Applying geostatistics to identification of spatial patterns of fecal contamination in a mussel farming area (Havre de la Vanlée, France)

Microbiological quality of shellfish production areas along the French coast is assessed through quantification of fecal coliforms, commonly used as indicators of fecal pollution. Their concentration is measured in filter-feeding molluscs, presumed to integrate highly fluctuating quantities of these germs from the surrounding filtrated water. In the context of a bacteriological monitoring network, knowledge of contamination spatial structures may help for further optimal sampling designs. This study aims at applying geostatistical techniques to describe and characterize the spatial structure of the fecal contamination in mussels (Mytilus edulis) over a given production area (Havre de la Vanlée, France), located in macrotidal waters. In comparison with common interpolation methods, the kriging estimator, a basic tool in geostatistics, presents the major advantage of providing unbiased estimates with known and minimum variances. It thus appears to be the most powerful procedure to produce contour maps for the concentration estimates and their standard deviations. Differences in mean concentration between the two conducted surveys are compared with differences in concentration found in continental inputs. Although two surveys are not sufficient to assess accurately spatio-temporal variabilities, the use of kriging reveals high spatial heterogeneity, with different structures and meteorological conditions from one survey to the other. This leads to questioning the validity of a monitoring based on a few sampling stations over a large area sampled monthly to estimate a mean bacteriological level.