Nutrient balance at chain level : A valuable approach to benchmark nutrient losses of milk production systems?

A nutrient balance approach is often used to quantify losses of nutrients, such as nitrogen and phosphorus, that contribute to environmental problems such as eutrophication. A nutrient balance generally is computed at farm level, implying that nutrient losses related to pre-farm processes, such as production of purchased feed, are neglected. Using a nutrient balance at farm level to benchmark livestock systems or individual farms that differ in, for example, amount of purchased concentrates, however, may lead to biased conclusions. To determine whether a nutrient balance that accounts for losses during production of purchased feed (i.e. a chain balance from cradle-to-farm-gate) are more suited to benchmark nutrient losses of milk production systems or individual farms than a nutrient balance at farm level, we analysed 19 Irish grass-based dairy farms and 13 Dutch concentrate-based dairy farms. For each farm, we computed nitrogen and phosphorus losses at farm and chain level (i.e. from cradle-to-farm-gate, including losses from purchased feed production), and expressed these losses per ton fat-and-protein-corrected milk. An independent T-test and Wilcoxon-Mann-Whitney test were used to examine the difference in mean losses at farm and chain level between Irish and Dutch systems. Regression analysis was used to determine if the ranking of the farms changes using a nutrient balance at farm or chain level. Results show that on average, Irish farms had higher nitrogen losses per ton milk than Dutch farms, both at farm (Irish = 20; Dutch = 8 in kg N/ton milk) and chain level (Irish = 22; Dutch = 11 in kg N/ton milk). Phosphorus losses per ton milk, on the other hand, did not differ between Irish and Dutch farms at farm (Irish = 0.3; Dutch = 0.1 in kg P/ton milk) or chain level (Irish = 0.8; Dutch = 1.0 in kg P/ton milk). Regression analysis revealed that the nutrient balance at chain level could be accurately predicted from the nutrient balance at farm level (R2 = 0.992 for N; R2 = 0.910 for P); whereas in case of phosphorus, the slope tended to differ between Irish and Dutch farms (p <0.10). Ranking 32 farms based on the nitrogen balance at farm or chain level, therefore, showed a similar pattern, whereas the ranking pattern based on the phosphorus farm balance differed from the pattern based on the chain balance. We concluded, therefore, that to benchmark nutrient losses of dairy systems, a nutrient balance at farm level can be used if differences in on-farm losses between systems are large, and pre-farm losses related to, e.g. production of purchased concentrates, are relatively unimportant. To benchmark individual farms, a nutrient balance at farm level can be used only if changes in pre-farm losses per unit change in on-farm losses are similar across farms. A chain level balance of a sample set, however, is required to verify these conditions.