Alimentary Habits of the Mexican Mud Turtle (Kinosternon integrum) in Tonatico, Estado de México
Abstract
Alimentary habits are critical to the ecology of all animals. They reflect resource availability and are correlated with important life history traits such as reproductive output and body size. We analyzed the diet of the previously unstudied Mexican mud turtle Kinosternon integrum at Tonatico, Estado de México, during 2003 and 2004. Analysis was conducted separately by sex and age (immature vs. adults) and seasons (rainy vs. dry). Gastric contents and fecal samples were used in combination for more complete results. Based on 57 samples (32 stomach flushes and 25 from feces), K. integrum is a generalist–opportunist in alimentary habits, feeding on 27 categories of food. Plant material, Coleoptera, Odonata, Diptera, and mixed animal matter were the most important components. Overall dietary diversity was similar between adults and juveniles and between the two sexes, but juveniles differed between seasons. In similarity analysis we found a shift in diet between seasons. Females shifted from being primarily carnivorous during the rainy season to being primarily herbivorous during the dry season, while males were carnivorous during both seasons. The data suggest that this turtle feeds opportunistically on available prey items rather than on a few preferred food items.
The feeding habits of turtles are diverse and range from strict herbivory to strict carnivory (Wilbur and Morin 1988). However, freshwater turtles are generally omnivorous (Burke et al. 2000). Typically the alimentary habits of turtles are determined by prey availability and abundance, with diets structured by higher frequency of animal than plant material, or vice versa (Bjorndal 1991). Sex and age have also been considered important determinants of the alimentary habits of turtles, with juveniles eating a higher proportion of animal matter than adults (Ernst et al. 1994) to allow more rapid growth (Lindeman 1996), and females sometimes consume relatively more animal matter than males (presumably to support reproductive expenditures; Ford and Moll 2004). Diet composition is a very important aspect of the local life-history of a turtle species; it reflects the resources provided by the environment toward the energetic needs of the individual during its lifetime (Pincheira-Donoso 2008), including reproduction (Hume 2005).
Kinosternid turtles are characterized by having a variable kinetic plastron and musk glands and inhabit a great variety of environments (Iverson 1999). The widespread New World distribution of these turtles could be explained, in part, as a consequence of their aestivation ability (Peterson and Stone 2000), their complex biogeographic history (Pritchard and Trebbau 1984; Iverson 1991), their variable plastron size (Iverson 1991), their diverse reproductive strategies (Iverson et al. 1991), and their diverse alimentary habits (Ernst et al. 1994). Many kinosternids are omnivorous, eating almost everything found in the water, including leaves, fruits, seeds, nuts, insects (aquatic and terrestrial), mollusks, crustaceans, aquatic vertebrates such as amphibians, other turtles (kinosternids included), and carrion (Hulse 1974; Punzo 1974; Vogt and Guzmán-Guzmán 1988; Ernst and Barbour 1989).
Alimentary habits have been studied in several kinosternids. Although Kinosternon oaxacae, K. chimalhuaca, K. angustipons, and possibly K. alamosae, have been considered mostly herbivorous (Legler 1966; Iverson 1986, 1989; Berry et al. 1997), other species such as Staurotypus triporcatus, S. salvinii, Claudius angustatus, Sternotherus depressus, S. minor, K. hirtipes, K. dunni, and K. creaseri have been considered carnivorous (Mahmoud 1968; Iverson 1988; Ernst and Barbour 1989; Iverson et al. 1991; Ernst et al. 1994; Espejel-González 2004). The rest of the family, including Sternotherus carinatus, S. odoratus, Kinosternon leucostomum, K. baurii, K. flavescens, and K. scorpioides have been considered omnivorous (Einem 1956; Moll and Legler 1971; Hulse 1974; Gibbons 1983; Mitchell 1988; Iverson 1989; Ernst et al. 1994; Morales-Verdeja and Vogt 1997; Campbell 1998; Ford and Moll 2004). In certain species alimentary habits have been shown to differ by sex, age, and season (e.g., K. herrerai; Aguirre-León and Aquino Cruz 2004).
However, for K. integrum, the turtle species with the most widespread distribution in Mexico (Iverson 1999), dietary patterns are unknown (Lemos-Espinal and Smith 2007). This species has a summer (rainy season) nesting period (Iverson 1999), beginning in July and ending in October (Macip-Ríos et al. 2009). Males are larger than females, with more elongated carapaces and wider heads (Ernst and Barbour 1989). This turtle aestivates in the dry season if water is not available, but some individuals remain active in permanent pools during the dry season (Iverson 1999; Macip-Ríos et al. 2009).
Our goal in this paper was to describe the alimentary habits of the Mexican mud turtle (K. integrum) at Tonatico, Estado de México, analyzing the diet of young vs. adults, males vs. females, and during the rainy vs. dry seasons. We expected to find 1) an overall omnivorous diet like many others tropical kinosternids, and 2) a shift in diet from animal matter in the rainy (reproductive) season to plant matter in the dry (nonreproductive) season because of the temporal dynamics of the ponds that the turtles inhabit, and because of the higher reproductive expenditures faced by adult males and females (Hume 2005).
METHODS
Turtles were collected from October 2003 to November 2004 at La Puerta de Santiago, Tonatico (18°48′N, 99°40′W), in southern Estado de México, near the states of Guerrero and Morelos. Tonatico is in the upper Balsas River basin hydrologic region, at 1640 m above sea level. Through the year, average annual temperature is 20°C, and annual rainfall averages 150 mm but is highly variable by month and year; the record maximum is 401.5 mm in August, and the record minimum is 0 mm from October to April (INEGI 2002). Turtles were collected in 4 temporary ponds with a range of 20–200 m in diameter and along a small permanent stream near the ponds. The 200-m pond was 4 km from the 50-m pond, while other two small ponds (30 and 20 m) were within 150 m of the 50 m pond. We considered the dry season to last from early October to mid-June and the rainy season from mid-June to late September, based on climate data obtained from the Servicio Meteorológico Nacional (México) and our field observations (Fig. 1).
Citation: Chelonian Conservation and Biology 9, 1; 10.2744/CCB-0782.1
The study site was visited monthly and turtles were caught by seine and hoop traps baited with fresh fish. The bait was enclosed in plastic containers to prevent turtles from consuming it. When ponds dried up (in October and November), we caught turtles only in the permanent stream and in the remaining water of the 200-m pond. During January and February no turtles were caught. Each turtle was marked by shell notching (Ferner 1979) and measured, weighed, and sexed by standard methods. In order to determine the diet we used stomach flushing (Legler 1977). Fecal samples were also collected when turtles defecated on capture. Despite the possible bias using stomach contents or feces data individually, the use of both techniques as an integrated approach is recommended to provide a comprehensive picture of turtle diet (Caputo and Vogt 2008). Feces and stomach contents were preserved in 70% ethanol for laboratory analysis. Samples were dried on paper and examined with a stereoscopic microscope; we separated animal and plant matter, identified each item to the lowest possible taxonomic level, and weighed each item on a semianalytical balance. Turtles were released where they had been caught, and no apparent harm was inflicted on them during the study.
Data were tabulated as follows: a) frequency of occurrence (percentage of the total number of stomach flushes or fecal samples in which each diet category occurred); b) numeric frequency (percentage of each item in each diet category in relation to the total number of categories across all samples); and c) percent by mass (percent of mass of each category of diet in relation to the total mass of all categories present (Hulse 1974; Aguirre-León and Aquino-Cruz 2004). Some samples such as grasses were counted as the number of leaves or roots found; algae were counted as clusters of 5-mm3 filaments. Comparisons were made between rainy and dry seasons and among sex and age categories (males vs. females vs. immatures) with regard to origin of the items consumed (animal vs. plant). We also used the index of relative importance (IRI; Hansson 1998; Gümüş et al. 2002) for the overall data of diet by sex and age. The values for the food items were plotted by season, sex, and age in order to analyze the ranks of food ingested by this population. The values of IRI were reported only for the complementary data of numeric frequency, frequency of occurrence, and percent by mass.
We used a contingency table to test the nonrandom presence of plant and animal matter frequencies in all diet categories by sex and age. Two-way analysis of variance (ANOVAs) were used when data achieved parametric assumptions. When those assumptions were violated, we used 2-way Kruskal-Wallis (Zar 1999; Marquez-Dos Santos 2001) to analyze variation in frequency of occurrence, percent by number, and percent by mass between seasons, by sex, and age. We also used the Shannon-Wiener index to quantify specialization in diet (Plummer and Farrar 1981; Vogt and Guzmán-Guzmán 1988). Data for seasons (rainy vs. dry) were compared using Wilcoxon tests, and finally, we used the Morisita simplified index to determine similarity in diet between seasons, sexes, and age classes (Krebs 1999; Aguirre-León and Aquino Cruz 2004). All statistical tests were performed in JMP ver. 5.0.1 (SAS Institute Inc. 2002), with an α = 0.05 (Zar 1999).
RESULTS
Of the 57 samples analyzed, 32 were from stomach flushing and 25 from feces; 31 samples were from the rainy season and 26 from the dry season. The diet of K. integrum in Tonatico included 27 food categories (Table 1). For both percent frequency and percent mass, plant material (grasses and other herbaceous plants), coleopterans, odonate larvae, dipterans, and mixed animal matter (undetermined) were the most important components. Other animal groups such as gastropods, annelids (leeches), arachnids, and amphibians were also found. Seeds of 7 plants and filamentous algae were also found in stomachs and feces.
Using the frequency of occurrence data we found that the kind of diet components (animal vs. plant) were ingested randomly among the 3 sex and age classes (χ22 = 1.13, p = 0.56; Table 1). For the percent of mass data, we also did not find significant variation in plant vs. animal matter ingested among sex and age classes (F2,53 = 0.26, p = 0.61). However, females showed a tendency to eat more plant matter in the dry season than males, and males ate similar amounts of animal matter during the year (Table 1). Females also showed an important shift in the amounts of animal and plant matter between seasons (Fig. 2); they switched their diet from mainly carnivorous in the rainy season to mainly herbivorous in the dry season. Males and immatures did the same but in a less dramatic way (Fig. 2). To confirm this suggestion we conducted a 2-way ANOVA on the ln-transformed data of percent of mass (only for animal matter) but did not find significant variation either between seasons (F1,31 = 3.04, p = 0.09) or among sex and age classes (F2,31 = 2.92, p = 0.07).
Citation: Chelonian Conservation and Biology 9, 1; 10.2744/CCB-0782.1
Percent of mass also did not differ among sex and age classes or between seasons (F2,53 = 2.36, p = 0.10; Table 1); however, these data describe the entire diet of this population on a detailed level and the amounts of mass ingested by sex and age. Excluding undetermined items, the data of percent of mass and frequency of occurrence (Table 1) indicated that the most abundant prey items consumed did not necessarily correspond to the higher amounts of biomass. We found significant variation in numeric frequency (F2,74 = 28.23, p < 0.05), which reflected differences between immatures in dry season and females in dry season. When food is presumably scarce, immatures fed more on filamentous algae, terrestrial insects that fell into the water, and snails (which were very abundant in this season). In addition, females extended their diet to include larger invertebrates and more plant material during the dry season.
The results of the IRI calculated for seasons and sex and age (Table 1) show that in the rainy season the mixed animal matter, mixed plant matter, and odoante larvae were the most important items consumed. For instance, males and females fed on the same categories in almost the same proportion, but immatures fed more on mixed plant matter. In the dry season mixed plant matter, Lemna sp., and mixed animal matter were the most consumed prey items. In the dry season there was an absence of odonate larvae, and immatures concentrated their diet on Lemna sp.; males on plant matter, animal matter, and seeds; and females on almost the same items as males. We also categorized the primary consumed items, secondary consumed items, and randomly consumed items, by the differences in IRI. We considered diet items with IRI values below 10 to be randomly consumed, between 10 and 15 secondarily consumed, and above 15 primarily consumed.
Diet diversity showed significant variation (H5 = 16.89, p = 0.004) among classes stratified by sex, age, and season (Fig. 3). Immatures had the lowest value in the dry season (2.08), followed by males in dry season (2.29), females in rainy and dry seasons (2.56, 2.68), males in rainy season, and finally immatures in rainy season with the highest value (2.78). The overall diversity (combined data of age and sex) did not differ statistically between seasons (Z1 = 1.27, p = 0.20; Fig. 3), but diet diversity had a tendency to decrease from rainy (2.79) to dry season (2.66) as the ponds dried up in the middle of the autumn (Fig. 2).
Citation: Chelonian Conservation and Biology 9, 1; 10.2744/CCB-0782.1
Morisita's index for frequency of occurrence showed similarities in the alimentary habits between sex and ages in the rainy season (Table 2). However, in the dry season the alimentary habits differed among females, males, and immatures from those during the rainy season, with an average similarity fluctuating around 0.60. Males and females showed a shift in similarity between seasons, but immatures showed a more dramatic dissimilarity between seasons (ranging from 0.56 to 0.63). In short, there was a shift in alimentary habits between seasons, becoming less diverse in the dry season when resources would be expected to be more limited.
DISCUSSION
Our results indicated that Kinosternon integrum could be classified as a generalist–omnivore. The diet of this turtle population is highly diverse and did not show differences in frequency of occurrence or percent mass, with no particular predominance of animal vs. plant matter. In addition, the IRI results point out the generalist trend in this population. These results confirmed our first prediction.
Hulse (1974) argued that most kinosternids are opportunistic in their alimentary habits, because they feed on every kind of animal matter found in water, as well as on plant matter (stems, leaves, and seeds) from the pond shore. Iverson and Berry (1979) suggested that K. integrum was an ecological generalist, but they did not provide information about diet. We found terrestrial insects in stomachs and feces (ants, land beetles, arachnids, and adult dipterans), and this observation suggested that they were ingested after accidentally falling into the water. However, some kinosternids are known to make occasional excursions on land to feed on plants and arthropods (Einem 1956; Mahmoud 1968). Carr and Mast (1988) suggested that the relatively large size of the head and jaws of K. herrerai let the turtle ingest a great variety of prey. Kinosternon integrum (males primarily) also had big heads and powerful jaws, which let the turtle feed on a great variety of prey items, but male diet variety was no greater than females, suggesting that head size may be more important for other activities (e.g., mating). Other species such as Staurotypus triporcatus, S. salvinii, Claudius angustatus, and Sternotherus minor also have big heads (Ernst and Barbour 1989; Ernst et al. 1994), and this pattern could be associated with carnivory, particularly molluscivory (Berry 1975).
There is evidence of opportunistic feeding in other kinosternids with a tendency toward carnivory in K. alamosae (Iverson 1989), K. arizonense (Iverson 1989), K. baurii (Einem 1956), K. chimalhuaca (Berry et al. 1997), K. creaseri (Iverson 1988), K. dunni (Ernst and Barbour 1989; Ernst et al. 1994), K. flavescens (Mahmoud 1968; Punzo 1974), K. herrerai (Aguirre-León and Aquino-Cruz 2004), K. hirtipes (Ernst and Barbour 1989; Ernst et al. 1994), K. leucostomum (Moll and Legler 1971; Vogt and Guzmán-Guzmán 1988), K. scorpioides (Moll 1990), K. sonoriense (Hulse 1974), Sternotherus minor, S. carinatus, K. subrubrum, and S. minor (Ernst and Barbour 1989; Ernst et al. 1994). Other species such as K. angustipons (Legler 1966) and K. oaxacae (Iverson 1986) show a tendency toward herbivory. No kinosternid is considered to be exclusively carnivorous or exclusively herbivorous. Diet composition in kinosternids seems to be localized and related to prey availability more than prey preferences, but few simultaneous data are available for kinosternids on prey density and diet (but see Vogt and Guzmán-Guzmán 1988).
The biomass (percent mass) analyses and IRI results showed that immatures were more herbivorous than adults. This is interesting because it contradicts the typical pattern in turtles of being more carnivorous as juveniles. Juvenile freshwater turtles typically eat more animal material, because that may allow more rapid growth and faster calcium gain for bone composition (Lindeman 1996), more rapid attainment of sexual maturity, and increased survivorship (Georges 1982; Hart 1983; Parmenter and Avery 1990; Gibbs and Amato 2000).
Prey item abundance and availability may be more important in kinosternids than a particular preference for prey type. In the dry season immature individuals consumed a low number of animal and plant items, with 50% of their diet made up of filamentous algae, mixed plant matter, and mixed animal mater. The mixed plant and animal categories included items that could not be identified to obvious taxonomic level, such as roots, leaf fragments, and insect parts that could be obscuring the real diversity of diet. However, the absence of odonate larvae in the diet during the dry season (although they were common in the diet during the rainy season) suggests a shift to other prey with higher availability during the dry season. Furthermore, the addition of gastropods and filamentous algae to the diet in the dry season (prey that are abundant when the ponds are drying) confirms our observation of a diet shift by prey availability. Ford and Moll (2004) found a similar seasonal diet shift for S. odoratus.
Females did not show significant variation in the biomass ingested (percent by mass) between seasons; however, data suggest a strong shift from carnivory in the rainy season (which is also the egg-laying season) to herbivory in the dry season (Fig. 2). Probably with larger sample sizes we would have found a statistically significant difference. This finding supports in part our second prediction, albeit only for females. Ford and Moll (2004) suggested a possible association of seasonal diet variation with reproductive investment in S. odoratus; however, other studies suggest that differences in alimentary habits between the sexes in freshwater turtles could be an effect of difference in habitat selection between males and females (Plummer and Farrar 1981; Hart 1983; Parmenter and Avery 1990). Because we caught males and females in about the same numbers in the ponds surveyed, our data suggest that habitat selection was not related to diet sex differences. Females ate more animal matter than males in the rainy (reproductive) season. In contrast, in the dry season males ate more animal matter than females. This trend could be driven by the energetic demand of reproduction in females (Hume 2005). However, still unclear is why males did not shift their diet between seasons. One possible explanation could be that males were more territorial than females and controlled the best food resources in the ponds throughout the year, or maybe the energetic demand of courtship and male–male competition (including territoriality) could also be highly costly year-round.
Another important issue related to the diet of K. integrum is the consumption of seeds. We did not perform tests of seed viability in fecal samples; however, Aguirre-León and Aquino-Cruz (2004) reported that K. herrerai also eats seeds and could be a seed disperser. Ford and Moll (2004) also suggested the same possibility in S. odoratus. Because K. integrum moves among the ponds, even as far as 1 km apart (R. Macip-Ríos, pers. obs.), we do not exclude the possibility of its being a potential seed disperser.
We found a decrease in dietary diversity from the rainy season to the dry season in immature turtles (from 23 prey items to 9 prey items), probably reflecting the condition of the habitat. Adults maintained dietary diversity across seasons. This could indicate stratification in feeding habits between adults and immatures, the latter possibly feeding on low-quality food or occupying different microhabitats. This likely reflects changes in the dynamics of prey diversity and abundance in temporary ponds (Williams 1997), with high peaks of diversity during the middle of the rainy season and decreases at the end of the season (Dodds 2002).
The overall diet diversity data do not show significant seasonal differences, but there is a tendency of diversity to decrease through the seasons (Fig. 3). Vogt and Guzmán-Guzmán (1988) reported the alimentary habits of K. leucostomum in two different habitats, and differences were related to prey abundance and perhaps availability, with a decline in dietary diversity between rainy season and dry season. These data suggest a shift in diet across seasons by sex and age, as Aguirre-León and Aquino-Cruz (2004) reported for K. herrerai. This suggests a pattern of resource segregation in the population, possibly based on male dominance, although further research is needed. In any case, our results confirm our second prediction only for immatures.
The study of diet habits and preferences in kinosternids certainly needs more attention. Research on lineage effects is needed to understand whether diet is primarily a response to local pressure (e.g., resource availability) or to evolutionary trends related to life history, physiology, and morphology. Other important issues such as dietary differences among sex and age classes and the effect of behavior (e.g., dominance relationships) in feeding behavior definitely need more attention.
Historical climate data for Tonatico, Estado de México (1963–1988). Bars indicate average rainfall per month (mm); line and squares indicates average temperature per month (°C).
Seasonal percent carnivory in the diet of Kinosternon integrum at Tonatico based on: A) frequency of occurrence, B) numeric frequency, and C) percent by mass. F = females, M = males, and I = immatures.
Shannon-Wiener comparison of diet diversity among sex, age, and overall data in rainy and dry seasons. Shared letters indicate no significant differences at α = 0.05.
