A comprehensive understanding of the natural history and ecology of imperiled taxa is essential for formulating effective recovery plans (Clark et al. 2002). One such aspect of a taxon's natural history is diet. Knowing and understanding the dietary habits of imperiled taxa is required if wildlife agencies wish to manage habitat in a way that ensures the perpetuity of nutritional resources for the species of interest both in natural and captive settings. For example, such knowledge is useful for identifying suitable habitat worthy of conservation (Fang et al. 2011; Carrión-Cortez et al. 2013) as well as for guiding the development of feeding regimes for captive assurance colonies (Kuchling and DeJose 1989; Gibson and Buley 2004).

The Rio Grande cooter (Pseudemys gorzugi) is a medium to large freshwater turtle native to the lower Rio Grande/Rio Bravo del Norte basin and its tributaries (Degenhardt et al. 1996; Dixon 2013; Rhodin et al. 2017). This turtle is classified as a species of greatest conservation need in Texas, listed as threatened in both New Mexico (New Mexico Department of Game and Fish 2018) and Mexico (Secretaría de Medio Ambiente y Recursos Naturales 2010), and is considered near threatened by the International Union for Conservation of Nature (Rhodin et al. 2017). Threats to the taxon include habitat degradation (Pierce et al. 2016), commercial collection (Bailey et al. 2008; Dixon 2013), roadway mortality (Walls 1996), recreational fishing (Waldon et al. 2017; Bassett et al. 2020), and wanton shooting (Christman and Kamees 2007; Suriyamongkol et al. 2019). Despite its conservation status, few studies have been conducted on P. gorzugi (Lovich and Ennen 2013). There are substantial concerns about the viability of extant populations (Pierce et al. 2016), but only 2 in-depth investigations of P. gorzugi diet have been published to date (Letter et al. 2019; Suriyamongkol et al. 2022).

The earliest report of P. gorzugi diet comes from Legler (1958), who documented the presence of vegetation in the stomach of a specimen collected from Blue Spring, New Mexico. In agreement with this account was Painter's (1993) observation of P. gorzugi eating the leaves of a cottonwood tree (Populus sp.) in the Black River of New Mexico. At the same drainage, Degenhardt et al. (1996) noted that P. gorzugi were eating the algae from the surface of submerged rocks and that a single specimen had consumed a crayfish. Recently, Letter et al. (2019) conducted the first in-depth study on the dietary habits of P. gorzugi by means of fecal content analysis in the Black River, New Mexico. Dicot vegetation, filamentous algae, and arthropods made up the greatest proportion of male, female, and juvenile diets, respectively. Other materials found in fecal samples included willow (Salix sp.), crustaceans, a feather, a fish vertebra, and monofilament fishing line.

More recently, Suriyamongkol et al. (2022) reconstructed the diet of P. gorzugi at 2 sites on the Black River of southeastern New Mexico using stable isotopes. The results of their study largely reflected the findings of Letter et al. (2019); for example, at their upstream site, netleaf hackberry (Celtis laevigata; seeds and leaves) constituted the majority of P. gorzugi diet (34%), whereas filamentous algae (20%) and cottonwood (19%) were the most important dietary items at the downstream site. They also compared the isotopic niche of P. gorzugi with the co-occurring red-eared slider (Trachemys scripta elegans) in the Black River and Berrendo Creek. At the Black River, the 2 species showed differing δ¹³C signatures, suggesting utilization of different basal dietary sources. At Berrendo Creek, T. s. elegans had an enriched δ¹⁵N signature, suggesting a more carnivorous diet relative to P. gorzugi—unsurprising given the larger amounts of animal matter typically found in stomach and fecal contents of T. scripta (Parmenter 1980; Dreslik 1999; Lindeman 2000; Bassett and Forstner 2021).

However, it has been routinely demonstrated that vegetation is often an important component of T. scripta diet throughout the known distribution of the species (Schubauer and Parmenter 1981; Pérez-Santigosa et al. 2011; Works and Olson 2018; Stephens and Ryan 2019). As T. scripta grow, plant matter generally constitutes a greater proportion of examined stomach and fecal contents (Clark and Gibbons 1969; Hart 1983). The inclusion of plant matter in the diet suggests that there is potential for a competitive interaction between T. s. elegans and P. gorzugi where the 2 species co-occur. Stable isotope ratios from animal tissue allow for a quantitative estimate of niche overlap between syntopic turtle species and can partially explain if and how species partition available resources (Lara et al. 2012; Balzani et al. 2016; Micheli-Campbell et al. 2017). Because the degree of isotopically defined niche overlap between chelonian taxa can vary across sites (Pearson et al. 2013; Suriyamongkol et al. 2022), investigations of the degree of niche overlap should be replicated throughout the known distribution of relevant species to develop a more complete characterization of resource use, the intensity of competition, and potential resource partitioning.

During the summer and fall of 2020, we studied the dietary habits and isotopic niche overlap of P. gorzugi and syntopic T. s. elegans at San Felipe Creek, Texas, USA—a portion of their distribution where dietary habits had not been previously examined. We used fecal content analysis and complementary stable isotope data to identify the most important food items. Such information can be used by management agencies to recognize suitable habitat worthy of conservation and by zoos and nonprofits to inform feeding regimes for captive assurance colonies. Our results also supplement the existing literature by characterizing resource partitioning between P. gorzugi and T. s. elegans in a unique habitat.

METHODS

Study Site and Turtle Capture. — We sampled P. gorzugi from 10 May to 2 October 2020 at San Felipe Creek in Del Rio, Texas. San Felipe Creek is a spring-fed creek located at the interface of the Chihuahuan, Tamaulipan, and Kansan biotic provinces of Texas (Blair 1950). The creek feeds into the Rio Grande at the Texas–Mexico border, is relatively isothermic (López-Fernández and Winemiller 2005), and has high water clarity. We captured turtles by hand while snorkeling a 1.35-km section of the creek from the Highway 90 overpass to the Johnson Street overpass (Fig. 1). Along this snorkel route, the creek margins are heavily vegetated with the nonnative giant reed (Arundo donax) while pecan (Carya illinoensis) and American sycamore (Platanus occidentalis) trees dominate the upper canopy. This section of the creek has been thoroughly modified by humans, including 2 impoundments and extensive channelization. Immediately adjacent to creek margins are numerous residential and commercial properties as well as public parks.

View Figure

• Download as Powerpoint
• Download Figure

When a turtle was captured, we measured midline carapace length and midline (minimum) plastron length (Iverson and Lewis 2018) using 80-cm Haglöf tree calipers. We photographed turtles from both a dorsal and ventral point of view using an iPhone 8. We marked all turtles by inserting a passive integrated transponder (PIT) tag (AVID Identification Systems, Inc.) into the anterior-most portion of the right hind limb (Roosenburg and Burke 2018). We also shell-notched turtles (Cagle 1939) in cohort based on year of capture with a Dremel tool equipped with a tile cutting bit.

Fecal Content Analysis. — Captured P. gorzugi were individually placed into 23-l plastic tubs with a small volume of water to maintain thermal stability and hydration. Water in the tubs was sourced from San Felipe Creek and was poured through a 1.24-mm sieve. Turtles were kept in tubs for up to 48 hrs or until they defecated—whichever came first. Water temperature was monitored once every 2 hrs to ensure that turtles did not perish from heat exposure. Fresh, sieved water from San Felipe Creek was added as necessary to maintain a water temperature between 19°C and 27°C. During the day, turtles were kept in a shaded area at the field site. If kept overnight, turtles were relocated to an air-conditioned building located approximately 3 km from the study site. Once a turtle defecated, it was released at the site of capture. Turtles that did not defecate after 48 hrs were likewise released at the site of capture. Tub contents were poured through a sieve with 1.24-mm mesh to retain fecal material. Solids in the sieve were transferred into 50-ml Falcon tubes with 95% ethanol.

Fecal material was examined in the laboratory with a Leica EZ4 dissection microscope (magnification range: ×8–×35) and identified and sorted to the lowest practical taxonomic level. Sorted samples were dried using vacuum filtration and a Buchner funnel. Samples were additionally patted dry with paper towels to ensure adherent ethanol did not influence volumetric measurements. The volume of each fecal content item category was determined using volumetric displacement.

The percent occurrence (%F) of each dietary component found in the feces of P. gorzugi was calculated. Percent occurrence is defined as the number of samples in which a dietary component is found divided by the total number of samples. Dietary diversity of sex classes (juveniles, males, and females) was calculated using the Shannon-Wiener measure (H′) of dietary diversity: H′ = –Σ pj log pj, where pj is the proportion of turtles consuming resource j. Turtles with carapace lengths ≤ 118 mm were classified as juveniles due to incompletely developed secondary sexual characteristics—specifically, long tails and foreclaws in male specimens (Readel et al. 2008; Gradela et al. 2017). This classification is consistent with Letter et al. (2019) to allow direct comparison between the 2 studies. Turtles with a carapace length > 118 mm were sexed by examining the aforementioned secondary sexual characteristics. The niche breadth index (H′) was then standardized on a scale of 0 to 1 by measuring evenness (J′) with the following equation: J′ = H′/(logn), where n is the total number of prey groups. To determine the importance of food items relative to turtle age and sex, we calculated the index of relative importance (IRI) as follows: IRI = V_iF_i/ΣV_iF_i, where V_i is the average percent volume of dietary component i measured across all samples and F_i is the percent occurrence of dietary component i.

Stable Isotope Analysis. — For every P. gorzugi and T. s. elegans that was captured, the terminal 0.5 cm of the medial toe claw on either hind foot was clipped using sterilized dissection scissors. Claws were stored individually in 1.5-ml Nunc tubes and kept on ice in the field. Samples were transferred to a –18°C freezer within 8 hrs and thereafter transferred to a –80°C freezer within 5 d. The following putative food items were targeted for sampling: riparian and aquatic vegetation, aquatic filamentous algae, and aquatic invertebrates. Plant and animal taxa that were visually most abundant in the environment were prioritized for sampling. Invertebrates and plants were placed in plastic tubes or Ziploc bags and kept on ice while in the field. Samples were transferred to a –18°C freezer within 8 hrs and thereafter transferred to a –80°C freezer within 5 d. Aquatic invertebrates were collected by hand and by sweeping a 2-mm mesh aperture dipnet through aquatic vegetation. Filamentous algae, riparian vegetation, and aquatic vegetation (macrophytes) were collected by hand. Only riparian plants immediately adjacent to, or physically overhanging, the creek were collected.

Aquatic plant samples and filamentous algae were rinsed with Milli-Q water to remove detritus and inorganic matter. Samples were examined with a dissecting microscope to remove aquatic invertebrates. Samples were dried at 60°C for at least 48 hrs and then placed in a fuming HCl chamber for 4–8 hrs to remove inorganic carbon. Samples were then dried for a minimum of 48 hrs at 60°C and then ground into a fine powder using a clean pestle and mortar. Riparian plant samples were rinsed with Milli-Q water to remove inorganic matter, dried for a minimum of 48 hrs at 60°C, and then ground into a fine powder using a pestle and mortar.

Gastropod and crayfish samples were dissected to remove foot and tail tissue, respectively. Muscle was dried at 60°C for a minimum of 48 hrs and ground into a fine powder using pestle and mortar. Turtle claws were scrubbed using a clean nylon brush and Milli-Q water to remove dirt and debris, dried at 60°C for a minimum of 48 hrs, and ground into a coarse powder using pestle and mortar. Lipids were extracted from claw material in 2-ml Nunc tubes with triplicate rinses of 2:1 chloroform:methanol and a final wash with Milli-Q water. Claw material was subsequently dried for a minimum of 48 hrs at 60°C.

All samples were encapsulated into tin capsules for analysis of δ¹⁵N and δ¹³C abundances and analyzed at the University of California at Davis Stable Isotope Facility using a PDZ Europa ANCA-GSL elemental analyzer interfaced to a PDZ Europa 20-20 isotope ratio mass spectrometer (Sercon Ltd., Cheshire, UK). Stable isotope values are reported as δ¹³C and δ¹⁵N in the sample relative to the standards of Vienna Pee Dee Belemnite for δ¹³C and atmospheric air for δ¹⁵N. Standard deviation relative to standards for measurement of δ¹³C was ± 0.2‰ and δ¹⁵N was ± 0.3‰.

We conducted all statistical analyses in Program R (R Core Team 2021). Linear regressions were conducted using midline carapace length (Iverson and Lewis 2018) as the explanatory variable and either δ¹³C or δ¹⁵N as the response variable to test for ontogenetic dietary shifts in P. gorzugi and T. s. elegans. We used randomization tests (replication = 10,000) to assess taxon (T. s. elegans vs. P. gorzugi) and sex (within each species) δ¹³C and δ¹⁵N differences. We used the package “MixSIAR” (Stock et al. 2018) to perform dietary mixing models using δ¹³C and δ¹⁵N data to determine the proportional contribution of food sources to the diet of P. gorzugi and T. s. elegans. We used a trophic enrichment factor of 2.7‰ ± 0.3 for δ¹⁵N (Seminoff et al. 2007) and 0.23‰ for δ¹³C (Aresco 2005). Because mixing models lose accuracy when important food sources are omitted, and models have an inability to differentiate among sources when the number of sources exceeds 5 (Phillips et al. 2014), fecal content data from this study and others (Thomas 1993; Bassett and Forstner 2021; L.G.B., unpubl. data, 2020) were used to a priori identify potential dietary sources included in the model. Of the food items for which we had isotope data, we selected those that had the highest average percent volume in turtle feces as sources. For P. gorzugi, this included filamentous algae, red swamp crayfish (Procambarus clarkii), and Myriophyllum sp. (Table 2). Filamentous algae were split into 2 sources: lentic filamentous algae and lotic filamentous algae. These were treated as unique sources because their mean δ¹³C signatures differed significantly (t = 19.483; df = 7.0073; p < 0.05). For T. s. elegans, this included filamentous algae (lentic and lotic), P. clarkii, and red-rimmed melania (Melanoides tuberculata) (Thomas 1993; Bassett and Forstner 2021; L.G.B., unpubl. data, 2020). Food items with a low average percent volume were omitted from the models. Mixing models were run with 3 Markov chains, each with 3 million iterations and a burn-in of 1.5 million iterations. Both models used the default uninformative prior α = (1, 1, 1, 1). Model convergence was assessed with the Gelman diagnostic ( < 1.05).

Table 2. Average percent volume of dietary items found in feces collected from male, female, and juvenile Pseudemys gorzugi. Asterisks (*) indicate food items which have not been previously reported in the diet of P. gorzugi. Daggers (†) indicate food items which are not native constituents of the flora and fauna of San Felipe Creek. Any turtle with a carapace length ≤ 118 mm was classified as a juvenile.

View Fullscreen

The trophic positions of P. gorzugi and T. s. elegans were estimated with a Bayesian approach using the R package “tRophicPosition” (Quezada-Romegialli et al. 2018). The model “oneBaseline” was utilized with filamentous algae designated as the baseline. Filamentous algae (lotic and lentic combined) were chosen as the basal source because they were one of the main items used in the dietary mixing models and constituted a substantial portion of turtle diets. The model was run with 20,000 iterations and 2 Markov chains. A trophic enrichment factor of 2.7‰ ± 0.3 was used for δ¹⁵N (Seminoff et al. 2007). The R package “nicheROVER” (Swanson et al. 2015) was used to estimate the percent of isotopic niche space overlap between P. gorzugi and T. s. elegans. The niche space of a group is defined as an area in isotope space where each individual of that group has a probability α of being found. Percent overlap is the percent probability that an individual belonging to Group A is found in the niche space of Group B (and vice versa). Niche overlap estimates were generated with 10,000 Monte Carlo draws with α = 0.95.

RESULTS

We captured a total of 35 individual turtles, including 22 P. gorzugi and 13 T. s. elegans. One P. gorzugi was captured twice, 1 T. s. elegans was captured 3 times, and 1 T. s. elegans was captured 2 times. Of the P. gorzugi fecal samples (n = 20), all were collected from unique individuals, except for 2 samples that were collected from the same female. Three of the 22 P. gorzugi captured did not defecate during the 48-hr holding period and therefore no sample material was obtained. We collected claw clips from all captured turtles; however, a malfunction of the mass spectrometer resulted in the loss of 6 samples with no data returned. We obtained isotopic data for 7 male, 10 female, and 1 juvenile P. gorzugi. For T. s. elegans, we obtained data from 3 males, 7 females, and 1 adult that we forgot to examine for secondary sexual characteristics.

Fecal Content Analysis. — Dissection of P. gorzugi feces identified 13 food items that had not been previously reported in the diet of this species (Table 1). These food items included 7 dicot plant taxa, bryophyte matter, and 5 animal taxa (Table 1). The most important dietary items for P. gorzugi based on frequency of occurrence (Table 1) were filamentous algae (100%), unidentified plant matter (90%), and dicot vegetation (85%). The most important dietary items based on average percent volume (Table 2) were filamentous algae (41.48%), unidentified plant matter (36.70%), and arthropods (5.83%). Average percent volume of filamentous algae ranged from 38.73% (in juveniles) to 45.89% (in males). Average percent volume of arthropods ranged from 0% (in males) to 9.25% (in juveniles). Indices of relative importance indicated that nonalgal autotrophs (i.e., macrophytes and terrestrial plants) and filamentous algae were the most important food items for male, female, and juvenile P. gorzugi (Table 3). Arthropods had a low IRI for juveniles (0.09), females (0.07), and males (0). Diversity (H′) and evenness (J′) scores slightly varied between sex classes, with juveniles having the lowest scores (H′ = 0.09; J′ = 0.2) and males having the highest scores (H′ = 0.16; J′ = 0.33).

Table 1. Frequency of occurrence of dietary items found in feces collected from male, female, and juvenile Pseudemys gorzugi. Asterisks (*) indicate food items that have not been previously reported in the diet of P. gorzugi. Daggers (†) indicate food items that are not native constituents of the flora and fauna of San Felipe Creek. Any turtle with a carapace length ≤ 118 mm was classified as a juvenile.

View Fullscreen

Table 3. Indices of relative importance (IRIs) for major dietary resources found in the feces of Pseudemys gorzugi collected at San Felipe Creek during summer and fall of 2020. Shannon-Weiner measures of niche breadth including diversity (H′) and evenness (J′) are provided in the lower 2 rows for the taxon and its sex classes.

View Fullscreen

Stable Isotope Analysis. — All linear regressions involving turtle carapace length and stable isotope signatures returned nonsignificant results (p > 0.05). All randomization tests comparing δ¹³C and δ¹⁵N values across sex classes within each of the 2 species also returned nonsignificant results. However, the randomization test comparing δ¹³C and δ¹⁵N values between species returned significant results (p < 0.05). Both the mean δ¹³C and δ¹⁵N values for T. s. elegans were generally enriched relative to those for P. gorzugi (Figs. 2 and 3). The dietary mixing model for P. gorzugi estimated the mean (± 95% CI) proportional contribution of the food sources as follows: 48.7% (40%–55.9%) for lotic filamentous algae, 25.2% (15.9%–34.3%) for lentic filamentous algae, 12% (0.8%–31.3%) for Myriophyllum sp., and 14.1% (0.6%–29.5%) for P. clarkii. The dietary mixing model for T. s. elegans indicated that lotic algae only contributed 13% (1.5%–26.2%) to the diet, whereas lentic filamentous algae contributed 31% (10.9%–51.2%), nonnative M. tuberculata contributed 30.4% (2.1%–61.3%), and P. clarkii contributed 25.5% (1.3%–61.4%) to the diet. Mean (95% Bayesian credible interval; CI) trophic position estimates for P. gorzugi and T. s. elegans were 2.919 (2.379–3.517) and 3.824 (3.133–4.544), respectively. The mean probability (± 95% CI) that an individual P. gorzugi would be found within the isotopic niche of T. s. elegans was 25.42% (1%–92%). The mean probability that an individual T. s. elegans would be found within the isotopic niche space of P. gorzugi was 4.27% (0%–16%).

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

DISCUSSION

Results of fecal content and stable isotope analyses indicate that P. gorzugi at San Felipe Creek are primarily algivorous. Volumetrically, filamentous algae were the most important food item for all P. gorzugi sex classes (Table 2). This finding is not concordant with the results from Letter et al. (2019) that reported an average percent volume of only 2.7% and 16% for filamentous algae in the feces of juvenile and male P. gorzugi, respectively. In contrast, the present study found an average percent volume of 38.73% and 45.89% for algae in the feces of juvenile and male P. gorzugi, respectively (Table 2). Therefore, algae are probably a more important dietary resource for the P. gorzugi population inhabiting San Felipe Creek than the population inhabiting the Black River in New Mexico. Algivory by P. gorzugi in San Felipe Creek was further supported by our stable isotope data, in that our dietary mixing model estimated that both lotic and lentic filamentous algae were the largest contributors to P. gorzugi diet, cumulatively accounting for > 70% of the diet. In contrast, Suriyamongkol et al. (2022) also performed stable isotope dietary mixing models and found that filamentous algae only constituted ∼ 8% and ∼ 20% of P. gorzugi diet at the upstream and downstream portions of the Black River, respectively. The difference in algae consumption between the Black River and San Felipe Creek may be driven by differential availability of algae. For example, Letter (2018) estimated percent occurrence of algae at his Black River study area to be 47%, whereas the percent occurrence of algae at the San Felipe Creek study area was 100% (Bassett 2021). This may be due to differences in turbidity between the 2 sites, as the Black River is more turbid than San Felipe Creek (Letter 2018; Bassett 2021), and high turbidity leads to light limitation and decreased algal growth (Wang 1974). The 2 sites also differ in regard to flow velocity. Although flow velocity at our San Felipe Creek study area varied spatially, with some portions lotic and others lentic, it was generally a more-lotic system than the area of the Black River sampled by Letter (2018). Flow velocity can improve algal growth by delivering necessary nutrients but can also constrain algal growth due to drag (Horner and Welch 1981; Stevenson 1996; Biggs et al. 1998).

In the present study, other food items including monocot vegetation, aquatic dicots, and arthropods were frequently consumed by P. gorzugi (Table 1), but in relatively small volumes (Table 2). Ingestion of this animal material by P. gorzugi may oftentimes be incidental while they graze on vegetation or algae. For example, amphipods, naucorids, and M. tuberculata are often found amongst filamentous algae and aquatic plants (e.g., Carolina fanwort [Cabomba caroliniana], coontail [Ceratophyllum demersum], and milfoil [Myriophyllum sp.]) that occur at San Felipe Creek (L.G.B., pers. obs.), which may lead to incidental ingestion. However, it is also possible that vegetation is being incidentally ingested while turtles target arthropods for consumption. Some instances of predation on animal taxa are most likely cases of targeted consumption. For example, lepidopterans and formicids found in the feces of P. gorzugi probably represent instances where insects fell onto the water and were opportunistically consumed by P. gorzugi. Similar evidence for the opportunistic consumption of animal prey by P. gorzugi exists from southeastern New Mexico (Letter et al. 2019; Mahan et al. 2020). However, the overall conclusion is that P. gorzugi at San Felipe Creek are primarily algivorous chelonians that incidentally and opportunistically supplement their diet with invertebrate prey.

Given that the composition of plant and animal species at San Felipe Creek are considerably different from those of the Black River drainage (Letter 2018; Bassett 2021), and that these 2 sites occupy unique biotic provinces (Blair 1950), we expected to find differences in the diet of P. gorzugi between these 2 sites. However, we anticipated that such differences would manifest as dietary constituents of differing taxonomic identities rather than differences in the primary trophic ecology of the turtle (e.g., algivory vs. herbivory vs. carnivory). Our indices of relative importance (Table 3) indicate that all P. gorzugi sex classes at San Felipe Creek are primarily algivorous and herbivorous, whereas Letter et al. (2019) found that males, females, and juveniles in the Black River were primarily herbivorous, algivorous, and carnivorous, respectively. These differences demonstrate how the dietary habits of the same species can vary substantially across its distribution and underline the importance of replicating dietary studies across the known distribution of chelonian taxa. Such differences are likely dependent upon the local abiotic conditions and the relative availability of different food sources.

Of all extant Pseudemys taxa, P. gorzugi has the westernmost native distribution followed by the Texas cooter (Pseudemys texana) and then the river cooter (Pseudemys concinna) eastward (Ernst and Lovich 2009; Powell et al. 2016; Rhodin et al. 2017). Our results are similar to those of previous studies that have investigated the dietary habits of P. texana and P. concinna. For example, Lindeman (2001) and Fields et al. (2003) both examined stomach and fecal contents of P. texana and found that samples were primarily composed of vegetation and filamentous algae. Additionally, many of the food items identified in the feces of P. gorzugi from San Felipe Creek have also been reported in the diet of P. texana including filamentous algae, C. caroliniana, C. demersum, Myriophyllum, and crayfish (Lindeman 2001; Fields et al. 2003; Ernst and Lovich 2009). Prior studies have likewise described P. concinna as primarily algivorous or herbivorous (Lagueux et al. 1995; Dreslik 1999; Lindeman 2000). For example, Lindeman (2000) found that algae accounted for > 99% of the ash-free dry mass of P. concinna fecal samples. Many of the food items identified in the feces of P. gorzugi from San Felipe Creek have also been reported in the diet of P. concinna including filamentous algae, crayfish, gastropods, hymenopterans, grasses, and C. demersum (Cahn 1937; Parker 1939; Buhlmann and Vaughan 1991; Lagueux et al. 1995; Dreslik 1999; Lindeman 2000). Our results therefore serve as additional evidence indicating that members of the genus Pseudemys primarily rely on algae and vegetation to fulfill bioenergetic demands.

Isotopic data for P. gorzugi and T. s. elegans indicated that each species in San Felipe Creek occupied largely unique niche spaces and had limited niche overlap. Trachemys s. elegans had δ¹³C and δ¹⁵N values that were significantly enriched relative to those for P. gorzugi (p < 0.05; Fig. 2), suggesting that T. s. elegans uses different basal dietary sources and has a more-carnivorous diet (Fry 2006). The possibility of T. s. elegans being more carnivorous is additionally supported by our trophic position estimates which ranked T. s. elegans (trophic position = 3.8) approximately 1 trophic level above P. gorzugi (2.9). Average estimates of isotopic niche overlap between T. s. elegans and P. gorzugi were small (∼ 4% and ∼ 25%), meaning that there is likely little competition occurring between these 2 chelonian taxa at San Felipe Creek. Our dietary mixing models suggest that T. s. elegans and P. gorzugi both consume filamentous algae and P. clarkii, but to differing degrees. For example, while P. gorzugi mainly feeds on lotic filamentous algae, T. s. elegans consumes more lentic filamentous algae. Additionally, the estimated proportional contribution of P. clarkii to P. gorzugi diet (14.1%) was much smaller than for T. s. elegans diet (25.5%). The 95% confidence ellipse for T. s. elegans in bivariate isospace was also considerably larger than for P. gorzugi (Fig. 2), indicating that T. s. elegans has a less specialized diet. Results of the current study stand in contrast to those of Suriyamongkol et al. (2022) in that the T. s. elegans population at San Felipe Creek had enriched values for both δ¹³C and δ¹⁵N relative to syntopic P. gorzugi (Fig. 2). At Berrendo Creek, Suriyamongkol et al. (2022) found that T. s. elegans were enriched for δ¹⁵N relative to P. gorzugi. At the downstream portion of the Black River, P. gorzugi and T. s. elegans shared similar δ¹⁵N values, but T. s. elegans had enriched δ¹³C values. Therefore, even among only these 3 sites, there is considerable variability in the direction and degree of isotopic niche space shared by these 2 species. Such differences among riverine systems are likely a function of differential composition, abundance, and diversity of available prey taxa, as has been demonstrated with the northern red-bellied cooter (Pseudemys rubriventris) and introduced T. s. elegans in Pennsylvania, USA (Pearson et al. 2013).

Of all the T. s. elegans we captured at San Felipe Creek, only 1 individual was melanistic. Interestingly, that individual had δ¹³C and δ¹⁵N signatures that were dramatically different from its nonmelanistic counterparts (Fig. 2). Given the sample size, it is impossible to make generalizations about differences in the trophic niche of melanistic and nonmelanistic T. s. elegans, but the difference observed in the present data set is nonetheless intriguing. Melanism is a condition primarily exhibited by male T. s. elegans that are relatively old (Tucker et al. 1995; Hays and McBee 2009), and future work may investigate variation in the isotopic signature of male T. s. elegans across age classes and varying degrees of melanism using a larger sample size. This will determine whether melanism corresponds with a unique dietary niche.

Our dietary mixing model estimated that the proportional contribution of M. tuberculata to the diet of T. s. elegans was relatively large (∼ 30%). Bassett and Forstner (2021) likewise found evidence that M. tuberculata constitutes a substantial portion of T. s. elegans diet at San Felipe Creek. Melanoides tuberculata is a thiarid gastropod native to the Old-World tropics (Brown 1980; Albrecht et al. 2018) that has established populations throughout the United States, including Texas (Murray 1971; Dundee and Paine 1977; Karatayev et al. 2009). At San Felipe Creek, M. tuberculata can be found in syntopy with the native freshwater gastropod Elimia comalensis, with the former occurring in much greater abundance. An interesting topic for future investigation would be quantifying to what degree, if any, T. s. elegans provides biotic resistance against this invasive snail species. Other aquatic turtles known to utilize M. tuberculata as a dietary resource include the white-lipped mud turtle (Kinosternon leucostomum) (Ceballos et al. 2016), the diamond-backed terrapin (Malaclemys terrapin) (Outerbridge et al. 2017), and the eastern musk turtle (Sternotherus odoratus) (Morrison et al. 2017, 2019). Additionally, a series of reports involving a sundry assortment of freshwater turtles exist that further demonstrate the proclivity of some aquatic chelonians to prey upon other invasive mollusk taxa (Lindeman 2006; Bujes et al. 2007; Patterson and Lindeman 2009; Sterrett et al. 2020; Vučenović and Lindeman 2021). Despite the growing body of evidence that aquatic turtle populations can consume large quantities of invasive mollusks, there is generally a lack of further research to determine whether such predation provides significant top-down control of relevant mollusk populations. A study involving M. terrapin and the marine gastropod Littoraria irrorata on the Georgia, USA, coast suggests that such top-down control likely exists (Silliman and Bertness 2002). We argue that the potential biotic resistance provided by aquatic turtles against invasive species warrants further research attention and should be considered when evaluating the ecological costs associated with turtle declines globally.

The current study represents the first detailed examination of P. gorzugi diet in the Edwards Plateau ecoregion and demonstrates that chelonian diets can vary considerably within different habitats across the species' range. We therefore encourage researchers to replicate diet investigations across the known distributions of chelonian taxa so that a more complete understanding of spatial variation in feeding habits and resource selection can be achieved. The data presented here also contribute toward our collective understanding of P. gorzugi ecology and equips wildlife agencies with information useful for recognizing suitable habitat worthy of conservation (Fang et al. 2011; Carrión-Cortez et al. 2013). Finally, our data can help guide the development of feeding regimes for captive assurance colonies and inform habitat management strategy (Kuchling and DeJose 1989; Gibson and Buley 2004).