Dendronotus robustus is often considered a circumpolar and deep-water species (Robilliard, 1970; Ekimova et al., 2015). However, based on the combined distribution of all specimens it has supposedly an extremely broad range encompassing the whole Arctic fringe of Eurasia and North America. This distributional hypothesis has never been tested using modern molecular approaches. Remarkably, the type locality of Dendronotus robustus is at an intertidal location at Grand Manan Island in the Canadian Eastern Atlantic (Verrill, 1870), whereas most specimens of supposed D. robustus come from considerably deeper waters, usually greater than 100 m. In spite of such obvious ecological contradictions, to assess a correct taxonomic name for the Swedish Gullmar Fjord population, we investigated both very shallow water populations previously identified as D. robustus from the Barents Sea and compared them with the deep-water specimens from almost the whole geographic range from North American Eastern Atlantic waters to the Arctic Laptev Sea.
Integration of molecular and morphological datanext section
We discovered significant molecular differences between shallow-water and deep-water specimens of putative D. robustus. These molecular differences agree with the disparity in the colour features in the first description of true shallow-water D. robustus from the intertidal (Verrill, 1870) and the deep-water specimens. Verrill (1870: 406) described the colour of D. robustus as “pale greyish, thickly sprinkled with small yellow spots”, whereas the colour of the deep-water Norwegian specimens, previously described by G.O. Sars (1878: 315) from 109 –183 m under the name of Dendronotus velifer G.O. Sars, 1878, is a “bright reddish with varied white spots” (“Color læte rufescens, punctis albidis variegates”).
As shown in the present study, such molecular and morphological differences between shallow-water and deep-water specimens persist over a very large biogeographic range, from the American Eastern Atlantic to the Arctic Laptev Sea. Significantly, a molecular and morphological data set has recently become available for a deep-water specimen from off Newfoundland (Valdés et al., 2017). This specimen matches, both its external features (i.e. rather bright and uniform reddish colour) and its molecular data, with our deep-water specimens from about 100 m and deeper from the Barents, Kara and Laptev seas. It does not match our shallow water Barents Sea specimens, which represent a different species with more greyish to brownish-yellowish colour. Although these differences are somewhat subtle, the greyish colour mentioned in the first description of true D. robustus from the Canadian Atlantic fits with the shallow-water species from the Barents Sea (Figure 3a, b, d, e), and not with the more uniformly red colour of the deep sea D. velifer from the various Arctic seas, including the Laptev and Kara seas (Figure 2a, b, h-j).
Therefore, both the morphological and the molecular data clearly suggest that two species are actually confused under the name D. robustus in the North Atlantic Ocean – one lives in shallow water from the intertidal to approximately 20 m depth and the other lives deeper, at depths of 50–300 m, most commonly at 80-150 m. The integrative, molecular and morphological data unambiguously suggest that the shallow-water and the deep-water specimens belong to different species, and therefore we hereby restore the name Dendronotus velifer G.O. Sars, 1878 for the deep-water species and retain D. robustus for the shallow-water species. Both species share a similar body shape and a soft-bottom habitat, usually with an almost omnivorous diet (including cnidarians, which is typical of other dendronotoids, but also polychaetes and even sunken terrestrial beetles; see Roginskaya, 1990).
Statistical analysis of the bathymetric distribution
Based on robust molecular data to distinguish between the shallow and deep species from other parts of the North Atlantic Ocean (Figure 4a, b), the historical specimens from Swedish Gullmar Fjord most probably belong to the deep-water species D. velifer and not to the shallow water D. robustus. This is because most of the specimens from the Gullmar Fjord were found in depths of 100–118 m, and never shallower than 53 m. This agrees with the general upper bathymetrical limit for D. velifer at 40–60 m. At some localities in the Artic seas where the layer of cold water can occur close to the surface, e.g. in the Kara Sea and the White Sea, D. velifer can be found in shallow depths, around 15–20 m, thus overlapping with the potential range of D. robustus. However, it is important to note that the true D. robustus was never found in Kara and White seas, except for the subarctic Barents Sea, which is influenced by the Gulfstream. Furthermore, in the Barents Sea D. robustus apparently never goes down to the shallowest depth (43 m) reported here for D. velifer. Thus, despite the possibility that the lower bathymetric limit of D. robustus might hypothetically overlap with the upper bathymetric limit of D. velifer at a general scale, these two species are unlikely to occur syntopically. Also, for the correct evaluation of the taxonomic placement of the Gullmar Fjord Dendronotus specimens, it is crucial that despite specimens of D. velifer (having general preference for deeper waters) in some localities are able to appear much shallower (for example following by low temperatures in shallow waters in many Arctic regions, e.g. in the Kara and White seas), D. robustus has never been recorded at depths greater than 20 m. Thus, along with morphological arguments, the results of the statistical analysis of the bathymetric distribution (Figure 4c) strongly support the hypothesis that it was D. velifer that occurred in the Gullmar Fjord and the southern Norwegian localities.
Further support for the hypothesis that the Swedish deep-water specimens belong to D. velifer is their reddish colour (Odhner, 1907: 19). However, in the same publication Odhner (1907) listed D. velifer as a synonym of D. robustus, without any specific discussion on the issue, probably due to the similarity of the broad body in both species. But Odhner evidently omitted noting the differences in colour (reddish versus greyish-yellowish colour), and importantly also the considerable differences in bathymetric distribution (100–300 m versus intertidal), and hence, in ecological patterns. Previously, Bergh (1900) had also uncritically synonymised D. robustus and D. velifer under the former name. Furthermore, Odhner subsequently reported D. velifer, mostly from some deep Norwegian fjords (e.g. Sandnessjøen, Nordland fylke, June 1938, coll. O. Björlykke, 200 m, see Odhner 1939, also Odhner, 1922, 1926), under the name D. robustus. Thus, it was accepted by most of the subsequent researchers and incorporated in regional lists and identification keys (e.g. Roginskaya, 1987; Martynov and Korshunova, 2011; Ekimova et al., 2015). Apart from the differences in coloration (which compared to some other Dendronotus species actually appear to be more stable in this species pair, but still may represent some difficulties for identification in the field) and considerable differences in the molecular data (Figure 4a, b), D. robustus and D. velifer can also be distinguished by radular features (Appendix; Figs. 1 2-3). In particular, specimens of D. robustus prove to have higher central (rachidian) teeth (Figure 3c, f) and considerably more denticulated lateral teeth, especially in smaller specimens (Figure 2c, f, k). Dendronotus velifer in contrast possesses lower central teeth and less denticulated lateral teeth. The lateral teeth in larger specimens of D. velifer are almost smooth with only traces of reduced denticles on some of them (Figure 1e), whereas smaller specimens of D. velifer may possess more denticulated teeth, which are considerably less denticulated (Figure 1j, k) than in D. robustus of comparable size (Figure 3g). These data on the Swedish Gullmar specimens are remarkably consistent with the original description of D. velifer by G.O. Sars (1878: 315), wherein he mentions of the lateral teeth: “uncini utrinqve 15, læves, vel vestigium modo indistinctum denticulorum hic et illic exhibentes” (= 15 lateral teeth, smooth with indistinct rudimentary denticles on the lateral teeth).
Dendronotus velifer also appears to have fewer dorsolateral appendages than D. robustus: even in quite a large specimen from the Laptev Sea (49 mm) there are only five pairs of dorsolateral appendages (Figure 2a, b), and likewise in specimens from the Gullmar Fjord, including a small specimen of 13.5 mm length (Figure 1a–h). In contrast, even in a subadult (19 mm) specimen of D. robustus (Figure 3d, e) from the Barents Sea, there are six dorsolateral processes. In the larger (35 mm) D. robustus from the Barents Sea, the number of dorsolateral appendages is seven. These data partly agree with the original descriptions of D. robustus and D. velifer: Verrill (1870) indicated six pairs of dorsolateral appendages in a 50 mm long specimen of D. robustus, whereas G.O. Sars mentioned five to six dorsolateral appendages for several specimens with a maximum length of 90 mm. According to these and our data, it is likely that D. velifer usually has no more than five dorsolateral appendages compared to the usual six appendages of D. robustus, even found in small specimens. As an exception, large specimens of D. velifer may have an additional sixth appendage, whereas D. robustus with half the size of the largest D. velifer possess up to at least seven pairs. Furthermore, the frontal digitate appendages (processes) on the oral veil are relatively much longer in D. robustus (Figure 3a-d) than in D. velifer (Figure 2a, b, h-j). Such relative differences, after additional testing, may prove to be reliable morphological features to distinguish D. velifer and D. robustus.
Dendronotus velifer differs from the three other congeneric species that can occur in the same geographic area (i.e. D. frondosus (Ascanius, 1774), D. lacteus (W. Thompson, 1840), and D. niveus Ekimova et al., 2015) by its wedge-shaped body, which is considerably expanded in comparison to the more slender body that is only slightly expanded anteriorly in the other three species. Furthermore, D. velifer has a considerably broader oral veil and it lacks lateral papillae on the rhinophoral sheaths, compared to the three congeners. Internally these three species also differ from D. velifer by their radular characters. Dendronotus velifer differs from a tropical Pacific Ocean species with a similarly wide anterior body, D. patricki Stout et al., 2011, by its reddish colour with scattered white dots and stripes (D. patricki is uniformly pinkish to reddish brown, without any white spots on the dorsum, except for the tips of the appendages) and D. velifer also differs from D. patricki by the less protracted cusp of the central tooth, more numerous lateral teeth per row, and by the presence of the reduced denticles on the lateral teeth (Figure 1e, j, k). Along the Swedish western coast D. lacteus and D. frondosus are common in the subtidal zone from the surface down to 30 m depth. None of these species have been confidently recorded from the deep section of the Gullmar Fjord, although the known bathymetric preferences of D. lacteus and D. niveus potentially allow these species to inhabit the deeper parts of the Gullmar Fjord. Should these species be found sympatrically with D. velifer in the Gullmar Fjord, it will be easy to distinguish them by body shape, colour and the radula. So far, true D. robustus has never been positively reported from the shallow areas of Norway or Sweden. Material from the Norwegian Trondheim Fjord, previously identified as D. robustus by Friele and Grieg (1901), was examined by Odhner and revised as D. frondosus.
Species differentiation, bathymetric distribution and biogeographic range
As shown above, the molecular analysis confirms that D. robustus and D. velifer are separate species, and that they are clearly differentiated by morphology. The bathymetric analysis also shows a clear separation in depth preference. The historical specimens from Gullmar Fjord have a morphology that is consistent with the recently collected Arctic D. velifer specimens and they show a clear preference to the greater depths around 100 m. Thus, these specimens can be confidently identified as D. velifer. A further finding relevant for this case, is the recent report of a single specimen identified as D. “robustus” collected from around 200 m depth off the North Atlantic coast of North America (Valdés et al., 2017) that fully agrees with D. velifer according to our molecular analysis. Therefore, it is possible that D. velifer might occur in more southward locations like the Gullmar Fjord if temperature and other ecological conditions permit. The isolated population of D. velifer in the Gullmar Fjord could therefore possibly have been a relict population from an earlier glacial period with colder water. At that time the species could have had a more southerly distribution.
The maximum depth of the Gullmar Fjord is about 118 m and many of the D. velifer specimens were taken from the deepest section of the fjord. The deeper parts of the fjord (90–118 m), inside the sill, have fully saline water and naturally stable conditions (Svansson, 1975; Filipsson et al., 2005). In the present study it is statistically proven (Figure 4c) that D. velifer differs significantly from D. robustus by its preference for greater depths and hence for lower seawater temperatures. Therefore, it can be supposed that small, statistically insignificant differences in the depth preferences between the Arctic specimens of D. velifer (Figure 4c) and the specimens from the Gullmar Fjord (Figure 4c) are due only to the depth limit of the Fjord. If the Gullmar Fjord were deeper, D. velifer would extend to depths of at least 200 m, and beyond. Two historical findings of single specimens in the fjords of Norway – the Oslo Fjord and the Hardanger Fjord – provide a link from the Gullmar Fjord in Sweden to the northwestern Norwegian coast, and further to the main range of D. velifer through almost all of the Arctic from the Barents to the Laptev Sea and further, to the Bering Strait.
Implications for nature conservation
The Gullmar Fjord has been impacted by bottom trawling and other disturbances since the 1900s. At about the midpoint of the fjord, close to the bay of Skår, one of Sweden’s busiest ferry services has been running since the late 1960s, operating the largest car-driven car ferries in the country. During the day these ferries ply back and forth across the fjord every 20 min. This causes constant turbulence of the bottom sediments on the slopes and, consequently has at least locally, a considerable impact on the fjord’s ecosystem. This impact, together with bottom trawling for shrimps in the habitat for D. velifer might have caused the decline and local extinction of the species. Furthermore, at some periods during later part of the 1970s and early 1980s there were some periods of low oxygen levels in the deep trench of the fjord, which must have added stress to the organisms living there. The collections of D. velifer from the 1940s were in the innermost and relatively shallow sections of the fjord, in which the bottom environment could have been comparatively less affected, at least for some time, and be more stable. It is unknown if such a stability still exists in places, which may imply that a population of D. velifer could still be present. This would then be in parallell with that of the long-absent aeolid nudibranch Flabellina borealis (Odhner, 1922), which was recently rediscovered in the fjord (Lundin and Malmberg, 2016).
The two historical records of single specimens of D. velifer in the Oslo Fjord in southeastern Norway and in the Hardanger Fjord on the southwest may represent a link from the Gullmar Fjord in Sweden to the northwestern Norwegian coast. The Bunnfiorden area has a threshold at 60 m, and an inner deep trench that goes down to 150 m. It is adjacent to parts of the city of Oslo and its southeasters suburbs. The water in the deep trench is of poor quality, hence we can hardly expect any D. velifer to have survived here since the early 1900s. The locality at Sunde, in the Hardanger Fjord on the southwestern coast of Norway is close to the mouth of the fjord, and thus it has better environmental conditions.
Although we cannot report living specimens of Dendronotus velifer from the Gullmar Fjord now, this study is the first that summarises all findings of this species and the first that uses morphological, molecular and ecological evidence to assess the taxonomic position of the predominantly Arctic D. velifer. Such an integrative assessment is important for conservation since there is an ongoing discussion on establishing a national marine park within the Gullmar Fjord. The relatively recent record of D. velifer in the Gullmar Fjord would certainly strengthen the arguments in favour of a marine park to protect the whole area.
Importance of natural history collections
The present study highlights the necessity of close linkages between historical museum specimens and actual biological patterns and processes including refinement of the taxonomic placement and biogeographic ranges. Therefore, it further contributes to a general recognition of taxonomy as a central biological discipline with high integrative potential for multipurpose applications. This is not the first case in which a study of historical specimens in museum collections can be applied to retrieve past species records and to reconstruct historical base lines regarding the distribution ranges of species, which can produce relevant information for the conservation of these species and their habitats (e.g. Shaffer et al., 1998; Pyke and Ehrlich, 2010; Rainbow, 2009; Drew, 2011; Ellis et al., 2011; Hoeksema et al., 2011; Lips, 2011; Hoeksema, 2015). The historical records of the present study should stimulate researchers to survey Gullmar Fjord and other localities for the present occurrence of D. velifer and for additional rarely encountered species that may be locally threatened.