Description of a new species from Clarias maclareni and phylogenetical analysis of Quadriacanthus (Monogenea, Dactylogyridae) species transfers between clariid and non-clariid fish hosts in Cameroon

Recently in Cameroon, two species belonging to Quadriacanthus: Q. anaspidoglanii Akoumba, Tombi & Bilong Bilong, 2017 and Q. euzeti Nack, Pariselle & Bilong Bilong, 2016 have been recorded on gill filaments of Notoglanidium macrostoma (Siluriformes, Claroteidae) in the Memou’ou River (Nyong Basin) and Papyrocranus afer (Osteoglossiformes, Notopteridae) in Lake Ossa, respectively. These records have been considered the result of lateral transfers from Clariidae to a Claroteidae host for the first case (parasitism of N. macrostoma by Q. anaspidoglanii) and from Clariidae or Bagridae to a Notopteridae host for the second (parasitism of P. afer by Q. euzeti). In this paper, the investigation of interspecific relationships among Quadriacanthus spp. parasitizing Clariidae, Bagridae, Claroteidae and Notopteridae in Cameroon resulted in the record of Q. anaspidoglanii from N. macrostoma, Q. euzeti from P. afer, a new record of Q. levequei Birgi, 1988 from Clarias jaensis in the Nyong River, and the description of Q. barombiensis n. sp. from Clarias maclareni in Lake Barombi Mbo. The newly identified species is characterized by having an accessory piece ending in one small hook and the median expansion of its dorsal bar with two filaments. Phylogenetic analysis based on 28S rDNA sequences confirms that the Quadriacanthus spp. parasitizing gill filaments of non-clariid hosts in Cameroon originate from lateral transfers from clariid fishes, and that Clariidae are ancestral hosts of these monogenean species.


Introduction
Monogenea have a direct life cycle. They are diversified and often host specific [12] and these characters make them an important asset to tackle the question of evolution of species or speciation [48]. As is the case for free-living organisms, speciation in parasites may occur "on-site" (sympatric/ synxenic) or on vicarious sites (allopatric/alloxenic) [10]. While the second type of speciation is common for free-living organisms following migration or population isolation, it is less easy for Monogenea; in fact, being strict host specialists, they are the least probable switchers. However, once host switching succeeds, they have a high probability for speciation [7,19,48]. Quadriacanthus (Monogenea, Ancyrocephalinae) was proposed by Paperna (1961) for Q. clariadis Paperna, 1961 from the gills of Clarias gariepinus (Burchell) sampled in Israel [32]. To date 38 species are recorded in this genus from Asia and Africa [13,45]. Although their majority (34 among the 38 known species) have been recorded from Clariid-hosts, the remaining four species have been recorded from non-clariid hosts: Quadriacanthus bagrae Paperna, 1979 from Bagrus docmak (Forsskål) and Bagrus bajad (Forsskål), both Bagridae [34]; Quadriacanthus euzeti Nack, Pariselle & Bilong  from Papyrocranus afer (Günther), Notopteridae [30], Quadriacanthus anaspidoglanii Akoumba, Tombi & Pariselle, 2017 from Notoglanidium macrostoma (Pellegrin), Claroteidae [2], and a fourth one, doubtful (see [17,34]) Quadriacanthus tilapiae Paperna, 1973 from Oreochromis esculentus (Graham), Cichlidae [33]. The presence of these Quadriacanthus spp. on gill filaments of non-clariid hosts raises the question of their origin. The recent study by Francová et al. [13] on Quadriacanthus parasites of catfishes in eastern Africa suggests that the record of Q. bagrae on a bagriid host is the result of a lateral transfer from a clariid-host and that Clariidae are ancestral hosts of Quadriacanthus. In Cameroon, the presence of Q. euzeti and Q. anaspidoglanii on non-clariid fishes was also considered to originate from lateral transfers between Clariidae or Bagridae to Notopteridae for the first [30] and from Clariidae to Claroteidae for the second [2]. Because clariids, bagrids and notopterids or claroteids live in sympatry in Lake Ossa [30] and/or in the Memou'ou River [2], it was impossible, without genetic data, to determine which group was the original host family of laterally transferred Quadriacanthus species. Therefore, the main topic of our work concerns the use of sequence data to test the origin of these species. Taking the example of Q. euzeti in Lake Ossa, we hypothesize that if this species comes from a clariid host, it will be phylogenetically close to Q. levequei Birgi, 1988 (which is morphologically close to Q. euzeti) hosted by Clarias jaensis Boulenger in this lake [42,43]. If Q. euzeti originates from a bagriid host, it will be phylogenetically close to Q. bagrae described on B. docmak, the sole Bagridae presently recorded in this lake. In the present study, we analyze three morphologically related Quadriacanthus species parasites of clariid and non-clariid fishes, namely Q. levequei, Q. euzeti, and Q. anaspidoglanii and add a new one, also morphologically similar. These four Quadriacanthus species are genetically compared to Q. bagrae and other Quadriacanthus species available in GenBank.

Material and methods
Specimens of the following four species: Clarias maclareni Trewavas [44] (n = 20) endemic to Lake Barombi Mbo (4°38 0 N, 9°22 0 E); C. jaensis (n = 15) from the Nyong River, Mbalmayo market (3°30 0 48.54 00 N, 11°30 0 04.83 00 E) and Sokamalem, Abong-Mbang (03°58 0 21.4 00 N, 13°14 0 53.3 00 E); A. macrostoma (n = 34) from the Nyong River, Mengong (2°58 0 31.64 00 N, 11°27 0 06.87 00 E), and P. afer (n = 10) from Lake Ossa (4°39 0 N, 9°24 0 E) (Fig. 1), were caught between January 2016 to February 2017 using gill nets, cast-nets, fishtraps or hook lines, and/or purchased from fishermen. They were immediately placed in a cool box containing ice, then transported to the laboratory where they were frozen at À21°C. In the laboratory, after thawing of the carcasses, the gill arches of fish specimens were removed by dorsal and ventral sections, then placed in a Petri-dish containing tap water. The parasites were dislodged from the gill filaments with a dissecting needle. Monogeneans were fixed individually between slide and cover slip in a drop of GAP (glycerin ammoniumpicrate mixture) [22]. After 24 h, preparations were sealed using Glyceel [4]. The identification was based on the morphology and the size of sclerotized parts of the haptor and the copulatory organs. The measurements, carried out according to Gussev [14] modified by N'Douba et al. [27] (Fig. 2), and drawings of the sclerotized parts of the haptor and copulatory complex, were made with the aid of a Leica DM 2500 microscope, LAS software (3.8), ImageJ 1. 53 K software and Corel DrawX4 Ò software, version 14.0.0.701. Measurements, in micrometers (lm) are presented as follows: mean (minimummaximum). Prevalence (P) and mean intensity (MI) were calculated according to Bush et al. [8]. Type material and vouchers were deposited in the helminth collection of the Royal Museum for Central Africa (MRAC) Tervuren (Belgium) under accession numbers MRAC 43425-43429. A principal components analysis (PCA) was performed using Statistica 6, with "standardized" measurements according to Messu et al. [26]. To prevent the influence of temperature or of development stage, we divided each by the length of hook II, which is supposed to keep larval size [36]. Twenty-four characters (among a total number of twenty-nine measured on each specimen, see Table 1) were retained for the PCA. Ten (10) specimens of each species included in this work were used in the PCA. For genetic purposes, fish were dissected in the field; gill arches were excised as mentioned above and stored in alcohol (95%) according to Justine et al. [18], then examined under stereomicroscope. Parasites found were mounted individually between slide and cover-slip in a drop of water and identified according to Birgi [6], Nack et al. [30] and Akoumba et al. [2]. After identification, each parasite was placed individually in an Eppendorf Ò tube containing 95% alcohol. PCR was performed on these specimens according to Marchiori et al. [23], directly without DNA extraction. Standard PCR was performed using primers specific to the D1-D2 domain of the large subunit region (LSU) of the 28S ribosomal gene: C1 (forward; 5 0 -AC-CCGCTGAATTTAAGCAT -3 0 ) and D2 (reverse; 5 0 -TGGTCCGTGTTTCAAGAC -3 0 ) [15]. The amplification consisted of three steps and began with 2 min at 93°C for initial denaturation, followed by 30 cycles: 30 s at 93°C, 30 s at 56°C for annealing, 1 min 30 s at 72°C for extension, with a final 5 min extension step at 72°C. The final concentration of different reagents was as follows: GoTaq Flexibuffer (Promega) 1Â, MgCl 2 2.5 mM, PCR nucleotide mix 0.2 nM of each DNTP, forward and reverse primers 1 lM of each, GoTaq (Promega) DNA polymerase 2 U, template DNA 0.2 lg (between 1.6 and 3 lL depending on the DNA extract concentration), nuclease-free water up to 20 lL. Sequencing was performed at the Genseq platform of ISE-M (Institute of Evolutionary Sciences of Montpellier) using the same primers as in initial PCR amplification. Purification was performed with an Agencourt Ò AMPure Ò PCR purification kit, following the manufacturer's recommendations. Sequences were aligned using the Muscle program and improved manually using molecular evolutionary genetics analysis (MEGA) software [41] version 6.0. The alignments were trimmed manually using the same software. Additional 28S sequences of seven  Table 2 for accession numbers). Three species parasitizing Siluriform fish, namely Synodontella zambezensis Douëllou & Chishawa, 1995, Schilbetrema sp. and Thaparocleidus mutabilis (Gussev & Strelkov, 1960) and Onchobdella aframae Paperna, 1968 parasitizing a Cichlidae were used as the outgroup; they were obtained from GenBank. Prior to analysis, an evolutionary model was selected by MEGA 6.0 using the Bayesian information criterion (BIC). The model with the lowest BIC score was considered to better describe the pattern. Neighbor-Joining (NJ), Maximum Parsimony (MP), and Maximum Likelihood (ML) analyses  were performed using MEGA version 6.0, assessing nodal support non-parametric bootstrap with 1000 replicates.
New sequences were generated from each identified species in the present study.

Quadriacanthus levequei Birgi, 1988
Type-host: Clarias pachynema Boulenger, 1909.   Seven pairs of hooks, pair IV with short handle, larger than pairs I, II, III, V, VI and VII, the latter pairs about subequal. Tubular male copulatory organ (MCO) enlarged at its basal zone and tapered at distal extremity. Accessory piece straight, slightly curved distally and ending in two small, rounded hooks, one  surmounting the other. Tubular vagina showing two reduced lateral expansions at its median zone.

Remarks
The morphology of dorsal bar with rectangular center and a median expansion stick-shaped showing two filaments at its end, the one of dorsal anchor, and the size of MCO and its accessory piece (compared to the measurements taken from the original drawings, see Table 1) of the specimens recorded in the current study on C. jaensis are similar to those of Q. levequei reported on C. pachynema by Birgi [6]. The differences observed between our measurements from the newly studied specimens, those taken from original drawings and those given in the original description (i.  [17].

Description
Adult worms 579 (410-730) long, 148 (102-208) large at level of ovary. Pharynx circular 30 (27)(28)(29)(30)(31)(32)(33). Dorsal bar with rectangular center, two lateral branches, stick-shaped median process with small circular median hole, and ending with two filaments. Dorsal anchor without handle nor guard, with regular curved thin blade and short point. Ventral bar V-shaped made up of two lateral medially articulated expansions. Ventral anchor with a blade curved in an arc and ending in a long point. Ventral and dorsal cunei triangular, dorsal cuneus being larger than ventral one (see Table 1). Seven pairs of hooks, pair IV with short and pear-shaped handle, larger than pairs I, II, III, V, VI and VII, the latter pairs about subequal. Tubular MCO large at basal zone and tapered at distal extremity, accessory piece slightly S-shaped ending in one small point. Vagina not observed.

Remarks
By its general morphology of haptoral structures and MCO: the stick shape of dorsal bar median process (1), tubular shape of MCO enlarged at basal zone and tapered at distal end (2), and s-shape of accessory piece (3), Q. barombiensis n. sp. resembles Q. levequei, Q. anaspidoglanii and Q. euzeti; but it can easily be distinguished from its congeners by: the morphology of the distal extremity of the accessory piece with one small hook versus two small hooks in Q. levequei (1), the dorsal bar postero-median process with two filaments versus none in Q. euzeti and Q. anaspidoglanii

Phylogenetic analysis
After trimming, the alignment of 616 positions (base pairs) was obtained, among these positions 335 variable sites were identified, 184 of which were parsimony informative. TN93 + G was selected as the best fit for our data. The analysis based on three different methods (NJ, MP and ML) produced a congruent tree topology (Fig. 6). All the Quadriacanthus spp. appeared clustered in one monophyletic group. Quadriacanthus kobiensis (Asian species) is well separated from African Quadriacanthus spp. and situated at the basal position of the tree. Considering African Quadriacanthus spp., two welldefined clusters were observed with high support. The first cluster (I) was formed by Q. bagrae, Q. clariadis, Q. fornicatus, Q. mandibulatus, Q. pravus and Q. zuheiri with high support. Within this cluster, Q. bagrae was sister species to Q. clariadis with high support. The second cluster (II) was formed by Q. levequei, Q. euzeti, Q. barombiensis and Q. anaspidoglanii with high support. Within this second cluster, Q. euzeti is separated from the other three Quadriacanthus spp. among which Q. anaspidoglanii was sister species to Q. barombiensis n. sp. and Q. levequei, both latter species being separated by 1% of Gamma-corrected genetic distance (Table 3, Fig. 6).

Discussion
Quadriacanthus barombiensis n. sp. is specific to C. maclareni, i.e. oioxenous [10], as is the case for the majority of known Quadriacanthus species [2, 3, 6, 27-29, 30, 45]. Quadriacanthus levequei was previously described from the gills of C. pachynema and considered oioxenous [6]; its new record on one congeneric host (C. jaensis) changes its host specificity status from oioxenous to stenoxenous [12], even better mesostenoxenous [9]. This enlargement of the host spectrum of Q. levequei in the Nyong River Basin may have been promoted by relative phylogenetic proximity [21,38] of C. jaensis and C. pachynema and/or local ecological conditions in the environment [30,35]. According to Teugels [43], C. maclareni is morphologically close to C. jaensis and both species belong to sub-genus Clarias (Platycephaloides), but contrary to C. jaensis which hosts four dactylogyridean species, namely Q. dageti Birgi, 1988, Q. teugelsi, Q. nyongensis Birgi, 1988 and Birgiellus calaris Bilong Bilong, Nack and Euzet, 2007 [5], C. maclareni hosts only one species: Q. barombiensis n. sp. We assume that when the ancestor of C. maclareni colonized Lake Barombi Mbo from the Memé River system, which played a major role in fish colonization of this lake [31,44], it could have hosted (1) several monogenean species, which have been lost due to environmental changes or following bottleneck events [37], or (2) only the ancestor of Q. barombiensis n. sp. Clarias maclareni being endemic in this Cameroonian volcanic line crater lake which shelters a Cichlid species flock [44], without parasite lateral transfer and/or intrahost speciation (synxenic [10]) favored by host population fragmentation, no increase of monogenean species richness has been possible [37,46]. The close relationship between host species: C. maclareni and C. jaensis (see Teugels op. cit.), and their respective parasite species: Q. barombiensis n. sp. and Q. levequei (see Table 3 and Fig. 6), is a good illustration of a co-vicariance followed by the co-speciation of both fish and their Monogeneans [7].  presumably C. jaensis, and early took place before the speciation of other Cameroonian Quadriacanthus (Q. anaspidoglanii, Q. levequei and Q. barombiensis n. sp.). This type of phenomenon (transfer from a distant host family) has been recorded by Messu Mandeng et al. [26] in Cameroon, where Cichlidogyrus Paperna, 1960 usually found on cichlid hosts transferred to a cyprinodontiform host. In addition, the basal position of Q. kobiensis, parasite of Clarias batrachus (Linnaeus) from Asia (where African clariids originate [1]), suggests that members of Clariidae are ancestral hosts of Quadriacanthus spp. [13] and that African Quadriacanthus species have an Asian origin too (which was suggested by Pariselle et al. [37] based on the presence of additional sclerites (cunei) associated with the anchors in Asian Siluriform monogeneans and Quadriacanthus species). Quadriacanthus euzeti, Q. anaspidoglanii and Q. bagrae from non-clariid hosts are distinguished from their close congeners (Q. levequei for the two first species and Q. clariadis for Q. bagrae) hosted by clariid species, by the morphology and size of sclerotized parts of the haptor, while the copulatory organs look similar [2,13,20,30,34]: e.g. the dorsal and ventral anchor blade length, the thickness of ventral bar and the length of dorsal cunei are always reduced in Q. anaspidoglanii, while they are bigger in Q. euzeti; Francová et al. [13] highlight that Q. bagrae differs from Q. clariadis by the length of the ventral bar and the size of dorsal anchor blade, longer in Q. clariadis. This observation   supports the adaptive nature of haptoral hard parts which are subject to selective pressure [16] such as gill morphology, encountered by these different Quadriacanthus spp. parasitizing distant hosts [26,40].