Until 1982, the geoemydid forest cane turtle, Vijayachelys silvatica, was known to science from only 2 specimens, collected in 1911, originally described as Geoemyda silvatica by Henderson 1912 and later transferred to the genus Heosemys. These specimens were obtained in dense forest near Kavalai, at about 450-m elevation, in the former Cochin State Forests, about 30 km east of Chalakudi in the Trichur District of the present Kerala State, southwest India. In 1982, Vijaya rediscovered this cryptic species (Vijaya 1982a, 1982b, 1983c). Subsequently, an expedition by the Madras Crocodile Bank Trust located an additional 20 individuals (Moll et al. 1986).

View Figure

• Download as Powerpoint
• Download Figure

The species was reassigned from the genus Heosemys to Geoemyda based on comparative osteological and penile morphology (Moll et al. 1986), but Ernst and Barbour (1989) chose to retain Heosemys as the generic name of the species. McCord et al. (1995), to determine generic relationships of a new batagurid from northern Sulawesi, performed a cladistic analysis on members of Geoemyda and Heosemys by using data from Moll et al. (1986) for, among other species, G. silvatica. Their analysis suggested that the clade that contained G. silvatica was a sister group to the clade that contained Geoemyda spengleri and Geoemyda japonica, which was in accordance with Moll et al. (1986). Subsequently, McCord et al. (2000) suggested that G. silvatica be returned to Heosemys. Finally, Prashag et al. (2006) performed a phylogenetic analysis of mitochondrial DNA sequence data on the genus Geoemyda sensu lato, and found G. silvatica to be a basal and isolated taxon, unrelated to any living Heosemys or Geoemyda. Consequently, they proposed the generic name Vijayachelys, in honor of the late Madras Crocodile Bank Trust turtle researcher, J. Vijaya, to whom the rediscovery of this species is attributed.

The study site is situated in Chalakudi District, Kerala, along the Nadukani–Thatavarra road, and is a narrow strip of hillside 30–100-m wide and about 4-km long. The total area of the study site encompasses 13,900 m². The Muppili River transverses the study site. Annual rainfall in the area is around 2500 mm, and altitude is about 800 m. The forest floor is covered with a thick layer of leaf litter under which V. silvatica shelter during daylight hours. The forest at the study site is evergreen and semi-evergreen leafy vegetation up to 3 m in height. Plant families characteristic of the region include the Dipterocarpaceae, represented by Dipterocarpus indicus, and species of Hopea, Shorea, Valeria, Vateria, and Clusiaceae, which include giant trees such as Mesua ferrea and Calophyllum sp. Also represented is the family Myristaceae (Mani 1974a; Puri 1960; Subramanyam and Nayar 1974). Forests from this region are floristically related to forests in Sri Lanka, Assam, Myanmar, and Southeast Asia (Mani 1974b; Puri 1960). Vijaya (1982a, 1982b, 1982c; 1983; 1984) provided descriptions of the cane turtle's habitat; however, precise locality coordinates are not provided here on purpose.

Here, we describe the basic habitat where Vijayachelys were found, in addition to new knowledge on the biology of this species. Specifically, we present information on habitat utilization, analyze sexual-size dimorphism, reproductive biology, relative abundance, and information on algal extent on shells, in addition to notes on diet.

METHODS

Straight-line carapace length, straight-line carapace width, plastron length, shell height, and body mass were measured on all animals found. The sex of adults was readily assigned by examining the plastron, which is concave in males, and the relative length of the tail proximal to the cloacal opening, which “exceeds the portion distal to it in males [while] in females the proximal portion is shorter” (Moll et al. 1986). For each individual, the habitat where the turtle was captured, shortest straight-line distance to a stream at the study site, extent of carapacial algal growth to the nearest 10%, and flexibility of the plastral hinge (Waagen 1984) were recorded. Turtles were marked by notching a unique combination of marginal scutes (Cagle 1939). To determine an estimate of age structure, growth rings (Zug 1991; Wilson et al. 2003) were examined at various locations on the carapace (pleural and vertebral scutes) and plastron (pectoral, humeral, abdominal, and femoral scutes).

An enclosure (Fig. 1) was constructed in native habitat for the purpose of confining animals to study aspects of their natural history difficult to observe otherwise, namely nesting biology and social interactions. The cryptic habits of V. silvatica and the thick leaf litter prevented the location of natural nests. Females were examined in the egg-laying season, and the condition of eggs within the oviduct was gauged by palpation of the inguinal pockets (Ewert and Legler 1978). After ascertaining the presence of eggs, oxytocin was injected into the subcutaneous tissue of the hind limb with a sterile 24-gauge needle, at a dosage of 1–4 units per 100 g BM, as recommended by Ewert (1985). The nesting season is known to be approximately between October and December (J. Vijaya, pers. obs.). Information on the diet of V. silvatica was gleaned from analysis of fecal samples, direct observations of feeding in the wild, and response to items offered in captivity.

View Figure

• Download as Powerpoint
• Download Figure

Statistical analysis was accomplished with the SPSS 10.0 software package for Microsoft. Analysis of variance (ANOVA) was used to assess differences among males, females, and juveniles in morphometrics, distance to water, and algal growth on the carapace. Pearson's correlation coefficients were used to analyze the relationship between the number of growth rings and CL, among other morphometric features. A χ² test was used to test for departure of the sex ratio among adults from 1∶1. An ANCOVA was used to control for body size when comparing algael extent between different size classes.

RESULTS

Morphological Variation

We obtained morphological measurements on 47 males, 62 females, and 26 juveniles, for a total of 135 V. silvatica observed between 1983 and 1984. A frequency distribution of SCL is presented in Fig. 2. The smallest sexable male had a 95.0-mm SCL, and the smallest sexable female had a 85.5-mm SCL. Females were more massive than males and were also significantly larger in all the features examined, except for SCL (Table 1). The largest animal measured was a female of 170 mm SCL, the largest male was 134 mm SCL.

View Figure

• Download as Powerpoint
• Download Figure

Table 1 Morphometric analysis of adult Vijayachelys silvatica.^a

View Fullscreen

Age Structure

The number of growth rings was positively correlated with SCL (Fig. 4), SCW (r  =  0.69, F  =  115.02, N  =  126, p < 0.001), PL (r  =  0.63, F  =  82.13, N  =  126, p < 0.001), SH (r  =  0.73, F  =  140.74, N  =  125, p < 0.001), and BM (r  =  0.73, F  =  137.90, N  =  126, p < 0.001). Most individuals were estimated to be more than 3 years old, when assuming that only one growth ring was deposited annually. The sex ratio among adult individuals was 47 males and 36 females (1.31∶1), which did not differ significantly from 1∶1 (χ²  =  2.06, df  =  1, p  =  0.180).

View Figure

• Download as Powerpoint
• Download Figure

Relative Abundance

The greatest number of turtles was found in October and the fewest in May (Table 2). In general, the fewest turtles were observed in months when search days were minimal (December) and highest when a large proportion of the month was spent searching (October).

Table 2 Relative abundance of Vijayachelys silvatica between May 1983 and January 1984.

View Fullscreen

Reproductive Biology

Mating was observed on the morning of 21 September 1983 and lasted for 1.5 hours. On 11 June 1983, male 12R was found mating with female 8R, at around 2000 hours. On 18 June 1983, at 1820 hours, male 11R was found mounted on a female (8R) in a corner of the enclosure. On the same day, this male mated again with the female for 43 minutes between 2117 and 2200 hours, and another male (12R) mounted female 8R between 2301 and 2315 hours. Heavy rainfall occurred on 3 November 1983 for 2 hours and appeared to induce 3 pairs of Vijayachelys to mate; this could have been induced by observer bias. Estimated ages of nesting females based on growth ring counts were between 5 and 13 years.

The first attempt at hormonal induction of ovipostion occurred on 7 January 1984. The time of injection was 1408 hours. At 1540 hours, a single egg was expelled. The single egg measured 43.0 × 22.5 mm and weighed 14.5 g. This female laid an additional 2 eggs on 10 February. Another female was injected with 6 mL oxytocin at 1543 hours on 2 January 1984. At 1603 hours, an additional 4 mL oxytocin was injected. A single egg was passed at 1641 hours. This egg measured 48.0 × 22.5 mm and weighed 15.5 g.

In one instance, a female nested without hormonal induction on 17 January 1984. This female was examined the day before, when palpation revealed oviductal eggs; hence, ovipostion occurred some time in the night of 16–17 January. An inspection in the area of the exclosure where the V. silvatica congregated revealed the nest. Two eggs were covered lightly with soil deep within a crevice where the turtles were found to shelter themselves. The nest temperature was 23.0°C, and temperature of the air at the time of collection ranged between 23.8° and 24.2°C. One egg was completely covered with soil, whereas the second, its position slightly above the first and more within the nest cavity, was partially embedded in soil. Both eggs were covered with leaves. Other egg and clutch details are given in Table 3.

Table 3 Summary of egg and clutch characteristics in Vijayachelys silvatica.^a

View Fullscreen

Algal Growth

Most V. silvatica (99 of 101 examined individuals) had algal growth on the carapace. The amount of algae on the shell varied among size classes. Males (X  =  52.5%, N  =  36) had algae on more of the carapace than females (X  =  36.32%, N  =  47), who had algae on more of the carapace than juveniles (X  =  6.75%, N  =  16). Because there was no difference in male and female SCL, but other morphometrics were much larger in females, we had to control for body size. We used mass as the covariate, because it was the most independent variable with respect to area (analysis of covariance, F  =  1.52, N  =  99). The proportion of algae on an individual was not significantly correlated with distance from the stream (r  =  0.11, F  =  0.954, N  =  80, p  =  0.33).

Distance to Water

Females (N  =  51) averaged 41.8 m, males (N  =  41) averaged 28.8 m, and juveniles (N  =  19) averaged 28.6 m from the stream. The distance to the Mupilli stream did not vary significantly among males, females, and juveniles (ANOVA, F  =  2.25, N  =  111, p  =  0.110).

Diet

One V. silvatica was observed in a dry area with a fecal deposition in the near vicinity, which included fruit matter, perhaps from the tree Fiscus eliptica. In another instance, Jaganathan offered a caterpillar to the Vijayachelys group in the exclosure, which was ignored. A male was then offered an earthworm, which was readily accepted. Several more earthworms were offered, and the male began to gorge on them. A male captured in the wild (130 mm CL) was observed feeding on a millipede. Another male (113 mm CL) was observed ingesting mud lumps before capture. Fruiting species of trees on which Vijayachelys were observed feeding included jungle pear (Eleaocarpus oblongus) and jack fruit (Artocarpers integricollia).

DISCUSSION

SCL was not statistically different between both sexes, but females were generally larger and heavier. Moll et al. (1986) measured a small female (116 mm CL) that weighed 222 g compared with the largest male measured by them (125 mm CL) that weighed 159 g. CL and BM of the largest male and female measured by Appukuttan (1991) were 128 mm, and 133 mm CL, and around 240 and 300 g BM, respectively. The largest male and female from the current study measured 134 mm and 170 mm CL, and 275 and 280 g BM, respectively.

Growth rings were used to obtain estimates of the population's age structure, because age (the number of growth rings) was correlated with SCL (see Fig. 3) and other morphometrics recorded. Growth rings were well demonstrated in juvenile individuals, but in adults this distinction was blurred, resulting in an estimate that varied by as much as 2 rings. This hinders age estimates, and this difficulty was reported by Sexton (1959) and Gibbons (1983). An assumption in our estimates of age is that measures of turtle size are related to age and that growth rings are annual. This is a relationship that we were unable to examine in any detail given that the study period was just 2 years. Wilson et al. (2003), in a literature review that encompassed 6 families of turtles, noted that studies related the number of growth rings to either size (presumably CL) or age of turtles. We observed high Pearson correlation values in V. silvatica that ranged from 0.63 for PL to 0.73 for SH and mass, within the range of 0.54–0.98 reported by Wilson et al. (2003) for correlations between growth rings and size for 6 turtles. The current study falls under “PSM” or “past the size of sexual maturity” category in the study by Wilson et al. (2003), wherein turtles can be aged past the age of sexual maturity. However, longer-term studies with individuals of known age are required to clarify the growth rings and age relationship, particularly in adults.

The larger number of juveniles collected in October may relate to the number of days searched (almost the whole month). The low number of juveniles represented in the sample (26) might reflect their cryptic behavior and coloration, rather than a scarcity of this age class. However, when the total number of turtles found per man hour per month was considered, the highest ratio was in May. This agrees with observations by field workers working in areas where the species occurs, who confirmed that Vijayachelys are incredibly hard to find at other times of the year (K. Kumar, pers. comm.).

In the present study, mating was observed in the months of June, September, and November, and oviposition was in January and February. Appukuttan (1991) observed mating in a pair of V. silvatica in June. Moll et al. (1986) discovered a clutch of 2 eggs in December from a captive group of V. silvatica at the Madras Crocodile Bank. Clutch sizes of females in the current study varied between 1 and 3 eggs. Data on egg length and egg mass, here averaging 43.87 mm and 15.74 g, respectively (Table 3), were similar to the 44.5 mm and 15.25 g reported for a single clutch by Moll et al. (1986). The response to injection of oxytocin in V. silvatica is akin to that observed in species laying relatively large eggs, “wherein the turtle becomes involved, gasping and retracting its head and forelimbs as the egg is passed out of the cloaca” (Ewert 1985).

Juvenile turtles had less algae than adults, which is consistent with findings from temperate and tropical regions (e.g., Edgren et al. 1953; Bergen and Renshaw 2008). Adult V. silvatica may associate closely with water or damp areas, which would encourage algal growth. The reason for this behavior is unknown, and may be related to thermal biology. Interestingly, the sympatric Indotestudo travancorica had no algal growth. Damp areas were also observed by Vijaya to have an abundance of snails, centipedes, and beetles living in the molding leaf litter and humus. Also, mushrooms grew in these areas. K. S. Appukuttan (unpubl. data, 1991) observed V. silvatica, in captivity, feeding on bananas, apples, tomatoes, insects, earthworms, and snails.

Although this article introduces new data on Vijayachelys, there is clearly more work to do on the biology of this species. Data sets from other localities where the species occurs would be useful for comparison with this particular population, especially with regard to long-term studies of known-age animals. This is urgent because the species is threatened by several dam projects that would flood significant portions of its habitat.