Diet Analysis of Subadult Kemp’s Ridley (Lepidochelys kempii) Turtles from West-Central Florida
Abstract
The present study is the first to characterize Kemp’s ridley turtle (Lepidochelys kempii) diets between the west-central Florida estuaries of Tampa Bay to Charlotte Harbor. We analyzed the stomach contents of 20 stranded subadult L. kempii and found that their diets were primarily carnivorous, consisting mostly of calico crab (Hepatus epheliticus), stone crab (Menippe mercenaria), and blue crab (Callinectes sapidus). In addition, we found uncommon prey items including fish and horseshoe crab (Limulus polyphemus), suggesting these L. kempii are opportunistic feeders. When separated into 2 size classes (25–45 [n = 9] and > 45 [n = 11] cm straight carapace length), there was a significant difference in the turtles’ diet composition. These new insights into the ecology and life history of a critically endangered sea turtle reveal essential prey resources used by a poorly studied yet crucial subadult life stage.
The conservation plans for endangered species will ideally integrate in-depth knowledge of the organisms’ life history and ecology. However, it remains challenging to design conservation plans for long-lived, migratory sea turtles that occupy a broad range of habitats and ecological niches throughout their lives.
Historically and presently, the efforts to protect and restore Kemp’s ridley (Lepidochelys kempii, Garmon 1880) populations focused on the nesting beaches by protecting nesting females and increasing the survival of nests or hatchlings (National Marine Fisheries Service [NMFS] et al. 2011). However, these turtles also face threats during the in-water life stages between the juvenile and reproductive adult phases. Therefore, conservation efforts must also take into account the precursor stages that are essential to meet a population target of 40,000 nesting females set by the Bi-National (United States and Mexico) Recovery Plan for the Kemp’s Ridley Sea Turtle (NMFS et al. 2011).
For Kemp’s ridleys and related sea turtles, matrix models identify the most-sensitive life stages in management actions to enhance overall population growth (Crouse et al. 1987; Heppell 1998). Conservation efforts aimed at restoring sea turtle populations also incorporate details from the juvenile and subadult life stages of a coastal–benthic period for the decade prior to maturation and recruitment to the breeding population (Schmid and Witzell 1997). Consequently, the necessary background that guides recovery plans must include an understanding of the essential habitats and food resources that support all the life history stages of Kemp’s ridleys.
Dietary analysis is a fundamental tool that reveals aspects of the enigmatic subadult life stage. The biological and geographic changes during ontogeny influence the turtles’ habitat preference and usage, including prey preference (Shaver 1991; Chaloupka and Zug 1997; Seney 2003, 2008; Schmid and Barichivich 2006). Kemp’s ridleys are generally accepted to be carnivorous, but diet analyses also indicate that the species is an opportunistic feeder, displaying habitat-specific feeding preferences (Hildebrand 1979; Lutcavage 1981; Shaver 1991; Burke et al. 1994; Witzell and Schmid 2005; Schmid and Barichivich 2006). Therefore, site-specific studies that address geographic gaps in knowledge contribute to a more-complete understanding of the species biology and life history. Previous diet analyses have examined Kemp’s ridleys of southwest and northwest Florida (Carr and Caldwell 1956; Schmid 1998; Witzell and Schmid 2005; Schmid and Barichivich 2006). However, studies remain lacking on west-central Florida, where subadult Kemp’s ridleys are known to occur.
The present study is the first to report on diets of L. kempii for the west-central region of Florida. We contrasted the diet of small and larger subadult turtles to describe possible changes in prey use and habitat preference during the juvenile and subadult life stages.
METHODS
Study Site
The west-central coast of Florida is bounded by the major estuaries of Tampa Bay to Charlotte Harbor and contains the coastal counties of Manatee, Sarasota, and Charlotte (Fig. 1). The continental shelf off this region’s coast is broad (90–110 km), gently sloping, and hosts a patchwork of offshore habitat and coastal benthic regions known as foraging grounds for juvenile, subadult, and adult Kemp’s ridleys (Poag 2015). The habitat types include sandy mud bottom, seagrass meadow, hard bottom (reefs), rock outcropping, and oyster bar (Culter and Leverone 1993). Fewer than 10 L. kempii nests are deposited annually along the entire southwest Florida nesting beaches, and the region provides important feeding and nesting habitat for green (Chelonia mydas) and loggerhead (Caretta caretta) turtles (Schmid 1998).
Citation: Chelonian Conservation and Biology 14, 2; 10.2744/CCB-1177.1
Sample Collection
Mote Marine Laboratory’s (Sarasota, Florida) Stranding Investigations Program (SIP) collected stranded Kemp’s ridleys during its operations in Sarasota County (n = 17) or in adjacent regions of southern Manatee County (n = 2) and northern Charlotte County (n = 1) over the period 2005–2012 (Fig. 1). All specimens were dead upon recovery and returned to the laboratory for examination and detailed necropsy. Stranding and morphometric data were collected according to the Florida Fish and Wildlife Conservation Commission Sea Turtle Stranding and Salvage Network data sheet. The straight carapace length (SCL) was measured from the nuchal notch to the posterior tip. The stomach or entire gastrointestinal contents were removed during this process and stored in 80% ethanol.
We separated samples into 2 size categories: 20–45 cm SCL and > 45 cm SCL. The categories were advised by studies of Kemp’s ridleys using growth models, mark–recapture techniques, and dietary information. Schmid and Barichivich (2006) noted distinct growth phases corresponding with geographic and ontological shifts. Specifically, they found a notable increase in growth rate at around 46 cm SCL, possibly indicating a habitat change at this size. A polyphastic growth model built by Chaloupka and Zug (1997) supported these data and suggested a shift in habitat use and onset of puberty around 40–50 cm SCL. Moreover, differences in the sizes of turtles between Schmid and Barichivich’s western Florida study and the present study, caused by variation in environmental conditions and nutritional uptake (known to affect growth rate), should be minimized because of the geographic proximity of the study sites (Bjorndal et al. 2003; Snover et al. 2007).
Sample Quantification
We rinsed stomach contents with tap water through a 5-mm-mesh sieve. Finer contents that passed through the sieve contributed < 1% of the total volume of stomach contents and were excluded from further analysis. We identified all retained prey items to the lowest taxonomic level possible with identification guides to regional marine fauna (Kaplan 1988).
We separated contents of each stomach sample by prey taxa. We obtained wet volume measurements of each taxon using a water displacement method in a 500-ml graduated cylinder. The strandings from 2005 to 2011 (n = 18) had only their stomach contents removed while the strandings in 2012 (n = 2) had their entire gastrointestinal tract contents removed. Removal of the entire gastrointestinal tract yielded a much higher volume of consumed prey because the esophageal and intestinal contents were also included. Because of this difference in sample volume, we excluded the 2, 2012 samples for volumetric calculations but included them in all other quantitative estimates.
Data Analysis
We calculated percentage by number (%N), frequency of occurrence (%F), and volume (%V) for each identified prey taxon. The %N of a prey taxon is the number of individuals of that prey taxon divided by the total number of individuals of all prey taxa found in all 20 turtle stomachs. The %F is the number of stomachs containing the taxon divided by the total number of turtles (n = 20). The %V of a prey taxon is the total wet volume of the taxon divided by the total wet volume of all Kemp’s ridley stomach contents. In a similar fashion, we separated samples into 2 size categories and calculated %N, %F, and %V of each prey taxon for the size categories.
We calculated a percent index of relative importance (%IRI) that incorporates %N, %F, and %V calculations and it was used to gauge the importance of each prey taxon:
where IRIi is (%Ni + %Vi) × %Fi for prey type i and 
is the sum of the IRI for all prey types i (Hyslop 1980). The %IRI was first calculated for each prey taxon within the entire sample of 20 Kemp’s ridleys and then again for each prey taxon that occurred within the 2 size classes separately.
To identify prey taxa that contributed significantly more to the diets of one size class than to the other, we calculated chi-square (χ2) tests for each prey taxon using %F as well as %V data.
We created cumulative prey curves for each size category to gauge whether an appropriate number of samples had been collected and, thereby, would accurately represent the entire range of prey species consumed by other Kemp’s ridleys of the same demographic. We constructed prey curves separately for each size category.
We conducted a canonical correspondence analysis (CCA) and complimentary Monte Carlo permutation test to elucidate any significant variation in overall diet between the size categories. We performed the analysis with prey taxa presence–absence data using R Studio version 0.95.263 (https://www.rstudio.com/), including the “reshape” 0.8.4 data manipulation package and the “vegan” 2.0-3 package. In addition, we estimated percent dietary overlap between the 2 size categories using Schoener’s Index (Schoener 1970):
where pxi is the proportion of prey category i in turtle size class x and pyi is the proportion of the same prey category i in turtle size category y. Index values greater than 60% suggest substantial dietary overlap (Wallace and Ramsey 1983).
RESULTS
The stranded Kemp’s ridleys (n = 20) ranged from 23.6 to 65.0 cm SCL, with a mean size of 45.9 cm SCL ± 3.1 cm SE (Fig. 2). The stomach contents contained 23 prey taxa belonging to 7 broad categories: 1) horseshoe crabs (%F = 15.0; %V = 7.7%); 2) crustaceans including mantis shrimp, identified crab species, and crab parts that could not be identified to the species level (%F = 90.0; %V = 180.0); 3) sea urchin (%F = 5.0; %V = 3.5); 4) fish that could not be indentified to the species level owing to advanced decomposition from digestion (%F = 25.0; %V = 8.7); 5) molluscs including gastropod and bivalve species (%F = 50.0; %V = 1.7); 6) plant matter (%F = 15.0; %V < 0.0); and 7) unidentified matter (%F = 50.0; %V = 15.4), which consisted primarily of soft animal tissue.
Citation: Chelonian Conservation and Biology 14, 2; 10.2744/CCB-1177.1
Of the 23 taxa, the greatest prey frequency was constituted by calico crab, stone crab, and blue crab (%F = 60.0, 55.0, and 55.0, respectively). The greatest amount for wet volume of prey was constituted by calico crab, stone crab, and blue crab along with unidentified matter (%V = 21.3, 14.7, 11.9, and 15.4, respectively; Fig. 3).
Citation: Chelonian Conservation and Biology 14, 2; 10.2744/CCB-1177.1
The smaller turtle categories consumed primarily fish (%IRI = 29.8) and stone crab (%IRI = 12.3), whereas the larger categories consumed primarily calico (%IRI = 30.5), blue crab (%IRI = 22.3), and stone crab (%IRI = 16.8). Unidentified matter was also among the top prey categories in both the smaller and larger size categories (%IRI = 38.3 and 6.7, respectively) (Table 1).
Six prey categories varied significantly in %V between the 2 size categories. Horseshoe crab (χ2 = 6.085, p = 0.014), blue crab (χ2 = 7.955, p = 0.005), stone crab (χ2 = 3.956, p = 0.047), and calico crab (χ2 = 8.144, p = 0.004) contributed significantly more in %V to the diets of the larger size category than to the smaller. Conversely, fish (χ2 = 33.424, p < 0.0001) and unidentified matter (χ2 = 20.855, p < 0.0001) contributed significantly more in %V to the diets of the smaller size category (Fig. 4). Additionally, blue crab and speckled crab occurred significantly more often within the larger size category than in the smaller (χ2 = 4.455, p = 0.035; χ2 = 4.000, p = 0.046, respectively).
Citation: Chelonian Conservation and Biology 14, 2; 10.2744/CCB-1177.1
The cumulative prey curves for both size categories reached moderate horizontal asymptotes, suggesting a reasonable number of Kemp’s ridleys had been sampled for estimating prey taxa richness (Fig. 5).
Citation: Chelonian Conservation and Biology 14, 2; 10.2744/CCB-1177.1
In addition, the CCA and Schoener’s Index, used to compare overall diet between the size classes, both indicated notable differences in prey consumption between the larger and smaller categories. The CCA and corresponding permutation test (599 permutations) showed a significant dietary difference (p = 0.028, F1,18 = 1.549) between the size categories (Fig. 6), which is visually represented by 2 distinct point clusters on the plot. A Schoener’s Index value of 56.4% supported these results by indicating no substantial dietary overlap between small and large size categories.
Citation: Chelonian Conservation and Biology 14, 2; 10.2744/CCB-1177.1
Lastly, 15 (n = 8 for 20–45 cm SCL, n = 7 for > 45 cm SCL) Kemp’s ridleys in this study were recovered during a severe harmful algal bloom (HAB; see “Discussion”). Given the small sample of turtles recovered outside of the HAB season (n = 1 for 20–45 cm SCL, n = 4 for > 45 cm SCL), which is dominated by turtles of the larger size class, a comparison of dietary composition of HAB and non-HAB turtles and broader conclusions of possible changes in turtle diet because of the HAB would be inappropriate. However, to determine if size class differences in dietary composition are confounded by collection during the HAB season, we conducted an additional CCA comparing dietary composition by size class of the 15 turtles recovered during the HAB season. Results of this analysis are consistent with those observed among all 20 turtles, with a significant difference (p = 0.015, F1,13 = 1.563) in dietary composition between small and large Kemp’s ridleys recovered during the HAB season (Fig. 7).
Citation: Chelonian Conservation and Biology 14, 2; 10.2744/CCB-1177.1
DISCUSSION
Overall Diet
We based the present study upon chance encounters of infrequently stranded species and therefore admit that a larger sample would be desirable. Nevertheless, the shape of the cumulative prey curves suggest that the study’s sample sizes were adequate to determine that these size categories of L. kempii were primarily carnivorous, feeding on decapod crustaceans living in the benthos. The calico crab, stone crab, blue crab, and unidentified matter (primarily soft animal tissue) were consumed most frequently and constituted the greatest volume. This primary finding of our study concurs with previous dietary studies of Kemp’s ridleys in the Gulf of Mexico (Shaver 1991; Werner 1994; Schmid 1995; Witzell and Schmid 2005; Seney 2008) and US Atlantic coastal waters (Lutcavage 1981; Burke et al. 1994; Seney 2003; Seney and Musick 2005) that also suggest Kemp’s ridleys consume primarily crab.
Despite this general agreement of our findings with earlier studies, there were also unique prey categories from our study that require further interpretation. The unusual presence of fish in 25% (5/20) of samples and horseshoe crab in 15% (3/20) may suggest opportunistic feeding patterns. For example, Witzell and Schmid (2005) examined fecal samples from 64 L. kempii in Ten Thousand Islands of southwest Florida and reported that 72.7% consumed a benthic tunicate species. Tunicates had never before been reported as L. kempii prey, and so these turtles were thought to be foraging opportunistically on this abundant food source.
Fish are considered uncommon sea turtle prey and thus credible scenarios may be considered. A first scenario with fish ingestion involves the massive fish mortalities during the HABs that are common in the eastern Gulf of Mexico (Glibert et al. 2005; Perrault et al. 2014). HABs along Florida’s southwest coast cause episodes of high mortality in fish, sea turtles, birds, bottlenose dolphins, and manatees (Gunter et al. 1948; Bossart et al. 1998; Foote et al. 1998; Kreuder et al. 2002; Flewelling et al. 2005; Gannon et al. 2009). The most prevalent type of HAB in Florida is caused by the dinoflagellate Karenia brevis, which produces brevetoxins (Baden 1989). HABs have recently become more frequent, long lasting, and widespread in the Gulf of Mexico, particularly along the western coast of Florida (Kirkpatrick et al. 2004). Fish that swim through the algal masses or ingest prey are contaminated with the toxin and often die (Glibert et al. 2005). Mass fish kills are a common occurrence during HAB seasons and potentially provide predators, such as sea turtles, an easily catchable, high-energy food source. Fifteen of the 20 Kemp’s ridleys in this study were recovered between February and August 2005, during one of the most severe episodes of HABs on record, with massive fish kills (Alcock 2007). While the small sample size (n = 5) of Kemp’s ridleys recovered outside of the 2005 HAB season precludes a confident assessment of the diets of HAB and non-HAB individuals, the 5 turtles in the present study that consumed fish were recovered during this period.
A second scenario involves possible interactions with trawler or hook-line fisheries. Previous Kemp’s ridley dietary studies that report the consumption of fish suggest the turtles were foraging on dead bycatch or bait from trawl fisheries activities, as sea turtles generally lack the speed and agility to catch live fish (Shoop and Ruckdeschel 1982; Werner 1994; Seney 2003, 2008). However, the trawl fishery in nearshore west-central Florida is not presently active, so we should more-closely examine alternatives to fish consumption of trawl fishery bycatch. A potential for interactions with a hook-line fishery is based upon SIP records over 20 yrs involving 36 Kemp’s ridleys documented with direct evidence of human interactions from hook and line fisheries or boating activities (G. Lovewell–SIP, unpubl. data, 2014). Entanglement with fishing gear and debris affected 36% (13/36) of L. kempii strandings and the remaining 64% showed clear signs of a boat strike; 35 of 36 Kemp’s ridleys were subadults and, if these animals are specifically targeting bait and depredating fishing lines, this already critically endangered species could be facing even greater conservation implications.
In addition, the L. kempii in the present study may have been feeding opportunistically on an abundant source of horseshoe crabs, a coastal benthic-dwelling species known to occur in Florida foraging habitat (Carmichael et al. 2004). Similar to fish, horseshoe crabs are also impacted by HABs, and the 3 Kemp’s ridleys that consumed this species were recovered during the 2005 HAB season. Horseshoe crab consumption by Kemp’s ridleys in Charlotte Harbor (Barleycorn and Tucker 2005) and the Chesapeake Bay (Seney and Musick 2005) also suggests opportunistic feeding behaviors and potentially niche partitioning with competing predators (Seney and Musick 2005). Lutcavage and Musick (1985) and Seney (2003) report suspected habitat partitioning among Kemp’s ridleys and loggerhead sea turtles consuming decapod crustaceans and horseshoe crabs in Chesapeake Bay and Virginia, respectively. Loggerhead sea turtles are abundant along the west coast of Florida and also consume benthic invertebrates. A comparative analysis of similar data on loggerhead diet in west-central Florida (which have been collected but not yet compiled), and historical data on prey and predator species abundance and distribution in the west-central Florida region, may help to discern whether our samples were from Kemp’s ridleys that were foraging opportunistically or selectively.
Diet and Turtle Size
The diets of the smaller and larger categories of ridleys were clearly separated in the present study, both by %F, %V, %IRI and by the top contributing prey taxa. The primary prey resources of the smaller L. kempii were fish and decapod crustaceans including stone crab, calico crab, and unidentified crab. The larger turtles also consumed decapod crustaceans, but the species ranks differed as calico crab, blue crab, stone crab, and spider crab. The size variation in diet may represent a product of different physical and learned capacities. For example, larger L. kempii have greater jaw dimensions, speed, and possibly more practice foraging on the fast-moving, large decapod crustaceans than do the smaller individuals. If these physical characteristics and learned foraging skills are valuable, then feeding opportunistically on dead fish instead of on the more common decapod crustacean prey may represent a resourceful shift in foraging strategy by the smaller Kemp’s ridleys. The importance of fish to the diets of the smaller size class is illustrated well by the CCA. Within the CCA biplot (Fig. 6), the clump of points corresponding to the smaller Kemp’s ridleys is a single, larger individual which represents the only individual within the larger size class to have consumed fish.
Size-related differences in diet of Kemp’s ridleys may reflect intraspecific niche partitioning in which the smaller and larger categories avoided competition by foraging upon different prey sources or in different ranges of the coastal habitat. While there are limited data about the local movements and ranges of subadult Kemp’s ridleys, a sonic transmitter tracking study of Kemp’s ridleys inhabiting Waccasassa Bay, Florida indicated no significant correlation between carapace length and home-range foraging area (Schmid et al. 2003). If similar foraging movements were occurring in the west-central Florida region, the smaller, fish-foraging Kemp’s ridleys were not necessarily using a spatially distinct or larger foraging habitat but, instead, exploited the same habitat range differently.
Our results must be interpreted cautiously because of the constrained sample size. Additional dietary studies, as well as the quantification of prey source distributions and abundances within the region, could provide valuable information regarding the habitat use and preferences of the subadult Kemp’s ridleys. Isotopic analysis of Kemp’s ridleys and their prey items (Schmid et al. 2013), tracking studies (Schmid et al. 2002, 2003), and other types of diet analyses such as analysis of fecal and stomach lavage samples (Werner 1994; Schmid 1998; Witzell and Schmid 2005; Schmid et al. 2013) have already provided important information regarding the diets and habitat use of these turtles in west Florida. These detailed analyses could also be employed throughout L. kempii range to compare and contribute novel insights. Such knowledge will inform conservation decisions on the protection of important foraging habitat for subadult Kemp’s ridley as well as on the regulation of fisheries activities that affect important prey sources for this critically endangered species.
Coastal stranding locations of 20 subadult Kemp’s ridley sea turtles used in this study. The turtles were separated into 2 size classes based on their straight carapace length (SCL): 20–45 cm (n = 9) and > 45 cm (n = 11).
Size distribution of the stranded Kemp’s ridleys (n = 20).
Percent frequency (%F, white bars) and percent volume (%V, black bars) contribution of each prey category identified within the stomachs of Kemp’s ridleys.
Percent volume (%V) of prey categories for the smaller (black bars) and larger (white bars) Kemp’s ridley size classes. Only prey categories that contributed > 1% to the total volume of a size class are shown. Asterisks indicate prey categories with a significant difference in %V between the 2 turtle size classes (* p < 0.05; ** p < 0.01; **** p < 0.0001).
Cumulative prey curves for stomach sample contents of Kemp’s ridleys (top) 20–45 cm in straight carapace length (SCL) (n = 9), and (bottom) > 45 cm SCL (n = 11).
Canonical correspondence analysis (CCA) biplot showing a significant difference in overall diets between the two Kemp’s ridley size classes (n = 0.028; F1,18 = 1.549). Each point represents an individual Kemp’s ridley within either the smaller (black points) or larger (white points) size classes designated by centimeters in straight carapace length (SCL).
Canonical correspondence analysis (CCA) biplot showing a significant difference in overall diets between smaller (black points) and larger (white points) Kemp’s ridleys recovered during a harmful algal bloom (n = 0.015, F1,13 = 1.563). Size classes designated by centimeters in straight carapace length (SCL).
Contributor Notes
Handling Editor: Jeffrey A. Seminoff
