Seahorses of the Hippocampus coronatus complex: taxonomic revision, and description of Hippocampus haema, a new species from Korea and Japan (Teleostei, Syngnathidae)

Abstract Morphological and molecular analyses were conducted on 182 specimens belonging to the Hippocampus coronatus complex (H. coronatus sensu lato), collected in Korea and Japan 1933–2015, in order to clarify the taxonomic status of the species within this complex. Three species are recognized based on the shape of the coronet, the number of trunk rings (TrR) and tail rings (TaR), and presence or absence of a wing-tip spine (WS) at the dorsal fin base. Hippocampus coronatus Temminck & Schlegel, 1850 (H. coronatus sensu stricto), is diagnosed by 10 TrR, 37–40 TaR, an extremely high coronet (55.7–79.0 % head length) with four tips on the corona flat (CoT), and one WS. Hippocampus sindonis Jordan & Snyder, 1901 is diagnosed by 10 TrR, 35–38 TaR, a moderately high coronet (36.3–55.4 % HL) with five CoT, and no WS. A new species, H. haema is described on the basis of 140 specimens, characterized by 10 TrR, 35–38 TaR, a moderately high coronet (34.1–54.9 % head length) with four CoT, and two WS. Hippocampus haema is only known from the Korea Strait, western Kyushu, and East/Japan Sea. Recognition of the three species is supported by differences in mitochondrial DNA fragments (cytochrome b, 16S rRNA, and 12S rRNA).

The species (or species group) H. coronatus sensu lato has been defined by possessing ten trunk rings, 34-40 tail rings, a bony armor, double gill openings (Lourie et al. 1999Kim et al. 2005;Kuiter 2009;Foster and Gomon 2010;Senou 2013;Lourie 2016), and a tall coronet on the head, which exhibits a wide range of height variation (Jordan and Snyder 1901;Mitani 1956;Lourie et al. 1999. Some authors have stated that this group includes two species, H. coronatus (sensu stricto), which has an extremely high coronet and a snout length ~2.33 times the head length, and H. sindonis, which has a moderately high coronet and a snout length ~3 times the head length (Jordan and Snyder 1901;Okada and Matsubara 1938;Matsubara 1955;Lourie et al. 1999;Senou 2002;Senou 2013), while others considered the variation in coronet height only as intraspecific variation (Mitani 1956;Araga 1984;Senou 1993). Based on variation in mitochondrial DNA (partial 12S rRNA), Mukai et al. (2000) suggested that the H. coronatus complex (H. coronatus sensu lato) consists of two genetically diverged groups.
Although the Korean seahorse (Korean name: Haema) has been identified as H. coronatus (Mori 1928;Chyung 1977;Kim et al. 2001;Kim et al. 2005), the height of its coronet and the number of tail rings appear to agree better with that described for H. sindonis (Jordan and Snyder 1901;Lourie et al. 1999Kim et al. 2013;Senou 2013;Han et al. 2014). In fact, H. sindonis has often been confused with H. coronatus (Lourie et al. 1999, and the height of the coronet in the type series of H. coronatus varies (Boeseman, 1947). These controversies have contributed to the uncertainty about the distribution of H. coronatus in both Korea and Japan, and led to its classification in the Data Deficient (DD) category of the International Union for Conservation of Nature and Natural Resources (IUCN) Red List, as there is a lack of information on population trends (Zhang and Pollom 2016). The present study aims to clarify the taxonomic status of Korean seahorses, redescribing H. coronatus and H. sindonis and describing a new species, all belonging to the H. coronatus complex.

Material examined
A total of 182 specimens of H. coronatus sensu lato collected from Korean and Japanese waters ( Fig. 1)

Morphological analysis
Procedures used for counts and measurements follow Lourie (2003) and are presented in Fig. 2 Meristic data were obtained from soft X-rays of the 182 H. coronatus sensu lato specimens. Measurements were obtained using the microscope-integrated Active Measure software (Shinhanoptics, Seoul, Korea). The coronet height was measured as CHMC (Lourie 2003) and CHGO (Temminck and Schlegel 1850;Jordan and Snyder 1901) (Fig. 2) so that our results could be compared to those reported in previous studies (Temminck and Schlegel 1850;Jordan and Snyder 1901;Lourie et al. 1999). Sexual dimorphism analysis was conducted on the 152 adults (80 females and 72 males). These are all the specimens over 53.9 mm, which is the minimum SL at maturation defined for H. coronatus sensu lato (Choi et al. 2006).

Molecular analysis
Tissue from the right eye ball or from the right-side of the tail was used to isolate genomic DNA from 22 specimens with moderately high coronets, collected in Busan, Tongyeong, Boseong, Soan Island, Maizuru, and Minami-ise, and from four specimens with extremely high coronets collected in Miura. Isolation was performed using an AccuPrep® Genomic DNA Extraction Kit (Bioneer, Daejeon, Korea), according to the manufacturer's instructions.
Etymology. The Latin word coronatus means crowned. The new Korean name, Wanggwan-haema means 'crowned seahorse', in agreement with the English and scientific names. In fact, Haema, which has the connotation 'common' and 'fish species belonging to the genus Hippocampus' in Korean, has been used to name seahorses commonly found in Korea, whereas Wanggwan-haema has been informally used to refer to H. coronatus in Korean. In addition, the word wanggwan [crown] is more suited for H. coronatus, whose coronet is considerably higher than that of H. haema. The Japanese name Tatsu-no-otoshigo literally means 'dragon's bastard child'.
Remarks. Temminck and Schlegel (1850) described H. coronatus based on five specimens. Boeseman (1947) designated one of these specimens RMNH.PISC.D 1543 as the lectotype. As a consequence the other three specimens RMNH.PISC.D 1541, RMNH.PISC.D 1542, and RMNH.PISC.D 1544 became paralectotypes, except that RMNH.PISC 3924 was reidentified as H. mohnikei (see remarks of H. sindonis below). However, two of the specimens described in Boeseman (1947), RMNH. PISC.D 1541 and 1542, have a moderately high coronet, not agreeing with the H. coronatus described in the present study and being more similar to H. haema (see species description below). The lectotype RMNH.PISC.D 1543 and the paralectotype RMNH.PISC.D 1544 have an extremely high coronet, which agrees with the present description of H. coronatus. Our 28 specimens have an extremely high coronet, a wingtip spine on the dorsal fin base, and CoT 4, as described and illustrated in Temminck and Schlegel (1850). The phylogenetic trees obtained in the present study also support the differentiation of these 28 specimens from H. sindonis and H. haema (Fig. 7).
The type series does not match Temminck and Schlegel (1850) The type locality of H. coronatus has not been established. Although it is thought to be Nagasaki (Eschmeyer et al. 2017), no specific locality information is provided for the type series or in previous studies (Temminck and Schlegel 1850;Boeseman 1947;Lourie et al. 1999). Seahorses are used historically as charm for safe-birth in East Asia (Korea, Japan, and China) and as a trinket in western culture (Lourie et al. 1999;Scales 2009). Thus, we cannot exclude the possibility that dried specimens might be from someone's folkloric collection (MacLean, 1973). This historical element might support that the type series was not caught in the Nagasaki area. Therefore, it is possible that collectors not only gathered specimens from Nagasaki, but Edo (present-day Tokyo) as well, which is the habitat of H. coronatus in this study (see Fig. 1; personal communication, M. van Oijen; MacLean 1973; Compton and Thujsse 2013;Nofuji et al. 2013).
Although H. coronatus sensu stricto was considered to be distributed along the coast of Japan and southern coast of Korea, we only found records from the Pacific Ocean. Mori (1928) reported H. coronatus off Korea for the first time, but the original data consisted only of checklists, not providing descriptions; thus, Mori (1928) might be reporting the occurrence of H. haema or H. coronatus. Therefore, the distribution of H. coronatus needs to be reviewed. In Korea and Japan, seahorse identification has been generally treated as a laborious task, leading to taxonomic controversy and misidentifications; thus, we recommend a careful revision of H. coronatus recorded from Korea and Japan. Senou (2002) and (2013) suggested that the publication date for H. coronatus was in 1847. However, based on Sherborn and Jentick (1895), Boeseman (1947), Mees (1962), Bauchot et al. (1982), and Eschmeyer et al. (2017), the year should be 1850.
Etymology. The specific name sindonis was derived from the name of M. Sindo, an assistant curator of fishes at Stanford University (Jordan and Snyder 1901;Lourie 2016). The English name was coined by Kuiter (2009). The Japanese name Hanatatsu literally means 'hana (flower or blossom, which indicates gorgeous) + tatsu (dragon, or the abbreviation of the word "Tatsu-no-otoshigo: seahorse")', and refers to the beautiful color and skin filaments of the species.

Hippocampus haema
Etymology. The Korean word Haema means 'seahorse', which connotes 'representative' and 'common'. Thus, the scientific and Korean names Haema were chosen to indicate that this seahorse is the one most commonly found in Korea. The Japanese name Himetatsu means 'princess seahorse' or 'dwarf seahorse', and refers to its lower coronet and smaller body compared to H. coronatus.
Remarks. Temminck and Schlegel (1850) described the extremely high coronet as follows: coronet height (CHGO, based on the inquiry of type specimens and on Jordan and Snyder [1901]'s description) of H. coronatus is identical to its SnL, 1/5 shorter than remaining HL (i.e., 4/9 of HL). All H. haema specimens present a moderately high coronet    (Fig. 6). The genetic distance between H. haema and H. coronatus is greater than that between species of the H. kuda complex (i.e., H. kuda, H. reidi, and H. ingens), supporting specific distinctness ( Fig. 7; Table 4).
Our data also suggest the existence of two subgroups, one from Korea and another from Japan: cyt b sequences of H. haema collected in these two areas consistently present two base pairs (bp) differences (0.3%-0.8% genetic distance). Based on molecular results, H. haema is more closely related to H. coronatus than to H. sindonis ( Fig. 7; Table 4), but based on coronet height and on the number of TaR, except for CoT and WS, it is more similar to H. sindonis (Tables 2 and 3).
Hippocampus haema was collected off the southern and southeastern coasts of Korea, but we were not able to collect H. haema off the western or northeastern coasts of Korea; only H. mohnikei was collected from all Korean waters. A few studies have reported H. coronatus from the western coast of Korea (Lee and Seok 1984;Hwang 1998;Hwang et al. 1998;Hwang et al. 2005), but these publications are mostly checklists, similar to that of Mori (1928), and H. mohnikei is not referred to in written records. Such inconsistency might be the result of misidentifications. The northern boundaries of H. coronatus in Korean waters determined in our study are similar to the distributions found by Choi et al. (2002) and Kim et al. (2005), who stated H. coronatus was limited to the southern coast of Korea, similarly to H. mohnikei. We found that the habitat of H. haema is affected by the Tsushima Warm Current (Briggs 1995;Nakabo 2009;Ishizu et al. 2017) and, therefore, H. haema might only rarely be found off the western and northeastern coasts of Korea.

Discussion
The NJ trees based on cyt b (670 bp), 16S rRNA (405 bp), and 12S rRNA (344 bp) recovered three monophyletic groups within the H. coronatus complex, all supported by high bootstrap probabilities (Fig. 7): viz. Hippocampus coronatus group, H. sindonis group, and H. haema group. This evidence strongly supports the existence of three species, H. coronatus, H. cf. coronatus, and H. sindonis, as suggested by Kuiter (2009). Lourie et al. (1999, based on the rings supporting the dorsal fin base (DsR), stated H. coronatus had '2 + 0 (TrR + TaR)' and H. sindonis had '2 + 1'. However, Jordan and Snyder (1901) described H. sindonis as '2 + 0' and H. coronatus as '2 + 1', which is the reverse. Moreover, all species within the H. coronatus complex described in the present study include '2 + 0' and '2 + 1' forms (Table 2). Thus, DsR is an inappropriate characteristic to diagnose the species studied here. Hippocampus coronatus has only one supraorbital spine whereas H. sindonis has two and H. haema has either one or two spines. Many ichthyologists have attempted to distinguish H. coronatus and H. sindonis based on color and skin filaments (especially Jordan and Snyder 1901). However, Curtis (2006) refuted the use of skin filaments on its key to distinguish H. hippocampus from H. guttulatus, as skin filaments grow irregularly in both species. Lourie et al. (1999) and Szabó et al. (2011) also suggested that color and skin filaments were affected by environment and/or growth, and therefore should be considered of limited diagnostic value. In the present study, several color and skin filament patterns were found in H. haema, which is in agreement with Mitani's (1956) data for specimens sampled from Maizuru Bay, Japan. This author interpreted these as intraspecific variations, but, given the results obtained in this study by molecular analyses, we do not agree that H. coronatus and H. sindonis should be treated as a single species. Hippocampus coronatus is ranked as DD in the IUCN Red List due to the lack of information on its population trends and to the uncertainty of its distributions, originating from taxonomic controversies (Zhang and Pollom 2016). Hippocampus sindonis is ranked as Least Concern (LC) because no major threat has been reported for its distribution (Fritzsche et al. 2010). The distribution of H. coronatus is similar to that of H. sindonis (i.e., southeastern coast of Honshu, Japan), and there is no data supporting its potential threat with distribution uncertainty. However, H. coronatus distribution has a narrower range than that of H. sindonis (Fig. 1), so it is more likely to be affected by human pressure. For these reasons, H. coronatus will likely be ranked above or equal to H. sindonis after further surveys of its population trends. To improve the conservation of these species, a better taxonomic understanding is required to resolve the DD rank of H. coronatus regarding the uncertainty of its distribution, as well as more data on its biology, habitat, and abundance. Previous studies considering the biology of H. coronatus conducted on local Korean areas (Choi et al. 2006(Choi et al. , 2012Huh et al. 2014;Park and Kwak 2015), might, in fact, indicate the biology of H. haema. Overfishing could potentially threat H. haema due to by-catch, given the species low density and patchy distribution (Choi et al. 2012;Zhang and Pollom 2016), and its wide distribution requires the study of populations across the entire area.  Lourie et al. (1999, Senou (2013), Lourie (2016), and the current study data.