INTRODUCTION
Following both the exponential increase in global movement of people within the past few decades and the global trade or transport of many plants and animals, the introduction of non-native species into new biogeographic areas has been considerably accelerated. Alien species may disrupt the delicate balance of the ecosystem and therefore cause drastic or irremediable changes to environments (Vitousek et al. Reference Vitousek, Mooney, Lubchenco and Metillo1997; Mooney and Cleland, Reference Mooney and Cleland2001). Accordingly, intentional or accidental biological invasions are considered a major threat to biodiversity second only to the destruction of natural environments (Vitousek et al. Reference Vitousek, Mooney, Lubchenco and Metillo1997). The success of introduced species depends on niche availability (Shea and Chesson, Reference Shea and Chesson2002), which may be driven by a combination of non-exclusive factors like competitive exclusion of resident species (Salo et al. Reference Salo, Korpimäki, Banks, Nordström and Dickman2007), lack of natural predators and/or enemies (Torchin et al. Reference Torchin, Lafferty and Kuris2001, Reference Torchin, Lafferty, Dobson, McKenzie and Kuris2003; Keane and Crawley, Reference Keane and Crawley2002; Clay, Reference Clay2003; Mitchell and Power, Reference Mitchell and Power2003), hybridization and introgression (Huxel, Reference Huxel1999; Ellstrand and Schierenbeck, Reference Ellstrand and Schierenbeck2000). In the context of biological invasions, parasitism must be also considered. Invaders may lose their parasites and subsequently outperform native species in their home range, or they may transmit their own parasites to naive host species (i.e. species not infected by exotic parasites), which in turn may have detrimental effects on survival rates of native species (Daszak et al. Reference Daszak, Cunningham and Hyatt2000; Anderson et al. Reference Anderson, Cunningham, Patel, Morales, Epstein and Daszak2004). This has been extensively documented from studies following translocation of vertebrate host-parasite complexes (e.g. Daszak et al. Reference Daszak, Berger, Cunningham, Hyatt, Green and Speare1999; Tompkins et al. Reference Tompkins, Sainsbury, Nettleton, Buxton and Gurnell2002; Taraschewski, Reference Taraschewski2006). Invasive parasites may, therefore, play a key role in animal and plant invasions (Prenter et al. Reference Prenter, MacNeil, Dick and Dunn2004).
The American red-eared slider, Trachemys scripta elegans, has been the most popular pet among turtles in the second half of the twentieth century. This and other American turtle species from the genera Apalone, Graptemys, Pseudemys and Chrysemys have been exported worldwide, especially to Asian and European markets. According to Telecky (Reference Telecky2001), about 52 million sliders were exported between 1989 and 1997. Because sliders can grow rapidly and attain a large size as adults (carapace length up to 300 mm), they become less attractive as pets. As a result owners, not aware of the environmental risk, have released them into the wild. T. s. elegans is now widespread in natural wetlands all over Western Europe and Asia (France: Servan and Arvy, Reference Servan and Arvy1997; Arvy and Servan, Reference Arvy and Servan1998; Spain: Da Silva and Blasco, Reference Da Silva and Blasco1995; Alarcos Izquierdo et al. Reference Alarcos Izquierdo, Flechoso del Cuerto, Rodríguez-Pereira and Lizana Avia2010; Valdeón et al. Reference Valdeón, Crespo-Diaz, Egaňa-Callejo and Gosá2010; Italy: Ficetola et al. Reference Ficetola, Thuiller and Padoa-Schioppa2009; Asia: Chen, Reference Chen, Koike, Clout, Kawamichi, De Poorter and Iwatsuki2006; Ramsay et al. Reference Ramsay, Ng, O'riordan, Chou and Gherardi2007) and is in fact considered as one of the worst invasive species (see the Global Invasive Species Database: http://www.issg.org/database/welcome/). Numerous studies have indeed reported that the invasive red-eared sliders are able to breed successfully in their new habitats (e.g. Arvy and Servan, Reference Arvy and Servan1998; Cadi et al. Reference Cadi, Delmas, Prévot-Julliard, Joly, Pieau and Girondot2004; Ficetola et al. Reference Ficetola, Monti and Padoa-Schioppa2002, Reference Ficetola, Thuiller and Padoa-Schioppa2009; Perez-Santigosa et al. Reference Perez-Santigosa, Diaz-Paniagua and Hidalgo-Vila2008; Kikillus et al. Reference Kikillus, Hare and Hartley2009) with the consequent threat of outcompeting indigenous species (Cadi et al. Reference Cadi, Delmas, Prévot-Julliard, Joly, Pieau and Girondot2004).
Two European freshwater terrapins are potentially endangered by the American slider, namely the European pond turtle Emys orbicularis and the Mediterranean turtle Mauremys leprosa. Both are listed in the Annexes II and IV of the European Union habitats directive and as “near threatened” in the IUCN Red List of Threatened SpeciesTM for E. orbicularis and “in danger” in the Liste Rouge de l'IUCN Français for M. leprosa. E. orbicularis is primarily distributed in the European and North African countries surrounding the Mediterranean Sea, while M. leprosa is mainly found in countries of North Africa, the Iberian Peninsula and Southern France (Bonin et al. Reference Bonin, Devaux and Dupré1998). Only recently have questions been addressed on the actual impact of American turtles on wild resident populations (Cadi and Joly, Reference Cadi and Joly2003, Reference Cadi and Joly2004; Polo-Cavia et al. Reference Polo-Cavia, López and Martín2008, Reference Polo-Cavia, López and Martín2009a,Reference Polo-Cavia, López and Martínb, Reference Polo-Cavia, Gonzalo, López and Martín2010a,Reference Polo-Cavia, López and Martínb; Reference Polo-Cavia, López and Martín2011). Some studies have shown that the exotic turtles were more competitive than E. orbicularis and M. leprosa in the use of basking sites within experimental pools (Cadi and Joly, Reference Cadi and Joly2003; Polo-Cavia et al. Reference Polo-Cavia, López and Martín2010b). A negative impact on the weight variations and survival rates of the European pond turtles was also demonstrated under experimental conditions in the presence of T. s. elegans (Cadi and Joly, Reference Cadi and Joly2004). Similarly, experimental results suggested that chemical cues released from invasive species could modify adversely the behaviour of M. leprosa (see Polo-Cavia et al. Reference Polo-Cavia, López and Martín2009a). Polo-Cavia et al. (Reference Polo-Cavia, López and Martín2009a) concluded that exotic sliders could ultimately contribute to the displacement of endemic turtles in natural environments. However, Cadi and Joly (Reference Cadi and Joly2004) did not exclude the possibility that E. orbicularis could be more sensitive to exotic pathogens transmitted from T. s. elegans, which would explain its lower fitness. This has been documented in Emys (formerly Clemmys) marmorata, an endemic endangered North American freshwater turtle, in which the herpes-like virus transmitted from introduced captive exotic turtles may be responsible of the decline of some populations (Hays et al. Reference Hays, McAllister, Richardson and Stinson1999) since that virus was able to kill captive individuals of the endemic turtle (Frye et al. Reference Frye, Oshiro, Dutra and Carney1977).
Trachemys s. elegans may carry its own parasites when released into natural environments but, to the best of our knowledge, only a single study assessed parasite transmission from American invaders to native European turtles and found transmission from indigenous to exogenous species (Hidalgo-Vila et al. Reference Hidalgo-Vila, Díaz-Paniagua, Ribas, Florencio, Pérez-Santigosa and Casanova2009). Among parasites reported from chelonians (Harper et al. Reference Harper, Hammond and Heuschele1982; Une et al. Reference Une, Uemura, Nakano, Kamiie, Ishibashi and Nomura1999; Du Preez and Lim, Reference Du Preez and Lim2000; Pasmans et al. Reference Pasmans, De Herdt, Dewulf and Haesebrouck2002; Eiras, Reference Eiras2005; Segade et al. Reference Segade, Crespo, Ayres, Cordero, Arias, García-Estévez and Iglesias Blanco2006; Hidalgo-Vila et al. Reference Hidalgo-Vila, Díaz-Paniagua, de Frutos-Escobar, Jiménez-Martínez and Pérez-Santigosa2007, Reference Hidalgo-Vila, Díaz-Paniagua, Pérez-Santigosa, de Frutos-Escobar and Herrero-Herrero2008, Reference Hidalgo-Vila, Díaz-Paniagua, Ribas, Florencio, Pérez-Santigosa and Casanova2009; Mihalca et al. Reference Mihalca, Gherman, Ghira and Cozma2007, Reference Mihalca, Racka, Gherman and Lonescu2008), species of the Polystomatidae (Platyhelminthes, Monogenea) are widespread among amphibians and freshwater turtles (Verneau, Reference Verneau2004). Polystomatids arose early in the course of vertebrate evolution and dispersed to ancestral freshwater chelonians in the Upper Triassic (Verneau et al. Reference Verneau, Bentz, Sinnappah, Du Preez, Whittington and Combes2002). Chelonian polystomes are divided into three genera, Polystomoides, Polystomoidella and Neopolystoma, based on the number of hamuli located between the posterior pair of suckers on the opisthaptor: two pairs, one pair and none, respectively. These parasites have a direct life cycle with free swimming infective larvae, i.e. oncomiracidia, and are mostly host and site specific (Verneau, Reference Verneau2004). Different polystome species have been recorded from the same chelonian host in three different microhabitats: the urinary bladder and cloaca, the conjunctival sacs under the eyelids, or the pharyngeal cavity, as it is the case, for instance, in the Southeast Asian box turtle Cuora amboinensis (Rohde, Reference Rohde1963, Reference Rohde1965; Richardson and Brooks, Reference Richarson and Brooks1987; Du Preez and Lim, Reference Du Preez and Lim2000). However Littlewood et al. (Reference Littlewood, Rohde and Clough1997) showed no evidence for intra-host speciation from a phylogenetic analysis. Because polystomes have been described from American terrapins in their native range (e.g. Wright, Reference Wright1879; Stunkard, Reference Stunkard1916, Reference Stunkard1924; Harwood, Reference Harwood1932; Platt, Reference Platt2000) as well as from wild populations of M. leprosa and E. orbicularis in Maghreb and Europe (e.g. Rudolphi, Reference Rudolphi1819; Combes and Ktari, Reference Combes and Ktari1976; Gonzales and Mishra, Reference Gonzales and Mishra1977; Knoeppfler and Combes, Reference Knoepffler and Combes1977; Mishra and Gonzalez, Reference Mishra and Gonzalez1978; Combes and Thierry, Reference Combes and Thiery1983), they are a good model to study parasitic transmission from the potentially invasive chelonian species. Our primary objective was to search for such a transfer within captive turtles of a Turtle Farm in Southern France where introduced American species such as Apalone spinifera, Chrysemys picta marginata, Graptemys pseudogeographica, T. s. elegans and T. s. scripta occur with the two native species in artificial pools. Due to the endangered status of both indigenous terrapins, the search of polystomes was conducted mainly from a non-invasive approach that relies on the presence of parasite eggs collected from infected turtles, and to a lesser extent from dissecting turtles. Phylogenetic systematics of polystomes recovered from captive animals was inferred from DNA sequences of the fast evolving gene, the cytochrome c oxidase I (COI), and from subsequent genetic comparisons with polystome species sampled from wild animals in their home range.
MATERIALS AND METHODS
Host sampling
Wild populations of E. orbicularis and M. leprosa were sampled from the province Languedoc-Roussillon in the South of France. Field surveys were conducted yearly from 2006 to 2009 in a small pond in the village of Leucate (42°50′32.21″N; 3°02′13.31″E) for E. orbicularis, and during 2008 and 2009 in a small pond in Canet-en-Roussillon (42°42′03.83″N; 3°01′18.91″E) and in the Tech River close to Le Boulou (42°32′02.12″N; 2°50′54.43″E) respectively for M. leprosa. Turtles were trapped overnight using catfish traps baited with fish and pork liver. Captured individuals were individually marked by a combination of cuts in peripheral scuts of the carapace following the procedure of the “Conservatoire des Espaces Naturels du Languedoc-Roussillon” for Capture-Mark-Recapture studies. Specimens of E. orbicularis, M. leprosa, T. s. elegans, T. s. scripta and A. spinifera were collected with landing nets in the Turtle Farm of Sorède (42°30′56.83″N; 2°57′26.76″E), mainly in 2009 and 2010. Ministerial authorisation numbers 06/71/AUT, 07/168/AUT and 09/247/DEROG for capture and sacrifice of E. orbicularis and M. leprosa, from March 2006 to November 2010, were obtained.
Parasite sampling
After capture, turtles were transported to the laboratory in Perpignan and placed in individual plastic boxes with water to a depth of about 5 cm. Polystome eggs were collected over three consecutive days by pouring the water through a set of sieves of 500 μm and 100 μm, respectively. The coarse material was collected on the 500 μm sieve while fine debris and eggs were collected on the 100 μm sieve. The contents of this sieve were then washed into a Petri dish and observed using a dissecting microscope. Polystome eggs were separated according to their shape, pipetted out and preserved in 70% ethanol until DNA extraction. Polystomes located in the urinary bladder or pharyngeal cavity release pear-shaped eggs, while parasites from the conjunctival sacs release diamond-shaped eggs (Fig. 1). After screening, all turtles were released in the exact location from which they were trapped except for two infected individuals of E. orbicularis from the natural pond of Leucate, and also two infected E. orbicularis and one individual each of M. leprosa and T. s. elegans from the Turtle Farm of Sorède which were dissected to study adult worms. Prior to dissection, turtles were euthanized by cardiac injection of 1 mL sodium pentabarbitone (Euthapent) diluted in 9 mL of luke warm water. Animals were dissected and the urinary bladder, accessory bladders and cloaca were removed intact. The head was severed well below the pharyngeal pouch and eyelids and nictitating membranes were lifted to enable examination of all crevices along the eye balls. The urinary bladder, the pharyngeal cavity, the conjunctival sacs and all potential sites were carefully examined using a dissecting microscope for the presence of polystomes. Adult worms were fixed under coverslip pressure in 10% neutral buffered formalin for morphological determination or in 70% ethanol for molecular analyses. All other adult polystomes used in this study were from our collections. Collaborators supplied some taxa while we collected others during field surveys conducted in many different countries and areas: e.g. Australia, Africa, Eurasia and South, Central and North America. Three other monogeneans infecting amphibian and fish species were used for outgroup comparison. GenBank sequences were already available for eight of these adult polystomes, the rest were sequenced as part of this study (see Table 1 for details).
Fig. 1. Polystome eggs recovered from infected turtles. Pear-shaped eggs are from parasites located either in the urinary bladder or in the pharyngeal cavity of their host whereas diamond-shaped eggs are exclusively released by parasites located in the conjunctival sacs.
Table 1. List of hosts and parasites investigated, geographical origin, source of polystomes, field and DNA samples and GenBank accession numbers
1: From Littlewood et al. (1997); 2: From Du Preez et al. (2007). TF means Turtle Farm; P means pear; D means diamond.
DNA Extraction, PCR and sequencing
Eggs and adult parasites were first ground with a micro-pestle and DNA extractions were carried out for one hour at 55°C in a final volume of 100 μL of Chelex 10% and proteinase K at 1 mg mL−1. Reactions were conducted at 100°C for 15 min and DNA samples were stored at 4°C until use for PCR. Amplification and purification of partial COI were done according to Verneau et al. (Reference Verneau, Du Preez, Laurent, Raharivololoniaina, Glaw and Vences2009) and Du Preez et al. (Reference Du Preez, Raharivololoniaina, Verneau and Vences2010), using the forward L-CO1p (5′-TTTTTTGGGCATCCTGAGGTTTAT-3′) and reverse H-COX1p2 (5′-TAAAGAAAGAACATAATGAAAATG-3′) primers (Littlewood et al. Reference Littlewood, Rohde and Clough1997). PCR products of about 440 bp were run in a 1% agarose gel and stained with ethidium bromide. DNA was purified using the Wizard SV Gel and PCR Clean-up System of Promega and sent to GATC (Biotech, France) for sequencing with both PCR primers. Sequences were edited and corrected with SequencherTM software (Gene Codes Corporation, Ann Arbor, Michigan, USA).
Phylogenetic analyses and molecular divergence level within species
Nucleic acid sequences from adult worms and polystome eggs were aligned using Clustal W (Thompson et al. Reference Thompson, Higgins and Gibson1994), which is implemented in MEGA version 4 (Tamura et al. Reference Tamura, Dudley, Nei and Kumar2007). DNA sequences were also translated to their corresponding protein sequences with the EMBOSS Transeq online software, following the alternative flatworm mitochondrial code, and aligned as before. The complete nucleic and amino acid sequence alignments were each subdivided in two sets of sequences depending on their origin. The first data-set comprised 35 haplotypes recovered exclusively from wild animals, while the second set included 55 haplotypes from both wild and captive turtles.
Maximum Likelihood (ML) analyses were performed on 345 nucleic acids characters without partitioning data-sets as follows. For the nucleic acids alignment including 35 haplotypes, a TVM + I + Γ model was selected by the Akaike Information Criterion (AIC) implemented in the program Modeltest 3.06 (Posada and Crandall, Reference Posada and Crandall1998), whereas a GTR + I + Γ was selected for the nucleic acids alignment including 55 haplotypes. Using these models, the search for the best ML trees was done following a heuristic procedure under the TBR branch swapping option with PAUP* 4.0b9 (Swofford, Reference Swofford2002). ML bootstrap support values were calculated with the same model of sequence evolution under the NNI branch swapping option after 500 replicates. Bayesian analyses were conducted using the software MrBayes 3.04b (Huelsenbeck and Ronquist, Reference Huelsenbeck and Ronquist2001), with four chains running for a million generations, sampling each 100 cycles. The Bayesian inferences were obtained using the selected models listed above for each data-set and Bayesian posterior probabilities were computed after removing the first 1,000 trees as the burn-in phase. Only ML analyses were conducted with the amino acid sequence alignment, which comprised 115 characters, using the PHYML online software (Guindon and Gascuel, Reference Guindon and Gascuel2003), under the amino acids substitution model LG and the NNI branch swapping option. Branch support values were performed after 500 replicates. Finally uncorrected pairwise genetic distances (p-distances) were estimated from the 55 nucleic acid haplotype sequences using MEGA in order to delineate the within species molecular divergence level.
RESULTS
Prevalence of infection in wild and captive turtles
Results obtained from polystome eggs investigation among specimens of E. orbicularis, M. leprosa, T. s. elegans, T. s. scripta and A. spinifera are summarized in Table 2. Among the wild 171 individuals of E. orbicularis captured between May 2006 and October 2009 in Leucate, all infected turtles produced pear-shaped eggs except one that presented diamond-shaped eggs. This specimen (Eol92) was dissected in 2007, and another (Eol48) in 2006. Three adult worms of Neopolystoma sp. were found in the urinary bladder of Eol48 and one in Eol92. No parasites were found in the pharyngeal cavity and, surprisingly, no parasites from the conjunctival sacs of Eol92 that released diamond-shaped eggs. Two infected E. orbicularis from the Turtle Farm were also dissected in 2006. The first specimen yielded three Neopolystoma sp. in the urinary bladder and 11 Polystomoides sp. in the pharyngeal cavity. The second specimen contained two Neopolystoma sp. from the conjunctival sacs. Similarly a single specimen of captive M. leprosa was dissected in 2006. It contained one Polystomoides sp. in the pharyngeal cavity and four Neopolystoma sp. in the conjunctival sacs. Finally one T. s. elegans dissected in 2009 in the Turtle Farm was infected with 46 specimens of Neopolystoma sp. in the urinary bladder and 46 individuals of Polystomoides sp. in the pharyngeal cavity.
Table 2. List of species investigated for polystome eggs and prevalence of parasite infection in wild and captive turtles
Organization of the haplotype diversity within polystomes
Sequences obtained from adult worms and polystome eggs were classified as corresponding haplotypes (from H1 to H52, Table 1). When adult worms had been previously identified, the species name is reported besides the haplotype in Table 1 and Figures 2a and 2b. Conversely, when worms were a new species, only the generic name is reported according to the number of hamuli (see Introduction). Haplotypes only found in polystome eggs are not assigned to a genus or species.
Fig. 2. a. Bayesian polystome phylogram resulting from analysis of 35 nucleic acid sequences (only polystomes from wild turtles). Species in red are from the urinary bladder, in blue from the conjunctival sacs and in green from the pharyngeal cavity. Boxes indicate species groups that were used to determine the molecular level of polystome species delineation. Values along branches indicate, from left to right, the Bayesian posterior probabilities and the ML bootstrap proportions resulting from analysis of nucleic and amino acids sequences, respectively. Letters A, B, C and D indicate polystome sub-lineages. N refers to Neopolystoma and P to Polystomoides. b. Bayesian polystome phylogram resulting from analysis of 55 nucleic acid sequences (polystomes collected from wild and captive turtles). Species in red are from the urinary bladder, in blue from the conjunctival sacs and in green from the pharyngeal cavity. Brown indicates polystome haplotypes recorded from captive animals. Values along branches indicate, from left to right, the Bayesian posterior probabilities and the ML bootstrap proportions resulting from analysis of nucleic and amino acids sequences, respectively. Letters A, B, C and D indicate polystome sub-lineages. N refers to Neopolystoma and P to Polystomoides. * indicates haplotype that is found from captive and wild turtles.
Phylogenetic relationships among polystomes
Bayesian phylograms obtained from the analyses of COI nucleic acid sequences with 35 and 55 haplotypes are shown in Figs. 2a and 2b, respectively. Because both topologies did not differ significantly from topologies inferred from ML analyses on nucleic and amino acid sequences, ML bootstrap values are given directly in Figs. 2a and 2b. Phylogenetic relationships among polystomes collected exclusively in wild animals show numerous polytomies (Fig. 2a). If we consider the basal placement of Polystomoidella sp1 (H23) as unresolved, thus six taxa and three sub-lineages (A, B and C) fall in a basal polytomy (Fig. 2a). Those six taxa, which infect turtles from disparate geographic localities (USA, Africa, Asia and Australia), and from distinct families of the suborders Pleurodira and Cryptodira, are found in the urinary bladder (in red, Fig. 2a), and in the conjunctival sacs (in blue) of their host. Clade A associates exclusively bladder parasites, but from geographically distant turtles (Asia and Australia) of distinct suborders and families. Similarly, clade B links only parasites of the conjunctival sacs, but from turtles of closely related families geographically less distant (Costa Rica and USA). Finally, clade C associates both parasites from the urinary bladder and pharyngeal cavity (in green). Whereas those parasites are from geographically distant turtles (Africa, Eurasia, South America and USA), all of them are from related families of the suborder Cryptodira. Within clade C, all North American bladder parasites form a monophyletic group (clade D). Concerning phylogenetic relationships among polystomes collected from wild and captive animals (Fig. 2b), additional haplotypes fall only in sub-lineages B and C and are tightly related to haplotypes previously recognized within polystomes of wild turtles.
Molecular species delineation within polystomes
Uncorrected pairwise divergences between haplotype sequences from eggs and adult worms collected in wild and captive turtles are reported in Appendix S1 (see Appendix S1, can be viewed at http://journals.cambridge.org/par). Adults of Neopolystoma orbiculare (see squared haplotypes H9 and H10, in Fig. 2a) diverge from each other by no more than 0·3%. The same level of divergence is also estimated between parasite worms belonging to the species Polystomoides oris (H11 and H12), and between specimens of Polystomoides sp4 (H50 and H51). A level of 0·9% is found between two adults belonging to Neopolystoma sp3 (H45 and H46). Similarly, sequences from worms or polystome eggs from M. leprosa in allopatry (Algeria and France), show divergences ranging from 0·9% to 1·7% (H25, H26 and H30). Conversely, a polystome collected from the urinary bladder of T. s. scripta in Florida (H48) differs by 1·7% to 2·0% from N. orbiculare individuals (H9 and H10) that were both sampled from the urinary bladder of C. p. marginata from Indiana. Polystomes collected from the pharyngeal cavity of T. s. elegans and T. s. scripta living in allopatry (H47 and H49) also differ by 2·6%. Summarizing, the genetic divergence within polystomes collected from the same body site across turtles of the same species, even in allopatry, ranges from 0 to 1·7%, whereas it ranges from 1·7% to more than 2·0% within polystomes collected from the same body site across turtles of different species or subspecies. These results suggest that the molecular species delineation can be fixed approximately to about 1·5%–2·0% divergence level in the COI, which is in accordance with the level found in amphibian polystomes by Du Preez et al. (Reference Du Preez, Verneau and Gross2007).
DISCUSSION
Modes of polystome diversification over geological times
From a sample of six chelonian polystome species, Littlewood et al. (Reference Littlewood, Rohde and Clough1997) showed that congeneric polystome species infecting the same body site from different host species were more related to each other than polystome species infecting different body sites of the same host species. Using a larger sample of polystome species recovered from geographically widespread chelonian host species belonging to different suborders and families, we demonstrate that patterns of polystome diversification are more complex (Fig. 2a). Our results show unambiguously that the two genera Neopolystoma and Polystomoides, based on the number of haptoral hamuli, are polyphyletic. Second, though our molecular marker could be highly saturated for tracking the phylogenetic route of chelonian polystomes since their origin in the Upper Triassic (Verneau et al. Reference Verneau, Bentz, Sinnappah, Du Preez, Whittington and Combes2002), the basal polytomy of parasites infecting either the urinary bladder or the conjunctival sacs of their host (Fig. 2a) confirms that polystomes could have originated and diverged very early. In fact, those parasites infect geographically distant turtles (Table 1) whose long isolation prevented host switching and speciation over recent geological time. The clustering of some parasites into clades that associate, respectively, polystomes of the urinary bladder (sub-lineage A, Fig. 2a) and polystomes of the conjunctival sacs (sub-lineage B, Fig. 2a) corroborates this hypothesis. Both clades, but mainly clade A, contains parasites from geographically and phylogenetically distant turtles. Third, the sub-lineage arising from the basal polytomy associates parasites infecting exclusively the urinary bladder and the pharyngeal cavity of their host (sub-lineage C, Fig. 2a). With the exception of Neopolystoma euzeti (H31), which is among the most basal taxa within that clade, all the remaining polystomes from the urinary bladder form a monophyletic group, suggesting a switch from the pharyngeal cavity to the urinary bladder. We can hypothesize that polystomes may have originated very early in ancestral chelonian hosts and specialized in particular body sites. During host evolution, they would have remained in their respective microhabitats and diversified following host speciation, but we demonstrate that host switching may also have occurred from one ecological niche to another, as exemplified for N. euzeti and the terminal clade of American bladder parasites (sub-lineage D, Fig. 2a).
Polystome species diversity in wild populations of M. leprosa and E. orbicularis
Polystomoides tunisiensis haplotypes (H25, H26 and H30) are reported in wild populations of M. leprosa in Algeria and France (Figs. 2a and 2b). Three other haplotypes (H27, H28 and H29) found among polystomes infecting captive M. leprosa are closely related to previous haplotypes (Fig. 2b). Genetic divergences estimated between each of these haplotypes range from 0·6% to 2·3%. In fact a single haplotype, H27, shows 2·3% divergence with H28, H29 and H30, but only 1·2% with H25 and H26. Though a threshold up to 2·0% was defined previously to delineate species using polystomes recovered from wild populations (see Results), it is obvious that H25 to H30 haplotypes belong to the same species. Therefore, P. tunisiensis would occur in wild populations and captive individuals of M. leprosa. This is also the case of N. euzeti (H31), which differs from H32 by only 0·6% divergence.
Haplotype H18, which infests E. orbicularis in the wild population of Leucate, has a genetic divergence with the most phylogenetically related haplotypes (H17, H36, H45 and H46, Fig. 2b) that range from 5·2% to 6·7%. The bladder polystome of E. orbicularis in Leucate (H18) can therefore be considered a new species. On the other hand, H21, which was recovered from a single individual of E. orbicularis in Leucate, corresponds to Neopolystoma sp6 (Fig. 2a) which occurs in the conjunctival sacs of G. pseudogeographica (Table 1). Only a single uninfected specimen of T. s. elegans has been recorded in the pond of Leucate in 2007, and G. pseudogeographica has never been found in natural environments during our field survey in Languedoc-Roussillon. This result questions the origin of that parasite from wild animals. It is very unlikely that after four years studying Leucate's population (20 different sessions of capture-mark-recapture), we missed specimens of G. pseudogeographica. Two explanations are possible for the presence of this exotic parasite in a wild population of E. orbicularis. Either G. pseudogeographica was introduced and disappeared from the pond after its parasite switched to E. orbicularis, or G. pseudogeographica was never there, and at least one individual of E. orbicularis has been introduced in that pond with an exotic parasite. Because only one turtle was found infected with an exotic parasite in Leucate, the second hypothesis seems the most plausible at the moment.
Polystome species diversity in captive turtles
Our results illustrate four cases of parasite transmission from exotic turtles to indigenous E. orbicularis and M. leprosa in the Turtle Farm. Among the four closely related haplotypes H14, H15, H33 and H34 in Fig. 2b, H14 was found from both endemic captive species, H15 from captive E. orbicularis, M. leprosa, T. s. elegans (pharyngeal cavity) and T. s. scripta whereas H33 and H34 were recorded only from captive M. leprosa. Genetic divergences for these four haplotypes range from 0·3% to 0·9% when compared to haplotypes H11 and H12 (P. oris) recorded from the pharyngeal cavity of wild C. p. marginata, suggesting the transfer of an American polystome species to captive European and American turtles. Similarly, among the three related haplotypes H19, H20 and H37 in Fig. 2b, H19 was found from captive E. orbicularis and M. leprosa, H20 from captive E. orbicularis, M. leprosa and T. s. elegans whereas H37 was only recovered from captive M. leprosa. Genetic divergences among these three haplotypes range from 0·3% to 1·7% when compared to haplotypes H9 and H10 (N. orbiculare) recorded from the urinary bladder of wild C. p. marginata, implying the transfer of a second American polystome species to captive European and American turtles. Because H17, which is found only in captive E. orbicularis and T. s. elegans, differs by 0·3% to 1·2% to H45 and H46 (Neopolystoma sp3) found in wild T. s. elegans, this is surely a third American polystome species that has switched to captive indigenous turtles. Finally, genetic divergences within three other closely related haplotypes, H21, H38 and H40, range from 0·9% to 1·5%. H21 (Neopolystoma sp6) was recovered from both captive E. orbicularis and M. leprosa, but also from wild E. orbicularis and American G. pseudogeographica, H38 was recorded from captive M. leprosa and T. s. elegans, whereas H40 was only found from captive M. leprosa. Therefore, this suggests the transfer of Neopolystoma sp6 to wild E. orbicularis and captive indigenous European and exotic American species. Because all of those American parasites have never been described from E. orbicularis and M. leprosa in natural environments, parasitic transmission from European to American turtles is unlikely. According to these results, polystomes appear less host specific in confined conditions than in natural environments. This was also documented by MacCallum (Reference MacCallum1918) who reported N. orbiculare among different host species in an aquarium in New York. In summary, host switching would have occurred from host to host within the same body site, within the urinary bladder for N. orbiculare and Neopolystoma sp3, within the pharyngeal cavity for P. oris, and within the conjunctival sacs for Neopolystoma sp6. These results shed light on the evolutionary history of chelonian polystomes that would have mainly diversified within the same microhabitat, following cospeciation and host-switching events.
Even if our study comprises the most thorough revision of chelonian polystomes so far we have found some haplotypes that do not match any known polystome species: H16, only found in E. orbicularis; H22 found in E. orbicularis and M. leprosa; H35 recorded from M. leprosa, T. s. elegans and T. s. scripta; and, H36 plus H39 found in M. leprosa. While H16 might correspond to the native polystome species infecting the pharyngeal cavity of E. orbicularis (i.e. Polystomoides ocellatum) which was originally described from wild turtles by Rudolphi (Reference Rudolphi1819), but not recovered in the wild population of E. orbicularis in Leucate, and H39 could be an undescribed polystome species of the conjunctival sacs of M. leprosa, H22, H35 and H36 cannot be yet ascribed to any known polystome species. Therefore the remaining three haplotypes probably reflect also undescribed species and may well represent instances of host switching.
Conclusions
Parasite transmission has been previously documented from indigenous wild populations of M. leprosa and E. orbicularis to red-eared sliders in Spain (Hidalgo-Vila et al. Reference Hidalgo-Vila, Díaz-Paniagua, Ribas, Florencio, Pérez-Santigosa and Casanova2009). Our study is the first to illustrate multiple cases of host switching between American turtles to E. orbicularis and M. leprosa in captivity. In addition, host switching is also illustrated here between captive American turtles in the Turtle Farm. Together these results demonstrate that American polystomes exhibit low host specificity, particularly when potential host species share the same habitats in confined areas, as demonstrated at least by P. oris, N. orbiculare, Neopolystoma sp3 and Neopolystoma sp6 that infest wild American turtles in their home range and captive American and indigenous European species. Though American and indigenous turtles in the Turtle Farm are found in distinct pools being located in small areas surrounded by fences of about 60 cm height, turtles can escape occasionally and transmit their own parasites. The small size of the pools (about 20 square meters each) and the relatively high densities of turtles (about 30 turtles per pool) would increase probabilities of host-parasite encounter, which in turn may facilitate parasite transmission and host switching from American to indigenous turtles. Nothing is known about polystome pathogenicity but other parasites (e.g. viruses) may have disastrous consequences when introduced to native host turtles (Hays et al. Reference Hays, McAllister, Richardson and Stinson1999). Because captive turtles of both species are currently used in numerous reintroduction programmes to reinforce wild populations (Bertolero, Reference Bertolero1999; Miquet and Cadi, Reference Miquet and Cadi2002; Cadi and Miquet, Reference Cadi and Miquet2004; Mosimann and Cadi, Reference Mosimann and Cadi2004; Bertolero and Oro, Reference Bertolero and Oro2009) it is important to search for the presence of exogenous parasite species in animals used in restocking programmes. Parasite surveys of wild populations of E. orbicularis and M. leprosa throughout their natural home range are now necessary to assess the risks of pathogens transmitted by red-eared sliders and other American turtles that are widespread in natural environments and occur syntopically with the indigenous species.
ACKNOWLEDGMENTS
We are grateful to Mrs Malirachs who gave us all facilities for turtle sampling in the “Vallée des Tortues” of Sorède, called Turtle Farm along the study. We are also indebted to Nacera Kaid who collected one specimen of M. leprosa in Algeria and to Tim Littlewood for his advice during the writing of the manuscript. We also thank two anonymous reviewers for their helpful comments on an early draft of the manuscript.
FINANCIAL SUPPORT
Funding for this resarch was supplied by grants from the CNRS and the NRF to OV, LDP and CP (PICS N°4837, PEPS N° FG/cb D156). LDP stays at the BETM in years 2009 and 2010 were funded by grants from the UPVD and the Languedoc-Roussillon Region.
