Triple barcoding for a hyperparasite, its parasitic host, and the host itself: a study of Cyclocotyla bellones (Monogenea) on Ceratothoa parallela (Isopoda) on Boops boops (Teleostei)

Cyclocotyla bellones Otto, 1823 (Diclidophoridae) is a monogenean characterised by an exceptional way of life. It is a hyperparasite that attaches itself to the dorsal face of isopods, themselves parasites in the buccal cavity of fishes. In this study, Cy. bellones was found on Ceratothoa parallela (Otto, 1828), a cymothoid isopod parasite of the sparid fish Boops boops off Algeria in the Mediterranean Sea. We provide, for the first time, molecular barcoding information of a hyperparasitic monogenean, the parasitic crustacean host, and the fish host, with COI sequences.


Introduction
Cyclocotyla bellones Otto, 1823 is a diclidophorid Monogenea, hyperparasite of cymothoid isopods of the buccal cavity of the sparid fish Boops boops (Linnaeus, 1758). It was first described in one of the earliest accounts on monogeneans by Otto (1823), who erected the genus Cyclocotyla Otto, 1823 for Cy. bellones collected from the skin of the dorsal face of the garfish Belone belone (Linnaeus, 1760) off Naples, Italy [62]. It was then frequently reported on isopods parasitic of B. boops and of Spicara spp. (see Table 1), with a single record on cymothoids of the carangid Trachurus trachurus (Linnaeus, 1758) [23].
The identity of the hyperparasite monogenean, the crustacean parasite-host, or the fish host was not confirmed via DNA sequencing in any of these instances. Hence, as part of an ongoing effort to characterize the parasite biodiversity of fishes off the Southern shores of the Mediterranean Sea [3, 6-12, 15-18, 20, 42], molecular characterization of the three members of the hyperparasite-parasite-host association is provided for the first time.
We do not show detailed illustrations or measurements of monogeneans in this paper because these will be provided in a future article; however, we provide a general illustration and a short description.

Materials and methods
Collection and sampling of fishes From 2017 to 2019, 624 specimens of B. boops and 45 Pagellus acarne (Risso, 1827) were collected from fish markets off Réghaia on the Algerian coast or directly from local fishermen in Bouharoun (36°37 0 N, 2°39 0 E). Fish specimens were transferred to the laboratory shortly after capture and identified using keys [26] and examined fresh on the day of purchase. The buccal cavities were carefully examined for isopods. Gill arches were also resected and placed in separate Petri dishes containing filtered sea water. The buccal cavity, isopods and isolated gills were observed under a dissecting microscope for the presence of monogeneans.

Collection of isopods and monogeneans
For the host fish B. boops, all isopods (sometimes with visible diclidophorid monogeneans) were removed from the buccal cavity using dissecting forceps. Monogeneans were isolated from the isopods with a fine dissecting needle. Other diclidophorids (Choricotyle sp.) parasitic on gills of another sparid, Pagellus acarne were removed from the gills using a fine dissecting needle ( Table 2). Note that amongst the polyopisthocotyleans collected on P. acarne, only Choricotyle sp. was included as it is remarkably similar in morphology to species of Cyclocotyla; Choricotyle sp. is a parasite of the fish so no isopod was involved in this particular association.

Morphological methods
Monogeneans were preserved in 70% ethanol, stained with acetic carmine, dehydrated in a graded series of alcohol for 15 min each: (70, 96 and 100%), cleared in clove oil, and finally mounted in Canada balsam. Monogeneans were identified on stained whole mounts. Isopods were identified with the help of Prof. Jean-Pierre Trilles (University of Montpellier, France).

Molecular methods
For complete traceability of the molecular study, special care was taken to ensure that hosts and monogeneans were labelled with respect to host-parasites relationships [3,7,8,10,11,39].
For three individual B. boops, the parasitic female isopod and one monogenean on this individual isopod were extracted ( Table 2). A tissue sample from the gill of the fish was taken and a pereopod was detached from each infected isopod and submitted to molecular analysis. For the monogenean, a small lateral part of the body was separated with a scalpel and submitted to molecular analysis, and the rest of the body was mounted on a slide as a voucher for drawing and deposition in a Museum collection. This ensures that the molecular identification of the host fish, the parasitic isopod and their monogenean parasite correspond perfectly at the individual host and parasite levels, and enable morphological assessment of sequenced monogeneans. Slides of monogeneans were deposited in the Muséum

Molecular barcoding of fish
Total genomic DNA was isolated using a QIAamp DNA Mini Kit (Qiagen, Courtaboeuf, France), according to the manufacturer's instructions. The 5 0 region of the mitochondrial cytochrome c oxidase subunit I (COI) gene was amplified with the primers TelF1 (5 0 -TCGACTAATCAYAAAGAYATYGG-CAC-3 0 ) and TelR1 (5 0 -ACTTCTGGGTGNCCAAARAAT-CARAA-3 0 ) [21]. PCR reactions were performed in 20 lL, containing 1 ng of DNA, 1Â CoralLoad PCR buffer, 3 mM MgCl 2 , 66 lM of each dNTP, 0.15 lM of each primer, and 0.5 units of Taq DNA polymerase (Qiagen). The amplification protocol was 4 min at 94°C, followed by 40 cycles at 94°C for 30 s, 48°C for 40 s, and 72°C for 50 s, with a final extension at 72°C for 7 min. PCR products were purified (Ampure XP Kit, Beckman Coulter, Brea, CA, USA) and sequenced in both directions on a 3730 Â l DNA Analyzer 96-capillary sequencer (Applied Biosystems, Foster City, CA, USA). We used CodonCode Aligner version 3.7.1 software (Codon Code Corporation, Dedham, MA, USA) to edit sequences, compared them to the GenBank database content with BLAST, and deposited them in GenBank (accession numbers in Table 1). Species identification was confirmed with the BOLD identification engine [67].

Molecular barcoding of isopods
Total genomic DNA from a pereopod was isolated using a QIAamp DNA Mini Kit (Qiagen, Courtaboeuf, France), according to the manufacturer's instructions. The 5 0 region of the COI gene was amplified with the "universal" primers LCO1490 (5 0 -GGTCAACAAATCATAAAGATATTGG-3 0 ) and HCO2198 (5 0 -TAAACTTCAGGGTGACCAAAAAAT-CA-3 0 ) [27]. PCR reactions were performed in 20 lL, containing 1 ng of DNA, 1Â CoralLoad PCR buffer, 3 mM MgCl 2 , 66 lM of each dNTP, 0.15 lM of each primer, and 0.5 units of Taq DNA polymerase (Qiagen). The amplification protocol was 4 min at 94°C, followed by 40 cycles at 94°C for 30 s, 48°C for 40 s, and 72°C for 50 s, with a final extension at 72°C for 7 min. Sequences were obtained as for fish and were deposited in GenBank (accession numbers in Table 1).

Molecular barcoding of monogeneans
Total genomic DNA was isolated using a QIAmp DNA Micro Kit (Qiagen). The specific primers JB3 (=COIASmit1) (forward 5 0 -TTTTTTGGGCATCCTGAGGTTTAT-3 0 ) and JB4.5 (=COI-ASmit2) (reverse 5 0 -TAAAGAAAGAACATA-ATGAAAATG-3 0 ) were used to amplify a fragment of 424 bp of the COI gene [13,48]. PCR reactions were performed in 20 lL, containing 1 ng of DNA, 5Â iProof HF buffer, 0.25 mM dNTP, 0.15 lM of each primer, and 0.5 units of iProof HF DNA polymerase (Bio-Rad). Thermocycles consisted of an initial denaturation step at 94°C for 2 min, followed by 37 cycles of denaturation at 94°C for 30 s, annealing at 48°C for 40 s, and extension at 72°C for 50 s. The final extension was conducted at 72°C for 5 min. Sequences were obtained as for fish and were deposited in GenBank (accession numbers in Table 1).

Trees and distances
For fishes, the phylogenetic analyses included three sequences of B. boops and one Pagellus acarne generated in this study, and other sequences of these fishes available in GenBank (Table 3), whilst Spicara maena (Linnaeus, 1758), a sparid previously reported as a host of Cy. bellones [5] was used as an outgroup.
For isopods, the molecular analysis was based upon mouth dwelling cymothoids, mainly species previously reported as hosts of Cy. bellones (Table 4). The body surface isopod Anilocra clupei Williams & Bunkley-Williams, 1986 was used as an outgroup. Table 2. Fishes, Isopoda, Monogenea, and their COI sequences. To ensure full traceability and respect of host-parasite relationships, for Cyclocotyla bellones one monogenean was collected from one parasitic isopod and each fish, isopod and monogenean individuals were sequenced. Choricotyle chrysophryi is a parasite of the fish, so no isopod was involved. All vouchers were deposited in the MNHN. For monogeneans, most sequences of Diclidophoridae available in GenBank were included in the phylogenetic analysis (Table 5), with three sequences of Cy. bellones and one of Choricotyle cf. chrysophryi obtained in the present study. A sequence of Plectanocotyle gurnardi (Van Beneden & Hesse, 1863), a member of Plectanocotylidae Monticelli, 1903, grouped with the Diclidophoridae in a previous phylogeny of Monogenea [38], was used as an outgroup. Molecular analyses were performed in MEGA, version 7 [45]. The trees were inferred using the neighbour joining (NJ) method [69] and the maximum likelihood (ML) method using MEGA7 [45]. Based on the best model, maximum likelihood was used for the best fitting tree according to the Hasegawa-Kishino-Yano with gamma distribution and invariant sites (HKY + G) for fishes; Tamura 3-parameter with gamma distribution and invariant sites (T92 + I) for isopods, and Tamura 3-parameter with gamma distribution (T92 + G) for monogeneans. The robustness of the inferred analysis was assessed using a bootstrap procedure with 1000 replications. Genetic distances, p-distance and Kimura-2 parameter distance (K-2-P), [43] were estimated with MEGA7 and all codon positions were used.
BLAST analyses of the COI sequences of monogeneans, isopods and fishes obtained in the present study were performed using NCBI and BOLD databases [67].
Body elongate, divided into three regions: a tapered anterior region; an enlarged and rounded middle region and a posterior region formed by the haptor. Haptor ovoid, bearing four pairs

Molecular identification of fish
The provisional identification of fish species using morphological characteristics was confirmed by the DNA barcoding approach. The obtained sequences were 652 bp long. BLAST analyses of the COI sequences of the present study with NCBI and BOLD databases showed sequence similarity values of 100% for B. boops and 99.85 % for P. acarne.
The ML tree is shown in Figure 2. The newly generated sequences of B. boops clustered in a well-supported clade (100% bootstrap). All sequences of P. acarne, including our newly generated sequence, clustered in a single robust clade.
Distances were computed using Kimura 2-parameter distance and p-distance. Our sequences of B. boops were identical (0% intraspecific variation). All Sequences of B. boops (available in GenBank plus our newly generated sequences) showed little to no variation (0-1%). The divergence among sequences of P. acarne was also low, and ranged between 0 and 1%.

Molecular information on isopods
Isopods were identified as Ceratothoa parallela (Otto, 1828). The newly acquired sequences were 658 bp long. BLAST analyses of the COI sequences with NCBI and BOLD databases showed sequence similarity values of 80.47-80.63% for "Ceratothoa sp."; however, no sequence identified at the species level was available from the databases.
Phylogenetic trees were constructed based on our newly generated COI sequences of Ceratothoa parallela, and combined datasets of Ceratothoa spp., mainly those previously reported as hosts of C. bellones. The analysis involved 13 nucleotide sequences, and there was a total of 409 positions in the final dataset. The topologies were nearly consistent among the ML (Fig. 3) and the NJ trees. The "mouth dwelling cymothoid clade" was well separated from the outgroup, which included body surface isopods. The mouth dwelling cymothoids were separated into two clades, one including C. oestroides and C. verrucosa, and one including C. parallela, C. collaris and C. oxyrrhynchaena. There was no intraspecific variation among our three sequences of Ceratothoa parallela. The mean intraspecies distance of sequences of Ceratothoa oestroides from three hosts, B. boops, Sparus aurata and Dicentrarchus labrax ranged between 0% and 2%, suggesting that the same species is harboured by the three fish species. Distances between individuals of congeneric species were high, the highest being between Ceratothoa oestroides from B. boops and C. parallela, 28%.

Molecular characterisation of monogeneans
The newly generated sequences were 402 bp long. The COI sequences of Cy. bellones were aligned with other diclidophorid sequences. For trees, the NJ and ML methods led to similar topologies; we show only the ML tree in Figure 4. The analysis involved 13 nucleotide sequences, and there was a total of 337 positions in the final dataset.  The three sequences of Cy. bellones reported in the present study formed a well-supported monophyletic lineage (96 bootstraps in ML, 100 in NJ). The phylogenetic analysis therefore supports Cy. bellones as a distinct species from Choricotyle cf. chrysophryi. The sequence of Choricotyle cf. chrysophryi nested within a Cyclocotyla + Choricotyle clade, but Choricotyle was paraphyletic. However, support values for most clades were low and this phylogeny is not discussed further.

Cymothoid isopods
The COI divergences observed between the newly generated sequences of Ceratothoa parallela sampled from B. boops ranged from 0 to 2% and did not exceed the 3% and 5% COI thresholds suggested for species-level divergences [34,35,70], confirming the morphological identification of the three specimens as conspecific. These findings can further be strengthened by additional COI sequences of this species from various localities and hosts.
Overall, the Cymothoidae are taxonomically challenging as descriptions of many species were originally based only on the morphology of few or sometimes single specimens, thus providing no information on polymorphism [76]. In addition, many misidentifications and incorrect data on species were generated when not considering the morphological variability ascribed to the parasitic lifestyle of cymothoids, polymorphism and sister species, as well as the variations in attachment site and the morphological adaptations [29,72,76]. In recent years, numerous genera and species described early were revised, eliminating some of the uncertainties and confusion [30,72,76]. Hence, every effort should be made to generate molecular data and deposit specimens in collections, which can be used subsequently.
Ceratothoa is a large genus, with 25 species presently accepted [31]. The species studied here, Ceratothoa parallela, was designated in 2015 as the type-species of the genus [55]. Unfortunately, not many COI sequences are available for members of this genus and of the family of Cymothoidae overall, and almost half are from unpublished works. COI sequences are known for 16 genera [33,36,41,57,[77][78][79][80]. 16S sequences are available for 15 genera [33,37,41,78,80]. 18S sequences are known for 4 genera (all from unpublished works) and 28S sequences are available for only 3 genera [33]. The complete mitochondrial genome is known for 2 genera of Cymothoidae from unpublished works, available only in GenBank

Cyclocotyla bellones
Genetic variations among our newly generated sequences of Cy. bellones were very low, 0-1%, a divergence lower than interspecific distances found in other studies of polyopisthocotylean monogeneans (1% vs. 10.2-15.0%; [7]). The divergence between sequences of Cy. bellones and Choricotyle cf. chrysophryi from the sparid P. acarne reached 18% and 16% in K-2-P and p distance, respectively, well above the 3% species threshold often admitted for Monogenea [1], and thus the separation between the two species is confirmed genetically.
In the present study, Cy. bellones was most exclusively on the isopods, mainly on the upper part of the pereon. These results agree with Euzet & Trilles (1961) who found Cy. bellones often attached to the crustacean (telson, pleon, rarely perion) and exceptionally in the buccal cavity of the fish, the palate and the internal edge of the upper lip [25]. We found a single specimen unattached in a Petri dish containing gills and isopods. We carefully examined the gills of B. boops during our four-year survey [6,8,11], but we have not observed any Cy. bellones attached to gills nor in the buccal cavity. Hence, we are confident that this monogenean attaches itself to the isopod and not the fish. Cyclocotyla bellones has been recorded on several hosts and from different localities (Table 1). It has been reported on cymothoids of the oral cavity of four fish species, B. boops, Spicara maena (Linnaeus, 1758), S. smaris (Linnaeus, 1758) and T. trachurus. Distributions of the previously mentioned host species overlap as all hosts co-exist in the Mediterranean Sea or in the Eastern Atlantic Ocean. The parasitic isopods have a larger distribution extending to the northern Indo-pacific and coexist with other fish hosts in the Black Sea, Mediterranean and Eastern Atlantic. Therefore, while we took a conservative position and tentatively identified the diclidophorid from cymothoids on B. boops as Cy. bellones, it is not unlikely that this hyperparasite is a species complex.

Conflict of interest
The Editor-in-Chief of Parasite is one of the authors of this manuscript. COPE (Committee on Publication Ethics, http:// publicationethics.org), to which Parasite adheres, advises special treatment in these cases. In this case, the peer-review process was handled by an Invited Editor, Jérôme Depaquit.