Genetic evidence supports a range extension for the Brazilian cownose ray Rhinoptera brasiliensis in the western North Atlantic

Abstract

We report 24 new records of the Brazilian cownose ray Rhinoptera brasiliensis outside its accepted geographic range. Sequencing of a 442-base pair portion of the mitochondrial NADH dehydrogenase subunit 2 gene for 282 Rhinoptera samples revealed eight records off the east coast of the USA and 16 from the eastern Gulf of Mexico. Both sexes of all life stages were documented in all seasons over multiple years in the Indian River and Lake Worth lagoons, Florida, indicating that their range extends further in the western North Atlantic than previously described.

The cownose rays (Rhinopteridae) are comprised of eight morphologically similar species in the genus Rhinoptera (Last et al., 2016). Rhinopterids are circumglobally distributed, with two recognized species in the western Atlantic: the Endangered Brazilian cownose ray Rhinoptera brasiliensis Müller 1836 (Vooren & Lamónaca, 2004) and the Near Threatened American cownose ray Rhinoptera bonasus (Mitchill 1815) (Barker, 2006). Both species are benthopelagic and indigenous to tropical and temperate shallow waters along continental shelves (Last et al., 2016). Historically, tooth series counts were used to distinguish between these two species, despite ambiguity arising from overlap in series counts (Bigelow & Schroeder, 1953). Recent research documented changes in tooth morphology with age, making this characteristic unreliable (Jones et al., 2017). More reliable methods of distinguishing R. brasiliensis from R. bonasus, such as cranial anatomy and number of spiral valve lamellae, necessitate either dissection or advanced analytical techniques, such as computed tomography or magnetic resonance imaging (Jones et al., 2017).

In lieu of clear, external diagnostic characteristics to distinguish Rhinoptera spp., species identifications in the western North Atlantic have largely relied on assumed geographic ranges. This approach, however, grossly underestimated the range of R. brasiliensis in the western Atlantic. The range of R. bonasus extends from New England in the United States to northern Argentina (Last et al., 2016). Historically, R. brasiliensis was considered endemic to Brazil, restricted to the waters of southern Brazil (Barker, 2006; Bigelow & Schroeder, 1953). Recent literature extended the range of R. brasiliensis to include Central America (McEachran & Carvalho, 2002), the southern Gulf of Mexico (Palacios-Barreto et al., 2017), and the northern Gulf of Mexico (Jones et al., 2017), using a combination of morphological and molecular methods in species identifications. This range extension of R. brasiliensis substantially increased overlap in the ranges of R. brasiliensis and R. bonasus, and given the absence of clear field characteristics, the range of R. brasiliensis is still poorly understood. Given this uncertainty, existing datasets of Rhinoptera spp. in the western Atlantic may reflect a species complex rather than single species, which could lead to the use of spurious data from biological and ecological studies in their management. Here, we used genetic methods to identify Rhinoptera spp. to evaluate the extent of occurrence of R. brasiliensis beyond its currently described range in the western North Atlantic, specifically in the eastern Gulf of Mexico (hereafter eGOM) and along the east coast of the United States (hereafter Atlantic).

A total of 282 archived or opportunistically collected Rhinoptera spp. tissue samples collected between 2002 and 2018 were analysed from 12 locations in the eGOM and Atlantic, spanning from Santa Rosa Sound, Florida, to Chesapeake Bay, Virginia (Figure 1). Rhinoptera spp. were caught throughout the year, primarily by gillnet, and fin clips were collected and stored in 95% ethanol. Total genomic DNA was extracted from ~15 mg of tissue using a Qiagen DNeasy DNA extraction kit (Hilden, Germany), according to the manufacturer’s protocol, with the exception that tissue samples were digested overnight. A 442-base pair region of the mitochondrial (mtDNA) NADH dehydrogenase subunit 2 (ND2) gene was targeted using a forward primer (RhinND2F1: 5’-GAACCCYTTAATCCTCTYCATC-3’) designed by McDowell and Fisher (unpubl. data) and a reverse primer (RayND2R: 5’-GGATTGATAGTACGCCTATGG-3’) designed for this study. Polymerase chain reaction (PCR) mixtures contained 25–50 ng template DNA, 10 mM Taq buffer (Invitrogen, Carlsbad, California, USA), 1.5 mM MgCl₂, 0.3 μM of each primer, 0.1 mM deoxynucleotide triphosphates (dNTP mix, Promega, Madison, Wisconsin, USA), 1 U of Taq polymerase (Invitrogen) and PCR-grade water for a final reaction volume of 25 μL. PCR cycling conditions consisted of an initial denaturation at 95°C for 5 min, then 35 cycles of 30 s denaturation at 95°C, 30 s annealing at 58°C, 30 s extension at 72°C, followed by a 5 min final extension at 72°C. Amplicons were cleaned using 3 μL of ExoSAP-IT (ThermoFisher, Waltham, Massachusetts, USA) and sequenced in the forward and reverse directions on an Applied Biosystems 3730XL DNA Analyser using a BigDye Terminator 3.1 Cycle Sequencing Kit (Applied Biosystems, Foster City, California, USA) following the manufacturer’s protocol except that all sequences were run using half reactions.

FIGURE 1.

Rhinoptera spp. sampling sites in the western North Atlantic. (a) Triangles represent general locations where the Brazilian cownose ray R. brasiliensis Müller 1836 and the American cownose ray R. bonasus (Mitchill 1815) co-occurred and circles represent general locations where only R. bonasus occurred. TAM, Tampa Bay; n = 22; ALT, Altamaha River, n = 2; PAM, Pamlico Sound and its tributaries, n = 66; CHE, Chesapeake Bay, n = 23. For locations of co-occurrence, inset maps b–d show sites where each species was caught, with overall sample size for each location indicated. R. brasiliensis, triangles; R. bonasus, circles; SSB, St. Sebastian River

Consensus sequences were generated by aligning the forward and reverse sequences for each individual in CodonCode v 9.0.1 (CodonCode Corporation, Dedham, USA). Resultant haplotypes were compared to mtDNA ND2 haplotypes generated, using the methods described here, for 10 specimens each of R. brasiliensis and R. bonasus collected from Mississippi and Alabama waters (United States), which were verified using a suite of 21 external morphological measurements and genetic analysis of the cytochrome c oxidase subunit I (CO1) gene (see Jones et al., 2017). The Jukes–Cantor substitution model (Jukes & Cantor, 1969) was determined to be the nucleotide substitution model of best fit for the data in jModelTest 2 v 2.1.10 using Bayesian Information Criterion (BIC) values (Darriba et al., 2012). Phylogenetic relationships between these haplotypes were inferred using a maximum likelihood approach with 10,000 bootstrap replicates and the Jukes–Cantor substitution model in MEGA v X (Kumar et al., 2018), with the bat ray Myliobatis californica Gill 1865 (GenBank accession no. KM364985) serving as an outgroup. Relationships among haplotypes were estimated using the maximum parsimony method of Polzin and Daneshmand (2003) and visualized as a median-joining haplotype network using Network v 10.1.10 (Bandelt et al., 1999).

The 282 tissue samples revealed 14 haplotypes that formed two distinct genetic clades. Three haplotypes representing 24 individuals clustered together with the single haplotype (RBRA1) sequenced for the 10 verified R. brasiliensis (GenBank accession nos. MT410205–MT410207) and 11 haplotypes representing 258 individuals clustered with the single haplotype sequenced for the 10 verified R. bonasus (RBON1) (GenBank accession nos. MT410194–MT410204) (Supporting Information Figure S1). The sampled R. brasiliensis and R. bonasus haplotypes were differentiated by 5.43–7.24% sequence divergence, based on the number of base pair differences (Supporting Information Figure S2). In contrast, the maximum sequence differentiation between sampled haplotypes within each species was much less: 0.68% and 1.13% for R. brasiliensis and R. bonasus, respectively (Supporting Information Figure S2). The minimum sequence differentiations between M. californica and R. brasiliensis and R. bonasus were 13.80% and 12.90%, respectively.

The 24 genetically identified R. brasiliensis were collected from eight locations in the western North Atlantic; eight were from two estuaries in the Atlantic and 16 were from six estuaries in the eGOM (Figure 1). In the Atlantic, five R. brasiliensis were caught at the mouth of the St. Sebastian River in the Indian River Lagoon, Florida (hereafter IRL) and three were caught in Lake Worth Lagoon, Florida. In the eGOM, six were caught in Apalachicola Bay, five in Santa Rosa Sound, two in Charlotte Harbor and one in each of St. Andrew Bay, St. Joseph Bay and Cedar Key. Individuals included males and females, juveniles and adults, and were caught across all seasons in multiple years between 2003 and 2018 (Table 1). Evidence of reproductive habitat use in the Atlantic was identified by the presence of two particular R. brasiliensis near the mouth of the St. Sebastian River, a gravid female caught in summer (August) and a juvenile male caught in winter (January). None were identified from estuaries in the Atlantic north of the St. Sebastian River in the IRL, including the Altamaha River, Georgia (n = 2), Pamlico Sound and its tributaries, North Carolina (n = 66), and Chesapeake Bay, Virginia (n = 23) (Figure 1).

TABLE 1.

Biological details for the Brazilian cownose ray Rhinoptera brasiliensis Müller 1836 genetically identified via a 442-base pair fragment of the mitochondrial NADH dehydrogenase subunit 2 gene from sites off Florida in the eastern Gulf of Mexico (eGOM) and along the east coast of the USA (Atlantic)

Region	Location	Date collected	Seasona	Sex	Life stageb	Disc width (cm)	Haplotype
eGOM	Santa Rosa Sound	2013–2017	NA	NA	NA	NA	RBRA1
		2013–2017	NA	NA	NA	NA	RBRA1
		2013–2017	NA	NA	NA	NA	RBRA1
		2013–2017	NA	NA	NA	NA	RBRA1
		2013–2017	NA	NA	NA	NA	RBRA1
	St. Andrew Bay	25 October 2013	Fall	F	Juvenile	55.0	RBRA1
	St. Joseph Bay	14 August 2012	Summer	M	Adult	90.0	RBRA1
	Apalachicola Bay	20 June 2013	Summer	F	Adult	88.5	RBRA1
		20 June 2013	Summer	F	Adult	71.5	RBRA1
		20 June 2013	Summer	F	Juvenile	49.0	RBRA1
		14 June 2012	Summer	F	Juvenile	48.0	RBRA1
		14 June 2012	Summer	M	Juvenile	40.0	RBRA1
		24 October 2013	Fall	M	Juvenile	49.0	RBRA1
	Cedar Key	6 December 2017	Winter	F	Adult	77.5	RBRA3
	Charlotte Harbor	17 October 2003	Fall	M	Juvenile	62.5	RBRA1
		15 December 2008	Winter	M	Juvenile	66.7	RBRA1
Atlantic	Lake Worth Lagoon	14 February 2018	Winter	M	Adult	74.0	RBRA1
		14 February 2018	Winter	F	Adult	78.4	RBRA1
		14 February 2018	Winter	F	Adult	80.2	RBRA1
	Indian River Lagoon	12 January 2018	Winter	M	Juvenile	61.3	RBRA1
		12 April 2017	Spring	M	Adult	87.0	RBRA5
		12 April 2017	Spring	M	Adult	87.5	RBRA1
		12 April 2017	Spring	M	Adult	85.1	RBRA1
		7 August 2017	Summer	Fc	Adult	93.2	RBRA1

Note:Specific dates and biological details were not available (NA)

^aSpring, March–May; summer, June–August; fall, September–November; winter, December–February.

^bLife stage of R. brasiliensis was inferred from disc width at 100% maturity for both sexes (>71.2 cm disc width) of R. bonasus (Mitchill 1815) from Charlotte Harbor, Florida by Poulakis (2013).

^cGravid.

The discovery of 24 R. brasiliensis based on mtDNA ND2 sequence data across sites in Florida waters further confirms its occurrence in the eGOM (see Carney et al., 2017; Jones et al., 2017; McDowell & Fisher, unpubl. data) and documents that the range extends further in the western North Atlantic than previously described, at least as far north as the St. Sebastian River in the IRL. Since mtDNA is maternally inherited, this finding assumes that these individuals are not hybrid crosses between female R. brasiliensis and male R. bonasus, which would require analysis of additional nuclear data. Hybridization has not yet been documented in cownose rays, but has been shown in freshwater rays (Potamotrygonidae, Crus et al., 2015) and fiddler rays (Rhinobatidae, Donnellan et al., 2015). While it is possible to detect patterns of introgression from mtDNA markers alone, robust evidence necessitates powerful datasets that typically use much larger DNA fragments (e.g., Rosenzweig et al., 2016).

These new records are part of a growing body of literature documenting the presence of R. brasiliensis in the western North Atlantic. Prior species accounts that used genetic identifications of R. brasiliensis in the Atlantic include four records with no life history data from Georgia in fall (October) 2007 (Quattro, unpubl. data), one of unknown size from North Carolina with an unknown capture date in 2008 (Naylor et al., 2012), and one from a deceased specimen of unknown size found on a beach in New Jersey in summer (August) 2017 (Stoeckle et al., 2020). There is one additional historic record from North Carolina, but since this identification was based on tooth series counts alone (Bigelow & Schroeder, 1953), it is not considered reliable. Despite these records, the Atlantic was not considered part of the accepted range of R. brasiliensis in subsequent accounts (e.g., Last et al., 2016), possibly because they were discounted as ‘vagrants’. The term vagrant was recently defined for elasmobranchs by Grant et al. (2019) as ‘individuals found outside of the species’ distribution, in a habitat not biologically utilized by the species’. Our work has shown the presence of both male and female R. brasiliensis of all life stages, documented across all seasons over multiple years as far north as the St. Sebastian River in the IRL in the Atlantic. This indicates that R. brasiliensis is unlikely to be a vagrant, at least in the Atlantic off Florida, and that this region is part of its range, although the northernmost extent of its range remains uncertain. Whether R. brasiliensis have long occurred in these waters but were obscured by the presence of morphologically similar R. bonasus, or this reflects a more recent range expansion, remains unknown.

Juvenile R. brasiliensis occurred near the mouth of the St. Sebastian River during winter, in St. Andrew and Apalachicola bays during summer and fall, and in Charlotte Harbor during fall and winter. The presence of juveniles in these estuaries along both coasts of Florida, as well as the presence of a gravid R. brasiliensis in the IRL during the time associated with their parturition (Rangel et al., 2017), suggests these bays may serve as nurseries. Juvenile R. bonasus are abundant in estuaries in the northern Gulf of Mexico and have been shown to tolerate low temperatures (e.g., <12°C, reported down to 8.3–9.1°C), potentially residing in estuaries over winter, and possibly until they reach maturity (Ajemian, 2011; Ajemian & Powers, 2016; Bigelow & Schroeder, 1953). Additional research is needed to better understand how and to what extent R. brasiliensis uses and relies on these estuarine habitats in the Atlantic and eGOM.

The opportunistic nature of the sampling regime in this study likely limited our understanding of the occurrence and frequency of R. brasiliensis throughout the study area. For example, for some locations four samples were collected at a single site in a single day and represented the same life stage, while elsewhere 52 samples were selected from archived samples to include all life stages, seasons and sexes over 9 years. Despite these differences, the presence of R. brasiliensis at locations outside of its accepted range, detected over multiple years, implies this species has a wider range than previously thought.

As documented here, the range of R. brasiliensis extends at least to the St. Sebastian River in the IRL in the Atlantic. More comprehensive, year-round surveys paired with genetic species identifications are still needed to clarify the northern extent of the range of R. brasiliensis in the Atlantic and seasonal patterns of occurrence in these waters. This range extension further increases range overlap between R. brasiliensis and R. bonasus, emphasizing the need to verify species identities using genetic methods prior to undertaking biological or ecological studies of either species. Furthermore, existing biological and relative abundance datasets require careful consideration as they may reflect a Rhinoptera species complex rather than single species. Future studies should use a suite of approaches, including tagging and molecular methods, to better understand and disentangle the biology and ecology of these morphologically similar, sympatric species.