Echinoids and Crinoids from Terra Nova Bay (Ross Sea) Based on a Reverse Taxonomy Approach

: The identification of species present in an ecosystem and the assessment of a faunistic inventory is the first step in any ecological survey and conservation effort. Thanks to technological progress, DNA barcoding has sped up species identification and is a great support to morphological taxonomy. In this work, we used a “Reverse Taxonomy” approach, where molecular (DNA barcoding) analyses were followed by morphological (skeletal features) ones to determine the specific status of 70 echinoid and 22 crinoid specimens, collected during eight different expeditions in the Ross and Weddell Seas. Of a total of 13 species of sea urchins, 6 were from the Terra Nova Bay area (TNB, Ross Sea) and 4 crinoids were identified. Previous scientific literature reported only four species of sea urchins from TNB to which we added the first records of Abatus cordatus (Verrill, 1876), Abatus curvidens Mortensen, 1936 and Abatus ingens Koehler, 1926. Moreover, we found a previous misidentification of Abatus koehleri (Thiéry, 1909), erroneously reported as A. elongatus in a scientific publication for the area. All the crinoid records are new for the area as there was no previous faunistic inventory available for TNB.


Introduction
The increasing application of integrated taxonomy, coupled with new modelling approaches, requires data to be findable, accessible, interoperable and reusable in the long term [1].
The most common challenges facing studies or the construction of biodiversity inventories are accurate species identification and the absence of detailed information on the distribution of taxa throughout the different geographical regions of the planet [2]. Morphology-based identification represents the classic approach to taxonomy but is strongly dependent on the level of experience and expertise of the identifier. This method is, thus, largely prone to mistakes whenever intraspecific variability has not been previously tested. However, the increase in molecular advances has made it evident that this approach comes with some inherent limitations [3]. Taxonomic discrepancies, such as synonymous or cryptic species, are extremely common when a traditional taxonomic approach is used. Neither molecular nor morphological taxonomic methods are sufficient on their own [4] and the number of examples where this integrated approach is applied to identify species is rapidly increasing (sea stars (e.g., [5][6][7][8][9][10][11]), brittle stars (e.g., [12]), holothurians (e.g., [13]), fish (e.g., [14]) and many more.
With the rapid accumulation of samples in museums and the co-occurring decline in taxonomic expertise in recent years [15], molecular tools, phylogenetics and coalescentbased analyses have become the practices used for species identification or The objectives of the study are: (i) to update the checklist for echinoids of the TNB area; and (ii) to evaluate the first comprehensive inventory for crinoids from the same location. To achieve this goal, we used an integrated approach using both DNA barcoding and morphological characters. The results will serve as a baseline for future works in ecology, monitoring and management of the study area. The current paper represents a further contribution of the Italian National Antarctic Museum (MNA), Genoa section, as the custodian of biodiversity data for the Ross Sea area. Many contributions to the Antarctic Biodiversity Portal have been published by the MNA over the years, with the aim of increasing the knowledge of the area [11,[40][41][42][43][44][45][46][47][48] (http://www.biodiversity.aq, accessed on 25 November 2022).

Materials and Methods
The samples available at Italian National Antarctic Museum (MNA), Genoa section, analysed in this study derive from the Antarctic Peninsula ( Figure 1a) and the Ross Sea sector, specifically the TNB area, which is part of both the marine protected area and the Antarctic Special Protected Area (n.161) (CIT 62) (Figure 1b).
Specimens were collected in the framework of several recent scientific expeditions performed in the Southern Ocean and which are now permanently stored and curated at MNA. The Italian National Antarctic Program (PNRA) expedition "XVII"

Sampling and DNA Extraction
A total of 92 samples, 70 belonging to echinoids and 22 to crinoids, were analysed and the distributional data considered here originated from 31 different sampling stations, ranging from 15 and 750 m in depth (Table 1). Sampling was performed through deployments of a variety of sampling gear comprising SCUBA diving (see Table 1 for details). Six pentacrinoid larvae (MNA-03760, MNA-03766, MNA-03795, MNA-03855 and MNA-07967) were included in the analyses and were obtained by examining biological materials (e.g., polychaetes tubes) on which they settled; one of these (MNA-09159) was found on the metal structures of Mooring L.
Whenever possible, following the collection and sorting phase, the live specimens were photographed by one of us (SS) to avoid the loss of potential diagnostic characteristics such as colours that would fade or disappear once the organism was fixed in ethanol. Samples were fixed in ethanol (95% Et-OH) or frozen (−20 °C) in order to preserve them for further genetic analysis. Sorting and classification on a morphological basis were performed at the MNA using the available literature and keys from Koehler (1926) [49], Clark (1967) [50], Moore (1983) [51] and Speel at al. (1983) [52]. All the samples were acquired as permanent vouchers at the MNA (https://steu.shinyapps.io/MNA-generale/, accessed on 21 November 2022). The clipped material from each sample was sent to the Canadian Center for DNA Barcoding using microplates (University of Guelph, Guelph, ON, Canada), which performed extraction, amplification and sequencing. Primers used for amplification were LCOech1aF1 (5′-TTTTTTCTACTAAACACAAGGATATTGG-3′) or EchinoF1 (5′-TTTCAACTAATCATAAGGACATTGG -3′) and HCO2198 (5′-TAAACTTCAGGGTGACCAAAAAATCA-3′). Sequences were uploaded to the BOLD platform (Barcode Of Life Data systems, http://www.boldsystems.org, accessed on 21 November 2022).  Manual taxonomic assignation was performed in BOLD (Accessed 16 November 2022) of which sequences are available for 5147 echinoid and 4291 crinoid specimens, representing, respectively, 307 and 203 species. Comparison was also performed in the National Center for Biotechnology Information (NCBI) database with BLAST (https://blast.ncbi.nlm.nih.gov/Blast.cgi, accessed on 21 November 2022) for definitive assignment. A correct identification was defined as a sequence match that exceeds 98% similarity to the reference database [53]. The taxonomic names and classification used in this study were obtained from the World Register of Marine Species (WoRMS) website (www.marinespecies.org, last accessed on 21 November 2022). Sequences were edited and corrected in CodonCode Aligner v9.0.1, developed by CodonCode Corporation in Centerville, MA, USA (http://www.codoncode.com/aligner/, accessed on 21 November 2022). The MUSCLE algorithm was used to align sequences, which is available within CodonCode Aligner, and result was visually inspected for accuracy. Odontaster validus Koehler, 1906 (GenBank accession number: ON103477) was selected as outgroup. The substitution pattern was determined by analysing the Bayesian information criterion (BIC) scores in MEGA X [43], and the T92 + G (Tamura 3-parameter + Gamma distribution [44]) model was found to have the lowest scores, indicating the best fit. Phylogenies were inferred using Bayesian, maximum likelihood (ML) and maximum parsimony (MP) approaches. Bayesian estimation of phylogeny was carried out using Mr Bayes [54,55]. Additionally, a generalized time reversible (GTR) model with gamma(G)-correction was used to avoid risk of obtaining unsupported results with under parametrization in Bayesian inference. Markov chain Monte Carlo (MCMC) algorithm with two simultaneous independent runs was performed starting from different random trees. Each run comprised four chains (one cold and three heated), which were sampled every 100 generations for a total of 2 × 10 8 generations. To ensure appropriate effective sampling size (ESS all > 100), Tracer v.1.6 was utilized. The final result trees for comparison were performed using FigTree v1.4.4 (http://tree.bio.ed.ac.uk/software/figtree/, accessed on 21 November 2022) for graphical representation. Sequences obtained in this study have been deposited in GenBank: accession numbers OR157781-OR157864.

Species Delimitation Methods
The assumption of using species delimitation methods dictates that two or more species are distinct by exhibiting a "barcode gap" [56]; that is, genetic variation between species (interspecific) greater than genetic variation within species (intraspecific) [57]. Four methods were conducted for primary species hypotheses to identify the number of molecular operational taxonomic units (MOTUs) within our dataset. The Barcode Index Number System (BIN) [58] and Automatic Barcode Gap Discovery (ABGD) [16] rely on pairwise sequence distances between specimens to determine the number of OTUs within a dataset. Standard BIN assignments are available on BOLD (http://www.boldsystems.org, accessed on 13 December 2022), but they are generated through the analysis of all barcode sequences on BOLD, meaning that the results are not strictly comparable with those obtained with other methods (because they are based on a more inclusive dataset). ABGD analysis was performed on the web interface (http://wwwabi.snv.jussieu.fr/public/abgd/, accessed on accessed on 15 December 2022), Kimura (K80) was used as the genetic distance with default range of 0.001 to 0.1 and was examined for intraspecific distances, while gap values from 1 to 1.5 were employed. The Generalized Mixed Yule Coalescent (GMYC) [59] differs strongly from the other methods because it is a model-based approach, aiming to discover the maximum likelihood solution for the threshold between the branching rates of speciation and coalescent processes on a tree. The tree-based methods employ a coalescent framework to independently identify evolving lineages without gene flow, each representing a putative species [60]. They can be performed using a single marker and are used to establish a threshold that identifies the separation of intraspecific population substructure from interspecific divergence. It therefore identifies those groups that may be candidate species [61]. The last species delimitation approach was implemented using a PTP process [62]. Here, we used the Bayesian implementation of the Poisson tree processes model (bPTP) [62], the ML tree was used as input. The bPTP analysis (species.h-its.org/ptp) runs parameters that were 500,000 generations of MCMC, a thinning of 100 and a 25% burn-in. In all the species partition methods used, the outgroup (Odontaster validus) was removed.

Morphological Identification
Following the "Reverse Taxonomy" approach [63,64], morphological analyses were conducted for a re-examination of our molecular results on available specimens. Observations were carried out under a stereoscopic microscope. For determination to species level, each sea urchin individual was identified according to the morphological features indicated in the taxonomic keys for Antarctic Echinoidea by Thomas Saucède (http://echinoidea-so.identificationkey.org/mkey.html, accessed on 13 January 2023). Crinoids were identified with available literature from Clark (1967) [50], Moore (1983) [51] and Speel at al., (1983) [52]. For echinoids, our morphological analysis focused on morphological skeletal features, such as accessory structures and spines. We particularly focused our attention on pedicellariae, which are defensive structures consisting of a head composed of two or more valves hinged to one another, a stem and sometimes a neck. The four main types of pedicellariae analysed were: globiferous, dentate, triphyllous and ophicephalous.
Given the taxonomic relevance of pedicellariae shape morphology for species identification, the small mandibular appendage that articulates on the test was removed from selected samples corresponding to putative species partition highlighted by the molecular analysis. The tissue portion was treated with sodium hypochlorite (NaClO) to remove organic matter. Subsequently, the skeletal elements obtained were washed with deionized water then after with water and ethanol (Et-OH). Proportions were increased until the skeletal elements were completely washed with 100% Et-OH. This made it possible to observe skeletal characteristics in detail under the stereomicroscope in order to obtain the correct identification of the species.
For crinoids, we compared the external morphological features. All diagnostic characters were analysed in detail, including the cirri, oral pinnules, genital pinnules, arm number, and segments of the cirri and arms under a stereomicroscope. Specimens identified in this study showed morphological characteristics corresponding to those described in the literature [39][40][41], and molecular species identification was cross-referred with the morphological result.

Results
A total of 92 specimens were analysed in the current study, and all were correctly sequenced to obtain a final COI sequence length of 628 bp. The COI dataset employed for analyses is reported as Supplementary Material (M1). Of the 92 sequences generated in this study, 70 belonged to echinoids and 22 to crinoids. All sequences were barcodecompliant (Supplementary File S2) and received a BIN, which aided species delimitation [58]. The other species delimitation methods recovered a different number of secondary species hypotheses (SSH) for sea urchin, but were all in agreement regarding the crinoid's investigation (Supplementary File S2). The most problematic method was bPTP because in the echinoids' SSH investigation, it showed an overestimation in species partition. The maximum likelihood and Bayesian analysis results were consistent and revealed 13 putative species of echinoids and 4 of crinoids (Figure 2 and Supplementary File S2).  [58]; ABGD: results from automatic barcode gap discovery method [16]; GMYC: species delimitation from generalized mixed Yule coalescent method [59]; bPTP: species delimitation using Bayesian Poisson tree processes method [62.

Molecular Results
Identification through barcoding requires specimens from the same species to cluster together using the barcode markers. Detailed and high-resolution trees' comparison (ML and Bayesian interference) with species partition method results are available in Supplementary File S3.

Crinoidea
The 22 crinoids analysed here were assigned to four morphospecies, all of them corresponding to described and well-known species. Our crinoid specimens were correctly grouped into four putative species by the species delimitation methods, showing consistency between the analyses. Posterior probability node values, which are shown on the tree (Supplementary File S3), range from 47% to 100% for ML tree reconstruction and a value included from 0.56 to 1 in Bayesian interference. In our samples, no corresponding sequence matched Abatus koehleri (Thiéry, 1909) (previously reported from the TNB area as A. elongatus), a species previously reported from Terra Nova Bay water [55].

Morphological Analysis
A total of 70 echinoid and 22 crinoid individuals were morphologically examined after primary species partition based on molecular screening following the "Reverse Taxonomy" approach. Clades 12 and 16 were assigned on a sole morphological base, as they did not match any sequence in the online databases.
The main descriptors to distinguish the species are given below for crinoids and echinoids, respectively, in Tables 2 and 3; the more in-depth descriptions of the species are reported in Supplementary Materials S4. The results of the molecular analysis were combined with the morphological results following the "Reverse Taxonomy" approach, and the species partition was consistent. To optimize the visualization and understanding of the results, the tree in Figure 2 was subdivided, highlighting the class of crinoids, in Figure 3, and echinoids, in Figure 4, with the available representative photos of selected specimens.    Figure 2, the portion analysed in detail in the image is highlighted in red. Scale bar: 1 cm in grey.

Faunistic Inventory Revision
In our analysis no samples corresponding, morphologically or molecularly, to A. koehleri, a species identified with classical morphology by Chiantore et al., 2006 and reported in that publication with the old name of A. elongatus [38] were found.
This was unexpected and we thus cross-checked all the available materials present in the Italian National Antarctic Museum (MNA, Section of the Genoa) collections. Unfortunately, only a small amount of previously studied and published material has been later given to the museum, preventing a general in-depth re-evaluation. However, sample MNA-00573 was found to belong to the bulk of specimens published by Chiantore et al., 2006, and still reported an original identification label indicating it as A. elongatus. During our study, this same sample was successfully sequenced, morphologically reviewed bringing to an undoubted identification of A. shackletoni Koehler, 1911. In the light of this result, we believe that the presence of A. koehleri, previously reported as A. elongates by Chiantore et al., 2006, in the Terra Nova Bay area has to be considered questionable. Hence this is the same for the published in the Southern Ocean Echinoid database (e.g., [33,34]) that are based on the same publication. This modifies the number of previous identified species from TNB area from four to three.
Overall, by combining molecular and morphological identifications, we found three more echinoids species, i.e., A. cordatus, A. curvidens and A. ingens, not previously reported in TNB, which bring the total number of echinoids present here to 6 species. The revised check list is given in Table 4, together with an updated depth range for the considered species ( Figure 5).  Figure 5. List of echinoids species found in Terra Nova Bay and updated depth range [38].

Discussion
The identification of species in an ecosystem is the first step in any ecological and conservation study, but even to provide a list of species for a given area is not an easy task. In recent years, one of the main hurdles in this kind of activity has become the chronic lack of experienced taxonomists. This fact, coupled to the generally time-consuming nature of morphological investigation, had the general effect of a significant slowdown of the duration of a given study.
The possible presence of cryptic species complexes represents another challenging aspect of biodiversity studies. These organisms are remarkably similar in appearance to other closely related species, resulting in them being virtually indistinguishable from the latter based on traditional morphological characters alone. Consequently, cryptic species are often overlooked, thus introducing serious biases in species richness estimates and in conservation efforts. Moreover, it is always possible that some of the unrecognized species in a sample collected in a newly studied area are new to science. This is a not a remote risk if it is considered that only 25% out of the 0.7 to 1.0 million marine species seems to have been described to date [65]. In terms of conservation efforts, this means that species could face extinction before they can be described [66].
A solution proposed to overcome all of the above issues relies on the use of molecular tools, such as DNA barcoding [3,19,57]. This method gained more attention in the last decade due to the increase in the speed of laboratory procedures, the availability of ad hoc software to better define species hypotheses and the high reproducibility of all these analyses.
The efficiency of identification through barcoding, however, depends on the quality and completeness of a reference database of sequences [67,68]. To this aim, projects such as the "The Barcode of Life" (BOLD) [69] directly and indirectly encourage large-scale molecular studies with a higher focus on quality. One of the main foundations of BOLD is the attention to voucher specimens that intrinsically provide opportunities for morphological and molecular studies using the same specimens as well as for subsequent cross-checking of identifications whenever disagreement is found. Another key point of a good reference library is the geographical coverage. In fact, data from few locations may only lead to an incomplete understanding of intraspecific genetic diversity, and a comprehensive DNA barcode library should include a broad range of species from as many locations as possible.
However, neither molecular nor morphological methods are sufficient per se for accurate taxonomy, and only the combined use of several methods provides a robust way to obtain a more precise estimation of species boundaries.
An ideal approach is, thus, that of "reverse taxonomy", where morphological analyses are performed after an initial molecular assessment [63,64].
In the case of echinoderm research, the application of "reverse taxonomy" has provided numerous benefits. Echinoderms, in fact, are known for their complex morphological features and challenging taxonomy, resulting in subjective and prone-tomisidentification species recognition [5]. This is especially true in cases where external morphology alone may not provide sufficient differentiation or when it is applied to largescale sampling activities or to museum collections where thousands of samples have to be processed.
Application of this integrated approach also allows accurate identification of damaged specimens or larval stages, which can be challenging or simply impossible to determine at the specific level by using traditional taxonomy alone.
The Southern Ocean is not an exception to all these problems, and whenever this combined approach was applied, unexpected outcomes emerged and even apparently complicated taxonomical situations were resolved (e.g., [11,25]).
In this contribution, we applied a "classic" reverse taxonomy approach on two classes of echinoderms, i.e., echinoids and crinoids, in order to verify the number of species present in our samples and to test its usefulness in enhancing our understanding of echinoderm biodiversity.
Molecular data were fully resolved by current available algorithmic methods for species delineation applied to DNA barcodes. Although this test involved low sample sizes, it provided an estimate of the relative efficacies of OTU designation via DNA-based methods and external morphology, speeding up the process of species identification.
Since in this case limited or no previous knowledge was available for the considered group, if the initial phase of taxonomic work would have been based on a pre-sorting based on external morphological characters alone, this step would only have slowed down the whole process and several species would not have been recognized.
Overall, in this study, we contributed nearly 100 new sequences, including sequences from two species not yet available in public databases (i.e., A. curvidens and A. cordatus), enhancing our understanding of echinoid and crinoid biodiversity not only for the TNB area, but also for the Southern Ocean in general. This is a significant result when considering that echinoderms represent a considerable biomass in marine habitats and play a major role in the Antarctic marine ecosystems [35,[70][71][72][73][74][75][76].
Species recognition was also possible for the six pentacrinoids larvae that were identified by comparing COI sequences against the reference sequences held in GenBank, confirming the importance of DNA barcodes in order to identify juvenile organisms and larval stages where morphological identification could be challenging.
Two putative new species (Antrechinus sp. and Ctenocidaris sp.), defined here based on COI sequences alone, need to be better characterized with integrated taxonomy to resolve their status. The sharing of this information may speed up comparisons with museum materials from other institutions, allowing, in the end, to formally assign a species name to these COI-based putative species.
The application of a reverse taxonomy approach has proven to be an efficient tool, even for checking the identity of old, already published, museum materials, highlighting the necessity of maintaining permanent repositories of scientific samples for future generations and comparisons.
This reinforces the pivotal role that museums play, not only as conservation centres for biological collections but also as hubs for information sharing. It is desirable that in the coming years, all available museum collections undergo molecular identifications to accurately assess species determinations and occurrences, and that all data are at the foundation of any monitoring activity.
Additionally, this work lays the foundations for future research on the diversity within the TNB area, now part of the Ross Sea Marine Protected Area, where a variety of monitoring activities are requested by the conservation measures of Annex 91-05/C [59].

Supplementary Materials:
The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/d15070875/s1. Supplementary Materials M1: COI sequence dataset produced in this study. Supplementary File S1: List of specimens analysed in this study and corresponding GenBank accession numbers for COI sequences; File S2: Species partition methods; File S3: Tree topology of the 92 samples analysed and species partition methods; File S4: Morphological description of identified species. The data presented in this study are openly available in GenBank (accession numbers: OR157781-OR157864). Institutional Review Board Statement: All Italian sampling activities in Antarctica were authorized by the Italian National Antarctic Program (PNRA).

Data Availability Statement:
In accordance with FAIR principles, the COI sequence dataset produced in this study (Supplementary Materials M1) can be found in the Supplementary Materials and is openly available in GenBank (https://www.ncbi.nlm.nih.gov/genbank/, accession numbers: OR157781-OR157864).