Nitrogen in river basins: sources, retention in the surface waters and peatlands, and fluxes to estuaries in Finland.

Nitrogen export from diffuse and point sources and its retention in the major river basins of Finland is quantified and discussed. The estimated total export from river-basins in Finland was 119,000 tonnes N a(-1) for the period 1993 to 1998 based on N export from different land use types defined in a GIS-based assessment model, incorporated with estimates of N inputs from atmospheric deposition and point sources. Agriculture contributes 38% of the total export, varying in the range 35-85% in the south-western basins and 0-25% in the northern basins. This estimate of N export from agriculture was based on regional N balances together with data from small agricultural research catchments. Forestry contributes on average 9%, with increasing dominance towards eastern and northern parts of the country: from 2% to 15% in the southern-mid-western Finland basins to 10-30% in the large northern basins. 'Background' N export from forests on both mineral and organic soils contributes 27% on average; in the northern basins it may contribute from 40% up to 90% of the total load. The estimate was calculated based on practically all data available from 42 small, experimental catchments in Finland. Of the total N input to Finnish river-systems, 0% to 68% was retained in surface waters and/or peatlands, with a mean retention of 22%. The highest retention of N (36-61%) was observed in the basins with the highest lake percentages. The lowest retention (0-10%) of N was in the coastal basins with practically no lakes. In the national N mass balance, 38,000 tonnes N a(-1) (32%) was estimated as lake retention and 4,000 tonnes N a(-1) (3%) as retention in peatlands. On the basis of mass balances and sensitivity analysis, retention was in most cases estimated to be in the range of 7.5-12.5 kg ha(-1)a(-1) in lakes and 0-1.5 kg ha(-1)a(-1) in peatlands. The model results were tested using the split-sample technique and uncertainty estimates for different data sources are provided and discussed.