A new Southern Ocean species in the remarkable and rare amphipod family Podosiridae (Crustacea: Amphipoda) questions existing systematic hypotheses


 The amphipod family Podosiridae is unusual in that it combines morphological elements of the disparate families Podoceridae and Eusiridae. Here, we describe a new species in the family from specimens collected from the Southern Ocean in the vicinity of the South Orkney Islands and South Shetland Islands. We present mitochondrial (COI and 16S) and nuclear (18S) nucleic acid sequences for this and a congeneric species and use these to investigate the phylogenetic placement of Podosiridae within the Amphipoda. Our results do not provide evidence for a close relationship between Podosiridae and Podoceridae or Eusiridae, suggesting that the superficial similarity between these families is the result of morphological convergence. Instead, it is likely that Podosiridae are more closely related to families within Amphilochidira, such as Stenothoidae. Definitive placement of Podosiridae in the Amphipoda awaits further specimen collection, additional nucleotide data (including sequences from the Hyperiopsidae and the Vitjazianidae) and a more directed analysis of relationships within this portion of the amphipod phylogeny.


INTRODUCTION
The family Podosiridae Lowry & Myers, 2012 is infrequently sampled and is represented by two described species: Podosirus vaderi Bellan-Santini, 2007, collected in 2002 from a hydrothermal vent community at 1680 m water depth in the Azores Triple Junction zone (Bellan-Santini, 2007), and Acutocoxae weddellensis Rauschert, 2017, collected at 694 m water depth in the Weddell Sea, south of Vestkapp (Rauschert, 2017 Between February and March 2016, four specimens resembling the Podosiridae were collected in the vicinity of the South Orkney Islands during the British Antarctic Survey research cruise JR15005 'SO-AntEco' (https://www.bas.ac.uk/project/so-anteco/; last accessed 8 November 19).
Here, we detail the morphology of these Southern Ocean specimens and describe them as a new species. We also present mitochondrial (COI and 16S) and nuclear (18S) nucleic acid sequences for these specimens and use this information to investigate the phylogenetic placement of the Podosiridae within the Amphipoda.

ColleCtion methods and loCations
The primary material for this study was collected during the British Antarctic Survey expedition 'SO-AntEco' (https://www.bas.ac.uk/project/so-anteco/; last accessed 8 November 19) between February and March 2016 on board the RRS James Clark Ross (cruise JR15005) (Griffiths et al., 2016). Podosirids were sampled at three stations at bathyal depths on the South Orkney shelf and slope (775-1139 m; Fig.  1; Table 1). Specimens were taken with a 2-m-wide Agassiz trawl (AGT), with a mesh size of 1 cm.
Collected specimens were fixed in 96% ethanol within minutes of arrival on deck (Griffiths et al., 2016).
An additional specimen collected off Clarence Island (South Shetland Islands) in 1937 ( Fig. 1; Table 1) has been found in the Discovery Collections held at the National Oceanography Centre, Southampton (http:// noc.ac.uk/facilities/discovery-collections; last accessed 8 November 19).

taxonomiC methods
Initial observations, dissections and pencil illustrations were made using an Olympus SZX10 stereoscopic microscope with an Olympus SZX-DA camera lucida attachment, and an Olympus BX51 compound microscope with a U-DA camera lucida attachment. Pencil drawings were scanned and inked digitally using Adobe Illustrator and a WACOM digitizer tablet (Coleman, 2003(Coleman, , 2009. Type material is deposited in the Natural History Museum, London, UK (NHMUK).
Setal and mouthpart classifications follow Watling (1989) and Lowry & Stoddart (1992, 1993, 1995. Measurements of inner and outer plates of maxilla 2 and the relative lengths and proportions of pereopods and gnathopods follow Horton & Thurston (2014). Use of the terms 'acute' and 'transverse' relative to the palm of gnathopod 2 follows Poore & Lowry (1997), where 'acute' describes the condition in which the included angle between the longitudinal axis of the propodus and the palm is < 90° and 'transverse' the condition in which this angle is ~90°. The species examined here exhibits modification of the propodus of pereopods 3-7. The term 'prehensile' is used where the propodus is expanded to a greater or lesser degree, forming a palm against which the dactylus can close. The palm thus formed is lined with setae noticeably stouter than those on the corresponding margin of the carpus.
The  (Table 1). All three left pleopods and pereopod 6 (excluding coxa) were dissected on ice using a sterile scalpel and forceps. Extractions were undertaken using a Qiagen DNeasy Blood and Tissue Kit following the standard 'Purification of Total DNA from Animal Tissues (Spin-Column Protocol)' (2 × 200 μL elutions). DNA concentrations after extraction were estimated using an Invitrogen Qubit 4 fluorometer: 'Event 68, Vial 3263' = 7.60 ng/μL; and 'Event 70, Vial 1499' = 9.54 ng/μL. DNA was also extracted, using the same methodology, from two specimens of A. weddellensis ('PS67' and 'PS81') kindly provided by Dr Claude De Broyer, Royal Belgian Institute of Natural Sciences.
Two ribosomal gene regions, 16S (~330 bp) and 18S (~2150 bp) and one protein-coding gene, cytochrome c oxidase subunit I (COI; ~650 bp) were amplified by polymerase chain reaction (PCR) and sequenced using one or more sets of primers (Table 2). Reactions were performed in 30 µL volumes, containing 2 µL of each primer (forward and reverse at 4 pmol/µL), 15 µL of Qiagen HotStarTaq Plus Master Mix, 5 µL of DNA template and 6 µL of double-distilled water. The PCR cycling protocols for all gene fragments were the same, except for the annealing temperature: an initial denaturation at 95 °C for 5 min, followed by 40 cycles of 94 °C for 45 s, the annealing step for 90 s, 72 °C for 1 min, and a final extension of 5 min at 72 °C. For COI and 16S, the annealing temperature was set at 43 °C. For 18S, the annealing step with the primers 18A1 mod and 1800 mod (Raupach et al., 2009) was at 56 °C, with additional primers for sequencing. Individuals PS67 and PS81 each generated two smaller 18S fragments (~1200 and ~880 bp long) rather than the single ~2150 bp fragment. The PCRs were performed on a Bio-Rad C1000 Thermal Cycler. PCR clean-up, sequencing reactions and clean-up, in addition to final Sanger sequencing, were undertaken by the Zoology Department sequencing facility at the University of Oxford. Forward and reverse sequences were assembled and cleaned using GENEIOUS v.6.1.8. All sequences are deposited in GenBank (see Supporting Information, Appendix S1 for accession numbers).
For the Bayesian analysis, MrBayes on XSEDE (v.3.2.6) was selected, and all options were specified using a MrBayes data block. Nebalia sp. was selected as the outgroup taxon, and trees were constrained to retain order-level monophyly. The optimal model of evolution and partitioning scheme were defined in accordance with the PartitionFinder results, with parameters unlinked and rates free to vary such that each partition could evolve under independent scenarios. Two runs were specified, with trees sampled every 2000 generations, using eight Markov chains with a heating value of 0.12. The two runs converged (standard deviation of split frequencies < 0.01) after 20 530 000 generations. A burn-in of 10% was selected, corresponding to stabilization of sample probabilities as determined using TRACER v.1.6 (Rambaut et al., 2013), and a 50% majority rule consensus tree was constructed from the remaining trees (Fig. 5).

Diagnostic description
Pereon dorsoventrally flattened; without setae. Head as long as or longer than deep; eyes present or absent. Antenna 1 peduncle with few or no setae; accessory flagellum absent; primary flagellum 5 or more articulate; callynophore absent; calceoli absent. Antenna 2 with sparse slender setae; flagellum longer or shorter than peduncle; flagellum 5 or more articulate. Mouthparts well developed, forming a subquadrate bundle. Mandible incisor dentate, straight; lacinia mobilis broad, apically dentate; molar present, medium sized, strongly triturating or forming an apically toothed conical flap; palp 3-articulate or reduced to  We are unsure whether Acutocoxae fits within this family, but for the sake of stability we have included it here until further data are available. In order to place the new species in the family Podosiridae, we have made significant amendations to the family diagnosis provided by Lowry & Myers (2012) (italicized statements in the Diagnostic description section above). A re-measurement of pereopods 5-7, based on Bellan-Santini's habitus illustrations of P. vaderi, suggests that these appendages are in fact subequal, contrary to the family diagnosis provided by Lowry & Myers (2012) (the relative lengths of pereopods 5-7, including coxae, in P. vaderi are 1:1.00:0.97; in Acutocoxae ogilvieae 1:0.97:0.94). Lowry & Myers (2012) indicate that coxae 4-5 are without lobes, but P. vaderi does possess a posterior lobe on pereopod 5. The two species show similarities in the gnathopods and pereopods, but this could be a result of convergent evolution. The structure of the head in the two genera differs markedly. The body of Acutocoxae is armed with lateral extensions of the pereonites, coxae and the head lobes, which are not present in Podosirus. The coxae of Acutocoxae are discontiguous, but there is some degree of overlap in Podosirus. The mouthparts of Podosirus and our new species show some similarities, but the mandibular palp in Acutocoxae is reduced to a single article, and the maxilliped has lost the outer lobes completely, has the inner lobes reduced and fused, and the palp is developed into an enormous raptorial organ. Our interpretation of the habitus illustration of Bellan-Santini (2007) differs from that of Lowry & Myers (2012) in that we believe that the pereon is dorsoventrally flattened rather than subcylindrical and the urosome is likely to be laterally compressed or subcylindrical. This can be confirmed only by examination of the type material, which we have not seen.

Remarks
Specimens of the new species exhibit the slender, dorsoventrally flattened body, moderate rostrum, lack of accessory flagellum, extended, raptorial maxilliped, acuminate coxae, elongate urosomite 1 and entire telson that are seen in the genus Podosirus. However, the genus Acutocoxae differs from Podosirus in the strongly produced lateral cephalic lobes, presence of eyes, ratio of antenna 1 articles, elongate peduncles of uropods 1-3 and uniramous uropod 3.
A revised diagnosis of the genus Acutocoxae Rauschert, 2017 is provided. Rauschert (2017) indicated that A. weddellensis possesses a rudimentary accessory flagellum. Our examination of the type material reveals a group of small setae, but no accessory flagellum. No accessory flagellum was found in our new species. Rauschert (2017) provided an illustration of the mandibular molar, which he described as having a 'strong elevation for grinding'. We have been unable to examine this, because both mandibles are missing from the type material. This is different from the toothed flap molar found in our new species. Eyes are prominent, hemispherical, and similar to those of some species in the family Melphidippidae Stebbing, 1899.

Etymology
This species is named as a noun in a genitive case after the maiden name (Ogilvie) of Imogen Catherine Rachel Ashford, wife to author O.S.A.

Sexual dimorphism
None.

Remarks
The new species appears to be similar to the only other species in the genus, A. weddellensis. We have studied all available type material of that species and found inconsistencies between the material and the illustrations and description provided by Rauschert (2017). Acutocoxae weddellensis is in the process of redescription, using new material (C. De Broyer, personal communication). The two species can be separated easily by the form of the propodus on pereopods 3-7, with the posterior margins of pereopods 3 and 4 and the anterior margins of pereopods 5-7 convex and expanded medially in A. ogilvieae and narrow and strongly concave in A. weddellensis.

Distribution
Southern Ocean (South Orkney Islands, South Shetland Islands).

phyloGeny
The conducted ML (RAxML and IQ-TREE) and Bayesian (MrBayes) analyses suggest, with high support, that the two species of Acutocoxae form a monophyletic group, but appear to be distinct species . This is in agreement with the morphological observations and supported by COI and 16S raw pairwise distances between the species of 10.7 and 3.2%, respectively (Tempestini et al., 2018).
Of the crustacean taxa included in our phylogenetic analyses, the genus Acutocoxae appears to be most closely related to the stenothoid amphipod species Antatelson walkeri (Chilton, 1912) and Stenothoe brevicornis Sars, 1882. This is inferred, with high support, from both ML and Bayesian analyses . In turn, in all phylogenetic trees presented, there is high support for a sister-group relationship between this podosirid/stenothoid clade and species of the amphipod family Stegocephalidae Dana, 1852 (Figs 3-5).
Deeper phylogenetic relationships differ among the three trees presented, and node support values are generally low. However, there is little support for a close relationship between Acutocoxae and any of the podocerid or eusirid species analysed (Figs 3-5).

DISCUSSION
Our ability to place Acutocoxae in phylogenetic context within the Amphipoda is somewhat restricted by the low support values for moderately deep nodes in the constructed phylogenies, which is reflected by differences in the phylogenetic results provided by the alternative analysis methods (Figs 3-5). Our phylogenies, coupled with the analysis of Bousfield (1983) and the supertree of Peracarida presented by Ashford et al. (2018), suggest that difficulties in resolving deeper nodes within the phylogeny of the Amphipoda might stem from rapid diversifications of amphipod lineages, possibly during the early Carboniferous period and later during the Mesozoic. Deeper nodes may be more completely resolved in future phylogenetic studies by incorporating both morphological and molecular data (Wortley & Scotland, 2006). However, even considering the problems in resolving deeper nodes here, there is little support for a close relationship between Acutocoxae and any of the eusirid or podocerid taxa included in our analyses, suggesting that the apparent similarity of the Podosiridae to the Podoceridae and Eusiridae (Bellan-Santini, 2007) reflects morphological convergence. Lowry & Myers (2012) also rejected a close relationship between the Podosiridae and Eusiridae and listed several characters that exclude Podosirus from Eusiridae, Calliopiidae and Pontogeneiidae. Our revision of the diagnosis of the Podosiridae does not significantly contradict their arguments. Instead, Lowry & Myers (2012) argued that the Podosiridae are most closely related to the family Amathillopsidae. Core characters shared between podosirids and amathillopsids include elongate raptorial maxillipedal palps, subchelate gnathopods, linear bases to pereopods 5-7, pereopods with elongate meri and a relatively elongate urosomite 1. Diagnostic characters shared between podosirids and amathillopsids include coxa 1 smaller than coxa 2, pereopods 5-7 (sub)equal in length, pleonites dorsally carinate and telson entire. However, podosirids differ significantly from amathillopsids by lacking an accessory flagellum, by the carpus of gnathopod 1 being longer than the propodus and by gnathopod 1 being smaller than gnathopod 2.
Congruence between ML and Bayesian trees, and high node support values, provide confidence that, of the peracarid taxa included in the phylogenetic analyses conducted here, Acutocoxae is most closely related to the amphipod family Stenothoidae Boeck, 1871. However, whilst Acutocoxae exhibits characteristics of some stenothoids, such as lack of an accessory flagellum, a reduced mandibular molar and palp, an absence of maxilliped outer plates, an enlarged gnathopod 2, coxa 4 without posterodorsal excavation, uniramous uropod 3 and entire telson, it is unlikely that Acutocoxae resides in Stenothoidae, because it lacks key diagnostic characteristics of the family; in particular, a small coxa 1 that is partly covered by the following coxae, and an enlarged, shield-like coxa 4 (Barnard & Karaman, 1991). It is possible that the Podosiridae is closely related to Stenothoidae or that the genus Acutocoxae resides in a family that is closely related to Stenothoidae but not included in our analyses. However, to test this hypothesis further would require a more directed and in-depth analysis of relationships within this part of the amphipod phylogeny, which is beyond the scope of the present study.
Based on a cladistic analysis of amphipod morphology, Lowry & Myers (2017) concluded that Podosiridae form a distinct clade in the infraorder Hyperiopsida, with the families Hyperiopsidae and Vitjazianidae as closest relatives. Although we cannot explicitly rule out this hypothesis, owing to incomplete taxon sampling (nucleotide data are not available in GenBank for members of either of these families at present), the seemingly close relationship between Acutocoxae and Stenothoidae suggests an alternative affinity for Podosiridae within the parvorder Amphilochidira. This would be in agreement with the conclusion by Lowry & Myers (2012) that Podosiridae might have a relatively close relationship to Amathillopsidae. The diagnoses of Amphilochidira and Hyperiopsida offered by Lowry & Myers (2017) are relatively similar (Table 3). However, the diagnosis of Acutocoxae we provide appears to have more similarities with Amphilochidira than it does with Hyperiopsida (Table 3). However, we concede that the definitive placement of Podosiridae within Amphipoda will require further specimen collection, additional nucleotide data (including sequences from Hyperiopsidae and Vitjazianidae) and a more directed analysis of relationships within this portion of the amphipod phylogeny.
Finally, the apparent close relationship between Stenothoidae and Stegocephalidae and the apparent distance of Stegocephalidae from the remaining lysianassidiran taxa in our phylogenies are noteworthy. This counters the current phylogenetic placement of Stegocephalidae within Lysianassidira (Lowry & Myers, 2017

SUPPORTING INFORMATION
Additional Supporting Information may be found in the online version of this article at the publisher's web-site: Appendix S1. Taxonomic identity and GenBank accession number for all genetic sequences analysed. Figure S1. Rooted maximum likelihood phylogeny, placing Acutocoxae ogilvieae in the phylogenetic context of 240 malacostracan taxa, estimated using RAxML v.8.2.12. Concatenated dataset of 18S SSU rDNA, 16S rDNA, cytochrome c oxidase I (COI) and histone H3 sequence data (2599 bp