Methods

The study described here was conducted between May 2004 and April 2005. A variety of collecting techniques, as previously described for freshwater turtles in the literature, were attempted to see which was the most effective for capturing A. cartilaginea. Hoop traps, commonly used to catch freshwater turtles in the Western hemisphere (Frazer et al. 1990; Legler 1960; Vogt 1980) were used. Native hoop traps, called bubu, were also used, in addition to another local fishing device called a selembau, which comprised a special system of scoops and nets attached to long poles. The device is stretched across a river or stream, scooping up whatever comes into contact with it (Fig. 2). Manual capture, otherwise known as muddling (Cagle 1942), was an effective albeit labor-intensive method of capturing softshell turtles. The technique involves wading through streams and probing areas of sand or mud and among roots with a stick, hands, or feet and could only be accomplished during times of low water level.

View Figure

• Download as Powerpoint
• Download Figure

Dietary analysis was conducted based on both stomach and fecal samples of living specimens. Dissection of stomachs to examine contents was not possible in this study because all turtles are protected under the Sarawak Wildlife Protection Ordinance (Sarawak Government Gazette 1998). Within 1 hour after returning from the field, stomachs were flushed by using the method developed by Legler (1977), which incorporated the modifications of Fields et al. (2000). Turtles were kept in plastic basins, which were 60 cm in diameter and 30 cm in depth, with some water. They were kept under observation until they defecated, usually 24 to 48 hours, and were released afterward at the place of capture. Each sample (stomach content or fecal matter) was washed in water by using a mesh strainer of 0.2-mm mesh size. These samples were immediately preserved in 70% ethanol for study.

Contents of the stomach and fecal samples were examined separately for each turtle. Types of food were sorted under low magnification by using an Olympus SZX9 dissecting microscope. The presence of each dietary group was recorded to the lowest identifiable taxon, typically to ordinal level of classification. In addition, 2 categories were used: unknown arthropods and unknown vertebrates. Any unidentifiable part of an insect or any other arthropod, i.e., wing, claw, etc., was grouped into the first category. The unknown vertebrates category represented any part known to be from a vertebrate, i.e., bone fragments, but unclassifiable further. The volume of each kind was determined by the displacement of water in either a 5-mL or 25-mL graduated cylinder, as described by Moll and Legler (1971). Descriptive analysis of data that pertained to contents of stomachs and fecal samples were expressed in 2 ways:

We tested for correlation between body size and the different food categories and volume found of either stomach contents or fecal matter by using the Pearson correlation test. Data were log-transformed. Analyses were performed by using SPSS for Windows ver. 10.1.

Results

In total, 18 animals were obtained, which comprised 3 juveniles, 9 females, and 6 males. Juveniles did not provide stomach samples and yielded little in terms of fecal matter and therefore, were excluded from the dietary analysis. Only 9 adult animals (4 males and 5 females) provided stomach contents; whereas, 13 animals (5 males and 8 females) provided fecal samples. Only 7 turtles provided both stomach contents and feces; whereas, 6 animals provided feces only, and 2 animals provided only stomach contents. Parasitic nematodes were found in 1 female and 2 male turtles, all from Loagan Bunut National Park. Furthermore, sand- or gravel-type substrate was found in all fecal samples. It is assumed that these items were not intentionally ingested and consequently were excluded from further analysis.

All 15 turtles that provided either stomach samples, feces, or both were captured evenly during the different seasons. Five animals, all female, were caught during the end of the northeast monsoon (wet season), which extends from November to March. All 5 had eaten plants, 4 had also eaten unknown vertebrates, and 3 had taken various arthropods and had fish parts in either their stomach or fecal samples. Five animals (3 male, 2 female) were captured during the southwest monsoon (dry season), which occurs from June to September. Four had plants in their stomach or fecal samples; 3 had fish and unknown vertebrate material; and only 1 provided arthropod parts. In the nonmonsoonal period (April, October), 5 animals (3 male, 2 female) were also caught. All animals had plant material and fish parts in their fecal or stomach samples; 4 animals had insect parts and parts of an unidentifiable vertebrate.

In total, 60% (n = 9) of turtles obtained in this study provided stomach contents. In all, 66% (n = 4) of the males provided stomach contents; whereas, 55% (n = 5) of the females provided stomach contents. They consisted of both aquatic and terrestrial plants, invertebrates, and the remains of fish. Muscle tissues and bone fragments from unidentifiable vertebrate species also occurred in stomach contents. Seven of the 9 samples contained plant material; although, only 5 samples contained animal parts. Frequency of occurrence of animals found in the stomach contents included unknown arthropods (1 female), dragonflies (Odonata) (1 female), spiders (Arachnida) (1 female), teleost fishes (Pisces) (1 female, 1 male), and unknown vertebrate remains (2 females, 3 males).

Percentage of volume within stomach contents is presented in Table 1. Plant material was present in 33% of the stomach samples; whereas, unknown vertebrate parts represented 48% of the total volume of stomach contents. The remaining percentages were teleost fishes (7%), unknown arthropod remains (4%), Odonata, and Arachnida, each comprised 4% of total volume. Percentage frequency of occurrence for food items found in stomachs of both male and female A. cartilaginea is presented in Fig. 3.

Table 1. Stomach contents of 9 Amyda cartilaginea from Sarawak, collected between May 2004 to April 2005.^a

View Fullscreen

View Figure

• Download as Powerpoint
• Download Figure

A Pearson correlation test was used to determine the relationship between body size and the number of identifiable prey types in the stomach. No correlation (r = –0.313, p = 0.412) was observed between these contents and body mass, which indicated that larger turtles would not necessarily take a greater type of dietary resources. The largest mass of an individual that provided stomach contents was 4 kg. Turtles contained food from 1 to 3 prey items in their stomachs. A Pearson correlation test was also used to check for any relationship between body size and the volume of stomach contents. There was no correlation (r = –0.518; p = 0.153) between these 2 variables.

Of the 15 adult turtles obtained, 13 (87%) provided fecal samples. Five (83%) of the males provided fecal samples; whereas, 8 (89%) of the females provided fecal samples. Items found in fecal samples expressed in frequency of occurrence are summarized in Table 2. They consisted of plant material (100%), arthropod remains (62%), flies (Diptera) (7.7%), beetles (Coleoptera) (7.7%), ants (Hymenoptera) (23%), dragonflies (Odonata) (7.7%), snails (Gastropoda) (7.7%), teleost fish (Pisces) (69%), birds (Aves) (7.7%), and unknown vertebrates (84%). The percentage frequency of occurrence of prey items found in feces is summarized for both sexes pooled in Fig. 4.

Table 2. Fecal sample of 13 Amyda cartilaginea from Sarawak, collected between May 2004 through April 2005.^a

View Fullscreen

View Figure

• Download as Powerpoint
• Download Figure

The percentage of volume of total contents in the fecal samples is presented in Table 2. Plant material had a total percentage volume of 56%, followed by teleost fish at 16%. Unknown vertebrates followed at 14% of the total volume, with birds at 6%, miscellaneous unidentifiable arthropods at 4%, Hymenoptera at 1%, and Diptera, Coleoptera, Odonata, and Gastropods each at < 1%. The small sample size does not permit statistical testing, but there seems to be no difference between the dietary preferences between males and females. All males and females that provided fecal samples contained plant materials. Three of the 5 males provided had insect parts; whereas, 5 of the 8 females had the same. Four of the 5 males provided unknown vertebrate material; whereas, 6 of the 8 females did the same. Interesting items recovered were bird feathers and bones found in a single adult male (field number KJ–126).

Similar to the results of the stomach contents, no correlation (r = 0.005, p = 0.987) was found between body mass and the number of identifiable prey items, which indicated that larger animals would not necessarily harvest a greater number of prey resource types. The largest animal that provided fecal matter for analysis was a female (field number KJ–31), which weighed 18 kg. Similar to the results of the stomach contents, there was no correlation (r = –0.101; p = 0.742) between body size and the volume of fecal matter.

Discussion

Overall, there appears to be a dearth of studies on diets of wild softshell turtles (reviewed by Moll and Moll 2004), perhaps because of their large size (hence, difficulty in capture and manipulation) and their presumed rarity or at least low densities relative to other organisms of similar size. Collection of large sets of field data on A. cartilaginea is difficult in Sarawak, where this species is actively hunted for food by nearly all indigenous peoples, and the species may consequently have become trap shy or otherwise not easily captured. In addition, it appears to be an extremely elusive animal. Although it may not move great distances, it cannot be seen from above or in the water, which is highly turbid (19–50 cm at Loagan Bunut National Park, 29–39 cm at Balai Ringin), and hence, with low or no visibility, the carapace color being either brown or black, apparently does not bask, at least on land, and does not enter hoop nets. Therefore, obtaining large data sets for this species is difficult and expensive.

The dominant foods found in both stomach and fecal analysis were plant materials, followed by unknown vertebrates, which indicated that A. cartilaginea is an omnivore. Tendencies toward omnivory were reported for another softshell turtle, Trionyx triunguis (see Branch 1988; Ernst and Barbour 1989), and the tendency to feed upon carcasses is known from other softshell species (Das 1995, Aspideretes gangeticus and Aspideretes hurum; Taskavak and Atatür 1998, Rafetus euphraticus; Akani et al. 2001, T. triunguis). Identifiable terrestrial vertebrates were found in 1 fecal sample in this study. Several white feathers, along with bone and muscle tissue that belonged to an indeterminate bird were recovered from an adult male. These remains may have come from an intermediate egret (Egretta intermedia) because this is the only bird species found at Loagan Bunut National Park with completely white plumage (Laman et al. 2006). Turtles have been known to take birds off the water surface for food. Pryor (1996) observed a Chelydra serpentina taking a semi-palmated sandpiper (Calidris pusilla) from the water's surface. In this article, Pryor also reviewed other accounts of C. serpentina predation on birds: laughing gull (Larus atricilla), lesser yellowlegs (Tringa flavipes), and possibly Forster's tern (Sterna forsteri), all migratory sea or shore birds. Unspecified species of birds have also been recorded in the diet of Trionyx triunguis (Gramentz 2005).

It is possible that the our softshell turtle took the intermediate egret off the water surface as it was foraging. Intermediate egrets are migratory birds and come to the wetlands of Borneo seasonally and at Loagan Bunut when water and fish are plentiful in the lake. This may represent a larger variety of food sources for these turtles during certain periods. Whether A. cartilaginea is preying upon terrestrial birds, as well as animal carcasses, it could be considered a keystone species in its contribution to recycling nutrients in the peat swamp ecosystem.

Results of the Pearson correlation test indicated that there was no correlation between body size and the number and volume of identifiable prey items in the stomach. Although the sample size was small (n = 9), these results may be related to problems with the stomach-flushing technique. It would be a safe assumption that larger turtles consume larger amounts of food. However, this was not what we found in our study. In fact, the largest sample (0.7 mL) was provided by a turtle that weighed 1.1 kg. Stomach flushing success is often inversely proportional to the size of turtle, and the volume of water injected with a hand-operated syringe may be insufficient for large animals (Legler 1977; Demuth and Buhlmann 1997). Without dissection, it is impossible to know whether failure to dislodge food from the stomach means that the stomach is empty or the technique is ineffective.

Sexual and ontogenetic differences in diet in the species remain unknown and appear to be negligible, but the sample in this study did not permit appropriate statistical testing. In addition, data were unavailable for juvenile turtles. Such differences have been highlighted in other trionychids, such as Apalone mutica (Plummer and Farrar 1981). In several other freshwater turtle species, there is a well-documented shift from carnivory to herbivory with increasing body size: Trachemys scripta (Clark and Gibbons 1969; Hart 1983), Emydura krefftii (Georges 1982), Pangshura tentoria (Das 1985), and Chelonia mydas (Bjorndal 1985). Further sampling with a wide range of size classes is needed to answer these questions.

There was no correlation between body mass and the number of prey items, nor between body mass and the volume of fecal matter. The sample size did not permit quantitative analysis of any seasonal differences. Seasonal differences in rainfall and water availability may influence the diet and behavior of turtles, especially in places where streams dry out completely, and turtles may be forced to aestivate or perhaps burrow and remain underground in a quiescent state (Jensen and Das 2008). Changes in the water level may influence feeding behavior, especially in areas where many of the streams dry out completely. During these periods, turtles were found buried in mud. All 3 individuals located in mud were found at Loagan Bunut National Park in small streams, buried at 0.3- to 0.5-m depth in wet mud in previously submerged stream banks. A smaller number of turtles may have had insects in their diet during the dry season because aquatic insects tend to breed during the wet season and, consequently, are less likely to be available during this period. Conversely, as the water levels decrease, fish are confined to smaller areas, which make them more easily harvestable prey for turtles. In addition, during the dry season, there is a large die-off of fish at Loagan Bunut; therefore, some of the fish intake may be from scavenging. Although the samples were small, these data are a noteworthy addition to the knowledge of the ecology of this turtle species.