The Minnesota Pollution Control Agency (MPCA) has developed draft eutrophication criteria for rivers protective of Minnesota’s aquatic life use. A multiple lines of evidence approach was used to develop the criteria. Conceptual models provide an overview of the focus of our research and the linkages we sought to establish. Several studies and data collection efforts provide the body of information used to develop the criteria. These data sources included the River Nutrient study which consisted of a series of efforts to assess and document relationships and patterns among nutrients, water quality and biological communities. In addition, data from MPCA’s biological monitoring program and USEPA's data system, STOrage and RETrieval (STORET), were used to develop the draft criteria. These studies have demonstrated significant and predictable relationships among summer nutrients, sestonic chlorophyll-a, and biochemical oxygen demand (BOD5) in several medium to large Minnesota rivers (Heiskary & Markus 2001, 2003). In addition, diel dissolved oxygen (DO) flux (based on submersible data recorders) also was strongly positively correlated to total phosphorus (TP) and chlorophyll-a concentrations. Our findings demonstrate significant relationships among several sensitive macroinvertebrate and fish metrics and TP, TN, chlorophyll-a, and DO flux. Biological thresholds for nutrients and associated stressors were determined using quantile regression and changepoint analyses using macroinvertebrate and fish data. The major steps or approaches that were used to develop draft river eutrophication criteria are summarized below. · Regressions described basic interrelationships among TP, TKN, sestonic chlorophyll, and DO flux based on the River Nutrient study datasets. Most relationships exhibited high r2 values and were highly significant. · Spearman correlation analysis, using the River Nutrient study data, provided an initial basis for identifying relationships among TP, TN, chlorophyll, and DO flux and fish and macroinvertebrate metrics. This provided a basis for identifying responsive metrics for each of these variables and helped to focus subsequent analyses. · Scatterplots helped visualize relationships among the more responsive metrics and the stressors and begin threshold identification. Statewide interquartile ranges for the biological metrics were used to place metric values in perspective and help discern where an important shift in the metric may be occurring relative to the stressor gradient. · More advanced statistical techniques including quantile regression and changepoint (regression tree) analyses were employed. These analyses are well-suited to the often wedge-shaped plots that are common with field-collected biological data. Based on the Spearman correlation analysis emphasis was placed on the more responsive biological metrics. These techniques were applied to both the river nutrient dataset and the larger biomonitoring and STORET datasets. Threshold concentrations were produced for statewide, wadeable vs. nonwadeable, and on a region-specific basis. · Relationships among nutrients, stressor variables, and the biology was further assessed by determining the levels of chlorophyll-a and total phosphorus associated with the BOD5 threshold concentrations; · A comprehensive review of the literature was conducted and literature-based thresholds provided further perspective on this issue. · Consistent with EPA guidance, data and relationships were analyzed in a regional context. Threshold concentrations ranges were placed in context with ecoregion-based frequency distributions compiled by MPCA for representative, minimally-impacted streams (McCollor & Heiskary 1993), a more recent compilation of stream TP data from STORET (period from 1996-2012), and IQ ranges from USEPA criteria manuals (USEPA 2000b, a, 2001). These data distributions reflect distinct regional differences in stream TP, BOD5, and other variables. This work combined with previous analysis of Minnesota’s ecoregional patterns resulted in defining three “River Nutrient Regions (RNR)” for criteria development. · All of the above was used to move from broad ranges of criteria to region-specific criteria (Table 1). In addition to these ecoregion-based criteria, we have proposed a numeric translator to address the impact of nuisance levels of periphyton that can limit aquatic life and aquatic recreational uses of Minnesota streams. This numeric translator is as follows: “Rivers shall have an algal biomass not to exceed 150 mg Chl a m-2 to avoid nuisance algal biomasses that interfere with important aquatic recreation designated uses.” This level is well supported in the literature (e.g., Welch et al. 1988, Dodds et al. 1997, Dodds & Welch 2000, Suplee et al. 2008b) and provides a good basis for defining impairment from excess periphyton. The river eutrophication criteria will be applied in a fashion similar to the previously promulgated lake eutrophication criteria. Prior to 303(d) assessment of a stream reach, water quality samples will be collected 6-8 times per summer for a minimum of two summers. These data will be combined with all available data for the most recent 10-year assessment period. Means will be calculated and compared to the criteria. For a stream reach to be listed as impaired it must exceed the causative variable: TP and one or more of the response (stressor) variables: sestonic chlorophyll, BOD5, DO flux, and/or pH. Stream reaches listed as impaired will be subject to development of a TMDL. The TMDL considers all upstream sources that contribute to the excess nutrients and the impairment. USEPA recommends that downstream protection is considered when developing nutrient criteria. This means that criteria need to be protective of both the assessed water (streams in this case), as well as downstream waters. In the case of river criteria, the downstream waters of concern would typically be lakes, reservoirs, or mainstem pools on major rivers. Based on a long history of lake restoration and watershed projects the proposed stream TP criteria are in the range of stream inflow values proposed as a part of restoration projects. One basis for this argument is comparing the stream criteria to the stream TP values used in the MINLEAP model. The MINLEAP model (Wilson & Walker 1989) has long been used as a basis for predicting in-lake TP for minimally-impacted lakes on an ecoregion basis. The model was regionally-calibrated and has long been used to help define in-lake goals for lake and watershed restoration projects. The corresponding regionally-calibrated NLF and NCHF ecoregion stream TP values used in the model are 52 μg/L and 148 μg/L, which are either equal to or higher than the proposed criteria for the North and Central RNRs (50 and 100 μg/L respectively). These stream values were deemed typical of representative, minimally-impacted watersheds for the two regions. This comparison suggests that the North and Central stream TP criteria will likely be protective of downstream resources. For perspective, about 50% of Northern RNR streams have TP <50 μg/L and 35% of Central RNR <100 μg/L. A similar MINLEAP-based comparison for the South RNR could not be made as the steam inflow TP used in the model was highly calibrated to account for extreme storm event loading and internal recycling within the lakes. However, the South RNR proposed criteria of 150 μg/L ranks near the 25th percentile for South RNR stream sites and should prove to be protective of downstream lakes and reservoirs in most instances. Ultimately, lake and reservoir nutrient TMDLs will determine the appropriate stream inflow TP required to meet water quality standards. The detailed data analysis and development of allocations assures this and would take precedent over existing stream eutrophication standards. We have demonstrated that the proposed river eutrophication criteria are protective of downstream waters in the case of Lake Pepin on the Mississippi River, where modeling has demonstrated meeting the river criteria should allow for attainment of Lake Pepin site specific criteria. The eutrophication criteria can also serve as a basis for protecting stream reaches that are currently better than the criteria via the implementation of nondegradation. As with other water quality standards there is an expectation that these are not “degrade down to” standards; rather waters that are currently meeting standards would be expected to continue to do so. The combination of the eutrophication standards and nondegradation language should assure that this is the case. These criteria represent a first step in a larger process. As Tiered Aquatic Life Use (TALU) standards are developed in future rulemakings there will be refinements to these criteria that reflect the more specific needs of the various tiered uses. One example is coldwater streams that will be addressed more specifically. However, in the interim, the region-based eutrophication criteria provide a basis for assessing the condition of Minnesota streams relative to excess nutrients. In turn, this allows for the development of strategies and policies to protect the condition of streams and to minimize and reverse the impact of excess nutrients on stream ecosystems.