To each his own: no evidence of gyrodactylid parasite host switches from invasive poeciliid fishes to Goodea atripinnis Jordan (Cyprinodontiformes: Goodeidae), the most dominant endemic freshwater goodeid fish in the Mexican Highlands

Background Goodeid topminnows are live-bearing fishes endemic to the Mexican Highlands (Mesa Central, MC). Unfortunately, in the MC, environmental degradation and introduced species have pushed several goodeid species to the brink of extinction. Invasive fishes can introduce exotic parasites, and the most abundant goodeid, blackfin goodea Goodea atripinnis Jordan, is parasitised by six exotic helminths. Poeciliids are widely dispersed invasive fishes, which exert negative ecological effects on goodeids. Poeciliids host several species of the monogenean genus Gyrodactylus von Nordmann, 1832, including pathogenic, invasive parasites. Here, we looked for evidence of Gyrodactylus species switching hosts from poeciliids to goodeids. Methods Fish were collected in rivers draining the MC into both sides of the continental divide. Hosts were screened for gyrodactylid parasites in localities where G. atripinnis and poeciliids occurred sympatrically. Gyrodactylus specimens were characterised morphologically (attachment apparatus) and molecularly (internal transcribed spacer region, ITS). A Bayesian phylogenetic tree using ITS sequences established relationships between gyrodactylids collected from goodeid fishes and those from parasites infecting poeciliids. Results Gyrodactylids were collected from G. atripinnis in six localities on both sides of the watershed where exotic poeciliids occurred sympatrically. Morphological and molecular analyses indicated the presence of four undescribed species of Gyrodactylus infecting this goodeid host. Gyrodactylus tomahuac n. sp., the most abundant and geographically widespread species, is described here. The other three Gyrodactylus spp. are not described, but their ITS sequences are used as molecular data presented here, are the only available for gyrodactylids infecting goodeid fishes. Morphological and molecular data suggest that two distinct groups of gyrodactylids infect goodeids, one of which shares a common ancestor with gyrodactylids parasitizing poeciliids. Conclusions No evidence was found of gyrodactylids switching hosts from invasive poeciliids to endemic goodeids, nor vice versa. Moreover, considering that G. atripinnis is known to host both Gyrodactylus lamothei Mendoza-Palmero, Sereno-Uribe & Salgado-Maldonado, 2009 and Gyrodactylus mexicanus Mendoza-Palmero, Sereno-Uribe & Salgado-Maldonado, 2009, with the addition of G. tomahuac n. sp. and the three undescribed Gyrodactylus spp. reported, at least six gyrodactylids may infect this host. This would make monogeneans the second most abundant parasite group infecting G. atripinnis, which to date is known to harbour 22 helminth species: nine digeneans, five nematodes, four cestodes, three monogeneans and one acanthocephalan.


Background
Human-mediated introduction of non-native (also referred to as alien or exotic) fishes has resulted in the establishment of breeding populations of exotic fishes in freshwater water bodies worldwide [1][2][3][4]. Introduced fishes may compete directly with native fishes, and several invasive fish species have been demonstrated to exert negative ecological effects following invasion [1][2][3]5]. Alien fishes can additionally act as carriers of non-indigenous parasites, which can negatively affect native hosts, as has been shown for several flatworm parasites of the class Monogenea Carus, 1863, whose short, one-host direct life-cycle facilitates establishment and invasion. Two compelling examples involve monogeneans from the genus Gyrodactylus von Nordmann, 1832 introduced along with their commercially-important fish hosts: Gyrodactylus salaris Malmberg, 1957, introduced in the mid-1970's from the Baltic Sea to Norway, which has caused a highly pathogenic epizootic among wild and farmed salmonid fishes [6]; and Gyrodactylus cichlidarum Paperna, 1968, a parasite which has achieved almost global distribution through the translocation of its "tilapia" (Oreochromis spp.) hosts and has caused mass mortalities of cultured fish stocks [7].
The ornamental fish trade is an important component of international commerce, with over 1.5 billion fish belonging to more than 4000 freshwater and 1400 marine species traded annually and generating revenue in excess of US $ 6 billion [8]. As is the case for species translocated for aquacultural purposes, anthropogenic introduction of exotic ornamental fishes has resulted in alien monogenean parasites causing epizootic outbreaks in confined fish populations, for instance in display aquaria [9]. Furthermore, monogeneans co-introduced to continental freshwater bodies with their exotic fish hosts can contribute to the loss of native fish biodiversity, following parasite host switches (HS) from the introduced, alien fish to indigenous fish hosts (reviewed in [10]). One group of ornamental fishes exemplifies the invasiveness of feral aquarium fishes and their role as vectors to monogenean parasites: the small, live-bearing freshwater fishes belonging to the cyprinodontiform family Poeciliidae, which were originally distributed in the Americas, Africa and Madagascar [11]. Poeciliids have been translocated worldwide with the aquarium trade (e.g. "guppies" and "mollies" from the genus Poecilia Bloch & Schneider, "swordtails" and "platys" from the genus Xiphophorus Heckel) and mosquito control programmes ("mosquito fish" from the genera Gambusia Poey and Poecilia), and now introduced, feral populations of exotic poeciliids are established in all continents except Antarctica [3]. Arguably, poeciliids are the most successful exotic fish family in Australia [12], a country with strictly enforced quarantine rules governing the import of live animals, with five ornamental species having established breeding populations in the continent: Eastern mosquito fish, Gambusia holbrooki Girard, green swordtail, Xiphophorus hellerii Heckel, Southern platyfish, Xiphophorus maculatus (Günther), guppy, Poecilia reticulata Peters, and dusky millions fish, Phalloceros caudimaculatus (Hensel) [13]. The same five poeciliid species plus mosquito fish, Gambusia affinis (Baird & Girard), have established reproductively in Indian continental waters [10]. In both cases, exotic poeciliid fishes have co-introduced their gyrodactylid parasites [10,13]; nonetheless, no evidence was found in Australia of gyrodactylids switching hosts from exotic poeciliid to native fishes. Trade of ornamental poeciliids has also contributed to the wide dissemination of their monogenean parasites: several gyrodactylid species have been recorded to infect aquarium fishes in Canada, UK, Czech Republic, Singapore, Korea, India, Australia, etc. [10,12,14]. In fact, five of the 19 species of Gyrodactylus known to infect poeciliid fishes were originally obtained from exotic fishes reared in aquaria (reviewed in [14]). Although their potential pathogenic effects on native fish populations have not been fully demonstrated, some gyrodactylids known to infect poeciliid fishes are remarkable for the extent of their geographical distribution following anthropogenic translocation, and for the range of host species they infect. For instance, Gyrodactylus bullatarudis Turnbull, 1956 has been recorded from six poeciliid fish species from the genera Gambusia, Poecilia, Pseudoxiphophorus Bleeker (syn. Heterandria), and Xiphophorus, infecting wild, captive and feral populations in North America, Europe and Australia (reviewed in [14]), and has also been shown to infect the non-poeciliid, cyprinodontiform killifish, Anablepsoides hartii (Boulenger) (syn. Rivulus hartii) [15].
Goodeid topminnows or splitfins are a family of livebearing cyprinodontoid fishes that evolved in the MC, and occur in Central Mexico and North America [17]; the subfamilies Goodeinae and Empetrichthyinae are principally distributed across these two geographic regions, respectively [19]. About 45 species from the MC are classified within the sub-family Goodeinae [20]. Goodeines were originally distributed in drainages along both the Pacific Ocean and Gulf of Mexico slopes; unfortunately, the MC is one of the most densely populated regions in Mexico and habitat destruction, water pollution, and introduction of exotic species have pushed several endemic fishes to the brink of extinction, with some goodeid species effectively extirpated from their natural habitats [21]. Only two goodeid species distributed in Central Mexico can be considered to be in no conservation risk category: the blackfin goodea Goodea atripinnis Jordan, and the dark-edged splitfin Girardinichthys multiradiatus (Meek) [22]. Goodea atripinnis remains common in many areas and is probably still the most abundant goodeid species overall, but its distribution and abundance have steadily declined over the last 30 years [20,21,23].
Poeciliids and goodeids are morphologically similar viviparous, cyprinodontiform fishes that exhibit comparable habitat use [17], and are now sympatric following the human-mediated translocation of poeciliids to the MC. Exotic poeciliids have been shown to modify the ecological structure, function and native species abundance of water bodies following invasion [1], which is usually attributed to competition for habitat and/or food [3]. Additionally, invasive fishes may prey on native species, as shown for Xiphophorus spp., which feed on eggs and juveniles and have been implicated in the extinction of the goodeids golden skiffia, Skiffia francesae Kingston, and banded allotoca, Allotoca goslinei Smith & Miller [5]. A further negative impact of invasive poeciliids that has been recorded in the MC is the heterospecific harassment of native goodeids; in particular, male guppies (P. reticulata) have been recorded to attempt forced copulations of female twoline skiffia Skiffia bilineata (Bean) [2].

Skiffia lermae
Meek Mich [24] of Gyrodactylus had switched host from introduced, alien poeciliids to endemic G. atripinnis. This was considered given that, as mentioned above, some species of Gyrodactylus have been shown to be extremely invasive following co-introduction with their fish hosts; and that poeciliids and goodeids are now sympatric in much of the MC and are likely to interact as they are morphologically and ecologically similar. We hypothesised HS were more likely from poeciliid to goodeid fishes than vice versa, because the first are very successful invaders and generally outnumber the endemic, endangered goodeids [4,16], although evidently HS could occur in both directions. Potential HS from poeciliids to goodeids could in principle be recognised, considering that all 19 gyrodactylid species known to infect poeciliid fishes have been characterised morphologically [25], and parts of the genome have been sequenced for others [14]. To test the HS hypothesis, several localities on rivers draining the MC into both the Gulf of Mexico and the Pacific Ocean were sampled, and gyrodactylid parasites recovered from both G. atripinnis and poeciliids were analysed by morphological and molecularbased approaches from those localities where this goodeid fish and poeciliid fishes were found in sympatry. Finally, analysing the gyrodactylid fauna of non-native poeciliid fishes in the MC enabled a preliminary assessment of the enemy release hypothesis, which posits that upon translocation, introduced species lose some of their natural enemies such as parasites, thereby gaining a fitness advantage -a topic which was recently addressed studying the Gyrodactylus fauna of native and introduced minnow populations in Norway [27].

Specimen collection
Fishes were collected by electrofishing at different localities in the Río Pánuco and the Río Lerma basins, which flow from the MC into the Gulf of Mexico and the Pacific Ocean, respectively. Different species of goodeid [G. atripinnis, G. multiradiatus, S. bilineata and Xenotoca variata (Bean)] and poeciliid [P. bimaculatus, P. gracilis,  [28]. After euthanasia, fish were preserved in groups separated by species and locality, and placed in labelled plastic bottles containing 96% ethanol until they were screened microscopically.

Specimen preparation
Fish fins, body and gills were inspected under a dissection microscope, and parasites found were dislodged gently using acupuncture needles. Representative gyrodactylid specimens were prepared as whole mounts in ammonium picrate glycerine following the procedure detailed by Malmberg [29] to study taxonomic features of the haptor (= opisthaptor, or terminal attachment organ), male copulatory organ (MCO) and pharynx. Further specimens had their haptors excised using a scalpel and were subjected to proteolytic digestion as described previously [30], to release the attachment hooks from enclosing tissue. The corresponding anterior portions of bisected bodies were stored at -20°C in 96% ethanol, individually labelled for subsequent molecular analyses. The hooks were mounted in a semi-permanent 1:1 formalin : glycerin solution and the edges of the coverslip were then sealed with the permanent mounting medium Pertex (Histolab Products AB, Gothenburg, Sweden).

Morphological analysis
For the morphological study, images of the haptoral attachment hooks of proteolytically-digested specimens were captured using a Zeiss AxioCam MRc digital camera interfacing with an Olympus BH2 compound microscope using a ×0. 75  GmbH, 1997) image analysis software. The 24 morphometric measurements are given in micrometres or in degrees (for angles) as the range followed by the mean in parentheses, and were selected from those described in Malmberg [29], Shinn et al. [31] and Paladini et al. [32].
To describe and compare the aspect ratio of the hamuli of different Gyrodactylus spp., the quotient between the mean hamulus proximal shaft width (HPSW) and the hamulus total length (HTL) was calculated. Measurements used to calculate the aspect ratio of G. lamothei, G. mexicanus and those of gyrodactylids recovered from G. atripinnis are new. Morphometric data of Gyrodactylus spp. infecting poeciliid fishes were taken from [25] and [14].

Molecular analyses
Individual genomic DNA of 17 ethanol-fixed bodies of excised specimens of Gyrodactylus spp. collected from G. atripinnis, and of three worms collected from G. multiradiatus were extracted using a DNeasy® Blood & Tissue Kit (Qiagen, Valencia, California, USA) following the manufacturer's instructions. The target region spanning the 3′ end of the 18S rRNA gene, ITS1, 5.8S rRNA gene, ITS2 and the 5′ end of the 28S rRNA gene subunit failed to amplify using the primers employed previously by Rubio-Godoy et al. [33]. New primers ITS1-fm 5′-TAG AGG AAG TAC AAG TCG-3′ and ITS2-rm 5′-CGC TYG AAT CGA GGT CAG GAC-3′ were designed to facilitate amplification of the target using the following PCR conditions: initial denaturation for 4 min at 95°C followed by 35 cycles of: 94°C for 30 s, 54°C for 45 s, 72°C for 1 min, with a final extension at 72°C for 7 min. PCR amplicons were visualised on GelRed (Biotium, San Francisco, California, USA) stained 1% agarose gels and then purified with ExoSap-IT (USB Corporation, Cleveland, Ohio, USA). Sequencing reactions were carried out with the use of Big Dye Terminator chemistry, incorporating the same primers as those used in PCR, and cleaned by filtration with Sephadex G50. The sequenced products were read on an ABI PRISM 3100 automated DNA sequencer (Applied Biosystems, Foster City, California, USA). Electropherograms were visually inspected with the use of FinchTV (Geospiza Inc., Seattle, Washington, USA), and overlapping fragments of forward and reverse sequences were assembled with the use of the computer program  [36]. Phylogenetic hypotheses of ITS sequences were inferred with maximum likelihood (ML) and Bayesian inference (BI). Prior to ML and BI analyses, the optimal model of evolution and parameter settings for the ITS dataset was selected in jModeltest 2.1.10 [37,38] using the Akaike information criterion (AIC) [39] and the TVM+I+G model was selected. The likelihood program GARLI 2.0 [40] was used for topology reconstruction under a GTR model, allowing the program to estimate the I and G parameters. Two independent likelihood analyses were conducted for each data set to ensure convergence. Analyses were terminated after 100,000 generations, with an additional 1000 bootstrap replicates used for evaluating nodal support; each replicate was terminated after 10,000 generations. Bayesian phylogenetic relationships were inferred using MrBayes v.3.2.1 [41], running 10,000,000 generations and sampling one tree every 100 generations. Parameter settings used were nst = 6 and rates = invgamma. Posterior probabilities for supported clades were determined by a 50% majority-rule consensus of the 180,002 trees retained after 10% 'burn-in'. Trees resulting of ML and BI analyses were rooted with G. pakan and G. teken, parasites of Astyanax aeneus, which occur sympatrically with poeciliids in their native distribution range in Veracruz. Finally, estimates of evolutionary divergence between ITS1, 5.8S and ITS2 sequences (number of base substitutions per site between sequence pairs) among species of Gyrodactylus parasitising goodeid and poeciliid fishes were obtained with MEGA7 [42]. All positions containing gaps and missing data were eliminated; there was a total of 677 positions in the final dataset.

Results
Goodeid and poeciliid fishes were found to be sympatric in several locations in both the Atlantic Ocean and Pacific Ocean watersheds of Mexico. In general, invasive poeciliids were more numerous than endemic goodeids at localities sampled in the MC (Table 3). In the current study, only cases where gyrodactylids were recovered from blackfin goodea, G. atripinnis in localities where both goodeid and poeciliid fishes were collected simultaneously are presented ( Table 3). As reported previously [14], poeciliids (P. mexicana) captured within their native distribution range in the Río Pánuco basin on the Gulf of Mexico slope were infected with G. pseudobullatarudis and G. xtachuna. None of the gyrodactylid species known to infect poeciliids were found on invasive fishes at the localities sampled within the MC (Table 3). Nonetheless, two invasive poeciliid species (P. mexicana and P. bimaculatus) collected at Araro, Michoacán, within the Río Lerma basin were infected with the cichlid fish pathogen G. cichlidarum (García-Vásquez A, Razo-Mendivil U, Rubio-Godoy M, unpublished). Further, undescribed species of Gyrodactylus were found infecting poeciliid fishes on localities draining into both sides of the continental divide (Table 2); these will be described following collection of further samples. Specimens of G. lamothei were recovered from the goodeid G. multiradiatus collected at San Nicolás Peralta, Estado de México, a locality within the MC in the Río Lerma basin. No specimens of G. mexicanus were recovered.
Morphological and molecular analyses indicated that four undescribed species of Gyrodactylus infect the blackfin goodea G. atripinnis (Table 3). In the current study, only one species is described, i.e. the most abundant and geographically widespread gyrodactylid species, which was collected from both sides of the continental divide (Fig. 1). The other three undescribed species of Gyrodactylus (referred to as Gyrodactylus sp. 1, sp. 2 and sp. 3) were found in limited numbers (i.e. 1, 2 and 4 specimens, respectively) and in only a few localities (for details see Table 3); these latter species are not described taxonomically in the present work. The findings of the study, however, broadly illustrate that the three undescribed species differ morphologically from other known gyrodactylids infecting goodeid and poeciliid fishes; and molecular data obtained from them are presented, because given the scarcity of genetic data on Mexican Gyrodactylus spp., these new sequences inform the phylogenetic relationships between gyrodactylid parasites infecting goodeid and poeciliid fishes. A summary of the specimens prepared from each locality and analysed by morphological and molecular means is presented in Table 3.

Remarks
Gyrodactylus tomahuac n. sp. is the third gyrodactylid described from G. atripinnis. This host has previously been found to harbour G. lamothei and G. mexicanus [24], as well as undescribed species of Gyrodactylus (Table 2). At the type-locality, Vega de Ramírez, Río Moctezuma, Querétaro, the only gyrodactylid species recorded was G. tomahuac n. sp., with 41 parasites found on ten hosts (Table 3). At other localities, G. tomahuac n. sp. burdens were lower; and G. tomahuac n. sp. was found concurrently infecting the same hosts as Gyrodactylus sp. 3 in two localities in the Río Lerma basin (Table 3). Gyrodactylus tomahuac n. sp. was found in four different localities in the Mexican highlands ( Fig. 1) at considerable distances from each other: Vega de Ramírez, Río Moctezuma is c.150 km apart from the other three locations in the Río Lerma basin where the parasite was recorded (i.e. El Fresno, San Miguel Tlaxcaltepec, and Araro), and these three sites are c.60 km apart from each other. More importantly, the parasite was recorded in river basins flowing into both directions of the continental divide: the type-locality (Vega de Ramírez) at the Río Moctezuma is within the Río Pánuco basin, which drains into the Gulf of Mexico; while streams at El Fresno, San Miguel Tlaxcaltepec and Araro are part of the Río Lerma-Santiago basin and flow into the Pacific Ocean. The morphology of the marginal hooks is considered the most informative character to differentiate between species of Gyrodactylus. The marginal hooks of G. tomahuac n. sp. clearly differ in shape from those of both G. lamothei and G. mexicanus (Fig. 3): the marginal hooks of G. lamothei possess a very long and forward-tilted shaft whose tip ends well beyond the toe of a rounded, almost elliptical base, in contrast to G. tomahuac n. sp. where the marginal hook shaft is upright and does not extend beyond the angular base of the toe; and the marginal hooks of G. mexicanus have an upright shaft ending in a downward-curved point almost as long as the shaft itself, and the marginal hook base has an extended bridge ending in a rounded toe, in contrast to the comparatively short shaft point and short-bridged, angular toe in G. tomahuac n. sp. Superficially, the marginal hooks of G. tomahuac n. sp. resemble those of two gyrodactylids infecting poeciliid fishes, G. bullatarudis and G. takoke (Fig. 4); however, features of the marginal hook base allow their discrimination. While the marginal hook base in G. tomahuac n. sp. has a clear bridge almost level with the union of the heel and the shaft which forms a sharp angle with an almost straight, downward inclined  Gyrodactylus tomahuac n. sp. can also be differentiated from gyrodactylids known to infect goodeid and poeciliid fishes, comparing the overall morphology of their haptoral sclerites. Based on the variation in the haptoral hook morphology of the Gyrodactylus species infecting goodeid fishes (Fig. 3), two groups can be roughly identified: G. lamothei and Gyrodactylus sp. 2 both have relatively slender hamuli and marginal hooks with forward-angled shafts extending well beyond the toe, while G. tomahuac n. sp., G. mexicanus and both Gyrodactylus sp. 1 and sp. 3 possess robust hooks, with comparatively wider hamulus proximal shafts (as shown by the ratio between the mean hamulus proximal shaft width (HPSW) to the hamulus total length (HTL; Table 5), and marginal hooks with upright, shorter sickle shafts. Further, G. tomahuac n. sp. can easily be distinguished from G. lamothei, G. mexicanus and Gyrodactylus sp. 2, as these have elongated sclerotised structures attached to the root of the hamuli (Fig. 3). The haptoral hook morphology of G. tomahuac n. sp. is quite distinctive from that of most gyrodactylid species infecting poeciliid fishes (Fig. 5), except for perhaps G. bullatarudis, G. gambusiae Rogers & Wellborn, 1965 and G. pictae, all of which have robust hamuli (Table 5); albeit these species can be easily separated from G. tomahuac n. sp. because they all have longer hamuli (HTL > 50 μm vs c.45 μm) and since the ventral bars of these gyrodactylids infecting poeciliid fishes are  Table 5), but the majority possess slender hamuli.

Phylogenetic analyses
The length of the new ITS sequences of 14 specimens of Gyrodactylus spp. (seven G. tomahuac n. sp., one Gyrodactylus sp. 1, two Gyrodactylus sp. 2, and four Gyrodactylus sp. 3) collected from G. atripinnis varied from 866 to 934 bp. ITS sequences obtained for G. lamothei were deposited in GenBank (GenBank acc. nos. KX555666-KX555668), and varied from 887 to 888 bp. One representative sequence of each of these mentioned taxa was compared with those downloaded from GenBank, producing an alignment composed by 24 sequences with a length of 1324 bp. Phylogenetic analyses did not include G. mexicanus, because no ITS sequences are available in GenBank for this species and we did not recover any specimens during the current survey.
ML and BI analyses of ITS sequences recovered identical phylogenetic relationships among the analysed species, with several well-supported nodes. ML analysis resulted in one tree (not shown) with a value of likelihood = -10,210.6007. The 50% majority-rule consensus tree (Fig. 6) obtained from the BI analysis recovered three main clades, in which the sequences of Gyrodactylus spp. infecting goodeid and poeciliid fishes do not constitute monophyletic assemblages; and the gyrodactylids from goodeid fishes appeared in two clades constituting five different lineages.
Sequences of G. tomahuac n. sp., Gyrodactylus sp. 1 and Gyrodactylus sp. 3 formed a main clade with very low branch support (0.5321 of pp). This clade is composed by one terminal branch (Gyrodactylus sp. 1) and one derived group containing G. tomahuac n. sp. and Gyrodactylus sp. 3 ( Fig. 6; red box). A second main clade contains two subclades, one of them constituted by the goodeid-infecting parasites Gyrodactylus sp. 2 and G. lamothei (red and yellow boxes), which appear as sister group to a cluster of mostly well-supported derived branches containing 11 species of Gyrodactylus infecting poeciliid fishes (green box). Finally, a third main clade has one terminal and two derived branches containing parasites infecting several non-related fish hosts: G. gondae infecting Lozano's goby Pomatoschistus lozanoi (de Buen); G. stephanus infecting common mummichogs Fundulus heteroclitus (L.); G. arcuatus infecting threespined sticklebacks Gasterosteus aculeatus L.; and two gyrodactylids from poeciliids, G. xalapensis and G. takoke (blue box).

Intra-and inter-specific nucleotide variation
Comparison of ITS1, 5.8S and ITS2 sequences among the 24 species analysed showed a great amount of nucleotide variation, ranging from 0.4 to 32.7%. The species pairs that showed these contrasting values were, respectively, G. lamothei and Gyrodactylus sp. 2, and both G. arcuatus and Gyrodactylus sp. 2, as well as G. jarocho and G. stephanus.
Likewise, new sequences of the gyrodactylids collected from G. atripinnis, representing G. tomahuac n. sp. and three unnamed species (Gyrodactylus sp. 1, sp. 2, and sp. 3) exhibited high sequence variation (Table 6). Nucleotide differences ranging from 6.8 to 31.3% were found among the 14 sequences of Gyrodactylus spp. infecting this goodeid fish. Values of nucleotide variation between the new species, G. tomahuac, and the other three unnamed species (Gyrodactylus sp. 1, sp. 2, and sp. 3) were 25.6 to 25.9%, 24.0 to 24.2%, and 6.8 to 7.1%, respectively. Intra-specific nucleotide variation was detected in G. tomahuac n. sp. and in Gyrodactylus sp. 3. Of the seven sequenced specimens of G. tomahuac n. sp. and four Gyrodactylus sp. 3, four and one specimens showed nucleotide variation, which ranged from 0.13 to 0.27%, and from 0.12 to 0.13%, respectively. Intra-specific nucleotide variation ranging from 0.13 to 0.45% was detected in the three G. lamothei specimens sequenced.

Discussion
The first objective of this study was to characterise the gyrodactylid fauna of G. atripinnis within its native distribution range in the MC; and in doing so, assess whether HS have occurred to this endemic host from alien poeciliid fishes, which have been extensively introduced to central Mexico.  No evidence was found of gyrodactylid HS from poeciliids to goodeids, nor vice versa, in the six localities where fishes of these families occurred sympatrically. In all instances, poeciliid fishes were more abundant than goodeid fishes (Table 3), and the former included species known to harbour several gyrodactylids: P. mexicana has the richest gyrodactylid fauna associated with poeciliids, as this host has been recorded to harbour six parasite species in Mexico; P. bimaculatus is a translocated species known to be infected by four gyrodactylid species; and the invasive P. reticulata has been registered as host to three gyrodactylids [14]. In their native Río Pánuco basin, P. mexicana were infected by G. pseudobullatarudis and G. xtachuna, two gyrodactylid species known to have a wide geographical and host range [14]. In the MC, poeciliid fishes were not infected by any of the gyrodactylids known to parasitise them in their native distribution range [14,25], some of which have been widely disseminated through the ornamental aquarium trade [10,13]. Although the current study was not designed to test the enemy release hypothesis, results suggest that upon translocation to the MC, poeciliid fishes have lost their gyrodactylid parasite fauna; however, this finding is based on the limited sample sizes of poeciliid fishes found sympatrically with G. atripinnis in a few localities only, and are by no means representative of invasive poeciliid populations established in the MC. Surprisingly, invasive P. mexicana and P. bimaculatus collected in the MC were found to carry the alien tilapia parasite G. cichlidarum (García-Vásquez A, Razo-Mendivil U, Rubio-Godoy M, unpublished).
Goodea atripinnis was found to be infected by G. tomahuac n. sp. and three undescribed Gyrodactylus Fig. 6 Phylogenetic tree of different Gyrodactylus species infecting teleost fishes. Shaded boxes indicate parasite clades infecting the following host groups: Goodea atripinnis (red), other goodeid fishes (yellow), and poeciliid fishes (green and blue). Fifty percent majority-rule consensus tree obtained from the Bayesian Inference analysis of the ITS1, 5.8S and ITS2 sequences from 23 species of Gyrodactylus. Bayesian posterior probabilities are shown above the nodes, and ML bootstrap support below the nodes. The phylogram is rooted with gyrodactylids infecting the characid Astyanax aeneus spp. This host was previously recorded to be infected by G. lamothei and G. mexicanum [24], suggesting that this endemic fish is parasitised by at least six gyrodactylids. The helminth fauna of goodeid fishes is wellcharacterised [24], and G. atripinnis is the host species known to harbour most parasites: nine digeneans, five nematodes, four cestodes, three monogeneans (including G. lamothei and G. mexicanum) and one acanthocephalan are recorded to infect this host. Considering six gyrodactylid species as parasites of this host, monogeneans represent the second most numerous helminth group known to infect this endemic fish. Bearing in mind that several reports have been made of the presence of undescribed Gyrodactylus sp. infecting G. atripinnis and several other goodeid fishes (Table 2); and the fact that along with G. multiradiatus, G. atripinnis remains a common and relatively abundant fish in much of the MC [22], it is foreseeable that several new species of Gyrodactylus infecting goodeid fishes will be found, potentially making monogeneans the most species-rich group of helminths infecting this host group.
Regarding the characterisation of the gyrodactylid fauna of G. atripinnis and other goodeid fishes, both morphological and molecular data support the hypothesis that two distinct groups of parasites infect these native fishes. Morphologically, some gyrodactylids possess robust hamuli, while others have slender hamuli with sclerotised plates located at the base of these sclerites. Very robust hamuli are the most salient feature of G. tomahuac n. sp., the species described here, a characteristic shared with G. mexicanus, and to a slightly lesser degree with the undescribed Gyrodactylus sp. 1 and Gyrodactylus sp. 3, both of which have stout hamuli (Table 5). Both G. lamothei and the undescribed Gyrodactylus sp. 2 possess slender hamuli, with elongated plates located at their roots; these plates were also drawn in the description of G. mexicanus [26]. It would be interesting to ascertain whether these structures are similar to the sclerotised plates found in Gyrodactylus proterorhini Ergens, 1967 [45], and to establish the phylogenetic relationships of G. mexicanus to other gyrodactylids once molecular markers are available.
The phylogenetic tree presented here is the most complete hypothesis of the relationships between gyrodactylids infecting goodeid fishes; to date, the only molecular data available for these parasites are those presented in this work (Fig. 6). Based on this hypothesis, Gyrodactylus spp. infecting goodeid fishes do not constitute a monophyletic group, as these parasites appear in two clades, which correspond to the two groups formed based on the morphology of the hamuli. The first, not well-supported clade includes G. tomahuac n. sp. as sister species to the undescribed Gyrodactylus sp. 3, and Gyrodactylus sp. 1 as a terminal branch; these three taxa all have relatively robust hamuli, albeit their marginal hooks differ (Fig. 3). Gyrodactylus lamothei and Gyrodactylus sp. 2 both have slender hamuli with plates at their roots, and appear in a separate clade as sister taxa to 11 gyrodactylid species known to infect poeciliid fishes, most of which also possess slender hamuli. Two gyrodactylids known to infect poeciliid fishes, G. takoke and G. xalapensis, appear in a further clade, which includes parasites infecting a variety of non-related fish hosts. Molecular characterisation of further gyrodactylids infecting goodeid and poeciliid fishes should shed light on the phylogenetic hypotheses outlined here: that two morphologically-distinct parasite lineages infect goodeids, one of which shares a common ancestor with several gyrodactylids infecting poeciliids.

Conclusions
No evidence was found of gyrodactylids switching host from introduced poeciliid fishes to endemic goodeid fishes in the MC; nor vice versa. Goodea atripinnis harbours at least six species of Gyrodactylus, including G. tomahuac n. sp., which is described here. Morphological and molecular data support the hypothesis that two groups of gyrodactylids infect goodeid fishes: one contains worms which possess robust hamuli including G. tomahuac n. sp., and the other encompasses parasites with slender hamuli and sclerotised plates at their roots such as G. lamothei -and phylogenetic analyses suggest that this second group shares a common ancestor with gyrodactylids infecting poeciliid fishes. Monogeneans may be the most species-rich group of helminth parasites infecting goodeid fishes.