Cryptic species in a well-known habitat: applying taxonomics to the amphipod genus Epimeria (Crustacea, Peracarida)

Taxonomy plays a central role in biological sciences. It provides a communication system for scientists as it aims to enable correct identification of the studied organisms. As a consequence, species descriptions should seek to include as much available information as possible at species level to follow an integrative concept of ‘taxonomics’. Here, we describe the cryptic species Epimeria frankei sp. nov. from the North Sea, and also redescribe its sister species, Epimeria cornigera. The morphological information obtained is substantiated by DNA barcodes and complete nuclear 18S rRNA gene sequences. In addition, we provide, for the first time, full mitochondrial genome data as part of a metazoan species description for a holotype, as well as the neotype. This study represents the first successful implementation of the recently proposed concept of taxonomics, using data from high-throughput technologies for integrative taxonomic studies, allowing the highest level of confidence for both biodiversity and ecological research.

Small ribosomal subunit (18S rRNA gene). Eight sequences of the complete 18S rRNA gene were successfully amplified and sequenced for the two species (Epimeria cornigera: n = 2, Epimeria frankei: n = 6). We found no insertions or deletions, resulting in a total length of 2,380 bp for both species. As a result of concerted evolution of rRNA gene clusters, we found no intragenomic or intraspecific variations. For all sequences of both species, average base frequencies were A = 0.22, C = 0.23, G = 0.28, and T = 0.27. In total, we observed 34 substitutions and a patristic distance of 2% between both species ( Fig. 3; Table S1). Most substitutions (82%) were located in variable regions, with V1 = 1 substitution, V2 = 5 substitutions, V4 = 14, and V7 = 8 (Table S1). Mitochondrial genome data: mitochondrial genome reconstruction. The lack of mitogenomic data from closely related species prevented a standard approach of mapping reads to a reference. We therefore implemented independent iterative mapping approaches for both Epimeria species. We selected three amphipod bait reference sequences: Gondogeneia antarctica (Chevreux, 1906) (Sars, 1900) (Genbank: NC_013819.1). Regardless of the employed reference, default k-mer values of 31 produced no mappable reads. We therefore carried out the reconstruction using k-mer values of 19 and 25. Both of these values allowed reads to map to our chosen bait reference sequences. However, due to the high level of rearrangements in these distantly related amphipod mitochondria, the correct orientation and therefore continuous sequence of  Maximum Likelihood phylogenetic tree of the partial COI sequence data set of 19 Epimeria species and Gammarus locusta as outgroup. Model of choice: HKY with a log likelihood score of −6124.0843, a discrete gamma distribution (+G, parameter = 1.0064) and a proportion of invariant sites (+I, 46% of sites). Numbers next to internal branches are non-parametric bootstrap values (in %). Only values above 90% are shown. Initial tree(s) for the heuristic search were obtained automatically by applying BioNJ algorithms to a matrix of pairwise distances estimated using the Maximum Composite Likelihood (MCL) approach, and then by selecting the topology with superior log likelihood value. The tree is drawn to scale with branch lengths measured in substitutions per site. All positions with less than 95% site coverage were eliminated. That is, fewer than 5% alignment gaps, missing data, and ambiguous bases were allowed at any position. There were a total of 542 positions in the final dataset. the Epimeria mitogenomes could not be recovered using a k-mer value of 19. This k-mer value resulted in multiple mapping initiation points along the reference, leading to a complete, yet fragmented sequence. In contrast, a k-mer value of 25 produced only a single initial mapping point on all three chosen bait reference sequences, allowing a single continuous sequence to be reconstructed and thereby overcoming the above mentioned problems caused by rearrangements. Regardless of the bait reference sequence used, the final consensus was identical, providing additional reliability to the sequenced genomes. Automated annotations showed the presence of all expected tRNAs, rRNAs and coding genes in an invertebrate mitochondrial genome. Although iterative mapping and consensus sequence reconstruction were performed independently on each species' genome, annotations showed very similar gene locations and lengths, providing additional support to our approach and final consensus sequences.
The complete mitochondrial genomes of Epimeria cornigera (14,391 bp) and Epimeria frankei (14,829 bp) are typical metazoan mitogenomes. Both consist of a single circular DNA molecule carrying 13 protein-coding genes, two subunits of the mitochondrial ribosomal RNA, 22 tRNAs, one for each amino acid except leucine and serine, which have two copies each, and a putative control region (Figs 4 and 5; Table 2). The heavy strands (H-strand) encode 25 genes (protein coding genes: 9, tRNAs: 16), whereas the remaining 12 genes (protein coding genes: 4, tRNAs: 6, rRNAs: 2) are on the light strands (L-strand). The gene organization of both genomes is similar to the supposed pancrustacean basic pattern 49 except for the location of various tRNAs and a translocation of a fragment containing two protein-coding genes and one tRNA (NADH6, Cyt b, tRNA Ser2 ). There are no gene  rearrangements between the two Epimeria species investigated in this study. Twenty three gene overlaps and 10 intergenic spacers were found for both mitogenomes. For Epimeria cornigera, the total length of overlaps is also 110 bp (ranging from 1 to 26 bp), whereas intergenetic spacers have values from 1 to 44 bp, resulting in a total length of 92 bp, respectively. For Epimeria frankei, the total lengths of overlaps are 110 bp, ranging from 1 to 26 bp, and 89 bp (1 to 44 bp) for intergenetic spacers. We found similar nucleotide compositions for both genomes, with frequencies of A = 0.36, C = 0.20, G = 0.10, and T = 0.34 for Epimeria cornigera and A = 0.35, C = 0.22, G = 0.10, and T = 0.33 for Epimeria frankei.
Mitochondrial genome data: ribosomal and transfer RNA genes. A total of 22 tRNA genes with lengths from 51 bp (tRNA Ser2 ) to 64 bp (tRNA Gly ) were found within both mitochondrial genomes (Figs 4 and 5; Table 2). The data revealed differences in lengths (+/−1 bp) for four tRNAs: tRNA Glu , tRNA Leu1 , tRNA Cys , and tRNA Thr . Sixteen tRNAs are encoded on the H-stand whereas six tRNAs are located on the L-strand. All tRNAs are capable of folding into a typically clover-leaf-like secondary structure except tRNA Ser1 and tRNA Ser2 . As previously reported for other crustacean taxa, these two tRNAs cannot form a secondary structure due to their lack of the dihydrouracil arms 50,51 . Both rRNA genes were found on the L-strand. They are located between the tRNA Leu1 and the putative control region and are separated by tRNA Val . Gene lengths of the 12S rRNA genes are 639 bp for Epimeria cornigera and 640 bp for Epimeria frankei, whereas the lengths of 16S rRNA genes are 1,049 bp (Epimeria cornigera) and 1,046 bp (Epimeria frankei). The nucleotide divergence (patristic distance) was 0.07% between the 12S rRNA genes and 0.08% between the 16S rRNA genes ( Table 2). 294 bp, Epimeria frankei: 737 bp) were identified between the 12S rRNA and tRNA Cys genes, which are considered to represent the putative control regions (Table 2). It should be noted that the read depth in this region was relatively high in Epimeria cornigera compared to the rest of the genome. This could indicate the presence of a repetitive element that could not be be properly assembled, All other 10 non-coding regions are smaller, ranging from 1 to 44 bp for both genomes. The A+T contents of the control regions are higher than those of other mitogenome regions, with 0.78 for Epimeria cornigera and 0.81 for Epimeria frankei. Furthermore, both control regions consist of several repetitive motifs in tandem that are different among both genomes.
Morphological distinction. Epimeria frankei can be distinguished from E. cornigera by the shape of coxal plate 5. In E. frankei, the anteroventral margin of coxa 5 describes a nearly straight line or is only slightly concave, whereas this margin is strongly concave in E. cornigera (less pronounced in males). Coxa 5 is strongly extended posteroventrally in both species, coming along with an only slightly concave posterior margin in E. cornigera (distinctly concave in E. frankei). Additionally, urosomite 3 can be used as a diagnostic character. In E. frankei, urosomite 3 is pronounced in a distinctly upwards pointed projection (Fig. 8D)  is characterised by an inconspicuous blunt hump (Fig. 14D) or is not pronounced at all. The most conspicuous difference between the two species, however, can be observed on live or freshly caught individuals (observed on approx. 100 indiviuals of each species): E. frankei features an even, brownish or grey colouration with hints of purple or even pink (Fig. 6). In contrast, E. cornigera is characterised by an eye-catching red-white colouration (Fig. 6). Unfortunately, the colouration of individuals vanishes quickly when being exposed to ethanol, resulting in yellowish or pale white specimens in both species.

Discussion
The current study fully implements the concept of integrative taxonomy, proposed as a framework to bring together conceptual and methodological developments, combining as many different data types as possible to characterise species in a multifaceted and detailed way 1,52,53 . Here, two sibling amphipod species are described using comprehensive morphological, genetic, and mitogenomic data. This is the first use of genomic data (here: complete mitochondrial genomes) as part of a species description and a redescription within the Metazoa, setting a new desirable standard for future species descriptions 35 . We are convinced that modern species descriptions should seek for this integrative addition of comprehensive complementary information, ensuring the highest level of confidence for both biodiversity and ecological research.
Biology and geographic distributions. Species  In fact, the current findings and the geographic information of the re-examined museum material suggest that E. cornigera s. str. may be restricted to the northern East Atlantic, whereas E. frankei seemingly exhibits a Mediterranean-Lusitanean distribution (Fig. 7). However, as also evidenced by the same type locality of the two species, both occur sympatrically on sandy bottoms of the North Sea and near the northern British coasts where water temperatures are probably suitable for both species.  58 . Consequently, there is continuous progress of using DNA barcodes as part of species descriptions in the Crustacea 24 . The analysis of our obtained DNA barcode data revealed distinct clusters for both species with a minimum interspecific distance of 8.69% (K2P) ( Table 1). This value is similar to interspecific distances known from other closely related Epimeria species, with minimum distances of 8.5% 59 . Furthermore, we found two lineages and two BINs (AAU2061, ACS7228) for Epimeria frankei (Fig. 1). However, as nuclear 18S rRNA gene sequences from specimens of both lineages were identical and no corresponding morphological differences were observed, these have to be considered as one species unless further data becomes available (Fig. 3). Therefore, the observed variability may be the result of phylogeographic effects in E. frankei [60][61][62] .
Phylogenetic analysis of the CO1 sequences. About 66% of the currently known species of Epimeria are found around Antarctica as part of a monophyletic lineage 59,63-65 , representing the best known example of a species flock within Antarctic amphipods 66 . This hypothesis is supported by our phylogenetic results: all studied species from Antarctica form a monophyletic clade with high bootstrap support (97%) (Fig. 2). Our analysis indicates a potential relationship of Epimeria cornigera and Epimeria frankei with two non-Antarctic species: Epimeria horsti and Epimeria bruuni. Both species are only known from waters around New Zealand so far 67 . However, this relationship is supported only by a low bootstrap value (40%). A more comprehensive analysis based on i) further, particularly non-Antarctic species, ii) more molecular data (both mitochondrial and nuclear) and iii) morphological characteristics is needed to reconstruct the phylogeny of this impressive amphipod genus in detail.

Mitochondrial genomes.
Because modern high-throughput sequencing technologies allow for much faster and easier analyses of genomes, the analysis of mitogenomes will likely increase dramatically in the near future 35 . Compared to other crustacean groups, the number of available complete mitochondrial genomes for amphipods is quite high (n = 29, 2017-06-21) and these display various patterns of gene arrangements 68 . Mitogenomes of both Epimeria cornigera and Epimeria frankei represent typical circular metazoan mitogenomes with an identical gene arrangement (Table 2) and a patristic distance of 0.12% between both genomes. Patristic distances between all protein-coding genes range from 0.12% (COI, COII) to 0.20% (ATP6), and in the case of rRNA genes from 0.07 (12 rRNA) to 0.08 (16S rRNA). With a length of 737 bp the putative control region of Epimeria frankei is significantly longer than in Epimeria cornigera (294 bp). This region is the largest non-coding part of the mitochondrial DNA in general and includes the major regulatory elements for its replication and expression 69 . Furthermore, it is the most polymorphic part of the mitochondrial genome with hypervariable sites 70 . Nevertheless, we found two conserved parts shared between both genomes, covering 80 and 190 bp, respectively. Our results support the concept that the control region is divided into three polymorphic domains, being separated by two stretches with no interspecific variability 71 . However, more data have to be evaluated to analyse this phenomenon in more detail.

Material and Methods
Sample collection and DNA extraction. In total, 75 specimens were collected with beam trawls in the North Sea between 2012 and 2014 as part of the German Small-scale Bottom Trawl Survey (GSBTS 72 ). The amphipods were preserved in ethanol (96%) directly after collection. All specimens were morphologically identified by JB and HN. For each specimen, total genomic DNA was extracted from one to three dissected pereopods using the QIAmp © Tissue Kit (Qiagen GmbH, Hilden, Germany) following the supplier's extraction protocol. Specimens were deposited in the collections of natural history museums (see below), whereas DNA extracts were listed and stored in the North Sea Fauna collection of the German Centre for Marine Biodiversity Research (DZMB) in Wilhelmshaven, Germany. Amplification and sequencing: DNA barcodes and complete 18S rRNA gene sequences. Polymerase chain reaction (PCR) was used to amplify the DNA barcode fragment using the primer pair LCO1490 and HCO2198 19 or the newly designed degenerated amphipod-specific primer pair CO1-LCO-AMP1 (5′-THT CDA CHA ACC AYA AAG AYA TYG G-3′) and CO1-HCO-AMP2 (5′-ATD ACT TCW GGR TGV CCR AAR AAY C-3′). For the latter, we added derived M13 forward and reverse tails to both primers to provide known primer-binding sites for sequencing 73 . Therefore, the complete primer sequences were 5′-TGT AAA ACG ACG GCC AGT THT CDA CHA ACC AYA AAG AYA TYG G -3′ for CO1-LCO-AMP1 and 5′-CAG GAA ACA GCT ATG AC ATD ACT TCW GGR TGV CCR AAR AAY C-3′ for CO1-HCO-AMP2 (with underlined M13 tail sequences). Amplification reactions were carried out on a Mastercycler pro S system (Eppendorf, Hamburg, Germany), using illustra TM puReTaq Ready-To-Go PCR Beads (GE Healthcare, Buckinghamshire, UK) in a total volume of 20 µl, containing 17.5 µl sterile molecular grade H 2 O, 2 µl DNA template with an DNA amount between 2 and 150 ng/µl and 0.25 µl of each primer (20 pmol/µl) for amplification. For this gene fragment (CO1), PCR thermal conditions included an initial denaturation at 94 °C (5 min), followed by 38 cycles at 94 °C (denaturation, 45 s), 48 °C (annealing, 45 s), 72 °C (extension, 80 s), and a final extension step at 72 °C (7 min).
Furthermore, the complete nuclear 18S rRNA gene was amplified for selected individuals of both species (Epimeria cornigera: n = 2, Epimeria frankei: n = 6). Detailed information about used primer pairs, amplification reactions and temperature profiles are provided in previous studies 74,75 . For verification, negative and positive controls were included with each round of reactions. Three μl of the amplified products were inspected for size conformity by electrophoresis in a 1.5% agarose gel with GelRed TM using commercial DNA size standards. The remaining PCR product was purified with the QIAquick © PCR Purification Kit (Qiagen GmbH, Hilden, Germany). Purified PCR products were cycle sequenced and sequenced in both directions at a contract sequencing facility (GATC, Konstanz, Germany) using the M13 sequence tails for the DNA barcodes or the same primers as employed in the PCR in the case of the 18S rRNA gene. For this marker, some additional internal primers were used for sequencing: Forward: AF700F_mod, 1000 F, 1250FN_mod; Reverse: 700 R, 1155 R, 1500R_mod 75,76 .
Complementary sequences were assembled with the Geneious version 9.0.4 program package 77 . BLAST searches were performed to confirm the identity of all new sequences 78,79 . Furthermore, Geneious was used to translate all COI sequences to amino acid sequences to check for presence of nuclear mitochondrial pseudogenes (numts).
Mitochondrial genome: library building, sequencing and raw read processing. DNA extracts were built into double-barcoded Illumina sequencing libraries using a published protocol set 82 , with some minor modifications 83 . The optimal number of cycles for amplification was established using qPCR to avoid saturation. Sequencing was performed on the Illumina Nextseq. 500 at the University of Potsdam, Germany, using 150 base pair (bp) paired-end (PE) reads. 25,368,699 PE reads were sequenced for the holotype of Epimeria frankei and 20,329,986 for the neotype of Epimeria cornigera. Illumina adapter sequences were trimmed from read ends, and all reads less than 30 bp were discarded using Cutadapt 1.4 84 . PCR duplicate sequences were removed using Fastuniq 85 . Forward and reverse sequences were interleaved into a single file using a perl script available in the MITObim package 86 .
Mitochondrial genome: sequence reconstruction. We carried out independent mitochondrial reconstructions for both species using MiTObim v1.8 86 , an iterative mapping wrapper script for the Mira v4.0.2 assembler 87 . We implemented a direct reconstruction without prior mapping assembly using default parameters with alterations to mismatch values, k-mers and reference bait sequences. The utilised bait sequences consisted of: Gondogeneia antarctica (JN827386.1), Metacrangonyx goulmimensis (HE860501.1), and Onisimus nanseni (NC_013819.1). The interleaved fastq files were treated as single end data. Reconstructions were conducted using k-mer values of 19, 25 and 31. We implemented a reconstruction protocol based on that found in Westbury et al. 88 by beginning with the most strict mismatch value (0%) and gradually increasing this value until the mitochondrial genome was complete. Mismatch values of 0-8% were implemented for Onisimus nanseni and Metacrangonyx goulmimensis whereas mismatch values of 0-6% were implemented for Gondogeneia antarctica. This resulted in 25 separate iterative mapping alignments.
Mira output maf files were converted to sam files and visualised using Geneious. For each independent iterative mapping alignment, a 20x minimum coverage was implemented for consensus building along with a strict 95% threshold for base calling. This resulted in 25 individual consensus sequences, nine constructed using Onisimus nanseni as reference, nine using Metacrangonyx goulmimensis as reference and seven from using Gondogeneia antarctica as reference. From this data, we constructed three reference-specific consensuses by aligning all consensus sequences produced via mapping to the same reference bait sequence, i.e. consensuses produced from different mismatch values but the same reference sequence, using Mafft 89 and a 50% strict consensus base call. These three reference-specific consensus sequences were then aligned in Mafft and a final consensus was built using a 50% strict consensus base call. We performed automatic annotations with MITOS 90 to check for the presence of genes and tRNAs in our final consensus sequence. The tRNAs were also annotated with ARWEN v1.2 91  Intra-and interspecific nucleotide distances of the analysed amphipods were based on the Kimura 2-parameter model (K2P) 92 , using the analytical tools on the BOLD workbench (align sequences: BOLD aligner; ambiguous base/gap handling: pairwise deletion). We also used BOLD to calculate base frequencies of all barcodes. All barcode sequences were assigned to the Barcode Index Number (BIN) system which is implemented in BOLD 81 . All sequences were aligned using MUSCLE version 3.6 93 with default settings. In a further step, a neighbor-joining cluster analysis (NJ) 94 was performed to reconstruct a phylogenetic tree based on K2P distances for all DNA barcode sequences using MEGA7.0.18 95 . Non-parametric bootstrap support values were obtained by resampling and analysing 1,000 replicates 96 .
In order to explore the phylogenetic position of Epimeria cornigera and Epimeria frankei, we reconstructed the phylogeny of the genus Epimeria in a maximum-likelihood framework using MEGA7.0.18. For the analysis, COI barcodes of the neotype of Epimeria cornigera and the holotype of Epimeria frankei were used as well as all available COI sequences of the DNA barcode fragment from NCBI GenBank with a length of at least 500 base pairs (n = 67, date of download: 01-02-2017), representing 16 of the 84 currently recognized species in the genus Epimeria (29%). A table of all included species and corresponding accession numbers is included in the supplementary information (Table S2). All sequences were aligned using MUSCLE version 3.6 93 with default settings and with the six sequences of Gammarus locusta as outgroup. The most appropriate model of sequence evolution was determined beforehand using the Decision Theory performance-based selection (DT) as implemented in jModeltest 2.1.1 97 . The Hasegawa-Kishino-Yano model (HKY85) 98 was shown to be the optimal nucleotide substitution model with the following parameters: nucleotide frequencies A: 0.32, C: 0.24, G: 0.15, T: 0.29; transition/ transversion ratio = 3.4401; gamma distribution shape = 0.885; and proportion of invariable sites = 0.439. Node support for individual clades was estimated using 1,000 non-parametric bootstrap replicates 96 . Small ribosomal subunit (18S rRNA). All eight 18S rDNA sequences of the two Epimeria species and one outgroup sequence of Gammarus locusta (AF419222) were aligned with MUSCLE version 3.6 using default settings. The software package MEGA7.0.18 was used to calculate nucleotide frequencies, interspecific distances based on patristic distances and to perform a NJ cluster analysis for a graphical representation of nucleotide divergence. Secondary structure elements were identified using the CRW site 99  Morphological analyses. Before examination, specimens were transferred to 70% ethanol and appendages were temporarily mounted in glycerin. Whole animals and appendages were then studied and drawn under a Leica M205c microscope equipped with a drawing tube. All material was later conserved in ethanol (95%). Pencil drawings were scanned and inked with the software Adobe Illustrator (CS6) on Wacom drawing tablets (Intuos 5, A4 and A3), following the methods described by Coleman 105,106 . Referring to the considerations of d'Udekem d' Acoz 107 and Krapp-Schickel 108 , the following terminology was applied in the descriptions: seta = slender, flexible articulated structure; spine = robust, inflexible articulated structure (synonymous to 'robust seta'); tooth = non-articulated, pointed ectodermal structure. Although officially listed in the collections of the Natural History Museum of Denmark, Copenhagen, Fabricius' holotype is missing (pers. comm. Jørgen Olesen) and was thus presumed as lost in the current study. Furthermore, there is no exact information on the type locality, which can only be assumed to be located in the northern North Sea. Therefore, types were chosen considering the geographical area where both species were commonly found sympatrically. This was further validated by the descriptions and drawings of Sars 55 , which seem to refer to E. cornigera s. str. All type material was deposited in the Natural History Museum, London, UK (NHMUK). Further specimens from the current study were transferred to the natural history museums of Berlin and Frankfurt including their original accession numbers of the  Diagnosis. Body dorsally smooth on pereonites 1-5; pereonite 6 roundly produced mid-dorsally; pereonite 7 with blunt mid-dorsal process; pleonites 1-3 strongly carinate, each with a blunt dorsolateral ridge; epimeral plates 1-3 biangular, with acute posterodistal teeth in epimera 2 and 3; urosomite 3 with distinctly upwards pointed projection; palp of maxilla 1 distally with 4 blunt spines and few subdistal setae; coxa 4 apical tip broadly rounded; coxa 5 anteroventral margin nearly straight or only slightly concave, apical tip of process broadly rounded; uropod 2 outer ramus distinctly shorter than inner ramus (about ¾).
Maxilla 2 (Fig. 10C) inner lobe broader and shorter than outer lobe, with numerous setae along the medial margin; outer lobe with a row of slender spines which are mediodistally serrate along with few short slender setae. Maxilliped (Fig. 10D,E) inner plate about 2/3 the length of 1 st palp article, apical ridge with blunt spines and stout setae; outer plate reaching about midpoint of palp article 2, ovoid, apical margin with progressively stouter, slender spines mediodistally, mediodistal region produced to a saw-like ridge; palp 4-articulate, medial margin of article 2 setose, article 3 with stout setae mediodistally, article 4 curved, with serrate medial margin.
Gnathopod 1 (Fig. 11A) coxa tapering distally and with broadly rounded apical tip, anterior margin nearly straight, posterior margin slightly concave and bearing a row of very fine setae; basis about as long as coxa, slightly curved anteriorly, posteromarginally with some long slender setae, setation of anterior margin sparse with few long slender setae proximally and a row of short slender setae mediodistally; ischium slightly wider than long; merus with longitudinal articulation of carpus, posterior margin oblique, with tuft of long setae posterodistally; carpus distally slightly expanded, with groups of long setae along the posterior margin and few short setae anterodistally; propodus about 0.8 of carpus length, posteriorly with tufts of setae; palm convex, with fine serration; dactylus curved, with serrate posterior margin.
Gnathopod 2 (Fig. 11B,C) coxa slightly longer than coxa 1, tapering distally with rounded apical tip, anterior margin sinuous, posterior margin distinctly convex medioapically with few very fine short setae mediomarginally; basis slightly shorter than coxa, nearly linear, distal half posteriomarginally with groups of long slender setae, sparse row of short setae along anterior margin with few long slender setae; ischium about as long as wide; merus and carpus of similar length and setation to gnathopod 1, but slightly narrower; propodus about as long as carpus, with tufts of setae along the posterior margin; palm obliquely convex, with fine serration and few setae on the lateral face (Fig. 11C); dactylus as for gnathopod 1.
Pereopod 3 (Fig. 12A) coxa subequal to coxa 2 but slightly longer and wider; basis about 0.7 as long as coxa, slightly sinuous but sublinear, posterior margin with groups of long slender setae on its distal half, anterior margin with sparse setation; ischium slightly longer than wide; merus length about 1.2 of carpus length, slightly curved posteriorly, with short setae along the posterior margin and tufts of setae antero-and posterodistally; carpus with tufts of stout setae posteromarginally; propodus about 1.2 the size of carpus, with 6 pairs of spine-like setae along the posterior margin; dactylus about half the size of propodus, curved.
Pereopod 4 (Fig. 12B) coxa strongly projecting ventrally and robust, anterior margin sinuous and strongly curved posteriorly, posterior margin excavate and with a centrally positioned rounded lobe, posteroapical margin strongly concave, apical tip broadly rounded; basis linear, posterior margin convex and with groups of long slender setae, anterior margin straight and with sparsely distributed slender setae; ischium about as long as wide, with a tuft of setae posterdistally; merus about 1.5 the size of carpus, sublinear but slightly curved posteriorly, with few short setae along the posterior margin and tufts of setae antero-and posterodistally; carpus with tufts of stout setae along the posterior margin; propodus about 1.2 the size of carpus, with 6 pairs of spine-like setae posteromarginally; dactylus about half the length of propodus, curved.
Pereopod 5 (Fig. 12C) coxa strongly extended posteroventrally, posterior margin evenly concave, anterior margin convex and describing a rounded right angle, anteroventral margin nearly straight, apical tip of process broadly rounded; basis rectangular, with posteriorly produced margin which expands to a bifid subrectangular lobe posterodistally, partly covering ischium when leg is retracted, anterior margin with long slender setae proximally and tufts of setae along the distal half; ischium longer than wide, lobate; merus about 1.2 the length of carpus, slightly curved anteriorly and slightly expanding distally, anterior margin with few very short setae, a single spine-like seta anterodistally and posterodistally; carpus slightly expanding distally, with 4 tufts of spine-like setae along anterior margin and posterodistally; propodus about 1.4 the size of carpus, with 5 pairs of spine-like setae anteromarginally; dactylus nearly half the size of propodus, curved.
Pereopod 6 ( Fig. 13A) coxa subrectangular, longer than wide with an angular lateral lobe pointing posterodistally, anterior margin with fine setae; basis slightly curved posteriorly, with lobe roundly expanding proximally on the medial side of the posterior margin and posteriorly produced lateral margin which expands to a bifid subrectangular lobe distally, posteromedial lobe with row of short slender setae mediodistomarginally, anterior margin with long slender setae proximally and tufts of setae along the distal half; ischium longer than wide, lobate; merus about 1.2 the length of carpus, slightly curved anteriorly and slightly expanding distally, very short setae along the anterior margin and medioproximally on the posterior margin; carpus slightly expanding distally, with 4 tufts of stout setae anteromarginally; propodus about 1.3 the length of carpus, with 6 pairs of spine-like setae anteromarginally; dactylus large, more than half the length of propodus, curved.
Pereopod 7 (Fig. 13B,C) on right body side missing, except for the coxa; on left side all articles complete; coxa slightly longer than wide; basis much wider than pereopod 6, posteriorly lobate, posterodistal lobe subacute and partly covering ischium, with minute tooth anterodistally, anterior margin with few short setae; ischium about as long as wide, with minute tooth on the anterodistal margin; merus slightly longer than carpus, slightly curved anteriorly, anterior margin with pairs of short stout setae, posterodistal corner bearing a blunt spine; carpus with 4 tufts of stout setae anteromarginally; propodus 1.3 the length of carpus, with 6 groups of spine-like setae anteromarginally; dactylus about half the length of propodus, curved.
Uropod 1 (Fig. 13D) peduncle subrectangular, with small spines along medial and lateral margins, spines increasing in size from proximal to distal; rami subequal, about 1.2 the length of peduncle, margins bordered with small spines.
Uropod 3 (Fig. 13F) peduncle shortest, with small spines along the sublateral margin, and short setae on the medial margin; rami subequal in length, spination similar to rami of uropod 1 and 2.
Telson (Fig. 13G) about 1.5 as long as wide, slightly notched to only 10% of its length, apices broadly rounded.
Variation. In general, males of E. frankei resemble their females except for smaller body sizes. They can, however, be distinguished from the females by some distinct morphological features of the trunk (Fig. 13H). In direct comparison, the female pereon segments are clearly broader than in males, which is particularly visible in dorsal view. Males, in return, tend to feature a proportionally more pronounced pleon. Furthermore, male coxal plates tend to be narrower and less tubby/rounded, resulting in a smaller coxal surface area. There is little intraspecific variation except for the sexual dimorphism described above. In the examined material, pereon segments 6-7 and pleon segments of specimens from the Mediterranean, however, exhibited higher mid-dorsal processes (Fig. 13I). Diagnosis. Body dorsally smooth on pereonites 1-5; pereonite 6 not produced or only slightly produced mid-dorsally; pereonite 7 with mid-dorsal process; pleonites 1-3 strongly carinate, each with a blunt dorsolateral ridge; epimeral plates 1-3 biangular, produced to subacute posterodistal teeth in epimera 2 and 3; urosomite 3 with only slightly produced rounded hump; palp of maxilla 1 distally with 6 blunt spines and few subdistal setae; coxa 4 apical tip subacute; coxa 5 anteroventral margin distinctly or strongly concave, apical tip of process subacute; uropod 2 outer ramus only slightly shorter than inner ramus (about 4/5).
Head (Fig. 14B) about as high as long; ventral lobus slightly rounded, anterior corner rectangular. Rostrum curved, ventral margin rounded, about as long as 1 st peduncular article of antenna 1. Eyes large and rounded, bulging.
Maxilla 2 (Fig. 16C) inner lobe slightly broader and shorter than outer lobe, with numerous setae along the medial margin; outer lobe with a row of slender spines which are mediodistally serrate along with few short slender setae. Maxilliped (Fig. 16D,E) inner plate about 2/3 the length of 1 st palp article, apical ridge with blunt spines and stout setae; outer plate reaching about midpoint of palp article 2, ovoid, apical margin with slender spines which get stouter mediodistally, mediodistal region produced to a saw-like ridge; palp 4-articulate, mediodistal margin of article 1 with long setae, medial margin of article 2 strongly setose, article 3 with setae medio-and laterodistally, article 4 curved, with serrate medial margin.
Gnathopod 1 (Fig. 17A) coxa tapering distally and having a rounded subangular apical tip, anterior margin sinuous, posterior margin straight to slightly convex and bearing a row of short very fine setae along medioapically as well as longer setae subproximally; basis about as long as coxa, distinctly curved anteriorly, posteromarginally with tufts of long slender setae, anterior margin strongly setose with fine setae along its entire length; ischium slightly longer than wide, with tuft of setae posterodistally; merus with longitudinal articulation of carpus, posterior margin oblique, with group of setae posterodistally; carpus distally slightly expanded, with dense setation along the posterior margin and a tuft of shorter setae anterodistally; propodus about 0.8 of carpus length, posteriorly with setae along the margin and on the lateral face; palm convex, with fine serration; dactylus curved, with serrate posterior margin.
Gnathopod 2 (Fig. 17B,C) coxa slightly longer than coxa 1, tapering distally with a rounded subangular tip, anterior margin slightly sinuous, posterior margin distinctly convex medioapically with a row very fine short setae along the margin; basis slightly shorter than coxa, slightly curved anteriorly, posteromarginally with tufts of long slender setae, long slender setae along the entire anterior margin accompanied with few shorter slender setae; ischium slightly longer than wide, with group of setae posterodistally; merus similar in length and setation to gnathopod 1; carpus shape similar to gnathopod 1, with tufts of long setae on the posterior margin; propodus about 0.8 the length of carpus, with transverse rows of setae along the posterior margin; palm obliquely convex, with fine serration and 2 long blunt spines subdistally on the lateral face (Fig. 17C); dactylus as for gnathopod 1.
Pereopod 3 (Fig. 17D) coxa subequal to coxa 2 but slightly longer and wider; basis about 0.8 as long as coxa, sublinear but slightly curved anteriorly, posterior margin with tufts of long slender setae, anterior margin with long slender setae and a field of very short fine setae which extends to the lateral face; ischium about as long as wide; merus length about 1.3 of carpus length, slightly curved posteriorly and distally expanding, with few short setae at the posterior and anterior margins as well as groups of setae antero-and posterodistally; carpus slightly expanding distally, with tufts of stout setae posteromarginally; propodus about 1.1 the size of carpus, with 7 pairs of spine-like setae along the posterior margin; dactylus about half the size of propodus, curved.
Pereopod 4 (Fig. 18A) coxa strongly projecting ventrally and robust, anterior margin sinuous and strongly curved posteriorly, posterior margin excavate and with a centrally positioned rounded lobe, posteroapical margin strongly concave, apical tip subacute; basis linear, posterior margin convex and with long slender setae extending to the lateral face, anterior margin straight and with shorter slender setae; ischium slightly longer than wide, with a group of setae posterdistally; merus about 1.5 the size of carpus and sublinear, with few short setae along the margins and groups of setae antero-and posterodistally; carpus with tufts of stout setae along the posterior margin; propodus about 1.3 the size of carpus, with 6 pairs of spine-like setae posteromarginally; dactylus less than half the length of propodus, curved.
Pereopod 5 (Fig. 18B) coxa strongly extended posteroventrally, posterior margin slightly concave, anterior margin convex and strongly curved posteriorly, anteroventral margin distinctly concave, apical tip of process subacute; basis rectangular, with posteriorly produced margin which expands to a bifid subrectangular lobe posterodistally, partly covering ischium when leg is retracted, anterior margin with numerous setae along its entire length; ischium longer than wide, lobate; merus about 1.4 the length of carpus, slightly curved anteriorly, with short stout setae postero-and anterodistally; carpus slightly expanding distally, with tufts of spine-like setae along the anterior margin and posterodistally; propodus about 1.5 the size of carpus, with 7 tufts of spine-like setae anteromarginally; dactylus on right pereopod damaged, but clearly less than half the size of propodus, curved. Pereopod 6 ( Fig. 18C) coxa subrectangular, longer than wide with a laterally protruding ridge anteriorly and an angular lateral lobe pointing posterodistally, anterior margin with fine setae; basis slightly curved posteriorly, with lobe roundly expanding proximally on the medial side of the posterior margin and posteriorly produced lateral margin which expands to a bifid subrectangular lobe distally, posteromedial lobe with many slender setae mediodistomarginally, anterior margin with long slender setae proximally and few setae along the distal half; ischium longer than wide, lobate; merus about 1.3 the length of carpus, slightly curved anteriorly and slightly expanding distally, with very short setae along the posterior margin and tufts of setae postero-and anterodistally;  carpus slightly expanding distally, with 5 groups of stout setae anteromarginally; propodus about 1.5 the length of carpus, with 7 groups of spine-like setae anteromarginally; dactylus about half the length of propodus, curved.
Pereopod 7 (Fig. 18D) on right body side missing; on left side all articles complete; coxa longer than wide; basis much wider than pereopod 6, posteriorly lobate, posterodistal lobe subacute and partly covering ischium, anterior margin with few short setae; ischium about as long as wide; merus slightly longer than carpus, slightly curved anteriorly, anterior margin with groups of short stout setae, posterodistal corner bearing two setae; carpus with 4 groups of stout setae anteromarginally; propodus about as long as carpus, with 7 groups of spine-like setae anteromarginally; dactylus about half the length of propodus, curved.
Uropod 1 (Fig. 18E,F) on right body side with damaged outer ramus, but undamaged on left body side; peduncle subrectangular, with small spines along medial margin, spine size slowly increasing from proximal to distal; rami subequal, about 1.2 the length of peduncle, margins bordered with small spines.
Uropod 3 (Fig. 18H) peduncle shortest, with small spines along the sublateral margin, and short setae on the medial margin; outer ramus slightly shorter than inner ramus, spination similar to rami of uropod 1 and 2.
Telson (Fig. 14E) about 1.5 as long as wide, slightly notched to about 15% of its length, apices pointedly rounded.
Variation. Similar to E. frankei, males of E. cornigera resemble their females, except for smaller body sizes.
The described morphological differences between the sexes in E. frankei (see above) also apply to E. cornigera (Fig. 19).