Parasite communities of the European Cod "Comunidades parásitas de Bacalao en aguas de Europa"

The structure of the metazoan parasite faunas and communities in Atlantic cod, Gadus morhua L., from six NE Atlantic regions [Baltic, Celtic, Irish and North seas, Icelandic waters and Trondheimsfjord (Norway) and two fish farms in Iceland and Scotland] has been studied. 1,254 fish were sampled in the NE Atlantic during 2002-2003. Altogether 57 parasite taxa were found. The predominant groups in regional parasite faunas were the trematodes (19 species) and the nematodes (13 species). Nine parasite species were found for the first time in cod (Diclidophora merlangi, Rhipidocotyle sp., Fellodistomum sp., Steringotrema sp., Schistocephalus gasterostei, Cucullanus sp., Spinitectus sp., Acanthochondria soleae and Chondracanthus ornatus). Contracaecum osculatum, Hysterothylacium aduncum and Echinorhynchus gadi were the only parasite species found to infect farmed fish. Regional parasite faunas showed lower richness with respect to the total list (c. 65%) with a notable decrease in the Baltic Sea and Trondheimsfjord (21 and 32%, respectively). Eleven species were present in all regions: Lepidapedon elongatum, Anisakis simplex, C. osculatum, H. aduncum, H. rigidum, Pseudoterranova decipiens, Ascarophis crassicollis, Capillaria gracilis, Corynosoma semerme, C. strumosum and E. gadi. Despite of the small size of the monogenean D. merlangi recovered in the present study, the Analysis of Principal Components showed that morphologically they are more similar to D. merlangi from whiting (Merlangius merlangus, type-host), than to other congeneric species from the North Atlantic, thus supporting their assignment to D. merlangi. The aspect and smaller size of the oöcytes suggested that D. merlangi on cod could not produce viable ova, although a specimen exhibited an egg with normally sized and shaped shell. The most species rich and abundant parasite communities were observed in cod from the open water regions (Celtic, North and Irish seas and Icelandic waters). Overall, parasite infracommunities exhibited lower predictability than component communities, due to the fact that only a restricted set of the species contributing to the similarity between component communities exhibited high abundance and dominated infracommunities. The multivariate techniques applied to examine similarity patterns of component communities across regions exhibited good agreement and detected distinct compositional segregation of those in cod from the two low-salinity regions. The highest homogeneity with respect to the composition and structure of parasite communities was observed Celtic, Irish and North Sea cods. Decay of similarity with geographical distance was observed in component communities but not in regional parasite faunas, the higher homogenisation of the latter being related to the migratory behaviour of cod and the domination of generalist parasites widely distributed. The spatial compositional autocorrelation exhibited by component communities and the substantially higher rates of similarity decay compared to other marine fish systems indicate that communities in cod are strongly constrained by the spatial configuration of locations and the dispersal abilities of parasites. Nested subset analyses revealed non-random patterns of faunal/community composition with poor faunas from low-salinity regions (Baltic Sea and Trondheimsfjord) nested in the richer faunas/communities from the high-salinity open water regions. The comparison of the learning behaviour of the three classification approaches, Random Forests (RF), Linear Discriminant Analysis and Artificial Neural Networks, using the same version of the parasite community data, revealed that RF appears as the best classifier. Anisakid nematodes, C. cirratus, D. varicus, H. communis, E. gadi, and C. adunca were selected as important for RF model development. The high accuracy of the predictive models developed for the Baltic and Icelandic samples indicate that the populations of these stocks can be confidently differentiated from the other stocks studied in the NE Atlantic. These results suggest that parasite community data can be used successfully to discriminate cod populations (putative stocks) of the NE Atlantic cod using RF. __________________________________________________________________________________________________ RESUMEN La composición y estructura de la fauna de parásitos metazoos del bacalao, Gadus morhua, es el objeto del presente estudio. Se recolectaron 1.254 peces durante 2002-03 procedentes de 6 áreas de pesca del Atlántico Nororiental (mares Báltico, Celta, de Irlanda y del Norte, Islandia y el fiordo de Trondheim), así como 2 granjas marinas situadas en Islandia y Escocia. Se recolectó una fauna rica formada por 57 taxones parásitos diferentes. Siete especies, Diclidophora merlangi, Rhipidocotyle sp., Fellodistomum sp., Steringotrema sp., Cucullanus sp., Spinitectus sp. y Chondracanthus ornatus son primeras citas en este hospedador. Once especies estuvieron presentes en las parasitofaunas de las 6 regiones naturales estudiadas: Lepidapedon elongatum, Anisakis simplex, C. osculatum, H. aduncum, H. rigidum, Pseudoterranova decipiens, Ascarophis crassicollis Capillaria gracilis, Corynosoma semerme, C. strumosum y Echinorhynchus gadi. Las comunidades parásitas de bacalao de las regiones de baja salinidad (mar Báltico y fiordo de Trondheim) estuvieron caracterizadas por las más bajas riquezas, abundancias y diversidades. La sincronización espacial de la composición de las comunidades componentes y las mayores tasas de declive de similitud exhibidas en comparación otras comunidades, indican que las comunidades componentes están fuertemente comprometidas por la configuración espacial de las regiones y la facilidad de dispersión de los parásitos. La ausencia de aleatoriedad en la composición de las faunas y comunidades componentes estuvo claramente relacionada con la salinidad: las comunidades empobrecidas de las regiones de baja salinidad estaban anidadas dentro de las faunas y comunidades más ricas de las regiones de mayor salinidad en mar abierto. La comparación del comportamiento de la metodología de clasificación de grupos, Random Forests (RF), con otros algoritmos, Análisis Discriminante Linear y las Redes Neurales Artificiales, utilizando los mismos datos recolectados en las cinco regiones del Atlántico nororiental, reveló que RF era la mejor herramienta de clasificación. Los nematodos anisáquidos fueron identificados como importantes para desarrollar los modelos junto con otras especies como Cucullanus cirratus, D. varicus, Hemiurus communis, E. gadi y Clavella adunca. Los buenos resultados de discriminación obtenidos, incluso para este pez migratorio, reflejan el elevado potencial de RF para desarrollar modelos predictivos usando datos que son complejos y ruidosos