Conventional studies of the diving behavior in sea turtles have mostly relied on data logging instruments such as time depth recorders (TDRs). Only rarely have attempts been made to study foraging turtles through direct in-water observation (Booth and Peters 1972; Houghton et al. 2003; Dunbar et al. 2008; Schofield et al. 2008), given the difficulty of working underwater (Forbes 1999; Hazel et al. 2009). Constraints chiefly include inadequate access to the study subject, variable visibility and sea state, physical danger and depth-related SCUBA diving restraints (Hooker and Baird 2001). Furthermore, turtles are often timid when approached by divers and may alter their behavior (Forbes 1999). At locations where these constraints can be eliminated or effectively managed, however, direct in-water observations may elucidate a fuller range of behavioral patterns than TDRs (Schofield et al. 2006) and serve to validate TDR data (Blumenthal et al. 2009).

Because the time an individual turtle can spend underwater principally depends on the amount of oxygen stored in its body and the rate at which oxygen is consumed (Kooyman 1989), data on respiratory rates and activity levels are critical to an understanding of sea turtle diving behavior and physiology. TDRs accurately record data on depth utilization and dive profiles (van Dam and Diez 1996a, 1996b; Blumenthal et al. 2008) but not on the underwater activities of turtles during dives (Seminoff et al. 2006), nor do they quantify respiratory rates during surface intervals.

In this study we present diving pattern information obtained over a 4-year period, based on direct in-water observations of a resident population of immature foraging hawksbill turtles (Eretmochelys imbricata) on a small coral reef in the Seychelles. Our study quantifies dive depth and duration, breath rates during surface intervals, and time spent in various categories of daytime activities. It also offers some subjective behavioral observations. Together, these provide new insights into the diving behavior and physiology of immature hawksbills. Understanding these interactions is particularly important given the Critically Endangered status of the hawksbill turtle globally (Mortimer and Donnelly 2008) and the precarious status of coral reef ecosystems (Spencer et al. 2000), which comprise their primary foraging habitat.

METHODS

Study Site

D'Arros Island (S 5°24.9′, E 53°17.9′), located in the Republic of Seychelles in the western Indian Ocean (Fig. 1), is part of a chain of small islands that comprise the Amirantes Group. In-water observations were conducted on an insular platform reef approximately 1 ha (10,000 m²) in size and located 100 m northeast of D'Arros Island (Fig. 1). This platform reef was selected on account of its accessibility (proximity to research station and shore access for SCUBA divers) and apparent high turtle-sighting frequency. The reef consists of a shallow platform (6-m to 10-m depth), which is surrounded by a steep reef slope that ends abruptly on a sandy plateau at depths of 20–26 m. Platform substrate is consolidated coral rubble, and benthos comprises primarily scleractinian hard corals. The surrounding reef slope consists mostly of unconsolidated reef rubble that is sparsely populated by gorgonian corals. Juvenile turtles were always encountered during dive surveys and ranged from 30 to 75 cm curved carapace length (CCL). The turtles feed predominantly on 2 species of cryptic sponge, which they excavate from the consolidated rubble substrate of the platform (von Brandis, unpubl. data).

View Figure

• Download as Powerpoint
• Download Figure

In-Water Observations

In-water observations were made at the study site from 2006 to 2009 by the senior author (RGvB), who conducted up to 4 shore-entry SCUBA dives daily (x ¯ duration  =  86 minutes, n  =  235) over a total period of 8 months (August–September 2006, January–February 2007, April–May 2008, and June–July 2009). Using SCUBA gear, systematic searches of the reef for turtles were conducted and upon locating a turtle, direct observations continued for as long as possible. Existing literature indicates that juvenile hawksbills display patterns of diurnal activity and nocturnal resting (van Dam and Diez 1996b; Blumenthal et al. 2008), but our study was unable to corroborate this because of difficulty in locating and approaching turtles at night.

All data were collected during dives that commenced between 0700 hours and 1750 hours and ended before or shortly after sunset. A turtle “identikit” was produced using facial profile photographs (Reisser et al. 2008; Schofield et al. 2008) and other distinguishing body marks. Although 9 individuals were commonly encountered on the reef, only the 4 most frequently observed turtles (named LG, PB, NS, and NM) were included in the present analysis because they showed little sign of disturbance when approached; moreover, they were observed almost daily. Activities of turtles during their dives were broadly classified as: 1) stationary foraging (excavation or ingestion of prey whilst maintaining position on the reef); 2) active foraging (probing or searching for prey whilst moving over the seabed); and 3) resting (stationary on seabed without foraging). Duration of each activity was timed and recorded using a stopwatch and writing slate.

Turtle depth was determined using the water pressure transducer of a standard wrist-worn dive computer (Suunto Mosquito, www.suunto.com/mosquito) but only recorded when the observer was horizontally aligned with and in close proximity (< 2 m) to a stationary turtle. Our methodology lacked the capacity of a TDR to produce continuous depth profiles of a turtle as it moved (swimming or crawling) over the seabed. Nevertheless, we believe our calculated dive depths are representative because stationary foraging (during which the observer was always < 2 m from the turtle) accounted for most of the turtle activity recorded.

Dive duration and surface intervals were measured using a stopwatch, and when visibility permitted, breaths were quantified by counting the number of times the head was lifted out of the water. In order to minimize disturbance to the turtle, the observer recorded data quantifying breath rates and surface intervals while remaining on the seabed rather than following the turtle to the surface. Curved carapace lengths were estimated by lightly applying a flexible measuring tape to feeding individuals at opportune moments when their vision was obscured. Additionally, field notes and photographic records pertaining to ascents, descents, surfacing, and buoyancy-related body positions were collected. Water temperature at the study site was recorded at 10-minute intervals during the entirety of 2007 using a StowAway TidBit data logger (Onset Computers, www.onsetcomp.com) with an accuracy of 0.4°C. The logger was secured at a depth of 6 m (low tide) on the eastern perimeter of the platform reef.

Data Analysis

Statistics describing mean and maximum dive duration (minutes) and dive depth (meters) were calculated. The nonparametric Kruskal-Wallis test was used to determine whether dive duration varied significantly between categorized depth ranges and the product moment correlation coefficient was used to determine how the 2 variables were related. Variation in mean duration and dive depth among the 4 turtles was examined using Kruskal-Wallis tests. Additionally, because these turtles represented different size classes (35, 44, 56, and 75 cm, respectively) the product moment correlation coefficient was used to evaluate whether body size was correlated with dive duration or depth.

A sample of 10 surface intervals (seconds) was obtained for each of the 4 turtles and averages compared by means of analysis of variance (ANOVA). The Product Moment Correlation Coefficient was used to test whether surface intervals were correlated with dive duration or body size.

ANOVA was used to test for variation in the number of breaths taken by the 4 turtles. In order to facilitate statistical comparison, foraging depths were grouped into 3 categories: 5–7 m, 8–10 m, and 11+ m. Next, ANOVA was used to test for variance in the number of breaths taken before descending to each depth category. Kruskal-Wallis tests were used to determine whether the duration of foraging dives varied significantly in the number of breaths taken before or after a particular dive. Finally, the relationship between the duration of surface intervals and the number of breaths was determined using the product moment correlation coefficient.

To evaluate the effect of time of day on dive patterns, mean depth and duration for dives conducted before and after 1200 hours were compared for 2 of the focal animals using t-tests. Data for the remaining 2 turtles were insufficient and skewed either toward the morning or afternoon period and did not conform to statistical testing.

RESULTS

Over the 4-year study period, turtles LG, PB, NS, and NM were under observation for 8, 33, 48, and 61 hours during which 19, 55, 53, and 60 dives were recorded, respectively (Table 1). Although all 4 turtles were sighted each annum, resighting frequencies varied considerably. Stationary foraging accounted for more than 76% of the activity of all 4 turtles during bottom times (Table 1). Although resting was one of the behaviors quantified in our study, none was observed and consequently all dives were foraging dives.

Table 1 Carapace length, sampling effort, activity during bottom times and dive information for the 4 most frequently observed turtles (LG, PB, NS, NM) at the study site. Standard deviations in parentheses.

View Fullscreen

Foraging dives ranged in duration from 10 to 62 minutes (x ¯  =  27.4, SD  =  9.3) and in depth from 5 to 16 m (x ¯  =  8.2, SD  =  1.6). Mean dive duration varied significantly (K₈  =  28.4, p < 0.01) but correlated positively with depth (r  =  0.95, df  =  8, p < 0.01; Fig. 2). Dive duration (K₃  =  66.5, p < 0.01) and dive depth (K₃  =  69.3, p < 0.01) varied significantly between turtles. Carapace length correlated positively with dive duration (r  =  0.995, df  =  3, p < 0.01) but not dive depth (r  =  0.355, df  =  3, p > 0.05).

View Figure

• Download as Powerpoint
• Download Figure

Whilst at the surface, turtles orientated their bodies horizontally with flippers extended laterally. Surface intervals ranged from 34 to 163 seconds (x ¯  =  81.5, SD  =  38) and varied significantly between turtles (F_3,36  =  95, p < 0.01, F_max < F_crit). The duration of surface intervals was positively correlated with carapace length (r  =  0.96, df  =  3, p < 0.01) but not with foraging dive duration (r  =  0.25, df  =  39, p > 0.05).

Breath rates were counted during 86 surface intervals and in each case, the turtle displayed consistent behavior. It lifted its head high out of the water, forcefully exhaled, immediately inhaled, resubmerged its head (apparently scanning its surroundings), then lifted its head again to repeat the process for however many breaths it took. With the final breath the turtle immediately descended. Breaths were evenly spaced over the surface interval, and turtles did not linger on the surface prior to or after respiration. The number of breaths taken between foraging dives ranged from 4 to 10 (x ¯  =  6.6, SD  =  1.3). No significant variation in number of breaths was found between individuals (F_3,183  =  1.75, p  =  0.16, F_max < F_crit) or foraging-depth categories (F_2,152  =  1.36, p  =  0.26, F_max < F_crit). The duration of foraging dives did not vary significantly in the number of breaths taken either before or after a particular dive (K₃  =  7.1, p > 0.05; K₃  =  8.7, p > 0.05, respectively). Predictably, there was a positive correlation between the duration of surface intervals and the number of breaths (r  =  0.67, df  =  39, p < 0.01).

For the 2 turtles with enough data to test statistically, neither dive depth nor duration differed significantly between the morning and afternoon dives: for turtle PB (t  =  0.06 and 1.28, respectively, df  =  53, p < 0.05) and NS (t  =  1.38 and 1.14, respectively, df  =  51, p < 0.05).

Mean water temperature at the study site was 29.09°C, ranging between a minimum of 25.63°C in July and a maximum of 31.05°C in April.

DISCUSSION

Diving patterns of immature foraging hawksbill turtles have been previously described at foraging sites in the Caribbean at Mona Island (van Dam and Diez 1996a, 1996b) and the Cayman Islands (Blumenthal et al. 2008), in Japan at the Yaeyama Islands (Okuyama et al. 2004), and in the Granitic Seychelles around Mahé Island (Houghton et al. 2003); data from all 6 studies are summarized and compared in Table 2. The present study and those of van Dam and Diez (1996a), van Dam and Diez (1996b), and Okuyama et al. (2004) are based on a relatively small number of animals, and results may not be entirely representative.

Table 2 Summary of the diving patterns of immature foraging hawksbill turtles at D'Arros Island and other study sites. Study methods included the use of time-depth recorders (TDRs) and in-water observations (IWO).

View Fullscreen

In all 6 studies, mean dive depth was relatively shallow (< 15 m) but maximum depths of 91 m at Cayman Islands (Blumenthal et al. 2008) and 72.1 m at Mona (van Dam and Diez 1996a) were recorded. Various factors probably determine the depth to which turtles dive, including distribution of food resources, physical conditions, and predator avoidance (Blumenthal et al. 2008). At D'Arros Island, the preferred food resources are often concentrated in narrow zones, which correspond to the depths at which turtles were most frequently sighted (von Brandis, unpubl. data). Mean dive duration at D'Arros was longer than that recorded at the Mona and Cayman Islands reef sites despite similarity in mean dive depths (Table 2). Dissimilarities in the mean body sizes of study subjects may in part account for this discrepancy (as body size was positively correlated with dive duration), but differences in respiration may have been more important. Specifically, turtles at D'Arros spent more time at the surface (x ¯ surface interval at Mona  =  50.5 seconds, and at D'Arros  =  81.5 seconds) and took more breaths (x ¯ no. of breaths at Mona  =  3.2, and at D'Arros  =  6.6). It follows that D'Arros turtles replenished their oxygen supplies more effectively and, conceivably, may have been capable of longer dives. Sea temperature is known to influence dive duration (Hazel et al. 2009), but temperatures were similar at the 2 sites: Mona, x ¯  =  27.72°C, range  =  4°C (van Dam and Diez 2002); and at D'Arros, x ¯  =  28.09°C, range =  5.42°C.

Average dive duration at D'Arros was 27.4 minutes, and although turtles occasionally remained submerged for periods of up to 62 minutes, activity patterns (and, thus, apparent metabolic rates) were consistent. Surface intervals and the number of breaths prior to and after longer dives did not differ significantly from those of shorter dives, which suggests that turtles were not physiologically challenged. The noteworthy disparities between average and maximum dive depth and dive duration in both the Caribbean and the Seychelles (Table 2) and the apparent ease at which long and deep dives are conducted suggests that immature hawksbills probably function well below their physiological limits during routine foraging activities and do not strive to maximize bottom times. This is in contrast to what has been suggested for leatherback turtles (Dermochelys coriacea; Eckert et al. 1989). Perhaps turtles that utilize deeper food resources more closely approach their physiological limits because of the increased energy expenditure required to travel greater vertical distances through open water and the stress associated with increased exposure to predators in such an environment (Hiethaus et al. 2007).

The positive correlation between dive depth and dive duration has been well-documented in sea turtles (van Dam and Diez 1996a; Hays et al. 2000, 2004; Hazel et al. 2009). Buoyancy-related constraints play an important role in this relationship because, in order to remain neutrally or negatively buoyant at shallow depths, turtles must restrict the volume of air in their lungs (Hochscheid et al. 2007; Hazel et al. 2009). This is reflected in the short mean dive durations recorded at the Mona cliff wall and the shallow reef at Mahé Island (Table 1). Turtles descending to greater depths can do so with greater volumes of air in their lungs, which enables them to remain submerged for longer periods. Assuming that turtles strive to maximize air volume in the lungs, neutral buoyancy should be achieved by the time the turtle reaches foraging depth. At D'Arros, mean foraging dive depth was 8.2 m, and turtles usually attained negative buoyancy by the time they reached a depth of around 3–6 m (von Brandis, pers. obs.). Despite resultant shorter dive duration, turtles may benefit from purposeful underinflation of the lungs in 2 ways: 1) negative buoyancy may enhance stability, thus improving foraging success during the extraction of cryptic food; and 2) turtles can access shallower resources during the same dive without having to surface to adjust lung volume.

The positive correlation between body size and dive duration in foraging hawksbills has been reported by several authors (van Dam and Diez 1996a, 1996b; Blumenthal et al. 2008). In sea turtles, lung volume increases proportionally to body mass (Hochscheid et al. 2007), and because the lungs store the majority of oxygen required for diving, larger turtles are capable of longer dives (Blumenthal et al. 2008). In addition, mass-specific metabolic rates of larger individuals are lower (Schmidt-Nielsen 1997), so less oxygen is required per kilojoule of energy. On the other hand, this study supported previous work (van Dam and Diez 1996b, 1997a) showing a positive correlation between body size and surface intervals, suggesting that larger turtles may be better able to replenish their oxygen levels during respiration and, thus, conduct longer dives. At Mona, Kontos and Eckert (1988) and van Dam and Diez (1997a) found that adults generally took more breaths during surface intervals than did juveniles. But, this was not the case at D'Arros (no significant difference between size classes in the average no. of breaths); instead, larger turtles simply held their breaths longer, suggesting more efficient oxygen uptake.

At Mona, turtles spent less time at the surface, took less than half the number of breaths between foraging dives, and averaged shorter dive durations than those at D'Arros. We suspect that D'Arros turtles may have higher energy demands. Turtles at Mona apparently graze predominantly on exposed prey (van Dam and Diez 1997b) while food resources at D'Arros are cryptic and require energy demanding excavation techniques (von Brandis, unpubl. data). So it is conceivable that, if Mona turtles were to increase their surface intervals and breath rates, they too would be capable of average dive durations well above the half-hour mark. One might assume that turtles are relatively more vulnerable to predation when surfacing and in fact, their behavior suggests this is so. They appear more wary during surfacing events, visually scanning their surroundings, and indeed are more easily frightened by divers at those times (von Brandis, pers. obs.). It follows that they might benefit from extended bottom times and fewer surfacing events. Our data indicate that it is within their physiological capability to extend their bottom times substantially, so it is unclear why immature hawksbills generally limit their foraging dives to less than half an hour. Perhaps the increased oxygen demands during sustained long dives do not support the metabolic rate that the turtles require to maintain optimal body condition and growth.

This study shows that direct in-water observations can yield turtle dive information that compliments the more extensive data obtained by animal-borne data logging instruments such as TDRs. The latter record dive profiles continuously and with greater detail and accuracy. TDRs also enable several turtles to be monitored simultaneously for periods in excess of a year, thus eliminating bias associated with direct observation during limited periods of time. Nevertheless, in-water observation can assist in the interpretation of remotely gathered data through instrumentation (Schofield et al. 2006; Blumenthal et al. 2009). In-water observation has enabled bottom time activity levels to be quantified (Houghton et al. 2003), dive profiles of TDR-fitted turtles to be correlated with in-water observations of their behavior (Blumenthal et al. 2008), and quantification of respiratory rates and activity levels of 4 individuals over a 4-year period (present study). Juvenile hawksbill turtles are particularly suitable for long term in-water observation studies because they: 1) typically occur in accessible neritic habitats (Musick and Limpus 1997), 2) often appear to have restricted home ranges, thus making frequent resightings possible (van Dam and Diez 1998), 3) are not easily disturbed while feeding (Blumenthal et al. 2009), and 4) can become habituated to the presence of nonintrusive observers (von Brandis, pers. obs.). In-water observations also have the potential to provide insights into such poorly understood topics as intra- and interspecific interactions, prey selectivity, rates of ingestion, feeding strategies, and habitat requirements.