Food & Vegetables

Towards a phylogenetic classification of Leptothecata (Cnidaria, Hydrozoa)

In this section we contrast the new taxonomic classification proposed herein with the traditional one (Figs 1, 2, 3, 4; Tables 2, 3, 4 and Supplementary Table S1; number of species for each group based on WoRMS database2) and discuss some macroevolutionary patterns that may be associated with the diversification of Leptothecata.

Molecular markers and evaluation of phylogenetic results

The general outline of our results is consistent with recent hypotheses41,45,46, but some particular aspects should be emphasized. The mitochondrial marker 16S accumulates greater nucleotide variation than the nuclear markers 18S and 28S, which may be seen by the differences in conserved sites, singletons and informative sites for parsimony, in relation to the total sequence size (Table 1). The 16S sequences are shorter and display a relatively greater proportion of filtered sites than 18S and 28S (Table 1). Thus, average support in 16S trees is expected to be lower for non-parametric methods, especially in the case of bootstrap.

Rogue taxa and “Lineages” found in the analysis of the 16S matrix have lower nodal support and greater tendency to change their phylogenetic position according to the level of filtering applied to the concatenated matrices. This is especially true for terminals with abundant missing data, such as H. gardineri and S. gigas (Figs 1, 2, 3; Supplementary Information, “Rogue taxa analysis” section). Taxonomic artifacts associated with 16S include long-branch attraction (LBA, also see Peña Cantero et al.46), such as between Siphonophorae and Plumularioidea, the group with the greatest variation in ribosomal data in Leptothecata (see Supplementary Figs S3-5 online). Previous studies attempted to get around this issue by not including Siphonophorae species as part of outgroup sampling41. Assuming a somewhat constant evolutionary rate, it is expected that basal lineages are more variable with greater saturation. While the use of 16S recovered most of the lineages of Plumularioidea, the LBA artifact results in the accumulation of nucleotide variation, causing non-historical convergence of Siphonophorae and Plumularioidea, which are groups with substantial differences in their life cycles, such as the loss of the medusa stage, strong polymorphism and colony specialization. Interestingly, these taxa bear superficial resemblance, such as the lack of the medusa stage, strong polymorphism and high zooid specialization.

Most non-Plumularioidea lineages are not monophyletic in the 16S analysis, in contrast with patterns from the nuclear markers 18S and 28S, both of which result in similar phylogenies, with 18S having lower resolution (Figs S6-8). The 28S provides a robust phylogenetic framework for Leptothecata due to the number of variable nucleotides, relative gene frequencies in the different matrices (with and without filtering) and support values in the individual phylogenies. Basal nodes in the 28S phylogeny and other phylogenetic groups (e.g., Sertulariida taxon novum) are recovered in the concatenated data matrices, with higher support when compared with nodal support of single-gene phylogenies.

Analyses of data matrices assembled under a range of criteria (combined or individual markers and filtering taxa and/or nucleotides at different levels) and employing different branch support methods, allow a better evaluation of the nodal stability by assessing congruence among results (e.g., Table 2; Supplementary Information “Rogue taxa analysis” online). Topologies change with the level of filtering, usually with a concomitant decrease in nodal support. This indicates that the highest filtering level “throws the baby out with the bath water”, i.e. eliminates information together with spurious phylogenetic signal. Intermediate levels of filtering seemingly strike a more reasonable signal-to-noise ratio. Additionally, simulations strongly indicate that non-parametric bootstrap is the branch support method most prone to type II error (i.e. failing to support a “true” clade)48. Therefore, by employing alternative support methods, our working hypothesis has recovered a greater number of significantly supported clades than what would be obtained by employing traditional bootstrapping (see Methods for details). All the taxonomic propositions in the following sections were thus based on this hypothesis (Fig. 1).

Outgroups and sister groups of Leptothecata

The sister group of Leptothecata remains unclear35,45, with different taxa being included or excluded in previous analyses of Leptothecata (e.g., exclusion of Siphonophorae41). Our topologies resulting from combined matrices, did not recover a well-supported sister group for Leptothecata. The only high nodal support for a sister group of Leptothecata was found in the 28S marker analyses (28S_N and 28_Nrw4), in which Bougainvilliidae (Filifera IV after Cartwright et al.35) is well-supported as its sister group (basal relationships similar to 16S18S28S). However, more data are definitely necessary to confirm the placement of this family.

Previous conflicts in leptothecate phylogeny and a new working hypothesis

Molecular studies based on ever-increasing number of species have contradicted taxa considered to be monophyletic under a traditional sense, such as Conica, Proboscoida and Plumulariida45,49. However, our results also corroborate some traditional taxa defined by their exoskeleton and specialized characters of polyps, as in the reformulated Plumularioidea (Fig. 4, Table 2). Ours is the most inclusive analysis of Leptothecata to date, which recovered well-supported clades based on multilocus data matrices and yielded a robust, working hypothesis (Figs 1 and 4; Tables 2, 3, 4). However, our proposal for new names is conservative and focuses only on the more stable and robust nodes, often corroborated by morphological synapomorphies, aiming to minimize changes in the current classifications (Figs 1 and 4). Instead of filtering “problematic terminals” from our main results, we considered that they should be included in the main results because these specimens/molecular markers are available in GenBank as any other sequence. We acknowledge that this strategy resulted in somewhat controversial or not well supported positions and therefore some of them are considered. Leptothecata incertae sedis (Lineages 1 and 2, rogue taxa), specific clades’ incertae sedis (Lineages 3-5) and only a few as dubious terminals (e.g, potential taxonomic misidentifications).

Lafoeida sensu novum

Lafoeida sensu novum comprises only the three species sampled here by now and is presently defined by having stolonal or erect colonies, polyps without abcaulinar caecum, hydrothecae without operculum and diaphragm (considering other potential genera) and fixed gonophores, usually forming coppinia (aggregation of gonothecae that can include nematothecae4).

Previous analyses had placed Hebellinae, Zygophylacinae and Lafoeinae into the Lafoeidae (cf. Marques et al.44) (Hebellidae and Zygophylacidae, respectively; Tables 3, S1, Figs 1 and 4), or related Zygophylacinae with Sertulariidae (cf. Marques et al.44). In our hypothesis, assumed to have a limited number of lafoeid species (Fig. 1, Table S1), Lafoeidae is the only nominotypic group of Lafoeida, not related to the “Hebellidae” and Zygophylacidae (see below), which is in agreement with previous studies35,43,44,45. Future addition of taxa will test that more properly.

Other molecular analyses indicate that Lafoeidae includes species of Filellum, Lafoea, Cryptolaria and Acryptolaria (Moura et al.43,50, present study), but its taxonomic range could be wider, because Grammaria (6 species) and Cryptolarella (1 species) were not included in our analyses. Altogether, the relationship of both genera with other Lafoeidae has been supported by morphology44,51. Another pending question about phylogenetic affinities for this family is its relationship with Clathrozoidae25,26, a family with two monospecific genera that has not been sampled in any published molecular phylogenetic analysis.

Hincksella sp. (traditionally considered as part of Syntheciidae) is the terminal that completes this clade. Morphological similarities between Syntheciidae and Lafoeidae species have already been identified52. Morphologically, the trophosome of Hincksella is similar to Lafoeida (e.g., Acryptolaria and Lafoea) because of the tubular hydrotheca. However, the proximal region of the hydrotheca of Hincksella has a distinct floor perforated by a hydropore53; additionally, species of Lafoeida, such as those of Acryptolaria and Lafoea, have coppinia (absent in Hincksella; but see Millard23 and Calder53 for opposing arguments on the affinity between Hincksella and Lafoeidae). Nonetheless, the hypothesis can only be tested with the inclusion of Parathuiaria (1 species) and Synthecium (23 species) in a broader analysis including multilocus data for Hincksella.

Additional references: Billard54; Rees & Vervoort55; Vervoort56; Migotto & Marques57; Schuchert58; Marques et al.59,60; Bouillon et al.30; Peña Cantero et al.51.

Stegella lobata and Melicertum octocostatum (Lineage 1 – L 1)

Stegella lobata belongs to the monotypic genus Stegella Stechow, 1919, which was previously associated with Campanulariidae61,62 and later included in Campanulinidae (Stechow 1919; originally as Stegella grandis63). It is characterized by pedicellate hydrothecae, rhizocaulomic colonies (hydrorhiza erect in parallel stolons), operculum with four flaps and fixed gonophores64,65.

Melicertum octocostatum belongs to the only genus of Melicertidae whose polyp stage is known. This family is divided in 4 genera and 8 species and is characterized by stolonal colonies with unusual morphology when contrasted with the typical “thecate” polyp: absence of hydrotheca around the hydranth and probably with no gonotheca surrounding gonophores30. Medusae of this taxon have no ocelli or other marginal sensory structure4, but the gonads are on the radial canals, which is a typical character of the medusa of Leptothecata. In our phylogeny M. octocostatum is the basalmost leptothecate species with medusae with gonads on the radial canals.

The clade defined by S. lobata and M. octocostatum (Melicertidae) has a basal position in Leptothecata (Figs 1 and 2), similar to Leclère et al.45 but different from Peña Cantero et al.46. This group varies in both its affinities and nodal support, depending on the analysis (Leptothecata incertae sedis; Figs 1, 2, 3). Hopefully, sampling a higher number of species and additional markers will elucidate the relative positions of S. lobata and M. octocostatum. The same goes for other unsampled families of the superfamily Dipleurosomatoidea (Orchistomatidae – 6 species, Dipleurosomatidae – 8 species, Table S1).

Additional references: Bouillon25,26; Govindarajan et al.40.

Laodiceida taxon novum

Laodiceida taxon novum includes two families with medusae in their life cycles (Laodiceidae and Tiarannidae), although the medusa stage is still unknown in several species. However, in species where it is known, the medusa is characterized by the presence of gonads along the radial canals in Laodiceidae and on the interradial walls of the stomach (or in adradial pouches) in Tiarannidae. Both groups have the pelagic stage with no statocysts but with cordyli4. The benthic stage has stolonal and erect (Stegopoma spp.) colonies, hydrothecae with operculum and diaphragm slightly developed or absent.

Monophyly of Laodiceida taxon novum and its phylogenetic position within Leptothecata are well supported (Table 2, Figs 1, 2, 3, cf. Leclère et al.45). According to our working hypothesis, Tiarannidae is not monophyletic and Laodiceidae becomes monophyletic only if Staurodiscus gotoi is not considered as part of the family (Fig. 1). Indeed, medusae of S. gotoi are morphologically similar to those of Hebella (Migotto & Andrade66; cf. Leclère et al.45; Moura et al.43), suggesting homoplasy of the marginal cordyli in Staurodiscus and species that truly belong to Laodiceidae.

Discussions on the validity of the families of Laodiceida taxon novum have been focused on the position of the gonads in the medusae4,25,26,66 and some uncertainties still remain because several genera are poorly sampled, their life cycles are not fully known and the polyp identification is complex, since their morphology is generally characterized as “cuspidella” type30. Also, Staurodiscus (type species Staurodiscus tetrastaurus Haeckel, 1879), whose monophyly remains uncertain, further confounds these issues.

Additional references: Pagès et al.67; Vito et al.68.

“Hebellidae”, Staurodiscus gotoi (Lineage 2 – L 2)

This is a non-monophyletic lineage including three species of “Hebellidae” (Anthohebella parasitica, Hebella scandens, Hebella venusta) and S. gotoi (see considerations on Laodiceida taxon novum), with high support for the relative position of these species (Figs 1, 2, 3). Similar results were recovered elsewhere with a smaller set of “Hebellidae” species45. Also, the proximity between S. gotoi and “Hebellidae” reinforced previous considerations about misinterpretations of life cycles in these groups. The phylogenetic proximity between Scandia gigas and Haleciidae (Fig. 2), contrasts with its placement close to Hydrodendron gardineri and H. mirabile in Fig. 3 (genus traditionally considered as Haleciidae; Figs 1, 2, 3, Table 3) and also with its placement in our working hypothesis, firmly set in Sertularellidae (rogue taxon; Fig. 1). As already remarked, such instability might be attributed to the 16S characters, thus we prefer to regard Lineage 2 as Leptothecata incertae sedis.

The phylogenetic basal position of L 2 and their life cycles (Hebella spp. and S. gotoi have medusae, Anthohebella has “swimming gonophores”) corroborate the medusa as an ancestral condition in Leptothecata life history (Fig. 1). According to this hypothesis, the absence of medusae in the life cycle of Leptothecata would be a derived condition. “Hebellidae” and S. gotoi have stolonal colonies with campanulate hydrothecae provided with a diaphragm or annular thickening and morphologically simple gonothecae44. While morphologically similar, other genera of “Hebellidae”, such as Bedotella (1 species), with fixed gonophores and Halisiphonia (5 species), which apparently has the medusa stage, need to be included in molecular and life cycle reconstruction analyses33,69.

Additional references: Calder53; Cornelius4; Boero et al.70,71; Migotto et al.72.


Statocysta are characterized by having statocysts, a typical equilibrium structure of the medusa stage45, indicative of a more active swimming. Two main lineages are well supported (Fig. 1): Campanulinida sensu novum and Eirenida taxon novum + Proboscoida sensu novum. Opercularella lacerata has a dubious position in our phylogenetic hypothesis (rogue taxa; Figs 1, 2, 3, Table 3). The traditional division of Leptothecata into Conica and Proboscoida is not supported because Statocysta combines parts of each one of these “taxa”. Proboscoida, in its traditional sense, might be re-interpreted as a derived clade within Conica (Figs 1, 2, 3, 4, Tables 3, S2). Future studies may infer alternative sister groups for Statocysta, especially considering those lineages included in Lineage 2 (“Hebellidae” and S. gotoi) and O. lacerata (rogue taxa).

Campanulinida sensu novum

This taxon includes species of several families and superfamilies of the traditional taxonomy (e.g., Mitrocomidae and Tiaropsidae), but its composition does not support the monophyly of Campanulinida in a traditional sense (cf. Bouillon25,26). Due to the difficulty of defining diagnostic characters for families and genera traditionally placed in Campanulinida4, the lack of information on life cycle and the difficulty in treating species with unrecognized medusae (e.g., Calycella and Campanulina; Bouillon et al.30), the monophyly for these genera and families have often been questioned. A clear example of the need to understand more about life cycle strategies in Hydrozoa is found in Lafoeina (Campanulinidae) and Cirrholovenia. The benthic stage of Lafoeina species are mistakenly mixed with species of other taxa (e.g. Cuspidella species), in some cases with unknown medusa stage and in others with young medusae strikingly similar to Cirrholovenia (genus of Lovenellidae, not represented in molecular studies72). Nomenclaturally, Campanulinidae is nominotypic, but rather a complex of species, whose problematic taxonomic status and identity have already been debated by other authors53. To solve the “campanulinid” taxonomy will be a major task in itself, beyond the scope of this manuscript. Therefore, we are aware that our hypothesis of Campanulinida sensu novum must be more broadly tested, including sequences of the type species of the genus Campanulina van Beneden, 1847 (Fig. 1).

None of the traditional lineages in Campanulinida is supported by our data, except the monophyly of Mitrocomidae (Figs 1, 2, 3, Tables 3, 4; Bouillon26). This is not surprising because there are no synapomorphies proposed for supraspecific taxa in the traditional Campanulinida1. In synthesis, morphology of the sampled species varies in the clade from the “cuspidella” type (stolonal colony, polyp with operculum but lacking diaphragm) to “generic campanulinida” (stolonal or erect colony, polyp with operculum and diaphragm). In those species with known medusa, this stage has four simple radial canals, many statocysts, hollow tentacles and no cirri (except Mitrocomella)4,26.

Eirenida taxon novum

This clade comprises some of the remaining families traditionally assigned to Campanulinida (e.g., Aequoreidae, Eirenidae, Lovenellidae28). Two groups are revealed in our working hypothesis: Eirenids I and Eirenids II.

– Eirenids I

If corroborated, this clade will eventually become the nominotypical taxon because it includes species of the family Eirenidae Haeckel, 187925,26 (non-monophyletic) and the genus Eirene Eschscholtz, 1829 (non-monophyletic), besides the type species of the genus, Eirene viridula (Péron & Lesueur, 1809). It also includes several genera of Campanulinoidea and Eutonina (one of the two valid species in the taxon).

In an attempt to overcome the difficulty in defining diagnostic characters in Hydrozoa, Bouillon25,26 proposed the non-monophyletic clade “Campanulinoidea”, characterized by “medusa without exumbrellar cirri” (with nomenclatural problems, N.B. the phylogenetic hypothesis in Bouillon25: p. 12). In this hypothesis, a non-monophyletic Campanulinoidea included the problematic “Campanulinidae” (Fig. 4. and Table 3). Also, Eirene is complicated to diagnose, requiring the presence of the medusa stage for an adequate taxonomic identification; additionally, the medusa stage may also be found without cirri. Other groups external to Eirenids I have medusae lacking cirri (e.g., Phialellidae) and therefore, the state of character “medusa without cirri” is homoplastic.

Polyps of Eirenids I are “campanulinida” type, with stolonal colonies (except Aequorea and Eirene), operculum formed by flaps without clear demarcation from the hydrotheca, hydranths with intertentacular membrane and hydrotheca with diaphragm. Medusa has gonads on the radial canals, distant from the manubrium, ocelli and cirri are absent and there are many closed statocysts (Cornelius4, Table 3). Internal relationships usually have weak support (Figs 1, 2, 3).

Additional references: Bouillon & Boero29; Govindarajan et al.73; Kubota74; Bardi & Marques75; Leclère et al.45.

– Eirenids II

This clade comprises two subclades with high support (Figs 1, 2, 3): “Eirenidae” (polyphyletic traditional family, see above for considerations about Eirenids I) and two species of Lovenellidae (Eucheilota menoni and Eucheilota maculata; another species of Eucheilota appears in Statocysta Lineage 3 along with Lovenella gracilis).

One of the two main clades in Eirenids II includes the monophyletic genus Helgicirrha (with 2 sampled species), Hydranthea (considered as Lovenellidae by Bouillon et al.30) and Eucheilota (Lovenellidae). Its sister group includes species of Eutima and Eugymnanthea japonica, nested amongst them76. Eirenids II is interesting due to both morphology and lifestyle of the species: those of Eugymnanthea and Hydranthea produce eumedusoids and inhabit the mantle cavity of bivalves.

Eirenids II have “campanulinida” colonies, with reduced hydrotheca and hydranths protected by a slender perisarc and intertentacular membrane; medusa with gonads on the radial canals, lateral cirri and closed statocysts30.

Proboscoida sensu novum and Lineage 3 (L 3)

Originally, Proboscoida included Bonneviellidae and Campanulariidae14 and later also Phialuciidae, being characterized mainly by the polyp with trumpet-shaped hypostome26. Except for Phialuciidae (monotypic family not included in this analysis), our topology provides support for phylogenetic proximity of the remaining Proboscoida, with the inclusion of Statocysta incertae sedis Lineage 3 (defined by Eucheilota bakeri+Lovenella gracilis and Eirene ceylonensis+Eirene menoni) as part of dubious results. In the Lovenellidae clade, our analysis groups E. bakeri with L. gracilis (Fig. 1). The origin of the GenBank sequences of E. bakeri is unknown to us and we suspect the material could have been incorrectly identified (part of conflicts are related to sequences’ metadata77). We and other specialists (P. Schuchert, pers. comm.) agree on this possibility and it would be necessary to examine the voucher to assure its identification. Historically, Lovenella has been associated to Campanulariidae (Hincks8, based on polyp characters), Lovenellidae (Russell78, based on medusa characters) and Campalecium (family Haleciidae, based on the “haleciid type” morphology of the polyp30). Hypotheses for Lovenellidae also included Eucheilota (Russell 1953), Cirrholovenia (Kramp 1959) and Hydranthea (see historical record in Miranda et al.79). Together with Eirenidae species, efforts on sampling additional molecular data for species of Lovenellidae are necessary to confirm its monophyly, or even to consider a possible resurrection of Eucheilotidae Bouillon, 1984. In our analysis, Proboscoida did not diverge firstly in Leptothecata as sister group of Conica, in contrast to the original proposal of the group (cf. Bouillon28).

Proboscoida sensu novum includes two highly-supported main clades, Campanulariida sensu novum and Obeliida taxon novum, the latter with high nodal support for apical groups as well (Figs 1, 2, 3, 4, Tables 3 and 4).

– Campanulariida sensu novum

This taxon includes species of the traditional families Campanulariidae and Bonneviellidae. It is characterized by stolonal and erect colonies, occasionally polysiphonic (as in Rhizocaulus), with medusae or eumedusoids present (e.g., Orthopyxis; Cornelius24). No genus is recovered as monophyletic in our main results. Therefore, unresolved systematic questions, already discussed in literature80, still remain, such as the monophyly of Campanularia and Orthopyxis, which is related with taxonomic interpretations based on the presence/absence of diaphragm and the shape of hydrotheca (e.g., Millard23).

Additional references: Berrill81; Cunha et al.80; Govindarajan et al.40.

– Obeliida taxon novum

Obeliida taxon novum embraces the remaining families of Proboscoida, namely Clytiidae and Obeliidae (Table 4). Obeliidae is monophyletic with high nodal support (Fig. 3, topology 16S18S28S_Nrw4). Clytiidae sensu novum is monophyletic as well, despite the presence of “Eirene brevistylus” (probably due to a long-branch attraction artifact on 16S sequences82 or to taxonomic misidentification) – presently this specimen is considered as incertae sedis. This clade is characterized by the presence of hydrothecae with diaphragm and a wide variety of reproductive strategies: presence of uniform medusae (Obelia), meconid medusae (Gonothyraea) and fixed gonophores (Laomedea)5,24,83.

Additional references: Cornelius83; Govindarajan et al.84; Lindner et al.85.

Billardia subrufa (Rogue taxa 4)

The taxonomy of Billardia Totton, 1930 remains uncertain and the genus has been morphologically associated either with Campanulariidae86, Lafoeidae and Syntheciidae23,24. Billardia colonies are erect, with no diaphragm or with an “annular thickening” at the base of the hydrotheca; gonophores are fixed and protected by a long, ringed gonotheca52,87. We obtained new sequences of the antarctic Billardia subrufa, similar to those of GenBank. Previous molecular analysis placed B. subrufa near to Lafoeidae, at the base of Leptothecata40,45. In our results, however, B. subrufa is at the base of Macrocolonia or proximal to Lafoeidae+Lineage 1 (Figs 1, 2, 3). The unstable position and low nodal support led us to consider B. subrufa as a leptothecate incertae sedis.

Additional references: Cornelius5.


Macrocolonia was proposed together with Statocysta, including groups with complex and erect colonies and encompassing species of Plumularioidea, Sertulariidae and Haleciidae45. It mostly coincides with the traditional proposal of Plumulariida and Haleciida (original Haleciida included the family Syntheciidae; Bouillon25,28; Fig. 4, Table 3), being characterized by “Hydranthes à hypostome conique, et dont le feuillet endodermique est différencié em deux régions distinctes, hydranthes rétractables dans leurs hydrotèques, Celles-ci ont une symétrie bilatérale, sont dépourvues de diaphragme mais leur plancher est perforé par un hydropore” (Bouillon25, p. 20)88. Macrocolonia has the greatest species richness within Medusozoa, with the medusa stage reduced and with fixed gonophores in most part of the species. Our hypothesis emphasizes the proposal of some new taxa, as well as readjustments in already defined taxa. We do not find it necessary to reinterpret Macrocolonia. Rather, we point out the relative positions of both Macrocolonia incertae sedis: Lineage 4 (Zygophylacinae) and Lineage 5 (Nemalecium lighti and H. mirabile), together with interesting evolutionary aspects concerning Staurotheca and Schizotricha.

Staurothecida taxon novum – Staurothecidae fam. nov. and Symplectoscyphidae fam. nov

At the base of Macrocolonia, Plumulariida, Sertularioidea and Sertulariidae are not monophyletic, which contradicts the traditional taxonomic view of these taxa within Leptothecata5. In our present understanding, a basal branch of Macrocolonia includes a weakly supported clade that we have named as Staurothecida taxon novum to preserve the taxonomic logic of Leptothecata. Staurothecida taxon novum includes two new well-supported families that support previously described patterns46: Staurothecidae fam. nov. and Symplectoscyphidae fam. nov.

Staurothecidae fam. nov. includes the type-genus Staurotheca Allman, 1888 and its species, which were previously attributed to Syntheciidae23,25,26,28,89,90 and Sertulariidae91. Morphologically, Staurothecidae and Staurotheca are characterized by tubular hydrothecae, with operculum and diaphragm arranged either in two longitudinal rows of opposite, sub-opposite or alternate hydrothecae or in decussate series. Male and female gonothecae are directly inserted at the hydrothecal base, with female gonothecae sometimes arranged on special supporting structures (Peña Cantero & Vervoort92, p. 2664).

Symplectoscyphidae fam. nov. includes the type-genus Symplectoscyphus Marktanner-Turneretscher, 1890 and the genus Antarctoscyphus, both of them historically related (cf. Peña Cantero et al.91). Like Staurotheca, Symplectoscyphus is also a monophyletic taxonomically stable taxon (e.g., Peña Cantero et al.46). We corroborated this taxon and we characterize Symplectoscyphidae fam. nov. by the hydrothecal aperture provided with three cusps and by an operculum of three valves. Nonetheless, the genera vary in colony morphology, branching pattern and gonotheca morphology (Peña Cantero et al.91, p. 24). Symplectoscyphus curvatus has to be taxonomically revalidated, since it is sister group of Antarctoscyphus species (Figs 1, 2, 3). Also, the identity of Symplectoscyphus turgidus specimen used here was confirmed by P. Schuchert (pers. comm.) and therefore, its placement within Sertularella has to be re-evaluated.

Additional references: Peña Cantero et al.46,90,91; Peña Cantero & Vervoort92,93; Peña Cantero94.

Haleciida, Halecioidea, Haleciidae sensu novum

Haleciidae sensu novum (and the redundant taxa Halecioidea and Haleciida) encompasses only species of Halecium Oken, 1815. It is a homogeneous and monophyletic taxon, sister group of Sertulariida taxon novum and Plumupheniida taxon novum. The complex taxonomic history of the taxa was reviewed by Calder53, Schuchert95 and Peña Cantero96. Some genera previously attributed to Haleciidae were placed in Plumulariidae and Lafoeidae due to the presence of nematothecae97, which are absent in species of Halecium23. In our topology, the genera Hydrodendron and Nemalecium have an alternative position, related to Plumularioidea (Lineage 5; Figs 1, 2, 3).

Halecium colonies are erect and ramified, often with free hydrophores. The hydrotheca is shallow, with conspicuous proximal desmocytes and with hydranth not retractable into it. The gonophores are fixed and protected in solitary gonothecae or aggregated in glomulus, usually dimorphic (cf. Calder53).

Additional references: Cornelius5; Schuchert95; Peña Cantero96.

Zygophylacinae (Lineage 4)

Zygophylacidae was considered several times to be a subfamily of Lafoeidae (Zygophylacinae) due to the presence of bilateral and tubular hydrotheca44,53. Here, the sampled species of Zygophylacinae occupy alternative positions in Macrocolonia as sister group of Sertulariida+Plumupheniida (Fig. 1), Haleciida (Fig. 2) and Staurotheca (Fig. 3). All these relative positions with low support are likely to be related to the exclusive use of 16S marker in our analyses for these species. Zygophylacinae has already been considered as Sertulariidae, related to Abietinaria, due to the form of the colonies, the presence of operculum (present in Abietinella) and the hydrothecal shape98. Despite being within Macrocolonia, the species of Zygophylacinae here analysed have characters not found in other members of Sertulariida, Staurotheciida or Haleciida sensu novum, such as the presence of nematothecae and gonothecae aggregated into coppinia or scapus44. Because of their incertae sedis position among Macrocolonia and features as defensive polyps (as present in Plumupheniida), a multilocus analysis would better define their position. Their basal “condition” called our attention to a speculative scenario, where Zygophylacinae resulted in a family group (as already discussed by Moura43), as a basal lineage of Macrocolonia, or directly related to Plumupheniida.

Sertulariida taxon novum

Traditionally, sertulariids are well-supported taxa defined by their colony shape and hydrothecal morphology, although these characters are plastic and integrative proposals have been questioned99. Sertulariidae was firstly split into three subfamilies (Sertulariinae, Thyroscyphinae and Sertomminae), mainly based on whether hydrothecae were sessile or pedicellate and on the presence-absence of abcaulinar cecum53,63. The current scope of Sertulariida taxon novum, however, includes several sertulariid species that are herein classified among families with new interpretations, such as Thyroscyphidae and Sertulariidae sensu novum, besides the new taxon Sertularellidae fam. nov. Other groups, on the other hand, were removed from Sertulariidae (see Staurothecidae and Symplectoscyphidae above). Overall, Sertulariida taxon novum is characterized by stolonal or erect monopodial colonies, with radial to bilateral symmetry on hydrothecae, hydrothecal margin with 3 to 4 cusps, operculum of 2 to 4 valves and fixed gonophores (except for Amphisbetia operculata (Linnaeus, 1758) and Sertularia marginata (Kirchenpauer, 1864), which are medusoid-producing species). Below, we present our new interpretations and descriptions within Sertulariida taxon novum:

– Sertularellidae fam. nov

This is a well-supported clade, including the highly speciose genus Sertularella Gray, 1848 with 131 nominal species distributed worldwide3. Colonies are erect, mono- or polysiphonic, branched or unbranched and hydrothecae have four marginal cusps and a pyramidal operculum with four triangular valves53. Our results call attention to a possible reinterpretation of S. turgidus as part of Sertullarellidae and supports Sertularelloides cylindritheca as a Thyroscyphidae species, as already discussed100. Further samples for these species will be of great value, mainly because until now there is only one specimen sampled for S. turgidus45(Fig. 1; see below).

Additional references: Vervoort98,101; Vervoort & Watson102.

– Thyroscyphidae

Morphologically, Thyroscyphidae includes species with pedicellate hydrotheca, with annular diaphragm and hydranths with a double basal annulus instead of abcaulinar diverticulum. Traditionally, the family includes Thyroscyphus, Parascyphus, Sertularelloides, Thyroscyphoides, Uniscyphus and Symmetroscyphus2,53. Historically, Thyroscyphidae is monophyletic, either considered as a separate clade from Sertulariidae (e.g., Calder53; Shimabukuro & Marques88), or as part of Sertulariidae (cf. Millard23; Bouillon26). We keep Thyroscyphidae at the family rank because of its high support and coincidence with other similar hypotheses45,100, although small changes in the position of Thyroscyphus might occur. We include S. cylindritheca (previously Sertularella cylindritheca) in Thyroscyphidae46,100, but this species will need further analysis, including Parascyphus, because of its uncertain affinities (cf. Calder53).

– Sertulariidae sensu novum

A redefinition of the taxa within Sertulariidae sensu stricto is required, due to our proposal involving new families (i.e., Sertularellidae, Staurothecidae and Symplectoscyphidae). In its current scope, many genera still need to be sampled and validated and even the richest genera (e.g., Sertularia and Dynamena) are not considered monophyletic in our hypothesis, as already presented and discussed45,100; morphological features, such as the abcauline diverticulum (cf. Calder53 p. 86–87), still have to be tested as synapomorphies. Sertulariidae sensu novum includes species with hydrothecae with non-pyramidal operculum, with one (e.g., Abietinaria, Diphasia, Idiellana, Salacia, Thuiaria) or two valves (the adcaulinar one sometimes divided into two, e.g., Amphisbetia, Dynamena, Hydrallmania, Sertularia). These characters are homoplastic in our working hypothesis; other characters also proved to be equally uninformative, such as fasciculation, hydrothecal arrangement and shape of hydrothecal aperture. It is noteworthy to highlight that the non-monophyly of Sertularia raises back the question about the validity of Tridendata Stechow, 1920. Indeed, Sertularia argentea (type species of Sertularia) and Sertularia perpusilla (type species of Tridentata) falls apart in our hypotheses (Figs 1, 2, 3) and several of the species usually assigned to the genus Tridentata (e.g., Tridentata marginata, Tridentata turbinata, Tridentata tumida) are close to the type species T. perpusilla (D.R. Calder, pers. comm.). The addition of new sertulariid taxa and sequences shall provide a better test for this hypothesis, but presently it surely cannot be discarded. Sertulariidae sensu novum has decreased its species richness, since traditional genera such as Sertularella, Antarctoscyphus and Symplectoscyphus (representing ~41{57f679433bdda16678ea619f315c9bc28ff40af1ef9e9f7b6fe14a3c8b72c25f} of the number of species in Sertulariidae sensu stricto3) are not here considered within the family.

Additional references: Berrill103; Calder53; Cornelius4,5; Migotto104; Marques et al.59.

Plumupheniida taxon novum

N. lighti and H. mirabile have unstable positions, but generally related to the clades Aglaopheniida taxon novum and Plumulariida sensu novum. Except for N. lighti and H. mirabile, Aglaopheniida taxon novum and Plumulariida taxon novum are characterized by uniseriate hydrothecae5,25,26,28,45,46 (Figs 1, 2, 3). Concerning the taxonomy, Plumupheniida has a different taxonomic comprehensiveness from Plumularioidea McCrady, 18593 because N. lighti and H. mirabile are included in this clade based on multilocus data (similar situation for Aglaopheniida and Aglaopheniidae; see below and Table 3).

Nemalecium lighti and Hydrodendron mirabile (Lineage 5 – L 5)

Traditionally considered as haleciids, N. lighti and H. mirabile were not recovered as a monophyletic group, or as closely related to Haleciida; therefore they are considered Plumupheniida incertae sedis. Nemalecium lighti is remarkable in having “swimming gonophores” with corpuscles that may be related to buoyancy105. This character is also found in some Plumulariidae (e.g., Macrorhynchia philippina cf. Gravier-Bonnet & Migotto105) and Aglaopheniidae (e.g., Dentitheca bidentata cf. Migotto & Marques57). The presence of “nematodactyls” in the ring of tentacles106 should be reanalyzed concerning its organization, because it may be homologous to nematophores53, which would corroborate the affinity of Nemalecium with the Plumupheniida. However, future analyses are still necessary to better investigate this question and to test the position of N. lighti with respect to the rest of Plumupheniida, especially concerning other traditional “haleciids”.

Hydrodendron mirabile is highly supported as sister group of Aglaopheniida and shares the presence of nematophores and nematothecae with Aglaopheniida+Plumulariida sensu novum. Defense polyps, therefore, are corroborated as a possible synapomorphic condition of Plumulariida+Aglaopheniida+H. mirabile (Fig. 1, 2, 3) and not as a convergence (Leclère et al.41; corrobating a posterior result45). The taxonomic position of H. mirabile, however, has to be tested with the inclusion of additional molecular data for H. gardineri and other terminals of Hydrodendron, a genus with ca. 25 nominal species2.

Additional references: Rees & Vervoort55; Vervoort56; Cornelius5.

Aglaopheniida taxon novum

Aglaopheniida has a high nodal support (see also references41,42,45). Some proposals are incongruent with our topology, because Gymnangiinae/Aglaopheniinae and Aglaopheniini/Cladocarpini107 are not monophyletic due to the position of Gymnangium gracilicaule and Lytocarpia sp. Also, all traditional genera of Aglaopheniidae, except by Macrorhynchia, are not monophyletic in our proposal (Fig. 1; see also previous analyses45). Aglaopheniida is characterized by having paired lateral nematothecae partially fused to the hydrotheca, all nematothecae one-chambered and immovable46. Part of the taxon has gonothecae protected by phylactocarps or corbulae5.

Additional references: Gravier108; Svoboda & Cornelius109.

Plumulariida sensu novum

Plumulariida sensu novum has a high nodal support and includes four families characterized by free nematothecae and bilateral symmetry of the hydrotheca in many species. Our cladogram shows Schizotrichidae as sister group of the rest of Plumulariida sensu novum (Figs 1, 2, 3), instead of sister group of Aglaopheniidae, as originally proposed46. Our proposal of the relationship among the other three families (viz., Halopterididae, Kirchenpaueriidae and Plumulariidae; Figs 1, 2, 3) is different from that previously proposed (cf. Leclère et al.41). The topology (Kirchenpaueriidae (Halopterididae, Plumulariidae)) (cf. Leclère et al.45; Peña Cantero et al.46) refutes the absence of cauline hydrothecae as synapomorphy for Plumulariidae, Kirchenpaueriidae and Aglaopheniidae45,110.

Kirchenpaueriidae is characterized by the absence of paired lateral nematothecae and the presence of one to two mesial nematophores, naked or with reduced nematothecae, associated with each hydrothecae, usually also reduced45. Kirchenpaueriidae was traditionally considered as sister group of other Plumulariida due to the simple morphology of the nematothecae90, which are similar to those of Haleciidae107. However, our phylogeny does not corroborate this hypothesis, since Schizotrichidae has a basal position in Plumulariida sensu novum. Our phylogenetic relationships within Kirchenpaueriidae are as previously thought111: Ventromma is a monophyletic group (cf. Calder 1997107; Peña Cantero & Marques112) not synonymous of Kirchenpaueria (cf. Millard23; Bouillon26), which differs from Ventromma by the presence of naked sarcostyles107. In our topology, Kirchenpaueria pinnata, Pycnotheca mirabilis and species of Oswaldella define a monophyletic group. Oswaldella is a monophyletic group with high support and geographic distribution restricted to Antarctica and Patagonia shelf112,113,114. Further studies with molecular and morphological data should focus on the taxonomic affinities of Halicornopsis, Ophinella and Wimveria.

As stated above, Halopterididae becomes monophyletic with the exclusion of Schizotrichidae (viz., Peña Cantero et al.46) and the inclusion of Polyplumaria flabellata5,107,110. Amendments of the diagnosis based on morphological characters of Halopterididae was proposed by Peña Cantero et al.46. The genera traditionally considered as Halopterididae are not monophyletic (e.g. Halopteris Allman, 1877, cf. Leclère et al.41,45; Peña Cantero et al.46) and the shape of the colony as the main character to define their taxonomical limits might be questionable110. Halopteris is divided into several high-supported clades. Halopteris carinata Allman, 1877, type species of Halopteris, has different morphology from many of the species traditionally assigned to the genus (except for Halopteris liechtensternii (Marktanner-Turneretscher, 1890)) and the hypothesis that genera of Halopterididae are not monophyletic can be supported by morphology as well as molecular data (Dale Calder, pers. comm.). The position of species of Antennella on the topology (Figs 1, 2, 3) supports multiple origins of simplified colonies from the pinnate ancestral form110. Four genera (i.e., Anarthoclada, Astrolabia, Nuditheca and Pentatheca) have to be included in further analysis, especially because they have already been considered as Aglaopheniidae30, Halopterididae107,110 and Schizotrichidae46.

Plumulariidae is defined by our molecular data as sister group of Kirchenpaueriidae+Halopterididae and it is also morphologically well defined46 (Figs 1, 2, 3). The taxonomy of the group, however, is troubled, because the genera are poorly defined and the species share many plastic characters107. Two well-supported clades were found in Plumulariidae: one includes the non-monophyletic genera Dentitheca and species of Plumularia (Plumularia spiralis, Plumularia habereri and Plumularia insignis); the other includes the 9 remaining Plumularia and 2 species of Nemertesia (Nemertesia antennina and Nemertesia ventriculiformis). In our phylogeny, Nemertesia is monophyletic and related to Plumularia species, a different relationship from previous study115. Overall, our results are congruent with the proposals of the current literature based on genetic data41,45,46.

Additional references: Millard116; Ansín Agís et al.117; Schuchert118.

Evolutionary scenarios, life cycle and biogeography

Traditional evolutionary scenarios for Leptothecata may be either inadequate or speculative because of the lack of data and sometimes because of a grounded phylogenetic framework. For instance, the hypothesis that erect, complex colonies without the medusa stage, such as those of Macrocolonia, would be an eventual energetic tradeoff for medusa suppression (discussed in Cornelius31), or the evolution of the “swimming gonophores”119, are not based on phylogenetic hypotheses.

However, one could say that the evolution of Leptothecata supports an interpretation of adaptive radiation120, both by having high rates of lineage diversification with morphological innovations (e.g., colony structure) and diversified ecological properties (e.g., benthic life cycle). On the other hand, extinction rates are difficult to estimate because of the limited fossil data, which severely restrict our knowledge on the historical distribution of ancestral lineages. Several alternative hypotheses may explain the rapid diversification of Leptothecata45,121,122. For instance, loss of the medusa stage, together with the reduction in the number of larvae, would result in phylopatric species with long-lived, sessile polyp stages (cf. Thiel & Gutow123).

Yet, when trying to explain potential advantages for presence or absence of the medusa stage, there are different interpretations. According to Leclère et al.45 (p. 510 our emphasis), the high number of convergent medusa losses/reductions observed among hydrozoans strongly suggests that getting rid of the pelagic stage is positively selected under some circumstances, leading Cornelius121 to emphasize a challenging paradox: “if medusa loss is advantageous and if it can evolve easily, then why have not all recent forms dispensed with the medusa long ago?” However, Cornelius stated that the paradox condition was the loss of the “advantageous medusa stage”: “The widespread loss of the medusa stage is a paradox in view of its presumed advantage and undoubted complexity” (Cornelius31, p. 249, our emphasis).

Leclère et al.45 proposed an explanation for the evolution of different life cycles strategies124. In their scenario, species with simple colonies (such as those typical of Statocysta) would invest more in the reproductive and dispersive pelagic stage, with a benthic stage in which the planula quickly settles on living substrates (with less competition and greater tolerance to environmental stress). On the other hand, species with erect, complex colonies, without medusa (like those typical of Macrocolonia), would strongly compete for the substrate and have limited dispersal abilities, which could result in low gene flow and consequent local speciation. The (homoplastic) appearance of species with medusoids in lineages that lost the medusa could, in turn, be explained by high rates of extinction of clades that do not acquire reproductive or dispersive benefits, limiting the phenomenon of colonial brooding44. Thus, it is clear that these ideas presume greater dispersal ability of the pelagic adult stages (cf. Gibbons et al.125,126 with Cornelius31,121,127).

Our data support the interpretation of dispersal and rapid settling in Statocysta, in contrast with competitiveness in Macrocolonia. Because of that, stasis (cf. Eldredge et al.128) may have predominated in lineages with medusa stage45. However, mechanisms associated with the speciation process in Macrocolonia would not be just a product of dispersal limitations. Other factors, even combined with those previously discussed, may have been important in the exploration of new potential ecological niches, genetic diversity and diversification success (e.g., parental investment129). The first mechanism is the type of fertilization (Miller130; discussed in Corynidae and Obeliinae by Panteleeva131), which is widely debated in other modular organisms such as tunicates and bryozoans132,133. Overall, hydrozoans with the medusa stage in the life cycle spawn gametes directly to the environment (e.g., Laodicea), while those without the medusa stage retain the unfertilized egg, presenting internal fertilization, sometimes with incubation (e.g., some Diphasia species). The second mechanism is associated with the strength and robustness of the colonies, especially those with a main axis with erect monopodial growth and branches that are strengthened by the fusion of the hydrotheca to the perisarc134. The third mechanism concerns the increasing in modular and morphological complexity, reflected in the high occurrence of nematophores/nematothecae in complex and reproductive defense structures135,136, such as the phylactocarps and the corbulae in Aglaopheniida. In this case, the polyp stage would live longer than the medusa stage, reducing the periodicity of the life cycle and increasing the rate of anagenesis31.

All these mechanisms modulate levels of endemism in different biogeographic scenarios of Hydrozoa. Cornelius31,137 discussed several hypotheses in favor of the prevalence of the polyp stage as the main mechanism of dispersal in Hydrozoa, such as the high-energetic cost of producing a medusa stage, the small-size of the hydroid colonies that release medusa, the different realms occupied by the two stages and the difficulties of interchanges between them, the large-size of the colonies that do not produce medusa and tend to perennate, the brief life span of the medusa stage and the low chances of two sexually different medusa to arrive at the same time in a long-distant area, to reproduce and to develop a new hydroid colony. In this context, we may hypothesize on one hand, that species with both polyp and medusa stages in the life cycle may have greater endemicity than those with only polyp or medusa stage in the life cycle31,121. On the other hand, species with both polyp and medusa stages in the life cycle are less common in remote areas of the deep sea and at distant places from the continental shelves138,139, as well as in regions with extreme environmental conditions (e.g., Antarctic, Arctic, hydrothermal vents). However, few biogeographic studies dealing with the level of endemicity and/or areas of endemism in Hydrozoa species were fully carried out, with this being an open question to be better explored through different biogeographic methods. Another interesting biogeographical pattern in Macrocolonia is seen in Staurothecida taxon novum, where Staurotheca, Symplectoscyphus and Antarctoscyphus are widely represented in Antarctic and sub-Antarctic waters, supporting a unique phylogenetic history. In contrast, other genera of Sertulariida taxon novum (e.g., Amphisbetia, Diphasia, Dynamena, Sertularia and Thyroscyphus) have an antipolar distribution. Results of ecological niche modeling140 of Bougainvilliidae show that, probably, the potential niche (and correlated areas) of distribution of certain species is the sum of ecological restrictions of polyps and medusae, which strongly limits distribution. Therefore, the presence of both stages in the life cycle results in narrower potential areas to be colonized. We conclude that reductionism in scenarios of advantages and disadvantages of the medusa stage are oversimplifications of complex evolutionary processes. It is also evident that the evolution of the life-cycle strategies has consequences for the understanding of species distributions and biogeography.

Final considerations

Evidences accumulated during the last decades on morphology and life cycle of leptothecates demand new interpretations on the evolution of the group, especially because traditional taxonomy is incongruent with the recent phylogenetic hypotheses. Monophyly of many traditional taxa (e.g., Sertulariidae) is not supported, although some patterns remain strong, such as orders Statocysta and Macrocolonia (Table 4). Therefore, we propose a phylogenetic taxonomical reorganization of Leptothecata down to the infraorder level, but considering only the well-supported clades. Although most of main traditional groups were sampled in this study, we emphasize that several taxa (e.g., from species up to the family level) have not been included yet in broad phylogenetic analyses and those clades must be intregrated to this classificatory framework in the near future:

Superorder Leptothecata Cornelius, 1992

Order Lafoeida Bouillon, 1984, sensu novum

Order Laodiceida taxon novum

Order Statocysta Leclère, Schuchert, Cruaud, Couloux, Manuel, 2009

Suborder Campanulinida Bouillon, 1984, sensu novum

Suborder Eirenida taxon novum

Suborder Proboscoida Broch, 1910 sensu novum

Infraorder Campanulariida Bouillon, 1984, sensu novum

Infraorder Obeliida taxon novum

Order Macrocolonia Leclère, Schuchert, Cruaud, Couloux, Manuel, 2009

Suborder Staurothecida taxon novum [includes Staurothecidae fam. nov. and Symplectoscyphidae fam. nov.]

Suborder Haleciida Bouillon, 1984 sensu novum

Suborder Sertulariida taxon novum [includes Sertularellidae fam. nov.]

Suborder Plumupheniida taxon novum

Infraorder Aglaopheniida taxon novum

Infraorder Plumulariida Bouillon, 1984 sensu novum

Related Articles

Back to top button