First report of lionfish prey from Western Florida waters as identified by DNA barcoding

Introduction

Native to the Indo-Pacific, lionfish (Pterois volitans and P. miles) have quickly become established in reefs and coastal habitats throughout the Western Atlantic since their introduction in the 1980s. Their settlement in this region has raised concerns of severe detrimental effects on local fauna, as lionfish have been shown to drastically reduce recruitment of native coral reef fishes (Albins & Hixon, 2008), exhibit high prey consumption rates (Green, Akins & Côté, 2011), prey on native fishes and invertebrates (Green, Akins & Côté, 2011; Côté et al., 2013; Dahl & Patterson, 2014; Harms-Tuohy, Schizas & Appeldoorn, 2016; Dahl et al., 2017) including the critically endangered wrasse Halichoeres socialis (Rocha et al., 2015), and drive declines in reef fish populations (Green et al., 2012; Albins, 2013, 2015). Controlled experiments exemplify these concerns, as they have shown lionfish to drive abundance declines and loss of species in Bahamian reefs at higher rates than native predators of similar sizes (Albins, 2013) and population declines of >50% in native species in experimental replicates (Ingeman, 2016). These findings underscore the importance of defining the prey preferences of lionfish in the Western Atlantic, specifically in areas where lionfish are relatively understudied.

In the Gulf of Mexico, settlement and spread of lionfish is thought to only have occurred over the past ten years. The first record of lionfish in the Gulf of Mexico (2006) is from a carcass found off the coast of Pinellas County, FL and is believed to represent a short-term individual (Schofield, 2010). In 2009 however, two lionfish were captured in the western Gulf of Mexico off the coast of Yucatan, Mexico (Aguilar-Perera & Tuz-Sulub, 2010), a record thought to be the first resulting from larval dispersal into the Gulf. Within a year of this capture, lionfish were reported along the U.S. coastline of the Gulf of Mexico. In August of 2010, lionfish were reported from waters near Cortez in Manatee County, FL (Schofield, 2010). By September of 2010, lionfish had been confirmed in the northern Gulf of Mexico, off the coasts of Florida and Alabama (Schofield, 2010).

Although widely studied in other areas of the Western Atlantic, particularly in the Caribbean, research into the predatory habits within the Gulf of Mexico remains comparatively sparse perhaps as a result of their relatively shorter invasive history. To date, documentation of lionfish prey species in the region have come from the Yucatan (Valdez-Moreno et al., 2012) and the northern Gulf of Mexico (Dahl & Patterson, 2014; Dahl et al., 2017, 2018). These studies, though varied in their sampling regime and identification approaches (i.e., molecular v. visual), indicate that lionfish in the region feed primarily on small fishes and to a lesser extent on some crustacean species with some variation associated with lionfish ontogeny. These findings are largely concordant with DNA barcoding results of lionfish prey items found in other Western Atlantic habitats, including Belize (Rocha et al., 2015), Puerto Rico (Harms-Tuohy, Schizas & Appeldoorn, 2016), and The Bahamas (Côté et al., 2013). Despite several identified taxa common among these studies, regional differences appear to exist between sites as each study has identified novel lionfish prey species. Thus, surveys of lionfish stomach contents in yet-to-be studied areas are needed to further our understanding of lionfish prey diets and preferences. This research would be valuable for understanding the local impacts of lionfish predation (Mizrahi et al., 2017) and for informing local management of lionfish (Bogdanoff et al., 2018). Towards this goal, we applied Cytochrome Oxidase I (hereafter COI) barcoding to prey fragments found in the stomachs of lionfish collected during the 2016 Sarasota Lionfish Derby where no previous studies on lionfish diets had been conducted and near the area where lionfish presence was first documented in the Gulf of Mexico.

Materials and Methods

Specimen collection

Lionfish were collected during the 2016 Sarasota Lionfish tournament in artificial habitat constructed of limestone rocks piled over the Gulfstream natural gas pipeline approximately 30 miles west of St. Petersburg, FL in 30 m water depth (Fig. 1). The limestone rubble topping sections of the pipeline creates an artificial reef in an otherwise unconsolidated sandy substrate with occasional natural hardbottom features. Lionfish were stored on ice until the removal of stomachs which was completed within 24 h of capture. Upon removal, stomachs were placed in individually labeled Whirl-Pak bags and stored at −20 °C until dissections were to be performed. Prior to dissections, the Total Length (hereafter TL) of each lionfish was recorded. Collections were carried out in accordance with all applicable local, state, and federal laws and regulations.

Results

A total of 70 stomachs were dissected from lionfish ranging in size from 140 to 400 mm total length (TL; mean_TL= 271.71 mm (SD = 69.86); median = 270 mm; mode = 269 mm; Interquartile Range: 221.5−332.5 mm). Twenty of the stomachs did not contain any observable prey fragments (28.6%). These individuals did not statistically differ in size from those which contained preys fragments (t(68) = −1.48987; p-value = 0.1410). In the remaining fifty stomachs (71.4%), we recovered 305 prey fragments (per stomach mean = 6.1, median = 4, mode =1). Of these, 184 prey fragments from 49 lionfish stomachs produced sequences meeting the minimum quality criteria (e.g., short sequences, evidence of cross-amplification). The remaining 121 fragments either failed to amplify despite several PCR attempts or produced sequences of low quality. This represents a 60.3% success rate of amplification. Most fragments were highly digested (RD: median = 4.5; mode = 5), preventing effective visual identification. Fragments that failed to amplify and/or produced low quality sequences were slightly more digested (median RD = 4.5; mode = 5) than those successfully amplified (median RD = 4; mode = 4); however, this difference was not significant (t(303) = 0.3253, p = 0.7452).

BOLD-based identifications produced species or genus level-matches for all but three sequences; 67.4% of all sequences were identified to species and 31.5% to a genus. Of these, sixteen fragments were identified as lionfish (P. volitans). Although cannibalism is known to occur in lionfish in the Western Atlantic (Valdez-Moreno et al., 2012; Dahl et al., 2018), we were unable to determine whether sequences identified as P. volitans were the result of cannibalism or cross-contamination. We thus removed these sequences from further analyses. Remaining fragments represented 21 fish species and six crustacean species. Fish prey species accounted for most obtained sequences (95.7%) and were present in the majority of lionfish stomachs (N_stomachs = 41). The most commonly observed fish species were Ptereleotris calliura (Blue goby; N_stomachs = 12; 24.3% of stomachs containing prey fragments), an unidentified species of Microgobius (N_stomachs = 10; 20.4%), Diplectrum formosum (Sand perch; N_stomachs = 7; 14.3%), Apogon aurolineatus (Bridle cardinalfish; N_stomachs = 6; 12.2%), Heamulon aurolineatum (Tomtate grunt; N_stomachs = 5; 10.2%). At the family level, the 21 fish species represented fourteen families. The most frequently observed fish families were Microdesmidae (Dartfishes; N_stomachs = 12; 24.5%), Apogonidae (Cardinalfishes; N_stomachs = 11; 22.4%), Gobiidae (True gobies; N_stomachs = 11; 22.4%), and Serranidae (Seabasses and Groupers; N_stomachs = 8; 16.3%). Crustacean species belonged to six crustacean families all in the Class Malacostraca. With the exception of Metapenaeopsis goodei (Caribbean velvet shrimp; N_stomachs = 3; 6.1%), all crustacean species were observed in a single stomach. Details of the occurrence of each species and family found in lionfish stomachs are presented in Table 1.

Discussion

Previous studies have shown lionfish in the Western Atlantic feed primarily on fish species and that a large degree of overlap in the taxonomic composition of prey exists across the region (Valdez-Moreno et al., 2012; Côté et al., 2013; Rocha et al., 2015; Harms-Tuohy, Schizas & Appeldoorn, 2016; Dahl et al., 2017, 2018; Sancho et al., 2018). By applying molecular approaches to identify prey fragments recovered from lionfish stomachs collected in the Gulf of Mexico off the South Florida coastline we report patterns broadly consistent with previous findings. Our results also demonstrate that adult lionfish in the study region preyed primarily on fish rather than invertebrates, as reflected in both the frequency in which each prey type was observed (i.e., 41 stomachs contained at least one fish prey item, while six stomachs had at least one invertebrate prey item) and the taxonomic richness of each prey type (e.g., 21 fish prey species compared to six crustacean species). Most prey species identified in this study were previously reported as lionfish prey in the Western Atlantic (Peake et al., 2018). The three higher taxa found in >20% of surveyed lionfish stomachs in this study were prevalent in lionfish stomachs in other areas of the Western Atlantic: with Apogonidae being prevalent in Puerto Rico (Harms-Tuohy, Schizas & Appeldoorn, 2016) and the northern Gulf of Mexico (Dahl et al., 2017), Gobiidae in Belize (Rocha et al., 2015), the Yucatan peninsula (Valdez-Moreno et al., 2012), the Bahamas (Côté et al., 2013), Biscayne National Park (Sancho et al., 2018), and the northern Gulf of Mexico (Dahl et al., 2017), and Microdesmidae in the Northern Gulf of Mexico (Dahl et al., 2017).

Such congruence in the taxonomic composition of lionfish prey amongst studies underscores that while generalists, these predators exhibit preferences for specific prey traits. Most fish prey species identified in this study share at least one of the following characteristics: they are reef-associated or demersal species with average sizes of <15 cm and are mostly active at night or during crepuscular times (Matheson et al., 2017). This is best exemplified by the most frequently observed prey taxa in this study, Ptereleotris calliura. This species was observed in ~25% of lionfish stomachs and is a small burrowing fish which does not exceed 12.5 cm in length (Robins & Ray, 1986). Similarly, Apogon species are small, nocturnal, reef associated fish (Hoese & Moore, 1998) and Microgobius are small reef-associated fish that may display burrowing behaviors (Birdsong, 1981). Diplectrum formosum while growing to relatively larger adult sizes than the above prey species, is a demersal species commonly associated with sandy and coarse gravel bottoms (Hoese & Moore, 1998). Lastly, Haemulon aurolineatum is a nocturnal reef-associated species that are <11 cm long in the juvenile stage (Collette & Talbot, 1972; Manooch & Barans, 1982). Such apparent preferences are in concordance with the findings of Green & Côté (2014), who using a trait-based model, identified nocturnal, small, shallow-bodied, solitary fishes found resting on or just above reefs as the most vulnerable to lionfish predation. Thus, our findings support the idea that lionfish, while generalists, exhibit a preference for small, nocturnal, reef associated fish prey in the Western Atlantic (see Peake et al. (2018) and references therein).

It is important to note that despite the overlap amongst results and past studies, we report patterns that suggest regional differences in lionfish diets may exist. We identified eleven putative novel lionfish prey taxa, nine species not identified in previous barcoding studies of lionfish stomach contents in the Western Atlantic [Diplogrammus pauciradiatus (Spotted dragonet), Halichoeres caudalis (Painted wrasse), Hypoplectrus floridae (Florida Hamlet), Phaeoptyx xenus (Sponge cardinalfish), Sparisoma atomarium (Greenblotch parrotfish), Synodus saurus (Atlantic lizardfish); Crustacean: Achelous ordwayi (Redhair swimming crab), Metapenaeopsis goodei (Caribbean velvet shrimp), Neogonodactylus bredini (Caribbean rock mantis shrimp)] and two unidentified species from genera known to be lionfish prey (Chaenopsis sp. and Microgobius sp.). Based on observations reported in the Smithsonian Tropical Research Institute Online Information System (Robertson & Van Tassell, 2019) as well as the Global Biodiversity Information Facility databases (www.gbif.org), most of these novel prey taxa appear to be more prevalent in Florida waters than in those previously surveyed by barcoding studies. These findings indicate at least a partial effect of differences in regional assemblages in lionfish prey composition. Indeed, although comparisons across barcoding studies of lionfish in the Western Atlantic are hampered by the application of different methodologies and the reporting of different frequency statistics, both prey composition and frequencies observed in our study appear to be the most similar to those reported by Dahl et al. (2017) from nearby Northern Gulf of Mexico waters (see Tables S1 and S2). Lastly, most prey species reported in this study are species frequently to commonly found in reef-assemblages in Florida waters (Smith et al., 1975; Darcy & Gutherz, 1984; Matheson et al., 2017). These patterns thus give credence to the Harms-Tuohy, Schizas & Appeldoorn (2016) suggestion that lionfish diets are a result of regional ichtyofauna and their preferences for specific traits. As such, continued research is important to elucidate lionfish diets in unstudied regions and habitats to understand local impacts on species and communities.

Similarly, future studies should investigate whether lionfish of various size classes prey upon different taxa. Although ontogeny appears unlikely to have affected results reported herein given the size distribution of lionfish sampled, ontogenetic shifts in diet are known to occur in lionfish with studies indicating a shift from an invertebrate-to a fish dominated diet with increase in size (Morris & Akins, 2009; Muñoz, Currin & Whitfield, 2011; Dahl & Patterson, 2014; O’Farrell et al., 2014; Peake et al., 2018; Sancho et al., 2018; Malpica-Cruz, Green & Côté, 2019). For instance, Peake et al. (2018) reported significant negative and positive relationships between lionfish size (as measured by standard length) and the contributions of shrimp and fish prey respectively to lionfish diets in the Arrecifes de Cozumel National Park. Similarly, Sancho et al. (2018) reported that in Biscayne Bay National Park, lionfish sized <180 mm TL exhibited diets dominated by shrimp prey as measured by Index of Relative Importance and those larger than 180 mm TL exhibited diets dominated by fish. Sancho et al. (2018) also reported some differences in prey composition across size classes (e.g., pleocyemate shrimp only preyed by smaller lionfish, dendrobranchiate shrimp by larger classes); however, the extent to which ontogeny affects the composition of lionfish prey remains unclear. Future barcoding studies may aid in elucidating this point as well as the drivers of any such changes.

Conclusion

In this study, we provide the first report of the lionfish prey-items from the Florida coastlines of the Gulf of Mexico. Our results are largely congruent with previous findings that suggest lionfish are generalists whose diet reflects regional species compositions and prey preferences for specific behavioral and morphological traits. Despite a limited number of samples and collection period, the results of our study were quite similar to the taxonomic composition of prey items reported by Dahl et al. (2017) from the Northern Gulf of Mexico, and suggest the former to be the most likely explanation. We report eleven putative novel prey species of lionfish, possibly reflecting the local prey availability in the habitat where lionfish were collected. We suggest molecular based efforts to understand the dietary habits of lionfish throughout their non-native region continue, with lionfish derbies serving as a cost-effective way to leverage public events into producing scientific information.

Supplemental Information