Morphological and molecular characterisation of Tristoma integrum Diesing, 1850 (Monogenea, Capsalidae), including its complete mitogenome

Capsalids are monopisthocotylean monogenean parasites found on the skin and gills of fish. Capsalines (subfamily Capsalinae) are large-sized capsalids, parasitic on highly prized gamefish, and species of Tristoma parasitise only the gills of swordfish (Xiphias gladius). We obtained specimens of Tristoma integrum Diesing, 1850 from swordfish collected off Algeria in the Mediterranean Sea. Here, we describe the specimens, including the key systematics characters of dorsolateral body sclerites. One specimen was used for a next generation sequencing analysis but a part of it, including the sclerites, was mounted on a permanent slide, drawn, and deposited in a curated collection. We characterised the complete mitogenome, the ribosomal cluster (including 18S and 28S) and additional genes such as Elongation factor 1 alpha (EF1α) and Histone 3. We also retrieved molecular information from the host tissue present in the gut of the monogenean and provide the sequence of the complete rRNA cluster of the host, X. gladius. The mitogenome of T. integrum is 13 968 bp in length and codes for 12 protein, 2 rRNA and 22 tRNA. Phylogenies of capsalids were generated from 28S sequences and concatenated mitochondrial protein-coding genes, respectively. In the 28S phylogeny, most subfamilies based on morphology were not found to be monophyletic, but the Capsalinae were monophyletic. In both phylogenies, the closest member to Tristoma spp. was a member of the Capsaloides. In an Appendix, we report the complex nomenclatural history of Tristoma Cuvier, 1817 and its species.


Introduction
The Capsalidae are a family of monopisthocotylean monogeneans, characterised by a size larger than most other monopisthocotyleans, with several species over 1 cm in length, and a typical morphology with a large discoid haptor and two prominent anterior organs [59]. Among the Capsalidae, the subfamily Capsalinae includes about 60 species that parasitise the skin and gills of gamefish [18]. Chisholm & Whittington (2007) summarised the challenges of the modern taxonomist studying these parasites "First, over a third of the currently recognised capsaline species were described during the 1800s. Thus, not only can the original descriptions be difficult to obtain, but often no type material to verify these descriptions exists. Second, because capsalines are usually large and obvious parasites on cosmopolitan gamefish species, many parasitologists have collected them worldwide. Therefore, some capsalines from the same host species have been described as different species when collected at different localities globally. In these instances, it is highly likely that these capsaline species are synonymous" [18]. However, the large size of the capsalines is currently an advantage for the modern taxonomist because they provide enough DNA to allow molecular studies with routine methods, and are even large enough so that only a part of the body can be used (and destroyed) for molecular purposes, while the rest of the body can be kept on a permanent slide for morphological work and be deposited as a voucher (hologenophore) in a curated collection.
One of the genera of the Capsalinae, Tristoma Cuvier, 1817, was described as early as 1817 by Cuvier [19] with the description of Tristoma coccineum Cuvier, 1817. Diesing (1835) published a monograph on Tristoma, which reported four species, with only one of them (T. coccineum) still currently included in the genus [22,23]. Another species, Tristoma integrum Diesing, 1850, was described later [24]. More than a century later, in 1968, Yamaguti [60] described two additional species, which Chisholm & Whittington (2007) considered as probably synonymous with the species above [18]. All species of Tristoma are parasites on the gills of the swordfish, Xiphias gladius, and are characterised by dorsomarginal sclerites in distinct transverse rows [18].
Because some capsalids have a pathological effect on fish of economic importance, and probably also because their large size makes it possible to obtain molecular data with relative ease, there are a number of sequences available [48], including several recently obtained complete mitogenomes [5,61,63,65]. However, prior to this study, the genetic references for Tristoma were rather scarce. They consisted of eight sequences representing three partial genes: 28S, Elongation factor 1 alpha (EF1a) and Histone 3 [45,48].
In a previous paper, we emphasised that many mitogenomes have been described, unfortunately, without any associated morphological study nor deposit of voucher material in a collection ( Table 3 in Ayadi et al., 2022 [2]). The aim of this work was to provide morphological arguments for correct identification of our material and associate it with as much molecular information as possible, including on the host.
In this study, we redescribe Tristoma integrum based on material collected in the Mediterranean Sea off Algeria, and we characterise its complete mitogenome, the complete ribosomal gene cluster including 28S and 18S, and additional genes including Elongation factor 1 alpha (EF1a) and Histone 3. The molecular work was performed on a specimen for which a lateral part of the body, showing the characteristic lateral sclerites, was mounted on a permanent slide and deposited in a museum collection, therefore allowing traceability; the sclerites were drawn to ascertain the specific identity. In addition, we could retrieve the DNA from the host from the digestive tract of the monogenean, which was used for the molecular analysis, therefore ascertaining the host identity with molecular tools.
In an Appendix, we provide details of the historical account of Tristoma and Tristoma integrum, their intricate nomenclature, and our chase for the original description of the species.

Collection and sampling of fishes
From 2017 to 2018, gills of 10 dead specimens of Xiphias gladius were collected directly from local fishermen in Bouharoun (36,370 N, 2,390 E) and Cap Djinet (36,877 N, 3,720 E). Fish specimens were processed shortly after capture and morphologically identified in the field using keys [28]. Gills were put separately in isolating bags, transferred to the laboratory in an ice bag, and examined at the laboratory on same day. The gills were cut into several parts, placed in Petri dishes in seawater, and observed under a stereomicroscope (Carl Zeiss™ Stemi™ DV4 Stereomicroscope, Oberkochen, Germany) for monogeneans. Synonyms, and scientific and common names of fishes are those provided in WoRMS [58] and FishBase, respectively [29].

Morphological methods for monogeneans
Monogeneans were removed from the gills using dissecting forceps. Monogeneans were heat-killed with freshwater then preserved in 70% ethanol, stained with acetic carmine, dehydrated in a graded series of alcohol (70, 96 and 100%) for 15 min each, cleared in clove oil, and finally mounted in Canada balsam. Monogeneans were identified on stained whole mounts.
Whole stained mounts were photographed using a Zeiss microscope equipped with camera.
Drawings were made with a Nikon Eclipse i80 microscope with DIC (differential interference contrast) and a drawing tube. Drawings were scanned and redrawn on a computer with Adobe Illustrator. The sclerotised structures (dorsomarginal spines and hamuli) were measured according to Figure 1. Measurements are in micrometres, and indicated as means and between parentheses, the range and number of measurements. The nomenclature of internal anatomy and sclerotised marginal body structures used by Barse & Bullard (2012) for Capsala laevis (Verrill, 1875) is adopted here [7].

Deposit of specimens
Voucher specimens of monogeneans were deposited in the Swedish Museum of Natural History, Stockholm, Sweden (Naturhistoriska riksmuseet), under registration numbers SMNH 209003-209011. Fish specimens were not deposited.

Traceability of specimens used for 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-parasite relationships [1-4, 11-15, 17, 33]. For one monogenean specimen (SMNH 209003), a lateral part of the body containing marginal spines was separated with a scalpel and mounted on a permanent slide and deposited as SMNH 209003, and the rest of the body was submitted to molecular analysis (and thus destroyed); this specimen is therefore a hologenophore according to Pleijel et al. (2008) [49].
Next generation sequencing, assembly, datamining and annotation The SMNH 209003 specimen was sent to the Beijing Genomics Institute sin Shenzhen, China. The institute took care of the DNA extraction and performed sequencing on a DNBseq platform, from which ca. 40 million clean 150 bp paired-end reads were obtained. Reads were deposited on Sequence Read Archive (SRA) and are available as part of the BioProject PRJNA947900 (https://www.ncbi.nlm.nih.gov/sra/ PRJNA947900). Assemblies were performed with SPAdes 3.15.5 [6] with two different k-mer of 85 and 125. Contigs of interest were extracted by data mining using blastn commandline [9], using as references the sequences indicated in Table 1. After circularisation and trimming, the mitogenome was annotated with the help of MITOS [8] and manually curated.
The position of the tRNA was verified using Arwen, with the -gcflatworm option [39]. The map of the mitogenome was obtained from the OGDRAW portal [40]. The annotation of the clusters of nuclear rRNA genes for both the parasite and its host were done with the help of Rfam [34].

Phylogeny
Two different phylogenies were performed. The first one was a multigene phylogeny performed on concatenated amino acid sequences of the mitochondrial proteins from the available mitogenomes of Capsalidae. The second was based on the dataset of partial 28S gene sequences from Perkins et al. [48] and Mollaret et al. [45]. For the multigene phylogeny, protein sequences were extracted and aligned separately using MAFFT 7 [36], trimmed using trimAl with the -automated1 option [16], and alignments were then concatenated using Phyutility 2.7.1 [53]. For the 28S inferred phylogeny, partial genes were aligned with the complete gene from T. integrum SMNH 209003 using the Clustal function of MEGA X [38] and trimmed manually at the endings. Evolutionary models were chosen according to ModelTest-NG v0.1.7 with default parameters [20]. All phylogenies were conducted using IQ-TREE 2.2.0 [43] with 1 000 ultrafast bootstrap replications. Type-host: Xiphias gladius, the Swordfish (Xiphiidae) [24]. Possible additional hosts: Tetrapturus belone Rafinesque (Istiophoridae), the Mediterranean spearfish [47]; Mola mola (Linnaeus) (junior synonym of Mola rotunda Cuvier) (Molidae) [47].
Specimens from Algeria, from the gills of Xiphias gladius. Vouchers deposited in the collection of the Swedish Museum of Natural History, Stockholm, Sweden (Naturhistoriska riksmuseet), SMNH 209003-209011. Hologenophore [49], lateral part showing marginal spines mounted on slide, the rest of body part cut off and used for molecular analysis, SMNH 209003.
Site on host: Gills. Description of specimens from Xiphias gladius from Algeria. Based on 10 specimens (9 whole specimens + hologenophore). Measurements in Table 2.
Body oval to rounded (Fig. 2), armed dorsally on its lateral margins with numerous transverse rows of marginal spines; each containing from 2 to 4 spines. Spines somewhat similar, 3 to 13 cuspids per spine (Fig. 1). Ventral body surface lacking papillae on most of its surface, but posterior body part covered ventrally with visible prominent papillae. Haptor circular, sucker-like, with a visible marginal membrane and a pair of haptoral anchors (Fig. 1). Haptor divided by seven radial weak septae into seven marginal more or less triangular loculi and a central one. Prohaptoral suckers oval to elliptical. Testes varied in shape and dimensions (Fig. 4E), confined to area between intestinal crura. Vas deferens ventral to ovary, originating somewhere at level of anterior testes (Fig. 4E), passing anteriorly on marginal part of ovary; folding anteriorly to vitelline reservoir then strongly winding before entering cirrus sac and uniting with prostatic duct (Figs. 4A-4C). Cirrus sac oval and elongated (Figs. 4A, 4C, 4D), thick walled with visible longitudinal muscles) (Fig. 3). Cirrus sac containing cirrus and accessory gland reservoir (prostatic reservoir) (Fig. 4D). Accessory gland reservoir (prostatic reservoir) fusiform, located at the base of cirrus pouch (Fig. 3). Cirrus cylindrical, muscular, thick walled, papillate for or half of its length (Fig. 3); ejaculatory duct thin, well visible (Fig. 3). Genital pore ventral, at midlevel of pharynx. Ovary rosette Seminal receptacle rounded to oval (Figs. 4B, 4E), immediately posterior to proximal portion of vaginal duct, connected to vitelline reservoir by a narrow-convoluted duct (Fig. 3). Vaginal opening dextral, ventral, posterior to male genital pore, with a visible glandular wall. Vaginal wall with several layers. Vitellarium including vitelline follicles and two sorts of ducts: lateral ducts and transverse vitelline ducts (Fig. 4E). Vitelline follicles extensive, abundantly developed along intestinal branches and occurring as well in spaces between testes. Transverse vitelline ducts large; lateral ducts smaller, reaching lateral body margin. Vitelline reservoir rounded (Fig. 4E), well visible, posterior to transverse loop of vas deferens (Fig. 3). Ovovitelline duct and eggs not observed.

The mitogenome of Tristoma integrum
A 14 003 bp long contig with a coverage of 64.7Â was retrieved after assembly (k-mer 125). Its AT-rich extremities showed redundancies, but also contained repeated sequences. It was trimmed to a circular contig of 13 968 bp, but due to the aforementioned presence of repetitions, this size should be taken with care. The mitogenome (GenBank Accession Number OQ355698) contains 12 protein-coding genes, 2 rRNA and 22 tRNA (Fig. 5). It was not possible to find a stop codon for the cox1 gene, which would overlap tRNA-Thr by 9 bp if terminated by a natural stop and showed no obvious clue of a premature termination by A residues to the mRNA. The cox2 gene is terminated by the presence of tRNA-Glu and the addition of A residues to the mRNA.

Multigene phylogeny and comparison with the mitogenomes of other Capsalidae
The best model of evolution returned by ModelTest-NG was the MTZOA+G4+F model. The multigene phylogeny associated T. integrum with Capsaloides cristatus Yamaguti, 1968 with maximum support at the nodes (Fig. 6). Both species were comprised of a bigger cluster that contained Capsala martinieri Bosc, 1811, Capsala pricei Hidalgo-Escalente, 1958 and Capsala katsuwoni (Ishii, 1936) Price, 1938; all these are members of the Capsalinae.
In Table 3, we compare the different features observed in the mitogenomes of the species belonging to the cluster highlighted in the phylogeny. It is worth noting that T. integrum and C. cristatus, which appear as sister species in the phylogeny, are also the only taxa sharing a putative overlap, of similar length, between cox1 and tRNA-Thr. However, possibly for this reason, the annotation of the cox1 gene of C. cristatus was noted as unverified in GenBank. The overlap between ND4L and ND4 seems to be a constant feature. Neobenedenia melleni (MacCallum, 1927) Yamaguti, 1963, which appears slightly outside the cluster mentioned above but clearly distinct from Benedenia spp., also displays this 28 bp ND4L/ND4 overlap [65]. Conversely, neither Benedenia hoshinai Ogawa, 1984 [35] nor any of the specimens of Benedenia seriolae (Yamaguti, 1934) Meserve, 1938 [37,48] harbour any such overlap. However, there is an intergenic sequence of ca. 118 bp between ND4L and ND4 in specimens of B. seriolae and this intergenic sequence is 282 bp in length in B. hoshinai.
The cluster of ribosomal RNA genes of Tristoma integrum and 28S inferred phylogeny A 6 273 bp long contig containing the cluster of rRNA genes was found with a high coverage of 805.91x. After verification on Rfam and trimming, the complete cluster was 5 839 bp long (GenBank: OQ349751), distributed as 1 985 bp (18S), 462 bp (ITS1), 153 bp (5.8S), 332 bp (ITS2), and 2 907 bp (28S). The 28S gene was compared with the partial 28S genes already available in GenBank for T. integrum. Our sequence of Tristoma integrum was found to be 99.02% identical with FJ972015 (409 bp from X. gladius collected off Italy [48]) and only 96.66% with AF131715 (329 bp from X. gladius collected off France [45]). However, in the case of AF131715, most differences were found in the 3 0 extremity of the sequence and were not shared with FJ972015, and are likely to be errors from early (2000) manual Sanger sequencing. After trimming of the dubious 3 ending, the percentage of identity rose to 99.68% between T. integrum SMNH 209003 and AF131715, and 98.80% with FJ972015. It is worth noting that all polymorphisms were found in the highly variable D1 divergent domain.
For the partial 28S inferred phylogeny, the best model of evolution returned by ModelTest-NG was the TVM+I+G4 model. Tristoma appeared non monophyletic since some members of Capsaloides were nested within its clade. The phylogeny reflected the polymorphisms detected in the 28S gene between T. integrum SMNH 209003 and other specimens of the genus. Sister-groups to the Tristoma clade were several members of Capsala and Nasicola klawei (Stunkard, 1962) Yamaguti, 1968 (Fig. 7). All these are members of the Capsalinae.

Other markers available for members of the genus Tristoma
The complete histone 3 and elongation factor 1 genes were also obtained from T. integrum SMNH 209003, based on the assembly made with k-mer 85.
The intronless histone 3 gene consisted of a 2 695 bp contig with a coverage of 70.4x. The gene itself, as deposited in GenBank (OQ354712), is 411 bp long. A megablast query on the whole gene returned as best result a partial 258 bp fragment obtained from Pseudonitzschia uku Yamaguti, 1965 (FJ972129). Manual alignment with MEGA X of the complete gene with the partial 148 bp fragment obtained by Perkins et al. [48] from T. integrum (FJ972131) also displayed 100% identity.
The EF1a gene was found inside a 1 693 bp contig (GenBank: OQ349179) with a coverage of 4.8x. The gene itself was 1 476 bp long, and contained 2 group I introns (39 bp and 72 bp), leading to a 1362 bp complete coding sequence. A megablast query returned as best result a partial 492 bp long fragment obtained from T. coccineum Cuvier, 1817 (FJ972071), with 98.18% identity. Tristoma coccineum was the only species of the genus Tristoma for which this gene was available in GenBank.

Host DNA and identification
From the pool of contigs obtained after assembly with k-mer 125, it was possible to retrieve an 11 425 bp contig with a coverage of 29.35x, which contained the complete cluster of rRNA of the host, X. gladius. After verification with Rfam and trimming, the complete cluster was 6199 bp long (GenBank: OQ349752), distributed as 1 838 bp (18S), 453 bp (ITS1),   host's mitochondrial genome. However, when datamining the results obtained with k-mer 85, 26 contigs accounting for 10 568 bp were found, with an average coverage of 3.7x. By merging several of these contigs, it was possible to obtain a 1 274 bp partial cox1 gene. Megablast query returned 99.29% identity with two full mitogenomes of X. gladius (AB470301 and AP006036) and 99.21% identity with two complete cox1 genes of X. gladius (HM071014 and GQ202122).

Identification of our specimens as Tristoma integrum
Given the organisation of the haptor, testes and ovary, our specimens are members of Capsalinae Johnston, 1929. By the unbifid posterior rays of the haptor and the shape of the dorsomarginal spines, they are members of Tristoma Cuvier, 1817  [51]. Currently, Tristoma includes four valid species all described from Xiphias gladius: T. coccineum Cuvier, 1817; T. integrum Diesing, 1850; T. adcoccineum Yamaguti, 1968 and T. adintegrum Yamaguti, 1968 [19, 24, 60]; however, we follow Chisholm & Whittington (2007) [18] and consider that the two species described by Yamaguti are probably junior synonyms. Euzet & Quignard (1961) remarked that the oncomiracidium was identical in both species T. integrum and T. coccineum [27]. They questioned the duality of the species in Tristoma and hypothesised that the morphological differences between the adults of the two species could be phenotypical differences due to their different positions on the gills. The availability of sequences from both species, which are different [48], now contradicts this hypothesis.
Tristoma integrum and T. coccineum are morphologically similar but can be distinguished by the number and organisation of spines and cuspids per spines, according to Price (1939) [51]. The rows of spines on T. integrum are numerous, while there are fewer rows of spines on T. coccineum. Additionally, in T. integrum the spines are similar in terms of cuspids per spine, whereas in T. coccineum the spines are dissimilar, especially with the most median spines having only one cuspid, and the outermost ones having 10 or more cuspids and the other having 2 to 7 cuspids [51]. Our specimens correspond well to T. integrum (Table 2), but we noted that spines had a slightly higher number of cuspids than the specimens described by Price. The description by Price [51] was on only two specimens.
Several hosts for Tristoma spp. have been mentioned in addition to X. gladiusthese probably need verification, possibly with molecular tools.

Molecular results
Relevance of the protein multigene for the phylogeny of the Capsalidae The tree (Fig. 6) shows a robust branch which includes only members of the Capsalinae. The Capsalinae includes only 4 genera (Capsala, Capsaloides, Tristoma and Nasicola) [18,48] so this tree, which includes 3 of the 4 genera, may be considered as providing significant information for this subfamily. The tree clearly shows the monophyly of the Capsalinae. Within this branch, Capsala is not monophyletic, with Capsala katsuwoni placed as an outgroup to all other capsaline. The rest of the species included in the tree are all members of the Benedeniinae, and the tree fails to recover the monophyly of this subfamily, with Neobenedenia melleni placed as sister-group of the Capsalinae. It should be noted that the tree includes members of only two subfamilies among nine recognised in traditional classifications (Perkins et al., 2009, their Table 1 [48]), and that the interest of our results is therefore limited. However, Perkins et al. (2009), in a molecular study based on various markers, noted that the single subfamily they could retrieve as monophyletic was the Capsalinae [48]; we therefore confirm their conclusion.

Relevance of the 28S tree for the phylogeny of the Capsalidae
The tree (Fig. 7) includes many more taxa than the protein tree. A major result is the monophyly of the Capsalinae, with representatives of all four genera of the subfamily (Capsala, Capsaloides, Tristoma and Nasicola) united together in a strong monophyletic clade. For all other subfamilies, results are much less in accordance with traditional taxonomy. A clade containing a Nitzschiinae and two Entobdellinae (Listrocephalos spp.) is sister-group to the Capsalinae. Three other branches appear to have robust support, but each of them is a mix of Trochopodinae, Benedeniinae and Entobdellinae, therefore precluding any comparison with phylogenies and classifications based on morphology. Again, this reminds us of the results of Perkins et al. (2009) [48] on a smaller taxonomical sampling, who concluded that they could only retrieve monophyly for the Capsalinae. We note that our limited results based on mitogenome proteins also found the Capsalinae monophyletic. The Capsalinae are all parasites of gamefish [18,48] and it is likely that they have evolved within this group of pelagic, high-speed animals.

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 Associate Editor, Jérôme Depaquit.
Acknowledgements. This work was funded by ISYEB, MNHN,  Dr. Helmut Sattmann commented "I tend to assume that the museum owns some originals and those were provided for scientific publication purposes that time. e.g., T. coccineum, published by Bremser 1824, is present in the museum's collection. My conclusion is, that T. coccineum is in the museum's collection, because it was used for publication, but T. integrum is in the national library, because it was not used for publication (neither by Bremser nor by Diesing)." The drawing by Joseph Zehner now digitised in the collections of the Austrian National Library has probably never been seen by parasitologists since the middle of the nineteenth century. We are showing it here (Fig. A1).
In view of the above, it is somewhat surprising, for a taxonomist accustomed to modern editions of the International Code of Zoological Nomenclature, that the species T. integrum could be considered available. The reason for this is certainly the complex reasoning made by Price (1936Price ( , 1939. Price (1936, summary of his thesis; 1939, full publication [50,51]) re-evaluated several species described in the nineteenth century, including Capsala martinieri, for which he provided a copy of the original description in French and of the drawings by Lamartinière (1797) and a description of new specimens [51]. For Tristoma, he considered that T. coccineum was valid. He discussed various interpretations of T. integrum and T. coccineum by several authors (Blanchard, Risso, Diesing, and Taschenberg) and finally concluded (Price, 1939, page 86 [51]) that "the species described by Taschenberg as T. coccineum must be regarded as a separate species and take the oldest available synonym which is Tristoma integrum Diesing". Sproston (1946) noted for T. integrum "Diesing, 1850, 429 (apparently a MS. Name, as synonym of T. coccineum Cuv.; Price, 1936, 12: revives the species name of Diesing)", and finally noted "the validity of Tristoma spp. is in great need for revision".
The Code [32] differentiates works before 1961 from those published during and after 1961: "Article 11.6. Publication as a synonym. A name which when first published in an available work was treated as a junior synonym of a name then used as valid is not thereby made available. 11.6.1. However, if such a name published as a junior synonym had been treated before 1961 as an available name and either adopted as the name of a taxon or treated as a senior homonym, it is made available thereby but dates from its first publication as a synonym". From the above, we conclude that Price (1936 [50]) acted as first reviser (Article 24 of the ICZN [32]) and that his decision to use a name first proposed as synonym is acceptable (Article 11.6.1). Finally, Tristoma integrum Diesing, 1850 is available according to the code. We therefore use Price (1939) for distinguishing the two species T. integrum and T. coccineum.
There is some irony here in that the figure made by Joseph Zehner in 1845 and reproduced here as our Figure A1, which cost us and several helpful colleagues some efforts to locate, has no relevance for systematics or nomenclature, as the decisions by Price were solely on the name and not on his examination of the figure.