This study examined productivity, nitrogen (N) flows and N balances in grassland-based systems of dairy production in Ireland. There were four stocking densities of dairy cows on grass ⁄ white clover pastures and four inputs of N as fertilizers, concentrates and biological fixation over 2 years; 2001 and 2002. Annual stocking densities were 1AE75, 2AE10, 2AE50 and 2AE50 cows ha )1 . Associated N inputs were 205, 230, 300 and 400 kg ha )1 respectively. There were eighteen cows per system. Cows calved within a 12-week interval in spring with a mean calving date of 28 February and lactation extended until mid-December in each year. There were no differences in annual milk yield (6337 kg cow )1 ; s.e.m. 106AE1), live weight or body condition score. Pre-grazing N concentrations in herbage increased (P < 0AE001) with increasing N input, whereas there were no differences in N concentrations in silage reflecting optimum N inputs for silage production. Grazed herbage accounted for 0AE64, silage 0AE26 and concentrates 0AE10 of annual dry matter consumed by the cows. Annual intakes of N ranged from 144 to 158 kg cow )1 and were mostly influenced by N concentration in grazed herbage. Annual output of N in milk and liveweight change was 38 kg cow )1 and was not different between systems. Annual N surpluses increased with increasing N inputs from 137 to 307 kg ha )1 , whereas the proportion of N inputs recovered in products declined from 0AE34 to 0AE24. More efficient N use was associated with lower N inputs and in particular lower N concentrations in grazed herbage.

This study examined the quantity of mineral N in soil and nitrate-N losses to groundwater from grasslandbased dairy production in 2001 and 2002. There were four treatments with different inputs of N, through fertilizers, concentrates and biological fixation, and four associated stocking densities. Nitrogen inputs were 205, 230, 300 and 400 kg ha )1 , respectively, and annual stocking densities were 1AE75, 2AE10, 2AE50 and 2AE50 cows ha )1 . There were 18 cows per treatment. Grazed herbage accounted for 0AE64, grass-silage for 0AE26 and concentrates 0AE10 of annual DM consumed by the cows. Quantities of mineral N (nitrate-N and ammonium-N) in soil were measured, following extraction in 2 M KCl (1:2 w ⁄ v) shaken continuously for 2 h, on three occasions between late September and early February each winter. Concentrations of nitrate-N in groundwater from wells inserted vertically to a depth of 1 m were determined throughout both winters. Quantities of mineral N in the soil increased (P < 0AE001) with higher N inputs and declined (P < 0AE001) with later sampling date. There were no relationships between nitrate-N concentrations in groundwater and N inputs, N surpluses, deposition of excreta-N at the soil surface and soil mineral N during both winters. Low losses of nitrate-N to groundwater were primarily attributed to high rates of denitrification associated with a heavy soil texture, wet anaerobic soil conditions, relatively high organic carbon contents throughout the soil profile and mild soil temperatures throughout the year. Uptake of N by herbage made an important contribution to low N losses over the winter.

Current policy instruments under the EU Water Framework Directive (WFD) to mitigate phosphorus (P) loss require that P use on farms is managed through regulation of farm gate P balances. Regulation at farm scale does not account for spatial variability in nutrient use and soil fertility at field scale, affecting the costs and effectiveness of farm gate measures. This study simulated the implementation of a P loss mitigation measure coupled with improving soil fertility so that farm productivity would not be compromised. The measure was simulated at field scale and the costs and effectiveness assessed at farm scale. Effectiveness was expressed as the time taken for excessive soil P levels to decline to levels that matched offtakes and this varied temporally and spatially within and between farms ranging from 1 to 8 years. Sub-optimum soil fertility was corrected on all fields across both farms, with applications of other soil nutrients and lime to protect productivity. An increase in costs ranging from 1.5-116% was predicted in the first two years of the measure on both farms after-which savings of 15-31% were predicted for each subsequent year until the measure was effective in year 9. Despite initial cost increase, there was no statistically significant difference in costs over the time taken for the measure to be effective, when compared to baseline costs. Successful implementation of measures should consider the impact on farm costs and time taken for measures to environmentally effective. Adoption of measures could improve if demonstrating to farmers that costs will not vary significantly from current practice and in time may results in savings if measures are paired with correcting soil fertility and increasing yields. This 'win-win' approach could be used into the future to ensure successful implementation and uptake of measures within the farming community.