Phylogeography of hydrothermal vent stalked barnacles: a new species fills a gap in the Indian Ocean ‘dispersal corridor’ hypothesis

Phylogeography of animals provides clues to processes governing their evolution and diversification. The Indian Ocean has been hypothesized as a ‘dispersal corridor’ connecting hydrothermal vent fauna of Atlantic and Pacific oceans. Stalked barnacles of the family Eolepadidae are common associates of deep-sea vents in Southern, Pacific and Indian oceans, and the family is an ideal group for testing this hypothesis. Here, we describe Neolepas marisindica sp. nov. from the Indian Ocean, distinguished from N. zevinae and N. rapanuii by having a tridentoid mandible in which the second tooth lacks small elongated teeth. Morphological variations suggest that environmental differences result in phenotypic plasticity in the capitulum and scales on the peduncle in eolepadids. We suggest that diagnostic characters in Eolepadidae should be based mainly on more reliable arthropodal characters and DNA barcoding, while the plate arrangement should be used carefully with their intraspecific variation in mind. We show morphologically that Neolepas specimens collected from the South West Indian Ridge, the South East Indian Ridge and the Central Indian Ridge belong to the new species. Molecular phylogeny and fossil evidence indicated that Neolepas migrated from the southern Pacific to the Indian Ocean through the Southern Ocean, providing key evidence against the ‘dispersal corridor’ hypothesis. Exploration of the South East Indian Ridge is urgently required to understand vent biogeography in the Indian Ocean.

Phylogeography of animals provides clues to processes governing their evolution and diversification. The Indian Ocean has been hypothesized as a 'dispersal corridor' connecting hydrothermal vent fauna of Atlantic and Pacific oceans. Stalked barnacles of the family Eolepadidae are common associates of deep-sea vents in Southern, Pacific and Indian oceans, and the family is an ideal group for testing this hypothesis. Here, we describe Neolepas marisindica sp. nov. from the Indian Ocean, distinguished from N. zevinae and N. rapanuii by having a tridentoid mandible in which the second tooth lacks small elongated teeth. Morphological variations suggest that environmental differences result in phenotypic plasticity in the capitulum and scales on the peduncle in eolepadids. We suggest that diagnostic characters in Eolepadidae should be based mainly on more reliable arthropodal characters and DNA barcoding, while the plate arrangement should be used carefully with their intraspecific variation in mind. We show morphologically that Neolepas specimens collected from the South West Indian Ridge, the South East Indian Ridge and the Central Indian Ridge belong to the new species. Molecular phylogeny and fossil evidence indicated that Neolepas migrated  across all three oceanic ridges in the Indian Ocean, as well as the global phylogeography of living eolepadid barnacles.

Morphological examination
The barnacles were dissected and the body, including six pairs of cirri, the oral cone, the caudal appendages and the penis, were examined by light microscopy (Zeiss Axio-scope and stereomicroscope Leica M80). The terminology used to describe eolepadid barnacles herein follows those in the previous studies [12,27], whereas the setal classification and description follow the more general terminology for barnacles overall [32]. Type and voucher specimens were deposited in the National Museum of Nature and Science, Tsukuba (NSMT) and the University Museum, the University of Tokyo (UMUT).

Comparison of capitular morphology between Kairei and Solitaire populations
To compare morphological differences between specimens taken from Kairei and Solitaire hydrothermal vent fields, the peduncle length, capitular height, height of rostrum and median latus, number of peduncular scales per whorl just below the capitulum region, width of scales (from three scales), size of scales projected from the peduncles (from three scales) were measured using a digital caliper (±0.1 mm). The angle of the tergal apex was measured from photographs showing the lateral view of the capitulum, using the image analysis software Sigma Scan Pro 5. For each specimen, the ratio of peduncle : capitulum length, the ratio of rostrum : median latus, the size of projecting scales and the tergal apex angle were obtained. Variation in each capitular character between the two populations was tested using either t-test or Wilcoxon Rank Sum test (when the normality assumption was violated).

Molecular phylogenetic analysis
Genomic DNA was extracted using DNeasy Blood & Tissue Kit (QIAGEN) from the adductor muscle of barnacle specimens. Partial sequence of the mitochondrial cytochrome c oxidase subunit I (COI) gene was amplified by polymerase chain reaction (PCR) using universal primer sets (LCO1490 and HCO2198, COI-3 and COI-6 [33,34]) and the Premix ExTaq Hot Start (TaKaRa). PCR was carried out in the following steps: initial denaturation at 94°C for 3 min and 35 cycles of denature (94°C for 30 s), annealing (50°C for 30 s) and extension (72°C for 90 s). PCR products were purified using Exo-SAP-it (USB, Affimetrix), following standard protocols. After BigDye reaction with BigDye Terminator v. 3.1, the products were sequenced using an ABI3130 automated sequencer (Applied Biosystems, Thermo Fisher). Electrophenograms obtained were checked by eye and assembled by Geneious v. 9 (Biomatters Limited) and registered to DNA Data Bank of Japan, with accession numbers LC350007-LC350015. The sequences obtained were aligned with eolepadid sequences available in the databases of the International Nucleotide Sequence Database Collaboration, using Clustal X included in MEGA v. 6.06 [35]. A total of 123 sequences from seven eolepadid taxa were used (4-45 individuals per taxa), with one sequence of the pollicipedid barnacle Capitulum mitella (Linnaeus [36]) as the outgroup. Electronic supplementary material, table S1 shows the full list of sequences used in this study. The model selection programme in the same software was applied to select the best model for the maximum-likelihood algorithm, which was the Tamura three-parameter + Gamma distribution model. MEGA v. 6.06 was also used to reconstruct the phylogenetic trees using the maximum-likelihood algorithm, with 2000 bootstrap replicates.
Order Scalpelliformes Buckeridge & Newman [38].            R/V Yokosuka cruise YK13-02 (principal scientist: Manabu Nishizawa), 11 February 2013. One lot of five juvenile specimens (UMUT RA32761), same data as above. Further specimens used for measurements and DNA sequencing: nine specimens from Solitaire vent field (same data as above) and 10 specimens from Kairei vent field (same data as holotype or paratype #1).
Distribution. Presently known from Kairei and Solitaire hydrothermal fields in the CIR (greater than 2500 m depth) and Longqi hydrothermal field in the SWIR (has also been referred to as the 'Dragon vent field' [6,16]). We consider the dredged material from 41°S site (site 21), SEIR [12] also represents the same species (see Discussion below).  Although the present new species is clearly genetically distinct from both of these species, the genetic distance between N. zevinae and N. rapanuii seemed insufficient for separation at species level [6]. However, the capitular arrangement of these species exhibits difference, with the rostrum being as high as the median latus in N. rapanuii, about the same in N. marisindica sp. nov. and higher than the median latus in N. zevinae [22]. The main difference among these three species is seen in the morphology of the mandibles. Neolepas zevinae has a tridentoid mandible in which the second tooth of the mandible has elongated teeth ( fig. 2i in [20]), N. rapanuii has a quadridentoid mandible with a small fourth tooth inbetween the third tooth and the inferior margin ( fig. 3d in [22]). In N. marisindica sp. nov., the mandible is tridentoid and without any small elongated teeth on the mandibular teeth. Among the three Neolepas   Morphological variations. The Kairei field population specimens were with orange-coloured peduncle and the capitulum coated with dark brown mineral deposits (figures 3a and 4); the Solitaire field population was whitish and without mineral deposits (figure 3b). A specimen (6 K-1327-R2-1) from the Solitaire field was dissected to demonstrate the variation in the external morphology. Compared to the holotype, the specimen from Solitaire field had a wider tergal apex angle, at 73°. The ratio of rostrum to median latus was 1.3. Twelve peduncular scales present per whorl at the region below the capitulum. Scales were approximately 2.4 mm wide and projected 1.65 mm out of the peduncle (figure 3b).
Arthropodal characters from the Solitaire field specimens are similar to those of the holotype (Kairei field). Six pairs of cirri: cirral counts of both anterior and posterior rami of each cirrus are similar between the holotype and the Solitaire specimen concerned ( figure 10 and table 1). Maxilla and maxillule of the specimen from the Solitaire field do not show great variation from the holotype ( figure 11a-d). Both maxillule and maxilla with simple type setae. Mandibles tridentoid and without extra small elongated teeth on the second tooth (figures 11d-h and 12a,b). Mandibulatory palp elongated with simple setae (figure 12c), labrum with a single row of small teeth ( figure 12d-h).
Comparing variations in capitular morphological characters between the Kairei population (10 specimens) and the Solitaire population (9 specimens), both populations shared similar peduncle characters: capitulum ratio (3 in Kairei and 3.8 in Solitaire), rostrum to median latus ratio (1.34 in Kairei and 1.45 in Solitaire), tergal apex angle (68 in Kairei and 71 in Solitaire) and the number of scales per whorl (20 in both populations; table 2). However, the Kairei specimens had significantly smaller scales (scale width 0.8 mm) when compared with the Solitaire population (1.54 mm; t-test, t = 3.5, d.f. = 17, p < 0.05; table 2, also figure 3a,b).

Molecular phylogenetic analysis
The reconstructed phylogenetic tree based on the maximum-likelihood algorithm is shown in figure 13. The relationships among eolepadid species were the same as previously shown [6], except for the additional OTUs of Vulcanolepas cf. parensis in Manus Basin [39], which was shown to share some haplotypes with L. longa in TOTO Caldera and Edison Seamount [6]. Neolepas marisindica sp. nov. from the three populations formed a single clade with previously reported sequences [6], which was sister to V. scotiaensis in the Southern Ocean. The N. marisindica sp. nov.-V. scotiaensis group is a sister group to EPR and Southern EPR populations of N. zevinae-rapanuii complex, whose outgroups consist of the undescribed Vulcanolepas species from the Lau Basin and the Tonga Arc [6] and V. osheai from the Kermadec Arc. the two CIR populations from Kairei and Solitaire hydrothermal vent fields of Neolepas marisindica sp. nov. examined in this study exhibited some differences in morphologies of capitulum and scales on peduncle, despite a lack of distinct sequence divergence in the COI gene between the two populations. The Solitaire hydrothermal field population, where diffuse flow venting was dominant, had larger scales with width greater than 1 mm compared with those from the Kairei hydrothermal field, where vigorous venting from black-smoker chimneys was dominant, whose scale width was approximately 0.8 mm. This difference was supported by statistical significance (p < 0.05; table 2). Morphological variations in neolepadines were also reported for Vulcanolepas scotiaensis in hydrothermal vent fields in the East Scotia Ridge, Southern Ocean, which exhibit a 'robust' form with short peduncle of peduncle : capitulum ratio as 1 : 1 in the site with low hydrothermal activity and a 'gracile' form with long peduncle of peduncle : capitulum ratio up to 20 : 1 in the site with active diffuse venting, but molecular analysis could not detect differences between the two [27]. The peduncular length is also variable in Vulcanolepas parensis, compared with the congeneric V. osheai and L. longa [27]. In the recent revision of taxonomy of Eolepadidae [27], the size of peduncular scales was used to discriminate Vulcanolepas and Neolepas, and the angle of tergal apex was considered diagnostic for Leucolepas and Neolepas. The presently examined specimens of N. marisindica sp. nov. exhibit intermediate characters between Vulcanolepas and Leucolepas in these two characters, respectively. The peduncular scales in the Kairei population are projected less than 1 mm out from the peduncle (table 2), which is within the diagnostic range indicated for Vulcanolepas [27]. Some individuals of N. marisindica sp. nov. had tergal apex angles of approximately 60°, which is supposedly a characteristic of Leucolepas (diagnostic tergal apex angle in Neolepas is 75° [27]).
The ratio of rostrum to median latus, as a key diagnostic character, is said to be 1.5 for Neolepas and 1 for other genera [27]. In this study, we found variations in the rostrum to median latus ratio among different specimens of N. marisindica sp. nov., which ranged from 1.0 to 1.8. The ratio of lengths of different capitular plates is clearly a continuous variable and it is highly problematic to treat these as the only diagnostic character for genus or even species identification, unless the range of variation is taken into consideration. We, therefore, suggest that the diagnostics and identification of Neolepas species is best relied upon investigation of arthropodal characters including mandibles, while also carefully considering their plate arrangement (with the intraspecific variation in mind), coupled with molecular DNA barcode analysis. The present phylogenetic analysis was consistent with previous molecular studies [6,27], showing a close relationship between V. scotiaensis and members of the genus Neolepas ( figure 13). This is different from taxonomic assignments based solely on hard part morphology, where V. scotiaensis was placed close to other Vulcanolepas species such as V. osheai [27]. These two species are then, in turn, sister to a clade consisting of N. zevinae and N. rapanuii, which appear to be genetically indistinguishable, at least using COI barcodes. This means V. scotiaensis is nested within the genus Neolepas. In addition, the mandible morphology of V. scotiaensis is actually very similar to those of other Neolepas species, as it has none or only minute longitudinal teeth. These results combined provide strong evidence that V. scotiaensis, in fact, belongs to the genus Neolepas, and therefore it is here formally transferred to Neolepas, as Neolepas scotiaensis (Buckeridge et al. [27]) comb. nov.
In contrast with high plasticity in the hard parts, morphologies of arthropodal characters are relatively stable and well supported by molecular phylogenetics. In this study, mandibles of the dissected individuals exhibited very similar morphological characteristics (figures 6e, 7e, 11e, 12e), whereas their hard part morphologies were more different ( figure 3). The morphology of mandibles of Neolepas from SEIR [12] was the characteristic of N. marisindica sp. nov., as it lacks small longitudinal teeth on the second tooth. Therefore, we here consider these specimens to represent a further population of N. marisindica sp. nov., extending its distribution to SEIR, at least to 41°S. The DNA barcoding sequences of Neolepas marisinsica sp. nov. collected from Kairei and Solitaire hydrothermal fields on the CIR could not be separated from the Longqi population previously reported from the SWIR [6], confirming the distribution of the present new species on the SWIR, at least as far as the Longqi field. Therefore, N. marisindica sp. nov. is the only species of vent animal so far confirmed to range across hydrothermal vents in all three Indian Ocean oceanic ridges-the CIR, the SWIR and the SEIR. The fact that the same haplotypes have been recovered multiple times from populations on the CIR and the SWIR indicates that N. marisindica sp. nov. probably has sufficiently high dispersal ability to contain a metapopulation connecting the CIR and the SWIR across the Rodriguez Triple Junction, while for the scaly-foot gastropod Chrysomallon squamiferum the triple junction is known to act as a dispersal barrier [40]. As no vent on the SEIR has been visited by a submersible, further investigation of vents on the SEIR and samples from there will certainly reveal valuable information on the biogeography of deep-sea hydrothermal vent fauna in the Indian Ocean.

Phylogeography of vent barnacles
The phylogenetic analysis of vent stalked barnacles here elucidated their historical migration patterns across a geological timescale. As V. cf. parensis in Manus Basin shared some haplotypes with L. longa in TOTO Caldera and Edison Seamount (figure 13), here we regarded V. cf. parensis in Manus Basin as misidentification of L. longa. Leucolepas longa from the Mariana Forearc and the Manus Basin diversified at the most basal branch in the neolepadines, subsequently Vulcanolepas osheai in Kermadec Arc, and then an undescribed Vulcanolepas from the Lau Basin diversified, and finally the monophyletic Neolepas ( figure 13). This renders Vulcanolepas paraphyletic. A previous tree published by Herrera et al. [6] combined three genes (28S, H3 and COI), however, with a different pattern with the basal branching being between a monophyletic Neolepas and a Leucolepas-Vulcanolepas clade. This node was highly supported in their study (0.88 and 100 for Bayesian posterior probability and bootstrap value, respectively). Considering that in our tree the node splitting Leucolepas from Vulcanolepas-Neolepas was not statistically supported (less than 0.70 in bootstrap probability), we interpret that the branching pattern observed in Herrera et al. [6] is more reliable. We, therefore, consider Vulcanolepas and Neolepas to be separate genera, following Herrera et al. [6]. On the other hand, the branching pattern within Neolepas (i.e. N. zevinae/rapanuii complex in the southern EPR, then N. marisindica sp. nov. in the Indian Ocean and finally N. scotiaensis in the Southern Ocean) was supported by high bootstrap probabilities (greater than 95 in bootstrap probabilities; figure 13).
The branching pattern, indicating close relationships between the species in Indian and Southern oceans compared with those in the southern EPR, is consistent with the pattern reported for the 'yeti crabs', squat lobsters in the genus Kiwa [7]. Neither Neolepas nor Kiwa has been reported from the Atlantic Ocean (except on the ESR of the Southern Ocean, which is technically in the extreme southern Atlantic), and their distribution and historical migration may be similar. The divergence between N. marisindica sp. nov. and N. scotiaensis was 1.7 Ma (95% HPD: 0.4-3.8) and the divergence between N. marisindica sp. nov.-N. scotiaensis and N. zevinae/rapanuii complex was 6.4 Ma (95% HPD: 3.0-11.2) [6]. The divergence between Kiwa sp. SWIR and Kiwa tyleri Thatje 2015 in Thatje et al. [41] from ESR, Southern Ocean was 1.5 Ma (95% HPD: 0.6-2.3) and the divergence between these two Kiwa species and Kiwa hirsuta from