Speciation in marine environments is usually a complex process involving geographic isolation and ecological adaptation (Bowen et al., 2013) mediated by an organism’s life history characteristics and biology (Craig et al., 2006). Some para- or sympatric lineages with incipient genetic differentiation might be considered as different species, not only populations (Avise, 2000; Potkamp, Fransen, 2019). This scenario has already been documented in some sharks (Corrigan, Beheregaray, 2009) and rays (Kashiwagi et al., 2012) that move extensively but are genetically restrained by environmental forces (Bowen et al., 2013). During the process of divergence, lineages incrementally acquire genotypic and phenotypic characteristics that make them distinct from each other, creating a grey zone where the definition of a species is ambiguous (De Queiroz, 2007). In such cases, a limited number of genetic loci are often insufficient to resolve taxonomic issues and semi-isolated lineages prevail until genomic studies are accomplished, making the delineation of species a challenging task. So, conservation should be a priority and not be constrained by a lack of clarity in species boundaries (Roux et al., 2016).

Dasyatid stingrays are globally distributed batoids that vary in size (from 23 cm to 2.2 m disc width), weigh up to 600 kg, and vary in disc shape from circular to rhombic. They primarily occur in coastal marine environments (down to 400 m), but can also be found in freshwater (Last et al., 2016b). Until recently, the genus Dasyatis Rafinesque, 1810 was indicated as a paraphyletic group of stingrays based on morphological data (Rosenberger, 2001), and subsequent studies based on the mitochondrial gene NADH dehydrogenase 2 (mt-nd2) corroborated this hypothesis (Naylor et al., 2012). Last et al. (2016a) and Lim et al. (2015) revised Dasyatidae based on morphological and molecular data and divided it into four subfamilies (Dasyatinae, Hypolophinae, Neotrygoninae, and Urogymninae). Moreover, what was previously known as Dasyatis was separated into eight genera (Dasyatis, Pteroplatytrygon Fowler, 1910, Taeniurops Garman, 1913, Bathytoshia Whitley, 1933, Hemitrygon Müller & Henle, 1838, Hypanus Rafinesque, 1818, Telatrygon Last, Naylor & Manjaji-Matsumoto, 2016, and Megatrygon Last, Naylor & Manjaji-Matsumoto, 2016) in the subfamily Dasyatinae, grouped by morphological similarities and molecular clusters.

Despite the resurrection of a monophyletic Hypanus, the most species-rich Dasyatinae genus around the American continent, relationships among its species and theirphylogenetic position within Dasyatidae were based on a single mitochondrial marker (mt-nd2), with few representatives per independent evolutionary lineage, and some missing ones due to lack of sampling (Last et al., 2016a).

Currently, Hypanus encompasses nine recognized species: H. americanus (Hildebrand & Schroeder, 1928), H. berthalutzae Petean, Naylor & Lima, 2020, H. dipterurus (Jordan & Gilbert, 1880), H. guttatus Bloch & Schneider (1801), H. longus (Garman, 1880), H. marianae (Gomes, Rosa & Gadig, 2000), H. rudis (Günther, 1870), H. sabinus (Lesueur, 1824), and H. say (Lesueur, 1817). Except for H. rudis from Guinea Gulf, on the western coast of the African continent, and H. dipterurus and H. longus from the Pacific Ocean, all other six species occur on the Atlantic coast of America. Even though six of these species were sampled and included in the analysis by Last et al. (2016a), the placement of H. marianae was not tested, and it was considered a Hypanus species based on morphological data, as well as H. rudis, which was recently corroborated as a Hypanus species by Petean et al. (2020), who also described a new one (H. berthalutzae).

Precise delimitation and identification of these stingrays are crucial for their conservation since they are frequently the targets of fisheries where they are harvested for food and clothing (Costa et al., 2015; Last et al., 2016b; Ceretta et al., 2020; Oliveira et al., 2021). More than half of Hypanus species are evaluated as threatened in the Red List of Threatened Species by IUCN: three are Vulnerable (H. berthalutzae, H. dipterurus, and H. longus (Charvet et al., 2020; Pollom et al., 2020a,c)), one is Endangered (H. marianae (Pollom et al., 2020b)), and one is Critically Endangered (H. rudis (Jabado et al., 2021c)); three are classified as Near Threatened (H. americanus, H. guttatus, and H. say (Carlson et al., 2020a,b,c)); and only one is clearly under no risk of extinction: H. sabinus (Least Concern, Carlson et al., 2020d)).

Another dasyatid genus, in the subfamily Urogymninae, with a similar pattern of species diversification in the Atlantic Ocean and facing risks of extinction is Fontitrygon Last, Naylor & Manjaji-Matsumoto, 2016, which currently contains six species (Last et al., 2016a). Four occur in western Africa: Fontitrygon margarita (Günther, 1870) (Vulnerable, Jabado et al., 2021a), F. margaritella (Compagno & Roberts, 1984) (Near Threatened, Jabado et al., 2021b), F. ukpam (Smith, 1863) (Critically Endangered, Jabado et al., 2021d), and F. garouaensis (Stauch & Blanc, 1963) (Critically Endangered, Jabado et al., 2021e) while two occur along the Northern coast of South America: F. colarensis (Santos, Gomes & Charvet-Almeida, 2004) (Santos et al., 2004) (Critically Endangered, Pollom et al., 2020d) and F. geijskesi (Boeseman, 1948) (Critically Endangered, Pollom et al., 2020e). Nevertheless, only three of these species were included in Last et al. (2016a) Dasyatidae revision due to a lack of samples, and both American species were provisionally positioned in Fontitrygon. Despite the incomplete taxon sampling represented, the phylogenetic relationships provided by those authors indicated that some members of Fontitrygon might be misclassified, leaving it as a possible paraphyletic genus.

A useful genetic marker for investigating phylogenetic relationships and species identities is the mitochondrial DNA (mtDNA) due to its high evolutionary rate, maternal inheritance, intraspecific polymorphisms, and genes arrangement (Avise et al., 1987; Harrison, 1989; Boore, Brown, 1998; Satoh et al., 2016). Even though a phylogeny based on mtDNA is a story of modifications on a small portion of DNA of maternal transmission, it has not been puzzled by recombination (Avise et al., 1987). Species in which females are more stationary than males, mtDNA can provide distinct information than nuclear markers due to biased dispersal by sexes (Moritz et al., 1987). However, studies on H. americanus from Central America have shown little to no philopatric behavior, with both males and females contributing to gene flow (Corcoran et al., 2013; Flowers et al., 2016; Schwanck et al., 2020). So, evolutionary studies on Hypanus stingrays based on mtDNA might tell a similar story to nuclear markers, to be further assessed. Recently, mitogenomes have been widely used for phylogenetic inferences in distinct metazoan clades: Diptera (da Silva et al., 2020), Rodentia (Abramson et al., 2021), Coleoptera (Nie et al., 2021), and Elasmobranchii (Palacios-Barreto et al., 2023).

Since the massive use of molecular methods, such as DNA barcoding (Hebert, Gregory, 2005), to identify and classify species, organisms with comparable phenotypes that could have been considered as unique species are recognized as genetically diverse, a concept known as “cryptic species”. Sáez, Lozano (2005) described these as “groups of organisms that are morphologically indistinguishable from each other, yet found to belong to different evolutionary lineages”. Sphyrna gilberti Quattro, Driggers, Grady, Ulrich & Roberts, 2013 and Squalus suckleyi (Girard, 1855) are examples of shark species with circumtropical distributions in which morphological analyses subsequently to identifications of genetic lineages corroborated the existence of more than one entity (Ebert et al., 2010; Quattro et al., 2013; Gaither et al., 2016).

A concept to be explored is that of taxonomic gap, in which there is a space between the extant biodiversity and what is actually known about it (Dubois, 2010; Raposo et al., 2020). This gap regards both the universe of unknown species and those susceptible to changes due to more studies. As many authors have said, “taxonomic stability is ignorance” (Dominguez, Wheeler, 1997; Benton, 2000; Dubois, 2010;) since with more data and analyses the gap might increase or decrease in a continuous progress of Science. It is indisputable that the lack of specimens that could serve as vouchers for each molecular sample could have consequences for taxonomy (Amorim et al., 2016) due to the impossibility of checking the morphology of all individuals. However, given the urgency in closing this taxonomic gap to recognize the world’s biodiversity before more extinctions take place, even tissue samples could corroborate the once-existing variety of species (Engel et al., 2021).

Due to the taxonomic uncertainties in Dasyatinae (Lim et al., 2015; Last et al., 2016a; Pavan-Kumar et al., 2022), the absence of a complete sampling of all Hypanus species in other published works, and the risks of extinction these stingrays are facing, our goal is to use mitogenomes to define the relationships among Hypanus species, identifying possible cryptic ones, and their relationships to other Dasyatinae genera. Afterward, species can be properly identified and (re)evaluated for adequate conservation measures.

Sampling, DNA isolation, and sequencing. To test the monophyly of the genus Hypanus and the subfamily Dasyatinae, we sampled 124 specimens from all nine valid species belonging to Hypanus, six representatives of almost all Dasyatinae genera (Hemitrygon akajei (Bürger, 1841), Telatrygon acutirostra (Nishida & Nakaya, 1988), Pteroplatytrygon violacea (Bonaparte, 1832), Batytoshia lata (Garman, 1880), Taeniurops grabatus (Geoffroy St. Hilaire, 1817), Dasyatis hypostigma Santos & Carvalho, 2004; except Megatrygon), and both Neotrygoninae genera (Taeniura lymma (Fabricius, 1775) and Neotrygon kuhlii (Müller & Henle, 1841)). As outgroup, we included representatives of each Fontitrygon species, subfamily Urogymninae (one sample of F. margarita, F. margaritella, F. garouaensis,and six of F. geijskesi; except F. colarensis and F. ukpam). Species distributions and sampling localities are provided in Tab. 1 (details in Tab. S1) and sample locations of Hypanus and Fontitrygon in Fig. 1. Valid names and distributions were obtained from Last et al. (2016a) and Eschmeyer’s Catalog of Fishes (Fricke et al., 2023). Nearly all tissues were collected in fish markets, making it unfeasible to preserve most of the specimens; however, we performed barcode analyses (described below) to compare clades to examined specimens deposited in collections. Lineages for which we could provide vouchers are H. americanus, H. berthalutzae, H. geijskesi, H. guttatus, H. sabinus,and H. say;even though there is no voucher for H. marianae, tissues came from the specimens identified and collected by Costa et al. (2022). Before mitochondrial gene capture, samples were genetically identified based on Sanger sequencing of the mitochondrial marker mt-nd2, as described by Petean et al. (2020), and compared to the database from Naylor et al. (2012). After capture, we performed barcode analyses comparing to data from GenBank (detailed as it follows) as another approach to verifying species’ identities. When we directly removed tissues from specimens through diving or trawling, we morphologically identified them. Data collection was under SISBIO permit 54254-3 and supported by Atlantis Divers in Fernando de Noronha, Brazil.

From genomic DNA extraction to the alignment of protein-coding gene sequences of mitochondrial genomes, all protocols and procedures followed Li et al. (2013, 2015) and Petean et al. (2020). Extracted DNA was sheared to 500 bp in an M220 Focused-ultrsonicator (Covaris, Inc., Wobuern, Massachusetts, USA) as the first step for library preparation, followed by the selection of > 200 bp fragments with solid-phase reversible immobilization beads (Li et al., 2015). We performed a series of reactions in each sample for mitochondrial gene capture using biotinylated RNA baits (Mycroarray, Ann Arbor, Michigan) (Li et al., 2013). For sequencing, we deployed an Illumina MiSeq Next Generation Sequencer and, from each read, removed low-quality reads (with Phred quality scores lower than 30; Illumina 2011, https://www.illumina.com/documents/products/technotes/technote_Q-Scores.pdf) and adaptors using Trim Galore 0.6.4 (Krueger, 2020) then mapped to the mitochondrial genome of a closely-related species, Hemitrygon akajei (NC_021132), from the GenBank using Geneious 7.9.1 (http://www.geneious.com). Finally, we used a pipeline (MitoAnnotator, (Iwasaki et al., 2013)) to annotate sequences, which are available on GenBank under accession numbers provided in Tab. S1.

Phylogenetic reconstructions. To study the relationships within the genus Hypanus, its relationships within the subfamily Dasyatinae, and to Neotrygoninae, the whole mitogenome sequences of all 135 specimens were aligned in GENEIOUS 7.9.1 using the MUSCLE algorithm (Edgar, 2004). After annotation we identified and excluded from all sequences the mitochondrial control region (CR), tRNA, and rRNA. Control region was deleted because it is highly variable among individuals and its coverage after mitochondrial capture was too low; RNAs regions were eliminated because the indels present in these regions make alignment difficult. The final alignment had 11,471 base pairs in 13 protein-coding genes.

To select the best-fitting model of molecular evolution we used PartitionFinder2 (Lanfear et al., 2017) and selected the best scheme for each protein-coding gene under Bayesian Inference Criteria: GTR+gamma+invariant sites for mt-nd1 and mt-nd5, HKY+gamma for mt-nd2, mt-atp8, mt-atp6, mt-coiii, mt-nd6, and mt-cytb, HKY+gamma+invariant sites for mt-nd3, mt-nd4l, and mt-nd4, and TN93+gamma for mt-coi and mt-coii. Maximum Likelihood analyses were conducted using RAxML version 8. (Stamatakis, 2014) in CIPRES Science Gateway (Miller et al., 2010), with bootstrap and consensus calculations based on a 1000-generation search of tree space.

Bayesian Inferences were carried out in BEAST 2.5 (Bouckaert et al., 2019) using Yule model as the prior tree as we are not considering known extinctions (μ = 0) and there is a reasonable sampling for analyses (ρ = 1) (Drummond, Bouckaert, 2015) in 1,000,000,000 generations with 5 chains resampled every 10,000. The software MEGA X (Kumar et al., 2018) was used to calculate uncorrected genetic p-distances and analyze intra- and interspecific genetic differences between Hypanus lineages.

Lineage delimitation methods. By analyzing the relationships among species, we noticed some valid species could be either paraphyletic or have long branches within them, suggesting possible distinct lineages. Therefore, we decided to do five species delimitation analyses within the clades of H. guttatus and H. say independently: multiple- and single-threshold Generalized Mixed Yule Coalescent (m-GMYC and s-GMYC, Fujisawa, Barraclough, 2013), multi-rate Poisson Tree Process (mPTP, Kapli et al., 2017), Bayesian Poisson Tree Process (bPTP, Zhang et al., 2013), and Assemble Species by Automatic Partitioning (ASAP,(Puillandre et al., 2021). For each clade’s data (H. guttatus and H. say) we selected an outgroup based on the results of our phylogenetic analysis and Last et al. (2016a): Hypanus marianae for H. guttatus and H. sabinus for H. say analyses. Most of these analyses (except ASAP) are performed on a tree topology: both GMYC methods rely on an ultrametric tree, which was built under a Bayesian Inference analysis in BEAST 2.5, and both PTP on a tree with nucleotides’ substitutions, built with a Maximum Likelihood analysis in RAxML version 8. For such phylogenetic analyses before delimitation ones, we used jModelTest2 (Darriba et al., 2012) to select the best molecular evolution model for each clade. The ASAP method depends on an alignment matrix instead of a tree. For more details on delimitation methods, refer to Petean et al. (2020).

DNA Barcode analyses. The mitochondrial protein-coding gene region cytochrome c oxidase subunit I (mt-co1) was identified and extracted from some sequences for comparisons to those available at GenBank (Tab. S2). The goal was to use the molecular clusters as support for the verification of taxonomic status when vouchers were available for at least one sequence in a clade. Sequences of Fontitrygon geijskesi sampled by this study were aligned with a sample from Guyana (GN17902) and four samples from Rodrigues-Filho et al. (2020) (GenBank numbers MN105749, MN105812, MN105813, MN105819), who provided a voucher for the species. The alignment was performed in MEGA X using the MUSCLE algorithm, which had 587 base pairs, 12 F. geijskesi samples, and one outgroup. The genetic p-distance was also performed in MEGA X and, to estimate species identities based on sequences’ similarities, a Neighbor-Joining (Saitou, Nei, 1987) tree was built in GENEIOUS 7.9.1 with 1,000 bootstrap replicas and edited in FigTree v. 1.4.4 (http://tree.bio.ed.ac.uk/software/figtree/).

The same barcode analyses were performed for the three clades containing species-complexes (H. guttatus, H. say, and H. americanus) identified by abovementioned analyses and Petean et al. (2020). Hypanus berthalutzae (GN18496) was used as an outgroup for all independent analyses, except for H. americanus, for which it was also the ingroup and H. guttatus (GN18434) was then used as an outgroup.

For H. guttatus, besides the 33 samples from this study, we included 52 mt-co1 sequences from GenBank, totaling 85 samples (and one outgroup) in 524 base pairs. To analyze the clade containing H. say, we extracted the mt-co1 region from mitogenomes of this species, H. dipterurus, and H. sabinus and added 13 mt-co1 sequences from H. say, two H. dipterurus, and eight H. sabinus from GenBank. The alignment had 40 samples (and one outgroup) in 547 base pairs. Finally, to investigate H. americanus, we not only added 38 mt-co1 sequences from GenBank, but we also included 23 H. berthalutzae, four H. longus, and four H. rudis, in a total of 77 samples as the ingroup in 599 base pairs.

Phylogenetic inferences. The genus Hypanussensu Last et al. (2016a) was recovered as monophyletic and sister to all other genera (except Megatryon, not sampled for this study) within the subfamily Dasyatinae (Fig. 2), which is a sister-group to Neotrygoninae. These results were already suggested by Last et al. (2016a) and are now corroborated by mitogenomes through the same resulting topologies by Maximum Likelihood and Bayesian Inference phylogenetic analyses with all nodes’ values higher than 88% bootstrap and 0.995 of posterior probability. Maximum Likelihood and Bayesian Inference trees topologies with all taxa and nodes values are available in Figs. S3 and S4, respectively.

There are clear unique lineages that correspond to valid species names: Hypanus berthalutzae, H. dipterurus, H. longus, H. marianae, H. rudis, and H. sabinus.Hypanus americanus, the Southern stingray, nonetheless, is not a single evolutionary lineage (Clade A in Fig. 2), as suggested by previous mitogenomes delimitation analyses and haplotype network based on mt-nd2 (Petean et al., 2020; Figs, 2–3, respectively), which showed 8 unsampled haplotypes and mutational steps between both lineages, while there are four between H. berthalutzae and H. rudis; and a phylogeographic study based on the mitochondrial control region by Richards et al. (2019) that found three populations of this species in the USA’s coast and Caribbean. The species H. marianae, which was not included in the previous molecular study (Last et al., 2016a), is a monophyletic lineage and sister to the clade containing H. americanussensu lato, H. longus, H. berthalutzae, and H. rudis. This clade is, then, closely-related to a group containing H. guttatus, which now is suggested to harbor two lineages: one distributed from Central America to the south of Brazil and another of specimens from Belize (Clade B in Fig. 2).

Two species of Hypanus occur along the Pacific coast, H. dipterurus and H. longus, both with similar evolutionary histories since they are independent sister-groups to Atlantic clades: H. say and H. berthalutzae + H. rudis, respectively. Moreover, within the clade H. say, there is a clear divergence of two lineages separated by the Peninsula of Florida (Clade C in Fig. 2).

Six representatives of Fontitrygon geijskesi, subfamily Urogymninae, which was not included in the Dasyatidae revision by Last et al. (2016a), formed a sister-clade to H. guttatus, within the genus Hypanus, but not Fontitrygon. So, for Hypanus to be monophyletic, this species should be reclassified as Hypanus geijskesi. The subfamily Urogymninae is then represented by three Fontitrygon species occurring in Africa, which formed a cluster: F. margarita, F. margaritella,and F. garouaensis.

Delimitation of lineages. Candidate species of both species complexes, H. guttatus and H. say, were analyzed by combining all five delimitation methods (Figs. 3–4); analyses of H. americanus complex were performed by Petean et al. (2020). The observed new lineages, sister to known species, are named as affinis to those they are closely related to. We kept the valid name according to the type-locality of each species: type of H. guttatus from Brazil, so H. aff. guttatus from Central America; and type of H. say from Egg Harbor (USA), H. aff. say from the Gulf of Mexico.

We selected those results which were more consistent among methods, with similar branches’ division, and those that provided the least number of lineages within a species complex to avoid over-splitting taxa due to mere genetic structure. mPTP and ASAP were the most conservative analyses suggesting only two lineages within each species complex; however, while bPTP agreed with mPTP in delimiting H. say two lineages, it was a less stringent method when analyzing H. guttatus data, pointing to 25 entities (almost one per individual). Both GMYC methods, single and multiple thresholds, resulted in similar groupings within each complex and proposed only one or two lineages more than we accepted.

Intraspecific average pairwise distances vary from 0.036% in H. longus to 0.26% in H. aff. guttatus (Tab. 2), while in interspecific average pairwise, the smallest distances are 0.82% between H. rudis and H. berthalutzae, 0.83% between H. americanus and H. aff. americanus, and 0.95% between H. say and H. aff. say (Tab. 3). Interestingly, the distance between two geographically distant species as H. longus, from the Pacific, and H. rudis, from Africa is only 2.4% (Petean et al., 2020), while H. sabinus has the highest distances to all other Hypanus lineages (11.74% from H. dipterurus is the smallest).

DNA Barcoding. By analyzing the protein-coding region mt-co1 of 11 samples of “Fontitrygon geijskesi”, we obtained a monophyletic group by a Neighbor-Joining analysis with 100% of bootstrap value (Fig. 5), a result similar to that using mitogenomes. Besides, the genetic distances among all sequences varied from 0 to 0.17%, with an average of 0.03% (Tabs. 4, S5). Given the genetic similarity of these sequences and since Rodrigues Filho et al. (2020), from which came four of those samples, could provide a voucher specimen for one of them, we have enough support to suggest that it is indeed a valid species. However, it should be considered as Hypanus geijskesi due to its close relationship to H. guttatus, as suggested by the abovementioned phylogenetic analyses.

Regarding the species H. guttatus, the scenario is convoluted: the lineage H. aff. guttatus identified by mitogenomic delimitation analyses was supported by the inclusion of more samples, as shown by the Neighbor-Joining tree with 89.2% of bootstrap value (Fig. 6). The genetic distance between these two samples (GN13939, GN13946) was 0.38%, which was the same distance between GN13946 and ten other samples (GN19470, MN105788, MN105792, MN105794, MN105808, MN105817, MN105869–71, MN105875), while the distance between GN13939 and the same ten samples was 0.76%. However, the distances between these two (GN13939, GN13946) and the other 73 were higher than 1.15%, distances between nine of those abovementioned (except GN19470) and the others varied from 0.76% to 0.95%, and distances between those 73 samples varied from 0 to 0.19% (Tabs. 5, S6).

Therefore, based on these results, we suggest that, besides the lineage H. aff. guttatus, there could also be some hybridization or incomplete lineage sorting driving the evolution of H. guttatus, which could be undergoing diversifications into distinct ecological niches (Nosil, Harmon, 2009); such processes could only be understood through more sampling and markers. Given the data we have, we can corroborate the lineage H. guttatus by examination of three vouchers by one of us (FFP) (GN18451, GN18458–9; deposited at the fish collection at Universidade Federal do Rio Grande do Norte under respective codes CIUFRN 4442, 4449–50).

Through the extraction of the mt-co1 region from samples of the species-complex H. say, the result agrees with our previous outcome: groups separated by the Peninsula of Florida (Fig. 7), with high bootstrap values for each clade, H. say and H. aff. say, of 97% and 97.5%, respectively. The genetic p-distances between both lineages varied from 0.93% to 1.09%, while within lineage values ranged from 0 to 0.31% (Tabs. 6, S7). For the lineage occurring on USA’s East coast, which should bear the name H. say, one of the samples (MH378605) came from a specimen deposited at the Smithsonian Fish Collection (USNM433289), from a place close to type’s location. This specimen was examined by FFP, who identified it as H. say, thus serving as a voucher for the lineage.

Sequences of the species H. dipterurus and H. sabinus were analyzed together with H. say complex, and analyses suggested the southernmost sample of H. dipterurus(MH 194454, from the Peruvian coast, Pacific Ocean) could be another lineage since it has an average of 3.84% of genetic distance to the other five samples of H. dipterurus from Baja California and California coast. Moreover, the inclusion of eight sequences of H. sabinus still leaves it monophyletic, with samples from both the East coast of the USA and the Gulf of Mexico. One of these samples (MT 455431) was extracted from a specimen deposited at the Smithsonian Fish Collection (USNM 426256), which was examined by FFP, hence could serve as a voucher for the clade.

Within H. americanus species-complex there seem to be two sympatric clades: one that should bear the species name, H. americanus (82.9% of bootstrap value) (Fig. 8), since mt-co1 sequences of some analyzed specimens by FFP at the Smithsonian Fish Collection (USNM 433102, USNM 433338–9) fall within it (KT 075327, MH 378683–4) and they were collected close to the species type-locality. The other clade is composed of three samples (two previously identified as H. aff. americanus by Petean et al. (2020), and one sample deposited at GenBank, MG837920). Genetic distance among these three H. aff. americanus samples is 0, and their distance to other H. americanus vary from 0.33% to 0.67%; while distances within H. americanus vary from 0 to 0.17%. Regardless of which H. americanus clade, sequences belonging to this species-complex have more than 1.5% distance to any other sequence belonging to H. berthalutzae, H. longus, and H. rudis (Tabs. 7, S8). Besides, some sequences previously identified and submitted to GenBank as H. americanus should be reallocated to H. berthalutzae (MK085594, MK085604, MK085629, MK085636, MK085638, MK085641, MK085657, MK085659, MK085662, MK085669, MK085672, MK085684, MK085742, MN105805, MN105821, MN105822, MN105823, MN105824, MN105839, MN105842, MN105845, MN105846, MN105847). Some of these sequences were used by Rodrigues Filho et al. (2020) to suggest the existence of two lineages of H. americanus in Northern Brazil; and one of them was described as H. berthalutzae (H. americanus 1). The clade they called H. americanus 2 is what we identified as H. americanus. The lineage identified by Petean et al. (2020) and hereby sustained as H. aff. americanus was not sampled by those authors.

Phylogenetic considerations. Hypanus was recovered as monophyletic by using mitochondrial genome sequences of all species previously attributed to it in addition to another species that was provisionally allocated in Fontitrygon by Last et al. (2016a) due to a lack of sampling and morphological similarities. These authors revised the family Dasyatidae and used the mt-nd2 gene to identify chondrichthyans’ lineages, as suggested by Naylor et al. (2012). Our results indicate the use of this marker is reliable for identifying Chondrichthyes species, especially when resources are unavailable for genome sequencing. Some lineages suggested to belong to Hypanus, but unsampled by Last et al. (2016a), were hereby included. Hypanus marianae was identified as a Hypanus species, as implied by morphological similarities by Last et al. (2016a), andit is a sister-species to the clade containing H. americanus species complex, H. longus, H. rudis,and H. berthalutzae, a recently described species (Petean et al., 2020) whose phylogenetic position was also corroborated by the inclusion of representatives of the whole genus.

For the monophyly of the genus Hypanus, the species “Fontitrygon geijskesi” should be considered a member of Hypanus, as it is found to be sister to H. guttatus. Due to morphological similarities, this species was expected to be related to its African congeners F. margarita, F. margaritella, and F. garouaensis (Last et al., 2016a). However, we suggest the reallocation of F. geijskesi to the genus Hypanus with a new name combination as Hypanus geijskesi (Boeseman, 1948). This result had already been observed by Rodrigues Filho et al. (2020) on a Neighbor-Joining analysis using the mitochondrial marker COI (mt-co1), in which “F. geijskesi” specimens resulted as a sister group to H. guttatus within the genus Hypanus; however, their analysis lacked a phylogenetic inference of its relationships to other Hypanus lineages. Through a combination of mt-co1 sequences by Rodrigues Filho et al. (2020) to those hereby provided, we noticed H. geijskesi intraspecific distances ranging from 0 to 0.17% and, as they have provided a voucher for one of their samples, we have support for the species reallocation. This change leaves the genus Fontitrygon with five species, of which four occur in the African continent and one in South America (F. colarensis, not sampled by this study).

As already suggested by Lim et al. (2015), Last et al. (2016a), and Pavan-Kumar et al. (2022), but including more representatives of the subfamily Dasyatinae, we also recognized its monophyly and close relationship to Neotrygoninae, with Hypanus as the sister-group to all other genera within the first subfamily; and the subfamily Urogymninae, represented by the genus Fontitrygon, supporting the subfamilies’ rooting.

Two Hypanus species that used to have the largest geographic distributions, H. americanussensu lato (Petean et al., 2020) from Massachusetts (USA) to São Paulo (Brazil) and H. guttatus from Mexico to Southeastern Brazil, are now recognized to encompass more than one lineage each. The lineage of H. americanussensu (Petean et al., 2020) occurring at the Brazilian coast was recently described as H. berthalutzae, a sister-species to H. rudis in Eastern Atlantic. What was left of H. americanus in Central and North America also represents more lineages, as suggested by our phylogenetic analyses (Clade A, Fig. 2), delimitations done by Petean et al. (2020) (Figs. 2–3), and phylogeographic studies by Richards et al. (2019) (Figs. 1–2). Richards et al. (2019) found three lineages of H. americanus occurring in the USA and Caribbean; however, based on our smaller sampling and distinct markers (that did not involve the mitochondrial control region used by those authors), we noticed two sympatric clades (Fig. 1A) where they named “Clade 3” (Richards et al., 2019) (Figs. 1–2) and we could not recover their “Clades 1 and 2”.

The second species with a wide distribution, H. guttatus, has also been shown to contain two lineages, as presented here and by Rodrigues Filho et al. (2020). Therefore, two wide-ranging marine coastal species were recently shown to occupy smaller areas than previously described, with currently valid species probably harboring more than one lineage and increasing the known diversity within the genus Hypanus.

Lineage delimitations. Based on five distinct lineage delimitation methods (sGMYC, mGMYC, mPTP, bPTP, and ASAP), we analyzed two clades within Hypanus that showed deeper divergences in phylogenetic analysis than what is expected for intraspecific evolution, since branches are longer between these lineages than within each one of them (Schwartz, Mueller, 2010).

Within the clade composed of H. guttatus, a species that supposedly occurs from Mexico to Southeastern Brazil, all analyses suggested a deeper divergence separating Central America’s samples from Brazilian ones than those within each area of occurrence. This scenario was also observed by Rodrigues Filho et al. (2020), besides a high genetic similarity among stingrays southern of the Amazon river mouth (0.19% of mt-co1 intraspecific distance, Tab. 5, . 3), which would be left under the valid name H. guttatus due to the species’ type-locality: “Brazil”. We infer that, even though H. guttatus is a marine and estuarine species that tolerates low salinities environments, the great freshwater and nutrients influx of the Amazon system may be a barrier for these stingrays, which resulted in isolated northern and southern lineages ( Rocha, 2003; Hoorn et al., 2010). It was also shown by Tosetto et al. (2022) that the small number of species in common between the Caribbean Sea and the Brazilian coast demonstrates their high isolation by the Amazon River Plume. This differentiation suggested by genetic analysis might be supported by morphological dissimilarities as well, such as morphometric differences in nostrils, spiracles, and caudal structures (FFP, pers. obs.; review in progress).

With regards to H. say, the lineage from the eastern coast of the USA has a different evolutionary history than that from the Gulf of Mexico, which is supported by all genetic analyses conducted here. Morphological differences could also justify this finding, such as differences in snout and caudal morphometries (FFP, pers. obs.; review in progress), leaving the lineage from eastern USA under the valid species name H. say according to its type-locality (New Jersey, USA), and that occurring in the Gulf of Mexico as H. aff. say until further analyses can be performed, and its taxonomic status evaluated. These results agree with the biogeographic proposals of Spalding et al. (2007) as both populations occur in the Warm Temperate North Atlantic province but in distinct ecoregions and separated by the Floridian one. The southernmost portion of the Florida Peninsula is known to have a detached ecosystem from the adjacent USA coast (Bowen, Avise, 1990). This can also be seen in H. americanus in which samples from the Bahamas (~50 Km from the US coast) are more closely related to those from the US Virgin Islands than those from Florida (Richards et al., 2019). Despite these results, gene transfer among lineages cannot be ruled out, which could result in hybridization and species not achieving reproductive isolation. As a consequence, there would be a disagreement between gene trees and species trees, a situation that might be underlying the evolution of freshwater stingrays Potamotrygoninae, as well as incomplete lineage sorting and diversification times (Fontenelle et al., 2021). These hypotheses should be further tested with nuclear genetic data (Petean and collaborators, working in progress).

The phylogenetic analysis of manta rays based on mitogenomes and nuclear exons (White et al., 2018) found pairwise distances between Mobula birostris (Walbaum, 1792) and M. alfredi (Krefft, 1868) as 0.4%; even though this distance might seem small for two species, their taxonomic identities as distinct species had already been suggested by morphometric, meristic, two mitochondrial, and one nuclear gene (Marshall et al., 2009; Kashiwagi et al., 2012). Likewise, the pairwise distances between the three cryptic lineages of Hypanus and the valid species to which they currently belong are small: between H. guttatus and H. aff. guttatus, 1.37%, H. say and H. aff. say,0.95%, and H. americanus and H. aff. americanus, 0.83%. Simultaneously, the largest distance between two Hypanus species is 13.55% in H. sabinus and H. marianae. Interspecific genetic distances within the genus vary from 0.82% (H. berthalutzae and H. rudis, Petean etal., 2020) to 13.55%, which is a high variation. However, intraspecific variations in Hypanus range from 0.047% in H. aff. say to 0.26% in H. aff. guttatus, which are values at least four times smaller than the interspecific distances.

There are many definitions of what “cryptic species” are and, even though there is still no consensus on their meaning (Struck et al., 2018), they could be “erroneously classified (and hidden) under one species name” (Bickford et al., 2007). Both sibling lineages to currently valid species H. guttatus and H. say are yet undescribed due to a lack of taxonomic studies and poor sampling. Subtle differences between species, with some of them being separated only by morphometrics, suggest a conservative morphology, making it more difficult to identify the species complex despite the allopatric pattern. Therefore, molecular markers such as mtDNA can be used to confirm species identification. Scenarios of cryptic speciation, with DNA sequences showing deep genetic divergences and morphological data revealing subtle diversity, have been observed in many non-elasmobranch fish clades: bonefish Albula Scopoli, 1777(Colborn et al., 2001),catfish Noturus Rafinesque, 1818(Egge, Simons, 2006), tubenose goby Proterorhinus Smitt, 1900 (Neilson, Stepien, 2009).

Genetic data have also been showing a higher species diversity in Elasmobranchs than formerly known, indicating the necessity of taxonomic work (Richards et al., 2009, 2019; Dudgeon et al., 2012; Borsa et al., 2016; Henderson et al., 2016; Sales et al., 2019; Fahmi et al., 2021; Gonzalez et al., 2021; Vilasboa et al., 2022; Kottillil et al., 2023). Through the combination of distinct tools, species’ hypotheses have been corroborated by independent studies, such as Gymnura van Hasselt, 1823 (Yokota, Carvalho, 2017, morphology; Rodrigues Filho et al., 2020, genetics; Vilasboa et al., 2022, genetics), Aetobatus Blainville, 1816(Richards et al., 2009, genetics; White et al., 2010, 2013, morphology and genetics; Sales et al., 2019, genetics), Rhizoprionodon Whitley, 1929 (Springer, 1964, morphology; Mendonça et al., 2011, genetics, Pseudobatos Last, Séret & Naylor, 2016 (Rutledge, 2019, morphology; Sandoval-Castillo, Beheregaray, 2020, genetics).

Our findings regarding these independent evolutionary units in Hypanus are only hypotheses of possible species as morphological and ecological data are recommended to be included since the use of exclusively molecular tools might lead to over or under-estimations of species (Carstens et al., 2013). This is due to species delimitation methods being unable to distinguish deep structure as a result of population-level processes or species boundaries (Sukumaran, Knowles, 2017). Therefore, to avoid failure, we are not describing any species until more data can be combined (Carstens et al., 2013).

Conservation. There is no threshold of genetic distances between lineages that should be regarded as populations and those that should receive a species status, since this is a faint boundary (De Queiroz, 2007; Roux et al., 2016). As mitochondrial evolution rates are slower in elasmobranchs than in other vertebrates (Martin et al., 1992), those mtDNA differences found here may represent distinct species. It is undoubtful that more studies are needed for a resolution of their taxonomic status; however, despite being categorized as species or populations, these lineages should be considered for conservation purposes (Henderson et al., 2016).

The urgency in identifying these lineages is because each entity in a threatened species complex might be even more endangered than the nominal species as a whole, and may need distinct conservation measures (Bickford et al., 2007). All current valid Hypanus species have been recently evaluated by elasmobranch specialists at IUCN (2020). Of the 13 evolutionary units identified in this study, only one is clearly under low risk, H. sabinus (Least Concern), while six are under some risk of extinction, with criteria used to evaluate each species threatened category in parenthesis: Hypanus berthalutzae (A2d), H. dipterurus (A2d), and H. longus (A2d) are Vulnerable, H. marianae (A2cd) is Endangered, and H. rudis (A2d) and H. geijskesi (A2d)are Critically Endangered. A concerning situation regards the three species-complexes (H. americanus (A2bd), H. guttatus (A2d), and H. say (A2bd)) since they had their threatened status recently evaluated and were considered as Near Threatened (Carlson et al., 2020a,b,c). However, after this and previous studies (Petean et al., 2020; Richards et al., 2019; Rodrigues Filho et al., 2020), we recognized each of them might be, at least, two evolutionary lineages. Therefore, the possible restriction of each clade’s geographic range could have an impact on their threatened categories, with consequences on management proposals. The recently described H. berthalutzae is the most recent example of this scenario since it was considered H. americanus and encompassed the largest species distribution within the genus. Soon after its description, the species was evaluated and already classified as Vulnerable (Charvet et al., 2020), thus demonstrating that lineages currently unknown can already be under threat before their formal description and evaluation as a species.

Throughout evolution, some lineages might be described as species due to population isolation, while others present high genetic variability and each may be named Evolutionarily Significant Unit (ESU) (Coates et al., 2018) in an attempt to identify independent entities for conservation and perpetuation of their evolutionary history (Diniz-Filho et al., 2013; Hoezel, 2023). These ESUs should be the focus of management efforts (Ryder, 1986; Waples, 1991, 1995; Moritz, 1994).

Currently, conservation aims mostly on valid species, ignoring genetic diversity. Due to the existence of several species concepts and species’ delimitation methods, there are many conflicts within taxonomy; besides, scientists do not comply on how to deal with ESUs leading to difficulties in actually applying measurements (Coates et al., 2018). Therefore, we suggest the evolutionary lineages hereby identified, even if not formally described as species, to be treated as ESUs and thus be the target of threat evaluation.

To conclude, based on 13 protein-coding mitochondrial genes, there is enough support for the monophyly of the resurrected genus Hypanus by Last et al. (2016a) after the description of a new species (H. berthalutzae) and the transference of Fontitrygon geijskesi to Hypanus, becoming Hypanus geijskesi due to its close relationship to H. guttatus. Besides the recognition of a cryptic species within H. americanus by Petean et al. (2020), we have also identified evolutionary lineages that represent currently known species, as well as suggested two putatively new ones not detected until now, which are sister-lineages to H. guttatus and H. say,thus reducing their geographic distribution, with possible impacts on their conservation status.

These results leave the genus Hypanus with 13 independent evolutionary units, of which 10 are valid species and three “affinis” to their siblings (H. aff. americanus, H. aff. guttatus, and H. aff. say). Further formal descriptions of these new lineages will have consequences on their conservation status since current areas of distribution of valid species will decrease with their division into more than one entity, leading to an urgency in evaluating their threatened status and proposing conservation measures, actions that could already begin with ESUs before descriptions. Even though we have delimited some evolutionary lineages within the genus, maybe more could be found with wider sampling. Finally, to rigorously evaluate these species complexes, morphological studies, the examination of type series, and ecological niche modeling should be performed to better define these stingray species and their geographic distributions.