Molecular assessment of Ulva (Ulvales, Chlorophyta) diversity in Vietnam including the new species U. vietnamensis

Species diversity of Ulva in Vietnam was investigated using three commonly used genetic markers, the nuclear encoded rDNA ITS region and the plastid encoded rbcL and tufA genes. Single locus species delimitation methods, complemented with morphological and ecological information resulted in the delimitation of 19 species. This diversity is largely incongruent with the traditional understanding of Ulva diversity in Vietnam. Only four species identified in this study, U. lactuca, U. reticulata, U. spinulosa, and U. flexuosa, have been previously reported, and seven species, U. ohnoi, U. tepida, U. chaugulii, U. kraftiorum, U. meridionalis, U. limnetica, and U. aragoënsis, are recorded for the first time from Vietnam. Seven genetic clusters could not be associated with species names with certainty. A new species, U. vietnamensis, is described from marine to brackish coastal areas from southern Vietnam based on its morphological and molecular distinctiveness from the currently known Ulva species. A comparison with recent molecular‐based studies of Ulva diversity showed that species composition in Vietnam is similar to that of adjacent countries, including Japan, China, as well as Australia. Our study emphasizes the importance of molecular data in the assessment of Ulva diversity, and indicates that a lot of diversity may still remain to be discovered, especially in tropical regions.


INTRODUCTION
Species of the green algal genus Ulva (Ulvales, Chlorophyta) are globally distributed in marine, brackish and freshwater environments where they often act as ecologically important primary producers (Nelson et al. 2003;Charlier et al. 2008;Ye et al. 2011). Several Ulva species have economic applications, including their use in integrated multi-trophic aquaculture (Neori et al. 2000) as fertilizers or biostimulants for plant growth (Dmytryk and Chojnacka 2018), ulvan extracts (Lahaye and Robic 2007;Alves et al. 2013), and alternative sources of food and biofuel for humans (Charrier et al. 2017). Under high nutrient conditions, the fast and persistent growth of Ulva species can cause problematic blooms, known as green tides (Nelson et al. 2003;Charlier et al. 2008;Ye et al. 2011). Despite their ecological and economic importance, species diversity in Ulva is incompletely characterized (Tran et al. 2022). Accurate species identification is fundamental in biodiversity and ecological studies, as well as in applied research. With more than 80 currently recognized species, and more than 400 names that are currently considered as synonyms (Guiry and Guiry 2021), Ulva is one of the most taxonomically challenging genera of green seaweeds. The few distinctive morphological features in combination with substantial morphological variability and plasticity make morphological species identification notoriously difficult (Kraft et al. 2010;Kiana et al. 2016;Xie et al. 2020). Recent taxonomic studies and investigations of species diversity in Ulva are therefore generally accompanied by molecular data. These studies have revealed a considerable conflict between traditional and phylogenetic species definitions (e.g., O'Kelly et al. 2010;Tonatiuh et al. 2019). Molecular studies of Ulva from subtropical and tropical regions are limited to Hawaii (O'Kelly et al. 2010), Mexico (Tonatiuh et al. 2019), India (Kazi et al. 2016), Iran (Kiana et al. 2016), the Eastern Mediterranean Sea (Krupnik et al. 2018), Japan (Ogawa et al. 2013), and Australia (Kraft et al. 2010).
Vietnam is located in Southeast Asia, bordering the Gulf of Tonkin, the South China Sea, and the Gulf of Thailand, and harbors a high diversity of macroalgae (Nguyen et al. 2013). The first taxonomic studies of Ulva from Vietnam dated back to Pham (1969) and were further completed by various studies (Nguyen 2007;Nhu Hau et al. 2012;Eduard et al. 2015). Nguyen et al. (2013) compiled these publications, and listed six foliose and 10 tubular Ulva species from Vietnam (Table S1). Most of these species, identified based on morphology, are taxa that were originally described from Europe, which casts doubt on their correct identities (O'Kelly et al. 2010).
In this study, we investigate Ulva diversity in Vietnam based on DNA sequence data of three commonly used markers, the nuclear encoded rDNA internal transcribed spacer region (ITS), the plastid encoded large subunit ribulose 1,5-bisphosphate carboxylase gene (rbcL) and the elongation factor Tu (tufA), and by applying different methods of sequence-based species delimitation, namely Automatic Barcode Gap Discovery (ABGD), Generalized mixed Yulecoalescent (GMYC), and Poisson tree processes (PTP). Molecular results were combined with morphological and ecological data to refine species boundaries. The estimated species diversity in Vietnam was compared with other regions.

Sample collection
Our sampling focused on brackish and coastal intertidal environments in both natural habitats and aquaculture settings along the coast of Vietnam between 2017 and 2019 ( Fig. 1). There were 176 Ulva specimens collected and analyzed. Habitat details of each specimen such as substrate and salinity, measured at a single time point by a refractometer, were recorded where possible (Table S2). For each specimen, a small piece was washed several times in clean seawater to remove epiphytes, blotted dry and preserved in silica gel for DNA extraction while the remaining part was mounted and dried on herbarium sheets and subsequently deposited at the Tropical Institute of Biology (VMN, Ho Chi Minh City, Vietnam) with duplicates in the herbarium of Meise Botanic Garden (BR, www.botanicalcollections.be) (Table S2).

Molecular data acquisition
Total genomic DNA was extracted following a modified CTAB method (Doyle and Doyle 1987). Three genetic markers were amplified using primers and PCR conditions described in Shimada et al. (2003) for ITS and Saunders and Kucera (2010) for rbcL and tufA. Because the amplification of ITS was often problematic compared to tufA and rbcL due to the presence of multiple bands, we were only able to determine 97 ITS sequences compared to 150 tufA and 148 rbcL sequences. All sequences were deposited in GenBank and accession numbers are listed in Table S2.
In addition to the sequences generated in this study, we included additional sequences downloaded from GenBank on 17 May 2020, summing up to 1346 tufA, 1560 rbcL, and 1441 ITS sequences of Ulva. Unique haplotypes were identified using CD-hit web server (Huang et al. 2010) and subsequently aligned using the MAFFT online web server (Katoh and Standley 2013).
For each genetic marker, two sets of alignments were created, one containing Ulva sequences from Vietnam only, and the other supplemented with GenBank sequences. The first set includes 52, 51, and 43 unique haplotypes for tufA (707 bp), rbcL (710 bp), and ITS (737 bp), respectively. The second set consists of 240, 539, and 533 unique haplotypes for tufA (601 bp), rbcL (1019 bp), and ITS (723 bp), respectively.

Species delimitation and identification
Three DNA-based species delimitation methods, ABGD, GMYC, and PTP, were used, which are reproducible and can reduce investigator-driven bias (Fujita et al. 2012;Leliaert et al. 2014). The GMYC method aims to detect switches between Yule speciation to coalescent process in an ultrametric phylogenetic tree (Fujisawa and Barraclough 2013). ABGD relies on the detection of a barcoding gap between intraspecific and interspecific divergence based on a distance matrix (Puillandre et al. 2012). PTP estimates species boundaries by modelling the speciation rate directly from the number of substitutions in a phylogenetic tree (Zhang et al. 2013). The three methods were applied to the first set of alignments containing only Vietnamese sequences, and the consistencies between different markers and methods were examined. ABGD was performed with relative gap width = 0.5 for rbcL and tufA, and relative gap width = 1.5 for ITS on the online platform (https:// bioinfo.mnhn.fr/abi/public/abgd/abgdweb.html) (Puillandre et al. 2012). All phylogenetic trees were reconstructed on the CIPRES Science Gateway (Miller et al. 2010). Maximum likelihood (ML) trees for the PTP analyses were constructed using IQ-TREE v.1.6.10 (Nguyen et al. 2015). Best-fit substitution model and rate heterogeneity were selected by ModelFinder (Kalyaanamoorthy et al. 2017) based on the BIC criterion integrated with IQ-TREE. Support values were estimated by ultrafast boot-strapping (1000 replicates). Ultrametric phylogenetic trees for the GMYC analyses were constructed using BEAST v.1.10 ) under a GTR + G + I substitution model, uncorrelated relaxed molecular clock, coalescent constant population size and other default priors. Each analysis was run for 10 7 generations with trees sampled every 10 4 generations. Tracer v.1.7 ) was used to inspect the convergence and acceptable effective sample size values >200 of each run. Maximum clade credibility trees were annotated with TreeAnnotator v.1.10.1 (Drummond and Rambaut 2007) using 75% remaining trees after burn-in. PTP analyses used the online web server (https://species.h-its.org/) (Zhang et al. 2013). Single threshold GMYC analyses (Fujisawa and Barraclough 2013) were executed using the Splits package (Ezard et al. 2009) in R v.3.6.0 (R Core Team 2018).
Operational taxonomic units (OTUs) of Ulva from Vietnam were identified by the majority rule, that is if clades were supported by the majority of the species delimitation methods implemented on the Vietnamese datasets (3 markers Â 3 methods) and if these clades were well-supported (bootstrap values >90 or Bayesian posterior probabilities >0.8) in the majority of phylogenetic analyses (3 markers Â 2 methods).
To aid species identification, that is, assigning species names to the identified OTUs, the same species delimitation procedures were applied on the second set of alignments comprising the Vietnamese and GenBank sequences with adjustments. Particularly, for rbcL and ITS, these analyses were limited to PTP and GMYC due to the non-overlapping GenBank sequences in rbcL and ITS alignments. Subsequently, the species identities of sequences that clusters with Vietnamese sequences were manually examined and curated accordingly to the most recent updates in Ulva taxonomy.
Phylogenetic tree based on a concatenated alignment A representative of each Vietnamese OTU and the most closely related species, identified in the analyses based on the second sets of alignment, were included in the third alignment of concatenated sequences (tufA + rbcL + ITS, Table S3). The inclusion criteria were based on the availability and the validity of species names, prioritized as follows: sequences from type specimens, sequences from publications, and GenBank sequences. The concatenated alignment was filled 76% at the gene Â specimen level. TufA sequences were least represented due to their limited availability on GenBank (Table S3). The concatenated alignment included 26 tufA (601bp), 40 rbcL (540 bp), and 34 ITS (577 bp) Ulva sequences. Ulvaria obscura (Kützing) Gayral ex Bliding (specimen voucher GWS003572), Umbraulva kaloakulau  For the ML and Bayesian inference (BI) phylogenetic analyses, the best-fit partitioning schemes and models of molecular evolution selected using the AIC criterion in PartitionFinder v.2.1.1 (Lanfear et al. 2017) were GTR + I +-G for each individual codon position of tufA and rbcL (6 partitions) and TRN + I + G for ITS (1 partition).

Morphological observations
Freshly collected specimens and herbarium specimens were examined, and the following morphological features were recorded: external morphology, including size and shape of the thallus, branches and holdfast, cell size, cell shape, arrangement of cells in different parts of the thallus, numbers of pyrenoids per cell, and chloroplast positions in the cell. Microphotographs were taken using an Olympus SC50 camera mounted on an Olympus BX41 microscope.

DNA-based species delimitation
Details of the alignments of Vietnamese sequences, and results of the species delimitation analyses based on these datasets are summarized in Table 1. The three species delimitation methods (ABGD, GMYC, PTP) applied to the three markers (tufA, ITS, rbcL) yielded different estimates of species diversity. Species delimitation based on the ITS dataset provided the highest estimates of species diversity, followed by tufA and rbcL. A pairwise distance matrix calculated by ABGD (Fig. S1) indicated that ITS has the highest sequence divergence, followed by tufA and rbcL.
The identification of OTUs based on the majority rule and clade support was generally straightforward except in the case of deciding whether to merge (supported by tufA) or split (supported by ITS) OTUs 8 and 9 (Fig. 2). Analysis of rbcL did not provide additional support for either scenario. In the assumption that ITS has a higher resolution in delimiting Ulva species, we opted to recognize two OTUs. In total, 18 Ulva OTUs in Vietnam were recognized.

Concatenated phylogeny and species identities
Species identification of the Vietnamese OTUs and identification of closely related species were based on the species delimitation analyses on the second set of alignments. Details of the species delimitation analyses are available on https:// zenodo.org/record/6821457. In general, we found considerably fewer GenBank tufA sequences falling in the same species cluster of Vietnamese OTUs compared to rbcL and ITS. Closely related sequences of Vietnamese OTUs were included in the concatenated phylogenetic analyses (Table S3).
The ML and BI trees based on the concatenated alignment had similar topologies (Fig. 3) Shimada et al. 2008) with low support (Fig. 3).
Among the 15 OTUs with a tubular morphology, five were identified as species that have been recently described based on DNA sequence data (Fig. 3, Table S4). OTUs 5, 11 and 13 formed well-supported clades with sequences of U. limnetica K. Ichihara     U. fenestrata Postels & Ruprecht) (specimen voucher LK-043, Kraft et al. 2010) and U. aragoënsis (Bliding) Maggs (specimen voucher HER_2_TC, Krupnik et al. 2018). The Australian specimen identified as U. stipitata var. linzoides, whose name is currently applied in Vietnam, and U. mediterranea Alongi, Cormaci & G. Furnari (now U. aragoënesis) were recognized as one species by Hiraoka et al. (2017), which is also supported by the current species delimitation analyses based on ITS and rbcL markers. OTU 7 clustered with U. torta (Mertens) Trevisan (specimen voucher GWS004454, Saunders and Kucera 2010) and U. clathratioides L. G. Kraft, Kraft & R. F. Waller (specimen voucher LK-009), of which U. torta has been reported from Vietnam. Although the conspecificity of these two species has been questioned (Kirkendale et al. 2013), we retained both names for OTU 7, pending further taxonomic research. OTU 9 clustered with Japanese specimen named U. simplex  (Table S4). OTUs 14 and 15 form a sister clade to U. meridionalis, and OTU 12 is related to U. tepida.

Morphological characterization
Morphological descriptions and photographs of the Ulva species from Vietnam are presented in Table S5 and Figs S2-S19 in the Supporting Information. OTU 4 contains morphotypes of U. reticulata (Fig. S10) and U. spinulosa Okamura & Segawa (Fig. S12). Previous studies have shown that the two species are indistinguishable by ITS and rbcL markers, but they express distinct morphologies and reproduction modes (Hiraoka et al. 2003a(Hiraoka et al. , 2003b. Therefore, we further split OTU 4 and recognized the two species, U. reticulata and U. spinulosa in Vietnam. Thus, in total, we recognize 19 species of Ulva in Vietnam based on our current sampling, and applying DNA-based species delimitation approaches, complemented with morphological data.
Below, we describe a new species of Ulva from brackish and marine environments in southern Vietnam.

Species delimitation
Among the three molecular markers used for species delimitation, tufA and ITS were found to be more effective compared to rbcL. The number of species estimated by the three methods based on the tufA and ITS datasets was more consistent compared to the analyses based on the rbcL data. These results vouch in favor of the use of tufA and ITS as markers in molecular-based species delimitation of Ulva species, although in our study, the amplification success of the ITS marker was lower than for the other two markers.
The disagreements in species delimitations between the markers partly result from differences in genetic variation. Although tufA has been favored for species delimitation of Ulva because it is more easily amplifiable than ITS and more variable than rbcL (Saunders and Kucera 2010), in our study, ITS showed the highest sequence divergence. In addition, all three markers failed to recognize the two species, U. reticulata and U. spinulosa which have been found to be reproductively isolated (Hiraoka et al. 2003a(Hiraoka et al. , 2003b, raising the question about the power of these markers to delimit the two closely related species. A higher resolution marker, such as 5 S rRNA that is frequently used to distinguish (sub)species in the U. prolifera O. F. Müller-U. linza Linnaeus species complex (Shimada et al. 2008;Cui et al. 2018), could also provide better insights about species boundaries in other clades of Ulva.
Presently, the rbcL public sequence dataset is the most comprehensive and in our study returned the highest numbers of matches with species names compared to ITS and tufA. Since ITS and tufA come forward as better candidate markers for species delimitation in Ulva, we recommend that future studies systematically sequence ITS and tufA in addition to rbcL to increment the sequence databases for these two markers.

Molecular data compared to traditional taxonomy
This is the first study assessing the species diversity of Ulva in Vietnam based on molecular data. The identification of 19 species (four foliose and 15 tubular species) stands in contrast to the 16 species (six foliose and 10 tubular species) previously recorded based on morphological criteria (Nguyen et al. 2013). Among the six foliose species, three (U. lactuca, U. reticulata, and U. spinulosa) were confirmed by molecular data complemented with morphological data. U. conglobata Kjellman, U. stipitata Areschoug (synonym of U. fenestrata), and U. papenfussii Pham-Hoang Hô, described from Hon Thu island and recorded in Ninh Thuan, Vietnam (Nguyen 2007), were not collected in this study. However, a foliose strain widespread across Vietnam was identified as U. ohnoi, a green tide forming species described from Japan (Hiraoka et al. 2003b).
Of the 10 tubular species previously identified in Vietnam, only two species, U. flexuosa and U. torta, were found in this study. However, there is an uncertainty in the name usage of U. torta. Our study confirms the analyses of Kirkendale et al. (2013), indicating that the molecular concept, that is, a group of samples that share similar DNA information, of U. torta is associated with U. clathratioides. Among the other 13 tubular Ulva species identified in this study and not previously recorded from Vietnam, five (U. tepida, U. chaugulii, U. limnetica, U. kraftiorum and U. meridionalis) were recently described from the Asia-Pacific region based on molecular and morphological data. One species, U. aragoënsis, was originally described from Europe (Bliding 1963) but has since been recorded from other regions, including the Asia-Pacific region. One species was identified as U. simplex (sensu Ogawa et al.). This Japanese specimen was recognized as U. simplex by the study of Ogawa et al. (2013). However, Hiraoka et al. (2017) pointed out dissimilarities in morphology between Japanese and European specimens of U. simplex, casting doubt on the identity of the Japanese specimens and subsequently referring to them as U. simplex (sensu Ogawa et al.). The six remaining species VN sp1 to VN sp6, could not be reliably associated to species names based on molecular or morphological data.
In contrast to the notion that U. compressa, U. intestinalis and U. prolifera were the most common species in Vietnam based on morphology-based identifications (Nguyen 2007), we did not find these species in our study. This may indicate that these species are absent in Vietnam, or that they are rare or have limited distributions. The latter may be more plausible, given that their presence has been molecularly confirmed in countries adjacent to Vietnam, including China, South Korea and Japan (Cui et al. 2018;Duan et al. 2012;Shimada et al. 2008).

Comparison of Ulva diversity with other regions
Localities of Ulva species occurred in Vietnam were extracted from literature and GenBank records from the second species delimitation analysis (available on https://zenodo.org/record/ 6821457). It indicates that Ulva species composition in Vietnam is largely shared with Asia-Pacific countries, especially Japan, China, and Australia. The majority of marine Ulva species in Vietnam turn out to be cosmopolitan species, of which U. ohnoi and U. tepida are the most common species found in Vietnam (Fig. S20a, b). Other species, U. spinulosa and Ulva VN sp6, were recorded from the Pacific basin (e.g., Japan (Shimada et al. 2003), South Korea (Kang et al. 2019), Hawaii (O'Kelly et al. 2010 and China (MF139302)). Ulva kraftiorum was, up to now, only known from Western Australia (Kraft et al. 2010).
Among marine and brackish species (Fig. S20c, d), U. aragoënsis is also found in the Mediterranean Sea, the Philippines (Guiry and Guiry 2021), and Australia (Kraft et al. 2010). Ulva chaugulii M. G. Kavale & M. A. Kazi, is currently known from the Middle East (Kazi et al. 2016;Kiana et al. 2016), China (Xie et al. 2020) and Eastern Mediterranean Sea (Krupnik et al. 2018). A common species along the Vietnamese coastline, U. meridionalis, is broadly distributed in China (Xie et al. 2020), Japan (Shimada et al. 2008;Horimoto et al. 2011), and the east and gulf coast of the United States (Melton III and Lopez-Bautista 2021).
Brackish species seem to have more confined geographical ranges (Fig. S20e, f). Only U. flexuosa is found in various localities in the world. Ulva simplex sensu Ogawa et al., is present in a few locations in Japan (Ogawa et al. 2013), China (Xie et al. 2020), and the United States (Hanyuda et al. 2018). Our study is the first report of U. limnetica, a freshwater species with restricted distribution in freshwater or low brackish environments from Japan, outside its currently known range. Yet it appears to be widespread in sheltered estuaries in Southern Vietnam in a wide salinity range from 1-24 ppt (Table S2).
Based on the unique diversity of Ulva found in Hawaii, O'Kelly et al. (2010) suggested that Ulva populations in tropical and subtropical regions consisted of species that are largely unique to these areas, and that the application of names based on types from temperate and boreal European and North American waters is inappropriate. Our study partly confirms this assumption. We found several species that may be endemic to Vietnam. However, more than half of the Ulva species recognized in Vietnam are found in other regions, including temperate regions in East Asia, Australia, and the United States. Ulva diversity patterns, especially in the tropics, have not fully been identified yet. New Ulva species, which we could not match with existing names, were discovered in our study. Further exploration of Ulva diversity, especially in the tropic, is thus necessary to gain better insights into the global patterns of Ulva diversity.

CONCLUSIONS
We provide new insights into the diversity of Ulva in Vietnam. The markers tufA and ITS proved to be more effective in delimiting Ulva species than rbcL. Nineteen species were recognized in this study in contrast with 16 species in previous studies based on morphological data. Ulva species in Vietnam identified based on DNA data are largely shared with its adjacent countries. This study provides a baseline to facilitate future studies on diversity or applied research on Ulva in Vietnam. special reference to an unusual morphology of U. meridionalis forming green tides. Eur. J. Phycol. 55: 412-25. Ye, N.-H., Zhang, X.-W., Mao, Y.-Z. et al. 2011. "Green tides" are overwhelming the coastline of our blue planet: taking the world's largest example. Ecol. Res. 26: 477-85. Zhang, J., Kapli, P., Pavlidis, P. and Stamatakis, A. 2013. A general species delimitation method with applications to phylogenetic placements. Bioinformatics 29: 2869-76.

SUPPORTING INFORMATION
Additional Supporting Information may be found in the online version of this article at the publisher's web-site: Fig S1. Pairwise distance distribution of the ITS, rbcL, and tufA data sets calculated by ABGD method.
Figs S2-S19. Morphological descriptions and illustrations of the Vietnamese Ulva species identified in this study.  Table S1. List of Ulva species that are currently known in Vietnam adopted from Nguyen et al. (2013) with their type localities, molecular concepts (of type specimens) and reference papers. Nomenclature/taxonomic revisions are shown when applicable. Table S2. Details of Ulva specimens collected in this study including species names, collection codes, alternative codes (if present), specimen vouchers, localities, sampling coordinates, ecology, collection date, GenBank accession numbers of tufA, rbcL, and ITS markers. Table S3. Sequences matrix (tufA, rbcL, and ITS) of Ulva specimens with their corresponding GenBank accession numbers in the concatenated phylogeny.