Adding pieces to the puzzle: insights into diversity and distribution patterns of Cumacea (Crustacea: Peracarida) from the deep North Atlantic to the Arctic Ocean

Sampling and study-area properties

The study area includes the northernmost part of the North Atlantic, extending across the GIN-Seas up to the Arctic Ocean. The main bulk of specimens included in this study was collected during the following international projects and expeditions: IceAGE (Icelandic marine Animals: Genetics and Ecology; Cruise M85/3 in 2011; Brix et al., 2014a; Meißner et al., 2018), which is a follow up of the BIOFAR (Biology of the Faroe Islands; Nørrevang et al., 1994; Gerken & Watling, 1999) and the BIOICE project (Benthic Invertebrates of Icelandic waters; Omarsdottir et al., 2013), and PASCAL (Physical feedbacks of Arctic PBL, Sea ice, Cloud and Aerosol; Cruise PS106/1 in 2017; Macke & Flores, 2018) onboard the RVs Meteor and Polarstern, focusing on remote shelf-break and deep-sea habitats within a depth range of 579–2,748 m (Fig. 1). Grant support and field permits are available under BR3843/3-1 and AWI_PS106_00. Additional specimens from the Norwegian Sea and waters off Svalbard sampled by the MAREANO program (Thorsnes, 2009) and the University of Bergen were included (Table 1). Cumaceans were sampled in large amounts in all projects.

Temperature and salinity were considered as hydrographic variables for the evaluation of water-mass characteristics, which were available for IceAGE areas (http://www.vliz.be/en/imis?module=dataset&dasid=6211) and PASCAL stations (https://doi.pangaea.de/10.1594/PANGAEA.881579) measured just off the sea floor with a conductivity-temperature-depth profiler (CTD) (see Brix et al. (2012); Brix & Devey (2019)). Each area and station was allocated to a defined water mass according to the definition of Schlichtholz & Houssais (2002), which is applicable for the Fram Strait region and, thus, the entrance to the deep Arctic Eurasian Basin. For the GIN-Seas around Iceland, the definitions as described by Hansen & Østerhus (2000), Brix & Svavarsson (2010) and Ostmann, Schnurr & Martínez Arbizu (2014) have been used as baseline (Table S1). The records were manually divided into eight ecoregions based on their predominate water-mass characteristics after the combined definitions of Curtis (1975), Spalding et al. (2007) and Piepenburg et al. (2011): Warm North Atlantic water mass (North Atlantic Ocean, ecoregion 4), intermediate Subarctic water mass (East Greenland Sea, 2; Norwegian Sea, 5; Barents Sea, 7) and cold Arctic water mass (Arctic Basin, 1; Kara Sea, 8; North Greenland Sea, 3; White Sea, 6).

Sampling data and sample treatment

Specimens were obtained using different types of benthic sampling gear. Most frequently applied was the Rothlisberg-Pearcy Epibenthic sled (RP-EBS, Rothlisberg & Pearcy, 1976; Brattegard & Fosså, 1991), equipped with a net of 500 µm mesh size and ending in a collecting cod end of 300 µm mesh size. Different from the standard deployment protocols as outlined in Brenke (2005), the sampling during the PASCAL expedition was conducted while the vessel was attached to an ice floe during a 2-week passive drifting according to the ocean’s current with an average drift velocity of 0.12 kn. Furthermore, the Camera-Epibenthic sled (C-EBS; Brandt et al., 2013), the Brenke-Sled (Brenke, 2005) and the Giant Box corer (BC; Hessler & Jumars, 1974) were deployed. A detailed description of the sampling design is given in Brix et al. (2014a). Once the deployed gear was on board, the haul was carefully floated in seawater and evenly decanted gently over a series of sieves with mesh sizes of 1/0.5/0.3 mm, washed with sea water on a sieving table and bulk-fixed in precooled 96% undenatured ethanol. All samples were treated as described in Riehl et al. (2014), ensuring that the samples stayed consistently cooled. The samples were sorted either directly on board or afterwards in the laboratories of the German Center for Marine Biodiversity Research (DZMB, Senckenberg am Meer, Hamburg, Germany).

Morphological specimen identification

A total of 947 specimens (Table S2) were determined to the lowest possible taxonomic rank, based primarily on original species descriptions (e.g., Hansen, 1920; Sars, 1900). Species identifications were conducted at the Department of Biological Sciences (University of Bergen, Norway) and DZMB Hamburg using either a ZEISS SteREO Discovery V8 or Leica MZ12.5 dissecting microscope. Dissected pereopods and mouth parts were assessed under a ZEISS Primo Star compound microscope. High quality pictures with depth of focus were taken with a Leica DFC400 digital compound microscope camera using the Z-stacking option in the Leica Application Suite imaging software. Current authoritative classification follows the catalogue World Cumacea Database (http://www.marinespecies.org/cumacea/, Watling & Gerken, 2019) in the World Register of Marine Species (WoRMS Editorial Board, 2019). Additionally, comparative museums’ material has been obtained from the Center of Natural History Hamburg (CeNak) and the University Museum of Bergen (ZMBN).

Molecular methods

Phylogenetic analyses

Raw sequences were assembled and manually curated in Geneious® version 9.8.1 (Kearse et al., 2012). Consensus sequences were generated and blasted against GenBank database to identify potential contaminant sequences (e.g., bacterial sequences). We further included 67 cumacean sequences published on GenBank, 51 sequences originating from North Atlantic cumaceans currently studied at the University Bergen as well as 16 from outside the study area (Table 2).

Due to the large number of substitutions and indels, the alignment of all sequenced species included many long, ambiguously aligned regions, which would compromise the following analyses. For this reason, we split the data into four subsets of more closely related (and thus more similar) sequences, based on morphological family taxa and a preliminary phylogenetic analysis on the complete dataset (dataset 1 in Fig. 2; Table 3). These family-based alignments had fewer ambiguities and gaps and were thus used for subsequent analyses. Alignments were calculated separately for each of these four subsets with MAFFT 7.402 (Katoh, 2002; Katoh & Standley, 2013) on the CIPRES Science Gateway version 3.3 (Miller, Pfeiffer & Schwartz, 2010) using the L-INS-I algorithm and subsequently trimmed manually in BioEdit© version 7.0.5.3 (Hall, 1999). One outgroup species (represented by a member of one of the respective other cumacean families) was included in the alignments for the phylogenetic analyses but removed to further improve the alignment for genetic distance analyses.

The best-fitting evolutionary model was identified in MEGA X (Kumar et al., 2018), resulting in the General Time Reversible Model with invariable sites and gamma distribution (GTR + G + I; Lanave et al., 1984; Rodriguez et al., 1990; Nylander et al., 2004) for all data sets. Subsequent phylogenetic analyses were performed with MrBayes version 3.2.7a (Ronquist et al., 2012) with ‘nruns = 4’ and ‘nchains = 6’, 50 × 106 generations and a sample frequency of 5,000. The first 25% of sampled trees were discarded as burn-in. The resulting consensus trees were visualized with FigTree version 1.4.4 (Rambaut, 2018a).

Two different species delimitation methods were employed, one tree-based (general mixed Yule coalescent, GMYC, Pons et al., 2006) and one distance-based (automatic barcode gap discovery, ABGD, Puillandre et al., 2012). The single threshold model of GMYC was performed in R (R Core Team) for each of the four subsets. The required ultrametric trees were calculated with BEAST 2.5 (Bouckaert et al., 2019), employing the Yule coalescent prior, enforcing the ingroup as monophyletic and running the analyses for 30 * 106 generations and sampling every 3,000th generation. Convergence was assessed in Tracer (Rambaut et al., 2018b) and the final tree produced with TreeAnnotator (Bouckaert et al., 2019), removing the first 25% of generations as burn-in.

The relatively large number of potential singleton taxa could be problematic for tree-based species delimitation approaches. For ABGD, uncorrected p-distances were precomputed without an evolutionary substitution model in MEGA X including transitions and transversions as substitution mutations and missing data was treated by pairwise deletion. Sequences shorter than 300 bp were excluded. The web-based version of ABGD was used (https://bioinfo.mnhn.fr/abi/public/abgd/abgdweb.html), setting Pmin = 0.01, Pmax = 0.1, the relative gap width (X) to 0.5 and the number of steps to 100.

Distribution maps

For a general occurrence range overview, distribution maps were created using the software R version 3.5.3 and the PlotSvalbard package version 0.8.5 (Vihtakari, 2019) for each species. A single map incorporates the available georeferenced records from the open-access portal OBIS (OBIS, 2019; https://mapper.obis.org/, accessed 19/09/2019) and either the type locality or reference localities from earlier publications. Additionally, new occurrence records, not integrated in the OBIS platform, were added from published literature (e.g., Watling & Gerken, 2005) as well as records from other publicly accessible occurrence record libraries (e.g., MAREANO platform, Table S3). The new OBIS dataset Icelandic Cumacea (ICECU) was created for IceAGE and PASCAL specimens (Uhlir et al., 2021; http://ipt.vliz.be/eurobis/resource?r=cumacea_pascal_iceage). Information on cumacean species sampled and identified by the MAREANO project are accessible for specific taxa via the species-list portal (http://webprod1.nodc.no:8080/marbunn_web/viewspecies).

Combined approach: morphological and molecular species delimitation

The 947 investigated specimens were assigned to 77 morphological species, representing all seven known families (Table S4). For 58 species, identification to a known species taxon was possible. In all other 19 cases, specimens were assigned to genus or family level, but clearly differed morphologically from all other species of these genera or families identified in our study. The largest number of species were assigned to the Nannastacidae (20 species), followed by the Diastylidae (19), the Lampropidae (14) and the Leuconidae (15). In terms of DNA quality and success rate, the Marine Animal Tissue Genomic DNA kit yielded the best results for the cumaceans and can, thus, be recommended for further studies on this taxon.

In total, 131 specimens were included in the genetic analyses, representing 54 of the 77 morphologically identified species (Table 2). The Bayesian phylogenetic analysis of dataset 1 in Fig. 2 (all taxa) resulted in the monophyly of the Cumacea (posterior probability (pp) = 1). Except for the Nannastacidae (pp = 1), families were not recovered as monophyletic (Figs. 3–6), but this was not surprising, as a single fastly evolving marker like 16S is not suitable to properly resolve such deep nodes.

The ABGD analyses delimited 53 genetic lineages (representing putative species). With few exceptions, lineages were separated by a clear barcoding gap, with the vast majority of intra-lineage distances being <1% and inter-lineage distances > 8% (mostly 15–45%) (Fig. S1; Tables S5–S12). Cases of intra-lineage p-distance exceeding 1% were Diastylis rathkei Krøyer, 1841 (Dia07, 4%; Tables S5, S6) and Hemilamprops cristatus G. O. Sars, 1870 (Lam05-A, 2%; Tables S7, S8). GMYC resulted in nearly identical species delimitation, only Diastyloides biplicatus G. O. Sars, 1865 (Dia11) and Diastyloides serratus G. O. Sars, 1865 (Dia12) were grouped together (because these are separated by a genetic distance of 18%, we consider this to be an artifact due to the high number of singletons (D. biplicatus is also singleton) and treat them separately in the following as suggested by AGBD).

Inconsistencies between morphological and molecular species delimitation occurred in eight cases, which are summarized in Table 4. In four cases, genetic divergence was higher than expected by prior morphological determination suggesting cryptic diversity. Cyclaspis longicaudata Sars, 1865 was split into two distinct lineages (Bod05, A–B; Table S9, S10) as were Leptostylis borealis Stappers, 1908 (Dia15, A–B; Tables S5, S6) and Leptostylis sp. 1 (Dia17, A–B; Tables S5, S6). Eudorella truncatula Bate, 1856 was split into three lineages (Leu04, A–C; Tables S11, S12). Conversely, Leucon (Leucon) aff. nathorsti Ohlin, 1901 (Leu09) and L. (L.) nathorsti (Leu08) were treated as two morphologically differing species based on prior determination following Hansen (1920) (Leu09 with rather pointy rostrum; two dorsolateral teeth on frontal lobe). However, the low genetic distance (1%) suggests that they belong to the same lineage (Tables S11, S12). Finally, two problematic cases highlighted the mismatch between morphological and genetic delimitation. First, Leptostylis longimana Sars, 1865 was genetically split into two morphologically cryptic lineages (Dia16, A–B; Tables S5, S6), of which Dia16-A was collected close to the species’ type locality on the continental shelf and a Norwegian fjord and Dia16-B from Arctic Polar Water (APW). Furthermore, Dia16-B was genetically identical to Leptostylis ampullacea Lilljeborg, 1855 (Dia14), which was collected in Icelandic waters (Norwegian Sea Arctic Intermediate Water, APW-NSAIW) more than 2,500 km further north. However, after re-examination these specimens could barely be distinguished based only on weakly discriminating morphological characters following G. O. Sars (1900; clumsier form of body) from specimens identified as the original Leptostylis longimana (Dia16-A). Second, one species (Hemilamprops cf. cristatus; Lam05-B) morphologically very closely resembled Hemilamprops cristatus (Lam05-A) but was eventually differentiated based on a shorter rostrum and smaller, but more teeth within the serrated dorsal crest in Lam05-A. Also, genetic analyses suggested two lineages with strong divergence (23% p-distance), however, the Lam05-A specimen (sequence ID ICE1-Lam018) from the Greenland slope in Subarctic waters (APW-NSAIW), morphologically assigned to Lam05-B, clustered together with Lam05-A from the Norwegian continental shelf. All other Lam06 specimens were from Iceland Sea Overflow Water (ISOW) in the Iceland Basin.

Biogeographical data mining in OBIS

Within the investigated area, a total of 11,714 occurrence records including 44,933 individual specimens were extracted from OBIS (9,151 records), literature and other databases (2,270), and the new ICECU dataset added 293 records (Fig. 7). Out of these, about 6,200 records are in shelf regions up to 250 m depth, about 3,900 records in shelf-break regions between 250–1,000 m and 639 records below 1,000 m in the deep sea, excluding about 780 records with no available depth information. More than half of the specimens (25,496) were classified on order level as ‘Cumacea indet.’, whereas 19,437 specimens were identified to family or a lower taxonomic level. In total, 109 known species are recorded, of which 18 species of five families were recorded for the first time on the OBIS platform within the ecoregions 1, 2, 4, 5 and 7 (Fig. 8). The amount of data and specimen records varied remarkably among the predefined ecoregions 1–8. Ecoregion 4 was assigned to North Atlantic water mass characteristics comprising the highest specimen count (20,101), followed by ecoregion 5 (11,664), 2 (7,667) and 7 (3,505), which are composed of a mixture of North Atlantic and Arctic water masses and were, thus, assigned to Subarctic water-mass characteristics. Arctic water-mass ecoregions 8 (858 specimens), 1 (232), 6 (490) and 3 (29) contributed the lowest specimen sampling effort. There was a general trend of decreasing number of taxa (species diversity) with fewer specimens following the northern extension of the North Atlantic Current (NAC). For example, out of 232 individual specimens in ecoregion 1, 203 were assigned to 30 different species, while ecoregion 8, one of the last ecoregions influenced by the NAC, had a higher sampling effort with 858 specimens, but a lower species diversity with 407 individuals assigned to 18 species.

Species distribution patterns within ecoregions

Overall, the composition of taxa was observed to change from a Northern Atlantic-boreal (ecoregion 4, 5, 7) to a typical Arctic community (2, 1, 8, 6, 3; Fig. 9). Investigating the composition of the most frequently occurring taxa in OBIS revealed that the taxon Diastylis rathkei was recorded in all ecoregions, followed by Campylaspis rubicunda Liljeborg, 1855, Diastylis goodsiri Bell, 1855 and Brachydiastylis resima Krøyer, 1846, which were recorded in seven of eight regions. Compared to other ecoregions of the same size, the Arctic Ocean and the area around Iceland are underrepresented in cumacean occurrence records. Currently available records in these ecoregions are restricted to 90 entries of 32 species. Within these records, the most frequently occurring species are Diastylis polaris, Sars, 1871, Leptostylis longimana, Platytyphlops semiornatus Fage, 1929, Campylaspis globosa Hansen, 1920, C. valleculate Jones, 1974, Leucon (Epileucon) spiniventris Hansen, 1920 and Platycuma holti Calman, 1905.

Faunistic mix in ICECU material

The morphologically and genetically investigated material of the ICECU dataset corresponded well with the earlier observed trend in the OBIS dataset of high species diversity in the North Atlantic (ecoregion 4) with 45 representative, ecoregion-specific species (Fig. 10; Table 5). With the northern extension of the NAC, the representative species number decreases to seven in ecoregion 1 (Arctic Basin) and five in ecoregion 2 (East Greenland Sea). The number of shared species occurring in more than one ecoregion decreases with distance: While the adjacent ecoregion 4 and 5 share five species, ecoregion 4 and 1 only share one species (Campylaspis intermedia Hansen, 1920). The two species Platysympus typicus G. O. Sars, 1870 and Leucon (Alytoleucon) pallidus G. O. Sars, 1865 were recorded in all ecoregions. Comparing regional distribution patterns of families, a seamless shift in the relative composition along the GIS-Ridge could be observed (Fig. 9). The highest number of investigated specimens and species was recorded in stations located south of the Ridge in the Icelandic and Irminger Basins, which have warmer and more saline water masses (ISOW, Labrador Sea Water (LSW), North Atlantic Water (NAW); ecoregion 4) and were characterized by the families Lampropidae and Nannastacidae. Stations north of the Iceland-Faroe Ridge, the Denmark Strait and on the Yermak Plateau north of Svalbard are influenced by colder and less saline water masses (APW, NSAIW, cold & warm Norwegian Sea Deep Water NSDWc & NSDWw; ecoregion 1, 2) and were mostly characterized by the families Diastylidae and Leuconidae. Representatives of the family Bodotriidae were only recorded in southern stations (Station 983, ISOW; 1,057, LSW; 1,072, NAW), as well as the Ceratocumatidae (1,057, LSW; 1,072, NAW) and the only representative specimen of the Pseudocumatidae G. O. Sars, 1878 (1,057, LSW).

In correspondence with the OBIS data, Leptostylis cf. longimana (Dia16-B; Fig. 11A) and Leptostylis ampullacea (Dia14; Fig. 11B) were the most frequently recorded species, occurring at 13 out of 21 stations, followed by Diastylis polaris (Dia06; Fig. 11C) and Leucon (Alytoleucon) pallidus (Leu05; Fig. 11D) recorded at five stations each. Other wide-ranging species, such as Platysympus typicus (Lam13; Fig. 11E) and Hemilamprops cristatus (Lam05; Fig. 11F) were found across multiple ecoregions. A majority of the species were only present in samples from one or two stations. For characteristic boreal and Arctic taxa investigated in this study, such as Leptostylis borealis (Dia15; Fig. 12A), L. (A.) pallidus (Leu05) or Hemilamprops pterini Shalla & Bishop, 2007 (Lam07; Fig. 12B), additional occurrence records were contributed within the expected ranges. The species Cimmerius reticulatus Jones, 1973 (Cer01; Fig. 12C) was the first record for the family Ceratocumatidae to be found in the North Atlantic. Interesting new records of species in Icelandic waters such as Hemilamprops cf. diversus Hale, 1946 (Lam06; Fig. 12D), previously only known from off South-eastern Australia, and Cumellopsis cf. puritani Calman, 1906 (Nan13), mostly recorded in the Mediterranean Sea, might either represent unexpected wide distribution ranges for these species or presence of morphologically closely related species, but new to science.

In order to minimize the taxonomic knowledge gap, especially in ecoregion 2, the focus of this study is on specimens collected within the regions 1, 2, 4 and 5, which represent all three water-mass categories (North Atlantic, Subarctic, Arctic) and differed remarkably in their taxa composition.

Representatives of the Arctic Ocean (ecoregion 1)

One representative specimen of the circumpolar species Diastylis spinulosa Heller, 1875 (Dia08; Fig. 12E) was detected from north of Spitzbergen. In congruence with the OBIS occurrence records, ecoregion 1 was dominated by representatives of the families Diastylidae, Leuconidae and Nannastacidae. The morphologically examined material from the Arctic Ocean included only four species: Leptostylis cf. longimana (Dia16-B) dominated in specimen number, followed by Campylaspis rubicunda (Nan08; Fig. 12F), Campylaspis intermedia (Nan06; Fig. 13A) and one specimen of Leucon (Alytoleuco) pallidus (Leu05). Campylaspis intermedia is known to be widely distributed in the whole Atlantic Ocean but was not recorded in the Arctic yet. Thus, the occurrence records in the present study from north of Svalbard extend the previously assumed distribution range significantly. Campylaspis rubicunda is known from the North Pacific, the North Atlantic and the Arctic Oceans and Leptostylis longimana is widely distributed in the cold areas of the Northern Atlantic. Nevertheless, none of the previously mentioned species is found strictly in ecoregion 1, but rather widely distributed and occurring in other ecoregions.

Representatives of East Greenland Sea (ecoregion 2)

The most common species in ecoregion 2 was Diastylis polaris (Dia06), which was sampled at five IceAGE stations composed of NSDWw and NSDWc. This observation corresponds with the occurrence records found in OBIS, which also include the morphologically closely related Diastylis stygia G. O. Sars, 1871, both being described as true Arctic species. The species Leptostylis borealis (Dia15), originally described south of Franz-Josef-Land, was recorded for the first-time off Iceland. Most representatives of the family Leuconidae in the IceAGE material were sampled in ecoregion 2, north of the Iceland-Faroe Ridge. This result suggested a restricted distribution range of sampled representatives of the family Leuconidae to colder and fresher water masses (e.g., APW, NSAIW and NSDW). The species Leucon (Leucon) profundus Hansen, 1920 (Leu11; Fig. 13B) and Leucon (Alytoleucon) pallidus (Leu05) were the most common species of this family in the investigated material. Vassilenko (1989) and Gerken & Watling (1999) described both with a circumpolar distribution range from the North Atlantic Ocean to the Canadian Arctic Ocean, but mostly found in cold-water stations (Watling & Gerken, 2005). However, by our samples, new occurrence records of these species were added in the Norwegian Sea waters influenced by the North-Atlantic current, suggesting an extension of the distribution range for warmer waters of the previously assumed cold-water species.

Representatives of the North Atlantic and Norwegian Sea (ecoregion 4 and 5)

Ecoregion 4 was dominated in specimens by Hemilamprops cf. diversus (Lam07) and Hemilamprops pterini (Lam08), which are taxa found predominantly in warmer waters, as well as Campylaspis sp. 2 (Nan10; Fig. 13C). In general, the genera Hemilamprops and Campylaspis reached highest record numbers both in terms of taxa and specimen count. The widely distributed boreal Atlantic species Hemilamprops cf. cristatus (Lam05-B) is recorded to occur in high abundances within ecoregion 4 and 5, though it was also sampled at one station in the Denmark Strait. Specimens in ecoregion 4 were mostly sampled in deep-sea habitats, while specimens in ecoregion 5 were mostly sampled in coastal and fjord habitats. Typical Atlantic Ocean species of other families only recorded south of the GIS-Ridge in warmer waters were Diastyloides atlanticus Reyss, 1974 (Dia10; Diastylidae; Fig. 13D), Bathycuma brevirostre Norman, 1879 (Bod01; Bodotriidae) and Cyclaspis longicaudata Sars, 1865 (Bod05; Bodotriidae; Fig. 13E), as well as Pseudocuma sp. 1 (Pse02), the only representative of the Pseudocumatidae in the ICECU dataset. Cimmerius reticulatus (Cer01), expanding its distribution northward from the Bay of Biscay, is the only representative of the Ceratocumatidae. The specimens of Eudorella truncatula sampled in ecoregion 4 (Leu04-A/-B; Fig. S2-U) and ecoregion 5 (Leu04-C) were all separated into distinct lineages, which was also observed in Cyclapsis longicaudata from ecoregion 4 (Bod05-B) and ecoregion 5 (Bod05-A).

Combined species delimitation approach - morphology and genetics

As only a handful of studies applied molecular methods for species delimitation within cumaceans, the combined approach in this study highlights the high quality and overall congruence in most cases of morphological and genetic analyses. Even though delimiting species based solely on one mitochondrial marker like 16S rDNA has been questioned as there is no universal threshold for species delimitation (Meyer & Paulay, 2005; Meier et al., 2006; Wiemers & Fiedler, 2007; Schwentner, Timms & Richter, 2011; Collins & Cruickshank, 2013), in our study a clear barcoding gap between 2% and 8% was observed among most inferred species, corresponding nicely with morphological taxonomic characters in the vast majority of cases. The 4% genetic distance observed between the geographically widely separated (>3,000 km) Diastylis rathkei (Dia07) individuals cannot be easily interpreted as either intra- or interspecific and might hint at a recent and/or ongoing speciation event. Similar observations have been made previously for 16S rDNA of cumaceans and other peracarids like isopods (Brökeland & Raupach, 2008; Held & Wägele, 2005; Rehm, 2007; Rehm et al., 2020; Riehl, Lins & Brandt, 2018). Herein, intraspecific distances were usually below 1%, though geographically widely distributed species featured up to 3% or 5%, respectively. Conversely, interspecific distances exceeded 7%. Also, here the higher intraspecific genetic distances of 5% were tentatively suggested to represent potential cryptic species (Rehm, 2007). Based on these findings we suggest that the rare cases of conflict between morphological and genetic data represent cases of cryptic diversity or extensive morphological variability.

Examples of morphologically cryptic diversity

The incongruence of 16S sequence data revealing high genetic diversity within a species (or even cryptic species) was observed in six cases (Table 4). This affects Hemilamprops cristatus (Lam05, A–B), Eudorella truncatula (Leu04, A–C), Cyclaspis longicaudata (Bod05, A–B), Leptostylis borealis (Dia15, A–B) and Leptostylis sp. 1 (Dia18, A–B; Fig. S2L). The intraspecific genetic distances observed within each of these taxa greatly exceeded those commonly observed within species of Cumacea and other peracarids. We therefore conclude that all of these cases indicate presence of morphologically cryptic species. Moreover, some of these cryptic species were not even recovered as sister species, but widely separated (Fig. 2). Further taxonomic studies will be needed to prove these cases.

In Cyclaspis longicaudata (Bod05, A–B), Eudorella truncatula (Leu04, A–C) and Leptostylis borealis (Dia15, A–B) the respective cryptic species were geographically well separated and usually occurred in different water masses. In the case of Hemilamprops cristatus (Lam05, A–B), morphological re-examinations showed weak differences in the rostrum length and the serration of the dorsal crest on the carapace (shorter rostrum and smaller, but more teeth in Lam05-A). However, one specimen (ICE1-Lam018) sampled on the Greenland slope in Subarctic waters and morphologically corresponding to Lam05-B, clustered genetically together with Lam05-A from the Norwegian continental shelf.

Our cumacean examples do support the finding in other peracarid taxa (amphipods and isopods) of either overestimations (Lörz et al., 2020) or underestimations of intraspecific divergence (Brix, Svavarsson & Leese, 2014b; Jennings, Golovan & Brix, 2019; Paulus et al., in press). This emphasizes that sampling from a geographically limited portion of a species’ range only risks missing relevant genetic variation, which blurs an important line between species-level and population-level diversity. Molecular species delimitation should, thus, include specimens sampled in the widest possible distribution range of the examined taxon (Knox, 2012).

Examples of morphological variability and taxonomic incongruence

Specimens identified as Diastylis polaris (Dia06) and Diastylis stygia were identical in 16S. Interestingly, Zimmer (1926) synonymized these species based on Ohlin’s (1901) and Stebbing’s (1913) conclusion of their conspecifity. In Zimmer, 1980 re-examined a specimen of D. stygia collected by the Russian Sadko Expedition (Zimmer, 1943) and separated it again from D. polaris as a valid species. Our study lays additional support for D. stygia being a synonym of D. polaris. Similarly, after re-examination of Platysympus typicus (Lam14) sampled off East Greenland and P. tricarinatus Hansen, 1920 from the Norwegian shelf, these two species names are probably representing the same species. Gerken (2018) called the assumed differentiating characters of more or less conspicuous folds on the carapace into doubt and suggested that less prominent characters might be owed to juvenile stages. Therefore, P. tricarinatus is assumed to be a synonym of P. typicus. Similarly, Leucon (Leucon) aff. nathorsti (Leu09) differed morphologically from L. (L.) nathorsti (Leu08) by the presence of two dorsolateral teeth on the frontal lobe of the carapace and a more acute rostrum and, thus, identified as a possible separate species. However, in this 16S analyses these two morphotypes proved to be identical, suggesting the presence of a single, morphologically variable species. As Leu08-specimens were sampled on the Norwegian continental shelf in North Atlantic waters and Leu09 was collected off Svalbard in Subarctic waters, an ecologically-driven morphological population variation might be implied.

Further, we found that the two morphologically almost indistinguishable species Leptostylis longimana (Dia16-B) and L. ampullacea (Dia14) grouped into a large unresolved clade forming a “Leptostylis longimana/ampullacea” species complex. Species in the genus Leptostylis are rather difficult to distinguish as there is a certain degree of phenotypic plasticity tied to sex and growth stages, and some morphological distinctions are quite subjective, such as ‘clumsier’ body form of L. ampullacea compared to L. longimana (Sars, 1900). A second “Leptostylis longimana”-clade was retrieved, based on specimens from coastal Norway, genetically well separated from the Iceland/Arctic “L. longimana/ampullacea”-clade by 26–27% p-distance. Based on the present data, the Iceland/Arctic specimens should be referred to L. ampullacea, originally described from Kullaberg (off Sweden). This further implies that the “true” L. longimana, originally described from the Oslofjord, a more coastal and/or southern distribution. Further studies including additional specimens and gene markers, as well as museum type material will be necessary to resolve the taxonomy in more detail.

Biogeographic integration