In animals without parental care the choice of nest site should be a particularly important contribution to lifetime reproductive success (Resetarits 1996; Mousseau and Fox 1998). For example, in reptiles nest site choice influences temperatures in developing eggs (Deeming 2004), which in turn can directly affect embryonic survival and offspring phenotypes, including offspring sex in species with temperature-dependent sex determination (TSD) (e.g., Janzen 1994; Doody et al. 2004). Moreover, recent research has widened the possibility that reptiles with TSD may be able to allocate offspring sex through nest site choice (Roosenberg 1996; Wapstra et al. 2007).

Studying nesting behavior may provide insight into what choices reptiles are making when choosing a nest site. For instance, knowledge of how many nesting areas are visited by gravid females may reveal their scope for producing viable hatchlings or a particular offspring sex. In another example, diel timing of nesting might influence a nesting reptile's ability to discriminate between sites with high and low reproductive success because of daily variation in substrate temperatures. Simply collecting nest site choice data in the absence of knowledge of associated behavior could thus lead to an incomplete diagnosis of how nesting turtles optimize offspring fitness.

Although nesting behavior has been documented in several turtle species (see reviews in Ehrenfeld 1979; Hirth 1980; Hailman and Elowson 1992), observations are biased toward sea turtles and temperate North American species (in particular the Emydidae) because of observability and the boreal bias (Ehrenfeld, 1979; Platnick 1991). More data from tropical nonmarine species are needed to understand the ecological and evolutionary implications of nesting behavior through comparative study. Moreover, diel timing of nesting is unknown for most turtle species (e.g., Ernst et al. 1994).

We used focal observations and remote camera systems to study the nesting behavior of the pig-nosed turtle, Carettochelys insculpta, for 3 years in the wet-dry tropics of northern Australia. Carettochelys is the single extant representative of the Carettochelydidae, one of the few turtle families with no published accounts of nesting behavior. We hypothesized that nesting behavior in C. insculpta would be similar to that of the softshell turtles (Trionychidae), the sister group to Carettochelys (Gaffney and Meylen 1988; Shaffer et al. 1997). Specifically, we addressed the following questions: 1) How many nesting areas does C. insculpta visit in search of nest sites? 2) What behavioral patterns are exhibited just prior to the nesting emergence? 3) At what time of day or night does nesting occur, and at what temperatures? Previous research has indicated that C. insculpta nests at night (Georges 1992; Doody and Georges 2000), but no direct observations have been published. 4) What is the duration and nature of the nesting sequence? We compare our findings to nesting in other turtle species, particularly North American softshell turtles. We conclude by generating hypotheses for why some turtles nest during the day and others at night.

METHODS

Carettochelys insculpta is a large freshwater species inhabiting rivers and associated habitats in southern New Guinea and northern Australia (Georges and Rose 1993; Georges et al., 2008). In Australia, Carettochelys is restricted to a few rivers and billabongs in the Top End of the Northern Territory. On the Daly River, Carettochelys nests on open sandy areas adjacent to rivers during the winter dry season from July through October (Georges 1992; Doody et al. 2003a, 2004).

We studied C. insculpta nesting behavior along a 30-km stretch of the Daly River near Oolloo Crossing, Northern Territory (14°04′40″S, 131°15′00″E), between 15 July and 15 October during 1997–1998, and again in October 2007. The Daly River is characterized by clear shallow water (< 2 m) during the dry season and turbid deep water (up to 20 m) during the wet season (Doody et al. 2002). We visited nesting beaches daily by motorboat, surveying for nests by probing the sand along crawls on suspected nesting beaches. Shaded air and water temperatures were recorded every 15 minutes by a Datataker™ 10 channel datalogger.

To document nesting times, we used Trailmaster™ remote camera systems (Doody and Georges 2000; Doody et al. 2001a). Each system consisted of an infrared beam transmitter, a programmable receiver with LED readout, and an automatic 35-mm camera (Olympus™). Transmitters and receivers were placed at opposite ends of a beach approximately 20 cm above the water. Each camera was attached to a stake emerging from shallow water. A time-stamp and photo recorded each turtle as it crossed the infrared beam upon emergence onto the beach and upon departure. Cameras were affixed close enough to the beach to allow clear identification of individuals in photographs. Turtles were marked with numbered cattle ear tags attached to the rear carapacial edge in a concurrent study (see Doody et al. 2002). Some individuals nested (nesting events) and others did not (nonnesting events). We distinguished nesting events from nonnesting events by locating nests during daily surveys in 1997–1998 (for more details, see Doody et al. 2002, 2003a, 2004).

To further document nesting behaviour, we made focal observations of 5 nesting turtles in 2007; in 3 of these, we obtained hand-held video footage to facilitate detailed analysis of the nesting sequence. Because previous research suggested that C. insculpta nests at night (Doody et al. 2003b), we typically watched for nesting turtles from dusk to approximately 2400 hours. Care was taken to remain hidden from emerging turtles and from saltwater crocodiles (Crocodylus porosus). In over 35 nights, we accumulated approximately 250 person-hours watching for nesting turtles from July and October in 1997 and 1998, and in 2007, we spent 6 nights on 3 nesting beaches over a 2-week period in early October.

RESULTS

Remote camera data revealed that C. insculpta nested at night, with the exception of one individual that nested at dusk (Fig. 1). Nesting times (X  =  21:01 ± 00:19:47 SE; range  =  18:32–00:48; n  =  20 events) were normally distributed (Fig. 1). Nesting events lasted 26.5 ± 6.43 SD minutes (n  =  20 bouts) and were significantly longer (single-factor ANOVA; F_1,147  =  299.1, p < 0.001) than nonnesting events (X  =  3.6 ± 4.94 SD minutes; n  =  128 bouts). Timing of nonnesting events (X  =  21:24 ± 00:08:48 SE; range  =  19:02–03:08; n  =  129) did not differ significantly from that of nesting events (Fig. 1; single-factor ANOVA; F_1,148  =  0.92, p  =  0.338). Nocturnal nesting meant that water temperatures were warmer than air temperatures during nesting forays (Fig. 2). The minimum air and water temperatures during nesting were 17.5 and 24.6°C, respectively.

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

Photographs of marked turtles documented individual use of multiple beaches, frequency of emergence onto beaches, and the number of nights individuals would emerge prior to nesting. Considering only cases of multiple photographs of an individual, 11 females emerged 5.9 ± 0.93 SD times (range  =  2–12) onto 1.6 ± 0.79 SD beaches (range  =  1–3). Most turtles using multiple beaches involved beaches that were within 1–3 km of one another, but one turtle used 2 beaches 5 km apart. Twelve individuals emerged onto beaches for 1.8 ± 0.72 SD nights (range  =  1–4), but these nights were not necessarily consecutive. The maximum time between the first and last emergence for an individual was 7 days (this represents one nesting because the internesting interval in C. insculpta is about 40 days (Doody et al. 2003a).

On 3 moonlit nights during the peak of nesting activity, we observed the behavior of turtles in the water along the nesting beach. On 2 such occasions, we documented a number of pre-emergence, intraspecific interactions. On both nights, 7–12 different turtles were seen simultaneously around the same beach, and 2 noteworthy behaviors were observed. First, when 2 turtles met face-to-face, they avoided one another by circling around one another. In contrast, when a turtle approached another from behind, the approaching turtle would often (n  =  14 observations) accelerate and collide with the other turtle, and the impact of the collision could be clearly heard from above water. The ramming behavior appeared to be deliberate and usually resulted in the receiver swimming quickly away. However, the receiver would usually return within a minute or two. Second, several smaller turtles (possibly subadult females or males) emerged and “nuzzled” the wet sand for 1–3 min and then quickly retreated to water. In all cases (n  =  12 times), the turtles were within 0.5 m of the water. These turtles were not captured to confirm sex because doing so would likely have disrupted the group, preventing further observations (J.S. Doody, pers. obs.).

Emergent females were wary while searching for a nest site. On 5 occasions, we observed crawling females collide with nesting turtles, invariably frightening the nesting individuals. In 2 cases, the nesting turtle had already constructed a nest chamber but did not return to nest on that beach that night. On the beach, females did not appear to make contact intentionally. Rather, contact seemed to reflect narrow beach widths and near-simultaneous emergence.

The nesting sequence was similar between the two turtles with complete observations (Table 1). Removal of surface sand and chamber excavation took the longest amount of time; whereas, egg laying and filling were relatively brief (Table 1). The following components were recorded for the two individuals and for additional turtles in the early stages of nesting.

Table 1 Times to perform component behaviors during a nesting bout for two individual Carettochelys insculpta (present study) and two individual softshells, Apalone spinifera (# 1  =  Breckenridge, 1960; #2  =  Cahn, 1937).

View Fullscreen

Approach

Females emerged in stops and starts, making their way into the nesting zone (0.5–5 m from water) within 1–2 minutes. During the pauses, turtles would often elevate the head a few centimeters above the sand, and sometimes side-to-side a few centimeters, possibly as vigilance behavior, or assessment of the beach or the elevation they had traversed. Locomotion involved alternating the front limbs as in most turtles, in contrast to simultaneous front flipper use in sea turtles.

Scraping

During many of the pauses, turtles would remove surface sand by flicking toward the body with alternating hind flippers. This action created a spray of sand up to 1 m high and 2 m behind the turtles. This behavior was sometimes followed by excavation of a chamber but often was not.

Chamber Excavation

This and concurrent nest surveys revealed that most nests were excavated with the turtle facing away from the river. Nest chamber construction involved the hind flipper clasping and removing sand toward the body. Removed sand was placed laterally to, and behind, the chamber. Turtles did not necessarily alternate flippers during chamber excavation, and one flipper would often excavate sand 2–3 times before the other flipper was used. The nondigging flipper was used to lift the body off the sand while the other flipper dug. The resultant chamber was generally flask-shaped. During the laying process turtles maintained a head-down position.

Egg Laying

Laying occurred immediately after chamber excavation. A slight rocking motion accompanied the deposition of each egg, which dropped into the chamber unaided. About 10 eggs are laid by C. insculpta at the site (Doody et al. 2003a). In the most easily viewed nesting sequence captured on video, the left hind flipper was positioned lateral to the chamber, bearing the weight of the turtle while the right flipper dangled inside the chamber next to the tail. After each egg was laid the right flipper moved down into the back of the chamber (posterior) and moved side-to-side while the ventral side of the flipper faced forward. The flipper may have been distributing the eggs more evenly throughout the chamber or moving any sand that was falling on top of the eggs.

Filling and Covering the Nest

Filling began just after the last egg was laid. The moist sand removed earlier was pushed and lightly compressed into the chamber with alternating hind flippers. Each time the turtle would move sand into the chamber with one flipper, the other flipper, placed just outside of the chamber, supported the weight of the turtle's posterior end (Fig. 3). The number of sweeps was similar among the 3 turtles for which video footage was obtained (16, 17, 19 sweeps). As this process was repeated, the turtle scraped drier sand from a more anterior position toward the filled nest, the flippers stretching farther forward (anterior and lateral to the body) until eventually a full extension of the flippers was achieved. This action produced an impression in the sand that superficially resembled bird wings. Each of the last 4–8 sweeps was associated with a single bobbing motion of the body that further compressed the sand into the filled chamber.

View Figure

• Download as Powerpoint
• Download Figure

Returning to the Water

After covering the nest, turtles immediately turned 180° and moved back into the water.

DISCUSSION

The movements associated with nesting and their duration may furnish rough estimates of effort in choosing a nesting beach and nest site. In the present study, C. insculpta emerged up to 12 times on up to 3 beaches; although, most emerged about 6 times over 2 nights onto 1–2 beaches. The number of beaches visited by nesting turtles may be influenced by the availability of cohesive sand, which influences nest site choice in the species (Doody et al. 2004). Some beaches in the present study lacked cohesive sand altogether, and this may have contributed to long-distance movements (up to 5 km in the present study) by some gravid turtles. In another study, one gravid female moved 6 km just prior to nesting, and home ranges of gravid turtles encompassed an average of 5 beaches (Doody et al. 2002). Collectively, these data indicate that considerable effort may be involved in searching for a nesting beach and nest site, especially relative to the apparent low energy uptake in C. insculpta during the dry season (Doody et al. 2002, 2003a).

Observations of agonistic behavior among conspecifics are currently unexplained. Social behavior has rarely been documented in turtles and is thought to be limited to dominance hierarchies (reviewed in Pearse and Avise 2001). During a concurrent mark-recapture study, we noted groups of C. insculpta feeding, moving, and thermoregulating together (Doody et al. 2001b, 2002, 2003b). In particular, groups of 2–8 gravid females moved together up and down the river in the weeks just prior to nesting. For example, 3 females observed together one day nested 2 nights later 7 km upriver. Perhaps turtles were taking advantage of the collective knowledge of several potential nesting beaches by moving together in a group. More focal observations on known sexes are needed to understand intraspecific interactions among individual C. insculpta.

Carettochelys insculpta on the Daly River nests mainly on sandy beaches and banks like members of its sister group, the softshell turtles (Webb 1962; Georges 1992). However, C. insculpta nests at night; whereas, softshells generally nest during the day (Webb 1962; Plummer 1976; Doody 1995). Nesting on open sandbars in the heat of the day in subtropical climate, as some softshells do, indicates that nocturnal nesting is not necessarily an adaptation to minimize the risk of heat stress. Rather, whether a turtle nests diurnally or nocturnally may be related to historical or current patterns of predation or perceived predation risk. Consistent with this idea, the South American turtles Podocnemis expansa and Phrynops geoffroanus, which are heavily harvested by humans, nest at night in some populations but during the day in others (reviewed in Pritchard and Trebbau 1984; see also Vogt, 2008). Uchida (1980) reported that Indonesian Eretmochelys imbricata nest exclusively at night in populated areas but by day in uninhabited areas. Similarly, some populations of P. expansa and Podocnemis unifilis, which were once known to nest in late afternoon, were known later to nest only at night, apparently because of human predation (Foote 1978). Pig-nosed turtles would have few predators as adults, given their large size (adults  =  6–12 kg in the Daly River; Doody et al. 2003a), and saltwater crocodiles (Crocodylus porosus) are probably their only major predator. Aboriginal people value C. insculpta as food, as do native New Guineans (Georges and Rose 1993) where the species probably originated (Cogger and Heatwole 1981). Another possibility is that the nocturnal timing of nesting may be explained by daily variation in activity. For example, softshells (Apalone spp.) exhibit diurnal activity and nesting (Webb 1962); whereas, the nocturnally nesting C. insculpta may be more active at night (Heaphy 1990; J.S. Doody and A. Pepper, unpubl. data). A multispecies comparative analysis would be necessary to examine any relationship between nesting times and activity times.

In the present study, C. insculpta nested in the first few hours of darkness, a time when surface (sand) temperatures have begun to decrease, but nest temperatures are near the maximum (Georges 1992). Why nesting is more or less restricted to the first half of the night is unknown but may be related to thermal properties of the nesting habitat. Nesting early in the night would facilitate detecting warm sand at a time when nesting areas can be difficult to see (C. insculpta seems to have poor vision out of water, J.S. Doody, pers. obs.). However, other nocturnally nesting species, such as sea turtles and the river turtle P. expansa, often nest after midnight (Pritchard and Trebbau 1984).

Our nonexhaustive review demonstrated that Carettochelys took less time to nest than did most other turtle species but was similar to that of the softshell Apalone spinifera (Table 1; Table 2). Hirth (1980) listed suspected influences on nesting duration in sea turtles, citing body size, nesting experience, reproductive readiness, width of the nesting area, compactness of the sand, and the density of nesters. We hypothesize that the relatively short duration of nesting in C. insculpta and at least some softshells is a result of the relative ease in excavating a nest chamber in sand and the close proximity to water of nesting beaches in these primarily riverine species, and this may be supported by short nesting duration in the sand-nesting Malaclemys terrapin (Feinberg and Burke 2003). The relatively large, flipperlike hind feet in the former two species also may hasten the nesting process.

Table 2 A nonexhaustive review of nesting durations in various turtle species gleaned from the literature. Durations include the total time spent out of water unless otherwise noted.

View Fullscreen

The short duration of nonnesting crawls was in part a result of the lack of lateral movement once females were on the beach. Remote cameras (photographing an individual crawling several times on the same beach) revealed that turtles adjust their lateral position on the beach on subsequent crawls, rather than moving laterally on the beach. The resultant courses of tracks in the sand were consistently U-shaped, unlike the often laterally meandering tracks of nesting softshells (Doody 1995).

The frequent occurrence of half-constructed or empty nest chambers in the present study was sometimes the result of incidental contact between nesting turtles. Previously, these empty chambers were thought to be “test holes”, as they are often termed in the literature (e.g., Shealy 1976). Although some empty chambers in lizards and turtles are abandoned because of obstructions such as rocks or roots, many are not. Why expend energy digging a complete hole at a satisfactory nest site, but then forego laying, only to return and nest a few centimeters from that hole in identical conditions? Incidental contact may better explain some abandoned chambers, and the occasional finding of eggs in an unfinished chamber supports this idea. In the present study, up to 5 turtles were photographed on a beach simultaneously, and up to 4 have been known to nest on a single beach in one night (Doody et al. 2003b).

In general, the nesting sequence seems to be conserved among turtles (Ehrenfeld 1979); although, there are important exceptions. For example, because nesting C. insculpta do not exhibit body-pitting, turtles are forced to nest in areas in which the shallow sand is cohesive or moist (Doody et al. 2004). Because of the conservative nature of most nesting components, here we highlight only component behaviors that differed notably between C. insculpta and other turtles.

Nesting C. insculpta used their hind legs to scrape and sweep sand prior to selecting a nest site. This is in contrast to most species, which use forelimbs for this behavior (Shealy 1976; reviewed in Ehrenfeld 1979; Pritchard and Trebbau 1984). We found no mention of such scraping in softshells. One of us saw scraping impressions in the sand along the tracks of the softshells Apalone mutica and A. spinifera (Doody 1995). However, it is unknown whether these scrapings were made by front or hind legs. Chamber excavation differed between C. insculpta and the softshell A. spinifera in 3 ways. First, C. insculpta often scooped sand with one flipper 2–3 times before changing flippers, as reported for A. spinifera by Cahn (1937; see also Jackson and Walker, 1997). However, Breckenridge (1960) apparently observed A. spinifera alternating hind legs to dig the chamber. Second, as it removes sand with its hind leg; A. spinifera apparently snaps the excavated sand back several meters (Cahn 1937; Breckenridge 1960); whereas, C. insculpta placed removed sand near the chamber. Third, while digging the chamber, C. insculpta remained head-down; whereas, according to Cahn (1937), an A. spinifera “held her head high, carefully watching”, while digging. Softshells are extremely alert while on land and in hand; whereas, C. insculpta, a species that does not typically move over land, is lethargic and less aware, often with head down.

Duration of egg laying was similar between C. insculpta and A. spinifera (Table 1). At least one C. insculpta used a hind flipper to distribute eggs (or sand) in the chamber after each egg was laid, a behavior reported for many nonmarine turtle species (reviewed in Pritchard and Trebbau 1984; Bager et al., 2007). The turtle P. expansa may use its tail to perform a similar task (Foote 1978; but see Alho and Padua 1982). Covering the nest generally comprises filling, packing, and concealment (Ehrenfeld, 1979). Like the softshells A. mutica and A. spinifera (Cahn 1937; Goldsmith 1944; Breckenridge 1960), and the green turtle Chelonia mydas (Carr et al. 1966), C. insculpta filled the nest chamber by scraping in the moist sand removed earlier with alternating hind flippers. However, in C. insculpta, each hind flipper dropped into the chamber and compressed the sand it had just scraped into the chamber. Such behavior may occur in softshells but has not been explicitly reported. However, packing behavior with the hind legs is known in other turtles (e.g., Shealy 1976). Concealment of the nest in turtles varies among species from depositing a pile of loose sand, to dragging the plastron to flatten the sand, to scratching sand onto the nest area (Ehrenfeld 1979). Softshell turtles seem to exhibit one or both of plastron packing and the scratching behavior (Cahn 1937; Goldsmith 1944; Ehrenfeld 1979). Similarly, C. insculpta exhibited packing behavior and a modified scratching behavior.