Streamflow Monitoring and Trends

All streamflow data at Wamego and De Soto were collected by the USGS using standard methods (Sauer and Turnipseed, 2010; Turnipseed and Sauer, 2010). Daily streamflow data have been collected at Wamego and De Soto since January 1919 and July 1917, respectively (U.S. Geological Survey, 2023). Daily streamflow data collected during January 1972 through December 2020 were used in combination with discrete water-quality data to compute flow-normalized and low-flow concentrations and loads of water-quality constituents of interest and characterize water-quality trends at Wamego and De Soto since the CWA was enacted in 1972.

Potential trends in streamflow could affect trends in flow-normalized water-quality concentrations and loads. Streamflow trends were examined using Quantile-Kendall plots (Hirsch, 2018). Each point on a Quantile-Kendall plot represents the trend in an annual order statistic. These annual order statistics range from the annual maximum flows (0-percent exceedance) on the left side of the plot to the annual minimum flows (100-percent exceedance) on the right side of the plot. Thus, a Quantile-Kendall plot shows the direction and magnitude of streamflow trends over the range of flow conditions. The daily streamflow values for each year are ordered by streamflow value, and the percent slope of the trend in the streamflow value at each rank was calculated. For example, the maximum daily streamflow value in each year was determined, the percentage slope was calculated for those maximum streamflows, and that percentage slope and the rank (expressed as an exceedance percentage) are plotted as a point on the Quantile-Kendall plot. These steps were repeated for the annual second-largest daily streamflow values and so on through to the annual minimum daily streamflow values. The slope is calculated using the Thiel-Sen slope estimator (Hirsch, 2018; Helsel and others, 2020) and a probability value (p-value) is calculated using the Mann-Kendall trend test (Mann, 1945). Quantile-Kendall plots that characterize potential trends in streamflow, shown in figures 4A–B, revealed no significant trends (p-values>0.1) in streamflow at Wamego and De Soto during 1972 through 2020.

Streamflow trends were also examined using streamflow duration curves showing streamflow conditions and frequencies of exceedance observed during the first (1972 through 1996) and second (1997 through 2020) halves of the study period at Wamego and De Soto (fig. 5). The second-half duration curves for both sites lie lower than the first-half curves, indicating that flows were lower during the second half of the period. However, trends in daily flows during the study period were not statistically significant (all p-values>0.1) using the Mann-Kendall trend test (fig. 4).

Water-Quality Sample Collection

The USGS water-quality samples used in this report were collected at Wamego and De Soto as part of multiple historical USGS projects and networks during 1972 through 2020 (Briggs and Ficke, 1977; Rasmussen and others, 2005; Foster and Graham, 2016; Graham and others, 2018; Williams, 2021; Lee, 2022, among others). USGS water-quality samples were primarily collected using depth- and width-integrated isokinetic sampling techniques (U.S. Geological Survey, 2006), except for fecal coliform bacteria samples, which were collected at the centroid of flow using a weighted basket. Occasionally, nonisokinetic sampling techniques (including single and multiple verticals or grab samples; U.S. Geological Survey, 2006) were used during extreme low-flow conditions or for safety purposes during extreme cold conditions. USGS discrete water-quality data are available from the USGS National Water Information System database (U.S. Geological Survey, 2023).

USGS samples collected during 1972 through 2020 were analyzed for total dissolved solids (USGS parameter code 70300), chloride (USGS parameter code 00940), ammonia (USGS parameter codes 00610 during September 1977 through September 1979; 00608 during October 1979 through October 1980; 00610 during November 1980; and 00608 during December 1980 onward), nitrate plus nitrite (USGS parameter codes 00630 during September 1977 through August 1979; and 00631 during September 1979 onward), total Kjeldahl nitrogen (USGS parameter code 00625), orthophosphate (USGS parameter code 00671), total phosphorus (USGS parameter code 00665), total suspended solids (USGS parameter code 00530), and fecal coliform bacteria (USGS parameter code 31616 during October 1974 through September 1976; and 31625 during October 1976 onward). USGS parameter codes for ammonia and nitrate plus nitrite consist of a combination of unfiltered (total) and filtered (dissolved) analyses because of differences in analytical procedures. The substitution of filtered analyses of ammonia and nitrate plus nitrite for unfiltered analyses was appropriate because the difference was minimal between unfiltered and filtered sample results according to the evaluation of methods described in the USGS Office of Water Quality Technical Memorandum 93.04 (U.S. Geological Survey, 1992). USGS samples of total dissolved solids, chloride, ammonia, nitrate plus nitrite, total Kjeldahl nitrogen, orthophosphate, total phosphorus, and total suspended solids were analyzed at the USGS National Water-Quality Laboratory using the methods described by Fishman and Friedman (1989). USGS samples of fecal coliform bacteria were analyzed by the USGS Kansas Water Science Center using the methods described by Myers and others (2014).

The KDHE water-quality samples used in this report were collected at Wamego and De Soto on a biweekly to quarterly basis during 1972 through 2020 as part of the Kansas Stream Chemistry Monitoring Program to characterize physical, chemical, and bacteriological conditions in the Kansas River (Kansas Department of Health and Environment, 2020). KDHE water-quality samples were collected at the centroid of flow using a weighted stainless-steel bucket (Kansas Department of Health and Environment, 2020). KDHE discrete water-quality data used in this report are available in Williams and others (2024).

KDHE samples collected during 1972 through 2020 were analyzed for total dissolved solids, chloride, ammonia, nitrate plus nitrite, total Kjeldahl nitrogen, orthophosphate, total phosphorus, total suspended solids, and fecal coliform bacteria, among other analytes. Unlike the USGS nitrate plus nitrite samples (reported as a combination of total and dissolved concentrations), the KDHE nitrate plus nitrite samples were reported as exclusively total, rather than dissolved, concentrations. However, filtration was required prior to analysis, particularly on high turbidity samples, to prevent damage to the analytical instrumentation (Tony Stahl, KDHE, written commun., August 27, 2023). Additionally, KDHE does not consider there to be a difference in total or dissolved nitrate plus nitrite concentrations from a regulatory or management standpoint (Tom Stiles, KDHE, written commun., August 18, 2023). Therefore, it was deemed appropriate to combine the USGS and KDHE nitrate plus nitrite datasets for the purposes of this report. KDHE samples of total dissolved solids, chloride, ammonia, nitrate plus nitrite, total Kjeldahl nitrogen, orthophosphate, total phosphorus, and total suspended solids were analyzed by the Kansas Health and Environmental Laboratory using methods described by the Kansas Department of Health and Environment (2020).

The combination of USGS and KDHE samples used for trend analyses in this report allowed for a larger and higher frequency dataset that reduced some uncertainties in estimated water-quality trends during the study period. These datasets were used to identify and investigate long-term water-quality trends of constituents of interest during 1972 through 2020. The constituents of interest in this report consisted of total dissolved solids, chloride, ammonia, dissolved inorganic nitrogen (computed by summing ammonia and nitrate plus nitrite), total nitrogen (computed by summing total Kjeldahl nitrogen and nitrate plus nitrite), orthophosphate, total phosphorus, total suspended solids, and fecal coliform bacteria. Dissolved inorganic nitrogen or total nitrogen results were not computed if both counterparts (ammonia and nitrate plus nitrite, or total Kjeldahl nitrogen and nitrate plus nitrite, respectively) were not available for a given USGS or KDHE sample. The number of samples and duration of sample collection during the study period for each constituent of interest are shown in table 1. USGS and KDHE sample results for these constituents were used to compute associated flow-normalized and low-flow trend likelihood results during 1972 through 2020.

Quality Assurance and Quality Control of Streamflow and Water-Quality Data

All streamflow data collected by the USGS at Wamego and De Soto during 1972 through 2020 were quality assured using standard USGS methods (Sauer and Turnipseed, 2010; Turnipseed and Sauer, 201040; Painter and Loving, 2015). These data were analyzed, reviewed, and approved following USGS guidance (U.S. Geological Survey, 2017).

Water-quality data preparation for the trend analyses described in this report involved data compilation, and identification and removal of inappropriate samples. Two USGS ammonia results and three USGS nitrate plus nitrite samples collected from De Soto had concentrations of zero, and, for reasons documented by Lee and Henderson (2020), these results were assumed to be missing and were not used in the trend analysis. If more than one sample of a constituent was collected on the same day (not including replicate samples for quality-control purposes), only the first sample of the day was retained in the trend analysis dataset. To reduce uncertainty caused by extended periods without sample collection, samples collected during 1972 through 2020 were not used in the trend analysis if no other USGS or KDHE samples were collected within 2 years of the sample. There were no other instances or justification of removal of USGS or KDHE samples from the trend analysis dataset. All USGS and KDHE data used in the WRTDS trend analysis are included in a USGS data release (Williams and others, 2024).

Quality-control samples, including concurrent replicates and field blanks, were collected at Wamego and De Soto by the USGS and KDHE to characterize variability and bias in sample results potentially introduced by equipment cleaning and rinsing procedures, sample-collection methods, sample-processing techniques, and analytical methods (Mueller and others, 2015). Relative percentage difference (RPD) among concurrent replicate sample pairs was used to quantify variability in constituent concentrations and was calculated by dividing the absolute difference of a pair of replicate samples by their associated mean concentration or density and then multiplying by 100 (Zar, 1999). RPDs were calculated for concurrent replicate pairs of samples only if sample times were within 30 minutes of each other. RPD could not be calculated for replicate pairs that contained one or more censored (data reported as less than the laboratory minimum reporting limit [MRL]) result. Median RPDs among USGS and KDHE concurrent replicate pairs of total dissolved solids and chloride were less than 2 percent (table 2). USGS nutrient (nitrogen and phosphorus species) median concurrent replicate RPDs were less than 7 percent (table 2). KDHE nutrient (nitrogen and phosphorus species) median concurrent replicate RPDs were less 10 percent, except for ammonia, which had a median RPD of about 29 percent (table 2), likely due to low concentrations that were less than the highest ammonia MRL (0.1 milligram per liter [mg/L]) during the study period. Median RPDs among USGS and KDHE concurrent replicate pairs of total suspended solids were less than 10 percent (table 2). Median RPDs were less than 40 percent for all fecal coliform bacteria replicate paired samples collected by USGS and KDHE (table 2).

Field blank samples were collected by USGS and KDHE to characterize bias potentially introduced by contamination from the sample environment, sampling procedures, and analytical methods (Mueller and others, 2015). All USGS total dissolved solids and chloride field blank sample results were less than two times the associated maximum MRL (table 2). All KDHE field blank sample results for total dissolved solids were greater than two times the associated maximum MRL; however, all these results were less than 7 percent of the associated environmental discrete sample concentration (table 2). One KDHE chloride field blank sample result was greater than 10 percent of the associated environmental discrete sample concentration. All USGS and KDHE nutrient (nitrogen and phosphorus species) field blank sample results were less than two times the associated maximum MRL, except for KDHE nitrate, total Kjeldahl nitrogen, and total phosphorus (table 2). Three KDHE nitrate field blank sample results were greater than two times the associated maximum MRL and greater than 10 percent of the associated environmental discrete sample concentration. Two KDHE total Kjeldahl nitrogen field blank sample results were greater than two times the associated maximum MRL and greater than 10 percent of the associated environmental discrete sample concentration. Four KDHE total phosphorus field blank sample results were greater than two times the associated maximum MRL and greater than 10 percent of the associated environmental discrete sample concentration. All USGS and KDHE total suspended solids field blank sample results were less than two times the associated maximum MRL (table 2). All USGS and KDHE fecal coliform bacteria field blank sample results were less than two times the associated maximum MRL (table 2).

Comparison of Water-Quality Data

Trend analyses in this report included samples collected and analyzed by USGS and KDHE personnel and laboratories. It is important to determine whether differences in sample collection and analytical procedures among agencies have the potential to affect trend analyses. However, without concurrently collected samples it is difficult to determine if observed differences are related to environmental conditions or agency-specific collection and (or) analytical procedures, particularly when sample concentrations are below laboratory reporting levels. The potential for the sampling agency and analytical laboratory to affect trend results across the range of streamflow conditions was indirectly characterized using analysis of covariance (ANCOVA) of the natural logarithm of sample concentrations relative to the natural logarithm of streamflow conditions. ANCOVA can be used to evaluate the effect of qualitative factors, such as sampling agency, on a dependent variable (such as water-quality constituents) within the context of other potentially influential independent variables (such as streamflow; Helsel and others, 2020). KDHE and USGS agencies were compared during years when water-quality sample data were available from both agencies and when KDHE samples had a USGS sample within 20 percent of the streamflow in which it was collected (table 3). In addition, to account for different laboratory reporting levels among agencies and sampling years, data below the median censoring level for a given constituent, and all results denoted as less than the laboratory reporting level, were not included in the analysis. ANCOVA was performed for a given site and constituent if the dataset had at least 20 sample pairs during the study period. Limited sample collection and censored data did not allow sampling-agency comparisons for ammonia and fecal coliform bacteria at Wamego, or for comparison of dissolved inorganic nitrogen, orthophosphate, and total nitrogen at both sites. To assess the comparability of nitrogen species, ANCOVA was performed on nitrate and total Kjeldahl nitrogen. The datasets for each study site were evaluated to determine if there was a significant (p-value<0.05) difference in the slope and intercept of constituent concentration versus streamflow regression relation lines between sampling agencies.

Of the 14 site and constituent pairs, 5 had significant (p-value<0.05) differences in the slope (nitrate at De Soto and total Kjeldahl nitrogen at Wamego) or intercept (nitrate at Wamego, and total suspended solids at Wamego and De Soto) of constituent concentrations among agencies relative to streamflow conditions (table 3). Total dissolved solids, chloride, and total phosphorus concentrations varied similarly between agencies with respect to similar streamflows; although not enough data were available to compare fecal coliform densities and ammonia concentrations at Wamego, the slope and intercept among USGS and KDHE for these constituents were relatively similar at De Soto. USGS-sampled total suspended solids concentrations were significantly larger than KDHE-sampled suspended-solids concentrations during similar flow conditions at Wamego and De Soto. The slope of the relation between total Kjeldahl nitrogen and streamflow was significantly larger for USGS than KDHE data at the Wamego site. KDHE nitrate concentrations were higher across the range of streamflows at Wamego, but the slope of nitrate concentrations across flows was higher for USGS data at De Soto; different results among sites may indicate that agency-specific factors other than nitrate analytical methods are influencing concentrations, or that more data are needed to determine if observed differences are consistent.

Because samples among agencies were not concurrently collected, these comparisons do not necessarily mean that trend results using data from both agencies are affected by sampling agency (or that sites/constituents without significant differences are not), but that the proportion of data among the sampling agencies across the period of record should be considered when interpreting trend results. For these reasons, all USGS and KDHE data described in the Water-Quality Sample Collection section were retained in the trend analysis dataset (Williams and others, 2024) used in this report.

Computation of Water-Quality Trends

Water-quality trends in concentrations and loads of total dissolved solids, chloride, ammonia, dissolved inorganic nitrogen, total nitrogen, orthophosphate, total phosphorus, total suspended solids, and fecal coliform bacteria densities were computed and analyzed using the Exploration and Graphics for RivER Trends (EGRET) R package (Hirsch and De Cicco, 2015). The EGRET R package uses WRTDS (Hirsch and others, 2010; Hirsch and De Cicco, 2015) to compute flow-normalized trends of water-quality concentrations and loads while removing the effects of seasonality and year-to-year variations related to streamflow.

WRTDS models estimate daily concentrations and loads of water-quality constituents by creating a weighted regression model for each day in the model period. The model inputs are time (in decimal years), daily discharge (streamflow), and season (time of year). Data points are weighted based on the distances in these three dimensions between the sample day and the model estimate day. Data points that are closer to the model estimate day have higher weights. There is a distance in each dimension called the “half-window width” (Hirsch and others, 2010) beyond which the weight goes to zero. The data points that are within the half-window width have nonzero weights which, increase depending on the closeness in time, discharge, and season between the sampled day and model estimate day. These weighted data points are used to fit a regression equation that is used for estimating the concentration for that day. Flow-normalized daily concentrations and loads are calculated by producing model estimates for each streamflow during the model period on the calendar date of the day being estimated (using the time and season of the day being estimated for all of them) and calculating the mean of those estimates. For example, the flow-normalized concentration for January 1, 2010, would be the mean of the estimated daily concentrations calculated using the daily streamflows from every January 1 in the model period and the time and season of January 1, 2010. Flow-normalized concentrations and loads are used to examine water-quality trends while accounting for the effect of variations in streamflow.

To estimate the uncertainty and likelihood of upward or downward trends in concentrations and loads, the Exploration and Graphics for RivEr Trends Confidence Intervals (EGRETci) R package (Hirsch and others, 2015) uses a block-sampling bootstrap method called the WRTDS bootstrap test (WBT; Hirsch and others, 2015) to produce 500 replicate models. Rather than relying on the results of a single realization of the model, the large number of replicate models created with the WBT method allow for distributions of model results to be used to estimate probabilities of trends and confidence intervals. The difference between the annual means of the flow-normalized model results from the first and last years of the model period was used to determine if a replicate model exhibited an upward or downward trend. The distribution of positive and negative trends from all the replicate models was used to estimate the probabilities of upward and downward trends. Because the discrete water-quality samples used in this report did not span the first and last full years (that is, January 1 through December 31) of a given constituent’s dataset, WBT estimates were not computed for the first and last year in which discrete samples were collected to avoid extrapolating in time. For example, samples collected in 2020 were used in the WRTDS model, but WBT results were available only through 2019. Likelihood descriptors, ranging from highly likely to highly unlikely, were then assigned based on trend probabilities (table 4).

Water-quality trends during low-flow conditions can be indicative of changes in point-source pollution. The WBT method also was used to estimate water-quality trend likelihoods at low flow by using a constant low-flow value (instead of using flow-normalized model results) to estimate low-flow concentrations and loads throughout the model period. The 10th-percentile (90-percent exceedance) streamflows during 1972 through 2020 were used for the low-flow values. The actual daily flows were used to develop the model, and then the model was used to make predictions of what the concentrations would have been if the flow was the 10th percentile value for every day. These streamflow values for the study period were 840 cubic feet per second (ft³/s) at Wamego and 1,200 ft³/s at De Soto. This approach followed the KDHE total maximum daily load methodology for determination of base-flow conditions (Kansas Department of Health and Environment, 2017), where water-quality constituents are more likely to be affected by potential point-source pollution.

Total Dissolved Solids and Chloride

Based on the WBT trend likelihood descriptors (Hirsch and others, 2015; table 4), flow-normalized total dissolved solids concentrations were likely to have an upward trend at Wamego (5.7-percent increase) between 1973 and 2019 and not likely to have a trend (upward trend is about as likely as not) at De Soto between 1974 and 2019 (table 5). During the same time periods, flow-normalized total dissolved solids loads were likely to have upward trends at both sites (4.9- and 6.3-percent increases at Wamego and De Soto, respectively; table 5). Low-flow (10th percentile of streamflows during 1972 through 2020 at Wamego and De Soto [840 ft³/s and 1,200 ft³/s, respectively]) total dissolved solids concentrations were likely to have downward trends at Wamego and De Soto (6.2- and 6.3-percent decreases, respectively) during the same time periods (table 6). Estimated annual mean, flow-normalized, and low-flow total dissolved solids concentrations with 90-percent confidence intervals at Wamego and De Soto are shown in figures 6A–B.

Flow-normalized chloride concentrations were not likely to have a trend at Wamego between 1973 and 2019 and were likely to have a downward trend at De Soto (7-percent decrease) between 1974 and 2019 (table 5). Similarly, flow-normalized chloride loads were not likely to have a trend at Wamego and were likely to have a downward trend at De Soto (6.7-percent decrease) during the same time periods (table 5). Low-flow chloride concentrations were likely to have downward trends at Wamego (13-percent decrease) and De Soto (12-percent decrease) during the same time periods (table 6). Estimated annual mean, flow-normalized, and low-flow chloride concentrations with 90-percent confidence intervals at Wamego and De Soto are shown in figures 7A–B.

Nutrients

All flow-normalized concentrations and loads of nitrogen species (ammonia, dissolved inorganic nitrogen, and total nitrogen) were very likely or highly likely to have downward trends at Wamego and De Soto. Similarly, all low-flow concentrations of nitrogen species were highly likely to have downward trends at both sites. Flow-normalized ammonia concentrations were highly likely to have downward trends at Wamego (94-percent decrease) and De Soto (97-percent decrease) between 1974 and 2019 (table 5). Similarly, flow-normalized ammonia loads were very likely to have a downward trend at Wamego (89-percent decrease) and were highly likely to have a downward trend at De Soto (94-percent decrease) during the same time period (table 5). Low-flow ammonia concentrations were also highly likely to have downward trends at Wamego and De Soto (98-percent decreases at both sites) during the same time period (table 6). Flow-normalized dissolved inorganic nitrogen concentrations were highly likely to have downward trends at Wamego (38-percent decrease) and De Soto (43-percent decrease) between 1978 and 2019 (table 5). Flow-normalized dissolved inorganic nitrogen loads were also highly likely to have downward trends at both sites (27- and 38-percent decreases at Wamego and De Soto, respectively) during the same time period (table 5). Low-flow dissolved inorganic nitrogen concentrations were highly likely to have downward trends at Wamego (44-percent decrease) and De Soto (67-percent decrease) during the same time period (table 6). Flow-normalized total nitrogen concentrations were highly likely to have downward trends between 2000 and 2019 at Wamego (28-percent decrease) and between 1975 and 2019 at De Soto (42-percent decrease; table 5). Similarly, flow-normalized total nitrogen loads were highly likely to have downward trends at Wamego (23-percent decrease) and De Soto (31-percent decrease) during the same time periods (table 5). Low-flow total nitrogen concentrations were highly likely to have downward trends at Wamego (36-percent decrease) and De Soto (55-percent decrease) during the same time periods (table 6). Percent decreases in flow-normalized and low-flow concentrations and loads of nitrogen species were generally larger at De Soto. Estimated annual mean, flow-normalized, and low-flow ammonia, dissolved inorganic nitrogen, and total nitrogen concentrations with 90-percent confidence intervals at Wamego and De Soto are shown in figures 8A–10B.

Flow-normalized concentrations and loads of phosphorus species (orthophosphate and total phosphorus) generally had upward trends at both sites during the study period. Low-flow trends in concentration were less consistent at the De Soto site, with results indicating likely downward trends. Flow-normalized orthophosphate concentrations were highly likely to have an upward trend at Wamego between 1996 and 2019 (180-percent increase) and were not likely to have a trend at De Soto between 1982 and 2019 (table 5). Flow-normalized orthophosphate loads were highly likely to have upward trends at Wamego (120-percent increase) and De Soto (39-percent increase) during the same time periods (table 5). Low-flow orthophosphate concentrations were highly likely to have an upward trend at Wamego (590-percent increase) and were likely to have a downward trend at De Soto (31-percent decrease) during the same time periods (table 6). Flow-normalized total phosphorus concentrations were highly likely to have upward trends between 1973 and 2019 at Wamego (120-percent increase) and between 1974 and 2019 at De Soto (49-percent increase; table 5). Similarly, flow-normalized total phosphorus loads were highly likely to have upward trends at Wamego (220-percent increase) and De Soto (150-percent increase) during the same time periods (table 5). Low-flow total phosphorus concentrations were highly likely to have an upward trend at Wamego (55-percent increase) and were likely to have a downward trend at De Soto (14-percent decrease) during the same time periods (table 6). Percentage increases in flow-normalized concentrations and loads were consistently larger at the Wamego site for orthophosphate and total phosphorus. Estimated annual mean, flow-normalized, and low-flow orthophosphate and total phosphorus concentrations with 90-percent confidence intervals at Wamego and De Soto are shown in figures 11A–12B.

Total Suspended Solids

Flow-normalized total suspended solids concentrations were highly likely to have a downward trend at Wamego (32-percent decrease) and De Soto (49-percent decrease) between 1974 and 2019 (table 5). Similarly, flow-normalized total suspended solids loads were likely to have a downward trend at Wamego (28-percent decrease) and were very likely to have a downward trend at De Soto (49-percent decrease) during the same time period (table 5). Low-flow total suspended solids concentrations were highly likely to have downward trends at Wamego (45-percent decrease) and De Soto (55-percent decrease) during the same time period (table 6). Percentage decreases in flow-normalized and low-flow concentrations and loads were consistently larger at De Soto. Estimated annual mean, flow-normalized, and low-flow total suspended solids concentrations with 90-percent confidence intervals at Wamego and De Soto are shown in figures 13A–B.

Fecal Coliform Bacteria

Flow-normalized fecal coliform bacteria densities were very likely to have a downward trend at Wamego (88-percent decrease) and were highly likely to have a downward trend at De Soto (93-percent decrease) between 1974 and 2019 (table 5). Similarly, flow-normalized fecal coliform bacteria loads were likely to have a downward trend at Wamego (76-percent decrease) and De Soto (85-percent decrease) during the same time period (table 5). Low-flow fecal coliform bacteria densities were highly likely to have downward trends at Wamego (79-percent decrease) and De Soto (97-percent decrease) during the same time period (table 6). Percentage decreases in flow-normalized and low-flow densities and loads were consistently larger at De Soto. Estimated annual mean, flow-normalized, and low-flow fecal coliform bacteria densities with 90-percent confidence intervals at Wamego and De Soto are shown in figures 14A–B.