A synoptic review of the family Dendronotidae (Mollusca: Nudibranchia): a multilevel organismal diversity approach

A synoptic review of the family Dendronotidae is presented based on morphological and molecular data. Three genera are recognized: Dendronotus, Pseudobornella, and Cabangus gen. nov. Two new Dendronotus species are described, D. yrjargul sp. nov. and D. nordenskioeldi sp. nov., which reveal fine-scale differences. Dendronotus yrjargul sp. nov. from mid-Norway and the Arctic regions is a sister species to the North Pacific D. kalikal. These two species are showing clear morphological and ontogenetic differences but are close in genetic distance. In contrast, Dendronotus nordenskioeldi sp. nov. from the Laptev Sea is externally similar to the white morphs of D. lacteus or D. frondosus, but according to the molecular data and radular Contributions to Zoology (2020) 1-61


Introduction
The evolutionary process is traditionally considered to be continuous (Darwin, 1859) and this concept prevails despite many years of controversial discussions (Eldredge & Gould, 1972;Mayr, 1982;Mallet et al., 2009;Coates et al., 2018). However, the species itself is supposed to be an unavoidably separate unit according to the nomenclature of current taxonomic practice, which in zoology is governed by the ICZN (1999). This is one of the greatest contradictions between the evolutionary processes and taxonomic classi cation, which has not been solved satisfactorily (Zachos, 2018). In many cases it has been shown that it is very di cult to assign a particular specimen to a particular species. Until now there are no unequivocal and universal criteria for species delimitations when the underlying evolutionary processes and phylogenies are taken into consideration (e.g., Zachos et al., 2013;González et al., 2018;Stanton et al., 2019;Platania et al., 2020), and even for the existence of species at all (Ereshefsky, 1998;Willis, 2017;Reydon & Kunz, 2019). At the conceptual level, more than 26 di ferent species concepts have been suggested (Zachos, 2016) but at a practical level even the most widely acclaimed criterion of reproductive isolation (Mayr, 1969(Mayr, , 1982 is not robust as has been shown on numerous occasions (e.g., McDonald and Koehn, 1988;Mallet, 2005;Kijewski et al., 2006;Harrison & Larson, 2014;Jančúchová-Lásková et al., 2015;Crossman et al., 2016;Fraïsse et al., 2016;Potkamp & Fransen, 2019).
With the dawn of the molecular era in taxonomy, the problem of hidden diversity emerged when molecularly divergent groups of specimens were discovered within morphologically highly similar species as de ned in the traditional way (Bickford, 2007;Korshunova et al., 2017aKorshunova et al., , 2019a. The discordant development of morphological structures and genetic apparatus per se is also a sign of evolutionary processes in progress. However, it is still possible to separate such cases by using molecular data and if there are extensive genetic di ferences, we may assume that evolutionary divergence has occurred among such species. Relatively rarerely noticed and much more interesting are those cases when despite relatively low genetic di ferences some species exhibit distinct morphological features (Mayer & Helversen, 2001;Wiemers & Fiedler, 2007). The current practice of de ning potential species by using molecular markers depending on various distances and factors (e.g., Ratnasingham & Hebert, 2013) referred to as "operational taxonomic units" ( s) does not consistently correspond to taxa at the species level. There is an unfortunate discordance in how the taxonomic practice currently works because it is based on a classi cation system consisting of rigid names that does not fully re ect information of the underlying evolutionary processes. A solution for the species problem using the separation between metapopulation lineages potentially without any distinguishing characters (de Queiroz, 2006(de Queiroz, , 2007 contradicts available biological evidence, e.g., that even clones show minor but detectable morphological and epigenetic di ferences. Therefore, to avoid an arti cial contrast between "cryptic" and "non-cryptic" species in a classi cation and to encompass various available data, the concept of the multilevel organismal diversity was proposed (Korshunova et al., 2019a). In the present study we apply this approach to the nudibranch family Dendronotidae.
Currently there is an accelerating discovery of ne-scale species-level diversity within the genus Dendronotus. In the last decade, 10 of 12 new Dendronotus spp. were described by our research group using both morphological and molecular data (Martynov 2015aEkimova et al. 2015;Lundin et al., 2017;Korshunova et al. 2016aKorshunova et al. , b, 2017bKorshunova et al. , 2019b. In this study, we present a modern taxonomic synopsis of the family Dendronotidae, which together with molecular phylogenetic analyses demonstrates both lower and higher degrees of evolutionary separations as a reliable example of multilevel organismal diversity. For instance, a newly discovered species from Norway (Dendronotus yrjargul sp. nov.) already has developed de nite signs of morphological and molecular divergence. This highlights the importance of considering broadly ontogenetic data in the framework of multilevel organismal diversity (Korshunova et al., 2019a). We therefore develop a general taxonomic framework for phylogeny and taxonomy of the family Dendronotidae, which for the rst time presents a synoptic review of all currently known species of the genus Dendronotus. Its also con rms the validity of the genus Pseudobornella Baba, 1932 and supports the proposal of the new genus Cabangus (see details in Synopsis). (Ronquist et al., 2012). Four Markov chains were sampled at intervals of 500 generations. Analysis was started with random starting trees and 5 × 10 6 generations. Maximum Likelihood-based phylogeny inference was performed using RAxML 7.2.8 (Stamatakis et al., 2008)) with 1000 bootstrap replicates. Final phylogenetic tree images were rendered in FigTree 1.4.2. Nodes in the phylogenetic trees with Bayesian posterior values > 0.96% and bootstrap values >90% were considered 'highly' supported. Nodes with 0.90-0.95% and 80-89% accordingly were considered 'moderately' supported (lower support values were considered not signi cant).
The program MEGA7 (Kumar et al., 2016) was used to calculate the uncorrected p-distances between all the sequences as well as mean intragroup and intergroup genetic distances. Automatic Barcode Gap Discovery ( ) (Puillandre et al., 2012) (https://bioinfo.mnhn.fr/abi/public/abgd/abgdweb.html) was used to estimate Dendronotus species divergence. Alignment from the marker were submitted and processed in using the Jukes-Cantor (JC69) and Kimura (K80) models with the following settings: a prior for the maximum value of intraspeci c divergence between 0.001 and 0.1, 10 recursive steps within the primary partitions de ned by the rst estimated gap, and a gap width of 1.
A multi-locus species delimitation analysis was performed using BPP v.3.1 (Yang 2015) for , 16S, and 28S genes. Two analyses (A and B) with di ferent population size (θs) and divergence time (τ0) priors were preformed, using the same settings and priors as in Martinsson & Erséus (2018) (X: θ 2,400, τ0 2,200; Y: θ 2,1000, τ0 2,200). All analyses were performed three times to con rm consistency between runs. We considered species delimited with a ≥ 0.90 in all analyses to be well supported. For clusters with a < 0.90, we accepted the best-supported more inclusive species. placed P. orientalis ( = 0.99) as sister to the genus Dendronotus. The H3 single tree shows polyphyly in genus Dendronotus if Cabangus regius would be classi ed as Dendronotus. For P. orientalis, the minimal mean intergroup genetic distance was found between P. orientalis and D. velifer (15.6%), and maximal between P. orientalis and D. lacteus (18.3%). For C. regius, the minimal mean intergroup genetic distance was found between C. regius and D. kalikal (16.2%), and the maximal between C. regius and D. frondosus (18.4%). The mean intergroup genetic distance within the clade including D. robustus, D. patricki and D. velifer ranged from 7.2 to 16.8%. The mean intergroup genetic distance within the clade "other Dendronotus" ranged between 2.1 and 14.2% (Appendix, table A2).
To de ne species, we used an integrative approach, including phylogenetic tree topologies, and analyses, pairwise distances as well as the haplotype network for the molecular data reconstructed using Population Analysis with Reticulate Trees (PopART).
To delimit the species based on single locus sequence data ( ), species delimitation was performed with the tool. In the full data set (88 Dendronotus specimens), results from initial partitions show 20 potential species: 16 reputed Dendronotus species + D. elegans from the type geographic region (North Atlantic), combined together with D. elegans ("as D. niveus") from the Arctic region + D. rufus, combined with D. lacteus + D. nordenskioeldi sp. nov. + D. yrjargul sp. nov., combined with D. kalikal. Generally, recursive partitions have more groups than initial partitions that are typically stable. Results from recursive partitions show 22 potential species, corresponding the potential Dendronotus species clades on the phylogenetic tree ( g. 1): 16 reputed Dendronotus species in additon to separate clades for D. nordenskioeldi sp. nov., D. yrjargul 1 Phylogenetic relationships of the species of the genus Dendronotus based on + 16S + 28S + H3 concatenated dataset inferred by Bayesian Inference ( ). Numbers represent posterior probabilities from and bootstrap values for Maximum Likelihood. Some branches are collapsed at species level. Dendronotus elegans from the type region (North Atlantic) are highlighted in dark purple. Abbreviations: h -holotype; p -paratype. Summary of species delimitation results from analyses for the dataset is shown.
Analyzing multi-locus genomic sequence data under the multispecies coalescent model was conducted for D. elegans (2 specimens), former D. "niveus" (16 specimens), D. yrjargul sp. nov (11 specimens). and D. kalikal (10 specimens). In analysis A, the three species model (D. elegans combined with the former "D. niveus", separate D. yrjargul sp. nov. and separate D. kalikal) is preferred with a mean of 0.94. In analysis B, the same three species model is also preferred with a higher mean of 0.99.
Results obtained by PopART for the and 16S genes showed a network of haplotypes that clearly clustered into three groups coincident with D. elegans combined with former "D. niveus ", D. yrjargul sp. nov. and D. kalikal ( g. 3).
Regarding the supposedly fast-evolving marker, genetic distance values between D. elegans from Newfoundland and D. elegans from Greenland is 0.61%. Maximal intragroup distance among specimens formerly identi ed as "D. niveus" from the White, Barents, Kara, and Laptev seas is 1.42%, whereas distances between the D. elegans and specimens formerly identi ed as "D. niveus" range from 0 to 0.94%, indicating without doubt that D. elegans and former "D. niveus" are the same species. Mean intragroup distance within the D. elegans clade is 0.59% whereas minimal intergroup distance (7.6%) was found between D. elegans and D. dalli (Appendix, table A2).
The molecular phylogenetic results conrm the morphological analysis data (see Systematic section and gs 1-8). Thus, D. elegans and "D. niveus" are the same species, whereas D. kalikal, D. yrjargul sp. nov, and D. nordenskioeldi sp. nov. are three separate species in the genus Dendronotus.

(cont.)
appendages with relatively short primary stalk, strongly branched secondary branches and elongate tertiary branches ( g. 4A-D). Basal colour is semitransclucent white to greyish, with characteristic two thin wavy subparallel brownish-orange to brownish lines running from the head nearly to the tail. There are no specimens with brownish basal colour. Upper part of the dorsolateral appendages covered with bright golden yellow to yellowish pigment. Foot bordered with thin brownish-orange line. Dorsal processes of jaws inclined posteriorly at approximately 40-48° to the longitudinal axis of the jaw body and 0.4 of its length ( g. 4E-I). Masticatory border of jaw with evident ridge-like denticles ( g. 4H, I). Radular formula: 38-50 × 0-9.1.9-0. Central (rachidian) tooth moderately denticulated and bearing up to 16 distinct denticles ( g. 4J-M). Denticles with deep furrows. Lateral teeth elongate, slightly curved, bearing up to 7 denticles ( g. 4J-N). Ampulla thickened kidney-shaped ( g. 6A). Prostate discoidal, consisting of up to 50 alveolar glands. Distal part of vas deferens moderate in length, transitioning to long, thick, curved copulatory organ. Bursa copulatrix large, rounded, with seminal receptaculum placed distally.
Distribution. From Norwegian Sea to Kara Sea.
Habitat and ecological observations. Den dronotus yrjargul sp. nov. is closely associated with the hydroid Abietinaria fusca (Johnston, 1847) (see Vervoort, 1942) which forms dark brown colonies that resemble small dry r branches (personal observations of VG and KJ). The hydroid usually inhabits somewhat deeper water, but currents and other conditions in the Norwegian waters make it reachable by scuba diving at circa 20 m.
We have only rarely observed the new species on other hydroids. Dendronotus yrjargul sp. nov. appears quite abundant in this location, up to 10 individuals per m 2 . The new species appears to have a one-year life cycle. Mating and egg laying occurs around March, and the large adult individuals eventually disappear throughout the spring, as does the hydroid they live on. During the summer, small individuals begin to reappear, along with the new growing phase of A. fusca. The new generation of D. yrjargul sp. nov. grow throughout autumn and winter.
Remarks. Externally, D. yrjargul sp. nov. readily di fers from all known species of the genus Dendronotus (synopsis given below), including the sister species from the North Paci c, D. kalikal (description for comparison is given below), by a combination of o f-white semitransparent body with two thin dorsal subparallel brownish lines and dorso-lateral appendages golden-yellow to yellowish ( g. 4A-D). Instead, D. kalikal invariably possesses a basal body colour covered with numerous brownish speckles and small spots, which make subparallel dorsal lines considerably blurred and indistinct ( g. 4O). Importantly, we investigated more than ten D. kalikal specimens from very distantly placed localities in the Bering Strait, Kamchatka and Kurile Islands ( g. 3A) and not one specimen possessed a mostly whitish semitransparent body with distinct thin subparallel dorsal lines, as in D. yrjargul sp. nov. ( g. 4A-D). The whitish colouration in D. kalikal consists of a background colour represented by areas in between the numerous basal brownish spots and stripes. Furthermore, considerable di ferences in the reproductive system and the maturation process between D. yrjargul sp. nov. and D. kalikal were revealed in course of the present study. All studied specimens of D. kalikal have an elongated bursa copulatrix ( g. 6B), whereas in all studied specimens of D. yrjargul sp. nov. ( g. 6A) the bursa copulatrix is rounded.
Furthermore, we discovered that there is considerable di ference in the patterns of maturation (and hence, ontogenetic patterns) between D. yrjargul sp. nov. and D. kalikal. Dendronotus yrjargul sp. nov. reaches a big size of 95 mm (live) in the adult state, and in specimens of 20-30 mm (live) the reproductive system was still very weakly developed. Instead, D. kalikal has a maximum length of 17 mm when preserved and already at lengths of 7-10 mm live specimens show a well-developed reproductive system. The molecular di ferences between D. yrjargul sp. nov. and D. kalikal are as follows. Regarding the marker, the maximal intragroup distance within D. kalikal is 1.74% and within D. yrjargul sp. nov. it is 0.85% (mean intragroup distances are 0.80% and 0.39%). The mean intergroup distance between D. kalikal and D. yrjargul sp. nov. groups is 2.1% (Appendix, Description. Body relatively narrow, preserved holotype 19 mm length ( g. 5). Five to six main branched appendages on oral veil. Five appendages on rhinophoral stalks; 12-14 rhinophoral lamellae. Lateral papilla on rhinophoral sheath present. Five pairs of dorsolateral appendages. Circa 20 lip papillae. Dorsolateral appendages with relatively short primary stalk, moderately branched secondary branches and tertiary branches ( g. 5A, B). Basal colour is semitranslucent white. No additional pigment. Dorsal processes of jaws inclined posteriorly at approximately 53° to the longitudinal axis of the jaw body and 0.42 of its length ( g. 5D). Masticatory border of jaw with ne denticles ( g. 5E). Radular formula: 43 × 3-11.1.11-3. Central (rachidian) tooth bears very weak denticulation that are di cult to separate into individual denticles (up to 20 and more variously varied denticles at least) ( g. 5F, G). Denticles without deep furrows. Lateral teeth elongate, slightly curved, bearing up to seven denticles (+ 2-3 reduced denticles may present) ( g. 4F-H). Ampulla thickened, kidney-shaped. Prostate discoidal, consisting of up to 20 alveolar glands. Distal part of vas deferens moderate in length, transitioning to long, thick, curved copulatory organ. Bursa copulatrix large, distinctly elongated, with seminal receptaculum placed distally.
Distribution. So far known only from the Laptev Sea.  (Ascanius, 1774) Diagnosis. Dorsolateral appendages with distinct tertiary branches. Oral veil with branched appendages. Jaws commonly with set of tightly packed ridge-like semi-discoid structures and also with denticles. Radula with at least more than eight lateral teeth in adult specimens. Central teeth with cusp integrated within lateral denticles (or within teeth body if denticles absent), the central cusp is not protruding. Denticle-bear part of central teeth separated from lateral sides by usually distinct shoulder, thus central teeth with squarish base and triangular top. Prostate with evident, conspicous alveolar glands, commonly represent by a discoid structure (varies from a attened disk to a compressed spherical structure), or very rarely non-discoid, linear (in this case alveolar glands also present). Copulative organ basically conical, with various ratios of length and width and degrees of apical tapering.
Remarks. The genus Dendronotus is restricted here since some species have been moved to other genera. See discussions and remarks under the genera Cabangus gen. nov. and Pseudobornella.
Species composition. Diagnoses and remarks for 27 included species are listed below. For each species, an extended diagnosis is provided when possible. The ne-scale diagnostics of the genus Dendronotus is a complex task since variation patterns of habitus and anatomical characters overlap in various species. By a careful listing it is possible to step-by-step ful ll the task and present a mutilelevel organismal diversity, though not in a strictly de ned form, as was targeted previously (e.g., Odhner, 1936;Robilliard, 1970), but in a complex form that is best suited to the evolutionary mosaic in character evolution. There are also still uncertanties regarding some particular characters as well as previous overinterpretations. First of all, previously the pattern of branching in the dorsolateral appendages was claimed to be of considerable taxonomic importance (Ekimova et al., 2015: Fig. 8). In this respect, it is necessary to state, that while some trends exist in a few species, importance of these characters were considerably overestimated partly due to the use of preserved specimens in the analysis. This can be illustrated by the D. kamchaticus case, when according to Ekimova et al. (2015: Fig. 8D), D. kamchaticus has bulbous short branches that di fer considerably from those in any other Dendronotus species. This resulted from the use of preserved specimens with considerably contracted branches, whereas the living adult specimens of D. kamchaticus have elongated branches of dorsolateral appendages, which are not fundamentally di ferent from the type species of the genus, D. frondosus (Korshunova et al., 2016a). The description of D. zakuro (similar to D. kamchaticus by radula morphology) is based on recently collected specimens with both shorter "bulbous" (see Martynov et al., 2020a: Fig. 3A) and elongate dorsolateral appendages (Martynov et al., 2020a: Fig. 3D, E), which is due to di ferences in physiological conditions. The preserved specimen of D. zakuro (see Martynov et al., 2020a: Fig. 3C) demonstrates very similar "bulbous" branches, which were incorrectly depicted as speci c for D. kamchaticus in its original description (Ekimova et al., 2015: Fig. 8D) (see details in Synopsis under D. kamchaticus). Dendronotus lacteus apparently has shorter tertiary branches, and a "curly" appearance of the dorsolateral appendages. However, variation in D. lacteus does not only concern ontogenetic stages, but also occurs among di ferent adult specimens. So, some large specimens of D. lacteus have quite long elongate branches of dorsolateral appendages, similar to those in D. frondosus (Korshunova et al., 2017a). Most species of the genus Dendronotus possess considerably branched dorsolateral appendages with distinct tertiary branches (MacFarland, 1966;Robilliard, 1970;Korshunova et al., 2016aKorshunova et al., , b, 2017aKorshunova et al., , 2019bMartynov et al., 2020a). Unbranched dorsolateral appendages were recently reported only for D. claugei from the deep-sea (Valdes et al., 2018), but due to the most basal position of this species ( g. 1) the generic position of D. claugei needs to be evaluated with more data. For another deep-sea species, D. comteti, the unbranched dorsolateral appendages were also indicated (Valdes et al., 2018: 413), however, according to the original description of D. comteti, it possesses both branched and unbranched dorsolateral appendages (Valdes & Bouchet, 1998: 320). Unbranched dorsolateral appendages can occur in posterior parts of some other Dendronotus species (e.g., in D. robilliardi, see Korshunova et al., 2016a). For some species, e.g., for D. dalli in Ekimova et al. (2015: 862), the indication of absent tertiary branches of dorsolateral appendages is incorrect. Dendronotus dalli de nitely has tertiary branches (Robilliard, 1970: 453;present study, g. 7). Instead, in the genus Pseudobornella the general appearance of dorsolateral appendages does not show as branches, but it is ctenidiumlike, because of very indistinct (or truly absent) tertiary branches ( g. 2). More details about the genus Pseudobornella are mentioned after the diagnoses of the species of the genus Dendronotus.
There are also various uncertainties and misinterpretations regarding internal characters that need to be addressed before diagnoses can be provided. For example, Robilliard (1970) suggested to apply several coe cients of jaw proportions. However, it is easy to incorrectly apply respective angle of measurements and receive very di ferent results, as for example in Ekimova et al. (2015) compared to the same species which jaws by Robilliard (1970). As an example, for D. dalli in Ekimova et al. (2015: 863) an angle of dorsal processes to the longitudinal axis of 25° was indicated, whereas in Robilliard (1970: 454) this was 60°.
In the present study we re-checked jaw proportions and presented updated values for the majority of species with su cient available information. This character still needs further testing, but some characteristic patterns for di ferent species are likely to occur. Similarly, the number of the prostatic alveoles once was thought to be a character that well characterizes di ferent species (Robilliard, 1970;Ekimova et al., 2016). However, our previous (Korshunova et al. 2016a, b) and present study show that this character is much more variable, and while there are clear trends in some species, it should generally be used with care. In the species diagnoses below, di ferences in shape, length and proportions of the copulative organ (penis) are indicated. While in some species the di ferences in penial morphology are clear, in other species this can be partly due to di ferent degrees of contraction after preservation. This character needs further investigation.
In the present study we show that the shape and size of the bursa copulatrix, at least among several Dendronotus species (e.g., between D. kalikal and D. yrjargul), are rather constant and can be used as a reliable diagnostic character. However, the exact position of the commonly very small receptaculum seminis can be very di cult to determine, but despite that, it will be given further attention in the diagnoses below.
Di ferent terms were applied for the structures on the masticatory edge within di ferent species of the genus Dendronotus. For example, MacFarland (1966) used "denticles" or "ridge-like" denticles and Robilliard (1970) applied "denticles", whereas Ekimova et al. (2015) used "rodlets". In this respect, the term "rodlets" was used for rod-like structures that constitute the body of jaws in other nudibranch groups. In the genus Dendronotus, the elements on the masticatory edges are usually termed as "denticles" when the bodies of jaws are entire. However, in the genus Dendronotus there is a special set of structures in addition to the denticles that appears as broad semidisks or ridges, which can additionally be covered with smaller spine-or denticle-like structures. It is likely that most species of the genus Dendronotus possess such structures. In addition to the semi-disks/ridges at the masticatory edges, more normally looking denticles can occur. Transitive denticles on masticatory edges were also occasionally reported for juveniles of D. frondosus s.l. (Colgan, 1914;Wägele & Willan, 2000). Therefore, in the present study we use two di ferent terms "ridge-like structures" and "denticles" when characterizing the masticatory edges of Dendronotus species. The few denticles seen in many cases are likely a precursor that further transits into semi-disk structures on the masicatory edges (see e.g., this clearly for D. kalikal, g. 4Q). Because the full row of denticles and/or ridges is not documented for all species due to diculties during preparation, the exact number for these structures can be uncertain. Though whenever possible, we already presented the number of denticles/ridges, for example the recently described Dendronotus species (e.g., Martynov et al., 2020a), in the diagnoses below we do not provide the number because it can be misleading, because only a part of the row was counted in various species. True rodlets can occur in addition, as part of the thin labial cuticle, which is separated from the masticatory edges. Such true rodlets have been well documented for the tropical representatives of the family Dendronotidae (Pola & Stout, 2008), and possibly also occur in many Dendronotus species, since the labial cuticle with rodlets were already described for several representatives (see e.g., MacFarland, 1966). However, these characters remain largely untested by modern methods and we do not include it in the diagnoses below. As an additional complication, denticles and especially ridge-like structures on the masticatory edges can be easily discarded during preparation, and therefore in some previous descriptions it was inappropriately indicated that masticatory edges are devoid of any denticles or ridges. For example, the absence of the denticles on the masticatory edges of D. frondosus were incorrectly mentioned in Ekimova et al. (2015), which was further used as a putative di ference between D. frondosus and D. primorjensis (Ekimova et al., 2016). However, in both species denticles on the masticatory edges are de nitely present (Korshunova et al., 2016a). The same is true also for other species, such as D. kamchaticus and D. kalikal. For D. kamchaticus the absence of any structures on the mascticatory edges was indicated in the original description. For D. kalikal the presence of merely "rodlets" was reported in the original description (Ekimova et al., 2015). However in reality denticles and ridge-like structures are present in both of these species (Korshunova et al., 2016a(Korshunova et al., , 2019b for D. kalikal see g. 4Q). Therefore, when the original descriptions denticles or ridge-like structures was not indicated for species that we could not check, we suppose the presence .
of at least one of these structures and indicate that in the diagnoses. Though a pair of salivary glands indeed inserts into the pharynx on the sides of the oesophagus, the knowledge of these is limited only to a few species (e.g., MacFarland, 1966;Valdes & Bouchet, 1998;Pola & Stout, 2008). The salivary glands are usually not gured in the original descriptions (an exception: Valdes and Bouchet, 1998) and were never consistently used for taxonomic diagnostics. We therefore do not include salivary glands in the diagnoses of the species of the genus Dendronotus.
The radular patterns within the genus Dendronotus still o fer one of the best diagnostic characters (Korshunova et al., 2017a). However in several species groups quite similar patterns of the central tooth can be observed. These similarities occur between groups/clades of closely related species (for example between species that closely related to D. frondosus, e.g., D. primorjensis and D. venustus, between D. lacteus and D. rufus, between D. dalli and D. elegans), but also among apparently more distantly related species such as D. kamchaticus and D. zakuro, and between D. albus and D. robilliardi (see g. 1). Despite these similarities, in a majority of these cases ne-scale di ferences in morphology of radula can be detected. At the same time, di ferent species which are grouped in the same subclade ( g. 1) can demonstrate considerably di ferent radular patterns, as for example the clade consisting of D. jamsteci, D. subramosus and D. albus (Robilliard, 1970;Pola & Stout, 2008;Korshunova et al., 2016a;Martynov et al., 2020a). Furthermore, the radulae of the morphologically very di ferent D. robilliardi and D. gracilis demonstrate similar patterns in the presence of a widened central tooth with numerous small lateral denticles without furrows Korshunova et al. 2016aKorshunova et al. , 2019b. The radula undergoes remarkable transformation during the development from early juvenile to adult. The central tooth in earlier juveniles is low and bears few distinct lateral denticles. The lateral teeth in early juveniles are very few in number (1-3) or absent, showing an indistinct denticulation or they are almost smooth. In individuals approaching the adults stage, the central teeth become elongate-triangular with distinct shoulders, lateral denticles become relatively smaller and numerous or completely reduced. Meanwhile, the lateral teeth increase in number and the denticles become more conspicuous. Since parts of juvenile radula may persist in some adults, this may create a considerable confusion regarding the taxonomic importance of radular characters. For instance, D. dalli usually has a smooth central teeth without denticles in adult stage, whereas for adults of D. dalli also denticles on central teeth were reported, which led to the incorrect synonymyzation of D. dalli with D. frondosus (Odhner, 1936;MacFarland, 1966). This decision was later changed in favour of the validity of D. dalli, but uncertanties over the taxonomic application of the putatively variable radula remained (Robilliard, 1970;Roginskaya, 1987). Such "variations" can be overcome by a consistent application of general ontogenetic principles in taxonomy, using the attribution of a particular set of characters to a particular ontogenetic stage (Martynov et al., 2015b. Thus, the application of a multilevel diagnostic within the genus Dendronotus (one that combines morphological characters of various levels and molecular data) instead of a search for immediate and clearly distinctive di ferences, requires a ne-scale, complex approach. Robilliard, 1972 Dendronotus albopunctatus Robilliard, 1972 MacFarland, 1966: 275-279, Pl. 40, Fig. 1;Pl. 46, Figs 1-4;Pl. 47, Figs 8-11;Pl. 48, Figs 7, 8;Pl. 49, Fig, 5;Pl. 50, Fig. 4;Pl. 51, Figs 6, 7. Dendronotus diversicolor Robilliard, 1970: 470-474, Figs 4-6, 25-28, Pl. 64, Figs 35, 36. Not D. albus sensu Robilliard, 1970: 466-470;Behrens, 1980: 72;Behrens and Hermosillo, 2005: 93;McDonald, 1983: 172;McDonald, 2009: 463 = D. robilliardi Korshunova, Sanamyan, Zimina, Fletcher, Martynov, 2016 Extended diagnosis. Body relatively narrow. Four to ve pairs of branched dorsolateral appendages, digestive gland penetrates only two pairs of dorsolateral appendages. Up to six appendages of oral veil. Three to seven appendages (external longest) of rhinophoral stalks. Lateral papilla of rhinophoral sheaths present. Rhinophores with 12-21 lamellae. Two to eight lip papillae. Basal colour translucent white to violet, dorsolateral appendage variable opaque orange or opaque white, including tip. Dorsal processes of jaws inclined posteriorly at approximately 50-60° to the longitudinal axis of the jaw body and 0.4-0.47 of its length. Masticatory borders apparently with ridge-like structures and denticles. Radula with up to 38 rows of teeth. Central tooth with up to 25 small distinct denticles without furrows. Up to nine lateral teeth with up to 14 denticles. Ampulla voluminous, folded. Bursa copulatrix large, oval. Seminal receptaculum very small and placed distally at a short distance from the vaginal opening. Prostate discoid with about 10-30 alveolar glands. The vas deferens is relatively short, bent, penis short, nearly straight, tapering to a blunt tip. Body length up to 73 mm. Extended diagnosis. Body relatively narrow. Five to six pairs of dorsolateral appendages. Six to eight appendages of oral veil. Five to six appendages (apparently equal in size) of rhinophoral stalks. Lateral papilla of rhinophoral sheaths present. Rhinophores with 15-18 lamellae. Lip papillae 15-25. Basal colour brownish with scattered distinct opaque white dots. Dorsal processes of jaws inclined posteriorly at approximately 55° to the longitudinal axis of the jaw body and 0.47 of its length. Masticatory processes apparently bear indistinct denticles (may possess ridge-like structures). Radula with up to 39 rows of teeth. Central tooth with up to 14 small denticles and reduced furrows. Up to nine lateral teeth with up to nine denticles. Ampulla folded twice. Bursa copulatrix is large and rounded. Seminal receptaculum small placed distally at a moderately short distance from the vaginal opening. Prostate discoid with about 25-30 alveolar glands. The vas deferens is moderate in length, penis long and twisted. Body length up to 19 mm.
Remarks. This recently described species is easily distinguished from its congenerics (Korshunova et al., 2016a). This is a true Arctic species that so far has been recorded only from the Laptev Sea and Kara Sea. Extended diagnosis. Body wide. Six to seven pairs of branched dorsolateral appendages. 10 to 15 appendages of oral veil. Five appendages of rhinophoral stalks. Lateral papilla of rhinophoral sheaths absent. Rhinophores with 15-25 lamellae. Lip papillae 50-70. Basal colour dull reddish-brownish with numerous distinct opaque white dots on notum, tips of lateral appendages, oral appendages. Dorsal processes of jaws inclined posteriorly at approximately 52° to the longitudinal axis of the jaw body and 0.56 of its length. Masticatory processes apparently bear indistinct denticles (may possess ridge-like structures). Radula with up to 41 rows of teeth. Central tooth with high cusp and up to 35 small distinct denticles without furrows. Up to 14 lateral teeth without or with few weak lateral denticles. Ampulla voluminous, with approximately three loops. Bursa copulatrix large, rounded to oval. Seminal receptaculum small placed distally at a short distance from the vaginal opening. Prostate discoid with about 21-25 alveolar glands. The vas deferens is long, penis elongate, narrow, curved. Body length up to 45 mm.

Dendronotus bathyvela
Distribution. Paci c coast of Northern Honshu, Japan, possibly may occur further in NW Paci c.

Bathymetry.
A shelf to deep-sea (bathyal) species, so far recorded at depths of 249-510 m.
Remarks. This species has recently been described by Martynov et al. (2020a). Until recently, only the single species D. robustus was recognized among the wide-bodied, shallow-water dendronotids of the North Atlantic and the North Paci c shelf (Ekimova et al., 2015). However, most recent data showed that "D. robustus" is a species complex, which was already separated into D. robustus s.str., D. velifer and D. bathyvela (Lundin et al., 2017;Martynov et al., 2020a).

Dendronotus claguei Valdés, Lundsten & Wilson, 2018
Dendronotus claguei Valdés, Lundsten, Wilson, 2018: 412-415, Figs 3B, 9. Diagnosis (original description). Body very narrow. Four to six pairs of dorsolateral appendages (reported as unbranched). Four appendages of oral veil. Four appendages of rhinophoral stalks (innermost appendages from each side longest). Lateral papilla of rhinophoral sheaths absent. Rhinophores with 11 lamellae. Lip papillae not reported in original description. Basal colour uniformly translucent white. Masticatory processes of jaws with irregular denticles and possibly with ridge-like structures. Radula with up to 32 rows of teeth. Central tooth with deep furrows and with up to 13 distinct denticles. Up to eight lateral teeth with up to ten denticles. Ampulla with three loops. Prostate compact with about nine alveolar glands. Bursa copulatrix large, rounded. Seminal receptaculum large placed distally at considerable distance from the vaginal opening. The vas deferens is moderate in length, penis relatively long, narrow. Body length up to 18 mm.
Distribution. Alarcón Rise, Gulf of California, Mexico, NE Paci c Ocean.

Bathymetry. 2369 m depth.
Remarks. For this recently described deep-sea species there is no additional morphological information and molecular data is very limited. In particular, the dorsolateral appendages need a further detailed study.

Dendronotus comteti Valdés & Bouchet, 1998
Dendronotus comteti Valdés and Bouchet, 1998: 320-323, Figs 1-3. Diagnosis (original description). Body narrow. Two or four pairs of partially branched, partially unbranhed dorsolateral appendages. Four to six appendages of oral veil. Four appendages (similar in size) of rhinophoral stalks. Rhinophores with seven to eight lamellae. Lateral rhinophoral papilla possibly absent. Lip papillae probably absent. Basal colour uniformly pale cream (preserved). Jaws reported as devoid denticles. Radula with up to 28 rows of teeth. Central tooth with deep furrows and with up to ten distinct denticles. Up to six lateral teeth with up to ve denticles. The reproductive system was not studied in the original description. Body length up to 5.5 mm (preserved).
Distribution. Middle Atlantic, Lucky Strike area in the Mid-Atlantic Ridge.

Bathymetry. 1685 m depth.
Remarks. This is the only known species that originated from deep-sea hydrothermal habitats, and which was considered as belonging to the genus Dendronotus according to its external and internal characters. Molecular data is necessary to determine its phylogenetic position. Dendronotus frondosus var. dalli Bergh, 1879-Odhner, 1907 Not D. dalli sensu Bergh, 1886, Knipowitsch, 1902, Roginskaya, 1987(= D. elegans Verril, 1880.

Dendronotus dalli Bergh, 1879
Extended diagnosis. Body relatively narrow. Four to eight pairs of branched dorsolateral appendages. Four to ve appendages of oral veil. Four to 12 appendages (equal in size or few posterior ones longer) of rhinophoral stalks. Lateral papilla of rhinophoral sheaths present. Rhinophores with 16-33 lamellae. Lip papillae 15-40 (and more). Basal colour uniform translucent gray, light yellow to orange, brownish or pinkish, with external opaque white pigment on tips of dorsolateral appendages. Dorsal processes of jaws inclined posteriorly at approximately 60° to the longitudinal axis of the jaw body and 0.45 of its length. Masticatory processes apparently bear denticles (may possess ridge-like structures). Radula with up to 51 rows of teeth. Central tooth completely smooth in adults. Up to 16 lateral teeth with up to 19 denticles. Ampulla relatively narrow with several (at least three) loops. Bursa copulatrix very large, oval. Seminal receptaculum small placed distally at a moderately short distance from the vaginal opening. Prostate discoid with up to 50 (and more) alveolar glands. The vas deferens is moderate in length, penis large, wide, with a blunt tip. Body length up to 14 cm.
Distribution. North Paci c (The Sea of Japan in the west and at least Washington State, USA in the east to at least Chuckchii Sea; Russia, in the Arctic (may possibly penetrate further to the west in the Arctic).
Bathymetry. Veri ed records from 5 to 50 m depth.
Remarks. Until recently, the North Atlantic D. elegans (see below) has been confused with the North Paci c D. dalli. The best external feature to distinguish these two closely related species is the presence of an external opaque white pigment on the apical parts of the dorsolateral processes in D. dalli, whereas in D. elegans such a pigment, if present, is internal. These species may co-occur in some Arctic localities, but this needs to be further investigated. For D. dalli an absence of tertiary branches of the dorsolateral appendages was incorrectly mentioned by Ekimova et al. (2015). There is a possibility that light-coloured varieties of D. kamchaticus were depicted under D. dalli, judging from photographs in for example Behrens (1980) (see also Korshunova et al., 2016a). In some studies, Dendronotus dalli was incorrectly considered a synonym of D. frondosus (Odhner, 1936;MacFarland, 1966) due to an underestimation of the importance of ontogenetic radula data (central teeth are denticulated in juveniles and usually smooth in adults).
We have investigated a considerable number of specimens and sequences of D. elegans from the North Atlantic and neighbouring subarctic regions. There are no con rmed records of true D. dalli in the North Atlantic and neighbouring regions. The indication of D. dalli from the North Atlantic (Genbank AF249800) in a  Robilliard, 1970: 452;Bleakney, 1996: 109 (D. elegans incorrectly mentioned as synonym of D. dalli).
Extended diagnosis. Body relatively narrow. Five to ten pairs of branched dorsolateral appendages. Six to nine appendages of oral veil. Four to ve appendages (equal in size or few posterior ones longer) of rhinophoral stalks. Lateral papilla of rhinophoral sheaths present. Rhinophores with 8-23 lamellae. Lip papillae 8-50. Basal colour uniform, pinkish to light orange, with internal opaque white pigment of tips of dorsolateral appendages. Dorsal processes of jaws inclined posteriorly at approximately 40° to the longitudinal axis of the jaw body and 0.43 of its length. Masticatory processes apparently bear denticles (which may possess ridge-like structures). Radula with up to 42 rows of teeth. Central tooth completely smooth in adults. Up to at least 13 lateral teeth with up to eight denticles (can be more). Bursa copulatrix large, oval. Seminal receptaculum small placed distally at a moderately short distance from the vaginal opening. Prostate discoid with up to 60 and more alveolar glands (their number was considerably underestimated in the original description of its new synonym D. niveus (see remarks below). The vas deferens is moderate in length, penis strong, conical. Body length up to 70 mm.
Distribution. North Atlantic (known southern limit -Cape Cod, USA) to the Arctic (including Barents Sea and White Sea, known northeastern limit is the Laptev Sea, Russia).
Remarks. Ekimova et al. (2015) overlooked important details in the original description of D. elegans Verrill, 1880, for which a big animal with uniform salmon-coloured body was used, and a special mentioning was made of the completely smooth central teeth (Verrill, 1880: 385-386). These characters are identical to those of the recently described D. niveus Ekimova et al., 2015. The available molecular data (Valdés et al., 2017; present study, gs 1-3) also show that specimens identi ed as D. niveus from o f the east coast of North America (relatively close to the type locality of D. elegans) are identical to specimens from the type locality of D. niveus in the White Sea. The holotype of D. elegans in the Smithsonian Institution (NMNH, 2020) is not suitable for molecular study. However, a microslide with radula of the holotype is . Dendronotus frondosus sensu Thompson and Brown, 1984 (partim.); not Ascanius, 1774.
Extended diagnosis. Body relatively narrow. Six to nine pairs of branched dorsolateral appendages, including two to three posterior ones. The dorsolateral appendages can generally be described as highly branched, giving this species a u fy appearance. Eight to 15 appendages of oral veil. Five to six appendages (posterior one very long) of rhinophoral stalks. Lateral papilla of rhinophoral sheaths present. Rhinophores with 11-17 lamellae. Lip papillae 20-49. Basal colour reddish-brown with opaque white stripes between dorsolateral processes to uniform light yellow with small brown spots. Dorsal processes of jaws inclined posteriorly at approximately 60° to the longitudinal axis of the jaw body and 0.43 of its length. Masticatory processes apparently with ne denticles. Radula with up to 37 rows of teeth. Central tooth with up to 20 small denticles without furrows, rarely completely smooth. Up to 11 (usually 9-10) lateral teeth with up to seven denticles. Ampulla with at least two loops. Bursa copulatrix large, rounded. Seminal receptaculum small placed distally at a short distance from the vaginal opening. Prostate discoid with a range of 35-40 alveolar glands. The vas deferens is very long, penis relatively narrow, bent. Body length up to 100 mm.
Distribution. NE Atlantic, so far known from the Netherlands, Norway, Sweden and the UK.

Dendronotus frondosus (Ascanius, 1774)
Dendronotus frondosus (Ascanius, 1774) -Odhner, 1936: 1105-1109  Extended diagnosis. Body narrow. Five to six pairs of branched dorsolateral appendages. Four to seven appendages of oral veil. Four to ve appendages (middle and posterior ones can be longer) of rhinophoral stalks. Lateral papilla of rhinophoral sheaths present. Rhinophores with six to 12 lamellae. Lip papillae four to 12. Basal colour brownish to reddish-brown, often with small white and yellow specks, but usually without opaque white stripes between dorsolateral processes to completely white translucent specimens. Dorsal processes of jaws inclined posteriorly at approximately 60° to the longitudinal axis of the jaw body and 0.4 of its length. Masticatory processes apparently bear ridgelike structures and denticles. Radula with up to 42 rows of teeth. Central tooth with deep furrows and with up to 14 (common range 8-12) distinct denticles. Up to 10 (usually up to eight) lateral teeth with up to seven denticles. Ampulla voluminous, folded. Bursa copulatrix large, oval to rounded. Seminal receptaculum small placed distally at a moderately short distance from the vaginal opening. Prostate discoid with range 16-30 alveolar glands. The vas deferens is moderate in length, penis relatively long, curved. Common body length no more than 50 mm.
Distribution. North Atlantic, both east and west parts, not distributed to the Arctic further than the easternmost border of the Barents Sea.
Remarks. Three common North Atlantic shallow water species, namely D. europaeus Korshunova et al., 2017b, D. frondosus (Ascanius, 1774), and D. lacteus, show an extremely similar range of habitus variation (see Korshunova et al., 2017b) and have been confused in older literature. Therefore, the veri ed distribution of D. frondosus can only be assessed using recent data (Korshunova et al., 2017b) that covers at least the North East Atlantic from Norway and also the Barents Sea and the White Sea to possibly France/ northern Spain. The range of D. frondosus does not extend further than the the true Arctic. Also, several deep-water records of D. frondosus (e.g., Odhner, 1939;Swennen, 1961;Thompson & Brown, 1984) most likely refer to D. lacteus, since according to our data D. frondosus was usually found shallower than 20-30 m depth. We have examined an extensive collection of various Dendronotus from the NW Paci c and NE Paci c and have so far not found any specimen from the North Paci c for which the identity could be con rmed as D. frondosus despite the presence of a few published records (e.g., Ekimova et al., 2016). In this respect, the In this study we also evidently show that the North Paci c D. kalikal and the North Atlantic and Arctic D. yrjargul sp. nov. are invariably placed in two distinct clades ( gs 1-3) with stable distinct di ferences in external and internal morphology ( g. 4) and without any gene ow between each other. Dendronotus kalikal thus does not occur naturally in the North Atlantic. A potential future introduction of some North Paci c species from the "D. frondosus megacomplex" may take place, but so far there are no reliable records of any North Paci c species in the North Atlantic. At least recently, a case of evident anthropogenic transportation of the dendronotid Pseudobornella orientalis was reported from Japan or China to NE Paci c (Agarwal et al., 2017) (for taxonomy of Pseudobornella see below). Thus, most previous records of this species from the Northwest Paci c refer to one of three recently described species, D. kamchaticus Ekimova et al., 2015, D. kalikal Ekimova et al., 2015, or D. primorjensis Martynov et al., 2015, and in the Northeast Paci c, previous records refer to D. venustus Stout et al., 2010or D. kamchaticus Ekimova et al., 2015 (see also Korshunova et al., 2016a, b).  Dendronotus gracilis Baba, 1949: 167, Pl. 35, Fig. 127, text-g. 109;Robilliard, 1970: 461-462;Nakano, 2018: 385. Diagnosis (original description). Body narrow. Four pairs of branched dorsolateral appendages. Four veil appendages. Five to six processeses of the rhinophoral stalks. Rhinophores with 18 lamellae. Basal colour bluish-white, with numerous scattered yellow spots, various processes opaque white. Masticatory processes of jaws with denticles. Radula with up to 41 rows of teeth. Central tooth with up to 20 small denticles without furrows, rarely completely smooth. Up to eight lateral teeth with up to nine denticles. Reproductive system unknown. Body length 25 mm (living, original description). Distribution. Japan, Paci c side of Honshu, Japan.
Remarks. This species is insu ciently known. Information on the internal morphology of specimens from the type locality is restricted to the original description . Additional data in Pola and Stout (2008) referring to D. gracilis specimens are from considerably remote locations o f the tropical island of Okinawa, whereas the type locality of D. gracilis is in the temperate Sagami Bay.
Their study lacks molecular data and may represent a separate species. The Okinawan specimen indicated in Pola and Stout (2008) has eight pairs of dorsolateral processes, six processess of the oral veil, and 12 rhinophoral lamellae, whereas according to the original description in  there are four pairs of dorsolateral processes, four processesses of the oral veil, and 18 rhinophoral lamellae, respectively. Taking into consideration that details of colour patterms between the original description of D. gracilis and Okinawan specimens are also di ferent as indicated by Pola and Stout (2008: 65), we cannot con rm the conspeci city of the Okinawan specimens with the original D. gracilis, even if we consider the potentially cooler waters in Okinawa at 69 m depth and that only a few specimens were involved in the comparison. Furthermore, Pola and Stout (2008) considered specimens from New Zealand as belonging to D. gracilis as well, which also show di ferences in colouration. However, in the light of a modern emerging paradigm of multilevel organismal diversity that challenges the previously concept of widely claimed polytypic species with broad ranges in favour of considerably smaller, often geographically restricted species, a potential presence of true D. gracilis in New Zealand can only be the result of an anthropogenic transportation, which is hard to explain in this case and needs veri cation. Fig. 7 Dendronotus iris Cooper, 1863: 59;MacFarland, 1966: 257-265, Pl. 47, Figs 12-18;Plate 48, Figs 1-6;Pl. 49, Fig. 4;Pl. 50, Fig. 1;Pl. 51, Figs 1-5;Robilliard, 1970: 446-450, Figs 4-6, 10-12, Pl. 63, Fig. 30;Behrens, 1980: 74;Behrens & Hermosillo, 2005: 94. Dendronotus giganteus O'Donoghue, 1921Pl. 4, Fig. 47;Pl. 5, Figs 57-59. Extended diagnosis. Body relatively narrow. Four to eight pairs of branched dorsolateral appendages. Two to four oral veil appendages. Two to five appendages (longest internal) of rhinophoral stalks. Lateral papilla of rhinophoral sheaths present. Rhinophores with 15-31 lamellae. 20-40 lip papillae. Basal colour grayish to reddish-orange, with white line bordered foot edge. Dorsal processes of jaws inclined posteriorly at approximately 40° to the longitudinal axis of the jaw body and 0.35 of its length. Masticatory processes bear denticles (possibly including ridge-like structures and denticles). Radula with up to 61 rows of teeth. Central tooth quite narrow, with deep furrows and with up to 18 distinct denticles. Up to 21 lateral teeth commonly smooth, occasionally with up to eight denticles. Ampulla relatively narrow with numerous loops. Bursa copulatrix moderate in size, oval, bent. Seminal receptaculum small, placed distally at a moderately short distance from the vaginal opening. Prostate very long, non-discoid, linear, with numerous alveolar glands. The vas deferens is short, penis moderately short, straight, with blunt tip. Body length of live specimens may reach more than 20 cm.

| . / -
Distribution. NE Paci c, from southern Alaska to California (but records from Baja California need to be con rmed). As an adult it feeds on cerianthid anthozoans (Wobber, 1970;Shaw, 1991), quite di ferent to other Dendronotus species.
Remarks. This species is clearly distinguished by a combination of external characters (uniform grayish to reddish-orange colour, massive dorsal appendages) and internal characters (linear instead of discoid prostate -unique for the genus), and also by adult behavioral patterns (special "rearing attack" feeding on species of Ceriantharia, e.g., Shaw, 1991). Dendronotus nanus Marcus et Marcus, 1967 was considered a synonym of D. iris (Stout et al., 2010), but in the absence of molecular data and presence of morphological di ferences (see below) we prefer to retain D. nanus as a separate species (see details below). Dendronotus iris possibly is the largest species of its genus. Extended diagnosis. Body narrow. Six to seven pairs of branched dorsolateral appendages. Six to nine appendages of oral veil. Four to ve appendages (posterior ones longest) of rhinophoral stalks. Lateral papilla of rhinophoral sheaths present. Rhinophores with 12-15 lamellae, 20-25 lip papillae. Basal colour whitish with brownish spots and streak. Dorsal processes of jaws inclined posteriorly at approximately 56° to the longitudinal axis of the jaw body and 0.47 of its length. Masticatory processess with ridge-like structures. Radula with up to 38 rows of teeth. Central tooth with up to 17 small, distinct denticles with furrows. Up to eight lateral teeth with up to nine denticles. Ampulla moderately narrow, with several loops. Bursa copulatrix large, oval. Seminal receptaculum small, placed distally at a moderately short distance from the vaginal opening. Prostate discoid with 50-60 alveolar glands. The vas deferens is moderate in length, penis conical, relatively thin. Body length up to 21.5 mm.
Bathymetry. A deep-sea (bathyal) species, so far recorded at 670-761 m depth (for discussion of the di ferent usage of the term "bathyal", see Rybakova et al., 2020).
Remarks. This species is recently described in detail . Bathymetry. 14-60 m depth.

Dendronotus kalikal
Remarks. Although this species was previously confused with D. frondosus, D. kalikal is very distinct from it by morphological features (mottled body, brownish colouration with indistinct but commonly present dorsal blurred subparallel stripes), radular patterns and molecular phylogenetic data. In the present study we provide evidence that in the North Atlantic to the Kara Sea there is a sister species, D. yrjargul sp. nov., separate from D. kalikal, see above. Extended diagnosis. Body narrow. Five to six pairs of branched dorsolateral appendages. Four to six oral veil appendages. Five to six appendages (apparently similar in size or posterior longer) of rhinophoral stalks. Lateral papilla of rhinophoral sheaths present. Rhinophores with 5-11 lamellae. Lip papillae 4-15. Basal colour semitransparent to brownish with irregularly scattered small brownish dots. Dorsal processes of jaws inclined posteriorly at approximately 70° to the longitudinal axis of the jaw body and 0.37 of its length. Masticatory processes possibly bear denticles and ridge-like structures. Radula with up to 44 rows of teeth. Central tooth with almost completely reduced denticles and furrows at adult stages, subadults may bear small denticles. Up to eleven lateral teeth with up to nine denticles. Ampulla moderately voluminous with several (at least two or three) folded loops. Bursa copulatrix large, rounded. Seminal receptaculum small, placed distally at a moderately short distance from the vaginal opening. Prostate discoid with circa 7-25 alveolar glands. The vas deferens is moderate in length, penis relatively long, slightly curved. Body length up to 30 mm.

Dendronotus kamchaticus
Distribution. NW Paci c from Kamchatka to Japan Sea, Russia, and NE Paci c in Salish Sea, Canada and USA.

Bathymetry. 7-17 m depth.
Remarks. Though this species previously was confused with D. frondosus, D. albus and D. venustus, D. kamchaticus is well delineated from those species according to the morphological features (semitransparent grayish with white-tipped dorsolateral | . / appendages and digestive gland often penetrating all branches of all dorsolateral appendages to body brownish colouration without dorsal subparallel stripes), radular patterns (usually almost smooth central teeth in adults) and molecular phylogenetic data ( gs 1, 2). In the original description branches of the dorsolateral appendages were incorrectly described as bulbous and short (see Ekimova et al., 2015: Fig. 8D), despite the presence of this species in the same paper (Ekimova et al., 2015: Fig. 16A, B) represented by photos of live paratypes with more elongated, not exactly bulbous, though still shortened branches. Further, in a specimen of D. kamchaticus from the Sea of Japan the branches were described as elongated and also very di ferent from the original description and Fig. 8D in Ekimova (2015). This signi cant discrepancy compared with the original description was explained: "because di ferent ontogenetic stages were used in the description and this study, as well as cerata morphology may be altered in preserved specimens" (Ekimova et al., 2016: 38). However, this discrepancy was not due to the di ferent ontogenetic stages. In the original description of D. kamchaticus, the type specimens were represented by an almost adult holotype (12 mm preserved length) with a mature reproductive system whereas seven paratypes were subadults (2-10 mm of preserved length). Most importantly, the drawing in Ekimova (2015: Fig. 8D) depicted the largest holotype, with an adult reproductive system and the arti cially bulbous branches were only seen in the preserved specimen, without any considerable ontogenetic modi cations. The largest specimen so far known of D. kamchatcius (adult, 30 mm) from the NE Paci c, also has considerably elongated branches (Korshunova et al., 2016a). This fully con rms that the branches were described and gured incorrectly in the original description. Fig. 7 Tritonia lactea Thompson, 1840: 88-89, Pl. 2, Fig. 3.

Dendronotus lacteus (Thompson, 1840)
Dendronotus lacteus Eliot, 1910: 161;Nordsieck, 1972: 68;Pruvot-Fol, 1954: 356;Thollesson, 1998: 191-193 Extended diagnosis. Body relatively narrow. Four to eight pairs of branched dorsolateral appendages which in adults commonly show speci c "curly" pattern though other specimens with more "typical" for genus Dendronotus elongated appendages may occur. Four to ten oral veil appendages. Four to six appendages (inetrnal middle and posterior ones longest) of rhinophoral stalks. Lateral papilla of rhinophoral sheaths present. Rhinophores with 10-14 lamellae. At least more than 15 lip papillae. Basal colour from uniformly dark red to uniform white, often with small brownish dots. Redddish and brownish specimens may have white stripes between cerata, white specimens milky, non-transparent in appearance. Dorsal processes of jaws inclined posteriorly at approximately 55-60° to the longitudinal axis of the jaw body and 0.52 of its length. Masticatory processes possibly bear denticles and ridge-like structures. Radula with up to 43 rows of teeth. Central tooth with up to 40-50 small denticles (commonly smaller), often with peculiar ribs, never completely smooth. Up to 16 lateral teeth with up to ve denticles. Ampulla moderately narrow with several windings or folded loops. Bursa copulatrix large, rounded to oval. Seminal receptaculum small placed distally at considerable distance from the vaginal opening. Prostate discoid with a range from 12 to more than 30 alveolar glands.
The vas deferens is relatively long in length, penis relatively long, bent. Body length up to 150 mm.
Distribution. North Atlantic and Arctic (eastern limit at least Laptev Sea, Russia).
Bathymetry. From circa 5 to 427 m depth; common at shallow, subtidal depths at around 10-20 m.
Remarks. Dendronotus lacteus together with D. frondosus and D. europaeus form a species complex that is di cult to distinguish based on external characteristics, but in which it is still possible to nd ne-scale morphological di ferences. In contrast, the radulae of these three species demonstrate distinctly di ferent patterns (Korshunova et al., 2017b).

Dendronotus nanus Marcus & Marcus, 1967
Dendronotus nanus Marcus & Marcus, 1967: 210-214, Figs 64-65;Farmer, 1980: 72;Robilliard, 1972: 428-430.  Marcus and Marcus (1967) in their original description specially highlighted that D. nanus di fers from D. iris by a considerably longer vas deferens in animals of only 13 mm length (an considered to be not fully mature), whereas D. iris that was studied anatomically by Robilliard (1970) had a considerably larger average body length 65-120 mm but a signi cantly shorter vas deferens. In the BOLD database there is a sequence of D. nanus (in GenBank identi ed as D. iris), but from a very remote location in British Columbia whereas the type locality of D. nanus is Baja California, Sonora, thus this available sequence does not represent a true D. nanus. Even if taken into account that the glandular part of vas deferens likely is underdeveloped in D. nanus (see Marcus & Marcus, 1967), the non-glandular muscular part of vas deference of D. nanus is considerably longer than in D. iris. Robilliard | . / -(1972) compared D. iris and D. nanus in detail and mentioned similarity of both species but did not synonymize them. All these characters need to be re-analysed before any further attempts at synonymizing these species should be done. Bathymetry. 1819-1822 m depth.

Dendronotus patricki
Remarks. So far only known from three specimens collected at the type locality. Fig. 7 Dendronotus primorjensis Martynov et al., 2015a: 60, Fig. 5A-G;Martynov et al., 2015c: 75-76;Korshunova et al., 2016b: 15-28, Figs 1, 2;Korshunova et al., 2016a: 20-32;Korshunova et al., 2019b: 6-8 Extended diagnosis. Body narrow. Five to nine pairs of branched dorsolateral appendages (including 2-3 smaller pairs). Five to twelve oral veil appendages. Four to ve appendages (internal middle and posterior ones longest) of rhinophoral stalks. Lateral papilla of rhinophoral sheaths present. Rhinophores with 8-12 lamellae. Five to 12 lip papillae. Basal colour nonuniformly reddish brown with few opaque white stripes between dorsolateral processes to more uniformly olive almost without white pigment, or almost lacking general pigmentation. Dorsal processes of jaws inclined posteriorly at approximately 45-47° to the longitudinal axis of the jaw body and 0.48 of its length. Masticatory processes bear denticles and ridge-like structures. Radula with up to 41 rows of teeth. Central tooth with deep furrows and with up to 18 (common range 12-14) distinct denticles. Up to nine lateral teeth with up to eight denticles. Ampulla moderately narrow with several windings or folded loops. Bursa copulatrix large, rounded. Seminal receptaculum small placed distally at a moderately short distance from the vaginal opening. Prostate discoid with range 12-19 alveolar glands. The vas deferens is moderate in length, penis relatvely long, curved. Reported body length up to 35 mm.

Dendronotus primorjensis Martynov, Sanamyan & Korshunova, 2015
Distribution. NW Paci c, common in the northern part of the Sea of Japan, at both the Russian coast and at Hokkaido on the Japanese coast (Korshunova et al., 2016b, 2019b).
Diagnosis (original description). Exact number of pairs of branched dorsolateral appendages unknown (at least, more than three). Living colour purple, preserved reddish brown. Circa seven oral veil appendages. Five appendages (described as equal in size, likely due to preservation) of rhinophoral stalks. Lateral papilla of rhinophoral sheaths present. Rhinophores with circa 20 lamellae. Several lip papillae (no exact count present in original description). Masticatory processes of jaws with small denticles. Radula with up to 44 rows of teeth. Central tooth with a very ne denticulation. Up to 14 lateral teeth with up to six denticles. Prostate and vas deferens unknown, penis long, straight, conical. Body length up to 20 mm.

Bathymetry. 31 m depth.
Remarks. According to its original description (Bergh, 1879), D. purpureus has 14 lateral radula teeth and is thus similar in this respect to the North Atlantic and Arctic species D. lacteus. However, D. lacteus of such a size has usually less than 14 lateral teeth, and also the central teeth usually have more distinct denticles. We therefore retain the North Paci c D. purpureus as a separate species, potentially a sister species to D. lacteus, until further studies are done. Fig. 7 Dendronotus robilliardi Korshunova, Sanamyan, Zimina, Fletcher & Martynov, 2016: 28-32, Figs 2, 3, 5. Extended diagnosis. Body narrow. Five to eight pairs (commonly more than six) of branched dorsolateral appendages, digestive gland penetrates at least three pairs of dorolateral appendages. Four to six appendages of oral veil. Four to six appendages of rhinophoral stalks. Lateral papilla of rhinophoral sheaths present. Rhinophores with 11-15 lamellae. Five to ten lip papillae. Basal colour white, dorsolateral appendage variable opaque orange or opaque white, including tip. Dorsal processes of jaws inclined posteriorly at approximately 50-60° to the longitudinal axis of the jaw body and 0.43-0.45 of its length. Radula with up to 43 rows of teeth. Masticatory processes with ridge-like structures and smaller denticles. Central tooth with up to 17 small, distinct denticles without furrows. Up to nine lateral teeth with up to eight denticles. Ampulla is hook-shaped to swollen kidney-shaped, with various degrees of folding. Bursa copulatrix large, rounded, with small seminal receptaculum placed distally at a short distance from the vaginal opening. Prostate discoid with about 19-30 alveolar glands. The vas deferens is short, penis relatively short, curved. Body length up to 35 mm.  Verrill, 1870: 405;Lundin et al., 2017: 303, 306, 309-315, Figs 3, 4. Non auct., not Ekimova et al. (2015) (mix-up with D. velifer G.O. Sars, 1878.

Dendronotus robilliardi Korshunova, Sanamyan, Zimina, Fletcher & Martynov, 2016
Extended diagnosis. Body wide. Six to seven pairs of branched dorsolateral appendages. 10 to 15 appendages of oral veil. Four to ve appendages of rhinophoral stalks. Lateral papilla of rhinophoral sheaths absent. Rhinophores with 7-14 lamellae. More than 20 lip papillae. Basal colour light grayish to yellowish with scattered opaque white and yellow spots. Dorsal processes of jaws inclined posteriorly at approximately 47° to the longitudinal axis of the jaw body and 0.55 of its length. Masticatory processes apparently bear denticles (may possess ridgelike structures). Radula with up to 31 rows of teeth. Central tooth with high cusp and up to 20 small distinct denticles without furrows. Up to nine lateral teeth with up to seven denticles. Ampulla moderately narrow, with approximately three to four irregular loops. Bursa copulatrix moderate in size, rounded to pear-shaped, with small seminal receptaculum placed distally near the vaginal opening. Prostate discoid with about 10 alveolar glands. The vas deferens is moderate in length, penis massive, broad, slightly curved. Body length of veri ed specimens up to 50 mm.
Distribution. North Atlantic (both western and eastern, Canada, Norway, Russia and USA), very shallow waters only, in the Arctic, the range does not extend further eastward than the Barents Sea.
Bathymetry. Intertidal to less than 20 m deep.

Distribution. North Eastern Paci c, Canada and USA.
Bathymetry. From intertidal to 20 m depth.
Remarks. Dendronotus subramosus is unique among the clades of narrow-bodied species since it has no lateral papillae on the rhinophoral sheath. However, phylogenetically D. subramosus is more closely related to the externally similarlooking deep sea species D. jamsteci (with the presence of rhinophoral sheath papillae), which was recently described o f Paci c Honshu (Japan) and also to the externally di ferent-looking NE Paci c shallow water species D. albus ( gs 1, 2). Such a morphological and molecular evolutionary mosaic is remarkable and contribute for the | .

Dendronotus velifer G.O. Sars, 1878
Extended diagnosis. Body wide. Four to six pairs of branched dorsolateral appendages. Six to twelve main appendages on oral veil. Four to ve appendages on rhinophoral stalks. Lateral papilla of rhinophoral sheaths absent or very small. Rhinophores with 12-15 lamellae. 30-35 lip papillae. Basal colour commonly light-to bright red with scattered white dots. Dorsal processes of jaws inclined posteriorly at approximately 65° to the longitudinal axis of the jaw body and 0.55 of its length. Masticatory processes with denticles. Radula with up to 36 rows of teeth. Central tooth with low to moderate cusp and up to 26 small distinct denticles without furrows. Up to 15 lateral teeth with which is smooth or with up to four rudimentary denticles. Ampulla twolooped. Bursa copulatrix large, rounded, with small seminal receptaculum placed distally at some distance from the vaginal opening. Prostate discoid with more than 50 alveolar glands. The vas deferens is moderate in length, penis thin, long. Body length up to 90 mm.
Distribution. North Atlantic (southern limit -Gullmar ord, Sweden) and Arctic (at least to the Laptev Sea), Norway and Russia.

Dendronotus venustus
Diagnosis. Body narrow. Four to eight pairs of branched dorsolateral appendages. Four to eight appendages of oral veil. Four to six appendages (posterior and medial ones longest) of rhinophoral stalks. Lateral papilla of rhinophoral sheaths present. Rhinophores with 8-14 lamellae. Lip papillae absent or up to eight. Colour forms include white specimens with yellow spots and brownish or reddish specimens with yellow dots and/or tubercles or large amount of white pigment, but with no white-tipped appendages. Dorsal processes of jaws inclined posteriorly at approximately 55-60° to the longitudinal axis of the jaw body and 0.41 of its length. Masticatory processess with ridge-like structures and smaller denticles. Radula with up to 48 rows of teeth. Central tooth with up to 12 small distinct denticles with furrows. Up to eight lateral teeth with up to ve denticles. Ampulla voluminous, folded. Bursa copulatrix large, pear-shaped. Seminal receptaculum small placed distally at considerable distance from the vaginal opening. Prostate discoid with few (5-12) alveolar glands. The vas deferens is long, possibly also partly prostatic since thickened on considerable length and then distally narrowed, penis thin, long. Body length up to 42 mm.
Distribution. NE Paci c, Canada and USA.
Bathymetry. Common in shallow waters, at depths of circa 5-20 m.
Remarks. Despite MacFarland's (1966) detailed description of D. venustus, it was almost universally considered a synonym of D. frondosus (Robilliard, 1970;McDonald, 1980;Behrens & Hermosillo, 2005) until molecular data showed that it is distinct from D. frondosus (Stout et al., 2010;Korshunova et al., 2019b). Robilliard (1970) did not doubt the synonymy of D. venustus, although he rarely observed more than 12 of the prostatic alveolar glands in the NE Paci c specimens, whereas in true D. frondosus their number is usually up to at least 30. There are also minor di ferences in the radular patterns and some details of colouration between D. frondosus and D. venustus (e.g., common presence of white forms with small yellow spots in D. venustus).
There is a possibility of the presence of hidden diversity within D. venustus. However, a specimen from Alaska included in our present study ( g. 7) matches the original description of D. venustus very well by external and internal features (MacFarland, 1966: 271-275, pl. 40 g. 2, pl. 46, g. 9, pl. 50 g. 3). These characters include the presence of yellow-greenish spots on a greyish to whitish body, a speci cally broadened, somewhat cheliform rst lateral tooth, and a long, broadened, apparently partly prostatic vas deferens in addition to a small discoid alveolar prostate. We therefore surely refer to this specimen as a true D. venustus in our present phylogenetic analysis ( g. 1). Five to seven appendages of oral veil. Circa ve appendages (posterior ones the longest) of rhinophoral stalks. Lateral papilla of rhinophoral sheaths present. Rhinophores with 10-11 lamellae. Lip papillae 20-30. Basal colour bright red to reddish-brownish with thin white broken lines between dorsolateral appendages, also scattered opaque white dots and speckles on dorsal and lateral sides, and on various appendages. Dorsal processes of jaws inclined posteriorly at approximately 70° to the longitudinal axis of the jaw body and 0.5 of its length. Masticatory processess with ridge-like structures. Radula with up to 36 rows of teeth. Central tooth completely devoid of denticles (except for the anteriormost juvenile radula in some specimens), only sometimes faint traces of furrows occur. Up to 12 lateral teeth with up to seven denticles. Ampulla moderately narrow, sligthly bent. Bursa copulatrix very large, oval. Seminal receptaculum small placed distally at considerable distance from the vaginal opening. Prostate discoid with circa 25 alveolar glands. The vas deferens is moderate in length, penis conical, curved. Body length up to 30 mm.

Dendronotus zakuro
Etymology. From the Indonesian word "cabang" meaning "branch" in reference to this genus as "dendronotids of the tropics" and to respect the great contribution of the Indonesian fauna to global marine biodiversity (e.g., Hoeksema, 2007).
Diagnosis. Dorsolateral appendages with distinct tertiary branches. Oral veil with branched appendages. Radula with at least up to nine lateral teeth in adult specimens. Central teeth with cusp integrated within lateral denticles. Denticle-bearing part of central teeth not separated from lateral sides, distinct shoulders absent, thus central teeth triangular in outline. Prostate represented by a thickened structure without evident alveolar glands and disk. Copulative organ conical.
Remarks. According to the present molecular phylogenetic analysis, the tropical Dendronotus regius Pola & Stout, 2008 comes as a most basal clade, distinct from all presently known species of the genera Dendronotus and Pseudobornella. Morphologically, D. regius di fers from any known species of the genus by the presence of a homogenous, ne prostate, which is externally smooth and does not show any alveoles. Pola and Stout (2008: Fig. 4) evidently did not indicate any alveols in the prostate of their D. regius, but in the text they (Pola & Stout, 2008: 48) mentioned a "large prostate with the proximal limit being marked by a closely set ring of alveolar glands". To clear this contradiction between this gure and the text in the original description, we checked several adult specimens of D. regius and con rm that there are no externally or internally conspicous alveols, otherwise so characteristic for a majority of true Dendronotus species. There is another striking di ference between both strictly tropical species of Dendronotus (D. regius plus D. noahi) and any other known species of this genus, namely that the denticle-bearing part of the central teeth is not separated from the lateral sides by a distinct shoulder. Thus the central teeth have a squarish base and a triangular top. This makes the radula of both species more similar (though not identical) to that of the family Bornellidae, whose species are also partly similar in external morphology. However, details of the stomach and the reproductive systems are di ferent. Therefore, taking molecular and morphological evidence into account, we propose here the new genus Cabangus gen. nov. with as type species Dendronotus regius Pola & Stout, 2008. The holotype of the only other tropical species described so far, D. noahi, is likely a juvenile since it possesses only four rows of lateral teeth and immature reproductive system. Although this species is now included in the genus Cabangus gen. nov., this needs further study. Externally, Cabangus spp. are di ferent from the common external appearance of Dendronotus spp. by a combination of a very narrow body and short dorsolateral appendages. By these characters, Cabangus gen. nov. is also somewhat similar to the family Bornellidae, which is phylogenetically distant from the Dendronotidae. Cabangus gen. nov. represents a basalmost clade within the Dendronotidae and hence may retain some features of the common ancestors with phylogenetically more distantly related families. Cabangus gen. nov. is also well supported by biogeographic data because the majority of its species inhabits Arctic and temperate waters, whereas Cabangus gen. nov. represents a distinct tropical lineage. (Pola & Stout, 2008)  Extended diagnosis. Body very narrow. Three to four pairs of branched dorsolateral appendages. Circa six appendages of oral veil. Four to ve appendages (posterior ones longest) of rhinophoral stalks. Rhinophores with 10-11 lamellae. Lateral papilla of rhinophoral sheaths present. Lip papillae possibly absent. Basal colour white or yellow, with blue, brown and reddish patches. Masticatory processes with denticles. Radula with up to 36 rows of teeth. Central tooth with up to 20 small distinct denticles with furrows. Up to nine lateral teeth with up to seven denticles. Ampulla voluminous, folded. Bursa copulatrix large, pear-shaped to oval. Seminal receptaculum small placed distally at a short distance from the vaginal opening (in the original description entangled duct from ampulla was incorrectly identi ed as "receptaculum seminis"). Prostate non discoid, thickened, without evident alveolar glands. The vas deferens is moderate in length, penis tapered. Body length up to 15 mm.

Bathymetry. Up to 45 m deep.
Remarks. See under Cabangus gen. nov.

Genus Pseudobornella Baba, 1932
Type species. P. orientalis Baba, 1932 ( g. 2) Diagnosis. Dorsolateral appendages without distinct (or with few very short) tertiary branches. One of the rhinophoral stalk appendages from both sides is extremely long, usually exceeding body length. Oral veil with long simple unbranched appendages. Radula with very small number of lateral teeth (presently known to be no more than two rows of lateral teeth). Central teeth with protruding cusp distinctly separate from lateral denticles. Prostate represents by narrow tube, without alveols and disk. Copulative organ partly attened and with a widened subcircular apical part.
Remarks. The genus Pseudobornella di fers from both Dendronotus and Cabangus gen. nov. by a unique combination of external and internal characters. These characters include a very long appendage of the rhinophoral sheath, absence of distinct tertiary branches of the dorsolateral appendages, resulting in the general appearance of dorsolateral appendages of Pseudobornella as ctenidium-like. Instead, all species of the genus Dendronotus possess well-de ned tertiary branches on the dorsolateral appendages, and no species has such a exceedingly long appendages of the rhinophoral sheaths. There is a deep-sea species, Dendronotus claugei (see above), for which unbranched dorsolateral appendages are reported, but this species is only known from a single specimen and needs further studies. According to the present phylogeny D. claugei may represent a separate genus, but more data are needed to support this. Furthermore, so far no single species of the genus Dendronotus and Cabangus gen. nov. have smooth unbranched appendages in the oral veil, whereas numerous reported Pseudobornella specimens invariably show strong unbranched appendages of the oral veil. Ultimately, the radular pattern of the genus Pseudobornella di fers from that in the adult stage of species of Dendronotus and Cabangus gen. nov. by a very small number of lateral teeth (so far reported no more than two). Adult specimens of the genera Dendronotus and Cabangus gen. nov. possess at least eight lateral teeth, and only very rarely in poorly known deep-sea lineages, the number of the lateral teeth rows can reach six in number. The tropical C. noahi was reported to have only four rows of lateral teeth, but only a single juvenile specimen is known (Pola & Stout, 2008). All investigated species of the genus Dendronotus showed the presence of 1-3 rows of lateral teeth during the early stages of their ontogeny (e.g., Martynov et al., 2020a), while only in the genus Pseudobornella such feature became apparent in adults stage. Furthermore, the shape of the central teeth of Pseudobornella is also di ferent from that in any adult Dendronotus and Cabangus gen. nov. species by the presence of a strongly protruding central cusp on the central teeth, which is distinctly separated from the lateral denticles.
A image of the radula of P. orientalis from the Sea of Japan was presented in Martynov et al. (2015a: 60, g. 5G). Radular characters coupled with the molecular data (including presenting for the rst time data on gene for the type species Pseudobornella orientalis, gs 1, 2) show that the genus Pseudobornella represents an early o fshoot of the family Dendronotidae. This evolution was fuelled by the paedomorphic reduction of the lateral teeth and juvenilization of the central teeth (for criteria of paedomorphosis Martynov et al., 2020b). A previous morphological cladistic study (Pola et al., 2009) placed the genus Pseudobornella outside the family Dendronotidae, thus highligthing the morphological peculiarities of this genus as inconsistent with those of other dendronotid genera. However, a solely molecular study suggested to synonymize Pseudobornella with Dendronotus (Pola & Gosliner, 2010). We con rm here the validity of the genus Pseudobornella Baba, 1932 using integrative evidences. Specimens of the type species Pseudobornella orientalis from the Sea of Japan ( g. 2) in the present study matched morphologically well with those in the original description of P. orientalis (Baba, 1932), including the characteristic small chocolate-brown spots and yellow lines on the dorsal side, and the shape of the radular teeth.
Species composition. This presently monospeci c genus only contains P. orientalis Baba, 1932.

General taxonomic and biogeographic overview of the family Dendronotidae: a revolution of ne-scale de ning of species
The family Dendronotidae and its type genus Dendronotus represent an emerging model for studying speciation, and ultimately for investigating general problems related to the "species concept". The total number of . / -| .
species belonging to the genus Dendronotus in restricted sense is 27 (a review of all the previously described and species considered valid with diagnoses is presented in the Synopsis section). Though some new species were recognized using morphological methods (e.g., MacFarland, 1966;Robilliard, 1970Robilliard, , 1972, the species composition of the genus Dendronotus remained quite conservative (Thompson & Brown, 1984;Roginskaya 1987Roginskaya , 1997. The genus Dendronotus was represented by just about 10 recognised shallow-water species worldwide (plus one deep-sea and two or three species of unclear status) until a considerable amount of hidden diversity was discovered recently in the shallow waters of both the North Paci c and the North Atlantic (e.g., Korshunova et al., 2017a;. The conservative approach in the taxonomy of the family Dendronotidae relied considerably on Mayr's (1942Mayr's ( , 1969 polytypic species concept and allowed the presence of broadly de ned "species" distributed over very large geographic distances. For example, until about the year 2000 only D. frondosus was recognised among the North Atlantic narrow-bodied Dendronotus spp. (Thompson & Brown, 1984). The wide geographic range of D. frondosus at that time also encompassed the entire North Paci c up to subtropical regions of the middle of Honshu (Japan) and China (e.g., Robilliard, 1970;Baba, 1993;Lin et al., 1986). Later this has showed not to be the case, and the supposedly super-polytypic and pan-geographic "species" D. frondosus has been split into at least 11 narrow-de ned species (see below and gs 1, 7). Each of these species occupies a particular geographic region. The natural range of true D. frondosus sensu stricto is limited solely to the eastern and western parts of the North Atlantic with the subarctic Barents Sea as easternmost limit, while it is absent in the Arctic region.

Dendronotus europaeus inhabits the northerneastern Atlantic in European waters only.
Dendronotus arcticus is reported only from the true Arctic, e.g., Kara and Laptev seas. Dendronotus lacteus occurs in northeastern Atlantic and Arctic with the Laptev Sea as easternmost limit. Dendronotus yrjargul sp. nov. inhabits mostly the Arctic (with the Kara Sea as easternmost limit) and the neighbouring northern European regions with middle Norway as southernmost limit. Dendronotus nordenskioeldi sp. nov. is so far only reported from the Arctic Laptev sea. Dendronotus kalikal is distributed in the northwestern Paci c until the Bering Strait. Dendronotus kamchaticus is limited to the Northern Paci c only, with scattered records from the northwestern and northeastern parts. Dendronotus zakuro is reported from the northwestern Paci c only, ranging from middle Honshu to Kamchatka Peninsula. Dendronotus primorjensis is restricted merely to the Sea of Japan and neighbouring localities in Japan. Dendronotus venustus is restricted to the northeastern Paci c. This considerable regionalisation in a majority of the species is remarkable since all of them have planctonic veliger larvae and may potentially disperse very widely. But in reality instead of a single "in nitely variable" species around whole Eurasia we have an enormously complicated radiation of evidently separate species. From these species, so far only D. lacteus shows a maximum range from the northern European seas to the Arctic. However, it also shows a considerable morphological variation compared to other species, which may lead to this species becoming further subdivided. A very broad geographic range (the North Atlantic, Arctic and North Paci c) was previously also indicated for Dendronotus dalli (Robilliard, 1970;Roginskaya, 1987). Mutilevel data for the putative North Atlantic and Arctic "D. dalli" show instead that it represents a | .
/ separate species D. elegans (as "D. niveus"). Dendronotus elegans is a predominatly Arctic species, which reaches cold water masses at the Canadian and US North Atlantic coasts. Dendronotus dalli instead is a predominantly North Paci c species, which reach Arctic seas into the Bering Strait and neighbouring waters, but it is absent in the subarctic North Atlantic regions. Importantly, many of these species could be separated by a careful application of morphological characters without aid of molecular data. The possibility of morphological separation is evident in the case of D. venustus from the NE Paci c. Dendronotus venustus was described in considerable detail and already became delineated from the resembling D. frondosus in the mid-20th century (published in MacFarland, 1966) without use of molecular techniques and scanning electron microscopy ( ). has been available to taxonomists since the 1970s and its use would enable differentiation of some of the Dendronotus species that were described in the 21st Century. However, this did not happen because all these characters could be considered as too subtle from the perspective of the past dominant paradigm. For instance, Thompson and Brown (1984) published an image of radula of a specimen identi ed as D. frondosus but which actually belongs to the recently described D. europaeus (Korshunova et al., 2017b). Even such morphologically well distinguished species as D. albopunctatus (Robilliard, 1972) was commented as very similar to D. robustus (see Rudman, 2007). While data on the multilevel diversity within Dendronotidae are currently accumulating, more and ner cases will be discovered in the future among already restricted "species". Nevertheless, a majority of the recently separated species can either be di ferentiated by radula data or by a combination of the habitus data, radular data and features of the reproductive system.
Notably, Robilliard (1970) noted di ferences in the reproductive system between European D. frondosus and his NE Paci c specimens of D. venustus that he at that time identi ed as "D. frondosus". At the same time, apparently key reproductive features (e.g., number of prostatic alveols and shape of ampulla) can vary signi cantly, as in the cases of D. albus or D. elegans. The separate characters should be therefore used with considerable care to distinguish particular species.
All these cases clearly con rm that not only molecular data, per se, changed the modern landscape of biodiversity research. The fundamental changes have occurred in the dominated paradigm of the species description. A major modern feature of this is the general denial of huge polytypic and pan-geographic species in favour of ne-scale interspeci c morphological di ferences in combination with limited geographic ranges. Within restricted "narrow species" a polymorphism (a basis for the polytypic concept) is often manifested in the occurrence of parallel morphs, which may obscure species diversity (see e.g., Korshunova et al., 2020b). Though Mayr was among the major developers of the population-based, variable, "non-typological" species concept, he also admitted the possible existence of a hardly detectable diversity of "sibling species" (Mayr, 1942;Yoder et al., 2005). Despite this, the gravity of "a morphologically super-variable species concept" was clearly unfavourable for seeking of ne-scale di ferences. This was one of the reasons why researchers for a long time omitted evidence for morphological di ferences in the "D. frondosus megacomplex". The polytypic concept became popular more than ve decades ago and at that time Mayr (1942) suggested a potential solution for how to unfold the apparent "chaotic species" of "typologists" into a polymorphic "biological" species. However this concept can now be considered as rather an obstacle than an aid in the discovery of real patterns of biodiversity. As a clear evidence for that, until very recently only drastically di ferent species within the genus Dendronotus have been recognised without doubts (Robilliard, 1970;McDonald, 1983), despite uncertanties in their delimitation (Robilliard, 1975). An especially remarkable example is the notorious problem of how to distinguish "D. albus" from "D. diversicolor", which has persisted since Robilliard (1970). This problem was in a profound confusion of the morphological characters rather than in the absence of molecular data and was solved only recently by the distinction of D. robilliardi (Korshunova et al., 2016a). The majority of the described species of the genus Dendronotus inhabit cold and temperate regions of the Northern Hemisphere and their diversity clearly declines towards tropical localities. Pola and Stout (2008) assigned two tropical dendronotids rst to the genus Dendronotus. However, the morphology of the male part of the reproductive system of the these tropical representatives is di ferent from those in the genus Dendronotus, and available molecular data for "D. regius" show it to be a separate with a more basal placement compared to any other dendronotid ( gs 1, 2). Therefore "D. regius" has become the type species of the new genus Cabangus gen. nov., which encompasses exclusively tropical dendronotids. This corroborates well with the new general agenda of the multilevel organismal diversity, which proposes narrowly de ned taxa not only on a species level, but also at the genus and family levels (Korshunova et al., 2017a, b, c;2019a). This prevents to produce non-diagnosable huge taxa that not only lack support from morphology but also from various other data, such as biogeography. The taxonomic diversity of the family Dendronotidae undergoes a process of understanding and more taxa in this family are expected to be discovered. For example, a not yet described diversity within the genus Dendronotus was discovered in the temperate waters of Southern Hemisphere (e.g., Burn, 2015).
Most of the previously recognised Dendronotus species (seven out of nine, see Robilliard, 1970Robilliard, , 1972 occur in the NE Paci c, while in the North Atlantic and NW Paci c only one or two species were commonly recognized. This caused a distorted picture of a biogeographic asymmetry in the Dendronotus diversity between the North Atlantic, northwestern Paci c and northeastern Paci c. One of the important results of the ne-scale species de nition is a considerable reduction of the "biogeographic asymmetry" among Dendronotus spp. between NE Paci c, NW Paci c, Arctic and North Atlantic localities. The resulting numbers of shallow-water and shelf species in such regions, as the North Atlantic and the NW Paci c, are impressive. I.e., in the North Atlantic and western subarctic regions (Barents Sea as limit) there are currently seven species, in the true Arctic (Kara, Laptev and East Siberian seas) there are also seven species so far described and recorded, while in the NW Paci c (Bering Strait as northernmost limit) there are eigth species and in NE Paci c there are 10 species. In all regions the number of species will likely increase further. This analysis clearly suggests that not only geographical proximity to the rich warm water fauna (as in the case of NE Paci c) is important, but also that ecological and other factors (including the history of faunal formation) have contributed substantially to an active speciation. The presence of at least seven species of Dendronotus in the Artic as one of the world's coldest regions is a clear evidence for this. All these data and implications have contributed to rapid and revolutionary advancements in the taxonomy of the family Dendronotidae (Thollesson, 1998;Stout et al., 2010;Martynov et al. 2015aMartynov et al. , c, 2020aEkimova | . / et al. 2015;Korshunova et al. 2016aKorshunova et al. , b, 2017aKorshunova et al. , 2019bLundin et al., 2017;present study).

Multilevel Dendronotidae diversity and di ferent degrees of evolutionary separation
The majority of dendronotids are putatively united under the genus Dendronotus, which represents a morphologically heterogenous assemblage of several distinct clades with a signi cant molecular divergence (see gs 1, 2; Synopsis; Appendix, table A2). The case of the family Dendronotidae is therefore particularly illustrative of multilevel di ferences among apparently taxonomically equally recognized "species". These apparently natural units are not equal by their morphological and molecular properties, and broader, ontogenetic properties. Hence, the term "species" is not equal over large phylogenetic distances but also among relatively small levels, such as genus and family. To make taxonomy consistent with the multilevel diversity among the family Dendronotidae, we separate a new genus Cabangus gen. nov. and con rm a separate status of the genus Pseudobornella. Notably, even among the "core species" that is still constitute the genus Dendronotus ( g. 1) our present study demonstrates various degrees of separation from species in a state of being morphologically well distinguished but molecularly low delineated (the D. yrjargul sp. nov. -D. kalikal pair) to morphologically subtle but molecularly well distinguished species (D. frondosus-D. primorjensis, D. dalli-D. elegans, D. lacteus-D. europaeus and D. nordenskioeldi sp. nov.) ( gs. 1-8). We can unambiguously present distinguishing characters only when these apparent groups currently known as species really belong to su ciently distantly related subclades, as in easily recognizable species such as D. iris or D. subramosus, or D. frondosus and D. robustus ( gs 1, 7). A very complicated mutilevel morphological and molecular diversity of the genus Dendronotus implies the presence of many taxa that are incomparably more close to D. frondosus as the type species of the genus (e.g., D. primorjensis, D. venustus) than to the much di ferent wide-bodied D. robustus (and its closely related species, e.g., D. velifer, D. bathyvela). Therefore, although a direct comparison between D. robustus sensus strictus and D. frondosus sensus strictus is possible, this is not taxonomically meaningful anymore considering the presence of such hardly distinguishable complexes. The similarities between the "Dendronotus robustus complex" and the "D. frondosus megacomplex" are thus similarities at genus or higher level. This needs to be adressed in a further study.
The common formulation of a procedure to nd di ferences within such multilevel organismal diversity, and hence the degree of separation between di ferent groups of biological organisms is a "species delimitation" despite the species concept itself having no universal fundamental agreement even in recent publications (Stanton et al., 2019). In the present study we found a remarkable case when the mean di ference of a group is slightly above 2% (normally indicative of a single species) but when morphological, ontogenetic and biogeographic evidence are examined the two Dendronotus groups deserve to be taxonomically marked as distinct species, namely D. kalikal and D. yrjargul sp. nov. ( gs 1-4, 6). At the same time D. nordenskioeldi sp. nov., despite its habitus similarity with D. lacteus, di fers substantially from that species by its central teeth with very weak denticles devoid of rib-like structures ( g. 5) and intriguingly, ( g. 1) it demonstrates signi cant molecular distance from all know species of the genus Dendronotus (Appendix, table A2). Despite the lack of consensus about the meaning of "species" we can observe some "groups of genetically and ecologically similar individuals" (e.g., Shapiro . A conclusion was reached that the "population-to-species" level represents a continuum (Mallet et al., 2009;Coates et al., 2018). Using these numerous available data on di ferent groups and our present data, we therefore disagree with the most recent suggestion to save "species" as a universal concept using a "weak realism" approach (Reydon & Kunz, 2019). Because even in its weaker form, "species" as a reality still implies the universal application of this concept despite that it encompasses groups that have di ferent degrees of separation, and hence di ferent morphological and molecular properties. The big discrepance between "species" or other taxon as a systematic unit and "species/taxon" as a natural entity still persists (e.g., Ereshefsky, 1998;Zachos, 2018). The recently proposed concept of the multilevel organismal diversity (Korshunova et al., 2017b(Korshunova et al., , 2019a) is helping to reduce this discrepance and make a transit from the previous paradigm of "polytypic species", by placing the organismal diversity in a broad ontogenetic framework, as a nescale complex of the morphological, genetic and epigenetic patterns and processes.