De l’observation des pratiques d'échantillonnage des eaux et sédiments continentaux à l'élaboration de guides techniques : approche AQUAREF et résultats

[Departement_IRSTEA]Eaux [TR1_IRSTEA]BELCAInternational audienceThe Water Framework Directive (WFD, 2000/60/CE) and the daughter directive (2013/39/UE) impose regular surface water and sediment monitoring in order to establish water bodies chemical and ecological status. This includes the monitoring of several organic and inorganic substances in the long term. Quality of the monitoring data is a major issue to ensure a reliable decision-making process in order to reach good chemical and ecological water status in the future. Environmental data acquisition includes sampling, sample pre-processing and analysis steps. These important steps of the analytical chain strongly influence the quality of the final results; so they should be controlled to obtain representative samples. Nowadays, sampling operations are still not fully harmonized for continental waters and their impact on the variability of the final result remains largely unknown. In fact, intercomparison trials are used by routine laboratories to evaluate their performances and improve their methods. They also permit to estimate, for a group of laboratories, the expected variability of results for the analysis of a same sample. But such trials are rare for sampling activities especially for continental waters. Since 2007, AQUAREF (the French National Reference Laboratory in support of aquatic environment monitoring) has been undertaking several intercomparison (or collaborative) trials on continental water and sediment sampling in order to improve our knowledge on current practices and on the influence of sampling on the overall monitoring data quality. Also, these trials enable to propose specific recommendations dealing with the improvement and the harmonization of sampling conditions. AQUAREF collaborative trials on sampling have two main objectives: 1) To observe, compare and evaluate sampling practices of routine operators. 2) To evaluate the impact of sampling operation on the variability of results for chemical substances. In 2014, AQUAREF carried out a collaborative trial on the sampling of stream sediments. A preliminary study, conducted in 2013, highlighted the need to select two different stations in order to fulfill the trial objectives. Indeed, to meet the first objective, and evaluate the global sampling variability, it was necessary to select one station representative of the routine sampling performed within the monitoring control network of the WFD (station A). While, for the second objective, in order to avoid any excessive matrix heterogeneity, another station presenting homogeneous sediment was required (station B). The trial was realized from the 25th to the 29th of August 2014. Nine teams of 2 sampling operators from 7 sampling organisms took part in the trial. Sampling organisms are laboratories or technical consultants. They were selected by the French Water Agencies which are responsible for WFD monitoring in France. Each sampling team had to collect a representative sample (at station A) in the same way as they usually do within the WFD monitoring control network; they also had to collect one sample in duplicate at station B. In total, 34 individual samples were collected during the trial. Prior to the trial, a survey on participant practices was realized. This survey included questions on sampling practices, sampling material and quality assurance system. Then, during the trial, several observers followed all the trial steps and completed a preset questionnaire, which enabled to establish a picture of actual sampling practices. The data treatment of the survey and the questionnaire permitted to fulfill the first objective of the collaborative trial. Then, to evaluate the impact of sampling operations, the foreseen methodology allowed us to dissociate the relative contribution of variability related to analysis, to sampling or to a “sampling operator effect”. Besides granulometry and total organic carbon, 83 substances (metals, PCB , PBDE , phthalates, alkylphenols, PAH , organochloride pesticides and organotins) were analyzed in the collected samples. In order to reduce the analytical variability, each parameter was analyzed by only one laboratory. The variabilities and the presence of aberrant values were estimated using statistical tests (i.e., robust analysis of variance RANOVA, Cochran and Grubbs tests). The analytical results of the samplings realized on station A (with heterogeneous deposit) were used to estimate the total variability of monitoring data. This study has shown that the differences in the nature of the collected sediment could increase the total variability by a factor of about 2. In general, AQUAREF intercomparaison trial results have highlighted, on the one hand, the importance of cleaning all sampling tools, the impact of sampling tool materials and, on the other hand, the variability contributed by a non-harmonized sampling protocol. For example, the use of plastic sampling tools or cleaning with detergent has been suspected to bring contamination by phthalates for sediment samples. The conclusions of these collaborative trials were systematically presented and discussed with the participants. The outcome of this work will enable updating AQUAREF technical guidance on sampling and analysis of continental water and sediment for the WFD