Diet of the Texas Map Turtle (Graptemys versa): Relationship to Sexually Dimorphic Trophic Morphology and Changes Over Five Decades as Influenced by an Invasive Mollusk
ABSTRACT
The diet of the Texas map turtle (Graptemys versa) in the Colorado River drainage was studied using 1) stomach flushing and collection of feces in 1998–1999 and 2) dissection of museum specimens collected in 1949 prior to invasion of their habitats by exotic Asian clams (Corbicula spp.). Feces were found to be better indicators of dietary diversity based on log–log comparisons of sample volume with turtle size and the fact that Corbicula were absent in stomach flushes of small female specimens that yielded abundant shell fragments in their feces. Adult males fed most heavily on aquatic insect larvae (primarily trichopterans and ephemeropterans) and showed little change in diet over 5 decades. Small females that overlapped adult males in body size had wider heads and alveolar surfaces of the jaws than did males and fed predominantly on Corbicula, snails, insect larvae, and algae. Females that were larger than males in body size showed a change from a diverse diet in 1949 that included sphaeriid clams, trichopteran larvae, sponges, bryozoans, and algae, to nearly exclusive feeding on the invasive mollusk Corbicula 5 decades later. Based on the present report and previous studies, it appears that 1 result of the invasion of rivers in North America by Corbicula is greatly decreased diversity of diet in adult female Graptemys of moderate head width (mesocephaly).
Map turtles (Graptemys) are a trophically diverse group of river-dwelling emydid turtles. The 12 species have been categorized into microcephalic, mesocephalic, and megacephalic groups of 4 species each, based on the trophic morphology of adult females (Lindeman 2000a). The 3 categories correspond roughly to the degree to which adult females feed on mollusks such as snails, clams, and mussels: females of microcephalic species ingest few mollusks, females of mesocephalic species tend to mix large quantities of mollusks with a variety of other prey, and females of megacephalic species feed nearly exclusively on mollusks (Lindeman 2000a).
Male Graptemys are much smaller than females in all 12 species, and rarely feed heavily on mollusks; their prey generally consists of softer-bodied aquatic invertebrates (Lindeman 2000a). Juvenile female Graptemys that are similar in body size to conspecific adult males have proportionally wider heads than the males (Lindeman 2000a). In males and small juvenile females of similar body sizes, the females tend to be more highly molluscivorous (Sanderson 1974; Porter 1990).
The invasion of eastern North American rivers by the exotic Asian clam (Corbicula spp.) began in the late 1950s (McMahon 1982) and has greatly altered the diet of females of many of the more highly molluscivorous Graptemys. Shively and Vidrine (1984) described a strong contrast in the types of mollusks eaten by adult female G. pseudogeographica in rivers in Louisiana with and without Corbicula, and G. pseudogeographica is also known to feed moderately heavily on Corbicula in Arkansas (Vogt 1981) and Kentucky (Lindeman 2000b). Adult females of 3 species have been described as feeding almost exclusively on Corbicula (Graptemys ernsti, Shealy 1976; Graptemys geographica, Moll 1980; Graptemys caglei, Porter 1990).
The Texas map turtle (Graptemys versa) is one of the least-studied North American turtle species with regard to basic natural history (Lindeman 2001). The species is endemic to the Colorado River and its tributaries in central Texas (Ernst et al. 1994). Young G. versa feed primarily on aquatic insects (Lehmann 1979, cited in Bertl and Killebrew 1983). Originally described as a “narrow-headed” species (Bertl and Killebrew 1983), G. versa was classified as having mesocephalic females (Lindeman 2000a). Diet of the species with regard to molluscivorous tendencies might thus be expected to be similar to the diets of the other 3 species in which females are mesocephalic and feed moderately heavily or nearly exclusively on mollusks: G. geographica (Lagler 1943; Vogt 1981; White and Moll 1992), G. pseudogeographica (Vogt 1981; Shively and Vidrine 1984; Lindeman 1997, 2000b), and G. caglei (Porter 1990). Indeed, Kizirian et al. (1990) reported gastropod shells in the feces of a female G. versa of record body size.
I studied the diet of G. versa using stomach flushing and collection of feces from specimens caught in 1998–1999 in the South Llano River, Texas, to provide the first detailed information on diet of the species. In this study, I also compare the diets of males, small females of body sizes similar to those of males, and larger females, with emphasis on degree of molluscivory. I compare these results with samples taken from the gastrointestinal tracts of museum specimens collected 50 years earlier, in 1949, prior to invasion of the river by Corbicula, to examine changes in diet (with regard to both relative abundance and types of mollusks) that have occurred since this change in the prey base occurred.
METHODS
Study Area
Trapping was conducted during May 1998 and May 1999 on the South Llano River 19 km south of Junction, Kimble County, Texas, on a 4.5-km section of river located between 2 crossings of State Highway 377 and extending downstream of the northern crossing (30°21′N, 99°54′W). The South Llano River is a spring-fed river with cool, clear water and a bedrock or large cobble substrate except in some muddier, deep, slow-current pools. The study site was a series of alternating slow, deep pools and shallow, fast-current riffles. Museum specimens collected in 1949 were captured on a private ranch immediately upstream of the study site, where the habitat is similar (pers. obs.).
Sample Collection
I captured turtles with fykenets (Vogt 1980) and basking traps (MacCulloch and Gordon 1978), and occasionally by hand. Standard measurements included midline plastron length (PL), head width (HW) at the upper jaw, and alveolar width (AW) of the upper jaw. I used stomach flushing (Legler 1977) and overnight collection of feces in plastic tubs of water to obtain dietary samples. I flushed stomachs using continuous streams of water introduced via plastic tubes of 3.8 or 6.5 mm diameter (for turtles < 100 or > 100 mm PL, respectively) to unanesthetized turtles restrained manually. I individually marked turtles (Cagle 1939) and released them at their sites of capture; occasional recaptures were not detained for dietary sampling. I dissected the gastrointestinal tracts of museum specimens collected on 30 April 1949 and examined prey remains.
Sample and Data Analysis
To test the reliability of sample procurement by stomach flushing vs. fecal collection, I examined the log–log correlation of sample volume with PL. The expected slope of this relationship is 3 because it compares a volume with a linear measurement. A slope significantly different from 3 would indicate that completeness of the sample obtained by a method varies as a function of turtle size. Variation in fullness of the gut would influence the r2 value of the log–log relationship of sample volume to PL but under the assumption that fullness of the gut is not allometric (i.e., variation in fullness does not covary with PL), such variation would not influence the expected slope of 3.
Stomach contents and fecal samples were sorted by prey taxa and volume of each prey taxon was measured to the nearest 0.1 mL by volumetric displacement in a 10-mL graduated cylinder. Samples displacing < 0.1 mL were visually estimated to be either 0.05 or 0.01 mL in volume. Samples from museum specimens were too highly fragmented and fragile to separate into prey taxa for volumetric displacement; instead, I visually estimated the percent of the total sample volume that each prey taxon present in a sample comprised (nearest 5% for abundant taxa, nearest 1% for sparsely-represented taxa).
For each of 3 classes of turtles (males, small females overlapping males in PL, and large females) within 3 sets of samples (stomach flushes from 1998–1999, feces from 1998–1999, and gastrointestinal tracts from 1949), prey taxa were summarized by mean percent volume (Vi) of individual samples and percent frequency recorded for the class (Fi). Use of mean percent volume rather than percent of total volume across samples prevents skewing of Vi values that may occur because of variation in sample volume, which can be extensive (Lindeman, unpublished data, 1993–1995); i.e., each sample from a turtle is weighted equally, rather than being weighted by sample volume. Values of Vi and Fi were used to calculate an Index of Relative Importance (IRI; Hyslop 1980 as modified by Bjorndal et al. 1997) for each prey taxon within each class of turtles within each set of samples, as follows:
Values of IRI sum to 100, and thus may be used comparatively to represent the importance of various prey items for different classes of turtles or different sample categories.
I used the unweighted pairwise group mean averaging algorithm (UPGMA; Sneath and Sokal 1973) to quantify the similarity of samples in pairwise comparisons of males, small females, and large females. The UPGMA algorithm generates pairwise similarity scores that range from 0 (no overlap of prey taxa) to 1 (complete overlap in taxa of prey and their relative abundance in the diet).
RESULTS
Trophic Morphology
Small females had significantly greater HW and AW than adult males after correction for PL in ANCOVA (for HW, F1,35 = 30.32, p < 0.0001, Fig. 1a; for AW, F1,35 = 54.36, p < 0.0001, Fig. 1b). The sexes did not differ in AW after correction for HW (F1, 35 = 1.33, p = 0.26, Fig. 1c). In each analysis the covariate was significant (all p < 0.001) and the interaction term of covariate with sex was not (all p > 0.60).
Citation: Chelonian Conservation and Biology 5, 1; 10.2744/1071-8443(2006)5[25:DOTTMT]2.0.CO;2
Relationship of Sample Volume to Body Size
The log–log relationship of sample volume to PL was significant for both stomach contents (r2 = 0.24, n = 39, p < 0.01; Fig. 2a) and feces (r2 = 0.39, n = 50, p < 0.001; Fig. 2b). Confidence limits on the slope of the relationship included the expected slope of 3 for both analyses (slope = 2.19 ± 1.30 for stomach contents, 2.91 ± 1.07 for feces).
Citation: Chelonian Conservation and Biology 5, 1; 10.2744/1071-8443(2006)5[25:DOTTMT]2.0.CO;2
1998–1999 Stomach Contents
Stomach flushing was successful on 39 specimens (1 unsexed juvenile, 36 mm PL; 21 adult males, range 57–73 mm PL; 7 small females, range 59–74 mm PL; and 10 large females, range 83–145 mm PL; Table 1, Fig. 3a). Males fed most heavily on insect larvae (ephemeropterans and trichopterans) and to a lesser degree on snails. Small females fed most heavily on snails, ephemeropteran larvae, and filamentous algae. Large females fed almost exclusively on Corbicula. Stomach contents of the juvenile consisted of 4 ephemeropteran larvae. UPGMA similarity scores were 0.49 for a comparison of males and small females, 0.08 for males and large females, and 0.06 for small and large females.
Citation: Chelonian Conservation and Biology 5, 1; 10.2744/1071-8443(2006)5[25:DOTTMT]2.0.CO;2
1998–1999 Feces
Feces were obtained from 50 specimens (1 unsexed juvenile, 44 mm PL; 25 males, range 57–79 mm PL; 8 small females, range 59–74 mm PL; and 16 large females, range 83–163 mm PL; Table 2, Fig. 3b). When compared to stomach contents collected from many of the same specimens, the major differences were the heavy incidence of Corbicula and lack of filamentous algae in samples from small females, and the lack of ephemeropteran larvae in males and small females, which was coupled with a rise in incidence of fragmented, unidentifiable prey remains. Feces of the juvenile contained 50 trichopteran larvae. UPGMA similarity scores were 0.68 for a comparison of small and large females, 0.60 for males and small females, and 0.04 for males and large females.
1949 Gastrointestinal Tracts
I dissected 21 specimens from the 1949 series, of which 19 contained colons packed with prey remains (7 males, range 64–74 mm PL; 12 females, range 110–148 mm PL; Table 3, Fig. 3c). No specimens contained prey in the stomach or small intestine. Males were almost exclusively insectivorous, with most prey being trichopteran larvae. Females also fed heavily on trichopteran larvae, as well as on sphaeriid clams, sponges, and bryozoans.
DISCUSSION
Sexual Differences in Trophic Morphology
Graptemys versa shows the typical pattern of sexual differences in trophic morphology described previously for the genus (Lindeman 2000a): juvenile females have wider heads than do adult males (Fig. 1a) and alveolar surfaces of females are greater in absolute width (Fig. 1b) but not in width relative to head width (Fig. 1c). In a previous analysis of specimens of G. versa from a variety of populations, differences in HW between the sexes were marginally nonsignificant (p = 0.058; Lindeman 2000a), but all other species showed significantly greater HW in females, and it is likely that the pattern is ubiquitous in Graptemys.
Sample Reliability
Although analysis of the log–log relationship of sample volume with body size demonstrated no significant deviation from the expected slope of 3 for either stomach flushes or feces, 2 findings indicate that feces may have provided a more accurate and useful indication of overall diet in Graptemys versa: 1) there was more variability in the relationship of stomach content volume to body size, with a slope value farther from 3, and 2) there were discrepancies in comparison of the prey taxa of stomach contents and of feces. The absence in feces of ephemeropteran larvae, which were common in stomach contents, is probably related to the increase in unidentified insects in feces, because these insects fragment during passage through the gut, making identification difficult; trichopteran larvae are much less prone to fragmentation (pers. obs.). However, the lack of Corbicula remains in the stomach contents of small females cannot be so readily explained. The high incidence of clam shells in the feces but not the stomach contents of small females suggests that the diameter of the tubing employed in flushing stomachs may have prevented smaller specimens from regurgitating clam shells. Filamentous algae present in stomach contents were absent from fecal samples, possibly because of disintegration during digestion.
Stomach flushing was successful for only 39 turtles, whereas feces were collected from 50 turtles, and at least 4 smaller individuals (< 100 mm PL) died following stomach flushing. Passage rates and the digestibility of prey types in the intestine may vary taxonomically, biasing the relative importance of some taxa in analysis of fecal samples upward or downward, but overall, fecal sampling seemed to be a more useful technique than stomach flushing for this species.
Sexual Difference in Diet
Females ate more mollusks than males. The importance of trophic morphology for molluscivory in Graptemys is evident in the finding that small females of male size, with their wider heads and alveolar surfaces, ate more mollusks than males did. Similar findings have been reported for 2 species with megacephalic females, Graptemys barbouri and Graptemys ernsti (Sanderson 1974; Shealy 1976). Comparative analyses suggest that the evolution of Graptemys has been characterized by a correlated response of changes in head and alveolar width in females but not in males, because of the more molluscivorous tendencies of females (Lindeman and Sharkey 2001).
Similarity scores comparing adult males, small females, and large females were probably more accurate for fecal samples than for stomach-flush samples for the reasons stated above. These scores place small females as more similar in diet to larger females than they are to males of similar body sizes. These differences in similarity scores are driven primarily by the preference for mollusks in females.
Comparison to Other Map Turtles
The diet reported for G. versa females is typical for mesocephalic Graptemys females, with heavy reliance on mollusks plus inclusion of sponges, bryozoans, aquatic insect larvae, and algae. Similar diets have been reported for the females of the other 3 species that have mesocephalic females (G. pseudogeographica, G. geographica, and G. caglei; Vogt 1981; Porter 1990; Lindeman 1997, 2000b). Adult male G. versa fed most heavily on trichopteran and ephemeropteran larvae. With less overall interspecific variation in the trophic morphology of male Graptemys, there is no pattern of male diet associated with the categorization of species as having micro-, meso-, or megacephalic females (reviewed in Lindeman 2000a); however, heavy reliance of males on trichopteran and ephemeropteran larvae has frequently been reported (Sanderson 1974; Shealy 1976; Moll 1980; Shively and Jackson 1985; Porter 1990; Kofron 1991; Lindeman 1997, 2000b).
Change in Diet Between 1949 and 1999
The first record of the exotic clam Corbicula for Texas rivers is from 1972 (McMahon 1982). Since its invasion of the South Llano River it has become the predominant prey item of larger females, accounting for a mean of 66% of each stomach-contents sample (IRI = 94) and 95% of each fecal sample (IRI = 99). Associated with this change in diet is a marked reduction in dietary diversity in adult females. Items found in samples from 1949 that were not found in the diet of large females in 1998–1999 included native sphaeriid clams, sponges, bryozoans, and algae. The diet of males has in contrast shown little change from 1949 to 1998–1999.
Megacephalic female Graptemys feed nearly exclusively on mollusks (Cagle 1952; Sanderson 1974; Shealy 1976). Although mesocephalic females also feed heavily on mollusks, some reports indicate a more diverse diet (e.g., Vogt 1981; Lindeman 1997, 2000b). Analysis of the diet of G. versa females over 5 decades shows a shift from the moderate molluscivory typical of mesocephalic females to nearly exclusive use of the abundant nonnative mollusk Corbicula. Moll (1980) reported a similar change in diet for G. geographica in Illinois. In another Texas endemic, G. caglei, Corbicula comprised 88% of the dry mass of stomach contents taken from dissected females (Porter 1990). Shively and Vidrine (1984) reported nearly exclusive feeding on Corbicula by 13 female G. pseudogeographica in 2 southwestern Louisiana rivers and found a more diverse diet in 1 specimen from an upstream location where Corbicula was less abundant. It is likely that populations of G. pseudogeographica from downstream localities in southwestern Louisiana and G. caglei have experienced a change in diet similar to that of G. versa associated with the invasion of their habitats by Corbicula. Why females in these cases have switched to nearly exclusive use of Corbicula while females in other populations have maintained a more diverse diet that includes Corbicula (e.g., G. pseudogeographica in Kentucky; Lindeman 1997, 2000b) is unknown, but may be related to differences in relative abundance of the invasive species.
The log–log relationships of (a) plastron length (PL) and head width (HW), (b) PL and alveolar width (AW), and (c) HW and AW for adult male and juvenile female Graptemys versa of similar body sizes. All measurements in mm.
The log–log relationships of sample volume (mL) to plastron length (mm) for (a) stomach flushes and (b) feces of Graptemys versa with best-fitting lines from simple linear regression.
Index of relative importance (IRI) values for major prey taxa in 3 classes of Graptemys versa from (a) field-collected stomach contents from 1998–1999, (b) field-collected feces from 1998–1999, and (c) gastrointestinal tracts of museum specimens preserved in 1949. No small females (plastron length ≤ 79 mm) were available from the 1949 series of museum specimens. COR = Corbicula; SPH = sphaeriid clams; GAS = gastropods; TRI = trichopteran larvae; EPH = ephemeropteran larvae; COL = adult coleopterans; INS = unidentifiable fragmented insects; POR = poriferans; BRY = bryozoans; ALG = filamentous algae.
