The material studied was collected using a large demersal trawl (supplemented occasionally with a small ground gear net that was deployed directly underneath the main net deployed) on banks and terraces to the west of the Outer Hebrides, Scotland, during the Rockall Haddock Survey (2010), the OFFCON Rockall Survey (2011), the Rockall Anglerfish Survey (2010, 2013 and 2014) and the Deepwater Time Series (2008, 2009 and 2011), and to the North East of Scotland during the North Sea Anglerfish Survey (2012). Further occurrences from similar surveys over the period 2005-2016 are used to improve the known distribution (Fig. 1; Suplementary material Table S1). Overall, these surveys covered a depth range of 42 to 2145 m.

All material analysed from the Scotia cruises was collected using a demersal fish trawl with both the codend and the full body of the net being thoroughly examined for specimens after each deployment. Pennatulaceans of all types, including the new specimens studied here, were rarely recorded in the codend but commonly encountered meshed in the wings or belly of the trawl. Observations from a total of 1541 trawl deployments were used to provide distributional data on the genus. Temperature data close to the seabed were recorded using either a vertical deployment of the Seabird 19+ CTD profiler or a Star-Oddi DST logger mounted on the headline of the net. Both gave temperatures of the water within 3 to 5 m of the seabed. The relevant information on materials and sampling stations of these and other comparative material used in this paper is compiled in Table 1.

Pennatulacean colonies were sorted, labelled and fixed in buffered formalin (5% in seawater). After the fixation period, colonies were preserved in 70% ethanol. A certain number of colonies were directly fixed in 100% ethanol for further molecular studies.

For comparative purposes, the morphology of additional Pennatula species was also examined. These colonies were collected during various benthic surveys and over different geographical areas: Antarctica (BIOROSS), the northeastern Atlantic-Arctic (BIOICE), the northeastern Atlantic (Scotia cruises, INDEMARES Chica), the southeastern Atlantic (BENGUELA VIII), and the Mediterranean (INDEMARES Alborán). A subsample of these were sequenced (Table 2).

The material from the Scotia cruises referred to here is deposited in the National Museum of Scotland (NMS), in the Natural History Museum in London (NHM), in the Museu de Zoologia de Barcelona (MZB) and in the collection of the research group Biodiversidad y Ecología Acuática of the University of Seville (BECA).

Sclerites of different parts of the colonies were prepared for SEM study employing the standard methodology described by several authors (e.g. Bayer and Stefani 1988), and permanent mounts were made for examination using light microscopy. Sclerite dimensions and illustrations are based on the holotype or indicated lot. Colony and sclerite terminology follows Bayer et al. (1983).

Total genomic DNA was extracted from ethanol (EtOH)-preserved specimens using the EZNA DNA kit (OmegaBiotech), following the manufacturer’s instructions. Two mitochondrial regions, mtMutS (=msh1) and Cox1, plus a nuclear region, 28S ribosomal DNA, were sequenced. These three markers are concatenated as this has been considered an octocoral barcode (McFadden et al. 2014). The start of the mtMutS region was amplified using the primers ND42599F and MUT3458R (France and Hoover 2002, Sánchez et al. 2003). The Cox1 region was amplified using the primers COII8068F and COIOCTR (France and Hoover 2002, McFadden et al. 2004). The 28S nuclear ribosomal gene (28S rDNA) was amplified using the primers 28S-Far and 28S-Rar (McFadden and van Ofwegen 2013). Each PCR used 0.5 U of DNA Stream Polymerase (BIORON), 0.2 mM of dNTPs, 0.3 µM of each primer and approximately 30 ng of genomic DNA, and was brought to a final volume of 25 µL with H2O. The mtMutS PCR was carried out using the following cycle profile: initial denaturation at 94°C for 2 min, 35 cycles of denaturation at 94°C for 30 s, annealing at 55°C for 30 s, extension at 72°C for 30 s, and a final extension at 72°C for 5 min. The Cox1 PCR used the same cycle profile with 58ºC as the annealing temperature and 40 s for extension duration on each of the 35 cycles. The 28S PCR used the same cycle as the Cox1 profile, but with 50ºC as the annealing temperature. PCR products were purified using the NucleoSpin® Extract II DNA Purification Kit, following the manufacturer’s instructions. Purified products were electrophoresed on an ABI PRISM® 3730xl genetic analyser, and sequence traces were edited using Sequencher™ v4.0.

According to the molecular phylogeny of Dolan et al. (2013) and Kushida and Reimer (2018), Pennatula species were included in Clade II. As no clear basal relationships have been demonstrated among the three pennatulacean clades, for the present study sequences of two ellisellids from GenBank were selected as out-groups. The set of new sequences obtained in this study and those from GenBank (see Table 2) were aligned using MUSCLE and then implemented in MEGA5 (Tamura et al. 2011). After alignment, the best nucleotide substitution model was selected using Modeltest implemented in MEGA 5 according to Akaike Information Criterion and hierarchical likelihood ratio test (hLRT) values. The phylogenetic reconstruction was obtained by applying maximum likelihood and Bayesian inference methods. The maximum likelihood method was carried out in MEGA 5 using the nearest neighbour interchange heuristic method and 1000 bootstrap replications. The selected nucleotide substitution model was T92+G+I for the concatenated mtMutS+Cox1+28S data set. The Bayesian inference was carried out with MrBayes v3.1.2 (Huelsenbeck and Ronquist 2001, Ronquist and Huelsenbeck 2003), using the substitution model GTR+G (lset nst=6 rates=gamma) and 107 generations, and discarding 25% of the initial trees.

In our mtMutS+Cox1+28S hypothesis (Fig. 5), the sequences obtained for Ptilella grayi n. sp. and Pt. grandis were reunited in a distinct and well-supported clade (Bst >90%, PP >0.9), far from those related to the available concatenated sequences of Pennatula species, which are also in a well-supported clade (Bst >90%, PP >0.9). In this phylogenetic hypothesis, species of different genera such as Actinoptilum, Acanthoptilum, Renilla, Ptilosarcus and Gilibelemnon are placed between Ptilella and Pennatula sequences with Ptilella diverging earlier than Pennatula.

Our results demonstrate that Ptilella and its closest morphological genus Pennatula are different by 19 substitutions in mtMutS, 27 in Cox1 and 110 in 28S. Moreover, 13 nucleotide mutations in mtMutS and 20 in Cox1 result in silent mutations, while a number of nucleotide mutations (6 in mtMutS and 7 in Cox1) imply a change of Aa (see Table S2).

An incomplete mtMutS sequence of Pt. inflata from Namibia was obtained during the molecular works carried out for this paper. Despite several amplifications, it was not possible to obtain a clean sequence that could be used in this study in order to characterize Pt. inflata. However, at least 17 of the 19 substitutions differentiating Ptilella from Pennatula in mtMutS were in agreement with its placement in Ptilella.

Pennatula grandis was succinctly described by Ehrenberg (1834: 290) based on an old specimen in the Museum of Berlin. Later on, Sars (1846: 17) described and illustrated Pennatula borealis based on material collected in Norway. The latter nomenclature and species were widely accepted thereafter by most of Sars’s contemporary octocoral researchers, such as Koren and Danielssen (1856), Milne Edwards and Haime (1857), Herklots (1858), Gray (1860), Richiardi (1869) and Kölliker (1869-72).

Milne Edwards and Haime (1857: 211) pointed out that Ehrenberg’s description of Pennatula grandis accorded well with that of Sars’s Pennatula borealis, using this last nomenclature. Richiardi (1869: 31) also considered Pennatula grandis a synonym of Pennatula borealis.

Gray (1870: 21) erected the genus Ptilella based on Pennatula borealis to include a number of distinctive morphological features including “pinnules membranaceous, broad, rounded, fringed with three close parallel series of short polype-cells on the edge”.

Kölliker (1872: 136) had the opportunity to examine the original material of Pennatula grandis deposited in the Museum of Berlin. He provided a diagnosis of the species as P. borealis and synonymized Ehrenberg’s species without being aware of the description of the genus Ptilella. However, this act was contrary to the principle of priority. Kölliker (1872) also described an additional specimen of 350 mm in length deposited in the collection of the Museum of Copenhagen.

Koren and Danielssen (1874: 422) summarized the discussion on the identity of P. grandis and P. borealis in favour of the former nomenclature by priority of Ehrenberg’s contribution. These authors also recognized the differences of P. borealis (as the type species of Ptilella) from all other Pennatula species. Their report is the first to use the nomenclature Ptilella grandis (Ehrenberg). This name was also adopted in a further contribution by the same authors (Koren and Danielssen 1877: 82).

Kölliker (1880: 4), however, did not recognize the genus Ptilella, considering that several rows of autozooids on the ventral border of the polyp leaves and the row of mesozooids on the dorsal edge of the polyp leaves could be included within the variability of the genus Pennatula. Verrill (1883, 1885) considered Ptilella a subgenus of Pennatula, as Pennatula (Ptilella) borealis Sars. Subsequent contributions by Grieg (1892: 10), Storm (1901), Jungersen (1904). Kükenthal and Broch (1910) and Broch (1913) do not consider Ptilella at all but simply mention the species as Pennatula grandis.

Despite this, Balss (1910: 54) started to form groups among the species in Pennatula that he considered valid, while Kükenthal and Broch (1911: 350) considered within the genus Pennatula a “Grandis-Gruppe” that brought together P. grandis, P. naresi and P. inflata by the presence of polyp leaves placed in oblique lines and polyps present along at least part of the dorsal margin of the polyp leaves. These authors described a poorly preserved specimen from the North Atlantic as Pennatula aff. inflata, thus clearly establishing the close relationship of P. inflata with P. grandis and P. bellissima. The intrageneric group “Grandis” (without taxonomic category) was not considered in subsequent important contributions (e.g. Kükenthal 1915, Hickson 1916: 181). However, Hickson (1937: 123) again mentioned a “Grandis-Group” when reporting P. inflata from the Maldive Archipelago, considering in that moment the possible synonym of P. borealis and P. bellissima as P. grandis, and even P. inflata as a synonym of the latter.

More recently, Ptilella was either considered a synonym of Pennatula (Williams 1995a: 126) or disappeared from the scarce specialized literature that was published (e.g. Altuna 2015). Thus, Pennatula grandis is reported in general contributions on North Atlantic benthic communities (e.g. Edinger et al. 2007, Hamoutene et al. 2008, Murillo et al. 2011, Baker et al. 2012, Baillon et al. 2012).

In the present contribution, the genus Ptilella is considered different to Pennatula because of the following set of characters:

1) the lack of zooids along the dorsal track; 2) the autozooids on the margins of polyp leaves being arranged in an oblique group of 3-4 (occasionally 2) polyps, giving the impression of several (3-4) series along the polyp-leaf border; 3) a distinctive thickening at the boundary between rachis and peduncle, often giving the appearance of an edged ring while exhibiting different sclerite morphology in the upper and lower part of this thickening; 4) the insertion of polyp leaves on the rachis being distinctly oblique and extending ventrally upward; and 5) DNA sequences (mMutS, Cox1 and 28S) being divergent from those of other pennatulacean genera (see phylogenetic analyses).

In accordance with the differences described, it follows that the diagnosis of Pennatula needs to be amended to include characters that separate it clearly from Ptilella.

Colonies pinnate. Symmetry of rachis bilateral throughout. Rachis-peduncle limit just under the lower polyp leaves, as a narrowing, without a distinct thickening or swelling with a prominent edged ring at the rachis-peduncle limit. Axis circular in section, present throughout length of colony. Polyp leaves present, usually large and conspicuous, deltoid, sickle-shaped or fan-shaped. Autozooids in one single row along the ventral edge of polyp leaves. Anthocodiae retractile into permanent spiculiferous calyces. Calyces tubular, with eight terminal teeth. Siphonozooids on rachis, and between polyp leaves. Mesozooids on rachis, or on basal dorsal margin of polyp leaves. Sclerites three-flanged needles on calyces, mesozooids, siphonozooids, inconspicuous three-flanged rods on surface of peduncle, and small ovals in interior of peduncle (mostly >0.1 mm long).

Although a revision of the species in the genus is needed, according to Williams (1995a) the following list of species can be tentatively considered valid in the genus Pennatula: P. aculeata Danielssen, 1860 (North Atlantic); P. delicata Tixier-Durivault, 1966 (Madagascar); P. fimbriata Herklots, 1858 (Japan, Philippines); P. indica Thomson and Henderson, 1906 (Indian Ocean); P. phosphorea Linneaus, 1758 (northeastern Atlantic and Mediterranean, Cosmopolitan ?); P. rubra (Ellis, 1761) (Mediterranean Sea); and P. prolifera Jungersen, 1904 (North Atlantic).

As we state above, our phylogenetic hypothesis based on mitochondrial and nuclear markers places the type species of Ptilella, (Pennatula grandis), far from those sequences from the type species of the genus Pennatula, (P. phosphorea), or at least, taking into account that the genus Pennatula is in need of revision from the sequenced species and specimens sharing the morphological characters defined for Pennatula phosphorea in this paper and the most relevant references (e.g. Kükenthal 1915, Kükenthal and Broch 1911) (see Figs 5 and 6).

In this paper, four Atlantic species are considered to be in the genus Ptilella (Pt. grandis, Pt. grayi n. sp. Pt. inflata, and Pt. bellissima) on the basis of a number of morphological features. Corroborative molecular information is currently available for two of them (Pt. grandis, Pt. grayi), and partially so for Pt. inflata (see above). Molecular grade tissue is, however, lacking from the old single specimen of Pt. bellissima due to the preservation method used.

In the late 19th century, Professor Albert R. von Kölliker described four Pacific sea pen species and ascribed them to the genus Pennatula: P. naresi Kölliker (1880: 2), P. pearceyi Kölliker (1880: 4), P. murrayi Kölliker (1880: 5), and P. moseleyi Kölliker (1880: 6). The original descriptions of these species agree with some of the characters considered here for Ptilella: 1) the obliquely inserted polyp leaves on the rachis; 2) at least two series of autozooids on the ventral edge of the polyp leaves; 3) the dorsal track free of zooids; and 4) mesozooids on the proximal dorsal part of the polyp leaves. While taking into account of the current state of knowledge of these species, their morphological features thus support their tentative inclusion in Ptilella here. However, it is important that their position be reassessed following an improved morphological and molecular understanding of these species in the future.

Ptilella grandis and Pt. grayi n. sp. are clearly differentiated from Pt. bellissima and Pt. inflata by their more rigid and spiculose colonies (especially visible on the upper and lower polyp leaf surfaces), a higher overall number of polyp leaves and a higher number of autozooids per polyp leaf. Moreover, the two former species have a much stronger separation between rachis and peduncle with an edged ring (see Figs 2A, S1A, and S2E for Pt. grayi n. sp.; Figs 7A and 8C for Pt. grandis). This structure is much more marked in Ptilella grandis and Pt. grayi n. sp. than in Pt. bellissima (see Figs 7C and 8I) and Pt. inflata (see Figs 7B and 8F), where the upper and lower surfaces are more continuous in structure.

Ptilella bellissima was described by Fowler (1888) based on a single specimen from the Bahamas (northwestern Atlantic) as Pennatula bellissima, now deposited in the Natural History Museum in London (BMNH 1880.6.28.1). More than a century later, Castro and Medeiros (2001: 154) described a new species of a large Pennatula (up to 350 mm in total length) from the Brazilian coast as Pennatula bayeri. After the comparison of both original descriptions and the holotype of Pt. bellissima, we believe P. bayeri to be a junior synonym of Pt. bellissima. As mentioned by Fowler (1888), Castro and Medeiros (2001), and as illustrated in the present paper, Pt. bellissima differs from Pt. inflata by its slender triangular and less fleshy polyp leaves, and well-spaced (sometimes difficult to follow) groups of 2-3 autozooids along the ventral edge of the polyp leaves. Castro and Medeiros (2001: 157) discussed the possibility of synonymy between Pennatula bayeri (=Pt. bellissima) and Pt. inflata, but we consider them as two clearly morphologically distinct taxa. It is important that future molecular studies (based on appropriately fixed material) should further explore the internal relationships of Ptilella species. Particular attention should be paid to further specimens of Pt. bellissima, in order to correctly evaluate the variability in the distribution of mesozooids around the insertion point of the polyp leaves and the autozooids.

The most immediately evident feature differentiating Ptilella grandis and Pt. grayi n. sp. is the colour: dark-red to red-brown both alive and preserved in the former but light fleshy to dirty white in the latter in the preserved state and light yellow to light pink in living specimens. In the examined material the maximum observed lengths of sclerites in Pt. grandis and Pt. grayi n. sp. tends to be overall slightly larger in the former species (calyces 1.70 vs. 1.40 mm; mesozooids 1.0 vs. 0.80 mm; siphonozooids 0.80 vs. 0.50 mm; polyp leaves 1.11 vs. 1.40 mm; dorsal track 0.50 vs. 0.40; rachis-peduncle limit 0.50 vs. 0.35 above edged ring and 0.16 vs. 0.17 mm below edged ring).

The molecular comparison of Pt. grayi n. sp. vs Pt. grandis revealed the following substitutions: at the positions 249 (A/C) and 672 (C/A) in mtMutS; at the positions 372 (A/G) and 624 (C/A) in Cox1; and at the positions 166 (C/A), 210 (A/G) and 519 (G/C) in 28S (see Table S3). In mtMutS, both substitutions imply the change of Aa (Met by Ile, and Phe by Leu, respectively), although in both cases all Aa are non-polar. In Cox1, both are silent substitutions. Although the genetic distance obtained in mtMutS did not exceed 0.30%, the more conserved mitochondrial marker Cox1 also showed differences between these species, which are additionally supported by the differences in the nuclear 28S.

According to our current knowledge, Pt. grayi n. sp. seems to have an overlapping but somewhat more geographically restricted and shallower bathymetric distribution than Pt. grandis (145-389 m and 90-2700 m in depth, respectively). Moreover, the new species is only known, for the moment, from the NE Atlantic, though Pt. grandis has been reported from both sides of the North Atlantic Ocean. The densities of Pt. grayi n. sp. on Rockall Bank are presently unknown and cannot be inferred from swept area estimates using numbers caught per haul, as it is likely that demersal trawl is an inefficient method of capturing large sea pens that are anchored in the sediment. Moreover, the exact design of trawl used varied with the survey that the specimens were obtained from.

On the polyphyletic family Pennatulidae and the familial relationships in Clade II

The traditional family Pennatulidae comprises six genera: Gyrophyllum, Pennatula, Ptilosarcus, Sarcoptilus, Crassophyllum and Pteroeides (Williams 1995a, 1995b). Dolan et al. (2013) showed a consensus phylogenetic tree suggesting a non-monophyletic Pennatulidae. Although the monophyly of Pennatulidae was not formally rejected by an SH test, in that study this family formed a non-natural group, placing the genera Pteroeides, Pennatula, and Gyrophyllum into three different clades (Clades I, II and III, respectively). Clade II based on two mitochondrial markers (mtMutS+ND2) included the genera Pennatula, Renilla, Distichoptillum and Protoptilum, the two former genera being sister groups. Later, results offered by Kushida and Reimer (2018) based on those same mitochondrial (mtMutS+ND2) again confirmed the polyphyletic nature of the current Pennatulidae, showing the same distribution of the genera Pteroeides, Pennatula and Gyrophyllum. In that study, Clade II again included the genus Pennatula as the sister group of Renilla, and both related to genera of different families: Echinoptilum (Echinoptilidae), Stachyptilum (Stachyptilidae), Distichoptilum and Protoptilum (both Protoptilidae), and Scytalium and Stylatula (both Virgulariidae). In our present phylogenetic analyses for Clade II we present a more complete molecular coverage, including two mitochondrial (mtMutS, Cox1) markers and a nuclear marker (28S), showing a well-supported genus Pennatula with Ptilosarcus as the sister group, with Renilla being the sister group of Acanthoptillum (see Fig. 5). Pennatula species were well separated from the resurrected genus Ptilella, which diverges early. Future sea pen studies utilizing a more complete set of taxonomic units (families, genera and species) and molecular markers may completely change our previous understanding of the distribution of genera among all the possible familial units in this specialized group of octocorals. We recognise the current incomplete and unstable state of taxonomic knowledge of the Pennatulidae and our placement of Ptilella within this family is to be regarded as tentative. We consider this as pending a thoroughly comprehensive review that will also achieve the segregation of diagnostic morphological characters across those well supported familial units.