Evaluation of the spatial uncertainty of agro-ecological assessments at the regional scale : The phosphorus indicator in northern Italy

Agro-ecological indicators are simple conceptual models to carry out agro-environmental assessments. Their use requires data that need to be obtained at low cost, frequently avoiding direct measurements. The quality of input data thus may limit the usefulness of the indicators. One source of uncertainty is the spatial interpolation of inputs, used to provide indicator values throughout an area. This paper explores the uncertainty of the inputs, and its effect on the output of one indicator, the phosphorus indicator (IP). The indicator evaluates the appropriateness of phosphorus (P) use by farmers, assigning bad scores to over- and under-fertilization. We evaluated its use for P management in the Sud Milano Agricultural Park (northern Italy). We used data contained in a large database of soil and farm properties as well as crop management information at the cadastral parcel level to calculate IP values. The uncertainty of a single input variable (extractable soil P) was tested to quantify the corresponding uncertainty of the indicator. The results show that the variability of IP is high and within 80% of the analyzed area excessive applications of P fertilizers are made, in particular in animal farms. In most cases, uncertainty is not relevant, as it is either very low, or (if high) it is related to extremely low indicator values: in these cases, the assessment of P management is unaffected by the uncertainty of the indicator. The results show that in this area P fertilizers should be applied at lower doses, or not applied at all. An extension service might help farmers with fertilizer management, reducing resource use, environmental pollution and costs. This study shows that uncertainty analysis is a crucial component of environmental assessments, and that the importance of uncertain input data needs to be evaluated on a case-by-case basis.