Feeding Ecology of Acanthochelys spixii (Testudines, Chelidae) in the Cerrado of Central Brazil
Abstract
We studied the diet of Acanthochelys spixii in a wild population in the Cerrado of central Brazil for 19 months, investigating ontogenetic, sexual, and interindividual variation. The diet consisted mainly of nymphs of Odonata, although other insects, amphibians, and plant material were also present. We observed no ontogenetic shifts in diet composition (e.g., no shift from carnivorous juveniles to herbivorous adults), which can be related to the high abundance of prey at the study site. There was no association between prey size and turtle carapace length, with larger animals still taking small prey. Dietary niche overlap was high, and there was no difference in niche breadth between sexes. However, differences in diet composition suggested differential habitat use, with males using more the periphery and females using more the center of ponds. Diet composition varied more among males than among females, which can result from higher diversity of prey at the pond margins, higher movement rates, or larger home range of males. The high frequency of empty stomachs (41%) reflected life-history characteristics of turtles (e.g., low metabolism, ectothermy, late sexual maturity, and great longevity). The importance of prey categories sensitive to pollution in the diet of A. spixii highlights the integrity of the study sites and the vulnerability of these populations to the rapid degradation of Cerrado biome.
The study of chelonian diets can explain variations in morphological characters (Rhodin et al. 1984), habitat choice (Plummer and Farrar 1981), and patterns of riparian vegetation composition (Moll and Jansen 1995). Many food sources are available seasonally, leading to seasonal dietary changes (Schoener 1971). For example, Chelodina rugosa feeds more upon nymphs of Odonata in the rainy season and more upon fish in the dry season, because fish become more aggregated in the dry season, making them more available to turtles, whereas nymphs of Odonata are more abundant in the rainy season (Kennet and Tory 1996). Further, omnivorous species can have a predominantly carnivorous diet during rainy periods and a predominantly herbivorous diet in dry periods, caused by the higher availability of aquatic insects during rainy periods (Clark and Gibbons 1969; Ottonello et al. 2005).
Ontogenetic shifts, from carnivorous juveniles to herbivorous adults, are relatively common in omnivorous freshwater turtles (Clark and Gibbons 1969; Georges 1982; Spencer et al. 1998; Allanson and Georges 1999). These changes are often combined with shifts in habitat use, from shallow to deep water, which in turn can affect food availability (Moll 1976; Hart 1983; Kennet and Tory 1996; Bouchard and Bjorndal 2006). In addition, there are energetic explanations for these shifts, because adults have low metabolic rates and tend toward herbivory, whereas juveniles need more protein to sustain their rapid growth (Pough 2004). In carnivorous animals, changes in diet throughout ontogeny reflect the rising ability to capture, subdue, and swallow larger prey. Thus, adults continue to ingest small prey, while also adding larger ones to their diet (Souza and Abe 1998).
Sexual variations in diet composition may result from sexual dimorphism in body size, which leads to differences in prey size and foraging behavior (Pough 2004). In freshwater turtles, females are often larger than males and, thus, feed upon larger prey (Georges 1982). Also, differential habitat use can lead to sexual variations in diet (Plummer and Farrar 1981). Moreover, sexual variation in diet can result from the greater energetic demand of females during the reproductive period (Ford and Moll 2004; Macip-Ríos et al. 2010). Females of Podocnemis unifilis consume more fish and mollusks than males, probably to satisfy calcium requirements for the production of eggshells (Fachin-Terán et al. 1995).
Although many animal populations have generalist diets, each individual can function as a specialist with a restricted diet (Bolnick et al. 2003). In this case, interindividual variation can work as a mechanism to reduce intraspecific competition (Van Valen 1965; Roughgarden 1972; Polis 1984). Existing models predict that generalist populations of individual specialists will occur under conditions of high food predictability (little or no seasonality) and high food availability and diversity (Van Valen 1965; Roughgarden 1974). Studies of population dynamics and resource partitioning consider conspecific individuals as ecologically equivalent, thus ignoring interindividual variation in diet. There is evidence that this kind of variation is common; however, it is underestimated and deserves more attention (Bolnick et al. 2003; Costa et al. 2008).
The black spiny-neck turtle, Acanthochelys spixii (Chelidae), is characterized by a dark gray to black carapace with a shallow dorsal groove and several elongated pointed tubercles on the neck (Duméril and Bibron 1835; Ernst and Barbour 1989). It inhabits swampy areas in Argentina, Uruguay, and Brazil, from Rio Grande do Sul to Distrito Federal (Ernst and Barbour 1989; Iverson 1992; Brandão et al. 2002). Sporadic observations suggest it has a carnivorous diet consisting of snails, tadpoles, frogs, fish, and insects (Coelho et al. 1975; Brandão et al. 2002; Achaval and Olmos 2003; Bonin et al. 2006; Estrades et al. 2008). This turtle does not leave the water except to migrate or lay eggs (Astort 1983; Bonin et al. 2006). There are few studies on the biology of A. spixii (Buskirk 1991; D'amato and Morato 1991; Monteiro-Filho et al. 1994), many of them conducted in captivity (Astort 1983; Lehmann 1988; Molina 1998). The northernmost populations, in highly fragmented areas of the central Brazilian Cerrado (Brandão et al. 2002), were recently subjected to demographic investigations (Fraxe Neto et al. 2011). The natural habitats of A. spixii are threatened by habitat loss and urbanization (Richard and Waller 2000; Ribas and Monteiro-Filho 2002; Bonin et al. 2006), and the species is listed as Near Threatened in the IUCN Red List of Threatened Species (IUCN 2010), which is a cause of concern (Bonin et al. 2006). Considering the global situation of freshwater turtles (Burke et al. 2000; Turtle Taxonomy Working Group 2010; Turtle Conservation Coalition 2011), the paucity of studies about chelids in Brazil (Souza 2004; Souza and Molina 2007), and the fast destruction of the Cerrado biodiversity hotspot (Myers et al. 2000; Klink and Machado 2005), this study aims at determining the diet composition of A. spixii in central Brazil, investigating ontogenetic, sexual, and interindividual sources of variation.
METHODS
Study Site
The study took place in two ponds at Parque Nacional de Brasília (PNB), Distrito Federal, Brazil: Lagoa do Henrique, with approximately 4-ha perimeter in the rainy season, and Lagoinha do Exército, with approximately 2.95-ha perimeter in the rainy season. The distance between the ponds is 8 km, and both are swampy areas with a high density of macrophytes. With an area of 42,389 ha, PNB is the largest biological reserve of Distrito Federal and part of the Cerrado Biosphere Reserve. Climate is markedly seasonal, with a dry season from May to September and a rainy season from October to April, receiving almost all of the 1500–2000 mm annual precipitation (Nimer 1989). The annual mean temperature ranges from 20°–22°C, being relatively constant throughout the year (Nimer 1989).
Field Methods
We captured turtles manually and with funnel traps baited with canned sardines (we perforated cans to prevent ingestion of bait). From each captured individual, we recorded the maximum carapace length (straight-line, with Mitutoyo® electronic calipers to the nearest 0.01 mm), sex, site, and date. Sex determination was based on characteristics of the plastron and cloaca (Dosapey and Montaño 2004). Animals with no definite sexual characters were considered juveniles (Métrailler 2005). Each captured turtle was individually marked with marginal scute notches (Cagle 1939). Stomach contents were taken right after capture by stomach flushing, applied until a mucus-coated mass came up, indicating that the stomach was completely flushed (Legler 1977). Stomach contents were preserved in 70% alcohol and identified under a stereomicroscope.
Stomach contents of animals captured more than once were considered different samples, because of the small number of captures. Individuals from both ponds were considered as belonging to a single population, because terrestrial migration is common in Acanthochelys and was previously observed in A. spixii (Lema and Ferreira 1990; Richard 1994; Lema 2002; Achaval and Olmos 2003). Stomach contents were sampled for 18 months, from September 2006 to March 2008. Traps were checked twice a week, and bait was replaced every 15 days. At Lagoinha do Exército, sampling occurred from April to September 2007, including nocturnal periods in August. Sampling at this site was interrupted because of repeated trap stealing. A simultaneous study on the activity of A. spixii was conducted in the same ponds, and captured animals were tracked with spool-and-line devices. Consequently, stomachs could be flushed at the time of capture during several days.
Statistical Analysis
The length and width of intact prey were measured with calipers (0.01 mm), and prey volume was estimated as an ellipsoid:
where w is prey width and l is prey length. The numeric percentage of each prey category ([total number of prey category i] / [total number of prey] × 100) was calculated for each individual and also for pooled stomachs. The percentage of occurrence of each prey category was calculated as ([stomach contents with prey category i] / [total number of stomach contents] × 100). The relative importance of each prey category was determined by the Importance Value Index (IVI):
where F% is the percentage of occurrence and N% is the numeric percentage. Prey volume was not included in IVI estimates because of the large number of fragmented prey. Dietary niche breadths (B) were calculated for each stomach using the inverse of Simpson's diversity index (Simpson 1949):
where p is the numeric proportion of prey category i, and n is the number of categories. Values of B vary from one (usage of a single category) to n (equal usage of all categories). Sexual differences in niche breadth were tested with rarefaction, using 1000 randomizations with replacement in EstimateS 8.0 (Colwell 2006), followed by a t-test (Brower et al. 1998).
To investigate sexual differences in dietary niche overlap, we used the following equation:
where p is the proportion of prey category i, n is the number of categories, and j and k represent the individuals (or groups) compared (Pianka 1973). Overlap (Øij) ranges from zero (no overlap) to one (total overlap). The IVI was used as a measure of prey usage. Next, the “Niche Overlap Module” in EcoSim 7.0 (Gotelli and Entsminger 2001) was used with 1000 randomizations. In this analysis, the null hypothesis is that niche overlap between sexes is larger than or equal to the simulated. The analysis was performed with randomization algorithms two (RA2) and three (RA3). RA2 replaces usage values of prey categories in the original matrix by random values between zero and one; however, it retains the zero structure of the matrix by replacing only values of the categories really used (Gotelli and Entsminger 2001). Therefore, RA2 is more suitable when some prey categories are not available to one of the sexes. RA3 randomizes usage values of prey categories between all possible categories, thus destroying the zero structure of the original matrix (Winemiller and Pianka 1990; Gotelli and Entsminger 2001). Therefore, RA3 is more suitable when all resources are available to all individuals. The analyses were repeated using only categories with IVI greater than 5% in at least one sex.
To assess the degree of individual variation in diet, we calculated pairwise dietary niche overlap values 1) among males, 2) among females, and 3) between males and females, using IndSpec 1 (Bolnick et al. 2002). Next, we used t-tests to compare 1) the mean pairwise overlap among individuals of the same sex with the mean pairwise overlap between individuals of opposite sexes, and 2) the mean pairwise overlap among females with the mean pairwise overlap among males.
To estimate prey richness, we built sample-based rarefaction curves for the pooled sexes and separately, for each sex, in EstimateS 8.0 (Colwell 2006) using 1000 randomizations without replacement. We calculated observed richness with the Mao Tau estimator, which involves data interpolation to smaller samples, reordering data at random. To estimate total richness, we used estimators that extrapolate sample data limits. Although Mao Tau represents the expected richness for a subsample of the total pooled categories, based on all categories actually discovered, extrapolation estimators estimate total richness including categories that were never collected (i.e., new categories that would emerge with additional sampling [Colwell et al. 2004]). We evaluated estimator performance by regressing mean estimated values (Mao Tau) against total richness (ACE, ICE, Chao 1, Chao 2, Jacknife, Jacknife 2, Bootstrap, and Michaelis-Menten), setting the intercept of the regression equation to zero (i.e., passing the regression line through the origin), following Brose et al. (2003). Precision was measured by the coefficient of determination (r2) of regressions, and bias was computed as the difference between the observed regression slope and the expected slope of one, which represents an unbiased and perfectly precise estimator. The best estimator was chosen as the most precise and least biased.
We used t-tests to assess intersexual differences in maximum and mean prey length, number of prey categories, and number of prey per stomach. To test for ontogenetic variation in diet, we used Spearman's correlations between turtle carapace length and the following variables: number of prey categories, number of prey items, niche breadth, and maximum and mean prey length per stomach. We performed correlation analyses for the pooled sexes and for each sex separately. To determine whether the proportion of stomach contents differed between sexes according to season (wet vs. dry), we used a Fisher's test.
We used logistic regression analyses (Tabachnick and Fidell 2001) to assess the dependency of empty stomachs on sex, carapace length, season, and capture method. Stomach contents from turtles that had spool-and-line tracking devices were considered from “manual captures” (the other method was using baited traps). To evaluate the importance of predictors, we used the chi-square statistic to compare the deviance residuals of a null model (no predictors) with those of larger models, formed by the sequential addition of predictors.
We used R 2.4.1 (R Development Core Team 2006) to run t-, z-, χ2, and Fisher's tests and correlation and regression analyses, all with a 5% significance level. We transformed original variables whenever necessary, to meet the assumptions of analyses, and also screened them for outliers, following Tabachnick and Fidell (2001). We transformed original variables to z-values and considered values with probability < 0.001 as outliers, resulting in 7 outliers, which were removed from subsequent analyses. Different ontogenetic stages (larva, pupa, nymph, adult) of the same prey category were considered as different prey categories, because they occupy distinct ecological niches. Plant material was not quantified; thus it was not included as a prey category in analyses that used prey abundance.
RESULTS
Captures, Sampling, and Diet Composition
From September 2006 to March 2008, we conducted 190 visits to the study site to check traps, change bait, and look for turtles. The technique was used in 38 distinct animals (14 females, 22 males, and 2 juveniles), and 19 stomach contents resulted from recaptured animals (Table 1). One of the juveniles was captured manually at night and the other with traps. All individuals were stomach flushed, but some had no material in their stomachs. The small number of sampled stomachs during the dry season precluded an assessment of seasonal variation in diet. There was no sexual difference in the proportion of sampled stomachs between seasons (F = 0.846, p ≅ 1).
We analyzed 188 prey items, 97 of which were too fragmented to be measured, corresponding to 29 prey categories that included frogs and tadpoles; larvae, pupae, nymphs and adult stages of insects; and plant material. Niche breadth was 8.14 ± 1.10 for the whole population, with a maximum possible value of 28 (because of the exclusion of plant material from niche breadth analyses), indicating that food resources were not used in the same proportion. The most precise and less biased estimator of asymptotic richness was Bootstrap (Table 2). In rarefaction curves, the observed richness of prey categories was always below estimated total richness (Fig. 1), indicating that additional sampling may reveal new prey categories. The total richness estimated for the whole population was 33 prey categories.
Citation: Chelonian Conservation and Biology 10, 1; 10.2744/CCB-0846.1
Dragonflies and damselflies (Odonata) were by far the most frequent and abundant prey, with the highest volume and importance index (Table 3). In this category, nymphs of Libellulidae were the most important, followed by nymphs of Aeshnidae (Table 3). The volumetric percentage of anurans was probably underestimated because only two adults could be measured; the others were in advanced stages of digestion. Macrophytes and algae occurred in some contents but always in low quantities. We also recorded some allochthonous (coming from outside the aquatic system) prey: adults of Odonata, Ploiariidae, Tenebrionidae, Carabidae, Homoptera, Formicidae, and Termitidae. Empty stomachs represented 41% of all flushed stomachs, and the logistic regression indicated that their occurrence was not associated with sex, carapace length, season, or capture method (χ2 = 1.287, p = 0.257).
Sexual Variation
Odonata was the most important category in both sexes; however, it was more frequent and important among females (Table 3). Also, plant material was more frequent among females. Larvae and adults of Anura were more important in the diet of males, whereas allochthonous prey occurred in both sexes. Niche breadth was 6.00 ± 1.04 for females and 7.90 ± 1.41 for males, but the rarefaction analysis indicated this difference was not significant (t = 1.16, p = 0.25). Niche overlap between sexes was 0.88, a value not different from the null expectation (p ≈ 1; same results for RA2 and RA3). When using only prey categories with IVI higher than five, the results remained unchanged (Øij = 0.89, p = 0.98, same results for RA2 and RA3). Nineteen prey categories were recorded in males and 18 in females, with a total estimated richness of 22 for both sexes. There were no intersexual differences in mean number of prey categories (t = −0.47, p = 0.64), mean prey length (t = −0.35, p = 0.72), maximum prey length (t = −0.15, p = 0.87), or total number of prey (t = −0.83, p = 0.42).
Mean pairwise overlap was 0.159 among males, 0.233 among females and 0.197 between males and females. There was no difference in mean pairwise overlap among individuals of the same sex and mean pairwise overlap among individuals of the opposite sex (t = 1.52, p = 0.13), indicating the absence of sexual dimorphism in diet composition. However, mean pairwise overlap among males was lower than among females (t = −4, p < 0.001), indicating a higher degree of individual variation in diet among males.
Ontogenetic Variation
Both juveniles and adults of A. spixii had a primarily carnivorous diet. There was no correlation, either considering both sexes together or separate, between carapace length and the following variables (p > 0.05 in all tests): 1) prey number; 2) number of prey categories (Fig. 2); 3) niche breadth; 4) maximum prey length (Fig. 3); or 5) mean prey length (Fig. 3).
Citation: Chelonian Conservation and Biology 10, 1; 10.2744/CCB-0846.1
Citation: Chelonian Conservation and Biology 10, 1; 10.2744/CCB-0846.1
DISCUSSION
Aquatic insect larvae (e.g., Ephemeroptera, Odonata, Plecoptera, Trichoptera, Diptera) are the main food source of carnivorous freshwater turtles (Moll 1990; Souza and Abe 1995; Kennet and Tory 1996; Allanson and Georges 1999). Indeed, nymphs of Odonata were the primary food of A. spixii. These nymphs employ diverse mechanisms to avoid predator detection: nymphs of Gomphidae bury themselves in the mud, whereas those of Libellulidae sprawl among the sediments; in Lestidae, Cenagrionidae, and Aeshnidae, nymphs climb vascular plants (Merrit and Cummins 1996). Some aquatic larvae and pupae of Diptera also are associated with vascular plants, others build channels among the sediments, and some are free-swimmers (Merrit and Cummins 1996). When insect larvae associated with plants are captured, turtles can accidentally ingest a small quantity of plant material. Nymphs of Odonata that climb vegetation were more prominent in female stomach contents, which could explain the higher proportion of plant material in the diet of females.
Because A. spixii feeds primarily in the water, the presence of allochthonous prey in the diet indicates opportunism, which is common among aquatic turtles (Moll 1976; Chessman 1984; Tomas et al. 2001). The occurrence of intact, large prey, such as frogs and nymphs of Aeshnidae, suggests that A. spixii uses suction feeding, which consists of negative buccopharyngeal pressure caused by volume expansion of the oral cavity (Pough 2004). This mechanism, combined with a sudden fast dart of the head, allows active predation in freshwater turtles by compensating for their reduced mobility and heavy shell (Aerts et al. 2001; Lemell et al. 2002).
Ontogenetic shifts in diet composition, from carnivorous juveniles to herbivorous adults, are common among omnivorous turtles (e.g., Parmenter and Avery 1990). We observed no such shift in the diet of A. spixii; likewise, Souza and Abe (1995) found no such shift in Hydromedusa maximiliani, which is also primarily carnivorous. Conversely, in Trachemys scripta, one of the best-studied freshwater turtle (Gibbons 1990), ontogenetic shifts in diet are known to occur (Clark and Gibbons 1969; Hart 1983), even though its diet is primarily carnivorous (Parmenter 1980; Parmenter and Avery 1990). The ontogenetic shift from carnivory to herbivory is assumed to result from 1) abundance of plants, 2) facility of ingesting plants by adults, 3) lower mass-specific nutrient requirements of adults, and 4) changes in foraging habitat, with adults using deeper water with lower densities of animal prey (Parmenter and Avery 1990). In habitats with large quantities of animal prey, this shift should be reduced or should not occur at all (Parmenter 1980). At our study site, food resources are abundant, especially Odonata nymphs, the main food resource of A. spixii. This can be the reason for the lack of ontogenetic variation in their diet.
Males of A. spixii at PNB use more the margins of the ponds, whereas females use more the central regions (GFH, unpubl. data). This could explain the greater importance of frogs and tadpoles, more common at the margins (Barreto and Moreira 1996), in the diet of males, and the greater importance of plant material, more concentrated in the center of ponds (GFH, unpubl. data), in the diet of females. A similar pattern was observed in Apalone mutica, where males used more the periphery and females used more the center of water bodies, resulting in intersexual differences in diet composition (Plummer and Farrar 1981) related to reduced exposure to predation in deeper waters and, therein, greater occurrence of females. Alternatively, Hart (1983) observed that small individuals of T. scripta frequently used shallow waters, whereas large individuals selected deeper waters. Similarly, larger individuals of H. maximiliani occupy deeper waters, whereas smaller individuals remained next to banks in slow and shallow waters (Souza and Abe 1998).
We observed no sexual dimorphism in mean niche breadth or in mean number of prey categories. Moreover, the most important prey categories were the same for both sexes, and niche overlap was very high, not differing from the expected by chance. Although some prey categories occurred in only one sex, they were not very important (IVIs ≤ 11.60). Most of these prey categories were found only in males, which was expected since there were more stomach contents from this sex. However, total estimated richness was the same for both sexes. These results indicate that males use the same prey categories as females, despite differences in habitat use.
Although we found no difference between mean diet overlap among individuals of the same sex and mean diet overlap between individuals of opposite sexes in A. spixii, mean overlap was lower among males, indicating higher interindividual variation in this sex. This could result from higher individual specialization in males, which can be a strategy to avoid competition (Van Valen 1965; Roughgarden 1972; Polis 1984). Nonetheless, resources must be limited for competition to exist, which does not seem to be the case at the study site, where nymphs of Odonata, anurans, and aquatic insects are abundant. The lower overlap among males is probably a consequence of wider microhabitat use. There are more types of microhabitats at the margins than at the center of ponds, where the concentration of macrophytes is higher. In addition, allochthonous prey are probably more common at the margins. Therefore, there may be a higher diversity of prey at the margins of ponds, leading to lower diet overlap among males. The higher capture rate and lower interindividual diet overlap among males may result from higher movement rates or larger home range in relation to females, as observed in other freshwater turtles (Graham 1995; Chelazzi et al. 2007).
Prey length did not vary between sexes, despite females of A. spixii being larger than males (Fraxe Neto et al. 2011). This lack of difference can occur when larger individuals continue feeding on smaller prey (Fig. 3). Larger animals tend to ingest larger prey, except when smaller prey are abundant (Schoener 1974). Moreover, there were few stomach contents from juveniles in this study. The same explanations fit the lack of correlation between prey length and turtle carapace length. Previous studies have shown that turtle size is directly related to prey size and inversely related to number of ingested prey (Moll 1976; Georges 1982; Chessman 1983; Parmenter and Avery 1990; Souza and Abe 1998). In our study, prey number did not vary between sexes, nor was it correlated to carapace length. In Phrynops geoffroanus, these relations also were lacking, which was interpreted as a consequence of the observed diet homogeneity in this species (Souza and Abe 2000). Acanthochelys spixii has a diet composed basically of Odonata nymphs, especially Libellulidae, which probably determined the lack of these relations.
Turtles are ectothermic animals with slow growth, late sexual maturity, and great longevity (Zug et al. 2001). The high incidence of empty stomachs in A. spixii was unrelated to sex, size, season, or capture method. This can be caused by low metabolism or the presence of tissue stores, such as fat, that allow animals to be active in a negative energy balance. Huey et al. (2001) showed that nocturnal lizards had higher incidence of empty stomachs than diurnal lizards (24.1 vs. 10.5%), which may reflect the difficulty in detecting prey in dim light or reduced activity of insect prey at night. Acanthochelys spixii is nocturnal (Lema and Ferreira 1990), but prey detection in turtles also can be aided by chemoreceptors or mechanoreceptors (Zug et al. 2001), which would not be impaired by dim light. Acanthochelys macrocephala was observed foraging actively even in complete darkness (Vinke and Vinke 2001). Moreover, Odonata nymphs, the main food of A. spixii at the study site, emerge at night (Merrit and Cummins 1996). Therefore, the high incidence of empty stomachs is probably a consequence of life-history characteristics of these animals. To the best of our knowledge, there are no studies that emphasize the incidence of empty stomachs in turtles.
Acanthochelys spixii is totally dependent on the aquatic environment, whose integrity is extremely important to the life history of this species. Aquatic insects often have aquatic respiration, being extremely vulnerable to low environmental quality (Merrit and Cummins 1996). Plecoptera nymphs need high quantities of dissolved oxygen and inhabit places with high microhabitat diversity, thus being used as bioindicators of habitat quality (Goulart and Callisto 2003). Immature plecopterans were found in the diet of A. spixii, which indicates the good quality of the studied habitats and that PNB has been complying with the goal of preserving Cerrado freshwater diversity. However, the conversion of Cerrado areas into agroecosystems has severely altered the species composition, functioning, and structure of freshwater habitats in this biome (Agostinho et al. 2005; Wantzen et al. 2006), with impacts on benthic communities (Wantzen 2006). The known distribution of the northernmost populations of A. spixii is in the southern portion of Cerrado, in the Paraná river basin, the most heavily impacted freshwater system in Brazil (Langeani et al. 2009), whose headwaters are confined to the most severely fragmented region of Cerrado, close to large urban centers (Cavalcanti and Joly 2002; Klink and Machado 2005). Odonata nymphs, the main food of A. spixii at PNB, also are vulnerable to pollution and siltation (Corbet 1980; Watson et al. 1982). This suggests that the expansion of agricultural activities, causing high rates of habitat loss and fragmentation in Cerrado, can bring risks to populations of A. spixii and associated freshwater communities.
Rarefaction curves of the number of prey categories in stomach contents of Acanthochelys spixii collected at Parque Nacional de Brasília. Curves were calculated for pooled contents and separately for each sex.
Relationship between total number of prey categories and turtle size (carapace length) in Acanthochelys spixii collected at Parque Nacional de Brasília.
Relationship between prey length (maximum and medium) and turtle size (carapace length) in Acanthochelys spixii collected at Parque Nacional de Brasília.
