Covariance in water- and nutrient budgets of Dutch peat polders: what governs nutrient retention?

Water and nutrient budgets were constructed for 13 low-lying peat polders in the Netherlands that varied in elevation relative to sea level (?0.2 to ?2.4 m below sea level), land use (7–70% of the total polder area covered by agriculture; largely dairy farming), and surface water prevalence (6–43%). Water balances were verified with chloride budgets and accepted when both met the criterion (total inflows ? total outflows)/(total inflows) <0.05. Apart from precipitation and evapotranspiration (overall means 913 vs. 600 mm), in- and outlet (171 vs. 420 mm) as well as in- and outward seepage (137 vs. 174 mm) were important items in the water budgets. Nutrient budgets, however, were dominated by terms related to agricultural land use (~60% of all inputs, 90% of N-removal and 80% of P removal) rather than water fluxes (8% and 5% of N and P inputs; 6 and 18% of outputs). After agriculture (200 kg N ha^?1 y^?1), mineralisation of the peat soil and atmospheric deposition appear to be important inputs (about 94 and 21 kg N ha^?1 y^?1). Major output terms were agricultural output (209 kg N ha^?1 y^?1) and denitrification (95 kg N ha^?1 y^?1). The average N budget was in balance (difference ~1 kg N ha^?1 y^?1), whereas P accumulated in most polders, particularly those under agriculture. The mean P surplus (15 kg P ha^?1 y^?1 in the 9 mainly agricultural polders) corresponds well with the accumulated difference observed elsewhere (700 kg P ha^?1 in the upper 50 cm in a nature reserve versus 1400 under agriculture) after over 50 years of dairy farming. Bulk retention of N and P in these polders is taking place in the peat soil, through temporary sorption to the matrix and N is lost through denitrification. In a principal components analysis combining land use, landscape pattern, water balance and nutrient budget terms, the three-first principal components explained 63% of the variability. The first component (PC) correlated strongly with the percentage of land under agriculture (r = 0.82) and negatively with the percentage covered by surface water (r = ?0.74). Most input and output terms of the nitrogen budget also correlated with this PC. The second PC covaried distinctly with the total area of a polder (r = ?0.79) and human population density at municipality level (r = 0.75). Phosphorus loads in inlet and outlet water correlated with this PC. This suggests that the variability in nutrient budgets among polders is largely governed by agricultural land use.