1 INTRODUCTION

The genus Doto Oken, 1815 comprises over 90 species of small nudibranchs (Molluscabase, 2021), seldom over a centimeter in length, often only half a centimeter or less. They are generally difficult to identify, with taxonomic uncertainties and species complexes (Lemche, 1976; Morrow et al., 1992; Pola & Gosliner, 2010; Shipman & Gosliner, 2015; Thompson & Brown, 1984), coupled with variation in pattern and a scarcity of defining morphological features. The genus is therefore often viewed as the one of taxonomically most problematic groups among the nudibranchs. The family Dotidae, including four genera, is placed within the Dendronotoidea, combining presence of slender body, smooth oral veil, cone-shaped dorsolateral appenadges and absence of the cuticular lining of stomach (Korshunova, Bakken, et al., 2020; Korshunova & Martynov, 2020; Wägele & Willan, 2000).

The species of Doto typically have stout dorsolateral appendages, forming pairs along the body. These are not cerata sensu stricto because of radically different structure and function. The appendages have neither cnidosacs nor apical pore, but instead have small tubercles in several circlets, placed one above the other. Some species have small, simple pseudobranchs placed close to the base of the dorsolateral appendages, on the side facing the dorsal midline. They can be distinguished from the tubercles by being transparent and having a different shape. The front of the head has a smooth and often inconspicuous oral veil that usually has two rounded lateral flap-like extensions. The rhinophores are smooth, fingerlike and are placed in distinct, often flared sheaths. The radula is narrow and uniseriate, and the jaws are frail and reduced, or even absent (Lemche, 1976; Lundin et al., 2020; Odhner, 1936; Thompson & Brown, 1984).

Species of Doto feed on thecate and athecate hydrozoans and are often claimed as specialized on certain hydrozoan species or genera (Morrow et al., 1992; Picton & Brown, 1981). They may have a notable impact on the marine benthic ecosystem because of the predatory pressure on hydroids, but to our knowledge no studies have been made on this. They do not prey upon the hydroid polyps per se, but slice through the perisarc of the hydroid stalks below the polyps and feed on the caenosarc fluid (Thompson & Brown, 1984). In the North Sea area, fully grown specimens are predominantly observed during late winter to early summer, but some species can be observed later in the season. Because of the taxonomic uncertainties of the species complexes, it is difficult to estimate more precise seasonal variation.

The type species of the genus, Doto coronata (Gmelin, 1791), was originally described by Johann Gmelin as Doris coronata (Gmelin, 1791, p. 3105). Lorenz Oken (1815) introduced the genus name Doto Oken, 1815 (Oken, 1815). Nils Odhner made the first attempt at a revision of the species of Doto in the world and suggested a division of the genus into three main subgroups, based on the coloration of the tubercles on the dorsolateral appendages (Odhner, 1936). Doto coronata had a prominent position in one of these subgroups and Doto fragilis (Forbes, 1838) in another. In the 1970, Henning Lemche discovered that Doto coronata constitutes a species complex, based on observations from samples of Doto collected at numerous localities in the North Atlantic, mainly in the northeast (Lemche, 1976). Based on morphology and species-specific preferences for hydroid prey, Lemche (1976) described five new species from the coronata complex; D. dunnei Lemche, 1976, D. eireana Lemche, 1976, D. koenneckeri Lemche, 1976, D. millbayana Lemche, 1976, and D. tuberculata Lemche, 1976; as well as redescribing older species and establishing a neotype for D. maculata (Montagu, 1804), thus doubling the number of known species from the North East Atlantic area. Picton and Brown (1981) acknowledged that Doto fragilis constitutes a species complex and described the species Doto hystrix Picton & Brown, 1981. Thompson and Brown (1984) indicated that Doto fragilis feeds on hydroids of several different genera such as Halecium Oken, 1815, Nemertesia Lamouroux, 1812, and Tubularia Linnaeus, 1758, but did not separate any new species from the D. fragilis complex, although “varieties” associated with food were mentioned. They concluded that D. fragilis was quote “typically found in association with Nemertesia and Halecium” (Thompson & Brown, 1984: 32); thus, the taxonomic value for prey preference was partially dismissed. Picton and Morrow (1994, 2016) did not separate any species from within D. fragilis either, but mentioned that D. fragilis in the British Isles consists of three different forms, feeding on different hydroids. Morrow et al. (1992) introduced genetic methods for studies of Doto and used electrophoresis to separate two new species from the D. coronata complex; Doto sarsiae Morrow et al., 1992 and Doto hydrallmaniae Morrow et al., 1992, also indicating that other morphs of D. coronata may represent yet other species. It would take another 18 years until genetic studies using Sanger sequencing methods were presented, including North Atlantic and Indo-Pacific Doto species (Pola & Gosliner, 2010; Shipman & Gosliner, 2015). Most species separated by Lemche were confirmed to be distinct, but not all, e.g., D. millbayana did not reveal significant molecular divergence from D. dunnei (Shipman & Gosliner, 2015). The North Atlantic Doto still contains unresolved species complexes with minor differences in morphology, but found on different hydroids, such as for D. fragilis, and more work needs to be done before the phylogeny of the whole group is elucidated. A promising method to untangle the species complexes is to delimit species using multi-locus analysis based on the multi-species coalescent (MSC) method (Knowles & Carstens, 2007; Yang & Rannala, 2010). This method has for nudibranchs recently been applied to the genus Amphorina (see Korshunova, Malmberg, et al., 2020).

2 MATERIALS AND METHODS

One of the aims of the present study is to test the species limits of the Doto species occurring in the North East Atlantic area, by using samples from Sweden, Norway and Northern Ireland (see Table 1, Table S1 and Figure 1 for sampling localities). We particularly wanted to investigate potential species hidden within our current understanding of D. fragilis and D. maculata, in relation to D. coronata. For this analysis, one mitochondrial and one nuclear gene were used. In addition, we are placing the species phylogenetically by analyzing our data together with already published data. Further, we test the results from the molecular analysis against morphology and radular ultrastructure for congruence, using the unified species concept (de Queiroz, 2007). Distance-based single-locus species delimitation was used to generate primary species hypotheses, which then were tested using a MSC-based multi-locus species delimitation method. In this model, genes evolve inside a species phylogeny, the branches are species, and their properties restrict the gene trees. One of these restrictions is that the divergence times between species have to be more recent than the coalescent times for any genes shared between them, assuming no genetic transfer after speciation (Rannala & Yang, 2003). This model can be used for statistical testing of species assignments (Fujita et al., 2012; Rannala, 2015) and has been shown to outperform distance methods (Yu et al., 2017). Thus, we could achieve a more robust model of taxonomy and phylogeny of the genus Doto in the Northeast Atlantic than previous studies, as a step on the way to a more conclusive model on a larger scale.

In the molecular study, a total of 82 specimens were included, representing 11 morphospecies of Doto, from the Skagerrak, the Irish Sea, the North Sea, the Norwegian Sea and the Barents Sea. Also, five out-group specimens from two species were used; Aeolidia filomenae Kienberger et al., 2016 and Dendronotus lacteus (Thompson, 1840) (see Table 1 and Table S1 for details). The phylogeny was estimated using our data combined with previously published data.

The Swedish specimens were collected at four different locations on the Swedish west coast, from south to north at the mouth of the Gullmar fjord close to Lysekil, the archipelago outside Smögen, the Väderö archipelago and finally at the Ide fjord at the border to Norway. The specimens from Norway were collected from south to north at the Oslo and Larvik area, at the mouth of the Sognefjord north of Bergen, at the Trondheimsfjord area, at Saltstraumen in Nordland and finally at the Finnmark area in the Arctic. The specimens from Northern Ireland were collected at Strangford Lough close to the Queens University Marine Laboratory at Portaferry. Specimens were deposited at the Gothenburg Natural History Museum (GNM), Gothenburg, Sweden, and at the NTNU University Museum (NTNU-VM) (Bakken et al., 2021), Trondheim, Norway.

2.3 Review of Doto prey specificity

Thompson and Brown (1984: 28–29) provided at least two different hydroid species as food for four out of 12 Doto species from the British Isles. Distantly related species such as D. fragilis and D. pinnatifida (Montagu, 1804) (Figure 2) were evidently indicated as feeding on the same hydroid species, i.e., Nemertesia antennina (Linnaeus, 1758) (Picton & Brown, 1981; Picton & Morrow, 2016; Thompson & Brown, 1984).

To test the for strict prey specificity within the genus Doto (e.g., Picton & Brown, 1981; Shipman & Gosliner, 2015), we compiled data on the hydroid associations of the Doto species from the available literature sources and own observations (Table 3). For pre-1976 sources (i.e., before Lemche in 1976 showed that “D. coronata” is a complex of different species), we generally did not include records of D. coronata from the same hydroids, as other species split by Lemche and other authors later on (Morrow et al., 1992; Picton & Brown, 1981) were claimed to be almost monophagous. In cases when a hydroid species was not indicated for some of these narrowly defined species, but for D. coronata or other pre-1976 species, we included such hydroids in Table 3. Clytia hemisphaerica (Linnaeus, 1767) was added to the list of food sources of D. coronata, and Nemertesia antennina was added to D. hystrix from our own observations (Table 3).

TABLE 3. List of hydroids associated with Doto species involved in present study

Hydroid species	Doto species	Reference
Abietinaria abietina (Linnaeus, 1758)	D. coronata (Gmelin, 1791) D. tuberculata Lemche, 1976	Lemche, 1976 Picton & Brown, 1981 Thompson & Brown, 1984
Aglaophenia pluma (Linnaeus, 1758)	D. coronata (Gmelin, 1791) D. koenneckeri Lemche, 1976 D. lemchei Ortea & Urgorri, 1978	Swennen, 1961 Picton & Brown, 1981 Thompson & Brown, 1984 Morrow et al., 1992 Present study
Aglaophenia sp.	D. koenneckeri Lemche, 1976 D. lemchei Ortea & Urgorri, 1978 D. millbayana Lemche, 1976 D. paulinae Trinchese, 1881 D. pinnatifida (Montagu, 1804)	Ortea & Urgorri, 1978 Urgorri & Besteiro, 1983 Just & Edmunds, 1985 Rudman, 2006 Shipman & Gosliner, 2015 Present study
Aglaophenia tubulifera (Hincks, 1861)	D. lemchei Ortea & Urgorri, 1978	Morrow et al., 1992 Picton & Brown, 1981 Thompson & Brown, 1984
Amphisbetia operculata (Linnaeus, 1758)	D. coronata (Gmelin, 1791) Doto eireana Lemche, 1976 D. pinnatifida (Montagu, 1804)	Hecht, 1896 Cornet & Marche-Marchad, 1951 Ortea, 1978 Thompson & Brown, 1984
Bougainvillia muscus (Allman, 1863)	D. coronata (Gmelin, 1791)	Miller, 1961 Thompson, 1964
Clava multicornis (Forsskål, 1775)	D. coronata (Gmelin, 1791)	Larsen, 1925 Jaeckel, 1952 Miller, 1961
Clytia hemisphaerica (Linnaeus, 1767)	D. coronata (Gmelin, 1791)	Present study
Coryne eximia Allman, 1859	D. coronata (Gmelin, 1791)	Swennen, 1961
Coryne muscoides (Linnaeus, 1761)	D. coronata (Gmelin, 1791)	Miller, 1961
Diphasia fallax (Johnston, 1847)	D. coronata (Gmelin, 1791)	Present study
Diphasia rosacea (Linnaeus, 1758)	D. coronata (Gmelin, 1791)	Picton & Brown, 1981 Thompson & Brown, 1984
Dynamena pumila (Linnaeus, 1758)	D. coronata (Gmelin, 1791)	Alder & Hancock, 1846 Lemche, 1976 Picton & Brown, 1981 Thompson & Brown, 1984
Ectopleura larynx (Ellis & Solander, 1786)	D. coronata (Gmelin, 1791) D. fragilis (Forbes, 1838)	Walton, 1908 Miller, 1961 Hamond, 1972 Swennen, 1961
Eudendrium ramosum (Linnaeus, 1758)	D. coronata (Gmelin, 1791)	Hamond, 1972
Eudendrium spp.	D. coronata (Gmelin, 1791)	Picton & Morrow, 1994
Garveia nutans Wright, 1859	D. coronata (Gmelin, 1791)	Picton, 1978
Halecium beanii (Johnston, 1838)	D. coronata (Gmelin, 1791)	Miller, 1961
Halecium halecinum (Linnaeus, 1758)	D. coronata (Gmelin, 1791) D. fragilis (Forbes, 1838)	Farran, 1909 Miller, 1961 Thompson, 1964 Hunnam & Brown, 1975 Picton & Brown, 1981 Present study
Halecium muricatum (Ellis et Solander, 1786)	D. fragilis (Forbes, 1838)	Picton, 1978 Picton & Brown, 1981 Thompson & Brown, 1984
Halecium spp.	D. coronata (Gmelin, 1791) D. fragilis (Forbes, 1838) D. pinnatifida (Montagu, 1804)	Walton, 1908 Hamond, 1972 Brown & Picton 1979 Urgorri & Besteiro, 1983 Present study
Halopteris catharina (Johnston, 1833)	D. maculata (Montagu, 1804)	Lemche, 1976 Picton & Brown, 1981 Thompson & Brown, 1984
Kirchenpaueria pinnata (Linnaeus, 1758)	D. dunnei Lemche, 1876	Lemche, 1976 Picton & Brown, 1981 Thompson & Brown, 1984
Laomedea flexuosa Alder, 1857	D. coronata (Gmelin, 1791)	Miller, 1961 Swennen & Dekker, 1987
Lafoea dumosa (Fleming, 1820)	D. coronata (Gmelin, 1791)	Miller, 1961
Nemertesia antennina (Linnaeus, 1758)	D. coronata (Gmelin, 1791) D. fragilis (Forbes, 1838) D. pinnatifida (Montagu, 1804) D. hystrix Picton & Brown, 1981	Jaeckel, 1952 Miller, 1961 Kress, 1968 Thompson, 1964 Lemche, 1976 Picton & Brown, 1981 Thompson & Brown, 1984 Present study
Nemertesia norvegica (Sars, 1873)	Doto fragilis, Doto cf. fragilis	Present study
Nemertesia ramosa (Lamarck, 1816)	D. cuspidata Alder & Hancock, 1862 D. fragilis (Forbes, 1838) D. millbayana Lemche, 1976	Miller, 1961 Thompson, 1964 Lemche, 1976 Just & Edmunds, 1985 Thompson & Brown, 1984
Obelia geniculata (Linnaeus, 1758)	D. coronata (Gmelin, 1791) D. paulinae Trinchese, 1881	Miller, 1961 Lemche, 1976 Schmekel & Kress, 1977 Picton & Brown, 1981 Thompson & Brown, 1984 Lambert, 1991 Martynov et al., 2006 Rudman, 2006 Shipman & Gosliner, 2015 Present study
Obelia dichotoma (Linnaeus, 1758)	D. coronata (Gmelin, 1791)	Swennen, 1961 Picton & Brown, 1981 Thompson & Brown, 1984 Lambert, 1991 Shipman & Gosliner, 2015
Obelia longissima (Pallas, 1766)	D. coronata (Gmelin, 1791)	Dekker, 1986 Urgorri & Besteiro, 1983 Martynov et al., 2006 Martynov & Korshunova, 2011
Rhizocaulus verticillatus (Linnaeus, 1758)	D. coronata (Gmelin, 1791)	Miller, 1961
Plumularia setacea (Linnaeus, 1758)	D. millbayana Lemche, 1976	Lemche, 1976 Picton & Brown, 1981 Just & Edmunds, 1985 Thompson & Brown, 1984
Schizotricha frutescens (Ellis & Solander, 1786)	D. hystrix Picton & Brown, 1981	Picton & Brown, 1981 Thompson & Brown, 1984 Present study
Sertularia argentea (Linnaeus, 1758)	D. coronata (Gmelin, 1791) D. dunnei Lemche, 1876 or D. millbayana Lemche, 1876 (recorded as Doto sp.)	Miller, 1961 Picton & Brown, 1981 Thompson & Brown, 1984 Shipman & Gosliner, 2015
Sertularia cupressina Linnaeus, 1758	D. coronata (Gmelin, 1791)	Swennen, 1961 Thompson, 1964 Thompson & Brown, 1984 Shipman & Gosliner, 2015
Sertularella gayi (Lamouroux, 1821)	D. tuberculata Lemche, 1976	Lemche, 1976 Picton & Brown, 1981 Thompson & Brown, 1984
Sertularia sp.	D. pinnatifida (Montagu, 1804)	McMillan, 1944 Moore, 1950 Kress, 1968
Symplectoscyphus tricuspidatus (Alder, 1856)	D. coronata (Gmelin, 1791)	Martynov et al., 2006
Synthecium sp.	D. lemchei Ortea & Urgorri, 1978	Thompson et al., 1990
Tamarisca tamarisca (Linnaeus, 1758)	Doto cf. fragilis	Present study
Thuiaria thuja (Linnaeus, 1758)	D. coronata (Gmelin, 1791)	Reid, 1846
Tubularia indivisa Linnaeus, 1758	D. coronata (Gmelin, 1791)	Jeffreys 1869 Jaeckel, 1952

3 RESULTS

From the total 87 specimens, including the out-groups, COI sequence data were successfully recovered from 79 specimens, and H3 from 48 specimens after phasing in the H3 dataset consisted of 53 sequences. The COI alignment was 658 base pairs (bp) long and the H3 alignment 328 bp long.

3.2 Phylogenetic inference

In both the COI and H3 gene trees (Figure S1a,b), Doto is monophyletic with maximal support, and the groups found in the respective ABGD analyses are recovered. In the H3 tree, there are no signs of hybridization, and all phased alleles of the same individual are closely related.

In the combined Bayesian tree (Figure 2, Figure S1), Doto is monophyletic with maximum support and is divided into four clades: (1) D. cuspidata + D. pinnatifida, which is the sister to the remaining Doto, (2) D. tuberculata, (3) D. hystrix + D. fragilis and (4) D. coronata, Doto sp. (= “D. cf. cuspidata”), D. dunnei, D. maculata, D. millbayana and D. koenneckeri. In the last clade, D. koenneckeri and one “D. cf. maculata” from Northern Ireland are found as sisters, but well separated, and the remaining forms a trichotomy consisting of one clade including D. dunnei, D. millbayana and one D. cf. maculata, another consisting of all D. coronata and all of the Swedish D. cf. maculata specimens, and the last clade with all Norwegian D. cf. cuspidata.

The concatenated Bayesian tree including the extended dataset of Doto and Kabeiro sequences (Figure 3) is similar to Shipman and Gosliner (2015, fig. 4) and Moles et al. (2016, fig. 7). Doto cuspidata is sister to D. pinnatifida, and D. formosa Verrill, 1875, is placed together with D. hystrix and D. fragilis. Our D. coronata + “D. cf. maculata” clade is placed together with additional D. coronata sequences (Figure 3, Figure S1), and our D. dunnei, D. millbayana and the Northern Ireland D. maculata are placed together with additional specimens of D. dunnei, D. millbayana, D. sp. A and the D. coronata specimens from Moles et al. (2016), our D. maculata specimen are placed close to one D. maculata and one D. africoronata Shipman & Gosliner, 2015 specimen, our Doto sp. (= “D. cf. cuspidata”) are placed together with D. paulinae Trinchese, 1881. The trees with only our data and the tree based on the extended data are mainly congruent; the only difference in the species included in both trees is the position of D. tuberculata, which is sister to the D. fragilis complex in the extended tree, and found in a tricotomy with the D. fragilis complex and D. coronata complex.

The ML tree (Figure S2) is congruent with the Bayesian trees.

3.4 Haplotype network

The network (Figure 4) shows that the three groups within D. fragilis are separated and that one of the groups also contains D. hystrix. In this group, the two mainly have separate haplotypes, only sharing one, but there is no clear structure within the group.

4 DISCUSSION

In this study, we use a combination of data sources and methods to test the species limits of northeast Atlantic Doto species. For a majority of the species the results are clear-cut. The exceptions are mainly in D. fragilis, where we find three groups, one that do not show any genetic separation from D. hystrix despite large morphological differences. The three groups are separated in COI, but not H3. It is possible that these groups represent separate species; however, more data are needed to confirm this.

This study is one of few studies on nudibranchs that delimits species using multi-locus analysis based on the MSC method. MSC analyses have been used successfully on a wide variety of taxa (e.g., Delić et al., 2017; Fossen et al., 2016; Leache & Fujita, 2010; Martinsson & Erséus, 2018), and it has been shown to outperform distance methods (Yu et al., 2017), which so far has been the standard for molecular species delimitation in nudibranchs. However, in this study we use the same data for the initial division of specimen into species hypothesis, and for the testing of them using MSC analysis, this is not optimal as it introduces issues of circularity (see e.g., Martinsson & Erséus, 2018; Yang, 2015), and the results should be interpreted with caution.

Specimens of Doto fragilis are found in three closely related clades, one that also includes D. hystrix. In the D. fragilis/D. hystrix clade, the D. fragilis from Northern Ireland groups with the Swedish D. fragilis “white morph” and the D. hystrix from Northern Ireland, Norway and Sweden. The spiky tubercles on the dorsolateral appendages of D. hystrix are distinctly different from the rounded tubercles of D. fragilis. Doto hystrix has until now only been reported from the hydroid Schizotricha frutescens, but are reported here also observed on Nemertesia antennina (Table 3). Species of Doto fragilis complex has been found on hydroids from the genera Halecium, Nemertesia and Tubularia, but so far not on Schizotricha frutescens (see Table 3). The second D. fragilis clade consists of most often red-colored specimens from Sweden. The D. fragilis from Shipman and Gosliner (2015) are also found in this clade (Mn33151). That specimen was from Wales and was feeding on the hydroid Nemertesia ramosa. The third D. fragilis clade consists of five specimens in the analysis: one brownish light white specimen from Sweden and four specimens from Norway.

Our review of the prey species recorded for Doto species shows that most species of the genus Doto feed on several different hydroids, and hence, the previously mentioned notion of general strict food specialization within Doto (e. g. Picton & Brown, 1981; Morrow et al., 1992, 1994) should be abandoned. Not even the species D. hystrix show strict monophagy, as previously thought (Picton & Brown, 1981), but has been documented feeding on both Schizotricha frutescens and Nemertesia antennina (present study, Table 3). It is possible that different populations within a species of Doto have specialized in feeding on separate hydroids. However, the composition of hydroid species is similar along most of the European Atlantic coast, and the fauna is dominated by widely distributed species (Medel & López-González, 1998), which points against differentiation between populations.

There are other examples of nudibranchs with similarly small genetic distances as the ones found between the clusters in D. fragilis, such as between some Indo-Pacific Chromodoris species, with interspecies COI distances as low as 2.0% (Layton et al., 2018) or between species of Felimare along the American Pacific coast with a minimal interspecific COI distance of 2.5% (Hoover et al., 2016), all of these represent cases of recent speciation. It is therefore possible that the clusters we recovered are separate species, but as we have no support from nuclear markers, and the BPP analysis with only H3 did not support a separation, we do not describe them in this paper. More data and analysis are needed to further test the separation of them. “Doto fragilis” is a species complex here found in three closely related clades, one that also includes D. hystrix, and all, except D. hystrix, are externally similar. These potential new species will be described following ongoing revisions of more Doto groups. Here, to clearly indicate the clade with true D. fragilis, we designate a neotype for it (GNM Gastr. 9061. Strangford lough, Portaferry, United Kingdom, date 20 May 2014, coordinates lat 54°22,00′ long 05°32,00′, depth 20–30 m, collector: Klas Malmberg, Figure 5). The specimen is deposited at the Gothenburg Natural History Museum in Sweden. The original types of D. fragilis are lost (Mollusca Type in Great Britain, 2020; Natural History Museum, 2020). Therefore, to define true D. fragilis we here designate a neotype for this species. According to ICZN 1999 article 75.3.6. “…evidence that the neotype came as nearly as practicable from the original type locality...” The neotype of D. fragilis originates from the Northern Ireland near Portaferry, just 60 km from the original type locality of D. fragilis on the western coast of Isle of Man (Forbes, 1838), thus fulfilling the ICZN requirements. According to the original description (Forbes, 1838), D. fragilis does not contain red-colored morphs.

The genetic markers used in our study show no separation between D. hystrix and D. fragilis, despite them being distinct morphologically and ecologically. Doto hystrix has long, pointed tubercles on the rim of the rhinophore sheaths (Figure 6) and, mostly feeds on the hydroid Schizotricha frutescens (Picton & Brown, 1981, Table 3), whereas the food preferences are wider for D. fragilis. There are also differences in the spawning between the two species. This can be seen as a parallel to what is reported from, e.g., some blue butterflies (Lepidoptera: Lycaenidae) where distinct species are mixed in COI (Wiemers & Fiedler, 2007) and also from recently described species from the eubranchid nudibranchs of the genus Amphorina (Korshunova, Malmberg, et al., 2020). However, the differences in ecology and morphology are large and points toward D. fragilis and D. hystrix being separate entities, in conflict with the genetic data. Therefore, we do not synonymize them in this study. Further studies are needed to better test whether the lack of genetic separation in combination with ecological and morphological differences are due to recent speciation, with a rapid evolution of morphological characters, or interspecific differences, possibly due to adaptation to different prey species.

The northwest Atlantic species Doto formosa groups with D. fragilis and D. hystrix, but unfortunately only H3 was available for D. formosa on GenBank. Closer studies of D. formosa are needed not only in sequence data, but in morphology and ecology. The original description of D. formosa by Verrill in 1875 is very meager and could as well cover D. fragilis as well. The results from our study show that these species form a clade, but more studies are needed to resolve the taxonomical status of the taxa involved.

Prey species is an indirect method for indicating morphologically similar species of nudibranch molluscs, as the species are partially specific in their choice of prey, through coevolution with the prey (see Goodheart et al., 2017). In some cases, this is justified, as, for example, the prey of D. maculata is the hydroid Halopteris catharina, but the concept of one Doto species/one hydroid species (with the exception of D. coronata) needs to be reconsidered (see below and Table 3). We also illuminate some problems in the D. coronata complex (see Shipman & Gosliner, 2015 for a discussion about this complex).

The analysis of the hydroid associations suggests that the evolution of morphological traits and that of food preference within the genus Doto were not associated processes. For example, the morphologically and genetically well-supported species D. pinnatifida, D. fragilis and D. coronata were found associated with the same hydroid species Nemertesia antennina, whereas D. koenneckeri and D. lemchei were found on the same hydroid species Aglaophenia pluma (Table 3). The present analysis also implies that possibly other species from D. coronata complex than solely D. coronata s.str. already were recorded from several various hydroids, but these records can still be hidden under the “D. coronata” name, as a precise species identification within this complex is a considerable challenge, even for experts.

The morphology of the radula is linked to the prey specialization in any molluscs, including nudibranchs. All the studied species of Doto have the radula formula 0.1.0 (Figures 7 and 8). Though the radula in the genus Doto does not show significant differences, our data on various species reveal several promising patterns. For every species, there is some minor differences in the details of the general shape of the teeth as well as the number and arrangement of the lateral denticles. The radula of Doto millbayana (Figure 8d2–d4) has a very distinct, sharp outline of the denticles and readily distinguishes from the majority of studied here species (Figures 7 and 8), supporting its status as a separate species. In addition, there is a slight similarity between D. coronata and the D. cf. maculata from Sweden in the small central cusp.

The specimens collected in Sweden and identified as Doto cf. maculata based on coloration, were never found on Halopteris catharina even though this hydroid species occurs in both the Skagerrak and the Kattegat areas, but instead on hydroids that could be identified during the scuba dives as belonging to the family Plumulariidae. There were also morphological characters that did not fully match D. maculata. This led us to suspect that the Swedish specimens could be a separate species, or at least not D. maculata, and this was confirmed by the genetic analysis. Contrary to the D. fragilis/D. hystrix case, between putative “D. cf. maculata” and real D. coronata there are no reliable morphological and ecological differences. The dorsolateral appendages of “D. cf. maculata” from Sweden is slightly different from that of D. coronata in the absence of a pigment dot on the apical tubercle. Further, the “D. cf. maculata” specimens lack the red markings at the inner side (toward the midline of the dorsal side of the animal) of the base of the dorsolateral appendages (Figure 6) that is commonly mentioned to be “typical” for D. coronata. However, D. coronata is a highly variable species (Martynov & Korshunova, 2011; Martynov et al., 2006) (Figure 6) and for D. maculata the apparent “key features” were ambiguously indicated in the redescription by Lemche (1976: 697 “Round reddish or dark brownish spots are placed on the tips of the tubercles on the cerata, except in many cases the end one”). Hence, coloration patterns should be used with care and always be checked against other morphological characters. Doto coronata evidently feeds on a variety of hydroids (Lemche, 1976; Picton & Brown, 1981; Table 3). This was supported by the molecular analysis (Shipman and Gosliner (2015), which showed that the true D. coronata has a wide prey specificity, as it was collected from the hydroids Sertularia cupressina, Obelia geniculata and O. dichotoma. The two species D. hydrallmaniae Morrow et al., 1992 and D. sarsiae Morrow et al., 1992 were not included in the study. According to an earlier work of Morrow et al. (1992), electrophoretic methods show that they represent distinct species, but are closely related to the D. coronata species complex; however, this is not yet confirmed by other molecular phylogenetic studies.

The study by Shipman and Gosliner (2015) could not resolve any genetic difference between Doto dunnei and Doto millbayana from sequences of COI, H3 and 16S, which led them to suggest that the two could be the same species. In our analysis, we got similar result from H3, but that does not provide any further support to synonymize the two species. However, we found minor radular characters supporting the separation. In the review of hydroid prey they have only been reported from their supposed prey species, so they appear monophagous. There is a distinct difference in the shape of the pseudobranchs (rounded in D. dunnei, pointed in D. millbayana). The dark red pigment spots on the dorsum are often numerous and partly fusing in D. dunnei compared to D. millbayana in which the pigment spots are fever and generally more separated. We therefore suggest that the two should remain separate species until further study.

One D. maculata specimen from Northern Ireland groups with D. africoronata Shipman & Gosliner, 2015 and the D. maculata from Shipman and Gosliner (2015) (Figure 6), unfortunately both specimens of D. maculata lack one of the genes, which could explain why they are not forming a monophyletic group, rather than being nested with D. africoronata.

As a conclusion, the multi-locus species delimitation is shown as a valuable tool in nudibranch systematics, and that D. fragilis is a species complex, including D. hystrix and D. formosa. Doto fragilis and D. hystrix have mixed COI haplotypes, possibly due to recent speciation. The present results are concordant with the concept of multi-level organismal diversity (Korshunova, Bakken, et al., 2020; Korshunova et al., 2019; Martynov et al., 2020).