The Souris River Basin (hereafter referred to as the “basin”) is an international basin in southeast Saskatchewan, north-central North Dakota, and southwest Manitoba (fig. 1). Maintaining good water quality in the basin is important for human health and ecological resources. Several communities in Saskatchewan, North Dakota, and Manitoba rely on the Souris River for all or part of their water supply, and the Souris River is an important source of water for waterfowl, fish, and other aquatic organisms. To monitor and maintain water quality in the Souris River, in 1991 the International Souris River Board (ISRB) adopted water-quality objectives (WQOs) for more than 40 constituents at two transboundary crossings or binational sites: Saskatchewan–North Dakota (Souris River near Sherwood, N. Dak., site 7 [U.S. Geological Survey (USGS) station 05114000; referred to hereafter as “Sherwood”]) and Manitoba–North Dakota (Souris River near Westhope, N. Dak., site 18 [USGS station 05124000, referred to hereafter as “Westhope”]; International Joint Commission, 2017; fig. 1). WQOs are a numerical concentration or narrative statement that is intended to support designated uses of water at a specific site (International Joint Commission, 2017). An International Joint Commission (IJC) review of WQOs in the Souris, Red, Rainy-Lake, and St. Croix River Basins recommended, “in instances where sustained exceedances of WQOs are observed, boards are encouraged to investigate the factors responsible for the exceedances, which may be the result of anthropogenic activities or natural system processes. If appropriate, boards should develop advice regarding potential mitigation and restoration solutions” (International Joint Commission, 2017, p. 5). Exceedances of a WQO are determined by comparing measured concentrations from samples collected at the binational sites against the WQO and are reported as the number and percentage of samples in a calendar year that have concentrations higher than the WQO. Sustained exceedances of WQOs for total phosphorus, sodium, sulfate, total dissolved solids (TDS), and total iron have been reported since the late 1990s at the two binational sites on the Souris River (International Joint Commission, 2017).

A basin-wide approach to trend analysis provides a broader understanding of how water quality has changed through time (temporal changes) and where the greatest changes are in the basin (spatial changes). Trend analysis of selected water-quality constituents in the Souris River was previously completed in 2000 and 2012 (Vecchia, 2000; Galloway and others, 2012) and was focused on the two binational sites on the Souris River, Sherwood and Westhope. Additional data have been collected since the previous studies, and to understand conditions at the binational sites, it is important to understand water-quality changes on a basin-wide scale. Also, because streamflow is highly variable in the basin and changes in streamflow affect water-quality conditions, it is particularly important to use a trend-analysis method that accounts for changes in streamflow.

Trends in water-quality concentrations can be affected by human-induced changes on the landscape or natural changes in land-runoff interactions that are driven by climate patterns and reflected by changes in streamflow (commonly referred to as “hydroclimatic variability”). In the primarily agricultural Souris River Basin, human-induced changes that are likely to affect trends are widespread changes in agricultural management such as fertilizer application, tilling practices, and crop types, as well as dam emplacement and artificial drainage. The Souris River is regulated for water supply and flood control by four reservoirs upstream from Minot, N. Dak. (International Souris River Study Board, 2021a). Dams built to create Boundary Reservoir and Lake Darling were completed prior to 1960, and dams built to create Rafferty Reservoir and Grant Devine Lake were completed in the early 1990s (fig. 1). Climate is highly variable in the basin, varying from year to year and decade to decade, and the climate fluctuates between wet and dry climate states (Kolars and others, 2016; Ryberg and others, 2016; International Souris River Study Board, 2021b). The shift from a dry climate state to a wet climate state is a larger regional phenomenon encompassing the Souris River Basin and other areas of Saskatchewan, North Dakota, and Manitoba (Kolars and others, 2016). In the Souris River Basin, a significant transition from a dry climate state to a wet climate state around 1970 was identified from statistical analysis of tree-ring chronologies and historical precipitation data in the basin (Kolars and others, 2016; Ryberg and others, 2016). The shift to the wet climate state has resulted in higher streamflow in the basin and, in 2011 historically unprecedented flooding was experienced in Minot, N. Dak. (Kolars and others, 2016). To assist the ISRB in assessing current water-quality conditions in the Souris River Basin and exceedances of WQOs at the binational sites, the USGS, in cooperation with the IJC, completed a comprehensive water-quality trend analysis in the Souris River Basin. The basin-wide approach includes descriptive water-quality statistics; water-quality trend analysis using a trend method that considers interannual hydroclimatic variability for selected ions, nutrients, and trace metals for many sites in the basin; and an assessment of exceedances of the WQOs for the Sherwood and Westhope sites.

Water-quality and streamflow or reservoir inflow or outflow data were compiled for 34 sites and 23 constituents from 1970 to 2020 in the Souris River Basin and used for descriptive statistics and water-quality trend analysis. Water-quality data for sites in the United States were compiled from the National Water Quality Monitoring Council Water Quality Portal (WQP; National Water Quality Monitoring Council, 2021) and streamflow data for U.S. stream sites were obtained from the USGS National Water Information System database (U.S. Geological Survey, 2021). For trend analysis of reservoirs, reservoir inflows or outflows were used as surrogates for streamflow. Water-quality data for Canadian sites were provided by Water Security Agency Saskatchewan (WSA), Manitoba Agriculture and Resource Development (MARD), and Environment and Climate Change Canada (ECCC). Streamflow data for Canadian sites were provided by ECCC or MARD. Twenty-three constituents with an established WQO were selected for analysis, and data generated or analyzed during this study are available as a USGS data release (Tatge and Nustad, 2023). At the binational sites, paired samples were evaluated for differences in the center of the data prior to combining data collected by different agencies.

Water-quality data from 1970 through 2020 were compiled for 34 sites in the Souris River Basin (table 1). Descriptive statistics for each constituent listed in table 2 are provided in appendix 1 for total dissolved solids and ions, nutrients, trace metals, and other measurements (tables 1.1, 1.2, 1.3, and 1.4, Tatge and Nustad, 2023). To visualize spatial variability across the basin, median concentrations for all sites for five constituents (TDS, sulfate, sodium, total phosphorus, and total iron) are discussed below and shown on figures 2–4.

Median TDS concentrations were low in the headwaters of the Souris River (Souris River near Bechard, Saskatchewan [WSA station SK05NB0574; hereafter referred to as “site 1”]) and three of the four sites (Souris River at Highway 39 near Roche Percee, Saskatchewan [WSA station SK05NB0198; hereafter referred to as “site 4”]; site 10; Pipestone Creek near Moosomin (PSC-152) [WSA station SK05NE0091; hereafter referred to as “site 29”]) with the highest median concentrations were measured in the upper basin (fig. 2, table 1.1; Tatge and Nustad, 2023). TDS concentration is a measure of the sum of major dissolved ions such as calcium, magnesium, sodium, potassium, sulfate, chloride, bicarbonate and carbonate, and many other constituents present in small amounts (Vecchia, 2005). For most sites, sulfate and sodium constitute about one-half of the TDS concentration (table 1.1). For example, the median sulfate concentration of 256 mg/L and median sodium concentration of 119 mg/L for Sherwood accounts for 49 percent of the median TDS concentration of 759 mg/L (table 1.1). Based on median concentrations, the smallest percentage of TDS from sulfate and sodium was about 30 percent for Lightning Creek near Carnduff (WSA station SK05NF0124; hereafter referred to as “site 21”), whereas the largest percentage of TDS from sulfate and sodium was about 64 percent for site 3. Median TDS concentrations in the Souris River Basin ranged from 337 mg/L at Gainsborough Creek at Provincial Trunk Highway 83 (MARD station MB05NFS019, hereafter referred to as “site 23”) to 1,170 mg/L at site 4 (table 1.1, fig. 2). The median concentration at site 4 is less representative than some of the other main-stem sites because, although the period of record is between 1974 and 2021, most of the 77 observations were collected between 2005 and 2021 (table 1.1; Tatge and Nustad, 2023). Three of the four sites with median concentrations greater than the binational TDS WQO of 1,000 mg/L were on the following tributaries: Des Lacs River (site 10), Moose Mountain Creek (site 5), and Pipestone Creek (site 29). Des Lacs River (site 10) was the tributary with the highest median TDS concentration, and main-stem Pipestone Creek sites (sites 28–30) all had median concentrations greater than 850 mg/L. For main-stem sites, concentrations were more variable in the upper basin, with median concentrations ranging from 372 at site 1 to 1,170 mg/L at site 4, and less variable in the lower basin, with most sites between 650 and 850 mg/L (table 1.1).

Like TDS, median sulfate concentrations were low in the headwaters of the Souris River (site 1), and two of the same sites (sites 4 and 10) in the upper basin had the highest median concentrations (fig. 2, table 1.1). Median sulfate concentrations across the basin ranged from 73 mg/L at site 1 to 494 mg/L at site 10. The three highest median sulfate concentrations were in the upper basin (sites 3, 4, and 10) and all were over 400 mg/L, and sites 4 and 10 were greater than or equal to the 450-mg/L sulfate WQO. Des Lacs River (site 10) was the tributary with the highest median sulfate concentration and Pipestone Creek sites (sites 28–31) all had median concentrations greater than 319 mg/L (table 1.1 and fig. 2). At main-stem Souris River sites, median sulfate concentrations generally were between 200 to 400 mg/L; however, there was more variability in the upper basin than the lower basin (table 1.1 and fig. 2).

Like TDS and sulfate, sodium concentrations were low in the headwaters of the Souris River (site 1), and two of the same sites (sites 4 and 10) in the upper basin had the highest median sodium concentrations (fig. 2, table 1.1). Unlike TDS and sulfate, sites on Pipestone Creek were not some of the highest median sodium concentrations in the basin. Median sodium concentrations across the basin ranged from 28 mg/L at the Antler River near Wauchope (WSA station SK05NF0125) to 226 mg/L at site 4. The two largest median sodium concentrations (sites 4 and 10) were in the upper basin and were greater than 200 mg/L (fig. 2). At main-stem Souris River sites, other than sites 1 and 9, median sodium concentrations were greater than the 100-mg/L WQO, with concentrations generally between 110 and 140 mg/L. In general, median sodium concentrations were more variable in the upper basin than the lower basin.

Median total phosphorus concentrations in the Souris River Basin were highest in the headwaters of the Souris River (site 1), and all sites had median concentrations greater than the 0.1 mg/L WQO (fig. 3). Across the basin, median total phosphorus concentrations ranged from 0.11 mg/L as phosphorus at Pipestone Creek near Whitewood (PSC-71) [WSA station SK05NF0125] and Plum Creek at Provincial Road 254 (MARD station MB05NGS085) to 0.52 mg/L as phosphorus at site 1 (fig. 3; table 1.2; Tatge and Nustad, 2023). Among the main-stem Souris River sites, other than site 1, median concentrations ranged from 0.18 mg/L at Sherwood to 0.31 mg/L at Souris River at Melita – Highway 3 (MARD station MB05NFS024, hereafter referred to as “site 25”). Many tributary sites had lower concentrations than the main-stem sites.

Median total iron concentrations were highly variable across the basin, ranging from 61.0 µg/L at Lake Darling to 1,420 µg/L at site 14 (fig. 4; table 1.3), and for main-stem sites median concentrations were greater than or equal to the 300-µg/L WQO. Because total iron is attached to particulate matter and is largely transported from the landscape into the stream or reservoir through surface runoff, the measurement of total iron concentration is more variable than many other constituents, and local geology and soil composition can affect concentrations. For example, site 17 had a range of concentration between 10.3 and 82,700 µg/L (table 1.3) and the median total iron concentration of 759 µg/L at site 12 was more than double the median total iron concentration at the site just upstream (site 11). Between sites 11 and 12, the geologic formation underlying the Souris River changes from Cannonball to Hell Creek and then to the Fox Hills Formation (not shown; North Dakota State Geospatial Committee and North Dakota Information Technology, 2022). Weathering of iron oxide concretions and nodules in the Hell Creek Formation (Biek, 2002) may contribute to higher median concentration of total iron at site 12.

Water-quality trends were analyzed for stream sites for a recent period (2009–19) and reservoir sites were analyzed for a common period of available data (2000–15). Stream sites with enough data were analyzed for a historical period (1976–2019).

Two measures of exceedance probability were evaluated for the binational sites: annual mean flow-averaged EP and flow-averaged EP. The annual mean flow-averaged EP is expressed as a probability value between 0 and 1 but can be interpreted as the proportion of time during the year concentrations are expected to exceed the concentration threshold, assuming average flow conditions. For example, if the annual mean flow-averaged EP for a given year is 0.25 (or one-fourth of the year), it is expected that the WQO would be exceeded about 25 percent of the time during that year (about 90 days), assuming normal flow conditions. The flow-averaged EP is computed for each 5-day time interval in the period of record and is interpreted as the chance of exceeding the concentration threshold during that time interval, assuming flow conditions were the same year after year. The annual mean flow-averaged EP and the flow-averaged EP during 1976–2019 are presented for TDS, sulfate, sodium, total phosphorus, and total iron. To compare how the seasonal patterns in exceedances have changed through time, the flow-averaged EP for 3 years (1988, 2005, 2019) are presented.

For Sherwood, the probability of exceeding the WQO for TDS, sulfate, and sodium increased between 1976 and 2019, especially for sulfate, which more than doubled from 0.13 or 13 percent of the year (or 47 days) to 0.34 or 34 percent of the year (or 124 days; fig. 20). The trend in annual mean flow-averaged EP for sulfate increased from 0.13 (13 percent of a year or 47 days) in 1976 (fig. 20B), 0.34 in 2019 (34 percent of the year or 124 days). The range in flow-averaged EP for TDS (approximately between 0 and 1, fig. 20A) and sodium (approximately between 0.25 and 1, fig. 20C) was larger than the range in sulfate (generally between 0 and 0.5, fig. 20B), which indicates more seasonal variability in TDS and sodium than sulfate (fig. 20).

Seasonal patterns of probability exceedances were noticeable in the flow-averaged EP for TDS, sulfate, and sodium (fig. 21). For TDS and sodium for all years, the highest probability of exceeding the WQO was in December and January when there was no surface runoff and groundwater inputs are likely to be a greater portion of the streamflow. Concentrations were least likely to exceed the WQO in late April and early May during runoff from the spring snowmelt, often when soils are mostly still frozen (fig. 21A and 21C). The pattern in the probability of exceedance for sulfate was less pronounced, with a more consistent probability of exceedance from July to October, a more gradual increase in November, and a more gradual decrease in February and March (fig. 21B). The more gradual decrease in exceedance probability for sulfate in February and March may be indicative of an additional source of sulfate or that there is a different process mobilizing sulfates. During June through October of 2005 and 2019, the pattern of probability of exceeding the WQO for all three constituents are the same suggesting that the same process is mobilizing the constituents. During June through October, evaporation is generally higher and given that sodium-sulfate evaporites are known to be present, it is likely that the dissolution of sodium-sulfate evaporites are mobilizing sulfate and sodium, and in turn TDS. In the winter months before spring runoff, TDS is likely being affected by other constituents present in groundwater inputs. By 2019, the seasonal pattern was much more pronounced for all three constituents, with higher probability of exceedance or fewer days of concentrations likely not to exceed the WQO during each season.

For Westhope, the probability of exceeding the WQO for TDS, sulfate, and sodium increased between 1976 and 2019, specifically, the probability of sulfate exceeding the WQO increased about seven times from 0.08 or 8 percent of the year to 0.58 or 58 percent of the year (fig. 22). Because sodium concentration is near the WQO most of the year, sodium had a different seasonal pattern than sulfate (fig. 23B, C). Between August and January, the flow-averaged EP did not vary because the concentrations were at or above the WQO for sodium (fig. 23C). Flow-averaged EP for sulfate follow a similar pattern to Sherwood and are likely being affected by the same processes (figs. 21B and 23B). For Westhope, between 1976 and 2019, the trend in annual mean flow-averaged EP for TDS increased from 0.22 to 0.5, sulfate increased from 0.08 to 0.58 percent, and sodium increased from 0.64 to 0.91 percent (fig. 22). For TDS and sodium, the increase in flow-averaged EP between 1976 and 2019 for Westhope was of similar magnitude to Sherwood, but instead of doubling like Sherwood, flow-averaged EP for sulfate increased about seven times from 0.08 to 0.58 (from 8 to 58 percent of a year or 29 to 212 days) at Westhope (fig. 22B). The range in flow-averaged EP for Westhope was largest for TDS, approximately between 0 and 0.75 in the early years and between 0.10 and 1 in later years (fig. 22A). For Westhope, a pattern in the probability of exceedances for TDS and sulfate were very similar, with the highest probability of exceedance in December and January and the lowest probability in late May and early June (fig. 23A and 23B). Probability exceedances increased slightly for TDS and sulfate in the summer months but sharply increased in October and November. For sodium, because the flow-averaged GMC exceeds the WQO from July through January, a seasonal pattern for the whole year is not evident (figs. 16B and 23C). For 2005 and 2019, the lowest probability of the flow-averaged GMC exceeding the sodium WQO was between late April and mid-May, which is about a month earlier than sulfate and TDS for the same years (fig. 23C). Also, the timing of the lowest probability shifted among years for sodium, with the lowest probability for 1988 a month later than for 2005 and 2019 (fig. 23C). The difference in the timing of the lowest exceedance probability for sodium and sulfate for 2005 and 2019 may be an indication of a different sulfate source or process other than dissolution of sodium-sulfate evaporites. Additionally, the shift in the lowest probability of exceedance for sodium from 1988 to 2019 may be another indication of different sodium sources driving exceedances.

For Sherwood, total phosphorus and total iron were highly likely to exceed the WQO and seasonal patterns of total phosphorus and total iron concentrations were generally similar with concentrations least likely to exceed the WQO between November and January (figs. 24 and 25). The annual mean flow-averaged EP for Sherwood for total phosphorus increased slightly from 0.75 (274 days) in 1976 to 0.85 (310 days) in 2019, but during the entire period, concentrations were likely to exceed the WQO most of the year (fig. 24A). For total iron, the annual mean flow-averaged EP decreased slightly from 0.76 (or 277 days of the year) in 1999, increased to 0.86 (314 days) in 2009, and decreased to 0.52 (212 days) in 2019 (fig. 24B). From the seasonal pattern of probability exceedances for Sherwood, total phosphorus concentrations were highly likely (probability near 1.0 for all years) to exceed the WQO during the summer months (June to September) and much less likely to exceed from November to January (fig. 25A). Total iron concentrations were least likely to exceed the WQO between November and January and highly likely (probability greater than about 0.7 for all years) to exceed the WQO during the summer months (June to September; fig. 25B). Higher probability of exceedances during the summer months may be related to surface runoff. Although the magnitude of the probabilities changed with time, seasonal patterns were nearly identical for all years.

Total phosphorus and total iron concentrations for Westhope were highly likely to exceed the WQO for most of the year (figs. 26 and 27). The annual mean flow-averaged EP for Westhope for total phosphorus decreased slightly from 0.95 (347 days) in 1976 to 0.92 (336 days) in 2019, but over the entire period, the probability of exceeding was close to 1.0 meaning concentrations were likely to exceed the WQO most of the year (fig. 26A). For total iron, there was little change in the annual mean flow-averaged EP between 1999 and 2019, with 0.66 (241 days) in 1999 and 0.67 (244 days) in 2019 (fig. 26B). From the seasonal pattern of exceedances for Westhope, the probability of exceeding the WQO for total phosphorus concentrations was near 1.0 during the summer months (July–September) and only decreased during October and November (fig. 27A). Total iron concentrations were most likely to exceed the WQO in February and March and least likely to exceed in June–July (fig. 27B). The difference between total phosphorus and total iron in the seasonal pattern of exceedances may be related to seasonal changes in the J. Clark Salyer National Wildlife Refuge Pools upstream which may be affecting total phosphorus and total iron concentrations differently. Although the magnitude of the probabilities changed over time, seasonal patterns were nearly identical for all years. Comparing Sherwood and Westhope, the seasonal patterns of exceedances for total phosphorus and total iron are different suggesting that different factors affect the concentrations at these two sites (figs. 25 and 27).

The probability of exceedances for ISRB WQOs, along with additional water-quality standards and objectives (for purposes of this discussion these will be grouped and referred to as “concentration thresholds”), is intended to provide a context for considering additional concentration thresholds during the historical period. Concentration thresholds listed in table 5 are generally designed to be protective of uses external to aquatic life and human health (for example, irrigation or drinking water). Depending on the designated water use, the concentration threshold is different (for example, more restrictive concentration thresholds are typically set for drinking water than irrigation). Annual flow-averaged EP is the measure of probability of exceedances used in this discussion for comparing against the concentration thresholds, but seasonal patterns may also be important for aquatic life. For TDS, the ISRB WQO is set at 1,000 mg/L, which is higher than concentration thresholds set by other jurisdictions. Compared with the TDS WQO, annual flow-averaged EPs for Sherwood and Westhope generally have ranged from about 0.2 to 0.5 during the entire period (fig. 28). If the most restrictive concentration threshold of 500 mg/L was used, TDS concentrations would be likely to exceed the concentration threshold for almost the entire period, and if 750 mg/L (approximately the median concentration for Sherwood and Westhope) was used, TDS concentrations would be likely to exceed the concentration threshold more than 50 percent of the year (or 0.5) during the entire period. For sulfate, the ISRB WQO of 450 mg/L is in the middle of the range of other concentration thresholds (table 5) and the probability of exceeding the WQO was about 0.05 to 0.6 (fig. 29). If the 500-mg/L concentration threshold was used, the annual flow-averaged EP for both sites during the entire period would be 0.5 or less. If a more restrictive concentration threshold of 250 mg/L was used, the probability of exceeding the sulfate concentration threshold would be about 0.3 to 0.5 until about 1998, after which the probability would increase to about 0.5 to 0.9. In contrast, if the sulfate ISRB WQO was more than doubled to a less restrictive 1,000-mg/L concentration threshold, sulfate concentrations generally would not have exceeded the concentration threshold (probability of less than about 0.05 at the end of the period) during the entire period. For sodium, the annual flow-averaged EP of the ISRB WQO of 100 mg/L for Sherwood and Westhope was more than 0.7 percent during the entire period (fig. 30). If the concentration threshold was 200 mg/L, the annual flow-averaged EPs for both sites would have been between about 0.25 and 0.30 during the entire period. For total phosphorus, the annual flow-averaged EP of the ISRB WQO of 0.1 mg/L for Sherwood and Westhope was more than 0.75 during the entire period (fig. 31). Using the more restrictive concentration threshold of 0.05 mg/L, total phosphorus concentrations would have exceeded the concentration threshold nearly 100 percent of the time (probability near 1.0) at both sites, and using the less restrictive concentration threshold of 0.15 mg/L, the annual flow-averaged EP for Sherwood would have exceeded the threshold between 50 and 70 percent of the time during the entire period and Westhope would have been between about 75 and 85 percent of the time during the entire period. For total iron, the annual flow-averaged EP of the ISBR WQO of 300 µg/L for Sherwood and Westhope was between about 0.50 and 0.75 during the entire period and using the less restrictive concentration threshold of 550 µg/L (median concentration for Sherwood), the probability of exceeding the total iron concentration threshold decreases to about 0.25 and 0.5 during the entire period (fig. 32). For the least restrictive concentration threshold of 800 µg/L (75th percentile concentration for Westhope), the probability of exceeding the total iron concentration threshold decreases even more to 0.15 to 0.3 during the period.

At the binational sites for the 43-year period of analysis, the annual flow-averaged GMC of sodium, total phosphorus and total iron was likely to exceed the WQO most of the time (generally more than 70 percent), whereas the annual flow-averaged GMC of TDS and sulfate was likely to exceed the WQO about one-half the time (20–50 percent for TDS and approximately 5–60 percent for sulfate). For sodium, total phosphorus, and total iron, exceedance of the WQO most of the time is not unexpected given that the flow-averaged GMC for these three constituents for most sites across the basin were greater than the WQO for most of the period (figs. 16, 18, 19, 20C, 22C, 24, 26). For TDS and sulfate, from 1976 to 1988, when streamflow was less likely to have been affected by the wet climate state, flow-averaged GMCs, which consider variability in concentration related to streamflow, were less than the WQOs and the annual mean flow-averaged EP was less than later years (figs. 14, 15, 22A, 22B). During 1988 to 2019, concentrations increased and rapidly approached or were likely to exceed the WQO. Concentrations vary by season and year, but given the abundance of some naturally occurring constituents (for example, sulfate) combined with long-term persistence, it is likely that concentrations also vary by decade or possibly longer. Thus, depending on the constituent and threshold, during the long-term it is likely that there will be extended periods when concentrations exceed the threshold for large portions of the year. If natural processes are affecting TDS and sulfate concentrations, concentrations would be expected to vary with time, and as a result, extended periods of concentrations greater or less than the WQO are likely depending upon climatic conditions.

A better understanding of the state of water quality across the Souris River Basin is beneficial to understanding and interpreting water-quality conditions at the two Souris River binational sites. Multidecadal monitoring of water quality and streamflow by multiple agencies at many sites across the basin was essential to this report. Although changes in water-quality conditions at the binational sites are the primary focus of international agreements related to the Souris River, water quality at the binational sites is affected by upstream water quality, and evaluating trends for other main-stem sites, tributary sites, and reservoir sites puts the binational sites into context of the rest of the basin. Although a better understanding of spatial and temporal changes in water quality in the basin was gained from this study, gaps in understanding of water-quality conditions were also identified.

Less information on changes in water-quality conditions was available for the headwaters of the Souris River, Moose Mountain Creek, Long Creek, and Pipestone Creek because sites were sparser in these areas and minimum data requirements for the trend method combined with data availability excluded many sites from trend analysis. Minimum data requirements of R–QWTREND are intended to ensure that observations are spread out among multiple years and among seasons within each year to obtain a reliable and representative trend model (Vecchia and Nustad, 2020). For some of the headwater sites and smaller tributaries sites in Manitoba and Saskatchewan, data availability did not match R–QWTREND data requirements for different reasons. For some sites, streamflow dries up during winter months, so samples are not collected. For several of the Saskatchewan sites, year-round streamflow and water-quality samples are collected, but within the periods analyzed for this study, 8 or 9 years of data were available. For some sites, a sample was available during winter months, but daily streamflow was not measured. A small amount of missing streamflow data can be filled in using R–QWTREND, but sites consistently missing streamflow during the same period each year, in this case winter months, cannot be estimated. In some instances, a streamgage was not co-located with the water-quality site. These sites were analyzed, but the trend model could be improved if the streamgage was co-located with the water-quality site because of less uncertainty in the flow. A more complete understanding of changes in the headwater sites and smaller tributaries sites in Manitoba and Saskatchewan may be achieved in the future by using a different trend method for sites without winter samples and streamflow, and by using R–QWTREND for sites with year-round streamflow and water-quality data when 10 years of data have been collected.

Fewer sites and a shorter period were available for comparison of trace-metal trends because of changes in laboratory-analytical methods and sample portion analyzed, and inherent variability in total iron likely affected the ability to detect significant trends. As a result of laboratory-analytical method changes in trace metals, only changes from 1999 onward were available. The sample portion (filtered or unfiltered) analyzed for trace metals varied depending on site. Most of the stream sites were analyzed for total (unfiltered) trace metals, except for several U.S. tributary sites (sites 3, 10, 13, 15, and 16). Canadian reservoirs (sites 2 and 6) were analyzed for dissolved (filtered) trace metals, whereas U.S. reservoirs (sites 8 and17) were analyzed for total trace metals. The unfiltered result for trace metals was selected for analysis in this study because the WQO is set for total iron at the binational sites. Given the mix of total and dissolved trace-metal data across the basin, the number of sites available for comparison of trace-metal trends was less than for other groups of constituents. For total iron, of the seven stream sites and two reservoir sites analyzed, all trends were nonsignificant except for one. Iron is associated with sediment and can result in high variability, which can make significant detection of trends more challenging. Reasons for collecting total trace metals instead of dissolved trace metals vary by site and agency, but collection of dissolved iron and subsequent trend analysis may result in an improved ability to detect significant trends because there is less variability in dissolved iron.

Although an abundance of data were available for the binational sites, differences in field-collection method, laboratory-analytical method, and field-collection location required careful consideration of comparability of constituent concentrations prior to trend analysis. To be confident that observed trends represented real environmental changes, it is imperative to address data comparability for each constituent. Constituents most affected by differences in field-collection method, laboratory-analytical method, and field-collection location tended to be those associated with particulate matter in the water. For example, significant differences were detected for total phosphorus for all binational paired datasets tested for this report. Although there were not enough data to test for field-collection location differences between Coulter, Manitoba, and Westhope, N. Dak., based on statistical testing of the other paired datasets, it is likely that constituents associated with particulate matter (total phosphorus, total suspended solids, total iron) are different between the two field-collection locations. Collecting water-quality samples at Westhope as the primary sampling location will provide the most consistent dataset moving forward, and collection of more paired samples from Westhope, N. Dak., and Coulter, Manitoba, would provide more information to test for differences owing to field-collection location. Paired samples between USGS and ECCC were valuable in evaluating comparability of data at the binational sites, but continuation of paired sampling is important to monitor any changes in the differences determined in this report.

The most consistent spatial and temporal change observed for this report was large and consistent increases in sulfate and TDS among tributary and main-stem sites since the late 1980s. Wetter climatic conditions that have persisted since about 1970, combined with naturally abundant sulfate in the soils, geology, and aquifers, has likely resulted in substantial changes in the hydrology and geochemistry of the basin and, in turn, increased the amount of sulfate in soils near the surface and the mobility of sulfate in shallow groundwater inflow to streams. Different basins undergoing the same natural (climatic variability) or human-induced changes (for example, population growth, agricultural activities, and artificial drainage) may have different changes in water quality because of differences in primary hydrologic flow paths and underlying physiography, geology, and soils (Capel and others, 2018). Two nearby basins, the Red River of the North and the Heart River (not shown), are experiencing similar natural and human-induced changes, and sulfate concentrations are also increasing in those basins, especially in more recent years. There are some differences in hydrologic flow paths, physiography, and soils among these basins, but there are also similarities, and the similarities are likely the reason for the consistent response. In the Heart River Basin, geochemical modeling indicated that sulfate evaporite minerals had the largest control on concentrations of dissolved ions in the Heart River (Tatge and others, 2022). Sulfate evaporite minerals are abundant in the Souris River Basin and based on consistent increasing sodium and sulfate trends for many sites from 1988 to 2005, it is likely that the dissolution of sulfate evaporite minerals are contributing to increasing concentrations of dissolved ions in the Souris River. In the Red River of the North Basin, the geology and soils are different in the Minnesota portion of the basin, but it is unknown if sulfate concentrations are increasing or decreasing because too few sulfate data were available for trend analysis (Nustad and Vecchia, 2020). In the North Dakota portion of the Red River of the North Basin, soils are generally similar to the Souris River Basin and sulfate concentrations have been increasing since the mid-1980s. In these three basins, although there are some differences in hydrologic flow paths, physiography, and soils, wetter climatic conditions have persisted in all basins for the last several decades. Human-induced changes may enhance sulfate increases, but consistently increasing sulfate concentrations among these three basins indicate that the increasing sulfate concentrations are likely caused by wetter climatic conditions.

From the exceedance probability analysis, the 43-year period of record provides long-term perspective on consistent exceedances of the WQOs in recent decades for TDS, sulfate, sodium, total phosphorus, and total iron. In approximately the first decade of analysis, sulfate and TDS were less than the WQO more than 80 percent of the time, but around the time of the onset of wetter climatic conditions, the annual flow-averaged EP started increasing and continued increasing until sulfate and TDS were expected to be more than the WQO about one-half of the year by 2019. In contrast, sodium, total phosphorus, and total iron concentrations have been consistently greater than the WQO, approximately 70 percent of the time, during the 43-year period. The 43-year period represents a period of natural changes, such as long periods of dry and wet climatic conditions, which resulted in a wide range of hydrologic conditions and human-induced changes, such as changing agricultural practices and increased artificial drainage. For sulfate and TDS, wetter climatic conditions combined with naturally occurring and abundant sources of sulfate likely contributed to sustained exceedances of WQOs in recent decades, and extended periods of concentrations greater than or less than the WQO are likely dependent on climatic conditions. For sodium, total iron, and total phosphorus, sustained exceedances of the current WQO likely will continue because most sites across the basin had flow-averaged GMCs greater than the WQO, and exceedances were consistently greater than the WQO during the 43-year period of analysis regardless of climatic conditions.

Further investigation into the factors causing increasing sulfate concentrations and a better understanding of reservoir dynamics would enhance the understanding of changes in water-quality conditions in the Souris River Basin. Like the Red River of the North Basin, further investigation into ancillary variables that reflect widespread (throughout much of the basin) natural or human-induced changes that could increase the amount of naturally occurring sulfate reaching the streams from various hydrologic pathways and sources would help more specifically identify causes of sulfate increases (often referred to as “trend attribution”). Possible ancillary variables might include temporal and spatial changes in water-table elevations of shallow groundwater aquifers and associated increases in naturally occurring sulfate in shallow groundwater discharge, artificial surface or subsurface drainage improvements, and spatial soil characteristics. Explanatory variables for sulfate are also likely to be closely related to those for TDS and chloride. Further investigation into reservoir nutrient dynamics, such as reservoir modeling could provide insight for understanding of the seasonal and interannual variability in water quality and would provide a better understanding of nutrient water-quality conditions downstream from the reservoirs.

The basin-wide approach of this report provided an improved understanding of water-quality conditions in the Souris River Basin, and results can be used to inform the current WQOs, inform potential changes to water management in the basin, and serve as a starting point for tracking future progress. Gaps in understanding of water-quality conditions can be closed through continued monitoring and further investigation into causes of changes in water-quality conditions identified in this report.

The Souris River Basin is an international basin in southeast Saskatchewan, north-central North Dakota, and southwest Manitoba. In 1991, the International Souris River Board adopted water-quality objectives for more than 40 constituents at two transboundary crossings or binational sites: Saskatchewan–North Dakota (Souris River near Sherwood, North Dakota, U.S. Geological Survey station 05114000) and Manitoba–North Dakota (Souris River near Westhope, N. Dak., U.S. Geological Survey station 05124000). Sustained exceedances of water-quality objectives for total phosphorus, sodium, sulfate, total dissolved solids, and total iron have been reported since the late 1990s at the two binational sites on the Souris River. To understand conditions at the binational sites, it is important to understand water-quality changes on a basin-wide scale. Also, because streamflow is highly variable in the basin and changes in streamflow affect water-quality conditions, it is particularly important to use a trend-analysis method that accounts for changes in streamflow. Trends in water-quality concentrations can be affected by human-induced changes on the landscape or natural changes in land-runoff interactions that are driven by climate patterns and reflected by changes in streamflow (commonly referred to as “hydroclimatic variability”). In the primarily agricultural Souris River Basin, human-induced changes that are likely to affect trends are widespread changes in agricultural management such as fertilizer application, tilling practices, and crop types, as well as dam emplacement and artificial drainage. Around 1970, there was a long-term natural (hydroclimatic) change in the basin in which a significant transition from a dry climate state to a wet climate state resulted in higher streamflow in the basin. To assist the International Souris River Board in assessing current water-quality conditions in the Souris River Basin and exceedances of water-quality objectives at the binational sites, the U.S. Geological Survey, in cooperation with the International Joint Commission, completed a comprehensive analysis for selected ions, nutrients, and trace metals for many sites in the basin that included descriptive water-quality statistics, trend analysis using a trend method that considers interannual hydroclimatic variability, and an assessment of exceedances of the water-quality objectives for the binational sites.

Water-quality and streamflow or reservoir inflow or outflow data were compiled for 34 sites and 23 constituents from 1970 to 2020 in the Souris River Basin and were used for descriptive statistics and water-quality trend analysis. Water-quality data compiled from multiple agencies collected during multiple decades introduce the potential for inconsistencies between data for the same constituent. Inconsistencies in data caused by field-collection and laboratory-analytical methods were primarily addressed through identifying known or documented changes in field-collection or laboratory-analytical methods, recensoring to a common censoring level, and normalizing data based upon paired statistical testing.

Median total dissolved solids, sulfate, and sodium concentrations were low in the headwaters of the Souris River and some of the highest median concentrations were measured in the upper basin. Three of the four sites with median total dissolved solids concentrations greater than the binational total dissolved solids water-quality objective of 1,000 milligrams per liter were on the following tributaries: Des Lacs River, Moose Mountain Creek, and Pipestone Creek. Des Lacs River was the tributary with the highest median total dissolved solids concentration, and main-stem Pipestone Creek sites all had median concentrations greater than 850 milligrams per liter. The three highest median sulfate concentrations were in the upper basin, and all were more than 400 milligrams per liter; concentrations for sites on Souris River and Des Lacs River were greater than or equal to the 450 milligrams per liter sulfate water-quality objective. At main-stem Souris River sites, all median sodium concentrations were greater than the 100 milligrams per liter water-quality objective, with concentrations generally between 110 and 140 milligrams per liter. Median total phosphorus concentrations in the Souris River Basin were highest in the headwaters of the Souris River and all sites had median concentrations greater than the 0.1 milligram per liter water-quality objective. Median total iron concentrations were highly variable across the basin, ranging from 61.0 at Lake Darling near Foxholm, N. Dak. (U.S. Geological Survey station 05115500) to 1,420 micrograms per liter at Souris River near Bantry, N. Dak. (U.S. Geological Survey station 05122000), and other than Souris River near Roche Percee, Saskatchewan (Water Security Agency Saskatchewan station SK05NB0198), median concentrations were greater than or equal to the 300 micrograms per liter water-quality objective for main-stem sites.

During the recent period (2009–19), the annual flow-averaged geometric mean concentration of total dissolved solids increased at all sites evaluated except for Souris River near Sherwood, N. Dak. (U.S. Geological Survey station 05114000), and by 2019, one-half of the sites had an annual flow-averaged geometric mean concentration greater than the total dissolved solids water-quality objective of 1,000 milligrams per liter. Annual flow-averaged geometric mean concentration of sulfate increased during the recent period at all sites, with most sites having mildly significant or significant increases and five sites with increases greater than 100 milligrams per liter. Changes in annual flow-averaged geometric mean concentration of sodium generally were small and nonsignificant, and most of the sites had concentrations greater than the sodium water-quality objective of 100 milligrams per liter. Changes in the annual flow-averaged geometric mean concentration of chloride were generally small and nonsignificant with about one-half of the sites decreasing and one-half increasing, but, unlike sodium, all sites had concentrations much less than the chloride water-quality objective of 100 milligrams per liter. The annual flow-averaged geometric mean concentration of total boron decreased for four of the six sites during the recent period, but all sites had concentrations much less than the 500 micrograms per liter water-quality objective. Annual flow-averaged geometric mean concentration of total phosphorus decreased for nearly all sites across the Souris River Basin during the recent period including Souris River near Sherwood, N. Dak., and Souris River near Westhope, N. Dak., but all sites had concentrations greater than the total phosphorus water-quality objective of 0.1 milligram per liter for the entire period. Small and nonsignificant changes in annual flow-averaged geometric mean concentration of total iron were detected at all sites, except Souris River near Sherwood, N. Dak., and by 2019 all sites had concentrations greater than the total iron water-quality objective of 300 micrograms per liter. Nonsignificant small decreases in annual flow-averaged geometric mean concentration of total barium were detected at six of the seven sites evaluated during the recent trend period. Overall, trends in total molybdenum concentrations in the Souris River Basin were generally small and nonsignificant, except for two sites that were significant. During 2000–15, for total dissolved solids, sulfate, and sodium, concentrations increased in the reservoirs and most were significant. Reservoir changes in total phosphorus were variable with a nonsignificant increase in Grant Devine Lake and mildly significant or significant increasing and decreasing trends in the other reservoirs. Nonsignificant increases in flow-averaged geometric mean concentration of total iron were observed at the two reservoir sites evaluated for trends.

During the historical period (1976–2019), large and consistent increases in total dissolved solids and sulfate have occurred since the late 1980s, with the largest increases and the most sites with mildly significant or significant increases generally occurring during the middle period (1988–2005). By the end of the historical period (2019), flow-averaged geometric mean concentrations for one-half of the sites increased to concentrations greater than the water-quality objectives for total dissolved solids and sulfate. Increases in sodium concentrations were large and significant or mildly significant at eight of 10 sites in the middle period (1988–2005) and were small and nonsignificant by the late period (2005–19). At the start of the historical period in 1976, flow-averaged geometric mean concentrations for 7 of the 10 sites were greater than the binational site’s water-quality objectives of 100 milligrams per liter for sodium, including the binational sites. By the end of the historical period (2019), 9 of the 10 sites had flow-averaged geometric mean concentrations greater than the sodium water-quality objective.

Similar to other basins in the region, such as the Red River of the North and Heart River, large and overall consistent increases since the late 1980s in total dissolved solids and sulfate in the Souris River Basin suggest that long-term natural (hydroclimatic) processes are large contributors to increases in the concentration of salts in streams and reservoirs associated with the onset of wetter conditions. Although the software package R–QWTREND removes the variability in constituent concentration because of interannual streamflow variability, variability in sulfate caused by longer-term hydroclimatic variability may not be captured because of changes in hydrologic pathways and changes in the contributions of sulfate from various natural sources. The concurrent increases in sulfate and sodium concentrations at all sites during the middle period (1988–2005) suggest that sodium-sulfate evaporite dissolution may be a factor contributing to increases.

Total phosphorus concentrations oscillated between increasing and decreasing during the historical period, with concentrations increasing by as much as 77 percent during the first trend period (1976–88) to the highest flow-averaged geometric mean concentration in 1988 for most sites and decreasing in the fourth trend period (2009–19) to the lowest flow-averaged geometric mean concentration in 2019 for most sites. During the historical period, changes in total iron concentrations were mostly nonsignificant and generally small, except for Souris River near Sherwood, N. Dak., and variability in total iron concentrations likely affected the ability to detect statistically significant changes in concentration.

The probability of exceeding the water-quality objective for total dissolved solids, sulfate, and sodium increased between 1976 and 2019 for the binational sites, especially for sulfate, which more than doubled for Souris River near Sherwood, N. Dak. and increased more than seven times for Souris River near Westhope, N. Dak. Like Souris River near Sherwood, N. Dak., total phosphorus and total iron concentrations for Souris River near Westhope, N. Dak., were likely to exceed the water-quality objective for most of the year, but unlike Souris River near Sherwood, N. Dak., seasonal patterns of total phosphorus and total iron concentrations were different, suggesting that different factors may affect concentrations at different times of the year for Souris River near Westhope, N. Dak. For sodium, total phosphorus, and total iron, exceedance of the water-quality objective most of the time is not unexpected given that the flow-averaged geometric mean concentration for these three constituents for most sites across the basin are greater than the water-quality objective for most of the period. If natural processes are affecting total dissolved solids and sulfate concentrations, concentrations would be expected to vary with time, and as a result, extended periods of concentrations greater or less than the water-quality objective are likely to occur depending upon climatic conditions.

A better understanding of the state of water quality across the Souris River Basin is beneficial to understanding and interpreting water-quality conditions at the two Souris River binational sites. Although a better understanding of spatial and temporal changes in water quality in the basin was gained from this study, gaps in understanding of water-quality conditions were also identified. Less information on changes in water-quality conditions was available for the headwaters of the Souris River, Moose Mountain Creek, Long Creek, and Pipestone Creek because sites were sparser in these areas, and minimum data requirements for the trend method combined with data availability excluded many sites from trend analysis. Fewer sites and a shorter period were available for comparison of trace-metal trends because of changes in laboratory-analytical methods and sample portion analyzed, and inherent variability in total iron likely affected the ability to detect significant trends. The most consistent spatial and temporal change observed for this study was large and consistent increases in sulfate and total dissolved solids among tributary and main-stem sites since the late 1980s. Wetter climatic conditions that have persisted since about 1970, combined with naturally abundant sulfate in the soils, geology, and aquifers, has likely resulted in substantial changes in the hydrology and geochemistry of the drainage basin and, in turn, increased the amount of sulfate in soils near the surface and increased the mobility of sulfate in shallow groundwater inflow to streams. For sulfate and total dissolved solids, wetter climatic conditions combined with naturally occurring and abundant sources of sulfate likely contributed to sustained exceedances of water-quality objectives in recent decades, and extended periods of concentrations greater than or less than the water-quality objective are likely to occur depending on climatic conditions. For sodium, total iron, and total phosphorus, sustained exceedances of the current water-quality objective likely will continue because most sites across the basin had flow-averaged geometric mean concentrations greater than the water-quality objective, and exceedances were consistently greater than the water-quality objective during the 43-year period of analysis regardless of climatic conditions. Further investigation into the factors causing increasing sulfate concentrations and a better understanding of reservoir dynamics would enhance the understanding of changes in water-quality conditions in the Souris River Basin.

The basin-wide approach of this report provided an improved understanding of water-quality conditions in the Souris River Basin, and results can be used to inform the current water-quality objectives, inform potential changes to water management in the basin, and serve as a starting point for tracking future progress. Gaps in understanding of water-quality conditions can be closed through continued monitoring and further investigation into causes of changes in water-quality conditions identified in this report.