flatbacks during their neritic phase are found in much the same habitats (shallow, turbid coastal waters; Zangerl et al. 1988; Walker 1991a), their diving behavior would show few changes during the first few weeks of growth and development. To a large extent, those predictions were confirmed because our data did not reveal any significant changes in either the type of dive profile shown or in their frequencies of occurrence during development. Instead, we found that significant changes in diving depth and duration were correlated with change in mass but not age (Fig. 6). We cannot explain why turtles of larger mass should make proportionally more shallow dives than smaller turtles. Their tendency to dive for longer durations may very well be a function of an increase in blood volume (and oxygen-storage capacity; Lutcavage and Lutz 1997) that accompanies their increase in mass.
On the basis of their body mass at emergence, body shape, and ecological circumstances we hypothesized that flatbacks would swim rapidly. However, the descent speeds shown by the turtles (excluding turtles 18 and 19) ranged between 1 and 20 cm/s, with averages < 9.00 cm/s (Table 2). These relatively slow descent speeds were comparable with those shown by neonate leatherbacks (3.4–16 cm/s, with a mean of 7.1 cm/s; Salmon et al. 2004) of comparable age (2–10 weeks) but of greater length and mass (60–126 g). We concluded that neonate flatbacks during their “routine” dives swim slowly, especially as they increase in mass. Doing so may be advantageous for several reasons, for example, to conserve energy by reducing drag and to more effectively detect prey (or predators) under conditions where water clarity is poor.
Turtles 18 and 19, however, behaved differently. These turtles were tested at the same site on the same day. Both completed many more V-dives during their 30-minute trial than any of the other turtles. A large proportion of these dives was deep (up to 11.5 m) and many dives were remarkably short (≤ 20 seconds; Fig. 7). Descent speeds averaged approximately 50 cm/s, but some were > 1 m/s (Table 2), even while towing a TDR. Relationships between depth and duration during these dives (Fig. 7) were clearly different from those made by other turtles during their routine V-dives (Fig. 5). The speeds involved were also faster than those recorded for any small juvenile marine turtle. For example, green turtles of comparable age but smaller mass (2–8 weeks; 35–70 g) had descent speeds that ranged between 8.2 and 41 cm/s and averaged 21.3 cm/s (Salmon et al. 2004). Green turtles were, until this study, considered the fastest of the marine turtles whose swimming speeds had been quantified (data summarized in Wyneken 1997).
Because we were unable to observe the turtles during their dives, we can only speculate as to what might have induced turtles 18 and 19 to swim so rapidly. The presence of predators near the surface was an unlikely cause because dives were shorter than usual and the turtles returned to the surface (and potential danger) frequently. Similarly, the presence of predators near the bottom was also unlikely because the turtles repeatedly made deep dives that also would have increased their vulnerability. Another possibility is that the 2 turtles behaved atypically because both came from the same nest and either possessed unique genes or were exposed to unique conditions during development. However, three of their siblings were tested at different times and at other locations; they showed “routine” dives. We, therefore, hypothesized that turtles 18 and 19 were by chance released at a site where they were exposed to an attractive stimulus (perhaps food) that elicited exceptionally vigorous diving activity.
Comparisons of Diving Behavior Among Species
How does flatback diving behavior and development compare with other species? Can the differences observed be correlated with ecology? Neonate diving behavior has now been studied in 3 marine turtle species, and in each, dive profiles vary in frequency, type, and temporal characteristics. In leatherbacks, 2–10 weeks of age, for example, V-dives dominate (Salmon et al. 2004) as they do in conspecific adults. W-dives were observed when young leatherbacks found jellyfish, began feeding, and moved up and down in the water column as they repeatedly approached the same prey for another attack. As neonate leatherbacks grow, the majority of their dives continue to be relatively shallow, but their deepest dives become deeper (Salmon et al. 2004), a change that probably reflected their increasing physiological capacity to store oxygen as well as the potential benefits that accrue with niche expansion toward a greater variety of deeper-dwelling gelatinous prey. This hypothesis is also supported by how the turtles dive; they descend almost directly downward and then ascend almost directly upward, maximizing their vertical but minimizing their horizontal dive displacement. Because these movements are accomplished with little change in swimming speed, the profiles of their dives generated a significant relationship between depth and duration (Fig. 8, top graph; statistical analysis reported in Salmon et al. 2004).
Citation: Chelonian Conservation and Biology 9, 1; 10.2744/CCB-0803.1
A similarly tight relationship between depth and duration also occurs in green turtles (Fig. 8, middle graph). With growth, larger neonate green turtles tend to dive significantly longer and deeper, but the magnitude of change between younger (2–4 weeks) and older (6–8 weeks) turtles is small because few of the routine dives by any turtle exceed 5 m. Green turtles also shift from mostly V-dives at 2–4 weeks to mostly U-dives at 6–8 weeks of development (Salmon et al. 2004). As a result, the dives of the older turtles do not change much in depth, even though they last longer. Thus, in contrast to leatherbacks, the behavior of neonate green turtles suggests that the resources they require for survival (food; shelter in the form of Sargassum rafts [Smith and Salmon 2009]) are located close to the surface. Young green turtles are capable of making much deeper dives (down to 18 m during the acclimation period) but do not dive deeply after the acclimation period (Salmon et al. 2004).
Flatback neonates show entirely different relationships between dive depth and duration (Fig. 8, bottom). We hypothesize that those differences arise as a consequence of hunting for food (and avoiding predators) in waters of limited visibility, an ecological condition that is unique for neonate marine turtles. Young leatherbacks and green turtles, for example, are found in an open ocean environment where water clarity is usually excellent. Not surprisingly, both species during this stage of development depend primarily upon vision to locate food (Constantino and Salmon 2003; unpubl. obs. on green turtles by M. Salmon). Leatherbacks also possess a specialized foveal region in their retina that may be designed to detect prey in the water column beneath them (Oliver et al. 2000).
Neonate flatbacks, however, must search for prey whose location (either in the water column or on the bottom) under most conditions cannot be determined from the surface. Even when food is located, flatbacks face another problem: how to find that site again after surfacing to breathe. Currents near the surface can displace the turtles horizontally in directions that are difficult (if not impossible) to determine. For these reasons, we believe that water turbidity has shaped flatback diving to optimize searches for food patches and to reduce the probability of losing contact with food once it is found.
We hypothesize that the “routine dives” of flatbacks function primarily to search for food. They are organized as mostly shallow V- and W-dives of relatively short duration interspersed with occasional deeper dives of variable duration (Figs. 5 and 6). This combination of shallow and deeper dives resembles a model of the most efficient method of searching for prey whose location in the water column cannot be predicted from the surface: the Levy walk (Sims et al. 2008). Computer simulations show that such a foraging strategy optimizes the probability of contact with prey and closely approximates the diving patterns shown by many diving predators (sea turtles, penguins, sharks, and some teleost predators) that face the same problem (diving “blind” to the location of prey). It, therefore, would be of interest to quantify enough dives of neonate flatbacks to determine whether they, too, fit the model.
Rapid swimming movements might be one mechanism used to maintain contact with locations where food is abundant. Another mechanism may be an ability to rapidly replenish oxygen stores while at the surface between dives. The dive records for turtles 18 and 19 indicate that rapid swimming occurs and suggest an ability to make consecutive, relatively deep dives in rapid succession (Fig. 7). Adult flatbacks possess blood hemoglobin with a greater affinity for oxygen at high partial pressures than the hemoglobin of loggerheads and, most likely, green turtles (Sperling et al. 2007). Neonates may also have these capabilities.
It might be argued that a better strategy for a small turtle that has located food is to continue feeding at the site for as long as possible before returning to the surface. That could be optimal for turtles that are larger. But, we suspect that, for a small turtle that is vulnerable to many predators, survival may be enhanced by retaining sufficient oxygen to sprint away from harm if it appears, take evasive action, and, in doing so, live to feed another day.
Inset, outline of Australia showing location of the Townsville area (square box) in northern Queensland. Below, the 7 testing sites where flatback diving trials took place. One site on the northwest side of Magnetic Island was used to avoid large waves generated by strong winds. Water depth varied between 9 and 12.5 m.
Flatback with a time–depth recorder (TDR) at the surface between dives. Inset, flatback with TDR covered by a foam sheath used to make the device slightly positive in buoyancy.
Change in mass with age (days since nest emergence at 0) by 33 neonate flatbacks from 4 nests. The dives of 22 of these turtles were analyzed in this study.
The 3 types of dive profiles (W, V, and U) shown by neonate flatbacks. These dives were made by turtle 8 at 3 weeks after emergence. Time between marks = 14 seconds.
Depth vs. duration plots of the 3 dive types. U-dives were short and shallow; W and V dives were longer and deeper.
Depth vs. duration plots for groups of turtles differing in mass (above) and age (below). The V-dives of turtles 18 and 19 are excluded from these plots (and presented in Fig. 7).
Dives performed during trials by turtles 18 (above) and 19 (middle). Time marks are at 10-second intervals. Below, depth vs. duration plot for the V-dives (n = 47) of these 2 turtles. Compare this distribution to the distribution of V-dives shown by all other neonates (Fig. 5). The latter were, on average, shallower in depth and longer in duration.
Depth vs. duration plots for all of the dives performed by neonate leatherbacks (top, n = 86 dives by 21 turtles) and green turtles (middle, n = 299 dives by 33 turtles) in a previous study (Salmon et al. 2004) and by neonate flatbacks (n = 192 dives by 21 turtles) in this study.
