Paraharmotrema karinganiense n. gen., n. sp. (Digenea: Liolopidae) infecting the intestine of serrated hinged terrapin (Pelusios sinuatus), east African black mud turtle (Pelusios subniger), and South African helmeted turtle (Pelomedusa galeata) and a phylogenetic hypothesis for liolopid genera

We herein describe Paraharmotrema karinganiense n. gen., n. sp. Dutton & Bullard (Liolopidae Dollfus, 1934) from specimens infecting the intestine of the serrated hinged terrapin (Pelusios sinuatus), east African black mud turtle (Pelusios subniger) (both Nwanedzi River, Mozambique), and South African helmeted terrapin (Pelomedusa galeata) (North-western Zululand, KwaZulu-Natal Province, South Africa). The new genus can be easily differentiated from the other accepted liolopid genera (Liolope Cohn, 1902; Helicotrema Odhner, 1912; Harmotrema Nicoll, 1914; Dracovermis Brooks & Overstreet, 1978) by the combination of having a linguliform body approximately 6–9 × longer than wide, tegumental spines/scales, a minute ventral sucker located in the anterior 1/7−1/8 of the body, deeply lobed testes that are transverse and abut the caeca (spanning the intercaecal space), a uterus that is lateral to the anterior testis (not ventral to the anterior testis), a lobed ovary that is dextral and nearest the posterior testis, and a vitellarium that does not extend anteriad to the level of the ventral sucker and that does not fill the intercaecal space. Nucleotide sequences of large subunit ribosomal DNA (28S) and internal transcribed space region (ITS2) from all analyzed specimens of the new species were identical, respectively; the 28S sequences differed from that of Liolope copulans Cohn, 1902 and from that of Harmotrema laticaudae Yamaguti, 1933 by 103 (8%) and 105 (8%) nucleotides, respectively. The 28S phylogenetic analysis recovered the new genus sister to a clade comprising L. copulans and H. laticaudae. A key to liolopid genera is provided herein. The present study comprises the first nucleotide-based phylogenetic placement of Harmotrema and first record of a liolopid from South Africa or Mozambique. It is the first proposal of a new liolopid genus in 43 yrs, and it documents a second liolopid genus from P. subniger while tripling the number of liolopid turtle hosts reported from the continent of Africa.


Introduction
Species of the seldom reported Liolopidae Dollfus, 1934 comprise 13 nominal species assigned to four genera (two species of Liolope Cohn, 1902;three of Helicotrema Odhner, 1912;four of Harmotrema Nicoll, 1914; four of Dracovermis Brooks and Overstreet, 1978; see also Niewiadomska, 2002) that collectively mature in the lumen of the stomach and intestine of ectothermic tetrapods (Table 1; Table 2). Infections have been reported from all continents but Europe, Australia, and Antarctica. During a recent parasitological expedition led by LdP and including HRD and SAB in South Africa and Mozambique, we collected specimens of a rather large trematode from the intestinal lumen of 3 turtles (Pelusios sinuatus (Smith, 1838), Pelusios subniger (Bonnaterre, 1789), Pelomedusa galeata (Schoepff, 1792)). We herein describe these trematode specimens as a new species, propose a new genus for this new species, and present a phylogenetic hypothesis for Liolopidae based on sequences of Table 1 Records for Liolopidae (type species in bold; cercarial infections*).

Specimen collection and preparation
Turtles were sampled during March 2020 from Karingani Game Reserve (KGR), Maputo province, Mozambique (24 • 20 ′ 8.09 ′′ S 32 • 15 ′ 42.0 ′′ E) and a roadside borrow pit filled with water in northwestern Zululand, Kwa-ZuluNatal Province, South Africa (SA) (27 • 00 ′ 52.8 ′′ S 32 • 08 ′ 30.1 ′′ E). The digestive tract of each turtle was excised intact, sliced longitudinally to expose the lumen, immersed in saline, and examined with stereo-dissecting microscopes each equipped with a fiber optic light source. Trematodes intended for morphology were observed microscopically, heat-killed on glass slides using a butane hand lighter under no coverslip pressure, fixed in 10% neutral buffered formalin (nbf), rinsed with water, stained in Van Cleave's hematoxylin with several drops of Ehrlich's hematoxylin, dehydrated through a graded series of EtOHs, made basic at 70% EtOH, exposed to a few drops of lithium carbonate and butyl-amine to ensure specimens are basic, dehydrated in absolute EtOH and xylene, cleared with clove oil, and permanently mounted on glass slides using Canada balsam . The resulting whole mounts were examined and illustrated with the aid of an Olympus BX53 with DIC (Tokyo, Japan) with drawing tube and a Ken-A-Vision X1000 Micro-projector (Raytown, MO, USA). Measurements were obtained with a calibrated ocular micrometer (as straight-lines along the course of each duct) and are reported in micrometres (μm) as the range followed by the mean, +/− standard deviation, and sample size in parentheses. Two specimens were processed for scanning electron microscopy (SEM) as per Bullard et al. (2019). Type and voucher materials of the new species were deposited in the National Museum of Natural History's Invertebrate Zoology Collection (USNM, Smithsonian Institution, Washington, D. C.). Turtle scientific and common names follow Rhodin et al. (2017), and higher classification of turtles follows Pereira et al. (2017).
A black-banded sea krait (Laticauda semifasciata (Reinwardt, 1837) (Serpentes: Elapidae)) was collected from Sakieda (Ishigaki Island, Okinawa Prefecture, Japan) on 8 September 2016 by Takahide Sasai, maintained alive in the Suma Aqualife Park, and necropsied there on 19 September 2016 by MU. Five adult specimens of H. laticaudae were collected from the intestine of that individual sea krait: one adult specimen of H. laticaudae was fixed in 95% EtOH and processed for DNA extraction and sequencing; 3 were stained in Heidenhain's iron hematoxylin, and 1 was retained in 90% EtOH.

Phylogenetic analysis
Total genomic DNA (gDNA) was extracted from 3 EtOH-preserved and microscopically identified specimens of the new species (2 adults of the new species from P. galeata in SA; 1 juvenile specimen of the new species from P. subniger in KGR) and from the EtOH-preserved and microscopically identified specimen of H. laticaudae from L. semifasciata using DNeasy™ Blood and Tissue Kit (Qiagen, Valencia, California) as per the manufacturer's protocol except that the proteinase-K incubation period was extended overnight, and 100 μL of elution buffer was used to increase the final DNA concentration. The nuclear large subunit ribosomal DNA (28S) and the internal transcribed spacer-2 region (ITS2) were amplified using the primer set of Orélis-Ribeiro et al. (2017). PCR amplifications were performed according to Dutton et al. (2019). DNA sequencing was performed by Genewiz, Incorporated (South Plainfield, New Jersey, USA). Sequence assembly and analysis of chromatograms were performed with Geneious version 2019.2.3 (http://www.geneious. com). All nucleotide sequence data were deposited in GenBank (OL413003− OL413009). The 28S phylogenetic analysis included 3 identical sequences from the 2 hosts (see above) plus the single available liolopid in GenBank (Baba et al., 2011). Other taxa included in the analysis were informed by Baba et al. (2011) andHernández-Mena et al. (2017). Sequences were aligned with the multiple alignment tool using fast Fourier transform (MAFFT) (Katoh and Standley, 2013) and trimmed to the length of the shortest sequence (1223 [28S] base pairs). JModelTest 2 version 2.1.10 was implemented to perform statistical selection of the best-fit models of nucleotide substitution based on Bayesian Information Criterion (BIC) (Darriba et al., 2012). Aligned sequences were reformatted (from .fasta to .nexus) using the web application ALTER (Glez-Peña et al., 2010) to run Bayesian inference   (Mehra, 1936) Brooks and Overstreet (1978)  (BI). BI was performed in MrBayes version 3.2.5 (Ronquist and Huelsenbeck, 2003) using substitution model averaging ("nst-mixed") and a gamma distribution to model rate-heterogeneity. Defaults were used in all other parameters. Three independent runs with 4 Metropolis-coupled chains were run for 5,000,000 generations, sampling the posterior distribution every 1000 generations. Convergence was checked using Tracer v1.6.1 (Rambaut et al., 2014) and the "sump" command in MrBayes: all runs appeared to reach convergence after discarding the first 25% of generation as burn-in. A majority rule consensus tree of the post burn-in posterior distribution was generated with the "sumt" command in MrBayes. The inferred phylogenetic tree was visualized using FigTree v1.4.4 (Rambaut et al., 2014) and further edited for visualization purposes with Adobe Illustrator (Adobe Systems).

Generic diagnosis
Body of adult extremely elongate, linguliform, approximately 6-9 × longer than wide, having tegumental spines/scales. Oral sucker slightly smaller than ventral sucker, ventral, subterminal, lacking unique spines/ scales. Ventral sucker minute, weakly muscular, in anterior 1/7− 1/8 of body, inter-caecal, not spanning inter-caecal space. Intestine comprising paired caeca extending sinuously posteriad approximately in parallel with lateral body margin and nearly reaching posterior body extremity, lacking diverticula and lateral out-pocketings, bowing laterad at level of testes. Gonads inter-caecal, in posterior half of body, posterior to male and female terminal genitalia. Testes deeply lobed, transverse (markedly wider than long), tandem, medial, closely flanked by caeca, delimiting male and female proximal genitalia, in posterior half of body. Cirrus sac massive (spanning breadth of inter-caecal space), transverse or oblique, between ventral sucker and anterior testis, containing bipartite seminal vesicle, pars prostatica, and spined eversible cirrus. Common genital pore sinistral, ventral to sinistral caeca, post-acetabular, pre-testicular. Ovary lobed, dextral, inter-testicular, inter-caecal, nearest posterior testis. Vitellarium co-distributing with caeca from posterior to level of ventral sucker posteriad to posterior body end, not extending anteriad to level of intestinal bifurcation, not extending mediad far beyond excretory ducts. Uterus extremely elongate, sinuous, lateral to anterior testis (not ventral to anterior testis), containing many operculate eggs. Excretory system comprising lasso configuration dextrally and sinistrally (excretory system having a pair of ducts, each having an anterior cyclocoel-like portion and a posteriorly-directed collecting duct); excretory pore dorsal (subterminal). Intestinal parasites of turtles.
Etymology: The specific epithet "karinganiense" (neuter) is for the type locality and honors the personnel of KGR for their generous logistic support and cooperation in conducting parasite biodiversity research in Mozambique.

Taxonomic remarks
The new genus can be easily differentiated from the other accepted genera of Liolopidae by the combination of having a linguliform body approximately 6-9 × longer than wide, tegumental spines/scales, a minute ventral sucker located in the anterior 1/7− 1/8 of the body, deeply lobed testes that are transverse and abut the caeca (spanning the intercaecal space), a lobed ovary that is dextral and nearest the posterior testis, a uterus that is lateral to the anterior testis (not ventral to anterior testis), and a vitellarium that does not extend anteriad to the level of the ventral sucker and that does not fill the intercaecal space. The new species differs from Helicotrema spp. by having testes in the posterior half of the body; from Liolope spp. by having a cirrus sac that does not abut the ventral sucker; from Harmotrema spp. by having transverse, deeply lobed testes that abut the caeca; and from Dracovermis spp. by having testes that are far apart and that are not limited to the posterior 1/3 of the body ( Table 2).
The new species is most similar to species of Harmotrema but further differs from all but one of them by having tegumental spines. The type species of Harmotrema (H. infecundum) and all congeners except Harmotrema indica Chattaparhyaya (1970) lack tegumental spines (Nicoll, 1914;Chattaparhyaya, 1970). Chattaparhyaya (1970) did not measure or draw a spine so we cannot know if the spines are similar/homologous to those of the new genus; we are skeptical that they are present in H. indica. The new species differs from H. indica and all congeners except H. infecundum by having a uterus that is lateral to the anterior testis; H. infecundum has a uterus that was illustrated as lateral to the anterior testis (Nicoll, 1914). The distribution of vitelline follicles further differentiates these taxa. In the new genus, the vitelline follicles do not extend anteriad beyond nor to the level of the ventral sucker, and the inter-caecal space posterior to the ventral sucker is predominantly void of vitelline follicles (Fig. 1). Harmotrema infecundum has vitelline follicles anterior to the ventral sucker (Nicoll, 1914); H. laticaudae has follicles that terminate at level of the ventral sucker and fill the inter-caecal space between the cirrus sac and ventral sucker (Yamaguti 1933); H. eugari has vitelline follicles that extend to nearly the level of the ventral sucker and fill the inter-caecal space (Tubangui and Masilungan, 1936); and H. indica has vitelline follicles that fill the intercaecal space between the ventral sucker and cirrus sac as well as between the posterior testis and excretory pore (Chattoparhyaya, 1970). The new genus further differs from Harmotrema spp. by having a large body >8 mm in length (vs. <7 mm), a minute ventral sucker not spanning the inter-caecal space (vs. ventral sucker spanning inter-caecal space), and a lobed ovary not abutting the posterior testis (vs. ovary not lobed, near to or abutting posterior testis). The new species also has a large uterus that can have >50 eggs (vs. 21 or less in Harmotrema spp.).

Discussion
The present study is the first record of a liolopid from South Africa or Mozambique, documents a second liolopid genus from P. subniger, which is the type and only known host for Liolope dollfusi Skrjabin (1962), and triples the number of liolopid turtle hosts reported from the continent of Africa (Table 1). The previous records of liolopid infections in turtles comprise a single species of Liolope (L. dollfusi) from a west African side-necked turtle (Pleurodira) plus two species of Helicotrema (H. magniovatum Odhner, 1912;H. spirale (Diesing, 1850) Odhner, 1912 from three neotropical turtles (two hidden-necked turtles (Cryptodira) and one side-necked turtle (Pleurodira)) ( Table 1).
A robust phylogenetic hypothesis for Liolopidae has not been tested. The only study focused on the phylogenetic relationships among liolopid genera is the cladistic analysis of Brooks and Overstreet (1978) (hereafter, BO). The phylogenetic analyses of Baba et al. (2011), Hernández-Mena et al. (2017, and the present study comprise the only nucleotide-based phylogenies that include a liolopid. None has tested monophyly of liolopid genera nor robustly examined evolutionary relationships among the various liolopid lineages. The nucleotide-based studies lack the taxon sampling to assess phylogenetic interrelationships, with only a single taxon and two additional taxa included in the present study. In their cladistic analysis based on 5 characters and the 4 accepted liolopid genera at that time, BO recovered Liolope sister to the other genera and Dracovermis sister to the crown group comprising Harmotrema and Helicotrema. They used this result to test hypotheses concerning host-parasite cophyly and biogeography. The cophyly analysis and its conclusions therein are problematic because (1) no tetrapod phylogeny is cited therein, (2) no cladistic matrix was provided (the character states assigned to the genera of Liolopidae can be inferred from the labelled cladogram but no matrix was published), (3) some character state assignments were erroneous, (4) exclusion of the turtleinfecting liolopid L. dollfusi was weakly justified (but excluding it was critical to supporting their hypothesis of cophyly), and (5) the other host records that would provide evidence to reject their cophyly hypothesis (i.e., the well-documented turtle-infecting liolopids) were wholly ignored or overlooked (see below). At least one example of an error cascade stemming from this work is evidenced by Niewiadomska (2002), who stated that no species of Liolope infects a turtle-perhaps relying upon BO to understand the diversity of hosts infected by liolopids.
Regarding liolopid-saurospidan cophyly, BO reported that, "The parasites' cladistic (genealogical) relationships reflect the phylogenetic relationships of their hosts." This was likely based on an antiquated understanding of the phylogenetic position of crocodylians (and Dracovermis was the focus of their paper) as closely related to lizards and snakes (Squamata), not as the sister lineage to birds (Aves) and member of the saurospidan crown group. Our current phylogenetic understanding of the natural history of reptiles and birds (Saurospida) (Chiari et al., 2012;Finn et al., 2014) is that frogs and salamanders (Amphibia) are the earliest branching lineage sister to the remaining saurospidans; with the lizards and snakes (Squamata) sister to the clade that comprises turtles (Testudines) sister to the crown group comprising crocodylians (Crocodylia) and birds (Aves). Mapping known liolopid infections onto that host phylogeny, Liolope spp. infect amphibians and turtles (which are not closely related); Harmotrema spp. infect water snakes, cobras, and sea kraits; Helicotrema spp. infect iguanas and turtles (each are members of early branching lineages that are distantly related to the crown group comprising crocodilians and birds); and Dracovermis infects crocodylians. The present study contributes a monotypic genus, Paraharmotrema, whose only species infects turtles.
The cladogram of BO clearly shows that the theorized phylogenetic relationships among liolopid genera therein does not allow for the acceptance of parasite-host cophyly under our modern understanding of vertebrate evolution. Most obvious is that crocodylians are apomorphic and sister to birds; they are not squamates (as presumed by BO). Hence, Dracovermis spp. (which infect crocodylians) must be recovered as apomorphic among liolopids if one is to accept parasite-host cophyly (no liolopid has been reported from a bird). No specimen of Dracovermis has been sequenced to date, but the phylogenetic position of this lineage is obviously intriguing. Cophyly would predict that Dracovermis spp. should be recovered as late branching, belonging to the crown group. Also obvious is that L. dollfusi, clearly diagnosed as a member of Liolopidae (see Niewiadomska (2002); inter-testicular ovary, 'strigeid-like' excretory system), infects a turtle but Liolope was recovered as stem lineage to the remaining liolopids in BO; clearly violating host-parasite cophyly (even from a coarse branching order perspective). In fact, these results alone reject monophyly of the turtle-infecting liolopids altogether. Interestingly, BO removed L. dollfusi from their analysis, regarding it as incertae sedis because they questioned the morphology of the oesophagus. However, this fluke is clearly a liolopid and infects a turtle. Hence, it is a lapse or intentional omission to exclude this host record in the context of host-parasite cophyly. BO made no argument for another genus assignment for this taxon within Liolopidae nor did they address this infection record in their discussion of host-parasite cophyly.
Regarding host records and the analysis of BO, as detailed above and including the new species, four liolopids (one, two, and one species of Liolope, Helicotrema, and Paraharmotrema, respectively; Table 1) infect turtles (Chelonia; a lineage sister to the clade including crocodylians and birds). These infections in turtles clearly violate a strict definition of liolopid-saurospidan cophyly (all records aside from the present study were published before 1978). In addition to dismissing L. dollfusi (see above), BO ignored or were evidently unaware of the other two turtleinfecting liolopids of Helicotrema that were known at that time (H. magniovatum; H. spirale). The existence of these turtle-infecting Helicotrema spp. further contradicts cophyly. In fact, based on the host records alone, one can see that Liolope and Helicotrema include species that lack phylogenetic host specificity, with each genus including species that collectively infect multiple saurospidan classes. This objectively makes a liolopid-saurospidan cophyly study seemingly pointless until more species are described.
The assignment of character states in BO is problematic and, in part, erroneous. Most important, not all species of the genera have all of the generic features presumably coded in their matrix. Their analysis was based on the minimum number of characters to run a cladistic analysis (n taxa -1: oesophagus (present or absent), body shape (body length <4 × body width or body length >4 × body width), anterior extent of vitellaria (preacetabular or postacetabular), tegumental spines (present or absent), and gonad position (all or some extending into middle third of body or all contained in posterior third of body)). Below, we treat the problems or lapses with their use of the oesophagus, vitellarium, and scales/spines.
Regarding the oesophagus, we regard all liolopids as having a duct, i. e., an oesophagus, that courses through the pharynx and connects the mouth and intestine. Published descriptions show that this feature can be variable within a liolopid genus and open to interpretation-thereby making it a dubious character (based on published information) for inclusion in a cladistic matrix. BO evidently coded Liolope as lacking an oesophagus (after dismissing the only liolopid species that clearly has an elongate oesophagus; see above). They coded the remaining liolopid genera (Harmotrema, Helicotrema, Dracovermis) as oesophagus present (thereby comprising a critical synapomorphy that grouped those genera sister to Liolope-the theorized ancestral lineage; which was a critical piece of evidence that supported their cophyly hypothesis). However, H. eugari has an intestinal bifurcation that stems from the posterior margin of the pharynx, indicating that an oesophagus extending posteriad from the pharynx is absent, despite all other members of the genus having such an extension. Additionally, the posterior extent of the oesophagus and the location of the pharynx relative to the intestinal bifurcation in Helicotrema spp. is indeterminate based on the published illustrations of those species (Diesing, 1850;Odhner, 1912; Travassos, H.R. Dutton et al. 1922). Moreover, BO's illustration of the body of Dracovermis occidentalis Brooks and Overstreet (1978) shows a laterally expanded portion of the intestinal bifurcation immediately posterior to the pharynx, i.e., no tube extends demonstrably posteriad beyond the pharynx (which could have been coded as "oesophagus absent"). To the contrary, Dracovermis brayi Brooks and Overstreet (1978) clearly has a long tube extending posteriad from the pharynx and connecting to the intestinal bifurcation ("oesophagus present"). BO included both species in their concept of the genus; seemingly inconsistent with their treatment of L. dollfusi.
Regarding the extension of the vitellarium, this too is inconsistent within liolopid genera. For example, BO coded Harmotrema as vitellarium post-acetabular; however, we have shown that the distribution of the vitelline follicles differentiates Harmotrema spp. from each other, i. e., they have an assortment of character states related to vitelline distribution (see Remarks). Additionally, they coded Dracovermis as having vitelline follicles that do not extend anterior to the ventral sucker; however, clearly, Dracovermis nicollii Mehra (1936) has vitelline follicles that extend far anterior to the ventral sucker. Its congeners have vitelline follicles that terminate at level of the ventral sucker or slightly anterior to the ventral sucker (Mehra, 1936;Tubangui and Masilungan, 1936;Baylis, 1940;Brooks and Overstreet, 1978). It would appear that their character state assignments, with exception to L. dollfusi and Liolope, were based on consensus among the congeners rather than confirming that each species of a genus has the character state they assigned to each genus included in their analysis.
Regarding tegumental scales (spines), BO evidently assigned the wrong character states to some of the genera in their analysis. Liolope spp., H. indica, and P. karinganiense have scales/tegumental spines, whereas the other liolopids lack them. Perhaps a lapse, BO (Fig. 4 therein) depicted an "evolutionary shift of character state" (character no. 4, presence/absence of tegumental spines) for Helicotrema; indicating that only Helicotrema lacks scales/tegumental spines. This is a critical error in character state assignment because it artifactitiously polarized Helicotrema from the remaining ingroup taxa, most of which lack spines. It also erroneously related Dracovermis to an earlier branching ancestor. Collectively, these issues underscore the fact that a robust phylogenetic hypothesis for the genera of Liolopidae is lacking and needed because the one study that has been published has fundamental problems.
Given the diversity of potential hosts and the wide geographic distribution of known infections, Liolopidae is likely taxonomically undersampled across salamanders, lizards, snakes, turtles, and crocodylians. Relatively little is known about liolopid infections especially in turtles, with only 3 publications total that have detailed an infection in a turtle (Odhner, 1912;Travassos, 1922;Skrjabin, 1962); only L. dollfusi and the new species reportedly mature in turtles exclusively (Table 1). This is astonishingly low sampling effort given the extremely high diversity of extant freshwater turtle species (Rhodin et al., 2017;Bullard et al., 2019).