Spatial Relationships between Polychaete Assemblages and Environmental Variables over Broad Geographical Scales

LISANDRO BENEDETTI-CECCHI1*, ; KATRIN IKEN2,; BRENDA KONAR2; JUAN CRUZ-MOTTA; ANN KNOWLTON2; GERHARD POHLE4; ALBERTO CASTELLI1; LAURA TAMBURELLO1; ANGELA MEAD5,; TOM TROTT6,; PATRICIA MILOSLAVICH3; MELISA WONG; YOSHIHISA SHIRAYAMA8,; CLAUDIO LARDICCI1; MARIA GABRIELA PALOMO; ELENA MAGGI

PLOS ONE

PUBLIC LIBRARY SCIENCE

Año: 2010 vol. 5 p. 1 - 10

1932-6203

This study examined spatial relationships between rocky shore polychaete assemblages and environmental variables over broad geographical scales, using a database compiled within the Census of Marine Life NaGISA (Natural Geography In Shore Areas) research program. The database consisted of abundance measures of polychaetes classified at the genus and family levels for 74 and 93 sites, respectively, from nine geographic regions. We tested the general hypothesis that the set of environmental variables emerging as potentially important drivers of variation in polychaete assemblages depend on the spatial scale considered. Through Moran´s eigenvector maps we indentified three submodels reflecting spatial relationships among sampling sites at intercontinental (>10000 km), continental (1000–5000 km) and regional (20–500 km) scales. Using redundancy analysis we found that most environmental variables contributed to explain a large and significant proportion of variation of the intercontinental submodel both for genera and families (54% and 53%, respectively). A subset of these variables, organic pollution, inorganic pollution, primary productivity and nutrient contamination was also significantly related to spatial variation at the continental scale, explaining 25% and 32% of the variance at the genus and family levels, respectively. These variables should therefore be preferably considered when forecasting large-scale spatial patterns of polychaete assemblages in relation to ongoing or predicted changes in environmental conditions. None of the variables considered in this study were significantly related to the regional submodel