Phylogeny of the genus Pinnixa White, 1846 (Crustacea, Brachyura, Pinnotheridae) and allies inferred from mitochondrial and nuclear molecular markers, with generic reassignment of twenty-one species

ABSTRACT We used mitochondrial 16S-NADH1 complex, mitochondrial 12S, and nuclear histone 3 genes to infer a molecular-based phylogeny, which allowed us to study phylogenetic relationships between species of Pinnixa White, 1846 sensu lato and other closely related pinnotherids. Polyphyly of Pinnixa s.l. was confirmed by maximum likelihood analyses. By our restricted definition, the genus Pinnixa sensu stricto is represented in these analyses only by its type species, Pinnixa cylindrica (Say, 1818). As a result of these molecular analyses, in combination with morphological studies, twelve species are reassigned to existing genera: Pinnixa faba (Dana, 1851), Pinnixa franciscana Rathbun, 1918, Pinnixa littoralis Holmes, 1894, Pinnixa schmitti Rathbun, 1918, and Pinnixa tubicola Holmes, 1894 are placed in the genus Scleroplax Rathbun, 1893, whereas Laminapinnixa miamiensis McDermott, 2014, Laminapinnixa faxoni (Rathbun, 1918), Pinnixa abbotti Glassell, 1935, Pinnixa arenicola Rathbun, 1922, Pinnixa floridana Rathbun, 1918, Pinnixa leptosynaptae Wass, 1968 and Laminapinnixa vanderhorsti Rathbun, 1924 are reassigned to Glassella Campos & Wicksten, 1997, the last two strictly on morphological bases. In addition, three new genera are erected to receive nine species: Rathbunixa n. gen. includes members of the Pinnixa pearsei Wass, 1955 – Pinnixa sayana Stimpson, 1860 complex, Pinnixa affinis Rathbun, 1918, and species in the Pinnixa californiensis Rathbun, 1894 – Pinnixa occidentalis Rathbun, 1893 complex; Tubicolixa n. gen. includes Pinnixa chaetopterana Stimpson, 1860 and the Pinnixa brevipollex Rathbun, 1898 – Pinnixa rapax Bouvier, 1917 complex; and Sayixa n. gen. is established for Pinnixa monodactyla (Say, 1818).


INTRODUCTION
Pinnixa White, 1846 is a long-standing pinnotherid genus, second in history only to the type genus of the family, Pinnotheres Bosc, 1802. As currently regarded, Pinnixa sensu lato (s.l.) is comprised of 51 species (the 50 species listed in the supplementary material of Palacios Theil et al. (2016) with the addition of P. hendrickxi Salgado-Barragán, 2015), which makes it also the second largest genus of the Pinnotheridae De Haan, 1833. Most of the species are from North to South American coasts, both Atlantic and Pacific, including the Gulf of Mexico and the Caribbean Sea. Six are, however, found in Indo-Pacific waters, with the distribution for one of them, Pinnixa penultipedalis Stimpson, 1858, reportedly reaching from Siberia to as far as the coasts of Mozambique (Schmitt et al. 1973). The genus has undergone partial revisions in recent decades resulting in reassignment of some species and descriptions of new genera (Manning & Felder 1989;Campos & Wicksten 1997;Ng & Naruse 2009;McDermott 2014;Palacios Theil et al. 2016), but it has never been subjected to comprehensive revision. About a third of the species within Pinnixa s.l. were described in the 19 th century and more than half in the early 20 th century, primarily in the course of studies based on museum materials. Most of these were collected during exploratory expeditions, especially along American coasts, as for example during the Albatross campaigns and similar efforts (Rathbun 1894(Rathbun , 1898Glassell 1935a, b). It is likely that these species were placed in Pinnixa s.l. because of their sharing a consistently much-wider-than-long carapace and a third ambulatory leg longer than the others. However, these characters are probably not synapomorphies but rather convergent adaptations to a symbiotic life within elongated habitats such as the tubes of polychaete worms, burrows of sipunculans, burrows of infaunal decapods, or the cloacal lumens of holothurians.
Preliminary molecular analyses have indicated that Pinnixa s.l. is polyphyletic in its present composition on the basis of the mitochondrial complex formed by part of the 16S gene, the tRNA-Leu, and part of the gene for NADH dehydrogenase subunit I (Cuesta et al. 2002;Palacios Theil et al. 2009). However, the number of species included in those molecular phylogenetic revisions was very limited. Here we increase the number of taxa analyzed, while adding another mitochondrial gene (12S) and the nuclear gene for histone subunit 3 (H3). On this basis we reexamine phylogenetic associations among species of Pinnixa s.l., as well as their relationships to other pinnotherid genera. Of special interest are the relationships of Pinnixa s.l. to other taxa within the subfamily Pinnixinae clarify its phylogenetic relationships to other pinnotherid genera, samples of other pinnotherid taxa available as sequence quality materials were included in phylogenetic analyses using a concatenated alignment based on three fragments (Table 1). Representative species of the families Gecarcinidae MacLeay, 1838, Grapsidae MacLeay, 1838, Ocypodidae Rafinesque, 1815, Sesarmidae Dana, 1851, and Varunidae H. Milne Edwards, 1853 were used as outgroups.
Dna extraction, amplification, anD Sequencing When specimens were large enough, the carapace was lifted at the posterior edge in order to obtain thoracic muscle tissue used for DNA extraction. Sometimes gills were also used but only after careful examination to rule out the presence of crustacean parasites. Total genomic DNA was extracted using the DNeasy Blood and Tissue Kit (QIAGEN, Valencia, CA, USA) or a standard DNA extraction protocol (Robles et al. 2007). Diluted total DNA was amplified by means of a polymerase chain reaction (PCR) following the recommendations of the Taq polymerase's manufacturers (AmpliTaq Gold® DNA Polymerase, Applied Biosystems, Foster City, CA, USA; M0273S Taq DNA Polymerase, New England BioLabs, Ipswich, MA USA; or DreamTaqTM Green DNA Polymerase, Fermentas, currently ThermoFisher Scientific, Waltham, MA, USA) in a Stratagene® Robocycler® Gradient 96 (Santa Clara, CA, USA). Three fragments were targeted: 1) a mitochondrial complex of about 830 bp consisting of part of the 16S rRNA gene, the tRNA-Leu, and part of the gene for NADH dehydrogenase subunit 1 (NADH1), 2) part of the mitochondrial 12S rRNA gene, and 3) part of the nuclear histone subunit 3 gene. These genes represent a diverse group of mitochondrial protein-or RNA-coding genes, as well as a nuclear protein-coding gene. They have been previously shown to be informative and have an adequate resolution to solve phylogenetic relationships between and among species and genera within Pinnotheridae (Palacios Theil et al. 2016;Tsang et al. 2018). In addition, they have proved to be obtained with relative ease using the primers indicated in Table 2, which shows also the length of the fragments obtained with each primer combination. PCR products were either sent to be purified and sequenced by Beckman Coulter Genomics (Danvers, MA, USA) or purified using SureClean (Bioline, Taunton, MA, USA), resuspended in water, sequenced with the ABI BigDye terminator mix, sequencing products being run on a 3100 Applied Biosystems (Foster City, CA, USA) automated sequencer.

molecular analySeS
The obtained sequences were assembled and manually edited when necessary with the program Sequencher 5.0 (Gene Codes, Ann Arbor, MI, USA). Preliminary alignments were assembled with BioEdit 7.1.3.0 (Hall 1999) and subsequently tested for accuracy with MAFFT (Multiple Alignment using Fast Fourier Transform, Katoh et al. 2002) on the website of the European Bioinformatics Institute (www.ebi.ac.uk, last visited 12 th Oct 2018). Poorly aligned positions were identified with Gblocks v. 0.91b (Castresana 2000) on the server of the Mediterranean Center for Marine and Environmental Research (CMIMA, molevol.cmima.csic.es/ castresana/ Gblocks.html, last visited 12 th Oct 2018). As a result, 87% of the positions were used for the 16S-NADH1 fragment, and 86% of the original positions for 12S. For histone 3 no Gblocks analysis was necessary and the whole fragment could be used. The three resulting alignments were concatenated for more accuracy in the analyses (Gadagkar et al. 2005). Total sequence length was 1445 bp, as a result of concatenating 776 bp from the 16S-NADH1 fragment, 340 bp from 12S, and 327 bp from the histone 3 gene. For the 16S-NADH1 fragment 593 bp corresponded to the 16S rRNA gene, 68 bp to the gene for tRNA-Leu, and 98 bp to the gene for NADH1, with a fragment of 14 bp corresponding to an intron located between the tRNA-Leu and the NADH1 genes. The alignment was submitted to the Cyberinfrastructure for Phylogenetic Research (CIPRES) web portal (www.phylo. org, last visited 12 th Oct 2018) for Randomized Accelerated Maximum Likelihood (RAxML-HPC2 on XSEDE) analysis (version 8.2.8, Stamakis 2014) with 1000 bootstraps, the maximum number of bootstraps allowed by the tool, and supplying the information for the partition, which included six fragments: the 16S rRNA gene (1-593 bp), the gene for tRNA-Leu (594-662 bp), a fragment corresponding to an intron (663-677 bp), the gene for NADH1 (678-776 bp), the 12S rRNA gene (777-1117 bp), and the histone 3 gene (1118-1445 bp). The analysis showed that GTR was the best nucleotide substitution model for all fragments. This model was subsequently applied for a Bayesian phylogenetic analysis, performed using MrBayes on XSEDE also on the CIPRES web portal (Ronquist et al. 2011). The analysis was run with four Markov chains for 10 000 000 generations, sampling one tree every 1000 generations and with the burn-in percentage set to 25% of the samples. The resulting trees were analyzed and edited with Mega 5.2 (Tamura et al. 2011). The sequences and the complete alignments were submitted to GenBank for public access ( Table 1). The majority of the sequences, those with GenBank accession numbers starting with EU and KU, had been made available by us as a result of previous publications (Palacios Theil et al. 2009, 2016, whereas the sequences with accession numbers starting with MN have been used here for the first time.

morphological examination anD illuStration
The specimens were examined under a Wild Heerbrugg dissecting scope, and selected characteristic parts illustrated with the aid of a Leica camera lucida. Smaller parts were examined under an Olympus BH2 compound microscope and a Nikon inverted compound microscope. Hand drawings were scanned and thereafter edited with the graphic design software programs Adobe Illustrator ® and Adobe Photoshop ® (Adobe Systems, San Jose, CA, USA). In some instances, previously published line illustrations were adapted for use, provided they were public domain or allowed by special permission. Measurements, where reported, were rounded to the nearest 0.1 mm.

RESULTS
genetic analySeS DNA extraction and sequencing were successful for all the specimens of Pinnixa included in the analyses, with the exception of the 12S sequence for Pinnixa chaetopterana Stimpson, 1860 (ULLZ 12537). This sequence was obtained only after repeated attempts and was not of good quality, probably due to the presence of multiple fragments of similar size, possibly indicating pseudogenes or contamination. About 10% of the positions in the obtained fragment were ambiguous, and the sequence was about 80% similar to other 12S sequences of P. chaetopterana, a species to which it was thought to be related based on morphology. For pinnotherids, the similarity between 12S sequences from different specimens of the same species is higher than 95%. Only the 16S-NADH1 and histone 3 fragments for the problematic specimen were included in the analyses. Results related to this sample must therefore be interpreted with caution ( Fig. 1). Attempts at obtaining the histone 3 sequence for Pinnixa latissima Coelho, 1997 (ULLZ 14136) were unsuccessful and for this sample only the 16S-NADH1 and 12S fragments were available. However, these fragments were of uncertain quality. In addition, no re-extractions could be performed without risking excessive damage to the only available specimen. For these reasons we chose to exclude this sample from the alignment. Additional samples were available that could not be included in the analyses due to our inability to obtain sequences. In these cases, DNA extractions were of poor quality or yielded low concentrations, and, although PCRs were attempted repeatedly for different genes, they did not succeed. These included specimens of Pinnixa cylindrica (Say, 1818) (USNM 1192250), Pinnixa latissima (USNM 1192248), Pinnixa lep-tosynaptae Wass, 1968 (ULLZ 14834), and some specimens that belonged to the Pinnixa faxoni complex, probably representing the poorly defined Laminapinnixa faxoni (Rathbun, 1918) or L. vanderhorsti Rathbun, 1922 (ULLZ 4430 and USNM 1192261).
Among all the taxa included in the analyses, the species of Pinnixulala formed a clade separate from both the subfamily Pinnotherinae De Haan, 1833 and the subfamiliy Pinnixinae ( Fig. 1). It included the species Pinnixulala valerii (Rathbun, 1931), the tentatively identified Pinnixulala petersi (?) (Bott, 1955), Pinnixulala retinens (Rathbun, 1918), and Pinnixulala sp., and none of the species of Pinnixa, or any other taxa which show a carapace wider than long, were closely allied to them.
The rest of the species within Pinnixa s.l. formed six subclades within the subfamily Pinnixinae. Five of the subclades formed a large clade, which included Pinnixa cylindrica (type of the genus Pinnixa), as well as P. monodactyla (Say, 1818); the P. chaetopterana, P. sayana and Scleroplax complexes (Fig. 1); and a subclade formed by those species of Austinixa included in the analysis. Support values for this clade were high (100/100). Within the clade, P. monodactyla separated from all other species at a basal node. Noteworthy were also the very small genetic distances observed among the included samples of P. cylindrica, even though these ranged from locations along the Atlantic and Gulf of Mexico coasts from North Carolina to Texas. The sixth subclade within subfamily Pinnixinae grouped some species of Pinnixa with Glassella costaricana Wicksten, 1982, Indopinnixa kumejima Naruse and Maenoso, 2012, Indopinnixa moosai Rahayu and Ng, 2010, Laminapinnixa miamiensis McDermott, 2014, and L. faxoni (Rathbun, 1918), forming the Glassella-Indopinnixa complex, with high support (95/100). It included the Atlantic species P. arenicola Rathbun, 1922, P. floridana Rathbun, 1918, Pinnixa sp., and the Pacific P. abbotti Glassell, 1935 (Fig. 1, Glassella-Indopinnixa complex). In addition to the aforementioned, an unidentified pinnotherid species from the Pacific coast of Mexico (ULLZ 9337), probably symbiotic with worms, also resolved within the subfamily Pinnixinae. It occupied a monotypic branch  Rathbun, 1918, P. littoralis Holmes, 1894, P. schmitti Rathbun, 1918, and P. tubicola Holmes, 1894, in addition to a juvenile specimen tentatively identified as P. scamit (?) Martin & Zmarzly, 1994. It was notable that genetic distances among the taxa included within this group were rather small when compared to the distances among species in other subclades ( Fig. 1, Scleroplax complex).
The Scleroplax complex was a sister clade to the P. sayana complex, which was represented by a highly supported clade (100/100) composed of the Atlantic species P. sayana Stimpson, 1860 andP. pearsei Wass, 1955, along with the Pacific species P. affinis Rathbun, 1898 and P. occidentalis Rathbun, 1894. The samples morphologically identified as P. pearsei and P. sayana within this group did not separate into two subclades but instead formed one polyphyletic clade, with small genetic distances among the taxa included ( Fig. 1, P. sayana complex).
Lastly, representatives of the western Atlantic species P. chaetopterana and P. rapax Bouvier, 1917 joined in a highly supported clade (100/100). Within this clade, P. chaetopterana and P. rapax grouped together, with no or low support (-/67) for the separation between P. rapax and the samples of P. chaetopterana from the different locations available. However, samples of P. chaetopterana from the Caribbean Sea (Venezuela and Belize) formed a highly supported subclade (98/99), as did the two samples from the northern reaches of the western Atlantic (North Carolina and Fort Pierce, Florida), although with lower support values (94/71). Four additional samples from the Gulf of Mexico also grouped together in topology, although this subclade showed significant support values only for the Bayesian analysis (-/92). The one sample of P. rapax was nested together with the Caribbean samples of P. chaetopterana, but again without significant support (-/67) ( Fig. 1, P. chaetopterana complex).
In the Glassella-Indopinnixa complex all American species included in the analyses, with the exception of an undetermined species from Panama (ULLZ 13337 and ULLZ 14141), formed a highly supported group (100/100), including among others those species recently reassigned to the genus Laminapinnixa McDermott, 2014, as well as Glassella costaricana. This group appeared as a sister clade of the two Asian species analyzed (Indopinnixa kumejima and I. moosai). The undetermined Panamanian species separated from them at a basal node within the complex (Fig. 1, Glassella-Indopinnixa complex).

morphological examinationS
Morphology of the members of the six clades resolved in the molecular phylogenetic analyses was re-evaluated in order to reverse-engineer arrays of morphological characters that supported these clade separations. Most commonly, these characters were found to include features of the carapace, the robustness and setation of the pereopods, shape of the male pleon, and proportions of the third maxilliped segments (Fig. 2).
Most members of the Glassella-Indopinnixa complex could be readily separated from the other taxa within the subfamily Pinnixinae by, among other characters, morphology of the third maxilliped. Except for Glassella costaricana, they all showed a third maxilliped with a long palp, where the club-shaped dactylus was as long or nearly as long as the ischiomerus and with it inserted at the proximal end of a stout conical propodus. The dactylus and the propodus were oriented in a wide angle, sometimes almost perpendicular, to each other (Fig. 2H). In Glassella costaricana the dactylus was strongly reduced, and it inserted at the distal portion of the propodus (Fig. 2G). In all other taxa examined for the subfamily Pinnixinae the third maxilliped showed elongate dactylus and propodus, similar in shape and size to each other, reaching to or past half the length of the ischiomerus (Fig. 2C, L, P, T, X). Additionally, the nine species in the Glassella-Indopinnixa complex shared a smooth but punctate carapace, relatively stout legs with dactyli shorter than the propodi (especially for P4 and P5; Fig. 2E), and a setose, somewhat elongate cheliped with relatively straight fingers and one or more rows of tubercles or granules on the outer surface of the palm, running along its length, the inferior one usually continuing along the fixed finger (Fig. 2F). Within the Glassella-Indopinnixa complex, Glassella costaricana, P. arenicola, P. abbotti, Indopinnixa kumejima, and I. moosai were easily distinguishable from each other on the basis of key characters from descriptive literature, as were males of Laminapinnixa miamiensis, L. faxoni, and P. floridana, three species with partially overlapping distributions. The latter three species could be easily distinguished by the shape of their pleon and the relative development of a gonopodal plate. However, the females could be discriminated only by subtle differences in the relative length of the pereopods dactyli, sharpness of the ridge running along the anterolateral margins, numbers, and positions of teeth on the posterior margin of the third ambulatory leg (P4), relative setation of legs and carapace margins, or coloration of the carapace. P. pearsei, P. sayana, P. affinis, and P. occidentalis were observed to share a carapace with defined regions and a sharp cardiac ridge, slender legs with a long slender merus, and smooth chelipeds with a strongly reduced or deflexed fixed finger. Their third maxilliped also had a club-shaped propodus and dactylus oriented nearly parallel to each other. (Fig. 2J-M). According to Wass (1955) the differences between P. pearsei and P. sayana are evident in a higher and straighter cardiac crest, a wider carapace, and a broader propodus of P4 in P. pearsei. However, we observed great variation in these characters, as well as in the morphology of the chelipeds (Fig. 3A-E). The size of the examined male specimen of P. occidentalis was smaller (cw = 7.25 mm, cl = 3.35 mm as opposed to cw = 9.5 mm, cl = 19.5 mm) and somewhat less granulate than indicated in the literature. However, carapace, legs, and especially the chelipeds, with a characteristic large blunt tooth on the margin of the fixed finger (Fig. 3F, with chela of P. affinis for comparison, Fig. 3G), matched the species description (Rathbun, 1894).  fig. 1, fig. 2c, a, with permission from Allen Press); H, I, Glassella faxoni (Rathbun, 1918)  Besides the sample used for the molecular analyses (ULLZ 8713), undoubtedly identifiable as a male of Pinnixa monodactyla, the only other specimen of this species available to us was an immature female collected with a box dredge in the northern Gulf of Mexico from about 39 m deep (ULLZ 8569). Despite the differences in size, sex, and origin, they showed obvious similarities. They presented a characteristic cheliped palm with a fixed finger reduced to a sharp spine, with an additional tooth at the base of the cheliped dactylus, and an elongated carapace with a tubercle on each anterolateral angle. Ambulatory legs were slender, cylindrical, with slender straight dactyli (Fig. 2N-Q).
Species in the Scleroplax complex shared a hard convex carapace and a third maxilliped with both the propodus and carpus long and spatulate. With the exception of P. scamit, they all had cylindrical legs (Fig. 2R-U). Unlike the other species in the complex, the examined juvenile specimen of P. scamit had slender legs, similiar to those observed for P. affinis and other species included in the P. sayana complex. In the Scleroplax complex, morphology supported the similarity between P. littoralis and P. faba, as indicated by Zmarzly (1992). They could be distinguished from each other only by the geometry of the cheliped fingers. This was easily observed for males, but for females the differences were again rather subtle. In females of P. littoralis the fixed finger of the cheliped is "slightly deflexed", and a gape is visible when the fingers are closed, as opposed to a "nearly straight" fixed finger and no gape in P. faba (Zmarzly 1992).
The specimens of P. chaetopterana and P. rapax were similar in having a carapace with clearly delimited regions, and in the relatively strong and pubescent chelipeds with a shortened or deflexed fixed finger in males, the slender dactyli of the ambulatory legs, and concentrations of pubescence on the pereopods and carapace margins. Their third maxilliped had both the elongate propodus and dactylus of similar size and shape (Fig. 2V-Y). The geographically separated samples of P. chaetopterana, similar to observations for P. pearsei and P. sayana, show variability in the morphology of the chelipeds, as well as in the relative length of the articles of the ambulatory legs, especially the dactyli. In addition, there were differences in number and sharpness of the granules and teeth on the edge of the subbranchial region as well as on the posterior surface of the P4 merus. The specimen of Pinnixa chaetoperana from Belize (ULLZ 12537) was a small juvenile, similar to P. chaetopterana in appearance of a carapace with clearly defined regions and in the denticulate meri of the fourth and fifth pereopods. The cheliped was similar to the female chelipeds for P. chaetopterana (see Fig. 2W). On the other hand, it was not as setose as most other specimens of P. chaetopterana.

remarkS
The morphological similarities among species transferred to this genus have in most cases been noted previously (Rathbun 1918(Rathbun , 1924Glassell 1935a;McDermott 2014), although their resemblance to Glassella of Campos & Wicksten (1997) has likely remained unnoticed because of the weight given to differences in the third maxilliped (Fig. 2G, H), often an important character in pinnotherids. However, there are other cases among pinnotherid genera for which striking variation in the third maxilliped has been observed among closely related species. In Calyptraeotheres Campos, 1990 some species lack a third maxilliped dactylus, whereas in others it is present (Hernández-Ávila & Campos 2006). The genus Laminapinnixa was erected by McDermott (2014) to accomodate L. miamiensis, a newly described species for populations that we and colleagues had long regarded on morphological evidence as closely related to Glassella faxoni n. comb. and G. vanderhorsti n. comb. The genus was diagnosed in part by the presence in males of a plate (therein termed "abdominal plate" = our gonopodal plate) derived from the pleon (posterior to the gonopods), which in the description and discussion by McDermott (2014) are portrayed as a structure that must be "lifted" to expose the gonopods during copulation. However, this structure is in all unmanipulated examples of G. miamiensis n. comb. that we have examined positioned to extend from its more posterior origin obliquely between the gonopods (derived from the first pleonal somite) and then anterior to them, where it usually broadens terminally and separates the gonopods from the sternum when the pleon is flexed (as shown clearly by McDermott, 2014: fig. 7B). This would appear to not necessarily separate them from exposure for copulation when the pleon is extended (or, as they appear when illustrated with the pleon removed, Fig. 5F). The broadened terminus of the plate in at least one related species can also perhaps move from anterior or posterior of the gonopods, provided the gonopods are flexed laterally, though this seems unlikely in G. miamiensis n. comb.
In the course of describing and investigating potential relationships of G. miamiensis n. comb. and seeking evidence of other species having this gonopodal plate, McDermott (2014) noted the unfortunate loss of types for potentially related species to which he had wished to make comparisons. By way of further explanation, most of these were among 98 pinnotherid specimens permanently lost to science when destroyed by the U.S. Postal Service, owing to the mishandling of a loan return shipment by a borrower in 2006. While this loss has also limited our own comparative efforts, one of us (DLF) and his late colleague Robert H. Gore had independently examined types and other now lost materials decades ago, at the time making rough-sketches of selected structures, several of which are herewith directly reproduced given the void they fill (Fig. 4). In addition, high-quality, previously unpublished illustrations by several Smithsonian Institution illustrators, contracted by Robert Gore or the late Waldo Schmitt (the latter for a never-published manuscript by W. L. Schmitt  Felder, 1979Felder, -1982 and E. S. Davidson), are in some cases annotated so as to be clearly identifiable with the now-lost types or other materials on which they were based (Fig. 5). From this evidence, it is clear that both G. faxoni n. comb. and G. vanderhorsti have forms of the gonopodal plate in mature males that we regard to be homologs of that in G. miamiensis, in addition to their sharing a number of other characters that group them with G. miamiensis and its herewith assigned congeners. The illustrated male gonopodal plate for Glassella vanderhorsti, published by McDermott (2014) but reproduced here from the original figures with credit and voucher indicated (Fig. 5F), was in fact based on the Zoological Museum Amsterdam (now Naturalis Biodiversity Center, Netherlands) holotype male (NCBN-ZMA.C RUS.D 242234). In a very similar but smaller topotypic male specimen (USNM 56903), also collected by van der Horst but now among the lost materials, the gonopodal plate was found to be somewhat longer but terminally very similar to the holotype male and like that of G. miamiensis in that its terminal reaches were positioned anterior to the gonopods (Fig. 4E, F). Only in G. faxoni n. comb. was the male gonopodal plate found to be small enough to be positioned largely between the gonopods or to freely move its arched terminal lobe from anterior to posterior of the gonopods (Fig. 4C, D). As evident, the cataloged specimens upon which our figures, sketches, and notes are based were the same as examined by Rathbun (1918Rathbun ( , 1924. This confirms personal communications of E. S. Davidson regarding gonopodal plates in these species, as were mentioned by McDermott (2014), who found the lack of Rathbun mentioning these plates in descriptions of G. faxoni n. comb. and G. vanderhorsti as reason to question their congeneric assignment with G. miamiensis. We suspect that Rathbun at the time attached little importance to gonopods as characters, especially among pinnotherids in the early years while she was working primarily with a hand-lens.
While the male gonopodal plate, or some ramification of it, may prove to unite most if not all species that we here assign to Glassella, our molecular phylogeny includes species in which it is at very least not reported to date, or for which intact male specimens are lacking. This remains the case for the generic type species, G. costaricana, known only from a female at the time of description. Our phylogenetic analyses included two additional females, but we also attempted inclusion of tissues from a very small mutilated male specimen (UF 18960) that appeared to be this species. Its identity as G. costaricana was confirmed by clear match of its 16S mitochondrial sequence to those of the females, but it was not included in the final phylogenetic analysis for lack of additional sequence data. Unfortunately, damage to its pleon and sternum obliterated evidence that might have made obvious the presence or absence of a gonopodal plate, leaving that question unresolved. We strongly suspect that the "enclosing sheath" of the gonopod in G. arenicola, as reported but only partially illustrated by Thoma et al. (2009), could represent yet another variation in this structure, and this species clearly groups with those that have the more obviously developed gonopodal plate. Glassella miamiensis and G. faxoni share the plate and are closely related both morphologically and genetically (Fig. 1), but we cannot yet determine if G. leptosynaptae and G. vanderhorsti (at least the latter of which also has the plate) are included in that same well-supported molecular genetic clade for present lack of sequence quality material.
Thorough study of the first pleonal somite in mature males for all other suspected members and close relatives of Glassella is required to determine if any homologous ramification of the gonopodal plate may have also in those been thus far overlooked. This is to be undertaken in the course of coming descriptions of the American species "Pinnixa sp. (ULLZ 13337 and ULLZ 14141)" from Panama (Fig. 1) and at least three additional new western Atlantic species that clearly represent Glassella in morphology, but for which we at present lack sequence quality materials. Further studies must also include molecular and morphological examinations of potentially related species, especially the eastern Pacific American species Pinnixa bahamondei Garth, 1957, P. darwini, Garth, 1960, P. hendrickxi Salgado-Barragán, 2015, P. pembertoni Glassell, 1935, and perhaps P. transversalis (H. Milne Edwards & Lucas, 1844. From our preliminary morphological observations of materials used in the present molecular study, we can state that a clear ramification of the male gonopodal plate is present in the eastern Pacific species G. abbotti, underpinning our inclusion of it in Glassella on the basis of more than solely molecular phylogenetics.
Further studies are also required to more thoroughly compare the Indo-West Pacific genus Indopinnixa to the American Glassella. We retain separation of these genera as sister clades, though only two of the seven species assigned to Indopinnixa could for the present be represented in the molecular genetic analysis. Furthermore, only one of these two species was represented by an intact male specimen, the other represented only by a donated tissues sample. The intact male of I. kumejima was stained and carefully examined, and no evidence of a gonopodal plate or ramification thereof could be found, suggesting this could be a character of use in separating at least some species of the two genera.
remarkS Morphological similarities among some species of this genus have been noted previously, though always between species sharing an ocean basin such as the eastern Pacific pair, R. affinis and R. occidentalis, and the western Atlantic pair, R. pearsei and R. sayana (Rathbun 1918;Wass 1955;Zmarzly 1992). We have observed great variability in the morphological characters that define R. pearsei and that are reported to differentiate it from R. sayana. Wass (1955) described the former species to separate specimens found in northwestern Florida from R. sayana, the distribution of which was known at that time to range from Massachusetts to Sarasota Bay, in southwestern Florida. Later records extended the distribution of R. sayana to Grand Isle, Louisiana, and Brazil (Schmitt et al. 1973). In addition, we have samples that fit the morphological characters of R. sayana from Corpus Christi, Texas. We also have collections of specimens matching the description of R. pearsei from Atlantic coast of Florida, Gulf of Mexico waters in southern Florida, and Gulf Shores, Alabama. All these samples are genetically very close in relationship (Fig. 1). This suggests that R. pearsei should be regarded as a junior synonym of R. sayana. However, the type of R. sayana is not extant, and the type locality is the mouth of Beaufort Harbor, North Carolina, a location we were unable to represent among collection sites for our samples of R. sayana, all of which are well to the south. Thus, we for now lack genetic evidence upon which to base genetic re-evaluation of these two taxa, and retain both names.
When Rathbun (1894) described Pinnixa occidentalis and P. californiensis she noted the resemblance between the two, but nonetheless treated them as separate species, though she later synonymized them (Rathbun 1918). However, more recently smaller and less granulate variations of R. occidentalis have been reported, indicating that this taxon should be treated as a "group of allied species" (Hart 1982). The specimen we examined is probably one of these variants. Whether or not some of these variants could possibly match the description of R. californiensis requires further investigation. For now, we elect to retain R. californiensis as a separate taxon, following Ng et al. (2008). The material of R. occidentalis included here was collected in Panama, expanding the southern limit of the species range, which was formerly Magdalena Bay, in Mexico (Schmitt et al. 1973). Despite the fact that we were unable to analyse additional samples of the R. californiensis -occidentalis complex, we provisionally assign both species to this genus, based on their long recognized relationship. Thorax transversely subeliptical, narrowing each side to the middle of the lateral edge, which is rounded, a tubercle each side marking the situation of the anterior lateral angles, surface punctured; orbits suborbicular; anntennae [sic], exteriors subequal to the breadth of the clypeus; hand oblong, somewhat quadrate; palm concave and ciliated in the middle, a spiniform angle instead of a finger, with a tooth at its base, and another at the base of the thumb larger; thumb abruptly incurved at base, rectilinear towards the tip, with an angle at the interior middle, tip acute, attaining the tip of the spiniform angle; feet, second, fifth and third pairs subequal, the latter rather larger, fourth pair larger, and with the fifth pair with somewhat dilated tibia; abdomen with a few larger punctures, terminal joint rounded at tip, entire, ciliated and attaining the tip of the geminate joints of the pedipalpi. Length three tenths, breadth one half an inch. This curious animal occurs in the Richmond Museum. Mr. J. Warrell, the proprietor of that interesting establishment, supposes it to be American, but whether from our eastern or western coast he could not say. It is particularity remarkable in having monodactyle hands, a character which in a very rigid arrangement would not only separate it from the genus Pinnotheres, but also from the preceding species as a distinct genus. The tibia of the fourth and fifth pairs of feet are somewhat dilated, but the corresponding tarsi are accidentally wanting in this specimen." DiagnoSiS. -Carapace transversally subeliptical, wider than long, punctate, narrowing toward rounded lateral edges; anterolateral margins each with single lobiform tooth or tubercle near or just anterior to lateral extreme. Third maxilliped with ischiomerus subtrapezoidal; propodus and dactylus elongate, longer than carpus; dactylus inserting near base of propodus, reaching beyond end of propodus. Chelipeds heavy, palm lacking longitudinal lines of setae; cheliped fixed finger strongly shortened, reduced to spiniform angle, with sharp tooth at base of dactylus. First two ambulatory legs (P2, P3) slender, P4 and P5 somewhat stouter; lengths P4 > P3 > P2 > P5. Male pleon subtrapezoidal, somites 4-6 constricted; telson oblong subellipsoidal, much wider than long.
etymology. -Named for Thomas Say, author of the type species of this new genus, and first author to describe pinnotherid species after Linneaus. material examineD. -In addition to the material included in the phylogenetic analyses (Table 1)

remarkS
In describing the species Pinnotheres monodacytlum, later transferred to Pinnixa, Say (1818) indicated that this taxon presented characters that "would not only separate it from the genus Pinnotheres, but also from the preceding species as a distinct genus". The "preceding species" he is referring to is Pinnotheres cylindricum, which would become later the type of the genus Pinnixa. He discusses in that work the differences between the two species and the genus Pinnotheres, but he chose to maintain both within the genus Pinnotheres. Later, in 1846, Adam White, assistant in the Zoological Deparment of the British Museum, established the genus Pinnixa for P. cylindrica on the basis of its carapace being much wider than long, its having a larger cheliped palm when compared to Pinnotheres, and on the relative lengths of the ambulatory legs. He, however, did not include what we herewith assign to Sayixa monodactyla n. comb. in the genus Pinnixa, most likely because he had not found the opportunity to examine it. According to Rathbun (1918), Sayixa monodactyla n. comb. had not been seen since the type was reported upon. Moreover, the type in Richmond Museum was, also according to her, probably not extant.
original DiagnoSiS by rathbun (1918). -"Carapace transverse, subpentagonal, hard, very convex, regions scarcely indicated, lower or true antero-lateral margin curving gradually into postero-lateral margin, not forming an angle with it as in Pinnixa. Ambulatory legs similar, third longest but not unusually long, fourth not noticeably reduced. Ischium of outer maxillipeds rudimentary, merus oblique, palpus three-jointed, the last joint articulating near proximal end of preceding joint. Only a single species known." DiagnoSiS of the genuS aS moDifieD by campoS (2006). -"Carapace hard, subheptagonal, highly convex dorsally, anterolateral margins not forming angle with posterolateral margins; MXP3 [= third maxilliped] slightly oblique, covers buccal cavity, ischio-merus subtrapezoidal, propodus extending to end of dactylus, both spoonshaped and larger than carpus. WL1-4 [= walking leg] of similar shape, third pair slightly longer, fourth not noticeably reduced." DiagnoSiS. -(Modified from Rathbun 1918 andCampos 2006 (4)  remarkS Genetic distances and the morphological differences observed among some of the species in this group are similar to those shown among conspecific populations in other pinnotherid genera, for instance Austinixa, Tumidotheres Campos, 1989, or Tunicotheres Campos, 1996. Furthermore, for some species there seems to be striking variation in key characters between juveniles and adults. For example, juveniles of P. littoralis and P. faba appear to be extremely difficult to discriminate (Zmarzly 1992). A more detailed investigation with larger sample sizes and markers appropriate to determine variability between populations of these species is required to clarify phylogenetic relationships within and among them. In addition, knowledge of host associations is required to accompany samples, as these taxa might represent species complexes of separate, but morphologically similar, populations that have adapted to different hosts, which may also be reflected in variations between inshore and offshore samples.
The only specimen of Pinnixa scamit available for molecular analyses and morphological examination was a juvenile (UF 11969), and had therefore been identified provisionally. Genetically it was closely allied to Scleroplax, however, morphologically it showed characters similar to those in Rathbunixa n. gen. It had long slender legs, somewhat compressed, and a sculpted carapace. Despite the results of the molecular analysis, we choose not to transfer Pinnixa scamit to the genus Scleroplax, until specimens definitively identifiable as P. scamit are available for analysis.  (2), ULLZ 4561 (2), ULLZ 5553 (2), ULLZ 6429, ULLZ 7395, ULLZ 7400, ULLZ 10286, ULLZ 14005 (2), ULLZ 14008 (6) remarkS Genetic and morphological differences between specimens of T. chaetopterana (Stimpson, 1860) n. comb. from Venezuela and  (Rathbun, 1918) n. comb., habitus, male paratype, cw 10.1 mm, USNM lot 7639; B, G. faxoni n. comb., left chela external surface, male holotype, cw 11.0 mm, USNM lot 7639; C, G. miamiensis (McDermott, 2014) n. comb., habitus, male, cw 4.7 mm, HBOI uncatalogued specimen from Indian River, Florida; D, G. floridana (Rathtbun, 1918) n. comb., habitus, male holotype, cw 6.7 mm, USNM 6996; E, G. vanderhorsti (Rathbun, 1922) n. comb., habitus, male holotype, cw 6.0 mm, Zoological Museum Amsterdam, now Netherlands Naturalis Biodiversity Center; F, G. vanderhorsti n. comb., gonopodal plate pleonal surface, male holotype, cw 6.0 mm, Amsterdam Museum. Belize and those from the Gulf of Mexico and North Atlantic coasts at minimum suggest population structure within this species. This taxon may represent a species complex, similar to that observed for some of the species of Scleroplax, with different morphotypes at the species and/or population level likely adapted to different habitats and/or hosts. However, most preserved samples available to us at present do not represent sequence quality materials. Additional studies with larger and more broadly representative sample sizes based on markers with resolution at the population level should be undertaken, along with more detailed collection information regarding habitat and hosts. Additional samples of T. chaetopterana n. comb. from Belize should further clarify the identification of that juvenile specimen, once at least a 12S sequence for can be obtained.
proviSional morphological key to american genera of pinnixinae ŠtevČiĆ, 2005 The present key must be regarded as provisional since many couplets require mature males, and these are not known or available for all species of each genus. Thus, it cannot be ruled out that exceptions to some of the applied characters may occur. It also includes one generic level taxon that remains to be named.
The holotypes of Pinnixa brevipollex Rathbun, 1898 (USNM 21593, near La Plata estuary, Argentina) and Pinnixa rapax Bouvier, 1917 (MCZ 10997, Gulf of San Matías, Argentina) require further study and comparison, along with molecular and morphological studies based on contemporary samples representing their putatively separate populations. These species have been suggested to be synonyms, but the holotypes remain to be compared (Fenucci 1975;Bezerra et al. 2006). Some authors suggest there are differences in the male pleon (Righi 1967), but the allegedly junior synonym P. rapax is still considered a valid species (Ng et al. 2008). This group may represent yet another species complex, and we elect to for now continue their treatment as separate taxa.

polyphyly of Pinnixa
The results indicated Pinnixa s.l. to be a highly polyphyletic genus, supporting inferences of previous studies based on molecular evidence as well as adult and larval morphology (Cuesta et al. 2002;Palacios Theil et al. 2009, 2016. The present analyses, being limited to sequence-quality specimens, could represent only 17 of 51 currently recognized extant species that are assigned to Pinnixa s.l., but our analyses did include the type of the genus, P. cylindrica. While 13 additional species in five genera presently within the subfamily Pinnixinae, as well as members of the family Pinnixulalinae, were available for inclusion, only two species belonging to the Indo-Pacific genus Indopinnixa could be studied. None of the seven species that are presently included within the genus Pinnixa and inhabit Indo-Pacific waters, including the Red Sea and the Persian Gulf, could be analyzed. Interoceanic phylogenetic associations among pinnotherid taxa in the subfamily Pinnotherinae have been observed in some cases, for example between species of the American genus Zaops Rathbun, 1900 andEuropean species in Nepinnotheres Manning, 1993, or between the European Pinnotheres and Asian species of Alain Manning, 1998(Palacios Theil et al. 2016. Similar relationships could be found here for Indopinnixa and others might become evident once these species are included in molecular and further morphological studies. In addition, specimens of Alarconia Glassell, 1938 must be analyzed. For Alarconia only two species are known, one from Pacific coasts of Mexico and another from Brazil, and their relationships remain in question. The seven described species placed in Indopinnixa are restricted to Indonesia, Hong Kong, and Japan. Nevertheless, the genetic evidence presented here shows their close relationship to the Pinnixa faxoni complex. This is in accordance with morphological similarities among these species that have been previously suggested by Naruse & Maenosono (2012).
No representatives of the subfamily Pinnixinae included in our molecular analyses were genetically closely allied to P. cylindrica, the type species of the genus. While sharing a carapace wider than long and the third ambulatory leg (P4) longer than the others, none of these subfamilial representa-tives grouped in the same molecular genetic clade with the type. Instead, they were separated at greater genetic distances, typically consistent with differences between genera. Among presently known members of Pinnixinae, including those unavailable for the present molecular phylogenetic analyses, none are known to be more similar morphologically to P. cylindrica than are P. lunzi Glassell, 1937 and P. monodactyla. However, no sequence quality material for P. lunzi was available to us and P. monodactyla did not in our analyses show grouping at the level of genus with P. cylindrica or with any of the species available and presently placed in Pinnixa s.l. Like the other taxa, P. monodactyla has a carapace that is wider than long and third ambulatory legs that are longer than the others. In addition, its third maxilliped is similar to those in P. cylindrica and those of Austinixa, Scleroplax, and the P. sayana, and the P. chaeopterana complexes. However, the morphology of the cheliped, shape and ornamentation of the carapace, and especially the male pleon in P. monodactyla differ from those of the aforementioned species. In P. monodactyla the fixed finger or thumb of the cheliped is more strongly reduced than in any of the other species, and has been replaced by a spiniform angle of the palm. Also, unlike for the other species that share its maxilliped form, P. monodactyla has a male pleonal telson that is wider than the subterminal pleonal segment.

taxonomic implicationS
While full understanding of group relationships must await access to additional sequence-quality specimens for a robust representation of morphological variants, present results indicate that eighteen species of Pinnixa as well as the three species in Laminapinnixa, L. faxoni, L. miamiensis, and L. vanderhorsti, warrant reassignment. Some species can be assigned to Scleroplax and others to Glassella, but three new genera are justified to accommodate those species most closely related to P. chaetopterana, those allied to P. sayana, and a third genus to receive P. monodactyla, as treated in the present paper.
is contribution number 1016 of the UL-Lafayette Laboratory for Crustacean Research, number 1112 for the Smithsonian Marine Station, Ft. Pierce, and number 192 for the Smithsonian CCRE program.