Morphological and molecular study on Yininemertespratensis (Nemertea, Pilidiophora, Heteronemertea) from the Han River Estuary, South Korea, and its phylogenetic position within the family Lineidae

Abstract Outbreaks of ribbon worms observed in 2013, 2015, and 2017–2019 in the Han River Estuary, South Korea, have caused damage to local glass-eel fisheries. The Han River ribbon worms have been identified as Yininemertespratensis (Sun & Lu, 1998) based on not only morphological characteristics compared with the holotype and paratype specimens, but also DNA sequence comparison with topotypes freshly collected near the Yangtze River mouth, China. Using sequences of six gene markers (18S rRNA, 28S rRNA, histone H3, histone H4, 16S rRNA, and COI), the phylogenetic position of Y.pratensis was inferred among other heteronemerteans based on their sequences obtained from public databases. This analysis firmly placed Y.pratensis as a close relative to Apatronemertesalbimaculosa Wilfert & Gibson, 1974, which has been reported from aquarium tanks containing tropical freshwater plants in various parts of the world as well as a wild environment in Panama.


Introduction
An explosive proliferation of unidentified, brackish-water heteronemerteans was observed in the Han River Estuary, South Korea, in the spring of 2013. Our morphological observation of the Han River ribbon worms indicated that they represent Yininemertes pratensis , a brackish-water heteronemertean known only by its original description from the Yangtze (Changjiang) River Estuary, China (for the nomenclature of the genus, see Sun and Lu 2008;Özdikmen 2009;Kajihara 2014). Outbreaks of Y. pratensis in the Han River Estuary were also observed in 2015, 2017, 2018, and 2019. Reportedly, the worms have caused severe damage (Lee 2015;Noh 2019) to local fisheries of glass eels, which are juveniles of Anguilla japonica Temminck & Schlegel, 1847, a valuable fishery resource in East Asian countries showing dramatic declines in recent years (Tzeng 1997;Tatsukawa 2003;Tseng et al. 2003). As causes for the eel declines, overfishing and habitat loss due to human activities (e.g., Chen et al. 2014) and oceanic-atmospheric factors such as changes in ocean circulation (Chang et al. 2018) have been suggested. To what extent the nemerteans have been contributing to the anguillid declines is not known. For glass-eel fisheries, fishermen set long, conical nets on the estuarine bottoms with apertures directing downstream. At the end of each net, ascending catches are to be concentrated mostly during flood tide. In the 2015 bloom, more than 90% of catches were worms, with none to only a few eels that were dead (Lee 2015) probably due to yet-unidentified neurotoxic substances (Kwon et al. 2017) in worm mucus within the concentrated net catches. These neurotoxins might have been discharged from epidermal cells and contained in the secreted mucus (cf. Tanu et al. 2004;Asakawa et al. 2013). To our knowledge, this is the first record of damage to fisheries directly caused by nemertean outbreaks, although a potentially indirect case is known. At certain Alaskan localities in the 1983-1984 and 1984-1985 brooding seasons of the red king crab Paralithodes camtschaticus (Tilesius, 1815), a widespread outbreak of the decapod-egg-predatory nemertean Carcinonemertes regicides Shields et al., 1989, and possibly Ovicides paralithodis Kajihara & Kuris, 2013 as well, caused high egg mortality (Kuris et al. 1991), which could have led to a subsequent decline in the red king crab population (e.g., Loher and Armstrong 2005). In addition, the milky ribbon worm Cerebratulus lacteus (Leidy, 1851) has been identified as an important threat to populations of the softshell clam Mya arenaria Linnaeus, 1758, which is one of the commercial bivalves in Atlantic Canada, although no outbreak has ever been reported for C. lacteus (cf. Bourque et al. 2001Bourque et al. , 2002. Facing a plethora of undescribed species with dwindling number of experts, some nemertean taxonomists agreed that taxonomic descriptions of ribbon worms will have to shift from traditional, internal-anatomy-based style to histology-free one with a combination of high-quality external images and molecular phylogeny (Strand and Sundberg 2011;Strand et al. 2014;Kajihara 2015;Sundberg et al. 2016). On the other hand, in the case of Heteronemertea, only about 10% of ~100 genera (Gibson 1995;Kajihara et al. 2008) have been represented by type species in terms of sequences for multi-locus analysis (Thollesson and Norenburg 2003;Andrade et al. 2012;Kvist et al. 2014Kvist et al. , 2015. Logically, until the rest of ~90 genera are also represented in the same manner, examination of internal morphology will remain indispensable to genus-level identification (e.g., Chernyshev et al. 2018). Moreover, most heteronemertean genera currently diagnosed are non-monophyletic. This has been repeatedly pointed out in previous studies (e.g., Sundberg and Saur 1998;Schwartz 2009;Puerta et al. 2010;Hiebert and Maslakova 2015). Therefore, as many type species of genus-group names (such as Yininemertes) as possible should be placed in molecular phylogenetic context for proper application of genus names in many other species of heteronemerteans as long as Linnaean binominal nomenclature is employed.
In this paper, we report the identity of Han River nemerteans based on morphological characteristics in comparison to the type material of Y. pratensis as well as DNA barcoding data from the type locality. Also, we infer the phylogenetic position of Y. pratensis among Heteronemertea based on a multi-locus molecular analysis.

Specimen collection and processing
Approximately 700 individuals of ribbon worms were collected from local fishermen's glass-eel nets for Anguilla japonica, set at about 37°36'08"N, 126°48'23"E, in Goyang, South Korea, approximately 40 km upstream of the mouth of the Han River (  Island, 31°34'39.4"N, 121°54'34.9"E, on May 14, 2016 (Figs 1A, C, 2C). Specimens from the Han River were anesthetized with 7% MgCl 2 solution before fixed in either 7% neutral-buffered formalin for morphological observation (~300 individuals) or 100% ethanol for DNA extraction (~300 individuals). Of these 12 specimens collected from Shanghai, nine were fixed in 70% EtOH for DNA extraction while three were used for taking photographs. Anterior portion of one formalin-fixed specimen from the Han River was dehydrated in ethanol series, cleared in xylene, embedded in paraffin (melting point: 56-57 °C), and transversely sectioned at thickness of 8 µm. Serial sections were stained with Mallory's trichrome method (Gibson 1994). Specimens were deposited in National Institute of Biological Resources Invertebrate Collection, Incheon, Korea (NIBR IV) and Invertebrate Collection of the Hokkaido University Museum, Sapporo, Japan (ICHUM) ( Table 1). For comparison, the holotype (DH005A) and a paratype (DH005C) of Y. pratensis deposited in Ocean University of China, Qingdao, People's Republic of China, were also examined.

Molecular phylogeny
Small pieces of tissue taken from 22 specimens collected from the Han River and seven specimens from Yangtze River were used for total genomic DNA extraction using DNeasy Blood and Tissue Kit (Qiagen, Hilden, Germany) following the manufacturer's instructions. Partial sequences of six gene markers (nuclear 18S rRNA, 28S rRNA, histone H3, and histone H4; mitochondrial 16S rRNA, and COI) were used for molecular analyses using the same primers published by Andrade et al. (2012). For PCR amplification, the following mixture was prepared in a total volume of 50 µL: 50 ng of template genomic DNA, 2.5 mM dNTPs, 5 µL of 10× Ex Taq™ buffer, 2 µL of each 10 pM primer, and 1 U (0.5 µL) of TaKaRa Ex Taq™ polymerase. Thermal cycling condition comprised an initial denaturation at 94 °C for 30 sec followed by 35 cycles of denaturation at 98 °C for 10 sec, annealing at 43-50 °C depending on primers for 30 sec, and extension at 72 °C for 1 min. A final extension step at 72 °C for 10 min was then followed. Amplified PCR products were sequenced using an ABI 3730 sequencer (Applied Biosystems, Foster City, CA, USA) from both directions. All sequences generated de novo in this study were deposited at GenBank (Table 2).
To assess phylogenetic affinity of the Han River nemerteans, maximum likelihood (ML) analysis and Bayesian Inference (BI) were carried out with 31 lineid heteronemertean species for which the aforementioned six gene sequences were available in public databases (Table 3). Outgroups were chosen to include Baseodiscus mexicanus (Bürger, 1893) and B. unicolor Stiasny-Wijnhoff, 1925(cf. Andrade et al. 2012Kvist et al. 2014). Sequence alignment was performed using MAFFT ver. 7 (Katoh and Standley 2013) with E-INS-i option for 18S, 28S, and 16S. For the protein-coding H3, H4, and COI, sequences were aligned straightforward without gaps. Sequences were edited and concatenated using MEGA ver. 5.2 (Tamura et al. 2011). Gaps and incompletely determined nucleotides accounted for 24.9% of the entire dataset of these sequences. Table 1. List of specimens identified as Yininemertes pratensis  in this study with catalogue numbers at the National Institute of Biological Resources Invertebrate Section, Incheon, Korea (NIBR IV) and the Invertebrate Collection of the Hokkaido University Museum, Sapporo, Japan (ICHUM) as well as their sampling date and locality.   PartitionFinder ver. 1.1 (Lanfear et al. 2012) was used to determine the best partition scheme for ML and BI. For BI, the most suitable substitution model for each partition was also selected: GTR+I+G for 16S and 28S; GTR+G for COI (1 st codon), H3 (1 st and 3 rd codons), and H4 (1 st and 2 nd codons); K80+I+G for 18S and H4 (3 rd codon); F81+I+G for COI (2 nd codon); HKY+I+G for COI (3 rd codon); and JC for H3 (2 nd codon). ML analysis was performed using RAxML ver. 8.0.0 (Stamatakis 2014) with a GTR+G model of nucleotide substitution for all partitions consisting of 1000 rapid bootstraps. BI was carried out using MrBayes ver. 3.2.3 (Ronquist and Huelsenbeck 2003;Altekar et al. 2004) with two independent Metropolis-coupled analyses (four Markov chains of 10,000,000 generations for each analysis). Trees were sampled every 100 generations. Values of run convergence indicated that sufficient amounts of trees and parameters were sampled (average standard deviation of split frequencies = 0.006616; minimum estimated sample size of tree lengths = 706.26; potential scale reduction factor of tree lengths = 1.001). Run convergence was also assessed with Tracer ver. 1.6 (Rambaut et al. 2014).

Morphology
The external feature of the Han River nemerteans agreed with the original description of Y. pratensis in that these worms were variously dark brown, brick red, and tinged with violet sometimes (Fig. 2A, B). Generally, their body color became paler posteriorly.  have reported that specimens from the Yangtze River Estuary sometimes show light-red, 4-10 transverse rings arranged on the body. Such ring arrangement was also found in specimens from the Han River Estuary (Fig. 2B) as well as topotype specimens (Fig. 2E, G) collected from muddy sediment with or without vegetation (Fig. 2C, D).
In specimens collected from the Han River, the proboscis was not branched, and reddish in color (Fig. 3A). Serially sectioned specimen (ICHUM 5260) showed  the following internal anatomical features: i) the proboscis had two muscle crosses (Fig. 3B), similar to that in the paratype of Y. pratensis (Fig. 3C); ii) the rhynchocoel outer circular musculature was not interwoven with the adjacent body-wall longitudinal musculature; iii) the nervous system had type-3 neurons (cf. Beckers 2015) along the inner portion of the brain (Fig. 3D); iv) the foregut wall had intraepithelial somatic muscle fibres that appeared to be circular or diagonal (Fig. 3E), similar to that observed in the holotype (Fig. 3F); v) the body-wall dermal glandular layer was not separated from the body-wall outer longitudinal muscle layer by connective tissue layer (Fig. 3E); and vi) the blood system comprised spacious cephalic lacuna (Fig. 3G, H), an alimentary vascular plexus (Fig. 3E), and a middorsal blood vessel.

Molecular phylogeny
Lengths of the six gene markers determined for Korean and Chinese materials were: 16S, 507-508 bp; 18S, 1000-1003 bp; 28S, 1132 bp; COI, 658 bp; H3, 331 bp; and H4, 160 bp. Resulting ML tree (ln L = −51290.378661) and BI tree (harmonic mean of estimated marginal likelihood for two runs = −52096.68) were topologically more or less the same, with Y. pratensis being a sister of Apatronemertes albimaculosa Wilfert & Gibson, 1974 in both trees with 100% bootstrap support value and 1.0 posterior probability (Fig. 4). The inter-specific K2P distance between the COI sequences of Y. pratensis and A. albimaculosa was 0.163-0.196. More basal relations between this clade (= Y. pratensis + A. albimaculosa) and other heteronemerteans included in this analysis were poorly resolved.

Population genetics
Median-joining and statistical parsimony networks were identical in shape, comprising eight haplotypes with a maximal difference of five mutations (Fig. 5). From 29 specimens analysed (22 from Korea, seven from China), a total of nine haplotypes were detected, of which two were shared by Korean and Chinese populations. Eleven of 22 sequences from Korea were represented by the same haplotype, which was also the main haplotype among the Chinese population (shared by five of seven Chinese individuals analysed). Eight COI haplotypes from Korea differed by 0.000-0.006 from each other in terms of both uncorrected p-distance and K2P. The Korean population showed higher values of nucleotide diversity and haplotype diversity than the Chinese ones (Table 4). Tajima's D and Fu's Fs values were all negative for the Korean population, the Chinese population, and the total population, although not significantly different from zero except for the Fu's Fs values for the Korean population and total population.  . Numbers near nodes are bootstrap values for maximum-likelihood analysis and posterior probability for Bayesian inference. Scale bar indicates the number of substitutions per site.    in the Han River and Yangtze River Estuaries.  (Rathke, 1799) and L. pseudolacteus (Gontcharoff, 1951) Coe, 1905(Zattara et al. 2019, Lineus pictifrons Coe, 1904(Coe 1932, and L. rubescens Coe, 1904(Coe 1930, have been documented. Asexual reproductive capacity may have evolved in more lineages than previously thought among heteronemerteans, possibly including Y. pratensis. Another hypothesis is that the Han River ribbon worms might have been introduced from other, unidentified localities. However, this hypothesis sounds rather unlikely, because the haplotype diversity in the Korean population (0.7316), which was greater than the Chinese one (0.5238), suggests that a stable population have existed in the Han River Estuary, probably since long before the first bloom observed in 2013. While Tajima's D and Fu's Fs values were overall negative, we cannot draw any robust conclusion about the population dynamics because most of the values were statistically not significant. Future study is needed to pinpoint possible environmental factors that are responsible for the Y. pratensis outbreaks, as well as to elucidate the species' basic biology for obtaining countermeasures against the economic loss to local glass-eel fisheries caused by such blooms. The family Lineidae McIntosh, 1874 currently contains about 90 genera and 370 species of heteronemerteans, which are morphologically characterized by having horizontal lateral cephalic slits and three apical organs. Most are marine, but six species (each in a monotypic genus) have been described from freshwater or brackish-water habitat. These are Planolineus exsul Beauchamp, 1928 from Indonesia; Siolineus turbidus Du Bois-Reymond Marcus, 1948 from Amazon; Hinumanemertes kikuchii Iwata, 1970 from Japan; A. albimaculosa from freshwater tanks in Germany (Wilfert and Gibson 1974), Austria (Senz 1993), USA (Smith 2001), and Japan , as well as in submerged logs and rocks in a pond in Panama (Kvist et al. 2018); Amniclineus zhujiangensis Gibson & Qi, 1991 from Zhujiang, China;and Y. pratensis from China and Korea present study). Our phylogenetic tree indicates that A. albimaculosa and Y. pratensis form a highly supported clade, suggesting that the remaining fresh-and brackish-water forms, especially those in Southeast and East Asia, may also belong to the same clade. At this moment, however, neither morphological nor molecular synapomorphy between A. albimaculosa and Y. pratensis can be perceived; for instance, the characteristic outer cephalic vessels in A. albimaculosa are not found in Y. pratensis. Both species are reddish in body color, but this may be due to convergent evolution, as freshwater monostiliferous hoplonemerteans in the genus Prostoma Dugès, 1828 also possess reddish body. Future studies with expanded taxon sampling, along with detailed morphological examination, should clarify the evolution of these freshwater heteronemerteans.