Zoogeographical patterns of flatfish ( Pleuronectiformes ) parasites in the Northeast Atlantic and the importance of the Portuguese coast as a transitional area

1 universidade de lisboa, Faculdade de Ciências, instituto de oceanografia, Campo grande, 1749-016 lisboa, Portugal. e-mail: jimarques@fc.ul.pt 2 universidade do Porto, Faculdade de Ciências, Departamento de Zoologia e antropologia, Praça gomes teixeira, 4099002 Porto, Portugal. 3 CiMar laboratório associado / CiiMar, rua dos bragas, 177, 4050-123 Porto, Portugal. 4 universidade de lisboa, Faculdade de Ciências, Departamento de biologia animal, Campo grande, 1749-016 lisboa, Portugal.

resuMen: Patrones zoogeográficos de parásitos de pleuronectiformes de la costa portuguesa y su importancia como área de transición.-los parásitos son reconocidos como una excelente fuente de información de la distribución de sus hospedadores.en este estudio, la fauna parasitaria de 20 especies de Pleuronectiformes, pertenecientes a cinco familias, que habitan la costa Portuguesa fue analizada y comparada con la conocida en otras cuatro áreas (Mar del norte, norte del atlántico nordeste, Mediterráneo y costa del noroeste africano) para determinar (1) sus patrones zoogeográficos y (2) el papel de la costa portuguesa como una provincia biogeográfica intermediaria.los macroparásitos de los pleuronectiformes encontrados en la costa portuguesa fueron recogidos de acuerdo con las técnicas estándar en parasitología y los de las otras cuatro áreas fueron recopilados de la literatura, totalizando 73 especies.ambos tipos de datos fueron posteriormente compilados en una matriz de presencia/ausencia.Hospedadores y macroparásitos fueron clasificados en categorías zoogeográficas, de acuerdo con su distribución y los patrones obtenidos mediante análisis multidimensional y de clúster.la zoogeografía de hospedadores y parásitos no fue totalmente concordante, aunque la de los endoparásitos fuera, en general, consistente con los patrones referidos para las especies marinas de vida libre.en cambio, sólo los ectoparásitos específicos reflejaron introDuCtion the last decade has seen increased interest in host-parasite relationships in ecological studies, with parasites providing important information about their hosts biology, ecology, phylogeny and population structure.since it is highly improbable that parasite species are distributed randomly among their hosts, the geographical variation of the occurrence and abundance of parasite species is regarded as an excellent source of information on the geographic distribution and migration routes of marine fishes (Mosquera et al., 2000;Poulin and Morand, 2000).However, though studies using parasites of marine fishes as indicators of extensive zoogeographical patterns are recognised as good models for studying biogeography (e.g.Carney and Dick, 2000;Poulin, 2003;Mejía-Madrid et al., 2007;Waltari et al., 2007;Pais et al., 2008), they are remarkably scarce.exceptions are some studies concerning the parasite fauna of economically important and well-studied fish species (e.g.rohde and Hayward, 2000), although results have led to different conclusions.blaylock et al. (1998) investigated the parasite fauna of adult Hippoglossus stenolepis (schmidt, 1904) from 15 localities along the north american coast and defined three zoogeographic zones consistent with the ones previously established using the distribution of fishes and invertebrates.Furthermore, by examining the ectoparasite fauna of Sebastes capensis (Valenciennes, 1833) along its distribution range (Chilean, argentinean and south african coasts), gonzález and Moreno (2005) found that parasite communities did not follow the biogeographic pattern known for other marine species in the southeastern Pacific.similar results were obtained by byrnes and rohde (1992) and Hayward (1997) for the ectoparasite fauna of four species of sparidae of australia and 26 species of sillaginidae from the indo-West Pacific, respectively.However, after examining the endoparasite fauna of S. capensis, gonzález et al. (2006) concluded that its biogeographic pattern was concordant with that of free-living organisms.
Pleuronectiformes (flatfish) are a convenient group for studying biogeography and host-parasite interactions since they are monophyletic in origin (berendzen and Dimmick, 2002;Pardo et al., 2005;Kartavtsev et al., 2007) but have evolved different ecological strategies and life-history patterns.this diversity and their cosmopolitan occurrence offer an excellent opportunity to investigate whether the biogeography of Pleuronectiformes' parasites follows that of their hosts or is highly dependent on the parasite itself, i.e. whether some kind of predictability of host-parasite associations (at the species or family level) can be depicted by evaluating parasite assemblages in several different areas.Comparisons of the parasite species infecting different flatfish species within the same area and infecting the same host species in different areas can be instruments for this investigation.
in a previous study comprising all flatfish species inhabiting the atlantic ocean, Marques et al. (2005) suggested that the parasitological relationships between the widely distributed families scophthalmidae, Pleuronectidae and soleidae, were influenced by the overlapping distribution of species of these families along the Portuguese coast.indeed, given its physical and biological characteristics, a "lusitanian marine province" has been suggested for the Portuguese coast (e.g.Dana, 1853).However, according to the current classification, it constitutes an important border area between the two major zoogeographic regions defined for the north atlantic: the cool temperate and warm temperate (briggs, 1974; gubbay, 1995).
in the present study, the parasite fauna of Pleuronectiformes inhabiting the Portuguese coast was investigated and compared with that known in four other areas (the north sea, north northeast atlantic, Mediterranean sea and northwest african coast) in order to 1) determine the biogeography of parasite taxa and 2) the importance of the Portuguese coast in the establishment of the patterns found.

Sampling and data collection
twenty species of Pleuronectiformes belonging to five different families (table 1) were obtained verdaderamente la distribución de sus hospedadores.estas diferencias reflejan la importancia de la ecología y dispersión de los hospedadores y de los factores ambientales en estos patrones.el carácter de transición de la costa portuguesa y la existencia simpátrica de hospedadores relacionados, contribuyen a que esta región desempeñe una función muy importante en la distribución de los macroparásitos de los pleuronectiformes del atlántico nordeste y del Mediterráneo, ya que promueven la adquisición de nuevas especies de parásitos o el mantenimiento de las relaciones históricas hospedadores-parásitos. from commercial fishing vessels seasonally along the Portuguese coast between January 2003 and July 2005.because all fish of each species were adults of a similar size range and sampled at an identical depth and habitat, they were pooled into one sample.samples collected in different areas (north, centre and south) and seasons were also pooled so that parasite assemblages of each host species could be representative of the entire area and comprise all species, including those with marked seasonal patterns.all sampled fish (480 individuals of each species) were then examined for macroparasites and these were collected, counted and identified to the lowest taxonomic level possible.the current distribution of macroparasites was obtained from the host-parasite database of the natural History Museum (uK), the european register of marine species (erMs -Marbef), the ocean biogeographic information system (obis) and peer reviewed published papers, and classified according to the biogeographic regions defined by briggs (1974) (Fig. 1a): arctic, north temperate, north subtropical, tropical, south subtropical, south temperate and antarctic.in order to investigate macroparasite distribution patterns in the distribution range of the flatfish species considered in the present study (obtained from Froese and Pauly, 2006), five areas were considered (Fig. 1b): the north sea, north northeast atlantic (from northern scotland -57º59'n, 4º57'W to the northwest coast of spain -41º21'n, 8º53'W), Portuguese coast, northwest african coast (from senegal to Morocco) and Mediterranean sea.Data on flatfish macroparasite infections in these areas were obtained from the literature (nicoll, 1915;Williams, 1959;MacKenzie and gibson, 1970;gibson, 1972;rodrigues et al., 1975a,b;Papoutsoglou and Papaparaskeva-Papoutsoglou, 1977;van den broek, 1979a,b;Carvalho-Varela and Cunha-Ferreira, 1987;Petter and radujkovic, 1989;renaud and gabrion, 1988;Zeddam et al., 1988;de Meeus et al., 1992;belghyti et al., 1993;Cordero del Campillo et al., 1994;lile et al., 1994;Petter and Cabaret, 1995;Van Damme and ollevier, 1996;el-Darsh and Whitfield, 1999;Palm et al., 1999;Álvarez et al., 2002;bartoli et al., 2005) and compiled in a presence/absence matrix together with the data resulting from the parasitological examination of individuals collected along the Portuguese coast.

Data analyses
as in most biogeography studies (e.g.Carney and Dick, 2000;McDowall, 2000;rohde, 2002), the present one compiles data from different sources. in order to diminish bias due to methodological differences, only those studies using standard parasi- tological methods, similar and large sample sizes and formally described macroparasites identified to the species level were considered.the validity of species' abundance and its evolution over time could not be determined from the consulted studies, so presence/absence data was selected as the best measure of diversity.although presented in table 1, infection data from host species presenting sample sizes lower than 30 individuals along the Portuguese coast, such as Lepidorhombus whiffiagonis (Walbaum, 1792), Phrynorhombus regius (bonnaterre, 1788), Zeugopterus punctatus (bloch, 1787) and Monochirus hispidus rafinesque, 1814, were discarded from the comparative analyses, since it has been shown that sample size is a major factor influencing the detection of parasitosis (Poulin, 1998;Marques and Cabral, 2007).Hosts infected by fewer than two macroparasite species were also discarded from the comparative analyses, as they proved to be differentiated from all other samples by not being infected.
Presence/absence data of ectoparasites and endoparasites were analysed separately, as these are influenced by different factors (MacKenzie and abaunza, 1998) and show considerable differences in their life-cycles -whereas that of ectoparasites is direct, endoparasites require at least one intermediary host.Multivariate analyses were performed at three different scales (host species, host family and area) using the bray-Curtis dissimilarity since this index accounts for joint presences.a multidimensional scaling (MDs) analysis, using the software PriMer version 5 (PriMer-e ltd., 2001), was applied to host and macroparasite data, in order to reveal the similarity between areas in a two-dimensional space.similarity between host species and host families was investigated by cluster analysis using the complete linkage agglomeration method, and their significance was tested by bootstrap analysis (1000 re-sampling steps) in Clustering Calculator version 1.0 (university of alberta, 2002) in order to reveal the association at the area or family level.results
according to the currently known distribution of the 40 macroparasites identified to the species level, only the digenean Derogenes varicus (Müller, 1784) and the copepod Caligus elongatus von nordmann, 1832 were truly cosmopolitan.on the other hand, the acanthocephalan Acanthocephaloides geneticus (buron, renaud and euzet, 1985) and the copepod Lepeophtheirus europaensis Zeddam, berrebi, renaud, raibaut and gabrion, 1988 had only been reported from flatfishes inhabiting the Mediterranean sea. the majority of macroparasite species had, nevertheless, a broad distribution in the north temperate atlantic region or in the subarctic and north temperate regions of the atlantic and Pacific (table 1).
the information collected for the macroparasite species infecting flatfish in the five areas considered in this study showed that most species were reported in more than one area and infecting more than one host.With a few exceptions, macroparasite species found in Pleuronectiformes were also found in other diverse and unrelated marine fish families (appendix 1, electronic supplementary Material).However, within each host, most macroparasite species were found in only one area and, for the more widespread species, one of the areas of distribution was usually the Portuguese coast (appendix 1, electronic supplementary Material).

Zoogeographic patterns
the multidimensional scaling (MDs) analysis carried out using the presence/absence data of hosts within each area revealed the north sea (ns) to be the most differentiated area and the Portuguese coast (PC) and the Mediterranean sea (Ms) to be the most similar ones (Fig. 2a).However, the MDs analysis performed on macroparasite data revealed a different pattern: the PC was very similar to the north northeast atlantic (na) with regard to endoparasite data (Fig. 2b), and differentiated from all other areas (widely separated from all other areas in the two-dimensional plot) with regard to ectoparasite data (Fig. 2C). the cluster analysis performed on the presence/absence data of endoparasites within each flatfish family (Fig. 3) showed an overall dissimilarity between families, although phylogenetically closer families (soleidae and Pleuronectidae; bothidae and Citharidae) had more similar assemblages, with all clusters being supported by high bootstrap values.the removal of hosts with only one ectoparasite species led to the exclusion of three of the families and the cluster analysis using ectoparasite data was therefore not performed.
When host species were clustered based on their endoparasite fauna (Fig. 4a), most samples were highly differentiated (76.2 bootstrap value associated with the dissimilarity of the seven major clusters), although some species of the same family inhabiting the same area showed small dissimilarities supported by high bootstrap values.the cluster of host species based on their ectoparasite fauna (Fig. 4b) revealed a significantly high similarity between all samples of Platichthys flesus (linnaeus 1758), regardless of the area where the host was collected, and between species of Solea from the Portuguese coast, the clustering of S. lascaris and S. solea being supported by a moderately high bootstrap value (66.7).(e.g. oliva and gonzález, 2005;Mejía-Madrid et al., 2007;Vinarski et al., 2007;Waltari et al., 2007;Pais et al., 2008).Focusing on widely distributed flatfishes, and covering most of their range, the present study identified some patterns in the host-parasite associations for this group of marine fishes.
according to the 'first law of geography', similarity between two observations decays as their geographic distance increases due to a decrease in similarity of their environmental conditions, the existence of geographical barriers between them, or both (nekola and White, 1999).since most endoparasite species infecting flatfish are acquired by ingesting invertebrate intermediate hosts whose dispersal is ultimately influenced by oceanic current patterns, the results obtained for the endoparasite fauna might be the reflection of similar environmental conditions between the Portuguese coast and the north northeast atlantic, which are the most geographically closest areas and have no geographical barriers between them.a decay of similarity in parasite communities over distance has been reported for other hostparasite assemblages (Poulin, 2003;Krasnov et al., 2005;oliva and gonzález, 2005).However, because parasites are influenced by the physical environment and the 'host environment', factors such as the host ecology and host species community composition must also play a role in the associations found. in the present study, most endoparasite species were recovered from the generalist soleidae (link et al., 2005), which feed on the most abundant Crustacea and Mollusca.similar environmental characteristics between the north northeast atlantic and Portuguese coast might therefore lead to similar feeding ecology of soleidae in both areas, also contributing to the similarity found.on the other hand, the distribution of marine ectoparasites, which have direct life-cycles, is mainly determined by the characteristics of the water mass during their brief free-living stage, and by the host's dispersal ability when they are adults (Mackenzie and abaunza, 1998;bush et al., 2001).this gives ectoparasites an advantage in biogeographic studies as their distribution is not confounded by factors related to intermediate host abundance (gonzález and Moreno, 2005), and it was expected that areas with similar host assemblages would also have similar ectoparasite assemblages.However, similarly to reports from other studies using marine fish ectoparasites (e.g.byrnes and rohde, 1992; Hayward, 1997;gonzález and Moreno, 2005), no congruence was found between the distribution of flatfish and that of their ectoparasites, with the Portuguese coast being highly differentiated from all other areas.Host distribution, abundance and behaviour, alone or in conjunction, might have contributed to this differentiation.
Host-parasite evolutionary relationships have been the subject of several studies (e.g.Mas-Coma, 1992;sasal et al., 1998;rohde, 2002;stireman, 2005) and a positive relationship has been found between the diversity of hosts and the diversity of parasites within one area and between parasite and host phylogenies.the present study is in agreement with these findings, since the most diverse family within the Portuguese coast, the soleidae, showed a higher number of parasite species than the least represented families, such as the Citharidae or the bothidae.results also indicate that more closely related families, such as the soleidae and the Pleuronectidae, have more similar macroparasite faunas, as evidenced by the cluster analysis performed on families of hosts.However, the high variability of the macroparasite faunas among hosts inhabiting different areas, evidenced by the low similarities obtained in the cluster analyses of both endoparasites and ectoparasites, suggests that host-parasite associations result mostly from environmental-driven evolution.under this scenario, no predictability of host-parasite associations can be made, except for specific parasites, such as Acanthochondria cornuta (Müller, 1776) and Lepeophtheirus pectoralis (Müller, 1776), whose distribution actually mirrors that of their host, Platichthys flesus (linnaeus, 1758), as revealed by the significant similarity between P. flesus ectoparasite assemblages regardless of the area where the fish were sampled.

Zoogeographic patterns
the endoparasite fauna of flatfishes inhabiting the Portuguese coast appears to be dominated by subarctic and north temperate atlantic species that also occur in other marine fish, their zoogeographical pattern being generally consistent with that of marine free-living species.the wide distribution of Monogenea and Digenea from marine fish and their considerable host specificity (Manter, 1966(Manter, ) led lebedev (1969) ) to distinguish 10 zoogeographical regions differing in the composition of their Monogenea and Digenea faunas. in this zoogeographic model, the Portuguese coast was included in the atlanto-Mediterranean region, which also comprised the Mediterranean and black seas and the atlantic coast of spain and Morocco. the results obtained in this study are therefore consistent with this model since most Digenea are common between the north northeast atlantic and the Portuguese coast or between these areas and the Mediterranean sea, resulting in a similar endoparasite assemblage between these areas, as revealed by the MDs analysis and, to some extent, by the cluster analysis performed on flatfish species.
although flatfish assemblages from the more southern areas were quite similar (Fig. 2a), some of the species that have their northern distribution limit along the Portuguese coast (Arnoglossus thori Kyle, 1913, Bothus podas (Delaroche, 1809), Dicologlossa hexophthalma (bennett, 1831), Microchirus ocellatus (linnaeus, 1758)) are not abundant in this area, resulting in low sampling sizes and their exclusion from the comparative analyses of macroparasite assemblages, and increasing the discrepancy between host and macroparasite infections results.ecological host-switching (Poulin, 1998;Page, 2003), through the contagious transmission of generalist ectoparasites between similar host species, or between widespread host species and those living near the limit of their distribution range, might be promoted by the high overlap of host species along the Portuguese coast.However, the relatively sedentary behaviour of adult flatfishes might contribute to a minimal dispersion of these ectoparasite species, resulting in a higher richness in this area and its differentiation from the other four considered.
Despite all the ecological factors pointed out so far, historical reasons cannot be ruled out to explain the results found in the present study.the origin and distribution of the great majority of our present species probably took place in tropical centres during the Pliocene and Pleistocene (briggs, 2006), and atlantic and Mediterranean marine faunas are no exception.results from a palaeontologic study conducted in the central region of the Portuguese coast (40ºn), indicates that the Pliocenic faunal association in this area had a typical atlantic affinity, with no fossil species from the Mediterranean being found and two exclusively atlantic fossil species being present (nolf and Marques da silva, 1997).after the mass extinction of Mediterranean fauna due to the Messinian salinity crisis, in the early Pliocene, this area was colonised by warm temperate atlantic species that are found here, and on the western coast of africa, their refuges during the Pleistocene glaciations, before they recolonised temperate eastern atlantic waters when conditions were favourable (almada et al., 2001;Domingues et al., 2008).these events offer a possible explanation for the similarity found between flatfish communities along the Portuguese coast, Mediterranean sea and northwest african coast, and also for the differences found between macroparasite assemblages: some parasite lineages established in flatfishes might have been lost when hosts migrated southwards (warm-temperate species) during glaciation peaks, due to extreme differences in environmental conditions or the absence of the required intermediate hosts to complete their life-cycles; glacial advances and retreats during the Pleistocene and the recent (less than 10000 years, after the last ice-age) colonisation of the Portuguese coast by some subtropical species (e.g.D. hexophthalma, M. ocellatus, also found in the warmer waters of the Mediterranean and northwest african coast) have not yet allowed the development of stable hostparasite relationships.this is supported by the fact that most macroparasite species found in the present study had subartic and north temperate distributions and infected hosts with temperate affinities, suggesting that similarities found between parasite faunas from Pleuronectiformes inhabiting the north northeast atlantic and the Portuguese coast are the result of long-lasting interactions between parasites and their hosts.
results of the present study point out the high importance of the Portuguese coast as a transition area in shaping macroparasite assemblages of Pleuronectiformes species occurring in the northeast atlantic and Mediterranean sea and in the establishment of host-parasite relationships, as suggested by Marques et al. (2005).the sympatric occurrence of closely related host species along this area, allows the colonisation of new, but similar, hosts by macroparasites specific at the genus or family level through evolutionary events, but also the continuum of host-parasite relationships involving specific parasite lineages.However, phylogenies of both parasites and hosts are essential in the study of the biogeography of the extant biodiversity (Cowie and Holland, 2006), and knowledge of the macroparasite faunas infecting Pleuronectiformes in more remote regions and of the rare Pleuronectiformes species occurring along the Portuguese coast are needed in order to clarify the relationships found in the present study.aCKnoWleDgMents this study was partially financed by the Fundação para a Ciência e a tecnologia (FCt), Portugal, through the grant attributed to J.F. Marques (grants sFrH/bD/8983/2002 andsFrH/bPD/26999/2006). this study was co-funded by the european union through the FeDer -Fisheries Programme (Mare).the authors would like to thank Dr. r. Poulin and Dr. a. Patterson for their comments on previous versions of this manuscript.the authors would also like to express their gratitude to Dr. enric Massutí, scientific editor of scientia Marina, and two anonymous referees for their comments and suggestions.all the experiments comply with the current laws of the country in which they were performed.

Fig. 3 .
Fig. 3. -Cluster analysis of the Pleuronectiformes families based on the presence/absence data of endoparasite species using the bray-Curtis similarity index and the complete linkage algorithm.numbers in cluster nodes indicate bootstrap values (1000 re-sampling steps).

Fig. 2 .
Fig. 2. -Multidimensional scaling analysis of the five areas, based on the presence/absence data of (a) Pleuronectiformes species, (b) endoparasites and (C) ectoparasites.abbreviations for areas are given in the legend of Figure 1.

Fig. 4 .
Fig. 4. -Cluster analyses of the Pleuronectiformes species within each area based on the presence/absence data of macroparasite species using the bray-Curtis dissimilarity index and the complete linkage algorithm: (a), endoparasites; (b), ectoparasites.acronyms are defined by five characters, with the first corresponding to the family, the following two to the species and the last two to the area.Family: C, Citharidae; b, bothidae; s, scophthalmidae; P, Pleuronectidae; l, soleidae.abbreviations for species are given in table 1 and for areas in the legend of Figure 1.numbers in cluster nodes indicate bootstrap values (1000 re-sampling steps).