Echantillonnage des prairies pâturées pour quantifier l'azote potentiellement lessivable : quel schéma pour quelle précision ?

Sampling scheme to quantify nitrogen leaching risk in grazed grassland: which scheme for which accuracy? In grazed grasslands, the heterogeneity of urine patches distribution needs adapted sampling scheme in order to evaluate N-NO3 - leaching risks. In such context, the aim of this paper is to define the accuracy that could be expected from the sampling scheme applied in this agro-ecosystem (one average sample including 30 core samples taken in the 0-30 cm soil layer) and the sampling scheme to be applied in order to reach a pre-defined level of standard error. The data bases mobilized were mainly 78 grasslands sampled in 2004 or in 2005 in 24 dairy farms and 8 grasslands sampled during four years (2004 to 2007) in the experimental site in Gembloux. Our results underlined the significant increase of the standard deviation with the concentration mean. So, the mean explained 63 and 49% of the standard deviation increase observed, respectively, in 2004 and 2005. In such context, the sampling scheme applied in the Sustainable Nitrogen Management Program (one composite sample of 30 core samples taken in the 0-30 cm soil layer) allowed to reach a precision of 10 kg N-NO3 -.ha-1 in 35% of the grazed grasslands sampled. To shift to three composite samples per hectare allows reaching such an accuracy in 90% of the parcels showing an average nitrite nitrogen content lower than 20 kg N-NO3 -.ha-1. For highest values, the samples number necessary to reach such a precision level increases quickly. In such context and in order to identify false positive parcels in audited farms, an important tolerance margin is defined around the threshold value. This value is defined yearly in farms integrating good practices in terms of nitrogen resources management. The use of alternatives indicators to evaluate N-NO3 -.ha-1 leaching risk is also evaluated.