INTRODUCTION
Parasites are often considered as good models to study the specificity of organisms with respect to a given resource (De MeeÛs et al. Reference De MeeÛs, Michalakis and Renaud1998), because the resource is easily defined as a host(s) or organ(s) used (Lymbery, Reference Lymbery1989; Timms and Read, Reference Timms and Read1999). While parasite taxa exhibit various ranges of host specificity (Sasal et al. Reference Sasal, Desdevises and Morand1998; Bush et al. Reference Bush, Fernandez, Esch and Seed2001), monogeneans are known to be highly host-specific (e.g. Rohde, Reference Rohde1994) and have a direct life-cycle that makes them practical for studies of factors linked to host specificity. Although studied extensively during the last decade, few things are known about these determinants (see discussion in Caro et al. Reference Caro, Combes and Euzet1997; Desdevises et al. Reference Desdevises, Morand and Legendre2002b). Ecological factors may ‘drive’ host specificity, which could be a way for parasites to aggregate on hosts for reproduction (Rohde, Reference Rohde1994). It is also acknowledged that host size is linked to host specificity in monogeneans (Sasal et al. Reference Sasal, Trouvé, Müller-Graf and Morand1999; Desdevises et al. Reference Desdevises, Morand and Legendre2002b; Morand et al. Reference Morand, Simková, Matejusova, Plaisance, Verneau and Desdevises2002), and that host immunity plays a critical role (Buchmann, Reference Buchmann1999). Early studies focused mainly on host factors, such as habitat characteristics favouring parasite specialization, but only a few studies have investigated intrinsic parasite factors that may act on host specificity (e.g. Gemmill et al. Reference Gemmill, Vinet and Read2000).
Recently, Jarkovsky et al. (Reference Jarkovsky, Morand, Simková and Gelnar2004) and Simkova et al. (Reference Simková, Verneau, Gelnar and Morand2006) showed a relationship between morphological features of parasites and some components of their ecological niche within hosts (see also Morand et al. Reference Morand, Page, Hafner and Reed2000). Monogeneans attach to their host via their opisthohaptor, which is composed of sclerified parts whose sizes and morphologies are characteristic for each parasite species. However, while size and shape of the different haptoral pieces involved in gill attachment could be under strong selection pressure (Rohde, Reference Rohde1979, Reference Rohde1989; Rohde and Hobbs, Reference Rohde, Hobbs, Cremin, Dobson and Noorhouse1986; Brooks and McLennan, Reference Brooks and McLennan1991; Simkova et al. Reference Simková, Desdevises, Gelnar and Morand2001a; Poulin, Reference Poulin2007), the role of morphological variability in host specificity, which was suggested as important in a previous study (Desdevises et al. Reference Desdevises, Morand and Legendre2002b), has not been fully investigated.
To limit the effects of phylogenetic inertia in the present investigation, we compared generalist and specialist species in a well-defined taxonomic group: monogeneans of the genus Lamellodiscus, which are gill parasites of sparid fishes. The pattern of host specificity within Lamellodiscus is well known, because this host-parasite system has been intensively studied in the Mediterranean Sea (Euzet and Oliver, Reference Euzet and Oliver1966, Reference Euzet and Oliver1967; Oliver, Reference Oliver1968, Reference Oliver1973, Reference Oliver1974, Reference Oliver1987; Euzet, Reference Euzet1984; Desdevises et al. Reference Desdevises, Jovelin, Jousson and Morand2000, Reference Desdevises, Morand and Oliver2001, Reference Desdevises, Morand, Jousson and Legendre2002a, Reference Desdevises, Morand and Legendreb; Neifar et al. Reference Neifar, Euzet and Oliver2004; Amine and Euzet, Reference Amine and Euzet2005). This genus contains strict specialist species that use a single host species, and generalist species, parasitizing several distinct host species (up to 6 in the study area).
We investigated whether size variability in haptoral parts is different between specialist and generalist monogenean species. It was hypothesized that specialist species exhibit lower haptoral variability (related to the ‘low’ variability encountered in gill size of their unique host) than generalist species that attach to a variety of different host species. To test this hypothesis, we selected strict specialists versus clear generalists using at least 4 host species, and conducted morphometric and molecular analyses. Morphometric studies of parasites are commonly conducted (e.g. Almeida et al. Reference Almeida, Rodrigues-Silva, Neves, Romani and Machado-Silva2007; Kuchta et al. Reference Kuchta, Vlcková, Poddubnaya, Gustinelli, Dzika and Scholz2007), particularly for monogeneans (e.g. Geets et al. Reference Geets, Appleby and Ollevier1999). However, thus far, they have mainly addressed taxonomic (Shinn et al. Reference Shinn, Gibson and Sommerville2001; Mariniello et al. Reference Mariniello, Ortis, D'Amelio and Petrarca2004; Rubtsova et al. Reference Rubtsova, Balbuena, Sarabeev, Blasco-Costa and Euzet2006) and ecological (Simkova et al. Reference Simková, Gelnar and Sasal2001b) questions. While interindividual morphometric variability within monogeneans is often mentioned (Mariniello et al. Reference Mariniello, Ortis, D'Amelio and Petrarca2004; Rubtsova et al. Reference Rubtsova, Balbuena, Sarabeev, Blasco-Costa and Euzet2006), and can be relatively high (Huyse and Volckaert, Reference Huyse and Volckaert2002; Mariniello et al. Reference Mariniello, Ortis, D'Amelio and Petrarca2004; Davidova et al. Reference Dávidová, Jarkovskỳ, Matejusová and Gelnar2005), it has never been used for investigating determinants of host specificity. As the same pattern of variability may also be encountered at the molecular level and would be independent of putative phenotypic plasticity, we also studied a ‘variable’ molecular marker (the first internal transcribed spacer of nuclear ribosomal DNA, ITS-1) in order to compare levels of morphometric and genetic variability within specialist and generalist species.
MATERIALS AND METHODS
Sampling
Host specimens were sampled in the Gulf of Lion, in the northeastern part of the Mediterranean Sea. Gills were isolated, and gill arches were separated and placed in containers of seawater. Monogeneans were removed under a dissecting microscope and identified based on morphological characteristics of the haptor and copulatory organ under a light microscope. After identification and measurements (see next section), individual parasites were fixed in 70% ethanol for subsequent molecular analyses. Parasite specialist species Lamellodiscus virgula and L. drummondi were both collected from Pagellus acarne, whereas L. erythrini and L. baeri were sampled from Pagellus erythrinus and Pagrus pagrus, respectively (Table 1). Parasite generalist species, L. elegans, L. ergensi and L. ignoratus, were sampled exclusively from 2 different host species, i.e. Diplodus vulgaris and D. sargus (Table 1), in order to minimize a possible host influence on parasite morphology. The parasite species selected were all ‘scattered across’ the phylogenetic tree of Lamellodiscus (see Desdevises et al. Reference Desdevises, Morand and Legendre2002b; Fig. 1B), thus avoiding the comparison of 2 groups of closely related species and generating confounding phylogenetic effects.
Fig. 1. Morphometric variables measured on the sclerified parts of the opisthohaptor of Lamellodiscus species: a, a′=total length of dorsal and ventral hook; b, b′=point length of dorsal and ventral hook; c, c′=blade opening of dorsal and ventral hook; d=distance between grip and hilt; f=grip length; g=hilt length; Bd=length of dorsal lateral bar length; Bv=length of median ventral bar.
Table 1. Pattern of host-parasite associations between sparid fishes and their gill monogenean parasites from the genus Lamellodiscus
Morphometric data
Morphometric data were measured microscopically with an optical micrometer at 400× magnification. Haptoral parts (Fig. 1) were measured according to Amine and Euzet (Reference Amine and Euzet2005). Only sclerified pieces were assessed because of the absence of deformation during dissection or measurement. Eleven variables (designated a, a′, b, b′, c, c′, d, f, g, Bd, Bv) were measured on the 102 parasites collected. A correlation matrix was then computed between all morphometric variables, in order to eliminate redundant variables for Pearson correlation coefficients of >0·9 (the kept variable was assumed to represent both). The remaining morphometric variables were assessed using a principal component analysis (PCA), to determine whether individual dispersions were different between specialist and generalist species. PCA and ordination plots were conducted using the program R Package 4.0 (Casgrain and Legendre, Reference Casgrain and Legendre2000).
Variances of each morphometric variable were compared among 2 groups: specialists and generalists. We computed weighted average variances for each variable within each group that were compared using an F-test (see Sokal and Rohlf, Reference Sokal and Rohlf1995) using the program XLSTAT 2007 (Data Analysis and Statistics Software for Microsoft Excel, Addinsoft, Paris, France).
Molecular data
Previously, it was shown that the ITS-1 was relatively variable in sequence among species of Lamellodiscus, but relatively conserved within species, with some inter-individual sequence variability (Desdevises et al. Reference Desdevises, Jovelin, Jousson and Morand2000). Therefore, the ITS-1 region was employed herein to ascertain genetic variability within species. Genomic DNA was extracted from whole individual monogeneans using Bio-Rad Chelex 100™ and proteinase K (10 mg/ml), according to the protocol of Hillis et al. (Reference Hillis, Moritz and Mable1996). Whenever possible, the same individuals were used for morphometric and molecular analyses. Genomic DNAs were isolated from 58 individuals and each dissolved in 100 μl of water.
For generalist species, sampling was by host species; 6 specimens of L. ignoratus were sampled from Diplodus sargus and 5 from D. vulgaris; for L. elegans, 4 individuals were sampled from D. vulgaris and 4 from D. sargus; for L. ergensi, 3 specimens were sampled from D. sargus and 4 from D. vulgaris. For specialist species, 19 individuals of L. drummondi and 5 L. virgula were sampled from Pagellus acarne, 8 specimens of L. erythrini were sampled from P. erythrinus, and 4 L. baeri individuals from Pagrus pagrus.
The entire ITS-1 region was amplified using the primers L7 (5′-TGATTTGTCTGGTTTATTCCGAT-3′) (Verneau et al. Reference Verneau, Renaud and Catzeflis1997) and IR8 (5′-GCTAGCTGCGTTCTTCATCGA-3′) that anneal to the 18S and 5·8S rRNA genes, respectively. PCR was carried out using standard reagents (GoTaq Flexi DNA Polymerase, Promega Corp., USA) in a volume of 25 μl containing 5 μl of genomic DNA (~25 ng), under the following cycling conditions: 5 min at 95°C, followed by 34 cycles of 1 min at 48°C, 1 min 30 at 72°C, 1 min at 95°C, and a final elongation of 10 min at 72°C. Subsequently, PCR products were resolved by agarose electrophoresis and detected following ethidium bromide staining and ultraviolet transillumination (Transilluminator 312 nm Ultraviolet, Model TC-312A, Spectroline, USA). The band of interest (~1 kb) was excised, purified using the Nucleospin Extract II kit (Macherey-Nagel GmbH & Co, Germany) and subjected to automated sequencing (in both directions) using the same primers as used for the PCR amplification (L7 and IR8). A total of 62 sequences (i.e. 58 new ones and 4 reference sequences with GenBank Accession numbers AJ276444–AJ276447; Desdevises et al. Reference Desdevises, Jovelin, Jousson and Morand2000) were aligned using the program MAFFT v.5 (Katoh et al. Reference Katoh, Misawa, Kuma and Miyata2002, Reference Katoh, Kuma, Toh and Miyata2005) and the alignment adjusted by eye using the program Se-Al v.2.0a11 software (Rambaut, A., 1996. Se-Al: Sequence Alignment Editor, available at http://evolve.zoo.ox.ac.uk/). Regions in which sequence heterogeneity was detected (and could not be aligned with confidence) were removed from the alignment before computing (with PAUP* 4.0b10 (Swofford, Reference Swofford2003)) uncorrected genetic distances within specialist and generalist species.
RESULTS
Morphometry
Based on the correlation matrix, variables b and b′ were removed from the subsequent numerical analysis, because they were highly collinear with variables a and b. The PCA was then conducted on the 9 remaining morphometric variables and the resultant ordination diagram is presented in Fig. 2. The two main axes represent 77·4% of the original variance (51·4% for Axis 1 and 26·0% for Axis 2). The most important variables, shaping the distribution of individuals in the morphospace, are ventral hook length (variable a, Pearson correlation coefficient (r)=0·98 with Axis 1) and grip size (variable f, r=0·76 with Axis 2). These variables were strongly correlated to several other variables and broadly define 2 groups of quite independent variables (see correlation matrix in Table 2). One variable is linked to hook size (correlated to variable a) and one to the haptoral muscles insertion (Kearn, Reference Kearn and Fallis1971), then mechanical strength of attachment (variables correlated to grip size (f): g, d, and Bv). On the ordination plot (Fig. 2), all but 1 specialist species, L. erythrini, clustered in small and well-defined areas, whereas the distribution of generalist species was more scattered. Variances and variance comparisons between specialists and generalists are shown in Tables 3 and 4, respectively. For each character, variance of the haptoral part within generalists was higher than that within specialists. All but 1 (d: distance between grip and hilt) of these variance differences were statistically significant.
Fig. 2. Ordination diagrams computed from the Principal Component Analysis performed on the morphometric variables described in Fig. 1, on 102 Lamellodiscus individuals. The main diagram is a distance bi-plot (preserving the original distances among individuals), the small diagram is a correlation bi-plot (angles indicate corrrelations between variables). Open symbols: specialists (□: L. baeri, △: L. drummondi, ○: L. erythrini, ⋄: L. virgula), filled symbols: generalists (●: L. elegans, ■: L. ergensi, ▲: L. ignoratus).
Table 2. Correlation matrix (Pearson correlation coefficient) between morphometric variables measured on the haptor of Lamellodiscus monogeneans
Table 3. Variances of morphometric variables measured on the haptor of Lamellodiscus monogeneans
Table 4. Comparison of variances between specialists (4 species, n=61 individuals) and generalists (3 species, n=41) for morphometric variables described on Fig. 1
Molecular data
The 58 sequences determined (deposited in the GenBank database under Accession numbers EU259002 to EU259059 were aligned and compared with the reference sequences (Accession numbers AJ276444–AJ276447; Desdevises et al. Reference Desdevises, Jovelin, Jousson and Morand2000). The length of the ITS-1 region varied from 340 to 729 bp. Length variation has already been reported among species within the genera Lamellodiscus (see Desdevises et al. Reference Desdevises, Jovelin, Jousson and Morand2000), Gyrodactylus (see Cable et al. Reference Cable, Harris, Tinsley and Lazarus1999; Huyse and Volckaert, Reference Huyse and Volckaert2002; Zietara and Lumme, Reference Zietara and Lumme2002) and Polystoma (see Bentz et al. Reference Bentz, Sinnappah-Kang, Lim, Lebedev, Combes and Verneau2006). ITS-1 sequences from individuals of the same species were aligned, as the sequence differences among species are too extensive to achieve an accurate alignment (precluding the reconstruction of a phylogenetic tree and comparison with an outgroup). Intraspecific variations (%) computed from these alignments are given in Table 5; variation is greater within the generalist species than within specialist species. Obviously, the >8-fold variation between specialists and generalists suggests a clear trend, although it cannot be tested for significance (i.e. distances are not independent data).
Table 5. Mean inter-individual uncorrected genetic distances (%) on ITS1 DNA sequences, with standard deviations
DISCUSSION
Variances of all morphometric variables investigated, and genetic distances, are clearly greater within species of generalist Lamellodiscus than within those of specialists. These results support the hypothesis that the use of a narrowly delimited resource by specialists is linked to a lower range of phenotypic and genotypic variation. Furthermore, this use strengthens the adaptive role of the attachment organ (Adamson and Caira, Reference Adamson and Caira1994) and its function towards host specificity (Simkova et al. Reference Simková, Verneau, Gelnar and Morand2006). The correlation found between host specificity and intrinsic parasite factors is in accordance with earlier results showing that host specificity in Lamellodiscus exhibits a significant level of phylogenetic inertia (Desdevises et al. Reference Desdevises, Morand and Legendre2002b), in that parasites with the same level of host specificity tend to be in the same clades.
Morphometric differences in attachment apparatus between specialist and generalist monogeneans have been reported previously for Dactylogyrus (see Jarkovsky et al. Reference Jarkovsky, Morand, Simková and Gelnar2004), where specialist species are more similar than generalists. This latter finding was interpreted to be an effect of a specialization with respect to the host. Similarly, Simkova et al. (Reference Simková, Desdevises, Gelnar and Morand2001a) showed that, within the same genus, a positive correlation between host size and haptor size exists for specialist species (while no significant correlation was found within generalists), suggesting a tighter adaptation of parasite specialists to their resource (i.e. the host species). Nevertheless, only the means of haptoral characters, not their variances, were considered in these studies, which may explain why some authors have failed to observe a link between host size and haptor-related variables within generalists, while this link holds for specialists alone (Morand et al. Reference Morand, Simková, Matejusova, Plaisance, Verneau and Desdevises2002). The intraspecific differences in variances that we found within Lamellodiscus are consistent with the results of previous studies of species of Dactylogyrus, and furthermore, extend to the individual level. Only mean values for individual species were compared in these previous studies and not individuals. In addition, because variability in attachment organ and genotype could point to a putative process determining specificity, the present results represent an important advance: more variable species could colonize a larger host spectrum.
However, an important question remains: is the intraspecific variability observed within Lamellodiscus specialists and generalists a cause or a consequence of host specificity? Generalist parasites could exhibit a higher variance simply because the different host species used (representing various niches) exert different pressures on parasite morphology, thus increasing morphometric variances as well as genetic distances. In this case, intraspecific variability would be a consequence of host specificity, i.e. the more a parasite species uses different host species, the more its intraspecific variance increases. Because we sampled generalists from only 2 different host species among their broad host spectrum, a putative influence of host species on phenotypic development of parasites has been controlled for. Indeed, if this influence were real, we should have observed the same level of intraspecific variability within specialist and generalist species infesting the same host species. However, this was not the case; intraspecific genetic distance computed between individuals from generalist species sampled from the same host species were comparable to distances computed between all individuals, and clearly higher than distances measured within specialist species (see Table 5). It is still possible that the host species from which generalist species were sampled possess intrinsic characteristics different from host species harbouring the specialist species selected. Specialists were collected on Pagellus spp. and Pagrus pagrus, whereas generalist species were sampled from Diplodus sargus and D. vulgaris. If Diplodus species present a higher variability in gill morphology or immune defence than Pagellus and Pagrus, or very different genetic structures, this may select for higher variability in their parasites, whatever their level of specificity. Presently, there are no data supporting this hypothesis, but it cannot be refuted. Specialist parasite species from these hosts would be required to test this hypothesis, but there is only 1 specialist species of Lamellodiscus known from 1 of these 2 hosts, namely L. mirandus from Diplodus sargus (Euzet and Oliver, Reference Euzet and Oliver1966), and it is a very rare species.
The alternative hypothesis is that an increase of morphometric variability in the attachment organ allows parasites to colonize more host species. This increase in variability would offer the parasites more opportunities to settle on different host species, and would then be a cause of host specificity. The case of L. erythrini, a clear specialist, then becomes quite interesting. This species exhibits a high morphometric and molecular variability compared with other specialist species, although the variation is lower than within generalists. However, if variability were driven by specificity, whatever its cause, L. erythrini should have either a narrower intraspecific variance or a larger host spectrum. It could be assumed that lack of host switching opportunities preclude it from capturing new host species, which is consistent with the observation that L. erythrini and L. baeri form a clade for which a co-speciation event with their ancestral host species has been hypothesized (Desdevises et al. Reference Desdevises, Morand, Jousson and Legendre2002a). Co-speciation is a process which is probably ‘driven’ mostly by the absence of host switching opportunities (see Poulin, Reference Poulin2007), and within monogeneans parasites is illustrated by the capture of new host species which seems to be clearly dependent on ecological constraints (Klassen and Beverley-Burton, Reference Klassen and Beverley-Burton1987, Reference Klassen and Beverley-Burton1988; Desdevises et al. Reference Desdevises, Morand, Jousson and Legendre2002a; Simkova et al. Reference Simková, Morand, Jobet, Gelnar and Verneau2004; Huyse and Volckaert, Reference Huyse and Volckaert2005). Subsequent speciation may or may not follow, leading to a new specialist species or simply increases the host range of the former switcher.
Finally, it has been suggested that some monogeneans display different haptoral morphologies, depending on the infected hosts, via phenotypic plasticity (Mo, Reference Mo1991a, Reference Mob). This plasticity could be an alternative explanation for the greater morphological variability observed within generalist species. However, it is always coupled to a greater genetic variability (in ITS-1), as is the case for the specialist species L. erythrini. This observation suggests that, if host-induced phenotypic plasticity exists, it is not the only cause of the variability measured in the attachment organ.
The high level of genetic variation within generalist species may also be explained by the occurrence of cryptic species infesting distinct host species. All of the species of Lamellodiscus investigated are clearly defined from a morphological viewpoint (Oliver, Reference Oliver1987; Amine and Euzet, Reference Amine and Euzet2005). However, the ITS-1 sequences derived from different individuals herein can be aligned within generalist species while this is not possible between distinct species (Desdevises et al. Reference Desdevises, Jovelin, Jousson and Morand2000). Similarly, we did not find any evidence in the present study that individuals within the 3 generalist species form clades that would be specific of the host species from which they were collected. If cryptic species exist, each of the new ‘molecular’ species should also be elevated to the status of generalist species. Another argument against the possibility of cryptic species is that all L. elegans individuals sequenced were different in their ITS-1 sequence, whereas this parasite is one of the most clearly defined species morphologically, with its male genital organ having a very specific and unique shape (Euzet and Oliver, Reference Euzet and Oliver1966; Oliver, Reference Oliver1968). As we cannot rule out the hypothesis of co-occurrence of cryptic parasite species within host species, the ITS-1 region and mitochondrial genes (Vilas et al. Reference Vilas, Criscione and Blouin2005), such as cytochrome c oxidase 1, should be analysed for additional generalist individuals sampled from their entire host spectrum.
In conclusion, the present findings suggest that host switching is more likely from parasite species exhibiting more intraspecific variability. Such findings have consequences for aquaculture, if more variable species do indeed have more opportunities to colonize new host species. For example, the generalist L. elegans enlarges its range to a new host, Sparus aurata, under aquaculture conditions (Mladineo and Marsic-Lucic, Reference Mladineo and Maršić-Lučić2007). Nonetheless, other factors do contribute to host specificity within monogeneans, such as host responses, particularly via host mucus (Buchmann and Lindenstrøm, Reference Buchmann and Lindenstrøm2002); therefore, immunological factors active during a host switch should be investigated. Similarly, because specificity involves host-parasite compatibility, the variability between host individuals within a species should also be explored in future investigations.
We thank Timothée Poisot for the drawing of Fig. 1, and Pascal Romans for invaluable help in fish sampling. We also thank two anonymous referees for useful comments.
