The Importance of Riparian and Aquatic Resources in the Diet of Dermatemys mawii in a Lentic System in Southern Mexico
ABSTRACT
The Central American river turtle (Dermatemys mawii) is one of the of the most critically endangered turtle species worldwide, as a result of habitat loss and heavy hunting for consumption. Research on D. mawii becomes more challenging as wild populations are disappearing, while basic ecological data are still required to inform conservation efforts. In this study, our aim was to analyze the diet of D. mawii based on sex, and compare the information provided by stomach flushing and fecal samples. We examined both types of samples from individuals captured in 2017 and 2018 in a lentic system in the Lacantun River Basin, Chiapas, Mexico. We identified 3 broad categories (aquatic macrophytes, riparian resources and invertebrates) and 12 different food items, of which four are new reports of vascular plants for the diet of this turtle. The principal items consumed by D. mawii were Pistia stratiotes (a free-floating macrophyte) and Ludwigia sp. (a riparian plant). A low index of relative importance (≤ 0.5) for invertebrates suggests that they may be ingested accidentally when feeding on plants. We did not detect significant sex differences in the relative contribution of food items between the two sampling methods; additionally, we found a large dietary overlap (Cλ ≥ 0.89) between sexes that indicates a similar use of resources in this lentic system. We found all 3 broad food categories in stomach flushing samples and only one (riparian resources) in fecal samples; however, we did not find significant differences in the number of food items. The use of both postingestion sampling methods provided a more comprehensive view of the diet of this herbivorous freshwater turtle.
Freshwater ecosystems are some of the most threatened ecosystems globally, primarily as a result of human alteration (e.g., impoundments, overfishing, loss of riparian vegetation, mining; Allan et al. 2021). The deterioration of these ecosystems has severely affected freshwater turtles, making them one of the most endangered groups of aquatic vertebrates (TTWG 2021).
Freshwater turtles may play different trophic roles as herbivores, carnivores, and omnivores in their communities (Iverson 1982). Turtles also play important ecological roles as seed dispersers, biological regulators of animal or plant communities, and metabolizers of organic matter, which in turn helps maintain water quality, nutrient cycling, and overall ecosystem functioning (Lovich et al. 2018). Thus, studying the dietary ecology of turtles is important because it can offer valuable insights into resource abundance, richness, and the potential impacts of human activities on freshwater ecosystems (Tucker et al. 2012; Cunha et al. 2020; Gacheny et al. 2021).
Historically, animal diet has been studied using direct observation in the field and by analyzing samples from stomach flushing or feces (postingestion samples; Luiselli and Amori 2016). Specifically, postingestion sampling enables the characterization of dietary patterns in relation to ontogenetic, sexual, seasonal, or habitat variations (Nielsen et al. 2018), thereby aiding in the prediction of responses to human disturbances (Arditi and Ginzburg 2012). However, dietary intake is difficult to assess, and any one method alone cannot perfectly assess the use of resources by consumers, whereas a combination of different analytical methods may help provide a more precise evaluation of trophic ecology (Major 1990; Caputo and Vogt 2008; Pincheira-Donoso 2008). For example, Caputo and Vogt (2008) conducted a study on the diet of Phrynops rufipes, using stomach flushing and fecal material samples. Their findings revealed that this turtle species is an opportunistic omnivore, consuming a larger quantity of aquatic animals than initially believed solely based on fecal sampling.
The Central American river turtle (Dermatemys mawii Gray 1847) is the only extant species within the family Dermatemydidae, and the largest (up to 60 cm straight-line carapace length) freshwater turtle in Mesoamerica where it is endemic to southern Mexico, Guatemala, and Belize (Legler and Vogt 2013; TTWG 2021). The destruction and degradation of natural habitats and the historical exploitation of the species for pets, food, and for medicinal markets, have led to a heavy decline in populations of D. mawii, currently categorized as one of the most critically endangered turtle species worldwide (TCC 2018). Generally, D. mawii inhabits large rivers with strong flow; however, they can also live in lentic and estuarine systems (Iverson and Mittermeier 1980). This turtle has been previously defined as totally aquatic, and herbivorous in all its life stages (Legler and Vogt 2013). Previous studies in lotic and lentic systems from Mexico and Belize showed that riparian plants constituted the most important food category for D. mawii (Moll 1989; Vogt and Flores-Villela 1992; Gil Alarcón 2008; Bishop et al. 2022). Animal food items were also found, but rarely, leading the authors to suggest that animal consumption may have been incidental (Moll 1989; Vogt and Flores-Villela 1992; Gil Alarcón 2008; Bishop et al. 2022). The state of Chiapas in southern Mexico is widely recognized as a critical area for the conservation of this endangered turtle (Vogt and Flores-Villela 1992; Guichard-Romero 2006; Legler and Vogt 2013). Despite this significance, to date little research has been carried out on the trophic ecology of D. mawii in this region (Vogt and Flores-Villela 1992). Furthermore, our understanding of potential sexual segregation in the diet of this species remains limited because this aspect has yet to be thoroughly investigated.
In this study we evaluated sex-related differences in the diet of D. mawii in a lentic system located in the Lacantun River Basin (LRB, Chiapas, Mexico) by comparing the information provided by stomach flushing and fecal samples. To our knowledge, this is the first study to assess sex-related differences in the diet of D. mawii using a combination of postingestion analysis in the LRB, and their ecological implications for aquatic ecosystems. The information provided by this study contributes to the understanding of its trophic ecology and can be useful for future conservation management programs of D. mawii in the region.
METHODS
Study Site. — The study site is located in the LRB, Chiapas, Mexico, within the Montes Azules Biosphere Reserve (Fig. 1). This basin has an approximate area of 9,700 km2, with the presence of lentic systems such as the Ocotal and Miramar lagoons as well as lotic systems including the Lacantun River (main tributary of the Usumacinta River), Miranda stream and Manzanares stream (Ramírez et al. 2022). Specifically, the study site is an oxbow lake (∼ 150 m above sea level) and located about 1 km from the Lacantun River (Fig. 1). The study site is surrounded by a tropical evergreen forest (Carabias et al. 2015), and some of the most common riparian species included Persicaria punctata, P. segetum, Ludwigia octovalvis, and L. peploides (Ochoa-Gaona et al. 2018). The climate is warm and humid, with an average annual temperature of 24°C, and two well-defined seasons: dry season from November to May (mean precipitation 106 mm), and rainy season from June to October (mean precipitation 415 mm; SMN 2022).
Citation: Chelonian Conservation and Biology: Celebrating 25 Years as the World’s Turtle and Tortoise Journal 22, 2; 10.2744/CCB-1577.1
The LRB has been identified as a priority area for conservation because of its high level of biological diversity (Ramírez et al. 2022). Particularly, the LRB hosts a great diversity of herpetofauna, including some taxa that have not been described (Muñoz-Alonso et al. 2018). However, this basin has continually been modified by changes in land use for livestock and agriculture (Vaca et al. 2012, 2019), affecting the wild populations of threatened species such as D. mawii (Legler and Vogt 2013; Muñoz-Alonso et al. 2018).
The populations of D. mawii in the LRB were monitored in previous studies (Vogt and Flores-Villela 1992; Guichard-Romero 2006; Muñoz-Alonso et al. 2018) and it was concluded that the main threats to their conservation were deforestation and the illegal extraction for trade. Recently the possible presence in the LRB of a lineage genetically different from other populations of D. mawii has been discussed (Martínez Gómez 2017). However, no long-term and standardized studies have been undertaken in the region to analyze current population trends, and no programs exist to protect the wild populations of D. mawii in the LRB (C.A. Guichard-Romero, pers. comm., July, 2023).
We do not present the coordinates of the study site herein because of the conservation context and the recommendation made by the Turtle Survival Alliance and International Union for Conservation of Nature Tortoise and Freshwater Turtle Specialist Group.
Turtle and Material Processing Protocol. — Turtles were collected in 2017 (August, September and December) and 2018 (April), using trammel nets (10 × 3 m, with a mesh of 15 × 10 cm). These nets were deployed for approximately 46 hrs each month. The nets were checked regularly (∼ every 2 hrs) to ensure the well-being of captured individuals and prevent damage, predation, or drowning. All captured turtles were marked using the shell-notch code system from Cagle (1939) and measured. For the purpose of this study, we only present body mass (BM; recorded using a spring scale, 0.01 kg accuracy) and the midline straight carapace length (CL) recorded using a camera (Canon model D70, Tokyo, Japan) and measured in tpsDig2 2.32 (Rohlf 2021).
In order to differentiate between the sexes, we took into account the following characteristics, according to Legler and Vogt (2013) and Ligon et al. (2019): 1) males exhibit a visibly longer and thicker tail compared with females; 2) in males, the cloaca is situated posterior to the marginal scutes, whereas in females it is positioned anteriorly; 3) the posterior notch of the anal shield is more pronounced in males than in females. Turtles ≤ 349.9 mm in CL were recorded as juveniles, and those ≥ 350 mm of CL as adults (Ligon et al. 2019).
We collected the material from stomach flushing following the method proposed by Legler (1977), along with fecal samples produced by freshly caught turtles. In both cases, the material obtained was preserved in 70% alcohol. For each sample, all food sources were separated, counted, and identified to the lowest possible taxonomic level. Additionally, the food items (i.e., the food source types, such as Pistia stratiotes, Ludwigia sp., nematoda, etc.) were assigned to 3 broad categories (i.e., aquatic macrophytes, riparian resources, and invertebrates). Volume was determined by fluid displacement using a graduated cylinder to the nearest 0.5 mL for the least abundant resources (e.g., invertebrates or seeds), and nearest 1 mL for the sources (e.g., aquatic and riparian plants) that were present in larger volumes.
Statistical Analyses. — Ten fecal samples were not included in the analyses because they did not present any identifiable food resources. Differences in CL and BM between adult females and males were tested with a Student t-test. We calculated the percent frequency of occurrence (%F), and volume (%V) for all food items in each sex class. The Index of Relative Importance (IRI) was calculated with a formula for herbivory adjusted by Bjorndal et al. (1997):
This index was calculated because the information on the %F and %V alone may overestimate or underestimate the importance of the food resources (Bishop et al. 2022). The IRI values integrate both indices (%F and %V) and allow a better comparison of the importance of food items, allowing the identification of potential differences among categories (e.g., sex).
We tested potential differences in feeding between adult females and males by applying an Heterogeneity Chi-square test for each postingestion sample using 1) the total number of broad categories found in each individual (NB), 2) the total number of food items found in each individual (NI), 3) the total volume considering all the food items identified in each individual (VI), and 4) the total volume of the principal food items in each individual (VpI); the principal food items (p) were selected according to the IRI values.
To evaluate dietary overlap between sexes, we calculated Morisita’s modified index of niche overlap (Cλ) by employing the frequency (f) of food items in adult females and males (Krebs 1989); the Cλ values can vary from 0 (no resources are shared) to 1 (the food resources are used in the same proportion; McCoy et al. 2020).
Potential differences between stomach flushing and fecal samples were assessed using a Heterogeneity χ2 test, comparing the composition (NB and NI) obtained in each postingestion sample. Normality was tested using the Shapiro-Wilk test, and the analyses were performed in Program R 3.1.2 (R Development Core Team 2014) with an α = 0.05.
RESULTS
Sampled Population. — We analyzed postingestion samples from 55 individuals whose gut and fecal contents were identifiable: 28 males (1 juvenile and 27 adults) and 27 females (12 juveniles and 15 adults).
From the stomach flush samples, we recorded data on 31 individuals, 1 male juvenile (CL = 344.6 mm, and BM = 5998 g), 11 adult females (CL, 471.2 ± 49.3, range 409.6 to 592.9; BM, 9758.7 ± 2909.9, range 5300 to 14,148), and 19 adult males (CL, 461.7 ± 51.4, range 381.8 to 560.7; BM, 9442.8 ± 2158.9, range 6477 to 15,700). No significant differences were found in CL (t = 0.49, p = 0.62) and BM (t = 0.33, p = 0.73) between adult females and males.
For fecal samples, we recorded data on 42 individuals, 13 juvenile females (CL, 275.1 ± 63.9, range 173.3 to 409.6; BM, 3069.6 ± 2470.7, range 563 to 8700), 10 adult females (CL, 470.7 ± 16.1, range 438.6 to 492.1; BM, 10,317 ± 2435.2, range 5300 to 12,650), and 19 adult males (CL, 460.8 ± 48.6, range 381.8 to 560.7; BM, 9276.0 ± 2169.6, range 6500 to 15,700). No significant differences were found in CL (t = 0.62, p = 0.53) and BM (t = 1.1, p = 0.24) between adult females and males.
Diet. — We recorded 3 broad categories and 12 different food items for both fecal samples and stomach contents (Table 1). The broad category “riparian resources” presented the largest variety of items in both postingestion samples (Table 1). The aquatic macrophytes, Pistia stratiotes and Pontederia sagittata, and the invertebrates were only detected in the stomach flushing samples (Table 1). Leaves accounted for the entire proportion of the aquatic macrophyte material found in the stomach contents. For the riparian plants, the turtles ingested leaves, fruits, and seeds (only for Mucuna argyrophylla, and Chamaedora sp.), whereas the invertebrates were ingested whole.
In the stomach flushing samples, we recorded all 3 broad categories (aquatic macrophytes, riparian resources, and invertebrates) and 10 food items (Table 2). According to the results of the IRI the most important items were P. stratiotes (aquatic macrophyte) followed by Ludwigia sp. (riparian resource; Table 2). The items in the broad category “invertebrates” presented low values compared with other resources (Table 2). For fecal samples we only identified the broad category “riparian resources”, and 6 food items (Table 3). For all the size and sex categories Ludwigia sp. was the most important item in the feces, followed by M. argyrophylla (Table 3).
Considering both methodologies, no significant differences were detected in the diet composition between female and male adults (NB and NI; Table 4). No evaluations were applied to assess NB differences between sexes using fecal samples because we only recorded riparian resources (Tables 1 and 3). Nonetheless, significant differences were observed in the overall volume consumed (VI) and when considering the principal food items (VpI). In both methodologies, females exhibited higher values than did males (Table 4). The Morisita’s index indicated that male and female turtles tended to make a similar use of food resources when considering either stomach flushing (Cλ = 0.89) and fecal samples (Cλ = 0.90).
Comparison of Postingestion Samples. — Comparing stomach contents and feces, we found significant differences in NB (stomach flushing, n = 31, 1.67 ± 0.87; feces, n = 42, only one broad category registered; χ2 = 2.46, p < 0.01), but not in NI (stomach flushing, 1.80 ± 0.90; fecal, 1.47 ± 0.74; χ2 = 3.42, p = 0.31).
DISCUSSION
Except for both aquatic macrophytes (P. stratiotes and P. sagittata), none of the plants documented in this study have been previously reported in the diet of D. mawii (Moll 1989; Gil Alarcón 2008; Bishop et al. 2022). The plants identified in both postingestion samples in our study correspond to common taxa of aquatic and riparian vegetation in the LRB (Carabias et al. 2015; Ochoa-Gaona et al. 2018). This agrees with previous studies suggesting that the diet of D. mawii is closely related to the plant sources available in their habitats (Legler and Vogt 2013; Bishop et al. 2022). We recorded a number of broad categories (NB = 3) and food items (NI = 12) consistent with findings from previous studies conducted in lentic systems (Moll 1989; Vogt and Flores-Villela 1992; Gil Alarcón 2008; Bishop et al. 2022). However, it is noteworthy that in lotic systems the turtles may rely on a wider range of resources (i.e., ≥ 19 food items; Moll 1989; Gil Alarcón 2008; Bishop et al. 2022). Allan et al. (2021) highlighted the contrasting abiotic factors (e.g., water velocity, depth, and ecosystem geometry) and biotic characteristics (e.g., the community composition of invertebrates, plants, and algae) that distinguish lotic and lentic systems, subsequently influencing the availability of potential food resources. For instance, Legler and Vogt (2013) reported that D. mawii in marshlands often exhibit a diet exclusively based on grasses as a result of their high abundance, while opportunistically consuming fruits when available. In contrast, individuals in riverine populations have the ability to move along lotic systems and access a wider variety of food sources, including leaves, fruits, lianas, and grasses.
In our study, the most important items identified for D. mawii were P. stratiotes (aquatic macrophyte), and Ludwigia sp. (riparian plant; Tables 2 and 3). Previous studies from lotic and lentic systems in the LRB (Vogt and Flores-Villela 1992) and Tabasco (Gil Alarcón 2008) in Mexico, and in Belize (Moll 1989; Bishop et al. 2022) emphasized the importance of riparian vegetation in the diet of D. mawii, while neither P. stratiotes nor other aquatic macrophytes contributed to their diets there. In the 1990s, P. stratiotes had only a limited distribution in the LRB (Ramírez and Lot-Helgueras 1992; SEMARNAP 2000). However, in the past 2 decades this aquatic macrophyte has been reported as a common species in the LRB in seasonal ponds, lakes, and streams with low current (Ochoa-Gaona et al. 2018) and is likely favored by anthropogenic eutrophication (e.g., from inputs of wastewater and fertilizers; Ghavzan et al. 2006). When possible, aquatic herbivores may selectively feed on aquatic rather than terrestrial resources (Newman 1991; Mandal et al. 2010; Tischler et al. 2019) because the former plants often have a comparatively higher nutritional value and are easier to digest as a result of a higher concentration of nitrogen (N) relative to carbon (C) in their tissues compared with terrestrial plants (Bjorndal and Bolten 1993; Tischler et al. 2019).
The high consumption of P. stratiotes by D. mawii suggests that the latter can act as a natural regulator of this aquatic macrophyte. Pistia stratiotes is considered a nuisance species throughout many tropical and subtropical regions (Neuenschwander et al. 2009) because it can quickly cover freshwater surfaces, preventing light penetration, and in turn, reduce the abundance and diversity of benthic algae, submerged macrophytes, and the associated benthic consumers (Howard and Harley 1997; Olkhovych et al. 2020). Additionally, P. stratiotes can provide microhabitats favoring disease vectors (e.g., mosquitoes), increase sedimentation rates, nutrient loads, and alkalinity, and degrade fish nesting sites (Howard and Harley 1997; den Hollander et al. 1999; Olkhovych et al. 2020).
In accordance with previous reports (Gil Alarcón 2008; Legler and Vogt 2013; Bishop et al. 2022), the occasional presence of invertebrates (only in the stomach flushing samples), and low values in %F, %V, and IRI (Table 2), suggest that these organisms were accidentally ingested by D. mawii in the study site. Some authors (Moll 1989; Ureña Aranda 2008; Gil Alarcón 2008; Zenteno Ruiz et al. 2010; Legler and Vogt 2013; Bishop et al. 2022) mentioned that D. mawii may function as a seed disperser, but no evidence has yet been shown about the effects of gut passage on germination. In this study, most of the seeds found in the postingestion samples appeared as inviable crushed remains, suggesting that the ability of the turtle to transport and spread viable seeds in the environment may be limited. The effect of endozoochorous seed dispersal by D. mawii should be addressed in future research on this species.
Consistent with previous findings in lotic and lentic ecosystems (Moll 1989; Gil Alarcon 2008; Legler and Vogt 2013; Bishop et al. 2022), our analysis revealed no significant differences in diet composition (NB and NI) between the sexes. Previous studies suggest that the diet composition of D. mawii may be primarily influenced by environmental conditions (i.e., habitat or season), rather than by biological aspects of the population (such as size or sex; Moll 1989; Gil Alarcon 2008; Legler and Vogt 2013; Bishop et al. 2022). Specifically, factors such as plant phenology, water level variation, current speed, and nutrient levels may have an influence on the composition of the diet of D. mawii (Moll 1989; Gil Alarcón 2008; Zenteno Ruiz et al. 2010; Legler and Vogt 2013; Bishop et al. 2022). Although adult females and males did not show any significant differences in body size (CL and BM), consumption (VI and VpI) of food resources was significantly higher in females, considering both methodologies (Table 4). Previous studies on other freshwater turtles similarly suggested that females may often ingest larger amounts of food than do males because females allocate substantial energy in reproductive processes such as egg production and movement to nesting sites (Congdon et al. 1987; Dood 1989; Doody et al. 2002; Legler and Vogt 2013).
In our study, stomach flushing revealed more broad categories (NB = 3) and P. stratiotes as the most important food item for D. mawii in the LRB (Table 2), whereas in the fecal samples only one broad category was registered (i.e., riparian plants), and aquatic macrophytes were not detected (Table 3). A potential explanation for this discrepancy is that more digestible food items with low concentrations of carbon in relation to nitrogen in their structure (e.g., aquatic macrophytes) are assimilated more efficiently and are therefore more difficult to detect in feces (Trites and Joy 2005). In this region, the average C/N ratio of P. stratiotes is significantly lower than in riparian species (M. Cazzanelli et al., unpubl. data, 2023). Thus, if only fecal samples were considered, the importance of Ludwigia sp. as a food source for D. mawii likely would be overestimated (IRI values for this plant in fecal samples were twice as high in females, and 3 times higher in males; see Tables 2 and 3). In other turtles, diet preferences were also inconsistent when only one method was considered. For example, Lima et al. (1997) characterized Phrynop rufipes as a palm-fruit specialist based only on fecal samples from small lotic habitats of Central Amazonia. However, based on stomach and fecal samples, Caputo and Vogt (2008) concluded that P. rufipes from similar habitats in the same region is an omnivorous species that feeds opportunistically on a wide variety of benthic macroinvertebrates and fishes.
In this study, large seeds of riparian species such as M. argyrophylla and Chamaedora sp. were only detected in fecal samples. Luiselli and Amori (2016) mentioned that large food items (e.g., seeds or whole fruits) cannot be obtained by the stomach flushing method on account of the constraint of the size of the esophagus. Thus, the use of feces makes it possible to identify some food items that may not be detected using only the stomach flushing method. Thus, the use of both methods provides more information and clarity regarding the diet, as well as the potential for seed dispersal.
In summary, P. stratiotes and Ludwigia sp. were the most important resources in the diet of D. mawii at the study site. We suggest that the high consumption of P. stratiotes by the turtle may indicate a significant ecological role in regulating the populations of this aquatic macrophyte. The results obtained in our assessments of composition (NB and NI) and niche overlap analysis indicated that D. mawii exhibits no diet differences between the sexes, although adult females demonstrated greater consumption (VI and VpI) of food resources in comparison with males. Finally, the use of both postingestion sampling methods may provide a more comprehensive view of the diet of freshwater turtles, especially in herbivorous species such as D. mawii. Long-term studies are needed to complete the characterization of the trophic ecology of D. mawii considering environmental, seasonal, ontogenetic, and sexual variations.
Location of the study area in the Lacantun River Basin, Chiapas, México. The shaded area in eastern Chiapas represents the Montes Azules Biosphere Reserve, and the square represents the region studied.
Contributor Notes
†Deceased
Handling Editor: Vivian P. Páez
