Approximately 4 billion people live within 60 km of the Earth's coastlines, influencing river flow rates, pollution, and eutrophication of estuaries (Kennish 2002; Greene et al. 2015). The diamondback terrapin (Malaclemys terrapin) has a life history that predisposes it to impacts from humans both on land and in the near-shore environment. Terrapins are unique among turtle species, being the only one to exclusively inhabit estuaries. They occur from Massachusetts along the Atlantic and Gulf coasts to Texas. Previous terrapin diet studies have been limited to temperate salt marshes in north Florida (Butler et al. 2012), South Carolina (Tucker et al. 1995), North Carolina (Spivey 1998), Maryland (Roosenburg et al. 1999), Virginia (Tulipani 2013), and New York (King 2007; Petrochic 2009; Erazmus 2012). Results of these studies showed that terrapins are secondary consumers that feed on benthic invertebrates and revealed the role of the terrapin as an integral component of the estuarine food web (Hurd et al. 1979; Tucker et al. 1995). In salt marshes, terrapins are dietary generalists that consume a variety of crustaceans, mollusks, and small fish (Tucker et al. 1995; Spivey 1998; Roosenburg et al. 1999; Whitelaw and Zajac 2002; King 2007; Petrochic 2009; Erazmus 2012; Tulipani 2013). However, much is still unknown about terrapin feeding habits and variation in diet across their wide geographical and ecological range, including terrapins inhabiting subtropical mangroves in the southernmost extent of their range.

The mangrove diamondback terrapin (Malaclemys terrapin rhizophorarum) is the only subspecies found exclusively in the subtropical waters off south Florida's coasts and keys. Aside from an opportunistic sampling trip to investigate terrapin diets in the Florida Keys (Denton et al. 2015), we have found that only 3 previous studies have examined the ecology of mangrove terrapins (Wood 1981; Baldwin et al. 2005; Hart and McIvor 2008). In addition to the mangrove terrapin, 2 additional subspecies can be found in south Florida's mangrove systems: the eastern Florida diamondback terrapin (Malaclemys terrapin tequesta) occurs from northeast Florida southward along the Atlantic coast to the Keys, and the ornate diamondback terrapin (Malaclemys terrapin macrospilota) inhabits islands in Florida Bay north along Florida's Gulf Coast. All 3 subspecies are part of one genetic “cluster” (Hart et al. 2014). Because the focus of the present study was to characterize the diet of Malaclemys in mangrove habitats, for simplicity, we categorized the 3 subspecies as South Florida (SF) terrapins throughout this article.

The objective of the present study was to provide insight into the foraging ecology of SF terrapins living in a subtropical environment. Previous terrapin diet studies in more temperate regions have shown immature females to have diets more similar to males, being of similar size, but dissimilar to the larger mature females that have more powerful jaws enabling them to consume a wider diversity of prey (Tucker et al. 1995; Petrochic 2009; Butler et al. 2012; Tulipani 2013). A shift in diet has also been reported in female terrapins, coincident with movements between habitats during the nesting season (Butler et al. 2012). Here, we investigated size- and habitat-related variation in SF terrapin diets for terrapins sampled within the boundary of Everglades National Park.

METHODS

Study Area

To limit anthropogenic influences, spatially distinct sampling sites and habitats were selected within the boundary of Everglades National Park (Fig. 1). Terrapins were sampled in forested mudflat creeks located north of Cape Sable on the southwest coast of Florida (Creek site) and on 2 islands in central Florida Bay (Island site). The Creek site is a tidal system consisting of 2 main channels with smaller bisecting creeks extending back approximately 2 km from the coast. The creeks are dominated by red (Rhizophora mangle) and black (Avicennia germinans) mangroves along the fringe and near-shore zone, transitioning to other salinity-tolerant tropical hardwoods toward the interior. Significant tidal fluctuations of 1.2 m are common and expose large mudflats during low tide. The Island site consisted of 2 islands (each < 1.5 km²) located in Florida Bay, > 50 km WSW of the Creek site (Fig. 1). Florida Bay is bounded by mainland Florida to the north, Cape Sable to Matecumbe Key on the west, and the rocky Florida Keys to the southeast. These muddy islands are comprised of red and black mangrove swamps, fringed with mangrove forests, and contain vast open spaces devoid of vegetation, which are typically inundated with water during the wet season (Enos 1989).

View Figure

• Download as Powerpoint
• Download Figure

Terrapin Capture and Sample Collection

All terrapins were captured with a dip net or by hand, and a GPS location was recorded. The Creek site was sampled in January 2012 over 4 consecutive days and on 2 d in September 2012. The Island site was sampled 10 times from May 2012 through June 2013. Terrapins were sampled, marked, photographed, and cataloged as in Hart and McIvor (2008). Seigel (1984) determined that females in central Florida reached maturity at a straight-line plastron length (SPL) between 135 and 143 mm; thus, we classified females with a SPL > 135 mm as mature and in the Large size class and terrapins with a SPL < 135 mm as Small. We held terrapins in 19-l buckets containing freshwater to induce defecation, after which we collected fecal samples in labeled centrifuge tubes containing 70% isopropyl alcohol. Terrapins were released at original capture sites within 24 hrs. In the laboratory, each sample was rinsed with deionized water, transferred to glass petri dishes, and allowed to air dry over several days or dried in a Boekel Scientific Incubator (133000) to reduce drying time. Dried samples were sorted for diet items, which were then identified to the lowest possible taxonomic level, and then weighed, with mass recorded to the nearest 0.0001 g using a Denver Instrument balance (APX-60). Concurrent to terrapin sampling, we collected specimens of potential prey items at each site to create a reference collection. Voucher specimens were used to identify prey items found in the fecal samples.

Diet Analysis

We identified samples to the lowest possible taxonomic level before grouping remains into 5 predetermined food categories (gastropods, decapod crustaceans, bivalves, barnacles, and fish; Table 1), enabling comparisons to previous studies. In addition to the 5 food categories and species data used in statistical analyses, we classified additional nonfood categories and subsequently excluded them from analyses. Similar to Butler et al. (2012), we combined crabs into one category and vegetation consisted of combined terrestrial and aquatic sources. Frequency of occurrence (FO) is an appropriate metric when individual prey items cannot be quantified because of fragmentation as in the present study (Rosenberg and Cooper 1990); thus, for each category, we calculated total dry mass (in grams) and FO within the mangrove system, separated by habitat and terrapin size class. We calculated FO from the number of samples that contained a food category, divided by sample size of terrapins in that particular habitat or size class.

Table 1. Frequency of occurrence (FO) of all food items identified from the fecal samples from south Florida terrapins. The food items are listed by categories (in bold) and species when available.

View Fullscreen

PRIMER Statistical Software 6.0 (Clarke and Gorley 2006) was used to calculate Shannon Diversity (H′) and Pielou's Evenness (J′) indices to describe diversity of food items for pooled samples, each size class, and each habitat using the FO of each prey category (Magurran 2004). The Shannon–Weiner index is calculated as H′ = −∑ p_j · log_e(p_j), where p_j is the proportion of the total sample belonging to ith species (Van Dyke 2008). Because H′ may range from 0 to ∞, the index was standardized using Pielou's (1975) evenness measure J′, calculated as J′ = H′/log(S). Constrained between 0 and 1, low J′-values indicate low diversity of prey consumed and, hence, a high degree of specialization (Krebs 1999).

We performed multivariate analyses in PRIMER 6.0 to test for similarities or dissimilarities of terrapin diet by size and habitat. Similarities were calculated by listing food items as present/absent, effectively giving equal weight to all categories, whether rare or abundant; thus, the data did not require additional transformation. Bray–Curtis similarity measures were calculated to construct similarity matrices based on presence–absence data (Clarke et al. 2006). We ran an analysis of similarities (ANOSIM) to test for differences in diets between habitats and size classes and reported the level of separation between groups as indicated by the Global R-value. Global R is constrained between 0 and 1, with values close to 1 indicating the diets of the groups were well separated with little overlap. When ANOSIM results indicated significant differences, we ran similarity percentages (SIMPER) analysis to determine percent dissimilarity and contribution of various prey categories responsible for differences between groups.

RESULTS

Terrapin Captures

During the 18-mo study, we captured 109 terrapins, with 57% providing fecal samples. Of 33 females that provided samples, 3 were juveniles (Island–1, Creek–2) in the Small size class (SPLs of 72, 90, and 121 mm). The 30 mature adult females in the Large size class had SPLs ranging from 145 to 172 mm, mean ± SD = 158 ± 9 (Creek–11), and from 145 to 184 mm, mean 157 ± 09 (Island–19). All 29 males that provided fecal samples were mature, with SPLs ranging from 96 to 119 mm, mean = 106 ± 5 (Creek–25), and from 104 to 113 mm, mean = 109 ± 4 (Island–4).

Dietary Analysis

The most commonly found categories in order of FO were gastropods (Melampus coffeus and Cerithidea scalariformis) and decapod crustaceans (Uca thayeri, Aratus pisoni, and Panopeus herbstii). Remains of several additional species were identified at lower frequencies, including mud snails (Assiminea sp.), bivalves (i.e., southern marsh clam, Polymesoda floridana, and Anomalocardia sp.), barnacles (Balanus sp.), false ceriths (Batallaria minima), and small fish bones and scales (for complete list of identified remains, see Table 1). Gastropod opercula were found in 29% of the samples and, although not identified to species, were frequently associated with shell fragments of C. scalariformis, Littorina angulifera, or B. minima.

Gastropods (27.81 g), bivalves (25.75 g), and crustaceans (25.06 g) comprised approximately 96% of the total mass across all analyzed samples (82.24 g; Table 1). Although vegetation (mangrove bark, algae, seeds, and rhizomes) was the most frequently occurring item (84%), dry mass of this category (2.17 g) contributed only 3% of the total mass across all samples. Barnacles (1.29 g) and small fish bones (0.15 g) constituted the remaining 2% of the analyzed food items. Although decapod crustaceans, gastropods, and bivalves were the dominant prey by both occurrence and mass, individual samples often contained multiple diet items within the fecal contents. Of the 59 samples, 22% contained 2 species, 32% contained 3 species, 19% contained 4 species, and 1 sample contained 5 different prey species.

Although vegetation had a high FO, as in previous studies (Tucker et al. 1995; Petrochic 2009; Butler et al. 2012; Erazmus 2012; Tulipani 2013), we presumed it to be incidental ingestion that may have been associated with the consumption of targeted invertebrate prey species and was excluded from our analyses. Several fecal samples also contained additional nonfood items excluded from the analyses (sand, manmade debris, parasites). Soft tissue and shell remains that were too fragmented for accurate identification were grouped into an Unidentifiable category that was subsequently excluded from analysis (n = 25, total dry mass 0.931 g). Insect remains were found in 22 samples; however, because of field conditions during sampling, it was difficult to determine whether they were ingested as food or drowned in the terrapin-holding-buckets; therefore, they were omitted from analyses. Additionally, ostracods, copepods, and other micro-invertebrates were identified infrequently but because of to their small size (< 1.5 mm), they were presumed to have been ingested incidentally and were excluded from further analysis. Three fecal samples contained only the excluded items; thus, analyses were performed on the remaining 59 samples.

Shannon–Weiner diet diversity and evenness indices at the level of 5 higher prey taxa showed terrapins from the Creek site and Small size class had slightly higher diversity of prey than did those from the Island site and Large size class, respectively. Evenness (J′) also showed prey items were distributed more equally at the Creek site and in the Small size class (Table 2). However, with values close to 1, all indices represent moderate to high levels of diversity in prey consumed (Krebs 1999), with the overall evenness of the diversity for SF terrapins representing little specialization among terrapins by size or habitat. When indices were recalculated by prey species rather than broader categories, diversity increased, and evenness decreased slightly for pooled samples from both mangrove habitats, although they are still suggestive of moderate to high levels of diversity in prey consumed by SF terrapins (Table 2).

Table 2. Shannon Diversity (H′) and Pielou's Eveness (J′) indices for south Florida (SF) terrapin diet diversity between size classes and habitats. Indices were calculated for 5 predetermined higher prey taxa (category data) and the species data (see Table 1). The lower the value of J′, the more specialized the feeding habits of a particular overlapping group (i.e., the lowest J′-value) indicates the least diversity of prey consumed and, hence, the greatest degree of specialization.

View Fullscreen

Although multivariate analyses of SF terrapin diets using species data revealed differences in diets by size class (ANOSIM, R = 0.14, p = 0.01), the low R-value indicates that size alone did not explain a large portion of the variance in the terrapin diet data. To further determine whether the differences in size class were attributable to habitat differences, we performed a 2-way crossed analysis of size and habitat (i.e., tested for a significant an interaction term), and results indicated that there was not a difference in diet by size (ANOSIM, R = 0.057, p = 0.208); thus, size classes were pooled for habitat comparisons. Terrapin diets differed between the Creek and Island habitats (ANOSIM, R = 0.48, p = 0.01). SIMPER analysis revealed terrapins collected from the Creek sites had a higher similarity to each other (38%), with crabs and Assiminea sp. comprising over 84% of their diet similarities, whereas the diets from terrapins at the Island sites were slightly less similar to each other (35%), with C. scalariformis and M. coffeus contributing to over 79% of similarities (Table 3). Additionally, SIMPER analysis indicated terrapin diets from the Creek and Island sites had an average dissimilarity of approximately 89%, with the dissimilar contribution of crabs, C. scalarifomis, M. coffeus, and Assiminea sp. accounting for almost 60% of dietary differences.

Table 3. SIMPER analysis revealed that terrapin diets varied within a site with similarities between only 34% and 39%, whereas terrapin diets within each site were more similar to each other than the diets of terrapins between the sites (over 88% dissimilar). These results suggest south Florida terrapins are generalists feeding on the resources available in the environment.

View Fullscreen

DISCUSSION

The majority of the fecal samples collected contained food items from multiple categories, as well as multiple species within particular categories, indicating that terrapins are foraging widely rather than selectively feeding on a few abundant prey types that they find in patches. In contrast to studies conducted in salt marsh habitats that concluded that Large terrapins have a more diverse diet (Tucker et al. 1995; Petrochic 2009; Tulipani 2013), the diversity and evenness was slightly higher for Small size classes relative to the Large size classes of SF terrapins. With both size classes of SF terrapins consuming prey from each category and having high levels of diversity, it does not appear that terrapins in mangrove habitats partition resources by species; however, our study design did not address the hypothesis in Tucker et al. (1995) that they partition food resources by prey size.

Although we found no significant difference in diet between size classes in this study, there were differences between size classes in the proportions of food categories consumed compared with other studies. Tucker et al. (1995) and Butler et al. (2012) found the salt marsh periwinkle, Littoraria irrorata, to be the gastropod most heavily consumed by mature females; yet we found the mangrove periwinkle, Littoraria angulifera, in the diets of only 2 Large terrapins. Petrochic (2009) found the mud snail Ilyanassa obsoleta to be the most frequently consumed gastropod; yet neither Tucker et al. (1995), Butler et al. (2012), nor we found I. obsoleta in samples. Denton et al. (2015) found C. scalariformis in 100% of samples, and it constituted approximately 97% of the total dry mass across all samples collected. Differences in gastropod frequency of occurrence in terrapin diets may be a function of availability and accessibility within the habitat. For example, mangrove periwinkles (L. angulifera) are frequently found higher up on mangrove prop roots, which may be accessible to terrapins only during high tide. In the present study, a different suite of mud snails constituted the diet of SF terrapins depending on size class and location. Large terrapins fed mostly on C. scalariformis and M. coffeus, whereas Small terrapins consumed smaller snails (Assiminea sp.) most frequently, followed by the larger M. coffeus and C. scalariformis. Although the particular species consumed differed across studies, consistent selection of gastropods across studies indicates their importance to terrapin diets.

Several crab species have been reported as terrapin prey within salt marsh habitats; large female terrapins fed on both large and small crabs, whereas smaller immature females and males typically fed on the smaller crabs (Tucker et al. 1995; Butler et al. 2012; Tulipani 2013). Crabs represented an important food resource for mangrove terrapins (present study) with several small crab species (U. thayeri, A. pisoni, and P. herbstii) being consumed by both Small and Large size classes. Frequency of occurrence of crabs for Small terrapins was twice that of Large terrapins; however, this could be attributable to differences in crab abundance and terrapin size classes sampled at each location. Based on qualitative counts, all 3 crab species were abundant at the Creek site, which also had a greater number of Small individuals sampled than from the Island sites where the larger blue crabs (Callinectes sapidus) were more abundant. Although previous work has reported blue crab fragments in mangrove terrapin samples (K. Hart, pers. obs.), there was no direct evidence that either size class of terrapin consumed blue crabs or other large crabs in this study. It is possible that terrapins ingested blue crabs at our sites, but species identification was not possible because of the degraded nature of the prey remains. Nevertheless, results from this study reveal the importance of crabs in the diets of mangrove terrapins, which is consistent with previous work that determined crabs to be the most (Tulipani 2013) or second-most frequently occurring food item (Tucker et al. 1995; Spivey 1998; Butler et al. 2012; Erazmus 2012; Table 4).

Table 4. Summary of this study (South Florida) compared with published diamondback terrapin diet studies ranking prey categories from most common (1) to least common (5) by maximum reported frequency of occurrence (FO); darker shading indicates a higher FO. Studies are listed left to right geographically from southern to northern sites along the Atlantic coast of the eastern United States. Studies include both size classes and both genders except Erazmus (2012), which included only Large nesting females. nr = prey type not reported; nv = prey type mentioned but no value given (modified from Tulipani 2013).

View Fullscreen

Even though the FO was low for bivalves, this prey class constituted the second highest dry mass (Table 1), emphasizing the limits of FO in determining the contribution of various prey items to diet and overall growth of terrapins. Bivalves such as the dwarf surf clam (Mulinia lateralis) and the soft-shelled clam (Mya arenaria) were the most frequently consumed prey items for terrapins from northeastern Florida and Jamaica Bay, respectively (Butler et al. 2012; Erazmus 2012), and southern marsh clam P. floridana were the dominant food item in fecal samples from terrapins in Florida Bay (n = 7; B. Mealy, pers. comm., August 2012). Although bivalves did not rank as high in terms of FO in the present study, we found P. floridana was the most frequently consumed bivalve (Table 1) for Large terrapins; in contrast, for Small terrapins, both P. floridana and Anomalocardia sp. were both frequently consumed.

The acorn barnacle (Balanus amphitrite) was identified in multiple samples from the Creek site, whereas a smaller unidentified barnacle species was found in only one sample from the Island sites. Fish bones were found in 4 samples; however, although terrapins may be quick enough to capture small intertidal fishes, it is not known whether they actively hunt fish or opportunistically consume them as carrion (Petrochic 2009), which could account for the small number of samples containing fish.

Diet diversity indices from the prey categories showed slight variation by both habitat and size class, with the higher diversity occurring in the Small size class and the Creek habitat. This result agrees with the qualitative diversity and abundance of potential food categories observed during this study. The lowest diet diversity occurred at the Island site where the fewest potential prey categories were observed near terrapin capture locations. Differences in the diet diversity and evenness indices in Island versus Creek habitats may be a function of overall site differences in prey availability and abundance rather than selection. Potential prey items for a species can vary throughout its range based on the flora and fauna associated with each habitat it occupies. The most frequently occurring diet categories at the Creek site were crustaceans (74%) and gastropods (68%) and throughout the system the mudflats were dominated by smaller crustaceans (U. thayeri and A. pisoni), and various gastropods were found in dense aggregations on the sediment surface. Barnacles (B. amphitrite) were attached to red mangrove prop roots on the high tide line throughout the system, limiting their accessibility to the terrapins, and bivalves were found less frequently, which may explain the lower FO for those prey categories. At the Island sites, however, where terrapin diets consisted mostly of gastropods (74%) and bivalves (63%), we observed an abundance of both along the edges of the islands, whereas we seldom saw crustaceans except for large C. sapidus. Barnacles (Balanus sp.) were present on mangrove roots, but they were much smaller and reduced in abundance compared with barnacles found in the creeks and were found in only 4% of samples. Quantitative analysis of the benthic community that comprises the terrapin diet is necessary to determine the availability and abundance of food items. Research on the various aspects of terrapin ecology and how potential anthropogenic stresses may influence terrapin foraging strategies can aid managers in developing plans to address potential impacts due to habitat degradation.

Well-designed diet collection studies are necessary for adequate representation of the food web for use in interpreting ecosystem models (Metcalf et al. 2008). Because prey remains in fecal contents are thoroughly crushed and gut retention times vary, it is not possible to estimate the abundance of a particular species per sample. Therefore, caution must be used when interpreting dry mass of a prey category or species because it does not accurately reflect mass of prey before ingestion, nor does it account for any soft-bodied prey items that were not identifiable through fecal analysis. Recently, stable isotope analysis has increasingly been used to investigate diet and foraging ecology in wildlife species (Newsome et al. 2009; Williams et al. 2014; Robertson et al. 2015) and can be useful when predators consume soft-bodied organisms that cannot be identified through gut or fecal remains. Analysis of different tissues can also provide dietary information over varying time scales. Concurrent to this study, we collected samples from terrapins, potential prey, and vegetation to investigate resource use over time using stable carbon and nitrogen isotopes. Results from the 2 studies will enhance our ability to identify temporal or disturbance-induced changes in resource use, requiring fewer sampling trips and lower costs.