Evidence for two morphologically cryptic species of Hysterolecitha Linton, 1910 (Trematoda: Lecithasteridae) infecting overlapping host ranges in Moreton Bay, Australia

Integration of morphological and molecular approaches to species delineation has become an essential part of digenean trematode taxonomy, particularly when delimiting cryptic species. Here, we use an integrated approach to distinguish and describe two morphologically cryptic species of Hysterolecitha Linton, 1910 (Trematoda: Lecithasteridae) from fishes of Moreton Bay, Queensland, Australia. Morphological analyses of Hysterolecitha specimens from six fish species demonstrated a complete overlap in morphometric data with no reliable differences in their gross morphological characters that suggested the presence of more than one species. Distinctions in ITS2 rDNA and cox1 mtDNA sequence data for corresponding specimens suggested the presence of two forms. A principal component analysis on an imputed dataset showed clear separation between the two forms. These two forms are partially separated on the basis of their host’s identity. Therefore, we describe two morphologically cryptic species: Hysterolecitha melae n. sp. from three species of Abudefduf Forsskål and one species of Parma Günther (Pomacentridae), with the Bengal sergeant, Abudefduf bengalensis (Bloch), as the type-host; and Hysterolecitha phisoni n. sp. from species of Pomacentridae (including A. bengalensis), Pomatomidae and Siganidae, with the black rabbitfish, Siganus fuscescens (Houttuyn), as the type-host.


Introduction
The Hysterolecithinae Yamaguti, 1958 (Lecithasteridae) is the second largest of the six lecithasterid subfamilies, comprising four genera, Hysterolecitha Linton, 1910, Hysterolecithoides Yamaguti, 1934, Machidatrema León-Règagnon, 1998, and Thulinia Gibson & Bray, 1979. Members of the Hysterolecithinae differ from those of other lecithasterid subfamilies in the possession of Juel's organ and a uterine seminal receptacle (Gibson, 2002). The type-genus, Hysterolecitha, is the richest of the four hysterolecithine genera, comprising 22 recognised marine and freshwater species. Species of this genus are distinguished from those of Hysterolecithoides, Machidatrema and Thulinia by an anterior fusion of the excretory ducts, absence of filamented eggs, and a weakly developed sinus-sac.
Species of Hysterolecitha have been reported from fishes from 21 families and most have been reported with oioxenous host-specificity (Table 1). Notably, while some species have been reported to be stenoxenic or euryxenic (infecting several host species of a single family or multiple families, respectively) or Table 1 Host information, host-specificity, and geographic distribution for the 22 recognised species of Hysterolecitha Linton, 1910.
Here, we describe two new species of Hysterolecitha from fishes of Moreton Bay, Queensland, Australia. These two new species are essentially cryptic relative to each other in that, despite being clearly distinct genetically while occurring in sympatry, they are effectively morphologically indistinguishable. Although these two species of Hysterolecitha infect an overlapping range of host species, they can be partially distinguished on the basis of their host range.

Specimen collection
Fishes were collected from Moreton Bay in southeast Queensland, Australia between 2015 and 2021 via line-fishing and tunnel netting. Fishes were euthanised via an overdose of anaesthetic (AQUI-SÒ, AQUI-S New Zealand Ltd, Lower Hutt, New Zealand). The gastrointestinal tract was removed and examined for digeneans using the 'gut wash' method (Cribb & Bray, 2010). Digeneans were washed in saline, fixed in nearboiling saline, and preserved in 80% ethanol. Multiple specimens were prepared as hologenophores for parallel morphological and molecular analyses (Pleijel et al., 2008).

Morphological analyses
Specimens were rinsed with distilled water, overstained in Mayer's haematoxylin, destained in a 1% hydrochloric acid solution, and neutralised in an 1% ammonium hydroxide solution. Specimens were then dehydrated in a graded series of ethanol solutions, cleared in methyl salicylate, and mounted in Canada balsam. Morphometric data were taken using a camera (Olympus SC50) mounted on a compound microscope (Olympus BX-53), and cellSens Standard imaging software. Measurements are in micrometres and are presented in Table 2. Drawings were made using a drawing tube attachment and digitised in Adobe Illustrator. Details of the two new species have been submitted to ZooBank and registered with Life Science Identifiers (LSID) to comply with the recommendations set out in the International Code of Zoological Nomenclature (ICZN, 2012). Specimens are lodged at the Queensland Museum (QM), Brisbane, Queensland, Australia.
For morphometric analyses, only characters from hologenophores with associated molecular data and paragenophores which were inferred as distinct based on host identity were used. The pre-ovarian length was used as a proxy for body length. To test the significance of some morphometric differences, Welch's ttest was used. A principal component analysis (PCA) was used to further explore the morphometric dataset in R (R Core Team, 2022) using the packages 'FactoMineR' (Lê et al., 2008) for the analysis and 'factoextra' (Kassambara & Mundt, 2020) for visualisation. As some specimens were damaged or incomplete, not all characters could be measured or included in the analysis. Standard PCA methods are not suitable for incomplete datasets; in this study, deleting individuals or variables with incomplete observations would decrease an already limited dataset and would reduce the statistical power of the analysis. To address this, the package 'mice' (van Buuren & Groothuis-Oudshoorn, 2011) was used to impute the missing measurements; these missing values were predicted using multiple imputations based on fully conditional specification, where each variable (or character) is imputed by a separate model. Only specimens with at least 50% of the original measurements were included in the imputation and analysis (n = 28). To test the significance of the PCA, the R package 'PCAtest' (Camargo, 2022) was used with 1000 bootstrap replications and 1000 random permutations.

Molecular analysis
Genomic DNA was extracted using a standard phenol/ chloroform extraction method (Sambrook & Russell, 2001) and sequence data were generated for one ribosomal DNA (rDNA) marker, the second internal  Amplification was conducted on a TaKaRa PCR Thermal Cycler (TP-690) and amplified DNA was sent to the Australian Genome Research Facility for purification and dual direction Sanger sequencing using the amplification primers. The ITS2 rDNA and cox1 mtDNA datasets were aligned separately in MEGA X (Kumar et al., 2018) using MUSCLE (Edgar, 2004), with UPGMA clustering for iterations 1 and 2. The final dataset for the ITS2 rDNA alignment contained 559 base pairs. The cox1 mtDNA alignment was translated (echinoderm/ flatworm mitochondrial code) and examined in Mesquite version 3.70 (Maddison & Maddison, 2021) for internal stop codons and to determine the correct reading frame. The alignment was trimmed after the correct reading frame was determined. All codon positions were then tested for non-stationarity in PAUP* version 4.0a (Swofford, 2003), and substitution saturation using the ''Test of substitution saturation by Xia et al.'' function (Xia et al., 2003;Xia & Lemey, 2009) implemented in DAMBE version 7.2 (Xia, 2018). Non-stationarity and substitution saturation were not detected, and as such, all codons were used in subsequent analyses. The final dataset for the cox1 alignment contained 474 base pairs. Neighbour-joining analyses were conducted for each alignment with the following parameters: ''Test of Phylogeny

Overview
Specimens morphologically consistent with the genus Hysterolecitha were collected from four pomacentrid species, the Bengal sergeant, Abudefduf bengalensis (Bloch), the Indo-Pacific sergeant, A. vaigiensis (Quoy & Gaimard), Whitley's sergeant, A. whitleyi Allen & Robertson, and the bigscale scaly fin, Parma oligolepis Whitley, one pomatomid species, the tailor, Pomatomus saltatrix (Linnaeus), and one siganid species, the black rabbitfish, Siganus fuscescens (Houttuyn). Initial morphological examination suggested that the collection represented a single species; however, the molecular data demonstrated the clear presence of two forms (Figure 1). A total of 38 ITS2 rDNA and 38 cox1 mtDNA sequences were generated. From the ITS2 sequence data, two forms were recognised, differing at 21 base positions. Corresponding cox1 sequence data differed at 89-91 base positions. Both forms showed intraspecific variation in the cox1 region, with one varying at a single base position, and the second at 1-7 base positions. Reexamination of hologenophores revealed some marginally diagnostic characters that could be used (albeit unreliably) to distinguish between the two forms, specifically the forebody length relative to the body length or the pre-ovarian length, and the development of the terminal genitalia. The two forms are partly biologically distinguished by their host; that is, siganid and pomatomid hosts were only infected by one form, whereas one pomacentrid, A. bengalensis, was infected by both.  The PCA of the morphometric data revealed a combination of characters which separated the two forms ( Figure 2 and Table 2). The PCA permutation tests revealed significant w (129.027, p = 0) and u (0.378, p \.001) values, indicating that the PCA was biologically significant. Based on cumulative variance and eigenvalues, the first five to eight principal components (PCs) would be retained for further analyses as they accounted for at least 80% of the total variation and had eigenvalues above one (Table 3). However, only PC1 and PC2 were statistically significant and were retained for subsequent analyses. Based on the loading values, a combination of 22 variables contributed significantly to PC1, explaining 35.4% of the total variation (95% confidence interval 30.2-44.8; p\.001) and a combination of six variables contributed significantly to PC2, explaining 14.9% of the total variation (95% confidence interval 12.3-21.9; p \.001). For list of significant variables see Table 2.
The integrated analysis of the morphological and molecular data (as well as the host-specificity) suggest that the two forms of Hysterolecitha in the collection represent two species. The forms do not agree with known species of Hysterolecitha and are described as new herein.

Description
Based on 19 hologenophores and eight paragenophores (from all hosts; see Table 2 for measurements). Body elongate, cylindrical, with hindbody wider than forebody, widest at level of mid-ventral sucker. Anterior end of body rounded, tapering distally. Posterior end rounded. Pre-oral lobe usually distinct. Oral sucker globular, subterminal, with anterior half generally broader than posterior half. Ventral sucker rounded, larger than oral sucker, with  Table 4 for accession numbers). ZooBank registration: urn:lsid:zoobank.org:act:4BC2369B-5BAF-4016-A16C-BB20E1526A09. Etymology: The new species is named for the first author's partner, Brody Phi Son Ly, in recognition of his constant support and encouragement.

Remarks
The current collection of Hysterolecitha specimens sampled from Moreton Bay fishes represents a case of sympatric, cryptic species that are associated with a partial overlap in host species ranges. Representative specimens from each individual host were prepared as a hologenophore and the subsequent genetic data associated with each hologenophore was used to tentatively identify paragenophores. This process resulted in collections of worms from single host individuals being identified as the same species. As the two species are essentially cryptic, we acknowledge that this division may not be completely reliable; definitive species identification is presently dependent on genetic data (and, partly, host identity). Based on this genetic delineation, specimens were compared morphologically to find a basis of distinction. There is a significant difference in the forebody length relative to: a) body length and b) pre-ovarian length between specimens of H. melae n. sp. shorter forebodies relative to their body length and pre-ovarian length, whereas H. phisoni n. sp. specimens generally have longer forebodies; however, this pattern is not completely consistent and is not observed across all specimens ( Figure 5). The specimens also differed in the development of their terminal genitalia; H. melae n. sp. generally has a well-developed sinus-organ and a smaller pars prostatica than H. phisoni n. sp. which has a poorly developed sinus-sac and a larger pars prostatica.

Discussion
Species delineation Bray et al. (2022) proposed a set of criteria for trematode species delineation which requires reciprocal monophyly in the most discriminatory molecular markers in combination with either morphological differences relative to other taxa or distinctions in host species ranges relative to those of closely related taxa. The recognition of two species of Hysterolecitha here is based on ITS2 and cox1 sequence data, and morphometric analyses that showed that the specimens formed distinct clusters. The strong reciprocal monophyly is the key evidence in justifying the recognition of two species. The two species are also partly distinguished by host range; specimens of H. melae n. sp. were found in only four pomacentrid species whereas H. phisoni n. sp. was found commonly in a siganid, uncommonly in a pomacentrid, and rarely in a pomatomid. Therefore, the interpretation of these data is that the present collection of Hysterolecitha from Moreton Bay comprises two genetically distinct species that are essentially morphologically cryptic.
The genetic differences reported here for H. melae n. sp. and H. phisoni n. sp. are generally comparable to other combinations of cryptic species. Recent studies have reported genetic differences of up to 21 base positions in the ITS2 region and up to 53 base positions in the cox1 region for cryptic species of lepocreadiids from the GBR (Bray et al., 2018;Bray et al., 2022) and monorchiids from the GBR and Japan (Wee et al., 2022). However, the genetic differences among some closely related, non-cryptic species has been reported to be much lower. For example, studies on morphologically distinguishable species of bivesiculids (Trieu et al., 2015;Cribb et al., 2022) and lepocreadiids (Bray et al., 2018;Bray et al., 2022) from GBR fishes have reported differences at one to two base positions in the ITS2 region and up to 52 base positions in the cox1 region. We conclude that use of a 'yardstick' approach to the interpretation of the significance of levels of molecular distinction is problematic. Instead, interpretations are best made on a case by case basis in the light of all available evidence. Here, we interpret the evidence as clearly indicating the presence of two species.

The genus Hysterolecitha
Previous reports of species of Hysterolecitha have been overwhelmingly based on morphometric data in isolation. A search for Hysterolecitha sequence data on GenBank returned only a single result, an 18S sequence of H. nahaensis (from an unknown host and locality), generated as part of a large phylogenetic study of the Hemiuroidea (Blair et al., 1998). This lack of genetic data associated with reports (and descriptions) makes reliable species delineation and identification difficult, especially for species as morphologically cryptic as H. melae n. sp. and H. phisoni n. sp. Without supporting genetic data, the current collection of Moreton Bay Hysterolecitha specimens would certainly have been considered a single euryxenous species with marginal intraspecific morphological variation. The genetic data, however, clearly indicates that the new specimens comprise two species with different forms of host-specificity (euryxenous and stenoxenous).
Of the known marine Hysterolecitha species, five have been reported from more than one locality, with H. nahaensis [see Yamaguti (1942), Yamaguti (1953), Parukhin (1989) and Zhokhov et al. (2018)] and H. rosea [see Linton (1910) and Wang (1982)] being reported as the most widespread. However, like the majority of Hysterolecitha species, the two new species are known from only a single locality, but unlike for most species of the genus, there is some evidence of absence. Hysterolecitha melae n. sp. and H. phisoni n. sp. have not been detected at other Australian localities, specifically the GBR, where two other known species (H. heronensis and H. nahaensis) have been found (Bray et al., 1993;Barker et al., 1994;Sun et al., 2012). While the distribution of Pomatomus saltatrix does not indicate it would be found in the GBR (Bray, 2022), the pomacentrid and siganid hosts of H. melae n. sp. and H. phisoni n. sp. have broad distributions that encompass the GBR (Randall et al., 1998;Parmentier & Frédérich, 2016;Bray, 2020). However, based on extensive collecting in Queensland waters for over 20 years, we have not detected either of the two new species (or the known species) outside of their respective known localities. That is, H. melae n. sp. and H. phisoni n. sp. have not been found in the GBR, and H. heronensis and H. nahaensis have not been found from Moreton Bay, despite the presence of suitable hosts for each species at these locations.

Cryptic species complexes in the Hemiuroidea
Digeneans have been reported to have some of the highest levels of cryptic diversity relative to other helminth taxa (Pérez-Ponce de León & Poulin, 2018). The findings of cryptic diversity here extend a growing list of cases within the Hemiuroidea. One reported case of cryptic hemiuroids was by Carreras-Aubets et al. (2011) who described a new species related to the lecithasterid Aponurus laguncula Looss, 1907. The new species was recognised for combined molecular, morphological and host distinctions. The new species was not strictly morphologically cryptic relative to A. laguncula, but it was sufficiently inconspicuous to have escaped recognition for over 100 years since the description of A. laguncula, although that species was suspected to constitute a species complex due to its euryxenous host-specificity and wide geographical distribution (Bray & MacKenzie, 1990;Bray et al., 1993). Another case was for the genus Hirudinella de Blainville, 1828 (Hirudinellidae). For this genus, the delineation of species is made challenging by the huge size of the specimens and the lack of reliable characters to separate species (Gibson & Bray, 1979); as a result, many nominal Hirudinella species are no longer recognised. Calhoun et al. (2013) used ITS1, ITS2 and 28S rDNA sequence data to show that specimens consistent with Hirudinella ventricosa (Pallas, 1774) Baird, 1853 comprised four species. Despite the clarity of the molecular data, only two of the species (H. ventricosa and H. ahi Yamaguti, 1970) could be formally recognised; two other species were not named due to limitations associated with the morphological features and additional genetic data. Recent molecular work on morphologically similar specimens of Lecithaster Lühe, 1901 (Lecithasteridae) collected from fishes belonging to multiple families has resulted in the recognition of two species that had been synonymised principally on morphometric similarity (Atopkin et al., 2020). Lecithaster sayori Yamaguti, 1934and L. salmonis Yamaguti, 1934[previously synonymised with L. stellatus Looss, 1907(see Manter & Pritchard, 1960 and L. gibbosus (Rudolphi, 1802) Lühe, 1901 (see Margolis & Boyce, 1969), respectively], were shown to be genetically distinct based on ribosomal markers. Most recently, adult and cercarial specimens morphologically consistent with Derogenes varicus (Müller, 1784) Looss, 1901 (Derogenidae) were shown to be genetically distinct, forming up to four lineages based on ribosomal and mitochondrial markers that were associated with different hosts and localities (Olson et al., 2003;Sokolov et al., 2021;Krupenko et al., 2022). Although the genetic distinctions are clear, the lack of corresponding adult morphological specimens has hindered the naming of these lineages as new species. The current findings of cryptic species of Hysterolecitha here are broadly consistent with the previous studies reviewed above. As is frequently the case, the distinct species here were largely associated with different host taxa. The most difficult problem of cryptic species, where combinations of species occur in the same host and the same locality (as partly occurred here), is not commonly reported.