Agri-environmental assessment of extractable soil phosphorus at the regional scale

Agri-environmental assessments at the regional scale are restricted by the amount and quality of input data. Such data may be available in public databases. In this study we estimated the extractable soil-P (ESP) in the Sud Milano Agricultural Park in northern Italy. Information about agricultural activities was integrated with ESP values across the area, using a large database of measured soil properties and crop management data collected from individual farms. Four interpolation methods were tested: two forms of ordinary kriging, kriging with external drift and a hybrid form (section-specific mosaic kriging). After splitting the dataset into three sub-sets, ESP was assessed at unsampled farms. High ESP levels were found in maize fields and on animal farms due to large amounts of P fertilizers applied in maize, particularly on animal farms. Using a reference threshold of 20 mg P per kg soil, most of the soils in the area were classified as being rich in ESP. As a result, it was concluded that P fertilization could be suspended in many cases for several years without crop yield decrease. Mosaic kriging showed a lower standard deviation of the prediction error and less smoothing and thus provided a better representation of the actual spatial variation.