The genus Rhachotropis has the widest geographic and bathymetric distribution of all amphipod genera worldwide. Molecular and morphological investigations of specimens sampled around Iceland and off the Norwegian coast allow the first insights into the relationships of North East Atlantic Rhachotropis. The 31 cytochrome oxidase subunit I (COI) sequences generated for this study were assigned 13 Barcode Index Numbers (BINs) in the Barcode of Life database (BOLD), of which 12 are new to the database. Molecular analyses of COI and 16S sequences could not confirm a theory that depth has a greater influence on the phylogeny of Rhachotropis than geographic distance. Although the North East Atlantic is a well-studied area, our molecular investigations revealed the genus Rhachotropis may contain cryptic species, which indicates a higher biodiversity than currently known. For example, the specimens which key to Rhachotropis helleri is a complex of three COI clades, two of which cannot be identified with morphological traits. One specimen of each of the clades in the cladogram was documented by high definition photographs. A special focus was on the visual morphology of the eyes, as this character shows interspecific differences within the genus Rhachotropis in response to fixation in ethanol. Detailed morphological investigation showed that some clades thought to be indistinguishable can be separated by minute but consistent morphological characters. Datamining Genbank to examine all registered COI-sequences of R. aculeata, the only previously known Rhachotropis BIN in the North Atlantic and sub-Arctic, showed R. aculeata to be subdivided by an Arctic and a North Atlantic population.

Amphipoda , Eusiridae , North Atlantic, IceAGE, NorAmph, COI, 16S

Eusiridae are fast moving predators with a worldwide distribution. The genus Rhachotropis has the widest geographic (all oceans) and bathymetric (0–9460 m) distribution of all amphipod genera (Lörz et al. 2012). Currently, the genus Rhachotropis (Eusiridae) contains 61 species (World Amphipod Database, Horton et al. 2017).

Expeditions to the North East (NE) Atlantic via the programs IceAGE (Icelandic Animals Genetics & Ecology), Mareano and several smaller Norwegian mapping projects (Norwegian and Spitsbergen economic waters) sampled various Amphipoda during the last 10 years. Almost all amphipod collections yielded members of Eusiridae. Rhachotropis was the dominant genus in eusirid samples, along with three other genera: Eusirus, Cleonardo and Eusirella.

As the material was suitable for DNA analysis we investigated the relationships of freshly collected Rhachotropis from the NE Atlantic to each other via analysis of the cytochrome oxidase I (COI) and 16S gene regions. We then set these new specimens in context to Rhachotropis from Norway and other parts of the world.

Twenty-one of the 61 described Rhachotropis species are known from the NE Atlantic and Arctic region: Rhachotropis aculeata (Lepechin, 1780); R. arii Thurston, 1980; R. distincta (Holmes, 1908); R. faeroensis Stephensen, 1944; R. gislii Thurston, 1980; R. gloriosae Ledoyer, 1982; R. gracilis Bonnier, 1896; R. grimaldii (Chevreux, 1887); R. helleri (Boeck, 1971); R. inflata (Sars, 1883); R. aff. kergueleni Stebbing, 1888; R. leucophthalma Sars, 1883; R. lomonosovi Gurjanova, 1934; R. macropus Sars, 1883; R. northriana d’Udekem d’Acoz, Vader & Legezynska, 2007; R. oculata (Hansen, 1887); R. palporum Stebbing, 1908; R. proxima Chevreux, 1911; R. rostrata Bonnier, 1896; R. thordisae Thurston, 1980; R. thorkelli Thurston, 1980.

Lörz et al. (2012) hypothesized that depth might have a greater influence on the phylogeny of the genus Rhachotropis than geographic distances. The large amount of freshly collected Rhachotropis from the NE Atlantic allowed us to test this theory. Here we consider this theory using new data from specimens from 110 m to 2750 meters depth in the NE Atlantic. We ask and discuss the following questions:

1) What are the phylogenetic relationships of NE Atlantic and Arctic Rhachotropis?

2) Do relationships among specimens from the shallow (Norwegian Channel) and deep (Icelandic Basin and Norwegian Sea) stations indicate biogeographic processes such as submergence or emergence?

3) Does depth have a bigger influence on the phylogeny of Rhachotropis than geographic distance?

The detailed description of the Icelandic study area is presented in the introduction of this volume (Brix et al. 2018). Large amphipod specimens were immediately sorted on deck, fixed in 98% ethanol and later transferred to 96% ethanol. Samples were decanted and the decant bulk fixed. The Norwegian amphipods were collected as parts of several mapping and teaching cruises and all examined through the Norwegian Biodiversity Information Centre financed project NorAmph at the University of Bergen. Material from the mapping projects “Mareano” and “Sognefjord” were sampled using RP-sleds where the material was decanted immediately and the decant bulk-fixed in 97% ethanol (Buhl-Mortensen et al. 2015). The high-arctic material from the Spitsbergen region was collected during student-cruises at the University Centre of Svalbard. Samples were sorted by teachers and students before being stored in 96% ethanol and kept in the cold-room for transport to land.

Overall the morphological differentiation of the Rhachotropis sampled in the NE Atlantic is mirrored in the differentiation of our mitochondrial DNA markers and reflected both in gene tree topology and genetic distances. However, both R. aff. helleri and R. inflata have diverged into groups that were unnoticed a priori from morphology and R. aff. inflata clusters with one of the latter lineages. The taxonomic status of R. aff. helleri versus R. helleri must be examined further with more data. There is also considerable COI divergence in the R. proxima group.

Our DNA sequence data are shedding new light on the species relationships of Rhachotropis, although based on one gene fragment only the phylogenetic trees should certainly be interpreted with caution. The difference between gene trees and species trees has been an important topic in theoretical phylogenetics since the seminal publication by Maddison (1997). The phenomenon of lineage sorting and deep coalescence certainly also has implications for our understanding of species, particularly with respect to the idea of divergent monophyletic lineages and of species delimitation based on a single mitochondrial marker (Degnan and Rosenberg 2006, 2009, Knowles and Kubatko 2010). While the DNA barcoding campaign has sparked an excess of new species discoveries, it should be remembered that gene trees are representation of genetic structure and that non-monophyletic or genetically distant subclades are not necessarily separate species (Sukumaran and Knowles 2017). We have found that there is generally good correspondence between morphological species assignments and genetic structure in Rhachotropis but that there are also several examples of haplotype clusters that are candidates for broader systematic examination with multiple genes and morphological analyses.

The following discussion is divided according to three questions asked.

1) What are the phylogenetic relationships of North East Atlantic and Arctic Rhachotropis?

The FastTree approach and the Bayesian method returned very similar tree topologies with minimum exceptions. In both cases there was strong support for most of the species clades and also for some sister species relationships. However, many of the deeper branches were less well supported, which should be kept in mind when inferring the evolutionary history of ecological and biogeographical events.

Only three species occurred in both sampling sets from our two collecting groups in Icelandic and Norwegian waters: R. aculeata (Fig. 6A), R. helleri (Figs 4B, 5B, 6B) and R. inflata (Figs 4D, 5D). Eight species were only sampled during the IceAge project, four species only during NorAmph. However, we assume that the species collected in certain depth strata could be found in the corresponding depth area of other areas. We assume that depth and temperature have a greater effect than other characteristics, e.g. sediment type; for example Rhachotropis helleri was found on the Reykjanes ridge in 289 m (Fig. 4B), the Lofoten area off Norway in 850 m, in the Sognefjord 307 m (Fig. 6B) and the Norwegian channel 203 m (see Table 1). The dominant sediment of the Reykjanes ridge is muddy sand with a small proportion of gravel (Meißner et al. 2014), whereas the R. helleri off Norway were collected on sediments dominated by sandy mud with little gravel.

Figure 4. 

Lateral view photos of IceAGE material representing different clades in the analyses. A Rhachotropis gisliiAMPIV004-17B R. helleriAMPIV233-17CR. aff. helleriAMPIV011-17D R. inflataAMPIV070-17E R. northrianaAMPIV225-17F R. oculataAMPIV228-17GR. aff. palporumAMPIV033-17HR. aff. proximaAMPIV005-17IR. cf. proximaAMPIV002-17J R. sp. B AMPIV009-17K R. thordisaeAMPIV007-17LR: thorkelliAMPIV078-17.

Weisshappel (2000) investigated the biogeographic distribution of Eusiridae species around Iceland and found a marked separation along the Greenland-Iceland-Faroe (GIF) Ridge. Weisshapel assumed that eusirids followed relatively sharp boundaries between water masses of different origin lying on and adjacent to the GIF ridge or within the area on each side of the ridge.

Weisshappel (2000) showed that most eusirid species are restricted to certain combinations of bottom water temperatures and bottom depths, 28 out of 36 were found within a single water mass. Different Eusiridae species show different biogeographic preferences. Eusirus holmii, for example shows a very narrow temperature distribution – it is only collected in waters colder than 1°C – but has a very wide depth range, 400 to 1600 m. R. aculeata is the opposite: it has a wide temperature tolerance (-1°C to +6°C), and a relative narrow depth distribution, 100-600 m. There seems to be some divergence (Fig. 3) between Atlantic and Arctic populations. Whereas the R. aculeata specimen from Spitzbergen Isfjord at 123 m cluster with the latter group, the R. aculeata from 580 m in the Norwegian Sea is closer to the COI haplotypes occurring in Canadian West Atlantic waters.

Figure 5. 

Dorsal view photos of IceAGE material representing different clades in the analyses. A Rhachotropis gisliiAMPIV004-17B R. helleriAMPIV233-17CR. aff. helleriAMPIV011-17D R. inflataAMPIV070-17E R. northrianaAMPIV225-17F R. oculata AMPIV 228-17 GR. aff. palporumAMPIV033-17HR. aff. proximaAMPIV005-17IR. cf. proximaAMPIV002-17J R. sp. B AMPIV009-17K R. thordisaeAMPIV007-17LR: thorkelliAMPIV078-17.

An even more temperature-tolerant species is R. inflata which only occurs in shallow waters above 400 m, but tolerating temperatures from 0°C to 9°C (Weisshappel 2000 and current study). Morphological separation of R. inflata (Figs 4D, 5D) and R. aff. inflata (Fig. 6D) is minute: a dent on urosomite 1 in contrast to a non-dented urosomite 1. When applying these two morphology based groups in computation of p-distance, the mean divergence is only 3.5 % (Table 3). However, COI data divides R. inflata into two distinct clusters, one also including R. aff. inflata, the latter which clusters with sequences of R. inflata from Nunavut. The mean p-distance between the two R. inflata clades is actually 15 %, so these clades are clearly candidates for critical taxonomic examination with multiple gene data. Not enough material was available for a study of the character development in different growth stages in both genders within the framework of this study.

DNA barcoding revealed three clades within a species that keyed out to Rhachotropis helleri. The original description of R. helleri was provided by Boeck (1871); one brief paragraph in latin with no illustrations. Sars (1895) illustrated R. helleri in detail. We regard the specimens illustrated and described by Sars as corresponding to the species named by Boeck. Further morphological investigation of our R. helleri samples revealed two helleri groups. The first group has a smooth dorsal seventh pereonite and eyes larger than 50 % of head (see Figs 4B, 5B, 6B) – this we named R. helleri as it closely matches the drawings provided by Sars (1895). The second group specimens bear a small distinct spine on their seventh pereonite and have a smaller eye, covering a maximum of 20 % of the head. We provisionally named these R. aff. helleri (Figs 4C, 5C, 6C). The specimens of the R. aff. helleri group split into two molecularly distinct clades (Fig. 1, Suppl. material 3) with a p-distance of 8.56 %. We were not able to find a morphological differentiation in this group.

We conclude that the two aff. helleri groups therefore either represent a single species, which contains widely-divergent COI sequence, or that the two clades may be two species, which are genuinely cryptic rather than pseudo-cryptic (Janosik and Halanych 2010; Brandt et al. 2014). Although there may be good reasons to flag potentially new species to attract further attention (Wägele 1994), we do not propose the formal erection of a new species here because the aff. helleri can be only distinguished by the presence-absence of one spine and eye size, characters that can vary with sex and body size, and the two aff. helleri can currently only be reliably identified by DNA sequencing.

R. macropus (Fig. 6F) and R. lomonosovi (Fig. 6E) show the smallest interspecific distance of 9.8 % (Table 2). R. macropus and R. lomonosovi are genetically (Table 3) and morphologically (Fig. 6E, 6F) very similar. R. macropus is known to be very numerous along the Norwegian coast and in Svalbard (d’Udekem et al. 2007). While generally found at moderate depths of 200-700 m, we have two specimens from 1230 m in our samples, which extends the depth range by 500 m. R. lomonosovi is a true Arctic species, common at depths of 1000-1500 m (d’Udekem et al. 2007). Our single representative of this R. lomonosovi is from the lower Arctic/ Norwegian northern coast (outer Lofoten area), from a depth of 823 m, but from a water temperature of -0.58°C.

Figure 6. 

Lateral view photos of NorAmph material representing different clades in the analyses. A Rhachotropis aculeataAMPNB077-13B R. helleriAMPNB276-15CR. aff. helleriAMPNB279-15DR. aff. inflataAMPNB524-17E R. lomonosoviAMPNB352-15; F. R. macropusAMPNB443-16.

2) Do relationships between the shallow (Norwegian Channel) and the deep (Icelandic Basin and Norwegian Sea) amphipods indicate biogeographic processes such as submergence or emergence?

Generally more species are currently known from the shelf and upper slope area. However, the observed depth pattern is heavily collection biased: areas with more stations show more species (Lörz et al. 2012). Bousfield and Hendrycks (1995) attempted a phylo-biogeographic analysis of Rhachotropis of the North Pacific. Some of the species, such as R. aculeata, R. helleri, R. macropus, R. oculata, R. inflata and R. distincta are known from both the North Pacific and the North Atlantic, but these distribution data are based on literature using morphological data only. Bousfield and Hendrycks (1995) postulated that species such as R. aculeata and R. macropus belong to more primitive forms. Characters considered plesiomorphic by Bousfield and Hendrycks (1995) were, for example, an elongate telson, long rostrum, strongly toothed pleon and urosome. These so-called primitive species have a large body size (10.5- 40 mm). In contrast the more advanced oculata-inflata groups have small body lengths (3.8-12 mm), smooth urosome and short telson. While the primitive large species were generally coastal shelf species, the species become smaller, more weakly spinose and anoculate in most abyssal members. Bousfield and Hendrycks (1995) assume submergence as the primary biogeographic pattern amongst the Rhachotropis species in general and amongst the North Pacific in particular. Our findings cannot confirm this. Neither can we agree that shallow = large, spiny species. Our most abundant species was collected at the shallowest station, R. northriana. This is a very small, not very spinose species with a relatively large telson. Several abyssal species are known to reach large body sizes, such as R. rossi (12 mm) or R. abyssalis (13 mm). Although some of the deeper nodes in our phylogeny (Fig. 1, Suppl. material 4) are not well supported, we also cannot agree that R. aculeata and R. macropus are the most primitive forms, as in our molecular analysis of COI and 16S they are not in a basal position. When observing depth distributions of Rhachotropis clades, there does not seem to a directional trend in terms of emergence or submergence of during speciation.

Eyes

The genus Rhachotropis is known to have a diversity of “eye phenotypes”. R. leucophthalma G. O. Sars, 1893 is a white-eyed species – this feature is so prominent that it provided its name; its eyes become colourless and hard to see in alcohol (d’Udekem et al. 2007), whereas the true arctic species R. lomonosovi Gurjanova, 1934 also has white eyes when alive, but they turn black in alcohol. (This is sadly not easily seen in Fig. 6F, as the specimens cuticulum has loosened from the ommatidia leaving the eye colour an indistinct grey.) Rhachotropis oculata has kidney-shaped eyes which keep their colour, eg. AMPIV228-17 Figs 4F, 5F and for the ommatidia Fig. 7B. (Our specimens match the description of R. oculata by Hansen, 1887, but differ in various characters from R. oculata pictured by Bousfield and Hendrycks 1995.)

Rhachotropis helleri (AMPIV233-17) has distinct red eyes, even after being preserved in 98% ethanol for more than 4 years (Fig. 7A). The ommatidia are clearly hexagonal in shape. After 6 years of preservation the eye colour and the outline fade, e.g. AMPIV011-17 Figs 4C, 5C. Here there is a clear difference between the R. helleri (largish eyes, and clearly defined Figs 4B, 5B, 6B) and the R. aff. helleri (see also Fig. 6C) which have much smaller eyes and no clearly visible edge of the eye.

Figure 7. 

Eyes of, A Rhachotropis helleriAMPIV233-17, ommatidiae (Photo) B Rhachotropis oculataAMPIV228-17 (CLSM), 7 stack-layers have been removed in the eye-region to reveal the ommatidia beneath the cuticulum.

R. northriana has distinct red eyes also clearly visible after being preserved in ethanol for several years (eg. AMPIV225-17, Figs 4E, 5E).

Another species which maintains clearly visible dark eyes in preservation is R. inflata (eg. AMPIV070-17 Figs 4D, 5D).

We therefore think it unlikely that the ancestral Rhachotropis was a blind inhabitant of the deep sea. We assume that submergence has led to the loss of eyes in truly deep sea or abyssal species such as R. thordisae (Fig. 4K) and R. thorkelli (Fig. 4L). The latter species prove that Rhachotropis can hunt without eyesight, however all species collected in the photic zone have well developed eyes.

3) Does depth have a bigger influence on the phylogeny of Rhachotropis than geographic distance?

Rhachotropis specimens are found in all major oceans of the world: Arctic, Atlantic Ocean, Mediterranean Sea, Caribbean Sea, Indian Ocean, Pacific Ocean and the Southern Ocean (Lörz et al. 2012). Rhachotropis specimens have been collected in all water depths, from the shelf (e.g d’Udekem et al. 2007, Lowry and Springthorpe 2005, Lörz 2015) to abyssal and hadal sampling sites (Thurston 1980, Lörz 2010), in trenches (Dahl 1959), as well as around hydrothermal vents (e.g. Bellan-Santini 2006). Based on molecular data (COI) Lörz et al. (2012) predicted that depth had a greater influence on phylogeny than geographic distance.

The genetic distance between shallow species such as Rhachotropis aculeata from 600 m and deep sea species such as R. thordisae from 2750 m is only 16 %, the shallow water R. inflata from 123 m and the deep sea R. thorkelli is 20.7 %. The two deep sea species R. gislii and R. thordisae were collected at the same station at 2750 m depth and show a similar genetic distance of 22 % as the two species R. cf. proxima and R. aff. inflata (21%) both later collected around 900 m (Table 3). The smallest genetic distance was observed amongst the species R. lomonosovi and R. macropus, which were collected in 823 m and between 300 m and 1230 m respectively. Potentially our sampled depth band in the North Atlantic is not wide enough to sufficiently illuminate biogeographic questions. It is likely that samples from abyssal and hadal depths will heavily influence the interpretation of distribution patterns. However, when we include Rhachotropis samples from GenBank not collected in the NE Atlantic the lack of correlation between genetic distance and depth as well as the lack of correlation between genetic distance and geographic distance continues. R. abyssalis was sampled at 3210 m and its genetic distance to R. cf. proxima from South Iceland at 900 m is only 19 % – the same distance as to R. rossi, which was collected below 3000 m in the Ross Sea.

The intraspecific distance is around 5 % in R. aff. helleri, and less than 1 % in R. aff. palporum, R. chathamensis and R. macropus, even though the latter was sampled in a depth range of 330–1260 m. Recent investigations by Brix et al. (2014) of the eurybathic isopod species Chelator insignis off Iceland presented a different picture: the isopods collected in different depth strata had high genetic distances over 20%, indicating narrow vertical distribution ranges of cryptic species and highlighting bathymetry as key player in speciation processes. Jażdżewska et al. (2018) provide the first insight into cryptic amphipods off Iceland revealed by DNA barcoding.

A characteristic species, R. aculeata (Fig. 6A), was selected for further investigation via COI sequences available online. For a better visualization of genetic distances a haplotype network was calculated from the COI alignment (Fig. 3). The network shows 13 haplotypes of R. aculeata, but the maximum distance is only seven mutation steps between haplotypes, proving R. aculeata to have a remarkable truly circum-arctic distribution. The haplotype network shows that the Svalbard specimen clusters with the population in the Canadian Arctic (Fig. 3). The geographic distance between these population is thousands of km. Even though Rhachotropis are known to be good swimmers, this is a tremendous distribution for an animal with no larval stage. Due to their predatory feeding behaviour it is unlikely that R. aculeata clings to drifting algae or ice.

Morphologically separated groups of Rhachotropis are well supported by the genetic markers COI and 16S, with possible cryptic species in Rhachotropis aff. helleri. We recommend a morphological study of allometry in this genus, where many species often are collected in large numbers. Our present data lead us to support the theory that Rhachotropis originated in shallow (photic) seas, and has subsequently submerged to greater (subphotic) depths, with loss of eyes for the abyssal species. The question about geographic versus bathymetric distance as a driver for genetic distance is harder to answer, as there is no clear picture for the entire genus.

The IceAGE material was collected during expeditions in 2011 and 2013 with the RV Meteor and RV Poseidon and sorted with the financial support of the Volkswagenstiftung “Forschung in Museen”. The Norwegian material was collected partly by the MAREANO programme financed by the Norwegian Government, partly by the AB-321 student-cruises of the University Centre of Svalbard in 2007, 2009 and 2015, and partly by the Sognefjord benthic mapping project financed by the Norwegian Biodiversity Initiative. DNA-sequencing of the Univ. Bergen specimens was facilitated by funding from the NORBOL consortium, and produced by the Canadian Centre of DNA-barcoding, Guelph, Canada. Photographic support was provided by T. Dalsgaard and F. Friedrich in Hamburg and K. Kongshavn in Bergen. Curatorial support has been provided by Antje Fischer DZMB, Kathrin Philipps-Bussau CeNak Centre of Naturkunde Hamburg and Jon Kongsrud University Museum of Bergen. The Norwegian Biodiversity Information Centre provided financial support for the work performed by Anne Helene Tandberg. We thank Prof. Wim Vader and Dr. Ania Jażdżewska for their critical comments to an earlier version of this manuscript.