The Chinese Box Turtle Cistoclemmys flavomarginata Gray 1863
by Michael J. Connor and Vickie Wheeler
Chinese box turtle, Cistoclemmys flavomarginata.
Photograph by Michael J. Connor.
This article discusses the natural history and the captive care, breeding and rearing of the Chinese box turtle, Cistoclemmys flavomarginata. This wonderful animal was formerly imported to the U.S. from Taiwan in large numbers for the bulk pet trade. Although direct exports from Taiwan have diminished considerably, quantities continue to enter the international pet trade via Hong Kong. In addition to collection for the pet trade, the species is under pressure from economic development which has spurred both habitat loss and harvesting for China's burgeoning food and medicinal products market. While it still survives in some numbers, like many of Asia's turtles the Chinese box turtle faces an uncertain future.
Cistoclemmys flavomarginata has been given a variety of common names including Chinese box turtle,
yellow-margined box turtle, golden-headed turtle and snake-eating turtle. Its generic status has been fluid. In his
original 1863 description, John Edward Gray assigned the species to Cistoclemmys. Later workers moved it to
Cyclemys in the early 20th century, and then on into the genus Cuora in the 1930's. More
recently, Roger Bour (1980) and Ren Hirayama (1983) presented extensive morphological evidence demonstrating that
flavomarginata and other turtles in the Cistoclemmys clade are generically distinct from the rest of
Cuora. Although criticized (McCord and Iverson, 1991), this hypothesis has never been formally disproved and the
revived name Cistoclemmys flavomarginata has slowly gained in acceptance over the years (e.g. Iverson, 1985;
Gaffney and Meylan, 1988).
The species occurs in southern China and various offshore islands. It has been divided into 3 poorly defined
subspecies (McCord and Iverson, 1991): the nominate taxon Cistoclemmys flavomarginata flavomarginata from
Taiwan; C. f. sinensis, a weakly distinguished taxon from southern mainland China; and C.
f. evelynae from the Ryukyu Islands that has also been proposed as a separate species (as Cuora evelynae).
All the animals discussed in this article are indistinguishable from C. f. flavomarginata as are
most specimens seen in the U.S.
Cistoclemmys flavomarginata is a small, highly domed turtle. In adults, the carapace and plastron are a
rich, dark brown except for the distinct creamy-yellow stripe down the vertebral keel, and the lightly pigmented lower
edges of the marginal scutes and adjacent plastral scutes that effectively form a creamy-yellow rim around the edge of
the plastron. In addition to the prominent vertebral keel, two broken lateral keels (one on each side) may be evident
along the pleural scutes. The carapace is joined to the plastron by ligaments and not the bony bridge found in most
other turtles. Because the plastron has a hinge between the pectoral and abdominal scutes the flexibility endowed by
the ligaments allows these turtles to close up very tightly. The skin of the limbs is brown. The top of the head is a
pale green marked on each side by a striking bright yellow line that goes from behind the eye to the back of the head.
The skin of the chin and neck is colored a delicate apricot to pink or yellow. There are five claws on the forefeet
and four on the rear. Sexing these turtles is not easy. Mature males have broader and larger tails than do females;
the tails may be so broad in males that they appear to be triangular in shape.
In his Turtles of Taiwan, Mao gives a secondhand account of one locality in southern Taiwan as: "a forested
hill, on the side of which was a shaded stream with extremely luxurious high-stemmed vegetation. Under the vegetation
of the damp bank, many turtles of this kind were concealed...about a hundred specimens were collected." He also cites
Gray's claim that C. flavomarginata are frequently seen in ponds adjacent to rice fields.
The island of Taiwan is dominated by a central mountain chain that rises to 4450 meters. Lowland (coastal) Taiwan
has a sub-tropical climate with cool, wet winters and hot wet summers. Higher elevations may stay cold year round. We
obtained climate data for 3 known C. flavomarginata localities on Taiwan from the National Weather
Bureau (ROC). I-Lan, in northeast Taiwan, has average minimum and maximum temperatures for January of 13.0° C (55.4°
F) and 19.2° C (66.6° F) respectively; and for July of 24.7° C (76.5° F) and 32.1° C (89.8° F). Annual rainfall is
2726 mm (107.3 inches). Hengchun, in the extreme south, averages minimum and maximum temperatures of 17.6° C (63.7° F)
and 24.2° C (75.6° F) in January, and 25.1° C (77.2° F) and 31.4° C (88.5° F) in July. Annual rainfall averages 2159
mm (85.0 inches). Taichung, on the drier west side of the island, averages minimum and maximum temperatures of 11.7° C
(53.1° F) and 21.9° C (71.4° F) in January, and 24.4° C (75.9° F) and 32.8° C (91.0° F) in July. Annual rainfall
averages 1710 mm (67.3 inches).
Southern China has a sub-tropical to temperate climate. Of the few mainland voucher specimens, the most northerly
is from Ganyu (Jiangsu), China. How much further north the range extends is unknown, but Ganyu has hot, humid summers
and cold, harsh winters. We obtained climate data for Ganyu from the U.S. National Climatic Data Center. Average
minimum and maximum temperatures for January are -4.58° C (24° F) and 4.18° C (39.5° F) respectively; and for July are
23.4° C (74.0° F) and 29.9° C (84.4° F). Annual rainfall is 909 mm (35.8 inches).
VW keeps her C. flavomarginata in a mixed group with other species of box and water turtle at liberty in her
specially modified back yard. The yard is divided into three main areas. A sunny, open area planted with shrubs, rocks
and driftwood has an above ground pond that the turtles can access by climbing up plastic rain gutters. Adjacent to
this area is a 26-foot by 13-foot wind-protected concrete patio that is used for basking on cooler days and where a
3-foot by 3-foot roofed turtle house is located that is heated to 24° C (75° F) in winter with a
thermostatically-controlled, dog kennel heating pad. A second, similarly heated turtle house is located in an adjacent
14-foot by 18-foot section that is roofed over with clear fiberglass and enclosed with lattice. This shady area is
extensively planted and has a waterfall with a catch basin, many water dishes and misters that are used on hot, dry
days. This is the preferred haunt of her C. flavomarginata. Off this section through a hole cut in the lattice is a drier 6-foot by 9-foot area surrounded by a 3-foot high brick wall. This is also well stocked with plants and grasses, and is the second preferred area for her C. flavomarginata.
For the first three years the C. flavomarginata were moved indoors each winter. Now, they are left out year-round, unless there is a very heavy winter storm when they are moved indoors because their favorite brumation site tends to flood. Occasionally a C. flavomarginata will be found in one of the heated houses in winter. Normally, however, they spend most of the winter dug down in the soil. They have bred successfully both when allowed to hibernate and when kept warm during the winter.
MJC houses his C. flavomarginata in 2 separate groups. A breeding pair is housed in a 9-foot by 13-foot enclosure that they share with breeding pairs of Emys orbicularis and Terrapene carolina carolina. They have lived in this enclosure since 1988 and have passed through every winter (including the recent El Niño bruiser) burrowed down into compost. A densely foliated lemon tree permanently shades half the enclosure. Water is always available from a small (3-foot by 4-foot) pond. The enclosure includes compost piles (mainly decomposing leaves and lawn clippings), a large tuft of deergrass (Muhlenbergia rigens) and rotting logs that are used as refuges. The other 1.2 young adults are housed in a 6-foot by 18-foot pen that they share with 2.2 Terrapene carolina major and an immature pair of Pseudemys concinna gorzugi. A banana plant and small shrubs provide shade all day, and the entire enclosure is shaded from the afternoon sun by an adjacent building. It is furnished in similar manner to that housing the breeding pair. The C. flavomarginata have been housed in this pen for 2 to 6 years. Both pens are watered daily from a sprinkler during the summer.
C. flavomarginata are highly omnivorous. VW's adults favor earthworms, frozen pinkies (defrosted), snails,
slugs, and mealworms. They also eat dry trout chow and moistened Whiskas dry cat food, canned cat food; fruits
including strawberries, bananas, cantaloupe, and papaya; and vegetables including grated carrots, corn on the cob, and squash. Leafy greens are ignored. Cuttlebone and parrot mineral blocks provide a source of calcium. Mineral blocks are preferred and eaten by females and young regularly. MJC offers as much Purina trout chow as they will consume 3 times a week, dry cat food once a month, and seasonal fruits from his garden (loquot, apple, fig, nectarine, and strawberry
guava) daily. Grapevine leaves and other sweet, green leafy vegetables and weeds are offered once a week. The
carefully tended compost piles in the pens are a ready source of invertebrates that the turtles hunt for, with June
bug larvae and slugs being principal prey. Occasionally they enter the pond and consume food (trout chow) meant for
the pond turtles but usually they eat on land.
Courtship in C. flavomarginata is both elaborate and elegant, and has been a constant source of interest and
amusement for both authors. Although it may involve some chasing and biting, by turtle standards courtship is
generally a gentle affair that requires the cooperation of both the male and the female. The most striking component
is the goose-like extension and swaying of the neck by the male. The male stretches out his neck towards the female's
head, placing his chin just above and parallel with the ground. He undulates his neck and in doing so may rub the top of his head on the chin of the female. This may be repeated several times. Sometimes the animals will then tussle:
first the male then the female will attempt to pursue and catch and overturn the other. The male may drool and
vocalize with a lip-smacking sound or a hiss. At some point this playful interaction becomes serious. The male will
start to bite the front edge of the female's carapace and shake her, or gently nip at her forelimbs. Once she is
stationary and has closed into her shell at the front he will move to her rear and mount.
Courtship occurs in the late spring, summer and fall. Neither author has witnessed evidence of courtship or copulation in the water. VW has observed that her adult males are aggressive towards other C. flavomarginata adult and juvenile males. She has seen neither interspecific aggression nor courtship, even though her animals are kept with many other species.
All of VW's females have nested during daylight and the nests have all been made at the same location. This site
receives only morning sun, and the soil is moist and fairly easy to dig. They may dig several holes before finally
nesting. The nests are about 10 cm (4 inches) deep. Nesting has occurred from March through August. The largest female
(867 g, 1.9 lb) lays multiple clutches of 2-3 eggs; in 1996 and 1997 she nested as follows:
Number of eggs
VW's smaller females may lay a single egg per clutch but most still lay two to three clutches per year at 1 to 2 month intervals during the nesting season.
Nine out of ten nests detected by MJC were constructed in friable soil in heavy shade at the base of the lemon tree in the enclosure. The tenth nest was located 2-feet away under a rotting Mulberry log. Although he has observed nesting directly on only 2 occasions (completed at 6:15 p.m. and 7:20 p.m.) the relatively shallow nests (dug to a maximum depth of 10 cm) are easy to locate by clearing away the leaves and other debris from the nesting area and gently probing the soil with fingers. In two nests, one of the eggs was at ground level. In the 7 nests detected
before 1997, the female had deposited one or two eggs per nest. In 1993 she nested, laid a single egg, and then two
days later nested again producing a second egg. Only the second egg was fertile. In 1994, 1995 and 1996 only single
clutches of 2 eggs were found. In 1997 she laid a single clutch of 3 eggs. So far this year (1998) she has laid 2
clutches of 3 eggs spaced 5 weeks apart.
Our combined experience has been that as a female grows, both the number of clutches and the number of eggs per clutch increases. The maximum clutch size of 3 eggs has been observed 7 times in 18 clutches (39%). Because the eggs are large, clutch mass forms a significant proportion of the female's total mass. Thus, the most recent clutch of 3 eggs (54 g) from MJC's 773 g female amounted to 7% of her body weight. Although this female laid two consecutive one-egg clutches two days apart in 1993, an internesting period of 3 to 5 weeks is the norm.
The hard-shelled eggs are cylindrical and relatively large: 38 to 52 mm (1.5 to 2 inches) long by 13 to 25 mm (0.5
to 1.0 inches) wide, and weighing 11 to 18.5 g. Mean values for 8 fertile eggs were 50 ± 2
mm long, 22.5 ± 5 mm wide and 18.0 ± 0.5 g in weight. Fertile
eggs usually develop a chalky, white band around the middle of the egg within a few days; rarely the whole egg will
chalk. For artificial incubation, VW buries the eggs in moist vermiculite (equal parts vermiculite and water by
weight) in a Tupperwareä container fitted with a lid punctured with a few holes for
ventilation and incubates them at 28.3° C (83° F). At this temperature, incubation takes from 68 to 85 days. For
artificial incubation, MJC buries the eggs in moist vermiculite (2 parts vermiculite to 1 part water by weight) in a
plastic shoe box with a loosely fitting lid placed in an incubator set at 28° C; but he usually leaves the eggs
undisturbed to develop in the nest. Hatchlings have emerged from eggs left to incubate in the ground after 75 and 90
days for two separate clutches with known laying dates.
Hatchlings and Hatchling Care
The beautiful hatchlings are considerably larger than Terrapene babies with carapace lengths of 31 mm (1.25
inches) to 44 mm (1.75 inches) and weights of 8 to 13 grams. They pip and emerge through one end of the cylindrical
egg, and are extremely active and alert. No intervention is usually required during hatching. In 1997, MJC found a
very agitated, newly emerged hatchling running around the enclosure with a constricting band of eggshell firmly
wrapped around its middle. After the hatchling was soaked in tepid water for 10 minutes the eggshell was cracked and
gently removed with tweezers. The hatchling thrived, and like most hatchlings began feeding within 5 days. The
hatchlings look like smaller and flatter versions of the adults but with relatively much longer tails. Overall
coloration is similar to that of the adults, but in many hatchlings and juveniles the creamy-yellow pigmentation on
the marginal scutes and edge of the plastron has a distinctly pink caste.
After the yolk sac has been absorbed, VW moves her hatchlings to a ten-gallon aquarium containing water to a depth
of about half an inch. The tank is furnished with a rock for them to climb out on to, lots of plastic plants for
cover, and a mineral block. Bloodworms are offered in the water to get them started eating. Canned cat food mixed with
moistened Whiskas dry cat food is placed on the rock. An under-tank heating pad and a spotlight keep them at 28° C
(82° F). The water is changed daily and the depth is increased as the hatchlings get stronger. By the time hatchlings
are six month old they are swimming in six inches of water but always have a place to get out of the water. This
method has produced strong, healthy, smooth-shelled juveniles. At six months, the young turtles are moved to a more
terrestrial environment in a screened cage outdoors. At this stage they have a carapace length of 60 mm (2.5 inches)
and weigh 80 to 90 grams. Growth slows after the turtles are moved outdoors.
MJC raises his hatchlings in individual plastic shoeboxes containing a half-inch of water. Half the area is filled
with sphagnum moss. This serves as both an "island" for the turtle to leave the water and a refuge that it can hide
under. The shoeboxes are placed inside 2-foot by 2-foot terraria and illuminated with Vitalite™ fluorescent tubes on a
12-hour light 12-hour dark cycle controlled by timers. The room housing the hatchlings is kept at a minimum daytime
temperature of 25° C (77° F). Initially, small earthworms are dropped into the water in front of the hatchling,
followed by a few pieces of trout chow. Once the hatchlings are eating it, trout chow is offered daily as the dietary
staple. Reluctant feeders are encouraged to eat by offering a small earthworm held in forceps; the worm is gently
whipped against the side of the hatchling's head until the now irritated hatchling snaps at and takes the worm. After
a couple of such feedings the hatchlings readily take earthworms and eventually anything that is offered from the
forceps. The water is changed daily, but maintained at a half-inch depth until the young turtles are moved outdoors
the following spring to a small pen that houses young box turtles. The first few C. flavomarginata hatchlings
were raised directly in the 2-foot by 2-foot terraria on a potting soil substrate (Supersoilä)
with cover provided by piles of sphagnum moss and tree bark. A shallow water dish was provided for drinking and
soaking, and the terrarium was misted every morning and evening. To ensure that all the hatchlings ate, they had to be
located and moved to the water dish for feeding. The shoebox method proved to be superior being less labor intensive,
more hygienic, avoids the risk of dehydration, allows numbers of hatchlings to be raised separately and provides an
environment in which they seemingly thrive.
We collected size data (weight, carapace length, plastron length, height and width) on all animals currently
maintained by the authors and an additional 4 from a third collection. These included 19 (5.9.5) smooth-shelled, normal-appearing specimens and 7 (3.4) deformed animals that showed grossly aberrant shell growth but were otherwise healthy. Weights were measured with Ohaus triple beam balances. Straight line carapace length, maximum shell height, shell height at the level of the hinge, shell width measured at the level of the hinge, lengths of the plastral fore-lobe, hind-lobe and total plastron length along the seam, were measured with calipers. The data was analyzed using Excel and Systat programs.
Figures 1, 2 and 3 are graphical plots of the relationships between carapace length and maximum height, width and weight respectively for the 19 animals showing normal growth. We have included these charts both for their informational value and to demonstrate how easily potentially valuable data can be collected using a few simple measurements. All the correlations were statistically highly significant (p<0.001). The lines drawn on the charts are
best fits calculated automatically using the least squares method. The equations for these lines can be used to predict expected height, width or weight from the carapace length. Shell deformities that are rarely or never seen in wild specimens are a common occurrence in captive raised C. flavomarginata, just as they are in other more widely maintained and bred species. Shell deformities vary considerably, but flat, abnormally-wide shell growth is typical. The equations relating carapace length to height and width could be useful guides in predicting if a young Chinese box turtle is growing normally. The relationship between carapace length and weight can be used to determine if a particular specimen is underweight for its size.
Figure 1: Carapace length vs. maximum height
There was an excellent correlation (R=0.997) between carapace length and shell height. The equation for the line is:
y = 0.5857x - 9.9742 where y is the maximum shell height in mm and x is the carapace length in mm.
Figure 2: Carapace length vs. shell width
There was an excellent correlation (R=0.992) between carapace height and shell width. The equation for the line is:
y = 0.6652x + 6.5117 where y is the total shell width in mm measured at the level of the hinge and x is the carapace
length in mm.
Figure 3: Carapace length vs. weight
There was an excellent correlation (R=0.983) between carapace length and (ln) weight. The equation for the line is:
y = 6.015e0.0324x where y is the weight in g and x is the carapace length in mm.
Since the carapace is usually longer than the plastron the ratio of carapace to plastron length is slightly higher
than 1. The ratios were relatively constant as can be judged from the small size of the standard deviations compared
to the respective mean values. In contrast, the deformed turtles tended to have longer plastrons than carapaces with
ratios of carapace to plastron lengths of less than 1. The ratios of the lengths of the plastral fore-lobe to the
hind-lobe were higher in juveniles than in adult males and females.
1.005 ± 0.007
0.881 ± 0.073
1.013 ± 0.014
0.720 ± 0.025
1.020 ± 0.015
0.742 ± 0.032
0.996 ± 0.093
0.769 ± 0.078
The only health problems have been in newly acquired animals that have suffered from poor husbandry or were recent imports. The owner of MJC's original pair claimed that they would not eat, but in reality the animals were starving and ate anything that they were offered. One of the young adult given to him 2 years ago is a grossly deformed female, suffering from "sombrero syndrome" with the flattened to sway-backed, wide shell and poor motor coordination typical of box turtles raised indoors on a diet of mealworms and supermarket salad. After 2 years of outdoor living and a more nutrient-rich diet she is now doing remarkably well and is able to scurry around her enclosure. A young male and female acquired separately 3 and 5 years ago had both been kept for several years in grade school classrooms in small glass tanks on Astroturf ("It was easy to clean."). They had both lost claws (and one of them a few toes), were dehydrated and malnourished. The rotting limbs responded to daily application of Silvadene ointment, and a healthier diet and outdoor housing took care of the rest.
VW has had problems with newly imported C. flavomarginata. Like most pet trade animals, they may require
extensive and expensive medical treatment when first acquired. The worst health problems were with the first pair: they were heavily parasitized and suffered a variety of infections. They were treated with antiprotozoals, vermicides, and six different antibiotics and had to be tube fed for almost a year. She now routinely deparasitizes any Asian box turtle that comes into her care with Flagyl and Oxfendazole.
Given the correct care and environmental conditions Cistoclemmys flavomarginata can thrive in captivity, and
the authors have successfully maintained reproducing groups of C. flavomarginata for over 10 years. They are
fascinating turtles to watch and interact with. Although some may take several years to get over their shyness they
eventually learn to tolerate human presence. Others quickly learn the association between humans and food, and will
greet their keeper with greedy interest.
We attribute our successes with this species to our attempts to provide habitats that mirror the natural
environment as much as possible. We both feel strongly that these animals should be housed outdoors, in roomy and
enriched environments where they have some ability to behave as they would in the wild whether that be by hiding,
basking, courting, swimming, hunting, or just feeling the wind on the backs of their shells.
Bour, R. 1980. Position systématique de Geoclemys paleannamitica Bourret, 1941
(Reptilia-Testudines-Emydidae). Amphibia Reptilia 1(2):149-159.
Gaffney, E. S. and Meylan, P. A. 1988. A phylogeny of turtles. In: The phylogeny and classification of the tetrapods.
Vol. 1., pp. 157-219, Claredon Press, Oxford, UK.
Hirayama, R. 1984. Cladistic analysis of batagurine turtles (Batagurinae: Emydidae: Testudinoidea); a preliminary result. Studia Geologica Salmanticensia, Vol. 1, (Studia Palaeocheloniologica 1): 141-157.
Iverson, J. B. 1985. Checklist of the turtles of the world with English common names. SSAR Herp. Circ. 14:1-14.
Mao, S. H. 1971. Turtles of Taiwan. Commercial Press, Taipei.
McCord, W. P. and Iverson, J. B. 1991. A new box turtle of the genus Cuora (Testudines: Emydidae) with taxonomic notes and a key to the species. Herpetologica, 47(4): 407-420.
Originally published in the Tortuga Gazette 34(10): 1-7, October 1998