*Article* **Discovery of Pelagic Eggs of Two Species from the Rare Mesopelagic Fish Genus** *Trachipterus* **(Lampriformes: Trachipteridae)**

**Hae-young Choi <sup>1</sup> , Hee-chan Choi <sup>2</sup> , Sung Kim 3,\*, Hyun-ju Oh <sup>1</sup> and Seok-hyun Youn 1,\***


**Abstract:** The ecology of the mesopelagic fish genus *Trachipterus*, which is rarely found in oceans, remains unclear. In this study, we found 22 eggs of *T. trachypterus* and *T. jacksonensis* around the Ulleung Basin of the East/Japan Sea during ichthyoplankton surveys from 2019 to 2021. The eggs were identified through genetic relationships with the genus *Trachipterus* based on partial sequences (COI and 16S) or concatenated sequences of 13 protein-coding genes and 2 rRNA genes of mitochondrial DNA. *T. trachypterus* eggs were discovered in all seasons, but more frequently during the winter. One *T. jacksonensis* egg that appeared during the autumn was the first in the northwestern Pacific Ocean. Identifying *Trachipterus* pelagic eggs would provide insight into their spawning ecology and biogeography.

**Keywords:** mesopelagic fish; mitochondrial DNA sequence; pelagic fish eggs; spawning; *Trachipterus jacksonensis*; *Trachipterus trachypterus*; Ulleung Basin

### **1. Introduction**

Members of the genus *Trachipterus* (Lampriformes, Trachipteridae), which comprise six species (*T. altivelis*, *T. jacksonensis*, *T. arcticus*, *T. fukuzakii*, *T. trachypterus*, and *T. ishikawae*) [1,2], have elongated and compressed bodies with large eyes and are distributed in several oceans [3,4]. They are rarely caught by deep fishing gear or found inshore [5,6]. The species of *Trachipterus* have been reported based on a few specimens [7–10].

The first nominal species of *Trachipterus* is *T. trachypterus* [11]. *Trachipterus* has been described using juvenile specimens, which reside in shallower waters than adults. Due to its rarity and the morphological changes at early developmental stages, *T. trachypterus* larvae of different sizes are often incorrectly identified [12]. *T. trachypterus* eggs were first reported as three specimens in the Gulf of Napoli [12]. The Mediterranean Sea, where *T. trachypterus* eggs, larvae, and adults have been found several times, is a known spawning ground [13].

Fish eggs are a key indicator of spawning and evidence of species intrusion [14,15]. Because most marine teleost fish release large numbers of pelagic eggs, their egg distribution density is higher than that of spawners [16]. Even in rare species, the probability of finding for eggs is higher than that for adult fish during the spawning season [15]. In addition, eggs with a shorter pelagic duration than larvae are in or close to spawning grounds [17]. The spawning grounds of the Japanese eel, *Anguilla japonica*, were revealed through DNA barcoding-based identification of their eggs after prolonged research [18].

**Citation:** Choi, H.-y.; Choi, H.-c.; Kim, S.; Oh, H.-j.; Youn, S.-h. Discovery of Pelagic Eggs of Two Species from the Rare Mesopelagic Fish Genus *Trachipterus* (Lampriformes: Trachipteridae). *J. Mar. Sci. Eng.* **2022**, *10*, 637. https:// doi.org/10.3390/jmse10050637

Academic Editor: Roberto Carlucci

Received: 15 March 2022 Accepted: 4 May 2022 Published: 7 May 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

DNA barcoding enables species-level identification of fish eggs [19]. Pelagic marine fish eggs are typically transparent and round [20]. Eggs of the same species may have morphologically different embryos depending on developmental stage [21]. Unlike dramatic changes in morphology, DNA remains constant throughout life history [22,23]. Intra- and inter-specific genetic distances, typically based on mitochondrial DNA (mtDNA) sequences (COI, 12S, 16S, etc.), are analyzed to identify fish eggs and larvae species [24–26]. dramatic changes in morphology, DNA remains constant throughout life history [22,23]. Intra- and inter-specific genetic distances, typically based on mitochondrial DNA (mtDNA) sequences (COI, 12S, 16S, etc.), are analyzed to identify fish eggs and larvae species [24–26]. We conducted COI barcoding of pelagic eggs collected from the East/Japan Sea from 2019 to 2021 and surveyed the literature that applied DNA barcoding to fish eggs

vealed through DNA barcoding-based identification of their eggs after prolonged re-

DNA barcoding enables species-level identification of fish eggs [19]. Pelagic marine fish eggs are typically transparent and round [20]. Eggs of the same species may have morphologically different embryos depending on developmental stage [21]. Unlike

*J. Mar. Sci. Eng.* **2022**, *10*, x FOR PEER REVIEW 2 of 12

We conducted COI barcoding of pelagic eggs collected from the East/Japan Sea from 2019 to 2021 and surveyed the literature that applied DNA barcoding to fish eggs near the study area. The eggs in this study were finally determined to belong to *T. trachypterus* and *T. jacksonensis* based on their COI, 16S rRNA, and mitogenome sequences. Here, we report *T. trachypterus* spawning and the first finding of *T. jacksonensis* based on eggs in the East/Japan Sea. near the study area. The eggs in this study were finally determined to belong to *T. trachypterus* and *T. jacksonensis* based on their COI, 16S rRNA, and mitogenome sequences. Here, we report *T. trachypterus* spawning and the first finding of *T. jacksonensis* based on eggs in the East/Japan Sea. **2. Materials and Methods** 

#### **2. Materials and Methods** *2.1. Pelagic Fish Egg Collection*

search [18].

#### *2.1. Pelagic Fish Egg Collection* Pelagic fish eggs were collected from 13 stations in the East/Japan Sea (Figure 1)

Pelagic fish eggs were collected from 13 stations in the East/Japan Sea (Figure 1) using a zooplankton net (mouth diameter: 80 cm; mesh size: 300 µm) during research cruises on R/V Tamgu 3 in four seasons (winter: February and March; spring: April; summer: June and August; autumn: October and November) from 2019 to 2021. The water depths at the stations are 126–2203 m. The net was towed obliquely from 10 m above the bottom to the surface at the stations with a water depth of less than 300 m (st. 103-05, 209-05, and 209-07). At the other stations, the net was lowered to a depth of 300 m and was towed as above. Samples were preserved in 95% ethanol. Temperature and salinity were measured using CTD (SBE 911plus, Sea-Bird Scientific Inc., Bellevue, WA, USA). using a zooplankton net (mouth diameter: 80 cm; mesh size: 300 µm) during research cruises on R/V Tamgu 3 in four seasons (winter: February and March; spring: April; summer: June and August; autumn: October and November) from 2019 to 2021. The water depths at the stations are 126–2203 m. The net was towed obliquely from 10 m above the bottom to the surface at the stations with a water depth of less than 300 m (st. 103-05, 209-05, and 209-07). At the other stations, the net was lowered to a depth of 300 m and was towed as above. Samples were preserved in 95% ethanol. Temperature and salinity were measured using CTD (SBE 911plus, Sea-Bird Scientific Inc., Bellevue, WA, USA).

**Figure 1.** Pelagic fish egg sampling stations in the East/Japan Sea, 2019–2021. Numbers with plus marks, station name; italic number, station lines. The base map is a Google Satellite map drawn using QGIS [27] through the XYZ Tiles tool (https://mt1.google.com/vt/lyrs=s&x={x}&y={y}&z={z}, accessed on 10 March 2022). **Figure 1.** Pelagic fish egg sampling stations in the East/Japan Sea, 2019–2021. Numbers with plus marks, station name; italic number, station lines. The base map is a Google Satellite map drawn using QGIS [27] through the XYZ Tiles tool (https://mt1.google.com/vt/lyrs=s&x=\{x\}&y=\{y\}&z= \{z\}, accessed on 10 March 2022).

Fish eggs were sorted from the samples using a stereomicroscope (M125C, Leica, Wetzlar, Germany). Among them, the 22 largest eggs were selected and photographed using a camera mounted on a stereomicroscope (SMZ18, Nikon, Tokyo, Japan).

### *2.2. Genomic DNA Extraction, PCR, and Sequencing and Sequence Analysis*

Genomic DNA was extracted from 22 eggs according to the protocol of the DNeasy Blood & Tissue Kit (Qiagen, Hilden, Germany). The COI gene of mtDNA was amplified using the primers VF2\_t1 (50 -CAACCAACCACAAAGACATTGGCAC-30 ), FishF2\_t1 (50 -TCGACTAATCATAAAGATATCGGCAC-30 ), FishR2\_t1 (50 -ACTTCAGGGTGACCGAA GAATCAGAA-30 ), and FR1d\_t1 (50 -ACCTCAGGGTGTCCGAARAAYCARAA-30 ) [28,29]. The 20 µL PCR mixture was composed of 10 µL of 2X DNA free-Taq Master Mix including PCR buffer, dNTPs mixture, and Taq DNA polymerase (CellSafe, Gyeonggi, Korea), 0.2 µL of each of the four primers, 2 µL of genomic DNA, and 7.2 µL of distilled water. The PCR program consisted of initial denaturation at 94 ◦C for 3 min; 35 cycles of denaturation at 94 ◦C for 30 s, annealing at 52 ◦C for 40 s, extension at 72 ◦C for 1 min; and a final extension at 72 ◦C for 7 min. The PCR products were sequenced on a 3730xl DNA Analyzer (Applied Biosystems, Foster City, CA, USA).

Other mtDNA regions of the 7 eggs among the 22 eggs were also amplified for comparison with those of related taxa. The 16S rRNA gene of one egg was amplified using the 16Sar (5-CGCCTGTTTATCAAAAACAT-3) and 16Sbr (5-CCGGTCTGAACTCAGATCACGT-3) primers [30]. PCR and sequencing methods were used for COI analysis, except that the annealing temperature was 56 ◦C. Six eggs were selected to analyze mitogenome sequences in the consideration of sample conditions; three eggs were collected in 2019 and 2020 each. To obtain sufficient DNA to analyze the complete mitogenome of the six eggs, the whole genome was amplified following the REPLI-g Mini Kit (Qiagen) protocol. The amplified products were sequenced using a NovaSeq 6000 (Illumina, San Diego, CA, USA). A total of 385,684,576–474,309,190 raw reads (length: 150 bp) were obtained from the six eggs. The reads were mapped to a reference sequence (GenBank accession number: NC\_003166, [31]) using Geneious R11 [32] mapper. The resulting consensus sequences were annotated using MitoFish [33] and Geneious R11. The three mitogenome sequences were constructed from three eggs (2002E3–E5), except for the d-loop region (mean coverage, 112 ± 119×–7866 ± 11,917×; 16,159–16,163 bp), but COI (1467–1551 bp) and 16S (1595–1683 bp) sequences were obtained from the other three eggs (1902E1, 1902E2, and 1904E1).

The COI, 16S, and mitogenome sequences from the 22 eggs were searched using BLAST [34] and BOLD systems [35] to identify related taxa. The sequences of the eggs, related taxa, and outgroups were aligned using Clustal Omega [36]. The sequences were used to analyze the maximum likelihood (ML) tree based on the best-fit substitution model [37–39] and Kimura 2-parameter distances in MEGA X (ver. 11.0.10) [40]. The egg sequences were submitted to NCBI GenBank under accession numbers OM527130– OM527151, OM527153, OM574770–OM574772, and ON231742–ON231747 (Table 1). We also investigated the literature using DNA barcoding for the species identification of eggs around the study area, and sequences from the eggs of Shin et al. [41] were used in this study.


**Table 1.** GenBank accession numbers of eggs of this study and literature.


**Table 1.** *Cont*.

Underlined sequences were analyzed from shotgun sequencing.

### **3. Results**

### *3.1. Genetic Identification of Eggs*

Twenty-two egg specimens were identified as *T. trachypterus* and *T. jacksonensis* according to genetic relationships based on COI, 16S, and mitogenome sequences.

The COI sequences of the 22 eggs and 3 eggs from the reference [41] and genus *Trachipterus* consisted of three distinct clades (between genetic distances: average ± standard deviation, 0.158 ± 0.057; min, 0.083; max, 0.227; Table S1), with either two or three species in the maximum likelihood (ML) tree (Figure 2). Of the three clades, COI\_Clade 1 had sequences from 24 eggs, *T. altivelis* and *T. trachypterus*, and COI\_Clade 2 had sequences from one egg of this study, *T. arcticus*, *T. jacksonensis*, and *Trachipterus* sp. Although each clade consisted of sequences from different species, the genetic distances within the clades (COI\_Clade 1, 0.009 ± 0.006; COI\_Clade 2, 0.010 ± 0.007; and COI\_Clade 3, 0.009 ± 0.005) were less than between the clades (0.158 ± 0.057), indicating that each clade represented the species.

Species of the two clades (COI\_ Clades 1 and 2; Figure 2a), including the eggs, were reanalyzed based on 16S rRNA and mitogenome sequences. Among the eggs of COI\_Clade 1, COI and 16S rRNA sequences for three eggs (1902E1, 1902E2, and 1904E1) and mitogenome sequences excluding the d-loop for three eggs (2002E3–E5) were obtained from mitogenome analysis (Table 1). The COI sequences of the six eggs were also located in COI\_Clade 1. Concatenated sequences (13 protein-coding genes and two rRNAs) of the three eggs (2002E3–E5) formed a clade with those of *T. trachypterus* (NC\_003166.1) (genetic distance, 0.007 ± 0.003; Table S2), and they were distinct from those of *Desmodema polystictum* and *Zu cristatus* of Trachipteridae (0.258 ± 0.012) (Figure 2b). The 16S rRNA sequences of the six eggs from the same samples of COI\_Clade 1 and one egg (MH144584.1) from [41] formed a clade with *T. trachypterus* and *T. altivelis* with very small genetic distances (0.002 ± 0.001; min, 0.000; max, 0.004; Table S3) (Figure 2c). Interestingly, the *T. trachypterus* 16S sequence (DQ027909.1, [42]) was distinct from 16S\_Clade 1, although it was analyzed from the same specimen with the *T. trachypterus* COI sequence (DQ027978.1) of COI\_Clade 1.

that each clade represented the species.

**Figure 2.** Maximum likelihood (ML) tree constructed using mitochondrial DNA sequences of pelagic fish eggs, *Trachipterus* species, and outgroups. (**a**) COI ML tree (based on the HKY + G + I model) including COI sequences from 25 eggs. Sequences of eggs with superscripts (bc, c) were also analyzed in the trees in (**b**) and (**c**). (**b**) Thirteen protein-coding genes and two rRNA genes ML tree (based on the GTR + G + I model), including concatenated sequences from the three eggs. (**c**) 16S rRNA ML tree (based on the K2 + G model), including 16S rRNA sequences from seven eggs. Bootstrap values (1000 replicates) over 50% are shown on the branches. Sequences shaded in gray were obtained using shotgun sequencing. **Figure 2.** Maximum likelihood (ML) tree constructed using mitochondrial DNA sequences of pelagic fish eggs, *Trachipterus* species, and outgroups. (**a**) COI ML tree (based on the HKY + G + I model) including COI sequences from 25 eggs. Sequences of eggs with superscripts (bc, c) were also analyzed in the trees in (**b**,**c**). (**b**) Thirteen protein-coding genes and two rRNA genes ML tree (based on the GTR + G + I model), including concatenated sequences from the three eggs. (**c**) 16S rRNA ML tree (based on the K2 + G model), including 16S rRNA sequences from seven eggs. Bootstrap values (1000 replicates) over 50% are shown on the branches. Sequences shaded in gray were obtained using shotgun sequencing.

COI\_Clade 3, 0.009 ± 0.005) were less than between the clades (0.158 ± 0.057), indicating

Species of the two clades (COI\_ Clades 1 and 2; Figure 2a), including the eggs, were re-analyzed based on 16S rRNA and mitogenome sequences. Among the eggs of COI\_Clade 1, COI and 16S rRNA sequences for three eggs (1902E1, 1902E2, and 1904E1) and mitogenome sequences excluding the d-loop for three eggs (2002E3–E5) were obtained from mitogenome analysis (Table 1). The COI sequences of the six eggs were also located in COI\_Clade 1. Concatenated sequences (13 protein-coding genes and two One egg (2110E1) formed COI\_Clade 2 with *T. arcticus, T. jacksonensis, T. altivelis*, and *Trachipterus* sp. comprised a clade with only *T. jacksonensis* (genetic distance, 0.000 ± 0.000) in the 16S ML tree (16S\_Clade 2; Figure 2c). The 16S sequence of *T. arcticus* (KJ128928.1) diverged sharply (0.107) from the clade of *T. jacksonensis* with the egg (2110E1; OM527130). The *T. arcticus* 16S rRNA sequence (KJ128928.1) was obtained from the same specimen as the *T. arcticus* COI sequence (KJ128643.1) in COI\_Clade 3.

#### rRNAs) of the three eggs (2002E3–E5) formed a clade with those of *T. trachypterus 3.2. Pelagic Eggs of Two Trachipterus Species*

COI sequence (DQ027978.1) of COI\_Clade 1.

(NC\_003166.1) (genetic distance, 0.007 ± 0.003; Table S2), and they were distinct from those of *Desmodema polystictum* and *Zu cristatus* of Trachipteridae (0.258 ± 0.012) (Figure 2b). The 16S rRNA sequences of the six eggs from the same samples of COI\_Clade 1 and one egg (MH144584.1) from [41] formed a clade with *T. trachypterus* and *T. altivelis* with very small genetic distances (0.002 ± 0.001; min, 0.000; max, 0.004; Table S3) (Figure 2c). Interestingly, the *T. trachypterus* 16S sequence (DQ027909.1, [42]) was distinct from 16S\_Clade 1, although it was analyzed from the same specimen with the *T. trachypterus* The average diameter of *T. trachypterus* eggs was 3.2 ± 0.1 mm (min, 2.7 mm; max, 3.6 mm) (Figure 3; Table S4). The diameter of *T. jacksonensis* egg was 2.2 mm, smaller than that of *T. trachypterus* eggs. The internal morphology of eggs preserved in 95% ethanol was difficult to investigate. The common characteristics of *T. trachypterus* eggs that could be confirmed were a narrow perivitelline space and a lack of oil globules. The developed embryos had melanophores on the head, dorsal side, and on the yolk sac around the middle of the body.

One egg (2110E1) formed COI\_Clade 2 with *T. arcticus, T. jacksonensis, T. altivelis*, and *Trachipterus* sp. comprised a clade with only *T. jacksonensis* (genetic distance, 0.000 ±

0.000) in the 16S ML tree (16S\_Clade 2; Figure 2c). The 16S sequence of *T. arcticus* (KJ128928.1) diverged sharply (0.107) from the clade of *T. jacksonensis* with the egg (2110E1; OM527130). The *T. arcticus* 16S rRNA sequence (KJ128928.1) was obtained from

The average diameter of *T. trachypterus* eggs was 3.2 ± 0.1 mm (min, 2.7 mm; max, 3.6 mm) (Figure 3; Table S4). The diameter of *T. jacksonensis* egg was 2.2 mm, smaller than that of *T. trachypterus* eggs. The internal morphology of eggs preserved in 95% ethanol was difficult to investigate. The common characteristics of *T. trachypterus* eggs that could be confirmed were a narrow perivitelline space and a lack of oil globules. The developed embryos had melanophores on the head, dorsal side, and on the yolk sac

the same specimen as the *T. arcticus* COI sequence (KJ128643.1) in COI\_Clade 3.

*3.2. Pelagic Eggs of Two Trachipterus Species* 

around the middle of the body.

**Figure 3.** Pelagic eggs of two *Trachipterus* species preserved in 95% ethanol. 1902E1–2108E1, *T. trachypterus*; 2110E1, *T. jacksonensis*. Scale bar, 1.0 mm. **Figure 3.** Pelagic eggs of two *Trachipterus* species preserved in 95% ethanol. 1902E1–2108E1, *T. trachypterus*; 2110E1, *T. jacksonensis*. Scale bar, 1.0 mm.

#### *3.3. Distribution of Trachipterus Eggs 3.3. Distribution of Trachipterus Eggs*

*T. trachypterus* eggs appeared in all seasons around the Ulleung Basin and Dokdo, and one *T. jacksonensis* egg appeared once in autumn (Figure 4). The occurrence frequency of *T. trachypterus* eggs was highest during the winter and lowest during the summer and autumn. The surface temperature of the stations where eggs were detected ranged from 10.0 °C to 25.9 °C (Table S4). The mean surface temperature of the study area was the lowest during the winter and peaked in the summer (Table S5). Unlike the surface, the bottom temperature was constant at approximately 1 °C. Thermocline was generated by the large difference between the temperatures of the surface and bottom. The depth and strength of the thermocline varied according to season and station. The *T. trachypterus* eggs appeared in all seasons around the Ulleung Basin and Dokdo, and one *T. jacksonensis* egg appeared once in autumn (Figure 4). The occurrence frequency of *T. trachypterus* eggs was highest during the winter and lowest during the summer and autumn. The surface temperature of the stations where eggs were detected ranged from 10.0 ◦C to 25.9 ◦C (Table S4). The mean surface temperature of the study area was the lowest during the winter and peaked in the summer (Table S5). Unlike the surface, the bottom temperature was constant at approximately 1 ◦C. Thermocline was generated by the large difference between the temperatures of the surface and bottom. The depth and strength of the thermocline varied according to season and station. The center of the thermocline was usually at a depth of 100–200 m (Figure S1). Salinity was approximately 33–34 psu, and there was no significant difference depending on season and depth, unlike the temperature.

center of the thermocline was usually at a depth of 100–200 m (Figure S1). Salinity was approximately 33–34 psu, and there was no significant difference depending on season

and depth, unlike the temperature.

**Figure 4.** Distributions of pelagic eggs of two *Trachipterus* species for the four seasons: (**a**) winter, (**b**) spring, (**c**) summer, and (**d**) autumn. *Tt*, *T. trachypterus*; *Tj*, *T. jacksonensis*; Sampling year of *Trachipterus* eggs: filled circle, 2019; fisheye, 2019 and 2020; filled triangle, 2020; filled square, 2021 of this study; filled diamond, 2017–2019 of Shin et al. [41]; no egg detection: cross, this study; empty diamond, Shin et al. [41]. **Figure 4.** Distributions of pelagic eggs of two *Trachipterus* species for the four seasons: (**a**) winter, (**b**) spring, (**c**) summer, and (**d**) autumn. *Tt*, *T. trachypterus*; *Tj*, *T. jacksonensis*; Sampling year of *Trachipterus* eggs: filled circle, 2019; fisheye, 2019 and 2020; filled triangle, 2020; filled square, 2021 of this study; filled diamond, 2017–2019 of Shin et al. [41]; no egg detection: cross, this study; empty diamond, Shin et al. [41].

#### **4. Discussion 4. Discussion**

DNA barcoding of fish eggs enables the investigation of the spawning ecology of various species [43,44]. Studies on fish eggs have been limited to a few species due to the difficulty in morphology-based egg identification [45]. Egg morphological characteristics were described based on larval fish hatching from eggs or eggs obtained from adult fish [46,47]. Recently, larval fish, which have more morphological features available for identification than eggs, have been analyzed using DNA barcoding to improve the accuracy of species identification [48]. However, DNA barcoding has limitations in determining species due to the lack of comparable DNA sequences, unsuitable DNA regions for barcoding, or sequences from misidentified specimens. DNA barcoding of fish eggs enables the investigation of the spawning ecology of various species [43,44]. Studies on fish eggs have been limited to a few species due to the difficulty in morphology-based egg identification [45]. Egg morphological characteristics were described based on larval fish hatching from eggs or eggs obtained from adult fish [46,47]. Recently, larval fish, which have more morphological features available for identification than eggs, have been analyzed using DNA barcoding to improve the accuracy of species identification [48]. However, DNA barcoding has limitations in determining species due to the lack of comparable DNA sequences, unsuitable DNA regions for barcoding, or sequences from misidentified specimens.

#### *4.1. Genetic Identification of Eggs 4.1. Genetic Identification of Eggs*

The first DNA barcode for species identification of 22 eggs in this study was the COI sequence, which contains a lot of information available to identify fish. The maximum likelihood (ML) tree based on the COI sequences of the 25 eggs, including those of The first DNA barcode for species identification of 22 eggs in this study was the COI sequence, which contains a lot of information available to identify fish. The maximum likelihood (ML) tree based on the COI sequences of the 25 eggs, including those of Shin et al. [41] and *Trachipterus*, showed three distinct clades of *Trachipterus* (Figure 2a). Each clade was regarded as a species by considering the genetic distances of the intra- and inter-clades of the COI sequences of fishes [23,28]. However, the coexistence of sequences from two or three species within each clade was problematic: COI\_Clade 1, *T. altivelis*, and

*T. trachypterus* with 24 eggs; COI\_Clade 2, *T. arcticus*, *T. jacksonensis*, *Trachipterus* sp., and one egg (2110E1); and COI\_Clade 3, *T. arcticus*, *T. jacksonensis*, and *T. trachypterus*. This could be because COI sequences were analyzed from misidentified samples or were not appropriate for distinguishing the species.

The *T. trachypterus* mitogenome sequence was the key to determining the species of COI\_Clade 1. The mitogenome sequences of three eggs (2002E3–E5) located in COI\_Clade 1 formed a clade with that of *T. trachypterus* (NC\_003166.1) (Figure 2b). In addition, genetic distances between the 16S rRNA sequences of the seven eggs (1902E1, 1902E2, 1904E1, 2002E3–E5, and DI\_7 (MH144584.1)) and *T. trachypterus* (NC\_003166, MH144581.1) were also very small (0.000–0.004; Table S3). These genetic relationships were a criterion for identifying 24 eggs, including those of Shin et al. [41] in COI\_Clade 1 as *T. trachypterus*.

COI\_Clade 2, including one egg (2110E1; OM527153), was determined to be *T. jacksonensis* based on the distinct 16S\_Clade 2 with sequences of *T. jacksonensis* and the one egg (2110E1; OM527130) in the 16S ML tree (Figure 2c). In addition, COI\_Clade 3 was identified as *T. arcticus* based on the *T. arcticus* sequences (COI, KJ128643.1; 16S rRNA, KJ128928.1) showing the same position in the COI and 16S rRNA ML tree (Figure 2a,c).

Interestingly, the *T. trachypterus* 16S sequence (DQ027909.1, [42]) from the same sample with COI sequence (DQ027978.1, [42]) in the COI\_Clade 1 was independently clustered in the 16S ML tree (Figure 2c). This indicates that the 16S rRNA sequence (DQ027909.1) was mishandled during the experiment or sequence analysis. In addition, *T. altivelis* (GU440558.1) of COI\_Clade 1 and *T. altivelis* (AY958674.1) of 16S\_Clade 1 would be derived from misidentification.

### *4.2. Trachipterus Eggs*

*T. trachypterus* is widely distributed in the Mediterranean Sea and Atlantic Indo-Pacific Oceans [5,6,9,49–60]. Its spawning area is the Mediterranean Sea, where eggs of *Trachypterus taenia* (synonym for *T. trachypterus*) have been found [12]. Lo Bianco [12] collected three live eggs of *T. taenia* from a sampling gear lowered to a depth of 100–150 m in the Gulf of Napoli in February, May, and October 1905–1907. Spawning was assumed to occur year-round. Three eggs (diameter, 2.90–2.95 mm) were in late developmental stages. In this study, we confirmed that *T. trachypterus* spawned in all seasons and peaked in winter based on eggs continuously collected during the survey periods. If the *T. trachypterus* eggs in this study had similar ecological characteristics to the eggs from the Gulf of Napoli, the spawning depth could be estimated as the thermocline layer.

*T. jacksonensis* is distributed in the Southern Hemisphere (Australia, New Zealand, Africa, Brazil, and others) [61–70]. Adults have not been reported in the Northern Hemisphere. The eggs detected once in autumn in this study were the first in the Northern Hemisphere. Similarly to *T. jacksonensis*, *Peristedion liorhynchus* has not been recorded in Korean waters, but its eggs and larvae have been recorded [15]. This suggests that eggs may be useful for the detection of rare species.

### **5. Conclusions**

In this study, we discovered *T. trachypterus* and *T. jacksonensis* eggs in the Ulleung Basin of the East/Japan Sea of which their sequences could be used for post verification. The Ulleung Basin is a spawning area for *T. trachypterus* along with the Mediterranean Sea; this is the first report of *T. jacksonensis* egg in the northwestern Pacific. Finding *Trachipterus* species eggs will elucidate their spawning ecology and geographic distribution.

**Supplementary Materials:** The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/jmse10050637/s1, Figure S1: Vertical distribution of temperature at stations where *Trachipterus* eggs appeared in this study; Table S1: Pairwise genetic distances of the COI genes of pelagic eggs of this study, *Trachipterus* species, and outgroups. Outlined values belong to each clade of the maximum likelihood tree in Figure 2a; Table S2: Pairwise genetic distances of the concatenated mtDNA sequences of 13 protein-coding genes and two rRNA genes of pelagic eggs of this study, *Trachipterus* species, and outgroups; Table S3: Pairwise genetic distances of the 16S rRNA

genes of pelagic egg of this study, *Trachipterus* species, and outgroups. Outlined values belong to each clade of the maximum likelihood tree in Figure 2c; Table S4: Information on *Trachipterus* eggs of this study and literature; Table S5: Mean and standard deviation values of temperature and salinity from 2019 to 2021 in the East/Japan Sea.

**Author Contributions:** Conceptualization, H.-y.C.; methodology, H.-y.C. and S.K.; validation, H.-y.C. and S.K.; formal analysis, H.-y.C. and S.K.; investigation, H.-c.C.; data curation, H.-y.C.; writing original draft preparation, H.-y.C.; writing—review and editing, H.-y.C., H.-c.C., S.K. and S.-h.Y.; visualization, H.-y.C.; supervision, S.-h.Y.; project administration, H.-j.O.; funding acquisition, H.-j.O. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the National Institute of Fisheries Science (NIFS) grant (Study on the ecophysiology and occurrence prediction of jellyfish; R2022052).

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The Sequence data that support the findings of this study are deposited in NCBI/GenBank (https://www.ncbi.nlm.nih.gov/genbank/) under accession numbers OM527130- OM527151, OM527153, OM574770-OM574772, and ON231742-ON231747.

**Acknowledgments:** We appreciate the captain and crews of the Tamgu3 research vessel for the survey. We are also grateful to the researchers in the NIFS for their assistance with sampling.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**


## *Article* **Feeding Strategy of the Wild Korean Seahorse (***Hippocampus haema***)**

**Myung-Joon Kim <sup>1</sup> , Hyun-Woo Kim <sup>2</sup> , Soo-Rin Lee <sup>3</sup> , Na-Yeong Kim <sup>1</sup> , Yoon-Ji Lee <sup>1</sup> , Hui-Tae Joo <sup>4</sup> , Seok-Nam Kwak <sup>5</sup> and Sang-Heon Lee 1,\***


**Abstract:** The feeding and spawning grounds for seahorses have been lost due to nationwide coastal developments in South Korea. However, little information on the feeding ecology of the Korean seahorse (*Hippocampus haema*) is currently available. The main objective in this study was to understand the feeding strategy of *H. haema* on the basis of DNA analysis of the contents of the guts. This is the first study on the feeding ecology of *H. haema*. Crustaceans were found to be major prey for *H. haema* in this study. Among the 12 identified species, arthropods were predominantly observed as potential prey of *H. haema* in this study. The *Caprella* sp. Was detected in all summer specimens followed by the *Ianiropsis* sp., whereas isopods were dominant, and amphipods accounted for a small proportion in winter specimens. According to the results in this study, there appears to be a seasonal shift in the major prey of *H. haema*. Moreover, a potential change in the habitats for adults was further discussed. Since this is a pilot study, further studies should be conducted for a better understanding of the feeding ecology of *H. haema*.

**Keywords:** wild seahorse; *H. haema*; feeding habits; NGS analysis

### **1. Introduction**

Seahorses are fascinating creatures for many people around the world due to their unique appearance and life-history characteristics that are different to those of common fish [1,2]. The overfishing of seahorses is occurring in some regions [3,4], and reckless coastal developments causing their habitat loss are also threatening their survival [2,5,6]. For this reason, the scientific community is making various efforts to reduce the loss of seahorse populations and preserve the *Hippocampus* species by adding them to lists or conventions, such as the red list of the IUCN (International Union for Conservation of Nature) and CITES (Convention on International Trade in Endangered Species of Wild Fauna and Flora). Until now, there is no record of seahorse overfishing along the Republic of Korea coastal area; however, the Korean coastal ecosystem, due to ongoing nationwide coastal development, has been losing its natural shelter abilities as a safe habitat and spawning ground for various fishes, including seahorses [2,7]. Despite these crises, very little ecological information on seahorses is currently available in the Republic of Korea [8,9]; thus, we do not know what ecological roles they play in coastal ecosystems.

Seahorses lead a sedentary life in coastal areas where the environment can be changed dynamically compared to the open ocean [10–13]. Five seahorse species have been reported

**Citation:** Kim, M.-J.; Kim, H.-W.; Lee, S.-R.; Kim, N.-Y.; Lee, Y.-J.; Joo, H.-T.; Kwak, S.-N.; Lee, S.-H. Feeding Strategy of the Wild Korean Seahorse (*Hippocampus haema*). *J. Mar. Sci. Eng.* **2022**, *10*, 357. https://doi.org/ 10.3390/jmse10030357

Academic Editor: Ka Hou Chu

Received: 8 January 2022 Accepted: 28 February 2022 Published: 3 March 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

in South Korean waters, mainly in the seaweed and seagrass beds throughout the southern coast [8,9,14–17]. Recently, *H. haema*, previously known as *H. coronatus*, was newly identified as a Korean species [18]. The first study on the morphometric characteristics and basic ecological data for the newly recorded Korean seahorse (*H. haema*) populations was conducted in Geoje-Hansan Bay [8]. However, no study has been conducted on their feeding ecology to date.

The study of the feeding habits of an organism is a cornerstone that can be used for efficient management, preservation, and artificial reproduction of specific organisms, and it is widely conducted for various fish [19–24]. Although several previous studies were conducted on the feeding habits of some seahorse species under laboratory conditions (*H. abdominalis* [25], *H. barbouri* [26], *H. erectus* [27], *H. guttulatus* [28], *H. hippocampus* [29], *H. kuda* [30], and *H. reidi* [31,32]), the information on wild seahorses is still insufficient [24,33–42].

A visual analysis (naked eye or microscope) of gut contents, often used in feeding ecology, can show what quantity of prey is present in the target fish's gut; however, this approach may overestimate some preys that have just been recently eaten or are difficult to digest. Often times, it is difficult to identify what kinds of prey they have eaten due to their easily digestible characteristics or their small size [39,42]. This difference in the digestibility of the prey items can cause some bias when evaluating their diet compositions [39,42]. Stable isotope analysis has the advantage of being able to perform nonlethal analysis of the target organism, but it is difficult to detect the recently eaten prey due to the time gaps in the turnover rate of isotopes [43,44]. On the other hand, metabarcoding has the advantage of being able to detect short-term feeding habits with a very small amount of sample, as well as to detect soft and highly digested items, which are not recognizable through morphological identification [42,45]. Moreover, metabarcoding is also a nonlethal method, depending on the sampling method (i.e., fecal sample [42] or flushing method [34,40]). Therefore, in this study, a DNA analysis method was applied to understand the feeding habits of small *Hippocampus* individuals.

In this study, the main objective was to understand the feeding strategy of the Korean seahorse species (*H. haema*) in the coastal environment on the basis of the contents of the guts in summer and winter, using molecular biological approaches. This is the first study on the feeding ecology of the Korean seahorse species (*H. haema*).

### **2. Materials and Methods**

### *2.1. Study Area and Samplings*

Among the samples collected by the Environ-Ecological Engineering Institute (EEEI) for the investigation of fish biota in the sargassum bed (*Sargassum piluliferum*), seven specimens were received frozen in summer (July 2017) and winter (January 2019) for a comparison of the feeding strategies of *H. haema*. The sampling area was Geoje-Hansan Bay, Republic of Korea (34◦4804000 N; 128◦3100300 E; Figure 1). The mean water depth of the sampling site was approximately 3 m. The two specimens taken from the samples reported in [8] were analyzed to confirm the possibility of feeding differences between each sampling group even in the summer season (Table 1). The body length and wet weight were measured at the home laboratory using a caliper (Mitutoyo, 0.01 mm, Kawasaki, Japan) and a scale (Mettler Toledo, 1 mg, Columbus, OH, USA) according to [46]. After those measurements, all specimens were stored at −80 ◦C in a freezer and brought to the Marine Ecological Laboratory in Pusan National University for further analysis. For all sampled specimens (except the two specimens from 2016), the length–weight relationship was calculated using the following equation:

$$\mathcal{W}t = a \times L^b{}\_{\prime}$$

where *Wt* is wet weight (g), *a* is the intercept, *b* is the slope, and *L* is the standard length (*SL* = straight line between snout and tail, mm). A *t*-test was conducted to verify the statistical difference between the measured values (*SL* and *Wt*) of the January and July groups (SPSS, version 12.0, Chicago, IL, USA). The two specimens in July 2016 were

deliberately selected as relatively large fish; hence, they were excluded from the average and statistical analysis (for a simple DNA comparison). liberately selected as relatively large fish; hence, they were excluded from the average and statistical analysis (for a simple DNA comparison).

where *Wt* is wet weight (g), *a* is the intercept, *b* is the slope, and *L* is the standard length (*SL* = straight line between snout and tail, mm). A *t*-test was conducted to verify the statistical difference between the measured values (*SL* and *Wt*) of the January and July groups (SPSS, version 12.0, Chicago, IL, USA). The two specimens in July 2016 were de-

*J. Mar. Sci. Eng.* **2022**, *10*, x FOR PEER REVIEW 3 of 13

**Figure 1.** Map of the study area in Geoje-Hansan Bay, Republic of Korea (South Korea). **Figure 1.** Map of the study area in Geoje-Hansan Bay, Republic of Korea (South Korea).


**Table 1.** Measurements of seahorses at each collection period and the *p*-value of the *t*-test.

#### *t Wt* 0.679 *2.2. Genomic DNA Extraction and NGS Library Construction*

*2.2. Genomic DNA Extraction and NGS Library Construction* The genomic DNA of *H. haema* was extracted after complete homogenization with Tissue LyserдⅡ (Qiagen, Hilden, Germany), by adding tissue lysis buffer to the gut of The genomic DNA of *H. haema* was extracted after complete homogenization with Tissue LyserTM II (Qiagen, Hilden, Germany), by adding tissue lysis buffer to the gut of each of the 16 specimens six times, according to the instructions of the AccuPrep® Genomic DNA Extraction Kit (Bioneer, Daedeok-gu, Korea). ND-1000 (Thermo Scientific, Waltham, MA, USA) was used for the assay and quantification.

each of the 16 specimens six times, according to the instructions of the AccuPrep® Genomic DNA Extraction Kit (Bioneer, Daedeok-gu, Korea). ND-1000 (Thermo Scientific, Waltham, MA, USA) was used for the assay and quantification. The NGS library of the gut contents of *H. haema* was constructed using primers (COIMISQ and NEXCOIMISQ, Table 2) targeting COI in the mitochondrial DNA region, The NGS library of the gut contents of *H. haema* was constructed using primers (COIMISQ and NEXCOIMISQ, Table 2) targeting COI in the mitochondrial DNA region, which was previously used for the diet study of *Dissostichus mawsoni* [47]. The blocking primer (Table 2) was prepared the region of the universal primer COIMISQF1 in the nucleotide sequences of *Amphipoda*, *Copepoda*, *Mysidacea*, and *Isopoda*, known as prey organisms, and modified with a C3 spacer at the 30 end to suppress annealing of the *H. haema* sequence [48]. For the first PCR reaction, a final volume of 25 µL mixed solution

which was previously used for the diet study of *Dissostichus mawsoni* [47]. The blocking

isms, and modified with a C3 spacer at the 3′ end to suppress annealing of the *H. haema* sequence [48]. For the first PCR reaction, a final volume of 25 μL mixed solution consisting of 10 ng of genomic DNA, 100 μM of COIMISQ primers, 5 μL of 100 μM Blocking primer, 2 μL of 10 mM dNTPs (Takara Bio Inc., Kusatsu, Japan), 0.2 μL of Ex Taq Hot Start Version (Takara Bio Inc., Kusatsu, Japan), 2 μL of 10× Ex Taq buffer (Takara Bio Inc., Kusatsu, Japan), 3% DMSO, and distilled water was used. The reaction conditions and process of

consisting of 10 ng of genomic DNA, 100 µM of COIMISQ primers, 5 µL of 100 µM Blocking primer, 2 µL of 10 mM dNTPs (Takara Bio Inc., Kusatsu, Japan), 0.2 µL of Ex Taq Hot Start Version (Takara Bio Inc., Kusatsu, Japan), 2 µL of 10× Ex Taq buffer (Takara Bio Inc., Kusatsu, Japan), 3% DMSO, and distilled water was used. The reaction conditions and process of the first PCR were as follows: initial denaturation at 94 ◦C for 5 min, followed by 94 ◦C, 48 ◦C, and 72 ◦C for 30 s, respectively, repeated for 15 cycles. The final extension was conducted at 72 ◦C for 5 min. Prior to the second PCR reaction, the amplicons were purified using an AccuPrep® PCR Purification Kit (Bioneer, Republic of Korea). For the second PCR reaction, a final volume of 20 µL mixed solution consisting of 5 µL of the purified amplicons, 100 µM of NEXCOIMISQ primers, 2 µL of 100 µM Blocking primer, 2 µL of 10 mM dNTPs (Takara Bio Inc., Kusatsu, Japan), 0.2 µL of Ex Taq Hot Start Version (Takara Bio Inc., Kusatsu, Japan), 2 µL of 10× Ex Taq buffer (Takara Bio lnc., Kusatsu, Japan), and distilled water was used. The reacting conditions were repeated for 18 cycles under the same conditions as that of the first PCR. The amplicons of the second PCR were identified by 1.5% agarose gel electrophoresis (630 bp) and purified by the AccuPrep® PCR Purification Kit. The libraries constructed by the Nextera XT index Kit (Illumina, San Diego, CA, USA) were quantified with a QuantiFluor® Fluorometer (Promega, Madison, WI, USA), before using the MiSeq platform (Illunima, San Diego, CA, USA) for NGS analysis.

**Table 2.** Primers used in this study.


### *2.3. Bioinformatic Analysis*

Some parts of the sequence data of the NGS analysis, which featured a short length (less than 100 bp) and low quality (QV < 20), were disregarded. According to [49], the merged reads (more than 470 bp) constructed with the Mothur software package v1.41.3 were used. After the primer sequences were elided, the OTU (Operational Taxonomic Unit) clustering was conducted with Usearch v8.1.1861 [50] at 97%, and the chimeras were discarded. The OTUs were classified as species (≥99%), genus (<99%, ≥90%), and unknown (<90%) on the NCBI GenBank database. Three specimens (No. 2 of 2017, Nos. 4 and 5 of 2019) were excluded from further analysis because no potential prey organisms were identified. The MEGA X Maximum Likelihood method was used for phylogenetic analysis [51].

### **3. Results**

### *3.1. Morphometric Measurements*

Summer and winter mean surface water temperatures are 24.9 ◦C and 9.3 ◦C, respectively, in the study area. A total of 16 seahorses were captured for this study. Table 1 shows the mean *SL* (mm) and *Wt* (g), as well as the results of the *t*-test for each collection period. Although no significant difference in mean *SL* and *Wt* for each collection period was observed, the July specimens had broader ranges in *SL* and *Wt* than the January specimens in this study. The length–weight relationship of each period's specimens is shown in Figure 2. The *b* values (slopes in Figure 2) of the July and January specimens were 3.237 and 3.034, respectively, which were greater than 3, indicating positive allometric growth.

### *3.2. NGS Analysis*

As a result of the DNA analysis from the guts of 16 individuals, all specimens were identified as *H. haema*. Of the 2,263,757 reads (mean of 174,135 reads per sample) identified by sequencing, 41.87% were identified as seahorse and bacteria. These reads were excluded from further analysis.

Thus, 47.4% of the nucleotide sequences of the analyzed prey organisms were identified to the lowest taxonomic level (species). Among the 12 identified taxa, 10 were arthropods, and the remainder were *Bryozoa* and fish (Table 3). Among the 10 arthropods, four orders (*Harpacticoida*, *Caprella*, *Ianiropsis*, and *Mysida*) showed a relatively high ratio of over 80% in gut contents of some specimens. In July 2016 and 2017, *Caprella* sp. (amphipods) was detected in all specimens, especially for No. 1 (84.35%) and No. 7 (67.37%), whereas *Mysida* was mostly found in two of the total specimens. The *Ianiropsis* sp. (isopods) was the second most common species. On the other hand, in the January 2019 specimens, isopods tended to be dominant in their prey items, with amphipods accounting for a small proportion (Table 3). *J. Mar. Sci. Eng.* **2022**, *10*, x FOR PEER REVIEW 5 of 13

**Figure 2.** Length–weight relationship between specimens collected for each period. The white cir-**Figure 2.** Length–weight relationship between specimens collected for each period. The white circles were taken in July 2017, and the black circles were taken in January 2019.

cles were taken in July 2017, and the black circles were taken in January 2019.

3.2. NGS Analysis **Table 3.** The results of genetic analysis of gut (up to the species level). Bold font indicates the results of further analysis up to the order level for the results showing "unknown" in the analysis up to the species level.

of over 80% in gut contents of some specimens. In July 2016 and 2017, *Caprella* sp. (amphipods) was detected in all specimens, especially for No. 1 (84.35%) and No. 7 (67.37%), whereas *Mysida* was mostly found in two of the total specimens. The *Ianiropsis* sp. (isopods) was the second most common species. On the other hand, in the January 2019 spec-

Since each specimen could be different in their energy budget requirements accord-

As a result of additional analysis of the 52.6% classified as "Unknown", clustering

was confirmed at the order level (Figure 3). Because Figure 3 does not indicate the proportion of gut contents but just all items identified up to the order level, some items with very low percentages (e.g., Algae, Calanoida, Ctenostomatida, and Perciformes) are not shown in Table 3 or Figure 4a. The difference in the gut composition in each collection period was clear in comparison at the order level rather than the species level (Figure 4a). The proportions of Harpacticoida and Amphipoda had high ratios in July 2016 and 2017, respectively, while Isopoda was dominant in January 2019. This result was consistent with

ing to their body size (growth), the seahorse's diet would be different by size in the summer with various sizes of seahorses. However, among the specimens collected in July 2017, the largest seahorse (No. 1, *SL*: 89.4 mm) had a higher proportion of *Caprella* sp. (84.35%) at the species level, while the smallest one (No. 5, *SL*: 35.5 mm) had mostly Amphipoda (96.2%) at the order level (Table 3). In both cases, the Amphipoda order showed a tendency to occupy the highest ratio, although it was not possible to compare them due


the similarity analysis in this study (Figure 4b).

to the difference in the lowest taxonomic level of identification.


**Table 3.** *Cont.*

Since each specimen could be different in their energy budget requirements according to their body size (growth), the seahorse's diet would be different by size in the summer with various sizes of seahorses. However, among the specimens collected in July 2017, the largest seahorse (No. 1, *SL*: 89.4 mm) had a higher proportion of *Caprella* sp. (84.35%) at the species level, while the smallest one (No. 5, *SL*: 35.5 mm) had mostly Amphipoda (96.2%) at the order level (Table 3). In both cases, the Amphipoda order showed a tendency to occupy the highest ratio, although it was not possible to compare them due to the difference in the lowest taxonomic level of identification.

As a result of additional analysis of the 52.6% classified as "Unknown", clustering was confirmed at the order level (Figure 3). Because Figure 3 does not indicate the proportion of gut contents but just all items identified up to the order level, some items with very low percentages (e.g., Algae, Calanoida, Ctenostomatida, and Perciformes) are not shown in Table 3 or Figure 4a. The difference in the gut composition in each collection period was clear in comparison at the order level rather than the species level (Figure 4a). The proportions of Harpacticoida and Amphipoda had high ratios in July 2016 and 2017, respectively, while Isopoda was dominant in January 2019. This result was consistent with the similarity analysis in this study (Figure 4b).

**Figure 3.** Phylogenetic tree of gut contents. **Figure 3.** Phylogenetic tree of gut contents.

**Figure 4.** (**a**) The order-level taxonomic ratio of gut contents analyzed with COI primer. (**b**) Similarity analysis of gut contents by collecting period. **Figure 4.** (**a**) The order-level taxonomic ratio of gut contents analyzed with COI primer. (**b**) Similarity analysis of gut contents by collecting period.

### **4. Discussion**

To compare the seasonal feeding habits of *H. haema* during summer and winter, 14 specimens were collected from Geoje Hansan Bay in July (seven) and January (seven), when the seasonal characteristics were distinct. All the specimens were found in *Sargassum piluliferum* [8] and identified as the same species by DNA analysis [18]. There was no significant difference in mean *SL* and *Wt* by the two different seasons, but there was a relatively wider range in the mean *SL* and *Wt* in the July specimens. In other words, more varied sizes of seahorses were captured in July compared to January. According to previous studies, the breeding season of seahorse is from late spring to autumn in Korea [8,9], and there is a positive correlation between the growth of the genus *Hippocampus* (*H. whitei*, *H. guttulatus*, and *H. zosterae*) and water temperature within a certain range [13,52,53]. Thus, the appearance of seahorses of various sizes in July could be due to the recruitment of newborns and the relatively fast growth rates during the summertime.

### *NGS Analysis*

A total of 12 taxa were identified in all gut contents except for three specimens without genetic information for any prey items. In comparison with other previous studies on the seahorse diet [24,34–42], Bryozoa was detected in a small proportion in this study, which might indicate that bryozoans were fed non-selectively with other prey.

Fish was detected only in one individual in this study. We presumed that other fish or eggs were misidentified due to the lack of NCBI data. *Pictichromis paccagnellae* is a tropical species which does not live in Korean waters. Although it is also assumed to be misidentified, it is likely that the genetic information could be other fish larvae or eggs, as actual feeding larvae have been reported in several different kinds of seahorses such as *H. abdominalis*, *H. reidi*, and *H. trimaculatus* [24,34,38].

Some "unknown" genetic information, which could not be confirmed even at the order level, suggests that there are still insufficient coastal zooplankton DNA library data for a variety of zooplankton in coastal areas.

Like other seahorse species reported previously [3,23,24,26,33–42,54–56], *H. haema* appears to feed mainly on crustaceans, according to the results in this study. However, the gut contents of *H. haema* were clearly differentiated by different sampling periods at the order level as shown in Figure 4a,b. *Caprella* was detected in most specimens in this study regardless of season, but a large proportion was recorded mainly in summer. In contrast, *Ianiropsis* accounted for a large proportion of gut contents in winter, showing a pattern opposite to that of *Caprella*. In fact, although *Caprella* appears in abundance on the coastal areas all year round [57–59], it is most often found in summer due to its strong tolerance to high water temperatures [57]. In comparison, large numbers of *Ianiropsis* are normally observed in winter [60].

Mysida species are generally observed throughout the year in coastal waters, which have an extreme seasonal temperature difference of more than 20 ◦C [61,62]. They stay and/or migrate in several swarms on the sandy bottom, near vegetation and rocks [61]. In general, seahorses forage accessible prey (slow and smaller than their mouths) within vegetation, but they occasionally hunt fast prey such as Mysida and Caridea on sandy bottoms [24,34]. The fact that Mysida was found only in some seahorses in this study indicates that several active individuals actually attempted to hunt them when the Mysida swarm reached their vicinity, but it was not easy for the seahorses to catch them due to the rapid reaction rate of Mysida. This can be supported by the low hunting success rates of *H. haema* toward a Mysida swarm observed in a breeding water tank (unpublished data).

Although Harpacticoida accounted for a large percentage of the specimens in July 2016, they occupied a very small portion for the prey items for *H. haema* in July 2017. Normally, Harpacticoids are a major item for some *Hippocampus* species (*H. zosterae*, [23]; *H. reidi*, [34]; and *H. subelongatus*, [54]). The seahorses can change their feeding capacity as they grow [3,23,24,63]. Indeed, in a laboratory environment, copepods (<1 mm) were eaten by all *H. haema* regardless of their size, whereas Artemia and Mysida (>1 mm) were

mainly eaten by large seahorses (unpublished data). Like *Caprella*, harpacticoids appear frequently in the late spring–summer period [64,65], which coincides with the breeding season of *H. haema*. Therefore, with a maximum size of approximately 1 mm, harpacticoids could be an important food source for growing seahorses with their small mouth sizes.

The coincidence between the gut contents of *H. haema* and the seasonally thriving zooplankton in our study indicates that the main diet of *H. haema* can be modified according to the availability of prey. The authors of [24] found that wild *H. abdominalis* collected from Wellington Harbor did not have a seasonal shift in their habitats, but there was a major seasonal change in their prey. According to a previous study that analyzed feces from *H. guttulatus* using a genetic method, the seahorse diet differs depending on the habitat [42]. Although some differences among species are expected, this ability of *Hippocampus* species to flexibly change their diets in response to spatiotemporal changes in their surroundings could be one of the survival strategies for adapting to dynamic coastal environments.

### **5. Conclusions**

Although specimens were collected once each in summer and winter, differences in feeding habits of *H. haema* could be clearly distinguished using DNA tools. However, due to the lack of genetic information on coastal zooplankton, it was difficult to identify up to the species level what seahorses mainly feed on. If coastal zooplankton can be identified up to the species level and their ecological characteristics understood, it could make a great contribution to understanding the role of the seahorse from an ecological perspective. In conclusion, this study is very important to provide the first information on the feeding behaviors of the Korean seahorse species, which is very valuable for their sustainable management and conservation in Korea.

**Author Contributions:** M.-J.K. and S.-H.L. conceived and designed the experiments; H.-W.K. and S.-R.L. performed the experiments; M.-J.K. and S.-R.L. analyzed the data; S.-H.L. and H.-W.K. validated the results; M.-J.K., N.-Y.K., Y.-J.L. and S.-N.K. investigated; M.-J.K., N.-Y.K. and Y.-J.L. curated the data; M.-J.K. wrote the original draft; M.-J.K. and S.-H.L. reviewed and edited the manuscript; S.-N.K. and H.-T.J. visualized the data; S.-H.L. supervised this research; S.-H.L. and H.-T.J. funded acquisition. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by a 2 year research grant from Pusan National University.

**Institutional Review Board Statement:** Ethical review and approval were waived for this study due to all fish samples were dead when we received them.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Acknowledgments:** The authors would like to thank the anonymous reviewers and the handling editors who dedicated their time to providing the authors with constructive and valuable recommendations.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**

