Förslag till bedömningsgrunder för näringsämnen i sjöar och vattendrag

Assessment criteria for classification of ecological status are of central importance to water management. The goal is for all waters bodies to achieve at least good status, defined as the maximum acceptable deviation from a reference value. Classification is primarily assessed using biological quality elements, with physicochemical variables primarily used to support and interpret biological classifications. In practice, however, physicochemical variables are often used for classifying water bodies. This is because for many water bodies only physicochemical data are available and also because physicochemical variables are directly linked to several types of impacts, and therefore provide a basis for designing and implementing best management measures.Current classifications of ecological status using nutrients has revealed a few shortcomings. The use of forest lakes ( 10% agricultural land in the catchment area), modelled root zone leakage from unfertilized and unharvested grassland multiplied by 0.5 has been used to estimate reference conditions. The factor 0.5 was arbitrarily chosen to account for retention between the root zone and the watercourse. There has also been no harmonization of assessments based on chemical (elevated nutrients) on biological responses.The aim of this project was to improve estimates of reference values for total phosphorus (TotP) in lakes and watercourses by taking into account the variability of physicochemical and land use/cover in the agricultural landscapes, as well as to harmonize class boundaries of ecological status between TotP and biological quality elements. In addition to developing the assessment criteria for TotP, assessment criteria for total nitrogen (TotN) were also developed to support management decisions and for comparisons made with other EU countries. Proposals for new class boundaries for ecological status were based on biological quality elements and using the concepts developed by ECOSTAT.The study comprised data for 210 watercourses and 146 lakes. The sites were located at subcatchment outlets to enable a comparison with source apportionment calculations from the Pollution Load Compilation program (PLC) and deemed to have sufficient data for model calibrations. Data from the national lake survey (n = 5080) were used for assessing performance of the lake models.The proposed approaches were developed using regression models calibrated with water chemistry and landscape parameters. The best predictors of TotP in watercourses were (ranked by explanatory power): water colour (absorbance), sulphate, sum of calcium and magnesium, average clay content in the catchment area, altitude, proportion of mire in the catchment area and proportion of surface water in the catchment area. Information on soil clay content was only available for watercourses in southern and central Sweden, therefore in order to be able to apply the model in northern regions clay content will need to be estimated. For lakes, the best regression model comprised mean depth, colour (absorbance), sulphate, magnesium and proportion of mire in the catchment area. As the clay content did not contribute significantly, these models can be used for all regions. Alternative models were developed for cases where some of the included measurement variables are not available; however, caution is advised as these models have larger errors compared to the proposed model.Corresponding models were developed for estimating reference values for TotN. For watercourses, the model comprised water colour (absorbance), calcium, northsouth coordinate, proportion of water and proportion of forest in the catchment area, altitude, east-west coordinate and catchment area (ranked according to how much they contribute to the model). The corresponding model for lakes comprised water colour (absorbance) calcium, mean depth, altitude and east-west coordinate.Since the water colour (absorbance) was included in models for both TotN and TotN, and as long-term trends of water colour and nutrients in waterbodies unaffected by eutrophication have been shown, the models will need to be continuously updated with the latest measurement data. This also makes it possible to base the assessments on the latest version of the PLC calculations.Class boundaries for total phosphorus were developed using a log-linear relationship between TotP and chlorophyll. In the assessment criteria for phytoplankton class boundaries differ among lake types. These differences may partly be due to random differences in selection of the sites or, for lakes in southern Sweden, that the proportion of agricultural land in the reference data base is greater than in other regions. Because the classification for TotP is site-specific, we propose to use the same class boundaries for all waterbody types. The mean value for the boundary between good and moderate status was close to the current limit, 0.5. Therefore, we propose to retain the current class boundaries for status classification for total phosphorus.Applying a corresponding methodology to watercourses by relating TotP to the IPS diatom index revealed weak correlations, providing no support for setting class boundaries. The weak association is probably because IPS was calibrated as a broader clean-water index and not solely for detecting nutrient impacts. We therefore propose using the same class boundaries for assessing impacts of TotP in watercourses and lakes and recommend that efforts be made to develop a nutrientbased benthic diatom index.For TotN, as studies have shown no significant relationships between TotN and biological quality elements in the P-limited lakes and watercourses in Sweden, the same class boundaries are proposed as for TotP.For watercourses, the proposed approach resulted in stricter assessments compared with the current regulations, with a larger proportion of sites classified as moderate or worse status. For lakes, on the other hand, a smaller proportion of the lakes had a moderate status or worse when the proportion of agricultural land was greater than 10%, while for lakes with less agricultural impact there was no difference between the two approaches. For watercourses, classification using TotN gave a smaller proportion of sites with moderate status or worse compared with TotP, while for the lakes, differences were trivial.In developing the proposed assessment criteria for nutrients, calculations were made solely using measured water chemistry and geographical data and therefore this approach is not directly dependent on model data for a status classification. This method is more aligned with the Water Framework Directive's requirements on how a status classification should be done; compared to the current approach where modelled leakage from unfertilized grassland is included in the assessment of agriculturally dominated watercourses. At the same time, the regression models were calibrated against the process-based modelling of background concentrations performed within PLC. This means that classifications of ecological status for nutrients are harmonised with Sweden's reporting to HELCOM and OSPAR. For lakes, the proposed approach resulted in more accurate assessments of sites heavily impacted by agricultural land use. For watercourses, the proposed approach resulted in more accurate assessments of all sites. The new approaches account for that agriculture soils are naturally nutrient rich and prone to erosion compared to other soils. However, we invite a broader discussion of whether it is reasonable to use unmanaged soils as the reference condition for nutrient leakage when the landscape has been cultivated for centuries or more.