Reassessment of Morphological Diagnostic Characters and Species Boundaries Requires Taxonomical Changes for the Genus Orthopyxis L. Agassiz, 1862 (Campanulariidae, Hydrozoa) and Some Related Campanulariids

The genus Orthopyxis is widely known for its morphological variability, making species identification particularly difficult. A number of nominal species have been recorded in the southwestern Atlantic, although most of these records are doubtful. The goal of this study was to infer species boundaries in the genus Orthopyxis from the southwestern Atlantic using an integrative approach. Intergeneric limits were also tested using comparisons with specimens of the genus Campanularia. We performed DNA analyses using the mitochondrial genes 16S and COI and the nuclear ITS1 and ITS2 regions. Orthopyxis was monophyletic in maximum likelihood analyses using the combined dataset and in analyses with 16S alone. Four lineages of Orthopyxis were retrieved for all analyses, corresponding morphologically to the species Orthopyxis sargassicola (previously known in the area), Orthopyxis crenata (first recorded for the southwestern Atlantic), Orthopyxis caliculata (= Orthopyxis minuta Vannucci, 1949 and considered a synonym of O. integra by some authors), and Orthopyxis mianzani sp. nov. A re-evaluation of the traditional morphological diagnostic characters, guided by our molecular analyses, revealed that O. integra does not occur in the study area, and O. caliculata is the correct identification of one of the lineages occurring in this region, corroborating the validity of that species. Orthopyxis mianzani sp. nov. resembles O. caliculata with respect to gonothecae morphology and a smooth hydrothecae rim, although it shows significant differences for other characters, such as perisarc thickness, which has traditionally been thought to have wide intraspecific variation. The species O. sargassicola is morphologically similar to O. crenata, although they differ in gonothecae morphology, and these species can only be reliably identified when this structure is present.

The genus Orthopyxis L. Agassiz, 1862 clearly illustrates the difficulties associated with taxa delimitation in the family. Many uncertainties exist concerning the validity of this genus e.g., [19,26,28,29,37,38], and it has been synonymized multiple times with the genus Campanularia Lamarck, 1816 based on their morphological similarities. In addition, species traditionally assigned to the genus Orthopyxis have very similar morphologies and few diagnostic characters, making delimitation difficult, particularly when only trophosomal characters are considered or available cf. [27,39]. Altogether, these practical issues-particularly the uncertain validity of the genus e.g., [19] (p.60) and many of its species e.g., [14,19]-demand different taxonomic approaches to reassess and establish species boundaries within Orthopyxis.
In the southwestern Atlantic, five species of the genus Orthopyxis have been recorded along the coast of Brazil by Vannucci-Mendes [40] and Vannucci [41,42], which were later re-identified as two species: Orthopyxis integra (Macgillivray, 1842) and Orthopyxis sargassicola (Nutting, 1915) [1,13,31] (Table 1). Vannucci-Mendes [40] and Vannucci [42] also recorded two species of Campanularia along the southeastern coast of Brazil, although both records are now considered dubious [8]. Unfortunately, a formal revision of these records is not possible, as most of the materials described by Vannucci have been lost [1]. Along the Argentinean coast, Blanco [43,44,45] recorded several species of Campanularia and Orthopyxis, some of which she subsequently re-identified as Campanularia subantarctica Millard, 1971 [46], which is currently considered to be a synonym of Campanularia lennoxensis Jäderholm, 1903 [47] ( Table 1). Other records of Campanularia and Orthopyxis for the southwestern Atlantic are listed in Table 1. Most of them are considered dubious, requiring a revision of species records in this region.
Currently, O. sargassicola and O. integra have been reported to occur in the southwestern Atlantic. In Brazil, O. sargassicola was recorded off the coast of Espírito Santo [10,48] and São Paulo states [1,49,50,51], and together with O. integra, it has been recorded along the coast of Rio de Janeiro [10,52,53], Paraná [54] and Santa Catarina states [13]. They are usually found in shallow waters, though have also been recorded in deeper areas of 35 and 70 meters [10,53], and frequently occur in epiphytic associations, often on macroalgae of the genus Sargassum C. Agardh, 1820 [1,13,50,51,54,55]. The species O. sargassicola, for instance, is among the most common and abundant species of hydroids in ephypytic environments in São Paulo and Paraná states [51,54]. In Argentina, O. caliculata (accepted as Campanularia integra, [46]) was recorded in Puerto Madryn, Chubut [43] and O. integra in Punta Peñas, Sán Julian ( [46], as C. integra); a third species, O. everta (Clark, 1876), was recorded by Blanco [44,45] along the coast of Argentina, but it was later re-identified as C. subantarctica by Blanco [46] and is now thought to be two different species [47,56] (Table 1). Studies with Orthopyxis from Argentina are restricted to their original records, in which species are generally reported in epiphytic or epizoic associations, from shallow waters to depths of 157 meters [43,46]. Species of Campanularia, on the   [3,4,6,9,131,132] Uruguay and Argentina --
The goal of this study was to reassess species boundaries within the genus Orthopyxis based on species models from the southwestern Atlantic. Furthermore, morphological characters associated with Orthopyxis are re-evaluated, one new species and one new record of Orthopyxis are described, and the intergeneric limits of Orthopyxis and Campanularia are reassessed.

Molecular data
Nuclear DNA and mitochondrial DNA were extracted using Instagene (Bio-Rad Laboratories, Hercules, California, USA), according to the manufacturer's protocol. Portions of the mitochondrial 16S ribosomal RNA gene and the cytochrome oxidase subunit I (COI) gene as well as the entire nuclear Internal Transcribed Spacer (ITS) region (ITS1, 5.8S ribosomal RNA gene and ITS2) were amplified by PCR and verified on 1.5% agarose gels (PCR conditions and primers are described in Table 3). PCR products were purified using the AMPure purification kit (Agencourt Bioscience Corporation, Beckman Coulter, Beverly, Massachusetts, USA), and purified products were prepared for sequencing using the Big Dye Terminator v3.1 Cycle Sequencing kit (Applied Biosystems, Foster City, California, USA) and the same PCR primers. The sequencing reactions were carried out using an ABI PRISM 3100 Genetic Analyzer (Applied Biosystems, Foster City, California, USA). Sequences were assembled and edited using Geneious (version 7.1 created by Biomatters, Auckland, New Zealand), and aligned using MAFFT [79]. The obtained sequences were compared with those deposited in GenBank using the Basic Local Alignment Search Tool (BLAST, [80]) to confirm genes and species of interest. Additionally, the ITS1 and ITS2 regions were extracted from the complete ITS sequences using the sequence from Hydra circumcincta [81] in GenBank (GU722663) as a guide to delimit the ITS1 sequences and the ITS2 Database [82] to delimit the ITS2 sequences. The coding sequences of COI were translated and compared with the complete mitochondrial genome of Laomedea flexuosa [83] (GenBank NC_016463) to ensure pseudogenes were not amplified. Since not all sequences of the same marker had the same length (see Table 3), some portions of the longer sequences were excluded from the alignments to adjust all sequences to the same length.

Phylogenetic analysis
Phylogenetic analyses were performed on (a) individual markers, (b) combined mitochondrial markers (16S+COI), (c) combined nuclear markers (ITS1+ITS2), and (d) the entire combined dataset (16S+COI+ITS1+ITS2), using maximum likelihood (ML) and parsimony (P) criteria. The datasets were built using unique haplotypes, and the combined datasets included only those specimens with sequences available for all markers (details of the analyses in Table 4).
Sequences of nuclear DNA with ambiguous sites (17 ITS1 and 22 ITS2 sequences) were treated using IUPAC ambiguity codes. The maximum number of ambiguous sites recorded for one sequence was five (the ITS2 sequence of a specimen from Penha, Santa Catarina), and 46% of the sequences had only one ambiguous site. Sequences with identical IUPAC codes at identical positions were considered as the same haplotype in the analyses. Phylogenetic analyses using parsimony (P) criteria were performed using the PAUP 4.0b10 [84] and TNT [85] programs. Analyses consisted of 1000 unweighted heuristic searches using a random algorithm and branch-swapping using the TBR (tree bisection-reconnection) algorithm. Gaps were considered as a fifth state. Branch support was estimated in TNT with bootstrapping on 1000 replicates. Phylogenetic analyses using Maximum Likelihood (ML) criteria were performed using PALM (Phylogenetic Reconstruction by Automatic Likelihood Model Selector, [86]) with the most appropriate model of nucleotide evolution for each dataset based on Akaike Information Criterion (AIC, Table 4). Branch support was estimated with bootstrapping on 1000 replicates. Phylogenetic p-distances (uncorrected) were calculated using the PAUP 4.0b10 program.

Morphological analysis
We performed Principal Component Analysis (PCA, [87]) on a correlation matrix based on 37 different measures of the trophosome (S1 Table) of the voucher specimens of O. caliculata and O. mianzani sp. nov. (the same specimens used in the phylogenetic analyses). For both species, we did not include any characters from the gonothecae in the PCA, as not all colonies presented this reproductive structure. This analysis was performed to better delimitate the species by assessing the degree of variation for their morphological characters and by identifying their most relevant diagnostic characters.

Nomenclatural acts
The electronic edition of this article conforms to the requirements of the amended International Code of Zoological Nomenclature, and hence the new names contained herein are available under that Code from the electronic edition of this article. This published work and the nomenclatural acts it contains have been registered in ZooBank, the online registration system for the ICZN. The ZooBank LSIDs (Life Science Identifiers) can be resolved and the associated information viewed through any standard web browser by appending the LSID to the prefix "http:// zoobank.org/". The LSID for this publication is: urn:lsid:zoobank.org:pub:280AC2D0-9DCE-4BCE-AF85-2586B3951522. The electronic edition of this work was published in a journal with an ISSN, and has been archived and is available from the following digital repositories: PubMed Central and LOCKSS.

Results
Nearly all the topologies obtained using the different datasets identified six well-defined clades with high branch support values. However, these topologies did present some incongruencies with respect to the phylogenetic relationships among these clades. The individual and combined nuclear datasets showed low resolution and low values for branch support, whereas the combined mitochondrial datasets showed higher resolution but also had low branch support (S1-S10 Figs.). The combined dataset involving all four markers revealed the best definition of the relationships among the lineages, with a higher frequency of well supported nodes (all six less inclusive clades with bootstrap = 99-100, Figs. 2-3). In addition, the 16S topologies showed the most congruent results (Figs. [4][5]. Therefore the topologies involving the combined and the 16S datasets represented the most robust hypothesis for our data and are used as our working hypothesis for discussions.

The genera Orthopyxis and Campanularia
The genus Orthopyxis was monophyletic according to the 16S topologies and the ML topology with the combined dataset, although with low support value (bootstrap<50, Figs. 2, 4-5).
Orthopyxis was not monophyletic in the P topology with the combined dataset, in which species assigned to Campanularia fell within Orthopyxis as a sister group to Orthopyxis caliculata (Hincks, 1853)+Orthopyxis mianzani sp. nov. (Fig. 3). Although not conclusive, Orthopyxis was monophyletic in the majority of our topologies, a hypothesis we follow in this study. However, this hypothesis requires further testing with the addition of more representatives from the genus Campanularia.
Campanularia was monophyletic only in topologies derived from the combined dataset. One of the lineages of Campanularia corresponds morphologically to Campanularia subantarctica Millard, 1971, and it is characterized by the deep hydrothecae with bluntly rounded marginal teeth, subhydrothecal spherule present; gonothecae oval-elongated arising from hydrorhiza, with distal aperture on top of a low collar [77,88], ([47], as C. lennoxensis). The second lineage of Campanularia is also morphologically similar to C. subantarctica, but we were   unable to identify this lineage to the species level due to the lack of gonothecae. Additionally, these two lineages showed genetic distances of up to 7.83% for mitochondrial markers and 26.38% for nuclear markers ( Table 5), indicating that they likely represent two distinct species.
The specimens of O. sargassicola and O. crenata recorded here have the general features of Orthopyxis, such as a thick perisarc (variable to some extent), campanulate hydrothecae, sinuous pedicels, and subhydrothecal spherule (Fig. 6). The presence of rounded hydrothecal cusps and a laterally compressed, completely ribbed gonotheca, are distinctive characters of O. sargassicola [1,13,31] (Fig. 6A-C, G), whereas O. crenata is characterized by low, rounded hydrothecal cusps and laterally compressed, smooth gonothecae [29,47] (Fig. 6D-F, H). Although these species can be readily distinguished by comparing their gonothecae, morphological variation in the size and shape of the hydrothecal cusps may cause these diagnostic characters to overlap when the gonotheca is absent, hampering identification. This is the first record of Orthopyxis crenata in the southwestern Atlantic, although previous authors may have overlooked this species due to its morphological similarity with O. sargassicola.
The species O. caliculata and O. mianzani sp. nov., although highly genetically divergent (Table 5), have similar morphologies that could be traditionally associated with Orthopyxis integra (Macgillivray, 1842). Both species have stolonal colonies, sinuous pedicels, subhydrothecal spherule, campanulate hydrotheca with rim even, and gonotheca roughly cylindrical, with wide aperture, truncated on top [19,78]. However, they are morphologically distinct with respect to characters usually assumed to show wide intraspecific variation, such as perisarc thickness and length of the hydrothecae and pedicels.
PCA performed using the morphometric data for O. caliculata and O. mianzani sp. nov. (Fig. 7) showed that the two lineages are clearly separated by perisarc thickness and polyp general dimensions. Specimens of O. caliculata have a thicker perisarc and smaller general dimensions (length and diameter of the hydrothecae, pedicels, and subhydrothecal spherule) of the polyp (Fig. 7). These results show that, although variable to some extent, perisarc thickness and polyp dimensions can be used to delimitate these species. Therefore, we believe the name Orthopyxis caliculata (Hincks, 1853) is the correct identification of one of these lineages, and we corroborate the validity of that species.
Remarks. Orthopyxis caliculata (Hincks, 1853) has been considered a synonym of O. integra (Macgillivray, 1842) by many authors. Levinsen [91] was likely the first to assign Hincks' species to O. integra (as Campanularia integra), arguing that he possessed colonies of O. integra that presented intermediate characters from both species, referring in particular to the thickness of the perisarc of the hydrothecae and the presence of annulations on the gonothecae. Many subsequent authors followed this proposal [19,29,78,92,93,94,95,96,97], also arguing that the characters used to distinguish these species are actually intraspecific variations of the same character.
Hincks [98] noted the shape of the hydrothecae and the presence of a "double cup" and "double" pedicel as the main characters that distinguish O. integra and O. caliculata. He subsequently amended his description by arguing that the appearance of a "double" hydrotheca and pedicel is a result of the considerable perisarc thickening in this species [26]. The widely accepted notion that these characters represent variations within the same species has prevented many authors from accepting them as informative (as stated above), although some authors who agree with Hincks [26,98] in regarding O. integra and O. caliculata as separate species point out characters such as the size and shape of the hydrothecae and gonothecae, as important differences between these species e.g., [27,28,99,100]. Indeed, the name O. caliculata is currently used as a valid name in some studies [101,102], based on similar opinions. Neither species was originally described with gonothecae [98,103], although subsequent descriptions of these species represented the gonothecae of O. integra as clearly different from those of O. caliculata. The gonothecae of O. integra is described as cylindrical, completely spirally grooved throughout, and truncated on top, whereas the gonothecae of O. caliculata is described as smooth, oval-elongated, laterally compressed, also truncated on top, and with a wide aperture [26,28,99]. Authors who advocate the synonymy of O. integra and O. caliculata consider both types of gonothecae as variations within O. integra (see [78]). Despite this, Millard [29] notes that she never recorded polyps of O. integra in South Africa with spirally grooved gonothecae, and many other records of O. integra include only specimens with oval-elongated, smooth gonothecae e.g., [13,97,104,105,106]. Indeed, cylindrical, spirally grooved gonothecae appear to be restricted to northern records of O. integra e.g., [38,107,108,109,110], as noted by Bale [99].
We studied non-type material of O. integra that presented spirally grooved gonotheca (USNM 17834 from Kara Sea, and 1106184 from Alaska, Aleutian Islands)-in contrast with the oval-elongated, smooth gonotheca of our material-and we have concluded that these two types of gonotheca indicate two different species. These non-type materials of O. integra also differ from our specimens of O. caliculata in the thickness of the perisarc of the hydrothecae and pedicels, as well as in the length of the hydrothecae, which is larger in O. integra (see comparisons on Table 6). Many of these differences have already been noted and discussed by Bale [99], and more recently by Calder et al. [102]. Our molecular analysis revealed two different lineages presenting the traditional morphological characters associated to O. integra. A reevaluation of the morphological characters of these two lineages demonstrates that their most consistent differences rely on characters previously considered to be intraspecific variations by many authors. Therefore, we conclude that the two completely different gonothecae morphologies should not be considered as variations within O. integra.
Considering this, we believe that none of the species described in this study, nor the records included in our synonym, should be assigned to O. integra; instead, they should be assigned to O. caliculata. We understand that it is not simple to delimit these two species morphologically; therefore, we did not include in the synonymy materials we could not access. The one exception is Vannucci's material [40,41], which is most likely lost (see [1]), for which we tentatively attribute the specimens she described with gonothecae to O. caliculata. The specimen Campanularia integra recorded by Blanco [43] consists of only one microslide with one polyp without gonothecae, and the hydrotheca of this specimen differs from the typical hydrotheca of O. caliculata, being more elongated and cylindrical, similar to the hydrothecae of many species of the genus Campanularia. It is unclear whether this morphology is a preparation artifact or an actual morphological difference, so we therefore decided not to include this record in the synonym of O. caliculata, pending more detailed study. However, the specimens of C. integra recorded by Blanco [46] correspond to the description of O. caliculata. Milstein [90] described specimens with gonothecae that also correspond to O. caliculata. The records of O. integra by Miranda et al. [13] came from localities very close to our records of O. caliculata, and examination of their material leaves no doubt that it should be assigned to O. caliculata.
Diagnosis. Hydrothecae, pedicels and gonothecae with thin perisarc. Lateral compression only on gonothecae, nearly no compression detectable on hydrothecae or pedicels, both usually longer when compared with other species of Orthopyxis. Reduced amount of sinuosities on pedicels, sometimes almost completely smooth. Gonothecae smooth and different from other Orthopyxis species with ribbed gonothecae.
Remarks. Although this species resembles several nominal species of Orthopyxis, it presents important morphological differences. With respect to the trophosome, it resembles that of the widely known Orthopyxis integra (Macgillivray, 1842), but they differ significantly in gonothecae shape (see remarks of O. caliculata; also see [26,28,111]). The gonothecae of O. mianzani sp. nov. is also very similar to that of O. caliculata (Hincks, 1853), but the length of the pedicels and hydrothecae in O. mianzani sp. nov. is 100 μm greater (on average) compared with O. caliculata (Table 6; Fig. 7), and its perisarc is, on average, two to three times thinner than that of O. caliculata (Table 6; Figs. 8 and 9). Indeed, a thin perisarc is a good diagnostic character for this species, as it does not appear to be as variable as in other species of Orthopyxis. Although there is some variation in perisarc thickness (2.5-12 μm on hydrothecae, 7.5-22.5 μm on subhydrothecal spherule and 7.5-12.5 μm on pedicels), it is never as thick as in O. caliculata or as described and illustrated for many other species of Orthopyxis, such as Orthopyxis pacifica Stechow, 1919, Orthopyxis angulata Bale, 1914 (see also [101]) and Orthopyxis compressima (Kubota & Yamada, 1992). Even among species currently considered to be synonyms of O. integra [19], the perisarc is frequently described as very thick or variable in thickness (e.g., Orthopyxis compressa Clark, 1877; Orthopyxis asymmetrica Stechow, 1919); in cases where the species is represented with a thin perisarc, other characters appear to differ from those of O. mianzani sp. nov., such as the gonothecae (e.g., Campanularia integriformis Marktanner-Turneretscher, 1890, Orthopyxis wilsoni Bale, 1914).
The slightly sinuous pedicels of O. mianzani sp. nov. may also prove to be a good diagnostic character, particularly for distinguishing this species from O. caliculata, as these sinuosities are never so marked as in the latter species. This character also differentiates O. mianzani sp. nov. from Orthopyxis clytioides (Lamouroux, 1824). The pedicels of O. clytioides, represented by Lamouroux [112] as real annulations, are quite different from the sinuosities found in O. mianzani sp. nov. and other species of Orthopyxis, such as O. integra and O. caliculata [26,28,78]. Orthopyxis clytioides, however, still has a doubtful taxonomic status and some authors suggest it may be related to the genus Obelia [19,99].
The specimens belonging to O. integra recorded by Grohmann et al. [53] in Rio de Janeiro, Brazil, closely resemble this new species, particularly with respect to the thin perisarc and shape of hydrothecae and gonothecae. They are tentatively assigned here to O. mianzani sp. nov., pending future study of the material of Grohmann et al. [53].
Silicularia, Orthopyxis sp. indet., and Campanulariidae sp. indet. Silicularia rosea Meyen, 1834 and unidentified specimens were only included in the 16S phylogenies, as we were unable to amplify COI fragments from these specimens. In the 16S phylogenies, Silicularia rosea has a basal position relative to the other genera. The highly supported clade Silicularia +Campanularia+Orthopyxis corroborates the close relationships between these genera, although this may have been affected by using a relatively distant root species (Obelia dichotoma, O. longissima).
The specimens from San Julián, Argentina (Campanulariidae sp. indet.) are morphologically similar to Orthopyxis mianzani sp. nov., but their ambiguous position among the different phylogenies (Figs. 4-5; S3-S4, S7-S10 Figs.) makes it difficult to determine their true identity. Considering only the 16S phylogenies, they occupied a basal position among Orthopyxis. The specimen from Caleta Olivia, Argentina (Orthopyxis sp. indet.) is morphologically similar to Orthopyxis crenata, but it lacks gonothecae, which would have allowed for better comparisons, and it also had an ambiguous position in the phylogenies, hampering its identification. This specimen, however, was consistently positioned among the species of the genus Orthopyxis. As reliable information for the identification of these specimens was lacking, they were left unidentified until more information is available to determine their taxonomic status.

Discussion
Our results reinforce the importance of using mitochondrial markers, particularly the 16S rRNA gene, for phylogenetic inferences at many taxonomic levels. The use of 16S to define genera and species is common in studies with the Hydrozoa [24,25,64,66,67,69,76,113], and its potential for barcoding has been demonstrated [71,114]. The resolution levels provided by this gene are also adequate for phylogenetic inferences among putative superfamilies, orders and even subclasses e.g., [65], including the Hydroidolina [16]. In this study, the phylogenetic signal from 16S proved crucial for defining the relationships among the species and genera in these analyses, corroborating the monophyly of the genus Orthopyxis and delimiting the four species that occur in the southwestern Atlantic.
By contrast, the nuclear ITS markers are not often used for phylogenetic inferences in studies of the Hydrozoa e.g., [81,115], being more common in studies of the Scyphozoa [116,117,118,119]. Species of the genus Aurelia [116,119] and many other invertebrates [120] (Insecta), [121] (Decapoda), [122] (Anthozoa) show great variability in the ITS region, and as a consequence, the ITS markers are generally considered inadequate for supraspecific phylogenetic inferences e.g., [121]. Our ITS analyses corroborate the results obtained with the mitochondrial markers by identifying the same six clades in nearly all analyses. However, the high genetic distance values of the ITS region (Table 5) provide important evidence that phylogenetic information based on ITS on more inclusive levels of the trees is inadequate.
Many molecular studies have characterized cryptic lineages, such as in the genera Aurelia (7-9 lineages with genetic distances of 13-24% for COI and 7.8-14.5% for 16S [116,118]) and Tamoya (2 lineages with genetic distances of 4.4-4.5% for COI and 2.1-2.5% for 16S [123]). Similar results were obtained for species of the genera Coryne, Turritopsis and Cordylophora, in which interspecific distances ranged from 12.35-15.3% for COI and 3.7-9.2% for 16S [67,69,72]. The genetic distances among the species O. sargassicola, O. crenata, O. caliculata and O. mianzani sp. nov. agree with those studies, ranging from 12.35-16% for COI and 7.81-10.2% for 16S. It is important to note, however, that specimens with the diagnostic features of the species O. integra, which are commonly reported in the study region, represented two different lineages, neither of which was diagnosed as O. integra after a reexamination of their morphological characters. Additionally, the commonly recorded species C. subantarctica appears to include two different lineages, although we could not assess the taxonomic status of these lineages due to the low number of specimens. The discovery of different lineages, sometimes in presumably cosmopolitan species, has been recurrent in the family Campanulariidae [14,23,76] and even in genera with extensive revisions aiming to establish interspecific limits (e.g., Obelia [18,20]).
Although it is possible to assess species boundaries in the genus Orthopyxis using molecular methods, this task is not straightforward using morphological characters, primarily due to wide intraspecific variation. Molecular studies involving morphologically variable groups reveal that morphological characters used to delimit species are frequently misinterpreted, and some traditional diagnostic characters are proving to be inadequate e.g., [124,125]. Despite this, many misleading assumptions regarding the variability of morphological characters in the genus Orthopyxis still remain, and conclusions are frequently based on partial or non-formal analyses, derived either from the study of relatively few specimens or from repetition of the opinions of different authors, which are sometimes not based on actual voucher specimens. Indeed, this appears to be the case for the species O. integra in the southwestern Atlantic. The intraspecific variation of O. integra has been widely documented [19,29,78,91,92,109], and this species is traditionally assumed to be cosmopolitan [19,28,97], but it is clear that the amplitude of intraspecific variation of certain O. integra morphological characters has been overestimated. Perisarc thickness, for instance, is an important diagnostic character for the species of O. caliculata and O. mianzani sp. nov. delimited in this study, although this character is frequently considered too variable to be relevant for diagnostic purposes [19,29,39,47]. Furthermore, we believe that other characters, such as the presence of annulations on the gonothecae, may also be useful diagnostic characters for different lineages within O. integra and that they should be investigated more closely. A worldwide revision of O. integra is particularly timely, as it appears many of its synonyms may in fact represent true species.
Orthopyxis sargassicola, a species widely known in the western Atlantic [1,13,28,31], also appeared as one of the lineages of Orthopyxis delimited here. We recorded this species along the southeastern coast of Brazil, and it is known to occur in different regions along the Brazilian coast e.g., [1,8,13]. There are no records of O. sargassicola in Argentina. Other records are from the Gulf Stream (type locality, [28]), east of cape Hatteras [108], and in Aruba, Bonaire and Curaçao [126]. Orthopyxis crenata, another lineage delimited in this study, is first recorded for the southwestern Atlantic. Previous records attributed to this species (Table 1) are misidentifications or still have a doubtful taxonomic status. Specimens of O. crenata were recorded for Brazil in the states of Ceará (Fortaleza), São Paulo (Ubatuba) and Santa Catarina (Penha and Laguna); other global records include Chile [47,127], New Zealand [39,97,128], South Africa [29] (as Campanularia crenata) and Japan [100] (as C. crenata). There have been many discussions of the variability of the hydrothecal cusps of O. crenata, which vary from slight crenations on the margin of the hydrotheca to well-developed cusps [19,29,39,47,97,100], commonly overlapping with the morphology of the cusps of the species O. sargassicola. Calder [31] highlighted the morphological similarities between these two species, which are distinguished by the presence of annulations on the gonothecae of O. sargassicola, and by their absence in O. crenata. Migotto [1] also noted that some of the specimens he identified as O. sargassicola from São Sebastião (SP), Brazil, had morphological similarities to O. crenata, particularly with respect to the hydrothecal cusps and medusoids. Neither species can be identified with any certainty in the absence of gonothecae, and therefore, the records of O. sargassicola without gonothecae in the southwestern Atlantic should be considered with caution.
Specimens assigned to the genus Campanularia here are morphologically similar to the species of Orthopyxis, from which they can be distinguished by gonothecae morphology. With respect to the trophosome, the specimens of Campanularia do not possess a thickened perisarc on the hydrotheca and pedicels, as is observed in many species of Orthopyxis. Galea et al. [47] considered Campanularia subantarctica Millard, 1971 to be a synonym of the species Campanularia lennoxensis Jäderholm, 1903 based on the argument that their specimens presented gonothecae features found in both species and that perisarc thickness is a variable feature in the Campanulariidae. As already discussed, Campanulariidae is well known for its morphological variability e.g., [19], but we show that perisarc thickness may be a relevant character for delimiting certain species, at least when included in a detailed analysis with a wide range of specimens. Additionally, descriptions of C. subantarctica for the study area resemble the specimens described by Millard [77] (e.g., with a thinner perisarc [88,129]). Considering this, we believe the proposed synonymy is premature without more complete evidence, and we regard C. subantarctica Millard, 1971 as a valid species, pending more detailed study.
The difficulties in identifying species of Orthopyxis and Campanularia in the study area are noteworthy, particularly considering the high number of nominal species described and the uncertain synonymies e.g., [1,6,13,31,40,41,42,43,44,45,46,130]. Our analysis corroborates the monophyly of Orthopyxis and delimits four species in the southwestern Atlantic, consistent with an assessment of their morphological characters. These findings are crucial to our understanding of the intergeneric limits and species boundaries in the family Campanulariidae. We believe that this integrative approach clarifies many taxonomic difficulties associated with the species of Orthopyxis, and we hope that it may serve as a model for the delimitation of other species within the Campanulariidae. Nacional de Historia Natural, Montevideo, Uruguay; and Allen G Collins and W Geoff Keel, National Museum of Natural History, Smithsonian Institution, United States, for providing museum specimens for study. Finally, we would like to thank Dale R Calder and an anonymous reviewer for their helpful comments on a previous version of this manuscript.