The Arakan forest turtle, Heosemys depressa (Anderson 1875), is a critically endangered (IUCN 2009), poorly known chelonian endemic to western Myanmar (Ernst and Barbour 1989; van Dijk 1993). From 1875, when the species was first described, to 1908, only 5 H. depressa were reported in the literature, all from Rakhine (formerly known as Arakan) State in western Myanmar (Iverson and McCord 1997). These specimens presumably originated from the Rakhine Hills ( =  Arakan Yomas), although specific locality data are lacking (Iverson and McCord 1997). More recently, we collected 21 H. depressa shells and a single living animal from villagers in central and southern Rakhine State (Platt et al. 2003a, 2003b). Additional living H. depressa and shells, all of unknown provenance, have been obtained from wildlife markets in southern China and Taiwan where the species is regularly encountered (Iverson and McCord 1997; Chen et al. 2009; G. Kuchling, pers. comm.). However, biologists have yet to observe H. depressa in the wild, and almost nothing is known regarding its ecology or population status (van Dijk 1997; Platt et al. 2003a, 2003b). Consequently, ecological studies and population surveys have been accorded high priority (van Dijk 1997; Platt et al. 2000). Here we report field observations of H. depressa in the Rakhine Yoma Elephant Range (RYER) of western Myanmar and make conservation recommendations. Because RYER is the only protected area within the currently recognized geographic distribution of H. depressa, Platt et al. (2003a) deemed the wildlife sanctuary a globally important site for the conservation of this critically endangered endemic species.

Study Area

RYER (lat 18°00′N, long 94°40′E) was established by the Myanmar Ministry of Forestry for the protection of Asian elephants (Elephas maximus). RYER encompasses 175,644 ha of the Rakhine Hills in Gwa and Thandwee Townships of southern Rakhine State (Fig. 1). Rakhine State is one of the most sparsely inhabited regions of southeast Asia (Salter 1983). The Rakhine Hills extend for 500 km along the western coast of Myanmar and represent the southern-most extension of the Himalayas (Salter 1983). These hills consist of high (to 1200 m in RYER), steep parallel ridges separated by streams that flow within deep, restricted valleys (Stamp 1930; Henderson et al. 1971). A narrow alluvial belt known as the Rakhine Coastal Strip (Roberts et al. 1968) abuts the western boundary of RYER. The climate of southern Rakhine State is influenced by the southwest monsoon and exhibits 3 distinct seasons: a wet season extending from early June to late October, a cool dry season from early November to mid-February, and a hot dry season from late February through late May (Henderson et al. 1971). Mean annual precipitation ranges from 4500 to 5300 mm (Roberts et al. 1968).

View Figure

• Download as Powerpoint
• Download Figure

Little specific information is available about the vegetation of RYER. In general, the Rakhine Hills are characterized by evergreen forests of poorly known floristic composition (Salter 1983). Extensive tracts of bamboo (Melocanna bambusoides) occur throughout the region, and the Ministry of Forestry estimates that approximately 75% of RYER is covered in dense stands of bamboo (Win Ko Ko 2003). Bamboo stands are considered an anthropogenic climax formation, developing in response to historic disturbances, such as shifting cultivation, fire, or a combination of both (Terra 1944; Soderstrom and Calderón 1979). Although dense bamboo often inhibits tree regeneration (Stamp 1924), large trees are scattered throughout bamboo stands in RYER. Mast fruiting of M. bambusoides in southern Rakhine State began during 2007, and, as typical for monocarpic bamboos (Janzen 1976), widespread culm (above-ground portion of bamboo plant) die-back was underway by 2008 (Fig. 2). An abundance of M. bambusoides seedlings and germinating seeds were noted beneath stands of dead culms in May–June 2009 (Platt and Khin Myo Myo 2009).

View Figure

• Download as Powerpoint
• Download Figure

Methods

We conducted field surveys in RYER and surrounding areas of southern Rakhine State from 21 May to 5 June 2003 and from 26 May to 8 June 2009. Surveys in both years coincided with the beginning of the annual wet season; intermittent rainfall, at times heavy, occurred daily in late May, followed by sustained torrential downpours in early June. During these surveys, we conducted open-ended interviews (sensu Martin 1995) with hunters, farmers, and other knowledgeable individuals regarding the natural history and local occurrence of turtles, hunting methods, and levels of exploitation. Such individuals are generally an excellent source of information on the local chelonian fauna (Thirakhupt and van Dijk 1994; Platt et al. 2004). In accordance with the format of an open-ended interview, we asked each informant a series of questions that included standard questions prepared in advance and others that arose during the course of conversation (Martin 1995). We searched for H. depressa by using teams of 10 to 12 people accompanied by 1 or 2 local hunting dogs. Dogs are widely used by turtle hunters in Myanmar, and our previous work indicated that dogs are much more efficient at locating concealed turtles than humans (Platt et al. 2003c). We searched for turtles during morning to early afternoon (approximately 0800 to 1300 hours) and late afternoon (approximately 1500 to 1900 hours), but adhering to this schedule often proved difficult because of inclement weather. Although local hunters maintained that H. depressa is most active after dark (see below), we elected not to conduct nocturnal searches because of the dense vegetation, our unfamiliarity with the extremely steep terrain, and the likelihood of becoming disoriented after nightfall. Captured turtles were measured (straight-line carapace length; CL) with tree calipers, photographed, and released at the point of capture. The geographic coordinates and elevation of each capture location were determined with a Garmin GPS 12. The sex of adult H. depressa was determined based on tail and plastral morphology; males are distinguished from females by their larger tails and concave plastrons (Iverson and McCord 1997). Juveniles could not be reliably sexed. If present, external parasites and feces were collected; the latter were macroscopically examined in the field, and dietary items were identified to the lowest possible taxonomic category.

Results and Discussion

We found 5 adult (1 male, 4 females) and 2 juvenile H. depressa during field surveys of RYER in 2003 and 2009 (Table 1). The two found in 2003 were located with the aid of dogs, but we found all turtles in 2009 without canine assistance. We encountered H. depressa during both years only in M. bambusoides thickets at elevations that ranged from 139 to 251 m above sea level (Fig. 3). Four turtles were found among bamboo with a dense layer of leaf litter; the other turtles occurred in bamboo thickets where the surface litter had been removed by dry season ground fires in March and April, probably rendering these turtles more visible to searchers. Two turtles were found climbing steep (approximately 60°) slopes and the remainder on flat terrain along ridges. Villagers inhabiting agricultural lands adjacent to RYER stated that H. depressa is also regularly encountered in second-growth forest typical of fallow fields. Hunters in central Rakhine State reported finding H. depressa in a variety of habitats, including bamboo brakes, evergreen forest, mesic ravines, open-canopy deciduous forest, second-growth forest, and even a sugarcane field (Platt et al. 2003b). Collectively, these observations suggest that H. depressa is not restricted to bamboo brakes but instead occurs in a variety of habitats. The apparent restriction of H. depressa to bamboo brakes in RYER is most likely because of the widespread occurrence of this vegetation type in the sanctuary. Similar to our earlier fieldwork (Platt et al. 2003b), we found nothing to suggest that H. depressa excavates burrows as suggested by Ernst and Barbour (1989).

View Figure

• Download as Powerpoint
• Download Figure

Table 1 Carapace length (CL), sex, and locality data for 7 Heosemys depressa found in the Rakhine Yoma Elephant Range of western Myanmar.^a

View Fullscreen

Hunters living near RYER contend that H. depressa is most active early in the wet season and can be difficult to find at other times of the year. Likewise, hunters in central Rakhine State considered May through July as the optimal time to search for H. depressa; at other seasons, turtles are so difficult to find that hunting was not deemed worthwhile (Platt et al. 2003b). Reports of dry season activity by H. depressa are generally associated with fire avoidance behavior. Anthropogenic dry season wildfires are common in the region, and villagers occasionally find H. depressa moving ahead of advancing flames. In March 2009, a forest ranger found 3 H. depressa crawling down a dry streambed to escape a brush fire in village lands adjacent to RYER. In central Rakhine State, fires are deliberately set to flush H. depressa and the elongate tortoise, Indotestudo elongata, from dense cover and toward waiting hunters (Platt et al. 2003b). We obtained feces from 4 (3 adults and 1 juvenile) H. depressa captured in 2009. At least 4 different dietary items were recovered from the feces of adults: seeds and intact fruits of Dillenia pentagyna (1 turtle), panicles of Melocanna bambusoides (2 turtles), unidentified vegetation (2 turtles), and unidentified fruits (1 turtle). Feces recovered from a juvenile contained only partially digested vegetation that could not be identified. Two adults were found beneath Dillenia pentagyna feeding on fallen fruits, although feces were not obtained from these animals. Hunters also reported that H. depressa consumes the soft endosperm of immature M. bambusoides seeds, and fruits of Erythrina suberosa, Ficus hispida, and Microcos paniculata. Turtles obtain fruit by foraging beneath large emergent fruiting trees scattered among the bamboo. Indeed, searching beneath fruiting trees, particularly D. pentagyna, is said by hunters to be the most reliable method for finding H. depressa. Hunters regarded late afternoon until 2100–2200 hours as the optimal period to find foraging turtles beneath fruiting trees, which suggests that H. depressa is at least partially nocturnal. We encountered H. depressa in mid to late afternoon (1504 to 1625 hours); no turtles were found during our morning searches.

The vernacular name for H. depressa in southern Rakhine is pyant chee zar leik, which literally translates to “turtle that eats rhinoceros feces” (pyant is a Rakhinese corruption of the Burmese word kyant, for rhinoceros; Maung Maung Kyi, pers. comm.). Sumatran rhinoceros (Dicerorhinus sumatrensis) formerly occurred in the Rakhine Hills (Christison 1945; Ansell 1947), and the vernacular name suggests that H. depressa consumed rhinoceros feces, partially digested plant material or fruit embedded within the feces, or insects inhabiting the dung pile (e.g., Guggisberg 1966). Because Sumatran rhinoceros were extirpated from the region about 50 years ago (Platt and Khin Myo Myo 2009), rhinoceros feces are no longer available as a food source for H. depressa. It remains unknown if the feces of other large herbivores that are common in the area (e.g., Elephas maximus or Bos frontalis) are consumed by H. depressa. Coprophagy has occasionally been noted among other terrestrial and aquatic turtles (Mares 1971; Goodman and Stewart 1998; Platt et al. 2009), which probably derive nutritional benefits from partially digested food items in the feces (Giovannetti 1982; Robbins 2001). To our knowledge, consumption of rhinoceros feces by other chelonians has not been reported; however, Melanochelys trijuga is known to deposit clutches in dung piles of the greater one-horned rhinoceros (Rhinoceros unicornis) (Dinerstein et al. 1987).

Virtually nothing is known about the reproduction of H. depressa in the wild. A hunter living in a village peripheral to RYER stated that he occasionally found “two or three” shelled eggs when skinning female H. depressa in June and July. Earlier we reported that a hunter in central Rakhine State captured a female H. depressa in February that contained 3 enlarged but unshelled follicles (Platt et al. 2003b). We found 2 small juvenile H. depressa 7 m apart in leaf litter on 3 June 2009. Umbilical scars were obvious on both turtles, and the carapace and plastron were pliable and not yet heavily ossified. These 2 juveniles (CL  =  73 and 75 mm) were slightly larger than 5 hatchlings measured shortly after emerging from the egg (CL  =  60.6 to 66.0 mm) at Zoo Atlanta but considerably smaller than a 2-year old turtle (CL  =  118 mm) (D. Lawson, Zoo Atlanta, unpubl. data). Based on body size and the absence of growth annuli, we speculate that these 2 juveniles probably emerged from the nest 2–3 months earlier (March–April 2009). Given an incubation period that ranges from 110 to 160 days (D. Lawson and B. Lock, Zoo Atlanta, unpubl. data), eggs are probably deposited late in the wet season (September to November), a time frame consistent with local reports of shelled oviductal eggs being found in June and July.

We collected 2 ticks: (Amblyomma geoemydae nymph and adult female Amblyomma supinoi) from 2 adult H. depressa in 2009. Our collection of A. geoemydae from H. depressa is a host record and constitutes the second report of an ectoparasite from H. depressa (Robbins and Platt, submitted). Amblyomma supinoi is considered one of the rarest ticks in southeast Asia, and our collection compliments an earlier report of this tick from H. depressa in central Rakhine State (Robbins and Platt 2001).

We found nothing to indicate that widespread commercial harvesting of H. depressa is underway in RYER. Although villagers frequently consume H. depressa meat and occasionally sell living turtles or plastrons to village-level wildlife traders, this harvest appears to be largely opportunistic; turtles are collected by villagers while gathering forest products, hunting other wildlife, or clearing fields. Some hunters occasionally seek H. depressa, but market demand is not particularly high, and prices paid by wildlife traders are low in comparison with more valuable species, such as Indotestudo elongata and Amyda cartilaginea (Win Ko Ko 2003; Platt and Khin Myo Myo 2009). Most of the turtles harvested appear to originate from village lands adjacent to RYER rather than from within the sanctuary. The interior of RYER is largely inaccessible because of the rugged terrain and lack of roads and trails. This inaccessibility, coupled with the negligible amount of money received from wildlife traders, makes hunting H. depressa an economically dubious endeavor.

Importantly, wildlife poaching in general appears minimal within RYER. We noted an abundance of large mammals within the sanctuary (Platt and Khin Myo Myo 2009), a component of the biota that is usually the first to disappear when unregulated wildlife harvesting is occurring. Furthermore, I. elongata remains common in RYER, and large adults were encountered frequently as we searched for H. depressa (Win Ko Ko 2003; Platt and Khin Myo Myo 2009). Indotestudo elongata is intensively harvested throughout Myanmar for both subsistence and commercial reasons, and commands a relatively high price in the illicit wildlife trade (Platt et al. 2007). Consequently, populations have been severely reduced or extirpated in many regions of the country (Platt et al. 2000, 2003c, 2007). Our frequent encounters with I. elongata in RYER would not be expected if turtle populations were being exploited within the sanctuary.

As a result of our fieldwork, we concluded that RYER offers excellent prospects for the future conservation of H. depressa and I. elongata for several reasons. RYER is a large, intact tract of undisturbed habitat that harbors unexploited populations of both chelonian species. Most importantly, human activity within RYER is minimal; the interior is extremely difficult to access, there are no temporary or permanent villages within RYER, and few people seem to enter the sanctuary. Indeed, villagers have little reason to enter RYER because natural resources, such as wild game, timber, fuel wood, bamboo, and other forest products, are present on village lands adjacent to RYER where they can be legally harvested. To our knowledge, RYER remains the only protected area within Myanmar where H. depressa is known to occur. Although it is possible that H. depressa ranges northward into the Chin Hills and protected areas within that region may harbor populations, this has yet to be verified (Platt et al. 2003b).

Although our fieldwork verified the occurrence of H. depressa in RYER and provided limited natural history data, much remains to be learned about the ecology of this enigmatic species. Because conservation measures are most effective when based on a thorough understanding of the natural history of a species (Dayton 2003), we recommend conducting multiyear field studies of H. depressa in RYER. Implementing effective in situ conservation measures for H. depressa at RYER will also require training for RYER staff, improvements in infrastructure (e.g., guard stations), and conservation assessments of the northern and eastern regions of the sanctuary (Platt and Khin Myo Myo 2009). Finally, we recommend coupling in situ conservation efforts at RYER with ex situ propagation of H. depressa at facilities in Myanmar and elsewhere. However, it is imperative that assurance colonies of H. depressa be founded by using turtles confiscated from wildlife traders or obtained in markets rather than by removing additional turtles from the wild.