Field-scale variability of soil test phosphorus and other nutrients in grasslands under long-term agricultural managements

Abstract

Field-scale variation of soil nutrients in grassland is becoming important because of the use of soil-nutrient information as a basis for policies such as the recently introduced EU Nitrates Directive. This study investigates the field-scale variability of soil-test phosphorus (STP) and other nutrients in two grasslands with a long-term history of poultry litter application. Two fields (field 1 for silage and field 2 for grazing pasture) were selected, and soil samples were collected based on 12 m by 12 m (field 1) and 15 m by 15 m (field 2) grids. Data were analysed using conventional statistics, geostatistics, and a geographic information system (GIS). In field 1, STP values ranged from 12.4 to 90 mg L–1 (average 38.5 mg L–1). In field 2, STP values ranged from 4.3 to 130.0 mg L–1 (average 21.4 mg L–1). Attention should be paid to long-term poultry application, as the average STP values in both fields were much greater than the recommended agronomic optimum STP status in Ireland of 8 mg L–1. Coefficient of variation values of soil nutrients in field 2 were much higher than those in field 1. Log-transformation and Box–Cox transformation were applied to achieve normality. Statistically significant (P < 0.01), positive correlations between P and other nutrients were found in both fields. Exponential and spherical models were fitted to the experimental variograms of STP in fields 1 and 2, respectively. Compared with the counterparts in field 1, soil nutrients in field 2 had larger ‘nugget-to-sill’ values, revealing that sheep grazing could weaken the spatial auto-correlation of soil nutrients. A grid of 60 m by 60 m was recommended for soil sampling in grassland, based on this study. High STP concentrations in field 1 were in the north-eastern side, which was related to uneven poultry litter application. Strong spatial similarity of low STP, magnesium, and pH values in their spatial distribution were found in field 2, confirming their strong statistical correlation.