Asian turtles have declined substantially over the past few decades, largely because of the over-collection of wild turtles for the food, pet, and medicine trades (van Dijk et al. 2000). This has drawn worldwide attention to the conservation peril that many of these species are facing; turtles are rapidly disappearing from the wild, making studies on the natural history and ecology of most species difficult, because many species are now so rare that they are simply unavailable for study. The paucity of information on these species has severely hampered conservation efforts, because understanding basic habitat requirements and behavior are integral components of successful conservation or management plans, including efforts to re-introduce individuals to the wild or establish and maintain captive populations to preserve genetic material (Dodd and Seigel 1991; Bertolero and Oro 2009).

The big-headed turtle (Platysternon megacephalum) is the sole member of the family Platysternidae and ranges throughout southern China, Laos, Myanmar, Thailand, and Vietnam (Ernst and Barbour 1989). The species is widely collected for use as food and traditional medicine (de Bruin and Artner 1999; Lau et al. 2000; Stuart and Timmins 2000; Shi et al. 2004; Gong et al. 2005a, 2006a; Zhou and Jiang 2008) and, consequently, is classified as Endangered by the IUCN (Asian Turtle Trade Working Group 2000). Big-headed turtles comprised 9% of all live turtles exported from China from 1998 to 2002 (Zhao and Jiang 2008) and do not breed readily in captivity (Shi et al. 2007). Field surveys have revealed low abundances; for example, in southern China, Gong et al. (2006b) captured only eight individuals in over 2000 capture-days of effort (also see de Bruin 1998), suggesting that this species is extremely difficult to find or that remaining populations are very small or severely depleted. Big-headed turtles have been the focus of some molecular work, both to explore their phylogenetic position (Parham et al. 2006) and to investigate genetic variation in natural populations (Ma et al. 2007; Zheng et al. 2008). Despite this, there is very little information available on their basic ecology.

Big-headed turtles inhabit cool rocky mountain streams (water temperatures 12°–17°C, although reports of 24°–28°C exist; Ernst and Barbour 1989; de Bruin and Artner 1999) and rarely bask (Ernst and Barbour 1989). Most of their time is spent in the water (Druzisky and Brainerd 2001), and consequently, body temperatures are generally low (22°–27°C in summer, 17°–21°C in autumn, selected body temperature is 25.5°C; cited in Zhang et al. 2009), and researchers track water temperatures rather than ambient temperatures (Shu et al. 2009). Big-headed turtles rely on their strong claws and long, agile tail to move over the rocky stream habitat and rarely swim (Druzisky and Brainerd 2001). They are thought to be nocturnal (captive individuals feed at night), and individuals often remain hidden beneath rocks or gravel during the day (Ernst and Barbour 1989; Kirkpatrick 1995; de Bruin and Artner 1999).

Aside from this basic information, studies of movements and microhabitat use in the field are lacking. Documenting this information is crucial for population monitoring efforts, as well as for establishing conservation and management programs. We translocated 16 big-headed turtles and followed them for up to 8 months to quantify movement patterns, microhabitat use, and general behavior. Our main goal was to understand whether males and females exhibited similar movement patterns and microhabitat use and whether this species would be a good candidate for additional translocation efforts.

METHODS

We purchased adult big-headed turtles from a turtle dealer in China and subsequently released them on protected private land in Quzhou County, Hebei Province, southern China (29°26′N, 118°21′E). These turtles were recently wild-caught adults, but the dealer would not disclose their origin. It is important to note that all turtles appeared alert and healthy prior to release (counter to most individuals found in food markets; K.A. Buhlmann and M. Lau, pers. comm.). The release site consisted of a rocky mountain stream varying in width from 1.1 m to 8.7 m and up to 2 m deep. Local villagers have reported seeing big-headed turtles at the release site, although we did not see any during our study. The climate is seasonal and temperate, with average monthly temperatures ranging from 5° to 29°C and monthly precipitation varying from 70 mm to 280 mm (www.weatherbase.com).

We radiotracked a total of 16 turtles (9 males, 7 females) from April to November 2007. Prior to release, we determined the sex of all turtles (Ernst and Barbour 1989) and took morphological measurements (Table 1). A transmitter (AI-2F, Holohil Systems Ltd., Canada) was bolted to the marginal scutes of each turtle (Fig. 1). Release times were staggered as follows: 6 turtles were released in April (3 males, 3 females); 4 were released in May (2 males, 2 females); and 6 were released in September (4 males, 2 females). Although individual turtles were tracked for varying lengths of time, roughly equal numbers of males and females were tracked simultaneously. We released turtles individually along a stretch of stream several hundred meters long and left them at the study site after the transmitters expired. We tracked turtles for 4–6 consecutive days twice per month (a total of 10–12 tracking days per month) using a LA12-Q radiotelemetry receiver (AVM Instruments, US). On each tracking day turtles were located once at dawn and once at dusk; by tracking at these 2 times, we could infer movement distances during the night (i.e., data collected at dawn) and during the day (i.e., data collected at dusk). Nocturnal tracking was not attempted because of logistical constraints.

View Figure

• Download as Powerpoint
• Download Figure

Table 1 Morphology of translocated big-headed turtles prior to release, shown by sex. Shown are mean values ± SE. Most measurements are standard for turtles except that “tail length” was taken as the distance along the tail from the junction with the body to the cloaca (which standardizes this measurement for individuals with partially missing tails).

View Fullscreen

We recorded the following information during each tracking event: time period (night or day); location (on land or in the water); behavior (e.g., basking, hiding); distance from the stream bank, and straight-line distance moved from the previous location. Also, we recorded microhabitat characteristics of each turtle location, including stream width and depth (when in the water), height of herbaceous vegetation directly above the turtle (when on land), vegetation cover above the turtle (%; estimated visually), refugia type (when hiding: silt, leaf litter, logs, rocks), and substrate type (classified as silt, grit, or scree [broken rocks]; see Gong et al. 2005b). We recorded distance moved and stream width to the nearest 0.1 m and stream depth and vegetation height to the nearest 0.5 cm.

We investigated movement patterns over 2 different time scales: 1) daily movements (separated into night vs. day); and 2) movements between consecutive field trips (i.e., movements over roughly 2-wk intervals; range: 9–15 d). Daily movements excluded the first day of each field trip and were calculated separately for night and day, and the movements between field trips included the period from the last tracking event of one field trip to the first tracking event of the next field trip. We conducted movement analyses 2 different ways: 1) by calculating the mean and maximum distances moved for each turtle (both daily and between field trips) and comparing these values between the sexes; and 2) by using all movement data to compare the sexes in terms of daily distance moved and distance moved between field trips. The first analysis helped control for individual variation in movement patterns as well as the length of time a turtle was tracked by including each individual turtle only once but was limited by sample size (i.e., the number of turtles we tracked). The second analysis had a much higher sample size but could be influenced by the length of time a turtle was tracked or individuals that consistently moved very short or long distances. Also, we graphed the mean and maximum distances moved each month to look for broad seasonal patterns. To test whether turtles took time to settle into their new environment, we looked for evidence of changes in distance moved over time by graphing nocturnal movement distances as a function of the number of consecutive daily distance movement points collected. Finally, we investigated the frequency that an individual turtle moved (i.e., the proportion of days tracked that each individual remained in one location) between the sexes by calculating the proportion of days each month that each turtle moved. These were also graphed monthly.

For all microhabitat use analyses, we included only unique locations and pooled data from all individuals to make broad comparisons between the sexes. We used contingency table analyses to test for differences in microhabitat use between the sexes in terms of categorical variables, including the number of locations when the turtle was visible, terrestrial habitat use, time period (night or day), and shelter-site use. We used analyses of variance (ANOVAs) to test for sex differences in terms of continuous data, including distance from the stream, location within the stream (i.e., distance from the stream bank while using stream width as a covariate to control for differences in stream width at each turtle location), and water depth. Prior to analyses, we checked that raw data met parametric statistical assumptions and used appropriate transformations of raw data when these were not met. All means are presented ± SE, and alpha was set at 0.05.

RESULTS

Morphology of big-headed turtles in our study was similar between the sexes in most respects (p > 0.11) except that males had significantly longer tails than females (F_1,14  =  26.28, p < 0.001; Table 1). The 3 largest turtles in terms of body mass and carapace length were males. We collected 1148 turtle observations, comprising 518 unique locations. Although we have no evidence of direct mortality or predation of any turtles, on 5 occasions we discovered transmitters without the turtle (or any signs of predation, such as bite marks on the transmitter). Because the transmitters were bolted to the marginal scutes (and thus unlikely to fall off by themselves), we assume that local villagers collected these turtles and removed the transmitters.

Movements and Activity

Using averages for each individual turtle, there was no difference in overall mean or maximum daily distance moved between the sexes or with respect to time of day (mean distance: sex, F_1,27  =  3.18, p  =  0.09, time period, F_1,27  =  0.22, p  =  0.64, sex · time period, F_1,27  =  0.40, p  =  0.53; maximum distance: sex, F_1,27  =  1.18, p  =  0.29, time of day, F_1,27  =  1.40, p  =  0.25, sex · time of day, F_1,27  =  0.04, p  =  0.85). The mean distance moved per day was 7.5 ± 1.07 m (range: 0.9–33.6 m), and the mean maximum distance moved was 28.72 ± 4.07 m (range: 1.2–89.6 m). Using all data (i.e., multiple movements for each individual turtle) gave different results (sex, F_1,399  =  6.76, p < 0.01, time period, F_1,399  =  5.01, p  =  0.03, sex · time period, F_1,399  =   0.14, p  =  0.71; Fig. 2a), indicating that males moved further than females and that both sexes moved further at night than during the day (Fig. 2a). However, we found no evidence that turtles moved further immediately after release at the study site; in fact, males and females initially tended to move very short distances during consecutive nights, which increased the longer individuals were tracked (Fig. 3).

View Figure

• Download as Powerpoint
• Download Figure

Overall, turtles showed very seasonal daily movement patterns and tended to move further in the warmer months (i.e., May to August; Fig. 4a). The distance moved between field trips (i.e., over roughly 2-wk intervals) was similar for individual males and females (for mean and maximum distances, p > 0.58); the mean maximum distance moved per individual during this time was 82.0 ± 15.1 m (range: 8.5–188 m), and the overall mean distance moved was 29.0 ± 5.1 m (range: 7.0–87.6 m). These results were qualitatively similar when the analysis included all recorded movements for each individual turtle (p  =  0.27). Seasonal movement patterns were again apparent, with turtles moving further during the warmer months (i.e., May to August; Fig. 4b).

View Figure

• Download as Powerpoint
• Download Figure

Turtles often spent multiple days in the same location and were more likely to change locations at night than during the day (sex, F_1,144  =  2.45, p  =  0.12, time period, F_1,144  =  6.60, p  =  0.01, sex · time period, F_1,144  =  0.22, p  =  0.64; Fig. 2b). These movement frequencies were highest from April to September and decreased from October to November (Fig. 5). Movement occurred during all types of weather conditions, including during rain and light snow flurries.

Microhabitat Use

Turtles remained hidden much of the time; however, males were visible significantly more often than were females (6.3% vs. 2.0% of observations, n  =  518 unique locations; χ²  =  6.0, df  =  1, p  =  0.01). The vast majority of turtle observations were in the stream, with turtles found on land only 7.5% of the time. Males and females both used terrestrial habitats seasonally (during spring and early summer), but this pattern was much more pronounced in females (9.6% vs. 4.5% of the time, n  =  518 unique locations; χ²  =  5.25, df  =  1, p  =  0.02; Fig. 6). There was no difference in the frequency of terrestrial locations between the sexes with respect to time of day (night vs. day; χ²  =  7.66, df  =  3, p > 0.05).

When on land, both sexes remained near the stream (n  =  36 unique terrestrial locations; F_1,34  =  0.55, p  =  0.55); the maximum distance a turtle was recorded from the stream was 5.8 m (mean distance  =  0.95 ± 0.24 m; 75% of all terrestrial locations were < 1 m from the stream). The one observation of a basking turtle occurred on land and was of a male located 0.7 m from the stream edge on an overcast day in June. All other terrestrial locations were of turtles hiding under rocks (80.6%), silt (8.3%), leaf litter (5.6%), or logs (2.8%). Although terrestrial behavior was relatively rare, it occurred in all months in which we tracked and did not appear to be related to weather conditions or water levels.

Both males and females used similar substrates while in the water (χ²  =  3.32, df  =  2, p  =  0.19); silt was used during 14.7% of locations, scree 17.6%, and grit 67.7%. Both aquatic and terrestrial retreats were covered by low-growing vegetation, which varied from 1.0 cm to 6.8 cm in height (mean  =  3.25 ± 0.2 cm) and ranged in canopy cover from 55% to 100% (mean  =  81.3 ± 2.1%; there were no sex differences in any of these microhabitat attributes nor differences between aquatic and terrestrial locations, all p > 0.25).

When in the water, females stayed closer to the stream bank than did males (ANCOVA with sex as the treatment, square-root transformed distance from land as the dependent variable, and square-root transformed stream width as the covariate; F_1,477  =  3.87, p < 0.05; Fig. 7). On average, females were located 57.4 ± 3.4 cm from the bank, whereas males were located 64.1 ± 3.1 cm from the bank (Fig. 7). Also, females used shallower water than did males (F_1,479  =  15.17, p  =  0.0001); females were found at an average depth of 25.3 ± 1.3 cm, whereas males averaged 32.3 ± 1.3 cm. Most (97.7%) in-stream locations were of turtles hidden beneath cover, but male turtles were visible more often than were females (3.9% of male observations vs. 0.4% of female observations; χ²  =  6.46, df  =  1, p  =  0.01). Turtles were observed actively moving a total of 7 times (1.5% of all in-water observations), and this activity took place in April, May, June, and August; 6 of these observations were of 5 individual males, and the seventh was of a female. On 4 occasions, 2 individual males were visible in the water sheltering beside rocks. All other in-water locations were of inactive turtles.

When in the water, males and females used different microhabitats (χ²  =  10.45, df  =  3, p  =  0.02); females were recorded more often beneath rocks (87.1% vs. 81.7%), more often hiding under silt (4.0% vs. 2.4%), less often beneath leaf litter (8.4% vs. 11.0%), and less often beneath logs (0.4% vs. 4.9%) than were males. Males and females used different substrates while in the water (χ²  =  8.37, df  =  2, p  =  0.015); females were recorded more often on silt (9.7% vs. 4.3%), less often on scree (77.9% vs. 87.4%), and more often on grit (12.4% vs. 8.3%) than were males.

DISCUSSION

Very little is known about movements and habitat use of big-headed turtles in the wild, and our radiotelemetry study has provided several important insights into their general behavior. Overall, translocated turtles remained hidden much of the time, were relatively sedentary, and spent most of their time in the stream. In terms of movements, some general patterns emerged: on a daily basis, males moved further than females (Fig. 2); both sexes moved further at night than during the day (Fig. 2a); a higher proportion of moves took place at night (Fig. 2b); and turtles moved further and more often during the warmer months (Figs. 4, 5). Perhaps surprisingly, in April turtles moved the shortest distances (Fig. 4) but, also, made more frequent movements than in other months (Fig. 5). Part of the reason for this could be that we initially released turtles in mid-April and, thus, tracked turtles for half as long as subsequent months. Additionally, because all turtles tracked in April were newly released, they may have been making frequent but short movements because of their unfamiliarity with their environment (e.g., Fig. 3).

View Figure

• Download as Powerpoint
• Download Figure

Although big-headed turtles have been reported to be nocturnal (Ernst and Barbour 1989; Kirkpatrick 1995; de Bruin and Artner 1999), our data indicated that they move both during the day and at night but move further distances and more often at night (Fig. 2). Note that, because we only tracked turtles during daylight (i.e., at dawn and dusk), tracking at night (when they are presumed to forage) could reveal important information about habitat use and movement patterns, especially regarding the amount of time that turtles are active or visible.

In terms of microhabitat use, big-headed turtles were very secretive and remained hidden most of the time. This species rarely basks (only 1 observation of basking was made in 517 unique turtle locations) and instead remains in the stream (or in very close proximity) much of the time. Females used terrestrial habitat more often than did males, but both sexes tended to use terrestrial habitats more often in the late spring and early summer period (Fig. 5) when females are presumably searching for suitable nest sites (e.g., Obbard and Brooks 1980; Baldwin et al. 2004; Litzgus and Mousseau 2004). Also, males and females used cover objects (rocks, logs, etc.) and substrates differently. The reason for these differences in shelter sites is unclear but could be biologically meaningful. We encourage future studies to focus more on the nocturnal behavior of this species, because our inferences on movements and microhabitat use are limited by our tracking schedule.

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

Taken together, these results strongly suggest that streams are very important for this species, and that their connectivity may provide critical movement and dispersal corridors. Also, the montane streams inhabited by big-headed turtles tend to be located in rugged terrain and, thus, are difficult to access. These types of habitats (with steep slopes) are not generally used for agricultural purposes and are at relatively low risk from large-scale habitat degradation. It is encouraging that this could mean that healthy populations may persist in remote areas that are difficult to access and with little collecting pressure or other human disturbance (e.g., Gong et al. 2006b).

Studies of terrestrial turtle species have revealed that translocated individuals often move much further distances than resident turtles, and such movements often result in animals moving outside of the recipient site (e.g., Tuberville et al. 2005; Rittenhouse et al. 2007; Hester et al. 2008; note that similar comparative data are lacking for semiaquatic and freshwater species). In contrast, the turtles in our study made relatively short movements (e.g., Figs. 2 and 4) and stayed near the release site. To our knowledge, this is the first published study of movement patterns and habitat use of translocated semiaquatic or aquatic turtles outside of Europe (e.g., Cadi and Miquet 2008) and the first published radiotelemetry study of turtles from mainland China. As such, there are no baseline data available on resident big-headed turtles or movements of translocated versus resident semiaquatic species with which to compare our data. Finally, our translocation experiment was not designed to establish a viable population but was a pilot study to help understand the behavior of big-headed turtles in the wild and to determine whether this species would be a good candidate for future projects.

There are potentially deleterious consequences of introducing turtles of unknown origin from the food trade back into the wild, especially in areas supporting resident turtles (e.g., introduction of disease, competition, territorial behavior, genetic pollution, etc.). Proposed translocation projects should follow established guidelines and protocols to increase the chances of successfully establishing or supplementing populations in biologically informed ways (e.g., International Union for the Conservation of Nature/Species Survival Commission Guidelines for Re-introductions, http://www.kew.org/conservation/RSGguidelines.html; Association of Zoos and Aquariums Guidelines for Reintroduction of Animals Born or Held in Captivity, http://www.aza.org/reintroduction/). Despite these caveats, our results are encouraging if translocation is to be considered as a long-term way to supplement or reestablish depleted big-headed turtle populations. Any such efforts will require careful planning, long-term monitoring, and management to ensure that 1) causes of the initial decline have been addressed, 2) the relocation site is protected, 3) enough animals are released to establish a viable population, 4) successful reproduction is occurring, and 5) survival of both offspring and translocated individuals is high enough to sustain the population indefinitely (e.g., Dodd and Seigel 1991). In Asia, this last goal may be the most difficult one because of the extraordinarily high harvest pressure on wild turtle populations.