Molecular phylogeny suggests synonymy of Sandalia bridgesi within Sandalia triticea (Gastropoda: Ovulidae)

The Ovulidae are the most ecologically diverse family of the order Littorinimorpha. And the Sandalia is a small genus containing only three species. In the present study, we collected 44 specimens of Ovulidae (Gastropoda: Cypraeacea), belonging to six genera and eight species, including 20 Sandalia bridgesi and 2 Sandalia triticea individuals. The cytochrome c oxidase I gene (COI), 16S rRNA, and ITS1-5.8S-ITS2 sequences were obtained from the specimens and compared with sequences downloaded from GenBank to calculate genetic distances and construct phylogenetic trees. The sequences of S. bridgesi and S. triticea exhibited a high degree of similarity, and the S. bridgesi does not formed a separate clade, supporting the proposition that S. bridgesi should be considered a synonym of S. triticea.

transparency and usually possesses a red or purple shell and light-coloured callosities. The middle portion of the dorsal side is normally light coloured, and the shell is roughly pear-shaped. Under ultraviolet (UV) light, yellow fluorescence can be seen in the terminal collars, callosities, dorsal mid-portion, and sometimes the entire shell. By contrast, S. bridgesi has a relatively uniformly coloured dorsal portion and a wider anterior part, and only exhibits fluorescence in a small area on either end of the shell under UV light.
DNA barcoding, which involves using a short DNA sequence for species classification, has received widespread attention since its advent (Meier et al. 2006). First proposed by Hebert et al. (2003), DNA barcoding was initially used as a tool for species identification. In the present study, we sequenced the cytochrome c oxidase subunit I (COI) gene, 16S rRNA and ITS1-5.8S-ITS2 (ITS) region for the construction of phylogenetic trees to elucidate the relationship between S. bridgesi and S. triticea. We obtained the sequence data for the internal transcribed spacer (ITS) regions of Ovulidae for the first time.

Material and methods Specimen collection
We collected 44 specimens of Ovulidae from depths of 0-6 m during spring low tides in Beihai City, Guangxi Province, and the cities of Xiamen and Quanzhou in Fujian Province between July and November 2020. Detailed information of the collected specimens is shown in Table 1. Specimens of the coral hosts in which the ovulids resided were also collected. All specimens were morphologically identified by experts and taxonomists in accordance with the identification keys published in the Fauna Sinica (Ma 1997 The specimens were observed and photographed on the dorsal, ventral, and lateral sides under a Leica S9D stereomicroscope ( Fig. 1 & Fig. 2). Specimens of Sandalia were preserved in 95% alcohol, and coral specimens were preserved using methods for the preparation of dried specimens.

DNA extraction and sequencing
Amplification was performed on three gene regions in each specimen, namely the mitochondrial markers 16S rRNA and cytochrome c oxidase subunit (COI) and the ITS region.
DNA was extracted from each specimen's muscle tissue using the DNeasy Blood & Tissue Kit (QIAGEN, China) following the corresponding protocol for animal tissues. Nucleic acid concentration in the DNA extracts was measured using BioDrop (BioDrop, UK). Due to the presence of inhibitors in the specimen tissues, all DNA extracts were diluted 50-500 fold before PCR amplification (Reijnen and van der Meij 2017;2019). Our experimental results indicated that the appropriate concentration for the diluted DNA extracts was approximately 0.2 µg/mL. Each PCR had a reaction volume of 50 µL and contained the following: 25 µL PCR mixture [Taq plus Master Mix II (Dye Plus)], 2 µL of each primer (10 µM), 5 µL (diluted) DNA extract, and 16 µL extra pure water. Table 2 shows the details of the PCRs performed for the three gene regions. Not all markers were successfully amplified for all specimens, successfully amplified COI, 16S rRNA, and ITS samples were sent to Sangon Biotech Co., Ltd (Shanghai, China) for PCR cleaning and sequencing. The sequences were submitted to GenBank (http: //www.ncbi.nlm.nih.gov) ( Table 1).

DNA data processing and molecular analyses
Joining and alignment of the sequences and trimming of ends with low signal strength were performed using DNAMAN v. 9 (Lynnon Biosoft, Canada) and SeqMan v. 7.1.0 (DNAStar, U.S.A.). The sequences were then converted to the .fas format using the EditSeq module in Lasergene (DNAStar, U.S.A.) (Burland and G 2000). Multiple sequence alignment was performed using the ClustalW tool in MEGA X (Kumar et al. 2018).
Phylogenetic trees were constructed, and genetic distances were calculated using MEGA X, with the precision of calculations set to eight decimal places. Models with the lowest Bayesian information criterion (BIC) scores, considered to best describe the substitution patterns, were determined for all sequences using the Find best DNA models (ML) option in MEGA X (Nei and Kumar 2000;Tamura et al. 2011). Phylogenetic trees were also constructed using the Kimura 2-parameter (K2P) method (Kimura 1980). All the data were set to 1000 bootstraps in the test for inferred phylogenies to estimate the internal stability of the tree nodes. As the results from the two different phylogenetic reconstructions were congruent, only the phylogenetic trees generated from models with the lowest BIC scores have been shown in the present paper.
Using pairwise distance data exported in table form from MEGA X, heatmaps were created with Microsoft Excel (Microsoft Office 2016) and Adobe Illustrator 2020.

Morphological data
Based on the photographs and descriptions provided by Lorenz (Lorenz 2009;Lorenz and Fehse 2009) and Hardy (2020a), the 44 specimens were identified as belonging to six genera and eight species. Twenty-two of the specimens were Sandalia species, among which 20 were identified as S. bridgesi, and two were identified as S. triticea based on differences in shell transparency, external appearance and colour.
White coral specimens were morphologically identified as Hicksonella guishanensis, and red coral specimens were morphologically identified as Melithaea japonica (Ma 1997;Mao and Jiang 1994). Table 1 shows the information and identification outcomes of the collected specimens.

Molecular data COI
Forty-one COI sequences were successfully amplified. After editing, the consensus length of all barcode sequences was 615 bp, and no stop codons, insertions, or deletions were observed in any of the sequences. All analysed sequences were larger than 600 bp. The sequences were compared with the COI sequence of S. rhodia [synonym: S. triticea (Sartori 2014)] downloaded from the National Center for Biotechnology Information (NCBI) (GenBank-MG450349). Phylogenetic tree construction was performed using the HKY+G+I model (Tamura 1992) (Fig. 3), and the root location was confirmed by selecting the COI sequence of Mauritia arabica (GenBank-JF693392) as the outgroup. The data processing methods used for the amplified sequences were also adopted for the two sequences downloaded from NCBI.

16S rRNA
Thirty-nine 16S rRNA sequences with lengths of approximately 520 bp were successfully amplified. After trimming, segments with lengths of 460 bp were obtained and compared with the 16s rRNA sequence of Cypraea gracilis downloaded from NCBI (GenBank MH571331) to find the best model using the Bayesian approach. Results indicated that the T92 (Tamura 3-parameter model)+I (Tamura et al. 2011) model was the most optimal evolutionary model with the lowest BIC score (Fig. 4).

ITS1-5.8S-ITS2
The amplified ITS sequences had lengths of 1200-1300 bp before trimming and approximately 1200 bp after trimming. Results of analysis using the Bayesian approach showed that the K2+G model (Kimura 1980) had the lowest BIC score. The ITS sequences' phylogenetic tree was unrooted due to the unavailability of outgroup sequences (Fig. 5). The greatest and smallest genetic distances between S. triticea and S. bridgesi were 0.0051% (20200722-T2 and 20200722-H1) and 0% (20200722-T1 and 20201117-H2), respectively, and the smallest genetic distance among the other specimens was 0.0980% (20201117-H6 and 20201117-H13).

Morphological data
S. bridgesi is a new species established based on morphological characters (Lorenz 2009). It differs from S. triticea mainly in shell transparency and length-to-width ratio. However, in our collected specimens, we observed the presence of a continuous transition in the length-to-width ratio and variations in transparency with observation angle, light intensity, and individual differences. Additionally, we noticed the presence of coevolution effects between Ovulidae and Gorgonacea. Shell and mantle colour also show a high degree of variability due to influences by various environmental factors and therefore, cannot be used as marker characteristics to distinguish between species (Schiaparelli et al. 2005). For instance, Diminovula culmen (Lorenz and Fehse 2009), Serratovolva dondani, and Crenavolva striatula (Hardy 2020c) exhibit diverse shell colour changes. In certain species such as Crenavolva aureola, coloured bands on the shell cannot be regarded as stable traits (Hardy 2020b). During our species identification process, both S. bridgesi and S. triticea specimens were irradiated with UV light with wavelengths of 395, 365, and 254 nm, but the yellow fluorescence reported by Lorenz (2009) could not be observed. Therefore, fluorescence may not serve as a stable trait in Ovulidae. Given the subjectivity and instability involved in morphological identification, the use of molecular data for taxa identification may be an effective method for resolving the issues described above.

Molecular data
COI barcoding has been widely applied in identifying species belonging to the class Gastropoda (Hou et al. 2013;Layton et al. 2014;QUINTERO-GALVIS and RAQUEL-CASTRO 2013;Stothard and Rollinson 1997). Research evidence has shown that COI is more capable of reflecting geographical differences than shell characters in certain taxa (Simison and Lindberg 1999). As Ovulidae evolve more rapidly than other Gastropoda taxa, they possess high phenotypic plasticity (Lorenz 2020; Reijnen and van der Meij 2017; Sánchez et al. 2016), leading to greater difficulties in classification solely based on morphology. In recent years, researchers have utilised COI and 16S rRNA in investigating the phylogeny of Ovulidae and found that both are capable of distinguishing specimens at the species level (Reijnen and van der Meij 2019; Sánchez et al. 2016;Schiaparelli et al. 2005), resulting in the discovery of synonymy among ovulid species (Reijnen 2015). Meyer and Paulay (2005) utilised barcoding in the analysis of the sequences of more than 2000 individuals in 263 taxa of the family Cypraeidae, the sister group to Ovulidae (Meyer 2003;2004;Rosenberg 1992), and found that identification of unknowns was 98% accurate with a neighbourjoining approach against an evolutionary significant unit (ESU) phylogeny. The correspondence between ESU definitions and traditional morphological taxonomy was high, with 255 ESUs (97%) recognised previously at either the specific or subspecific level, indicating that an ESU is a taxonomic unit equivalent to or smaller than a species. Therefore, traditional taxonomy within Cypraeidae at the species or subspecies level is supported by molecular data in addition to independent morphological criteria.
From the phylogenetic tree constructed using COI sequences (Fig. 3), it can be observed that the sequences of S. bridgesi and S. triticea were clustered on the same clade, indicating the absence of significant genetic differentiation between the COI sequences of these specimens. Other clades were also well-supported, which is in agreement with the findings of Meyer et al. The smallest genetic distance among the COI sequences of specimens other than S. bridgesi and S. triticea was approximately 4.8 times that of the greatest genetic distance between S. bridgesi and S. triticea, clearly demonstrating the high degree of similarity between the COI sequences of S. bridgesi and S. triticea.
The phylogenetic tree constructed from the 16S rRNA sequences showed that different specimens could be clearly distinguished at the species level using 16s rRNA (Fig. 4). A study by Schiaparelli et al. (2005) showed that the minimum and maximum interspecific divergence values (obtained using the Jukes-Cantor model) of the 16S rRNA distance matrix between ovulid species were 0.03 and 22.3%, respectively. In the present study, the smallest genetic distance among species other than S. triticea and S. bridgesi was 0.0325%, supporting the findings reported by Schiaparelli et al. The greatest genetic distance between S. triticea and S. bridgesi was 0.0064%, which was only approximately one-fifth of the smallest genetic distance among other specimens. Therefore, the 16S rRNA data further support the synonymy between S. triticea and S. bridgesi.
Being a non-transcribed spacer region, the ITS region is subject to smaller selective pressures and generally undergoes rapid evolution (Odorico and Miller 1997). It is commonly used for analysis at the population and species level because of its high degree of sequence variation (Harris and Crandall 2000;Hillis and Dixon 1991). Therefore, ITS1-5.8S-ITS2 provides higher discriminating power at lower taxonomic levels. Among the ITS sequences obtained in the present study, the smallest genetic distance among specimens other than S. triticea and S. bridgesi was approximately 19 times that of the greatest genetic distance between S. triticea and S. bridgesi, representing a significantly larger intraspecific-interspecific genetic distance ratio compared with COI and 16S rRNA. This indicates that genetic differentiation did not occur even in the rapidly evolving ITS1-5.8S-ITS2 gene region between S. triticea and S. bridgesi. In the ITS phylogenetic tree (Fig. 5), S. triticea was convincingly clustered with S. bridgesi while the other clades were well supported.

Conclusion
In conclusion, the COI, 16S rRNA, and ITS1-5.8S-ITS2 data of the ovulid specimens collected in the present study indicated the absence of genetic differences between S. bridgesi and S. triticea. Both the phylogenetic trees and heatmaps ( Fig. 6-Fig. 8) showed a high degree of similarity between S. bridgesi and S. triticea, suggesting that the morphological differences between the two species may be caused by phenotypic plasticity than genetic differences, which is not unusual among molluscs. As the extremely high level of phenotypic plasticity in ovulid species results in much ambiguity in morphology-based classification criteria, the analysis of species evolution through molecular approaches is of great significance to the elucidation of evolutionary history (Schiaparelli et al. 2005). Based on the present knowledge, it is evident that striking phenomena of convergence and homoplasy characterise shell morphology in Ovulidae and that a molecular framework is necessary to recognise phylogenetically related groups.