Pasture-based grazing systems contribute to the excessive nutrients found in some streams in south-eastern Australia. This study investigated phosphorus (P) exported in runoff from a rain-fed dairy pasture (Darnum) and 4 bays of irrigated dairy pasture (MRF). Runoff was monitored for 7 years at Darnum and 2 years at the MRF to identify factors associated with the variation in total P (TP) concentrations between events.

The flow-weighted mean annual P concentrations in runoff varied between 3.3 and 28.2 mg TP/L for Darnum and 6.2 and 31.5 mg TP/L for the MRF. The relationships between TP concentrations in runoff and days between fertiliser application and runoff, days between grazing and runoff, and total storm flow were examined using an additive component model that explained 61% and 70% of the variation in log-transformed TP for Darnum and the MRF, respectively. The interval between application of fertiliser and runoff and the effect of year were highly significant and explained most of the variation in TP. Grazing and fertiliser application were identified as the major factors that may affect TP concentrations that the land manager can control (preventable). The estimates of year effect (i.e. the component of TP not explained by the other variables and over which the land manager had no apparent means of control) ranged from 1.60 mg (s.e. 1.99) to 7.14 mg (s.e. 1.90) TP/L in non-drought years (>45 kL/ha runoff annually). The year effect averaged 5.7 and 6.9 mg TP/L for Darnum and the MRF, respectively. It is shown that an additive component model provides a useful structure for investigating similar, field-scale data.