What canmetazoanparasites reveal about the taxonomy of Scomber japonicus Houttuyn in the coast of South America and Madeira Islands?

*Instituto de Investigaciones Oceanológicas, Universidad de Antofagasta, P. O. Box 170 Antofagasta, Chile, ‡Universidade da Madeira, Centro de Estudos da Macaron esia, Campus da Penteada, 9000-390 Funchal, Portugal, §Museo de Historia Natural, Universidad Nacional Mayor de San Marcos, Apartado 14-0434, Lima 14, Peru, kDepartamento de Parasitologia Animal, Universidade Federal Rural do Rio de Janeiro, Caixa Postal 74.508, CEP 23851-970, Serop edica, RJ, Brazil and {Programa de Doctorado en Ciencias Aplicadas, mención Sistemas Marinos Costeros, Universidad de Antofagasta, Antofagasta, Chile


INTRODUCTION
Parasites have been used successfully not only as tools to understand the population structure of the host species but also to clarify doubtful systematic relations between closely related hosts. Szidat (1955) used the parasite fauna of hake (genus Merluccius) to identify their migratory routes that explained their current distribution. Later, Kabata & Ho (1981) proposed a new model to identify the origin and dispersal of hake but on the basis of only parasitic copepods. Similar results were obtained by Fern´andez (1985), who studied the parasites of the southern hake Merluccius australis (Hutton, 1872), and Fern´andez & Dur´an (1985) on the basis of biological information from the sanguinicolid digenean genus Aporocotyle. A similar biogeographic approach has been developed for Acanthopagrus and Sillaginidae, among other fishes (Byrnes, 1987;Byrnes & Rohde, 1992;Hayward, 1997). In a similar way Rohde & Hayward (2000) analysed the importance of oceanic barriers in the distribution of Scombridae and their ectoparasites (Copepoda and Monogenea), concluding that the East Pacific Barrier represents an efficient barrier for almost all the studied Scombridae, but a less effective barrier for species of the genus Scomber. All the cited examples refer to congeneric (or members of the same family) species living in discrete habitats.
Parasites have been used also as biological markers, mainly as tags for the identification of stocks (subpopulation units) and migratory movements (MacKenzie & Abaunza, 2005). Under the traditional population studies approach, parasites are used as a tool for the identification of discrete subpopulations of the same species but living within a continuous habitat. Few fish species show the alternative distributional model: the same species with an extended geographical distribution but living in discrete habitats. One of such species is chub mackerel Scomber japonicus Houttuyn, 1782, a widespread, circumglobal, pelagic species which is divided into geographically disjunct populations ( Fig. 1) (Scoles et al., 1998). This distributional pattern closely resembles the distribution of different species of the pelagic genera Trachurus and Engraulis and the demersal Merluccius. The main characteristic of the chub mackerel is its circumglobal distribution as a unique species, whereas the same distributional pattern is evident for other marine fishes but at the generic level. Data on the metazoan parasite fauna of the chub mackerel, specifically the highly specific parasites of S. japonicus from the south-eastern Pacific (Peru and Chile), western Atlantic (Brazil) and Eastern Atlantic (Madeiran waters, Portugal) were used in order to evaluate if the four chub mackerel populations studied can be considered as members of the same species or must be considered as different taxonomic units.

MATERIALS AND METHODS
Samples of chub mackerel were taken during, 2002 and 2003 from Callao, Peru, Antofagasta, Chile, Rio de Janeiro, Brazil, and Madeira, Portugal. Geographic coordinates, sample size and specimen sizes for the sampling localities are given in Table I. Fish were frozen (À18°C) and transported to the laboratory for parasitological analyses. After thawing, the fish total length (L T ) was measured to the nearest 10 mm, dissected, sexed and examined for metazoan ectoparasites and endoparasites. Parasites were separated, stored in 5% formalin, and counted by species. For each parasite species the population descriptors (prevalence and mean intensity) were calculated according to Bush et al. (1997). Two multivariate analyses were used to detect associations between metazoan parasite fauna and locality (data reduction: correspondence analysis, CA) and a classification analysis (multivariate discriminant analyses, MDA) to determine if parasites were good discriminants of locality. Analyses were performed following the recommendations of Wilkinson (1990) using SYSTAT Ò as the statistical tool.

RESULTS
A total of 373 specimens of chub mackerel were analysed and produced 13 757 parasite specimens belonging to 34 taxa (Monogenea ¼ 5, Digenea ¼ 10, larval Cestoda ¼ 4, Nematoda ¼ 6, Acanthocephala ¼ 5, Copepoda ¼ 2 and Isopoda ¼ 2). Anisakis identified as Anisakis type I and type II (Berland, 1961) from Madeira, Anisakis sp. from Callao and Anisakis sp. from Rio de Janeiro and Antofagasta were pooled and considered as Anisakis sp. because recent studies have demonstrated that the taxonomy of members of Anisakis based on morphology is confused (Mattiucci & Nascetti, 2006). Sibling species in Anisakis from the Atlantic and Pacific waters was first demonstrated by Nascetti et al. (1986). The discovery of at least five species of Anisakis in S. japonicus from Madeira (Pontes et al., 2005) using molecular techniques justifies this decision.
Mean L T of host fish differed significantly between localities (ANOVA, d.f. ¼ 3,369, P < 0 Á 001), with chub mackerel from the Pacific being larger than those from the Atlantic. For each locality, parasite richness did not co-vary with host size (Madeira r 2 ¼ 0 Á 0075, Rio de Janeiro r 2 ¼ 0 Á 0064, Callao r 2 ¼ 0 Á 0118 and Antofagasta r 2 ¼ 7 e À5 ). Thus, L T of the host was not considered as a confounding factor for parasite species richness. For all localities, except Antofagasta, 100% of fish were parasitized by one or more parasite species. For Antofagasta, 92% of the fish were parasitized. An ANOVA (d.f. ¼ 3, 369, P < 0 Á 001) following log 10 (n þ 1) transformation of the data indicated that parasite richness differed significantly among the four localities. The a posteriori Tukey test (Zar, 1996) showed that parasite richness of samples from the Pacific (P > 0 Á 05) and the Atlantic do not differ significantly (P > 0 Á 05). The parasitic species found from chub mackerel in each locality, and the prevalence and mean intensity of infection are given in Table II. From Table II it is evident that parasites can be included in three groups: parasites limited to the Atlantic (i.e the monogeneans Grubea cochlear and Kuhnia scombercolias, non-Didymozoidae digeneans except Nematobothrium scombri and the acanthocephalan Rhadinorhynchus pristis), parasites limited to the Pacific (the monogenean Kuhnia sprostonae, the didymozoids Ovarionematobothrium saba and Didymocystis sp. the isopods Ceratothoa gaudichaudii and Rocinela sp.) and parasites common to both oceans (the monogenean Kuhnia scombri and the copepod Clavellisa scombri).
Multivariate analyses (CA and DMA) were performed for the whole sample, for those parasites with prevalence >5% and those with prevalence >10% in at least one locality. For both analyses results did not differ. In CA the variance explained by the first two factors for the whole sample, parasites with prevalence >5% and parasites with prevalence >10% were 82 Á 8, 83 Á 2 and 84 Á 6%, respectively. In a similar way the DMA also did not show dramatic differences: Wilks' lambda ¼ 0 Á 0091, 0 Á 0095 and 0 Á 0010, for the whole sample, parasites with prevalence >5 and >10%, respectively. The correct assignation was high and similar: 92, 93 and 92%, respectively. The results of both analyses corroborate that parasites correspond well with localities and can serve to discriminate the origin of the samples with high resolution. (Figs 2 and 3)

DISCUSSION
As stated by Stepian & Rosenblatt (1996) there are many closely related marine fish species in temperate north and south Pacific, but absent from the tropical region of Central America. Representatives of Merluccius, Trachurus and Scomber are good examples, but other pelagic fishes like anchovies and sardines also meet this distributional pattern. Stepian & Rosenblatt (1996) concluded that the anti-tropical distribution at the species level for representatives of Merluccius is well supported by genetic evidence. For the genus Trachurus the genetic evidence is weak in support of specific differentiation but recently Poulin et al. (2004) and C´ardenas et al. (2005) demonstrated that representatives of Trachurus from the Pacific coast of North and South America belong to different Trachurus species. Finally and as stated by Stepian & Rosenblatt (1996), Scomber does not represent distinctly different species, but two isolated populations of S. japonicus.   Table I).
Prevalence of infection (P) and mean intensity of infection (MI) are given Parasite species

Madeira
Rio de Janeiro Callao Antofagasta L, Immature stages. *parasite species with records only from Scomber spp.  C´ardenas et al. (2005) suggested that Trachurus species entered the Pacific across the submerged Isthmus of Panama. It is not surprising that the hypothesis that explains the presence of Merluccius in the Pacific coast of America refers to the crossing of the submerged Isthmus of Panama, as suggested by Quinteiros et al. (2000), Kabata & Ho (1981) and Ho (1989) using molecular analyses, parasitological data and a cladistic analysis, respectively. Unfortunately, Fig. 1 (Scoles et al., 1998;Infante et al., 2006). An important biogeographic characteristic of this genus is the global distribution of S. japonicus (Fig. 1) with a distribution that closely resembles those of species of the genus Trachurus, Engraulis and Merluccius among others. The metazoan parasite fauna of chub mackerel from the Atlantic (Madeira and Brazil) and the Pacific (Peru and Chile) coast of America includes a total of 34 species (Table II), of which eight species (six ectoparasites and two endoparasites (Table II) are specific to Scomber spp. and at least one (Ovarionematobothrium saba) is specific to S. japonicus in Japan and the Pacific coast of South America.

FIG. 2. Plot of the correspondence analysis. (See
Parasites of S. japonicus from the studied localities can be pooled in three categories: parasites found only in the Atlantic, parasites found only in the Pacific and parasites common to all localities (Table II). Only two parasites specific to Scomber (the monogenean K. scombri and the copepod C. scombri) are present in all the localities. According to Kabata (1979) the copepod C. scombri 'offers one of the rare instances among parasitic copepods of discontinuous distribution'. This comment also applies to K. scombri. The presence of both ectoparasites demonstrates a common origin for the four studied populations of S. japonicus, but the presence of K. scombercolias in the Atlantic population and K. sprostonae in the Pacific one suggests an early separation of both host populations. Recently, Costa et al. (2007) recorded just one specimen of K. sprostonae in one of 151 specimens of S. japonicus from Madeiran waters. Mazocraeids have been defined as primitive parasites specific to Clupeiformes and the presence of mazocraeids (Grubea spp. and Kuhnia spp.) are considered to be the consequence of a host shift (Mollaret et al., 2000;Jovelin & Justine, 2001), but specificity for the new host apparently remains as a characteristic of this family of monogenean worms.
Didymozoids are common parasites of pelagic and oceanic fishes and four species were found in the samples. Ovarionematobothrium saba and Didymocystis sp. are present only in the Pacific coast of South America, Halvorsenius exilis is restricted to the Atlantic (but not Brazil) and N. scombri occurs in both Atlantic and Pacific host populations (but was not found in Peru). Didymozoid digeneans also can show high host specificity. The presence of specific (or specialist) parasites limited to the Atlantic and Pacific in the same host species raises some questions: either some parasites speciate at a higher rate than their host, or the S. japonicus populations from the Pacific coast of South America and Atlantic coast of South America and Madeira really represent cryptic species, that in turn can explain the presence of different parasite species. Merluccius spp. and Engraulis spp. harbour the monogenean Anthocotyle merlucci and P A R A S I T E S O F C H U B M A C K E R E L 551 Pseudanthocotyloides heterocotyle, respectively, along their geographical distribution as a genus, suggesting that the hosts speciate at a rate higher than their monogenean parasites.
The current distribution of S. japonicus in the Atlantic and Pacific Ocean can be explained by a migratory route that closely resembles those of the genera Merluccius and Trachurus, specifically using the passage over the submerged Isthmus of Panama. But, why did Merluccius and Trachurus (and other fish species) produce different species in an anti-tropical distributional pattern along the coast of South America whereas Scomber did not? A speculative point of view suggests a very low speciation rate for Scomber. Both, the genera Trachurus and Merluccius have speciated since the Isthmus of Panama closed as a passage between the Atlantic and Pacific Ocean. Any other possible explanation must consider a migratory route throughout the Pacific Ocean. Rohde & Hayward (2000) suggested that the eastern Pacific was a good barrier for most scombrids but not for the genus Scomber. According to Zardoya et al. (2004) marine migratory fishes with high dispersal capabilities and large effective population size have high levels of gene flow and consequently a low degree of differentiation. From this perspective, Mediterranean populations of mackerel (S. scombrus and S. japonicus) are genetically well differentiated. Specifically, the population of S. japonicus is considered as panmictic in the Mediterranean Sea and adjacent Atlantic waters. In addition, Nesbo et al. (2000) pointed out that the distribution and level of divergence of different genetic lineages in highly mobile species can reflect both present and historical events of gene flow. The presumed high mobility of S. japonicus has not been proved (at least for the Pacific coast of America). Moreover, it is well known that populations from the South Pacific and North Pacific coast of America do not overlap (Stepian & Rosenblatt, 1996;Scoles et al., 1998) and gene flow between both populations remains undefined.
Unfortunately, the most comprehensive study of the phylogeography of the genus Scomber does not include samples from the Pacific coast of South America, Brazil and Madeira (Scoles et al., 1998). The particular distributional pattern of S. japonicus in addition to the parasitological evidence, based on the presence of highly specific parasites for Atlantic and Pacific populations of S. japonicus, strongly supports the recommendation of a revision of the taxonomy of S. japonicus that may actually be a complex of species along its extended and discontinuous distribution. Recently Infante et al. (2006) have suggested, on the basis of nuclear DNA sequences, the phylogenetic differentiation between S. colias, a species from the Atlantic (Gulf of C´adiz, Spain) and S. japonicus from the Pacific (Kochi Prefecture, Japan). In addition, they strongly suggest the need for a revision of the classical taxonomic status of mackerels of the genus Scomber, a suggestion also proposed by Collete et al. (2001) who give evidence that supports the division of Scomber into two genera. The suggested revision is now supported not only by molecular evidence, but also by parasitological evidence. The agreement between molecular and parasitological evidence reinforces the use of parasites as adequate tools to clarify taxonomic problems of their hosts.
We thank R. Poulin (University of Otago, New Zealand) for reading and commenting on an early draft of the manuscript. J.L.L. was supported by a fellowship 552 M . E . O L I V A E T A L .