Purpose and Scope

This report provides a status update on water quality in the lower Boise River watershed in southwest Idaho. Numerous water quality improvement projects and TMDLs have been implemented since the previous watershed assessment in 2009–2012 by Etheridge and others (2014). This update is needed to provide information for the evaluation of the effectiveness of these improvement projects and help watershed managers prioritize sites and constituents to maximize returns on future water-quality improvement efforts. We present constituent concentrations and loads from 29 synoptic sampling events of select tributaries and the mainstem Boise River during the irrigation and non-irrigation seasons in water years 2019–2023. For constituents and sites with adequate data, we also assess trends in concentrations and loads over a 15–30-year period. The sampled tributaries were Fifteenmile Creek, Mill Slough, Mason Creek, Indian Creek, Dixie Slough, Conway Gulch, and Sand Hollow Creek (a tributary to the Snake River). Mainstem Boise River sites were located at Eckert Road (upstream from Boise), near Middleton, and near Parma. Fivemile Creek and Tenmile Creek, the two main tributaries to Fifteenmile Creek, were also sampled. All sites were sampled for SSC, TP, and OP, and Boise River sites were also sampled for TN, nitrate plus nitrite, E. coli, and periphyton chl-a (table 1). All years are water years (October 1—September 30) unless specified otherwise.

Description of Study Area

The lower Boise River watershed and the monitoring locations in the current study were described in detail by Etheridge (2013) and Etheridge and others (2014) and are briefly described here. The lower Boise River extends 103 kilometers (km) from the outflow of Lucky Peak Dam to the confluence with the Snake River. Along this reach, the river flows through the cities of Boise, Garden City, Eagle, Star, Middleton, Caldwell, Notus, and Parma, Idaho. We collected samples at three mainstem locations spanning most of the length of the lower Boise River. From upstream to downstream, these were the Boise River at Eckert Road (U.S. Geological Survey [USGS] site 13203760), the Boise River at Middleton (USGS site 13210050), and the Boise River near Parma (USGS site 13213000; fig. 1; table 1; U.S. Geological Survey, 2024). The Boise River at Eckert is located upstream from most urban and agricultural influences and is considered relatively pristine (Etheridge and others, 2014). Between Eckert and Middleton, urbanization increases and the Boise River receives effluent from the City of Boise’s Lander and West Boise wastewater treatment plants. No major tributaries enter the Boise River between Eckert and Middleton, but water is withdrawn via several large diversions, reducing the streamflow in this reach during the summer irrigation season. Downstream from Middleton, several major tributaries enter the Boise River, including, from upstream to downstream, Fifteenmile Creek, Mill Slough, Mason Creek, Indian Creek, Conway Gulch, and Dixie Slough (fig. 1; table 1). These tributaries drain large areas of agricultural land and, in some cases, receive municipal wastewater effluent (for example, Indian Creek, Mill Slough, Conway Gulch, and Fifteenmile Creek via its tributary Fivemile Creek), and have previously been identified as important sources of sediment and nutrients to the lower Boise River (Etheridge, 2013; Etheridge and others, 2014). The Boise River near Parma, near the confluence with the Snake River, is located downstream from these major tributaries and therefore represents the mix of agricultural and urban influences in the watershed. Samples were also collected at Sand Hollow Creek, which receives wastewater from the City of Parma. Although Sand Hollow Creek discharges to the Snake River, it is generally considered to be part of the lower Boise River system and is included in the lower Boise River TMDL (Idaho Department of Environmental Quality, 2015a).

Sample Collection and Analysis

Sites were sampled six times per year from 2018 to 2023 (five in 2022), targeting a range of hydrologic and management conditions including high and low flows during the irrigation and non-irrigation seasons, and the spring “freshet,” or first flush. In 1994–2017, samples were collected on less regular frequencies which varied over time and by site and constituent. At Boise River near Parma, TP samples were collected at high frequency (as often as every 2 days) starting in 2004 and continuing through 2023 as part of a related study in partnership with the City of Boise (King and Yoder, 2025).

All samples were collected and processed following the U.S. Geological Survey (USGS) National Field Manual (U.S. Geological Survey, variously dated) and previous lower Boise studies (Etheridge, 2013; Etheridge and others, 2014). Water samples were collected using the equal-width increment method where possible, and multiple vertical samples or grab samples were collected in cases where wading was not possible or where velocities were too slow. Water samples were collected using a 1-liter high-density polyethylene bottle with a depth-integrating sediment and water-quality sampler (model DH-81, DH-95, or D-95, depending on conditions) and homogenized in a plastic churn. Water samples from Dixie Slough were collected by the City of Boise using similar methods. Most of the high-frequency TP samples at the Boise River near Parma were collected using an automated refrigerated sampler (Teledyne ISCO, Lincoln, Nebraska) from the riverbank. Samples from the automated sampler were compared against 25 concurrent samples collected using the equal-width increment method during varying hydrologic conditions. The median relative percent difference between TP concentrations collected using the two different methods was 1.7 percent, confirming that samples collected using the autosampler were representative of the whole-channel conditions (King and Yoder, 2025). Periphyton chl-a samples were collected three times per year (October, February, and August) at the three mainstem Boise River sites following methods described by Moulton and others (2002). Scrapings from 11 submerged rocks were composited, then a subsample of known volume was filtered through a 0.7-micrometer (µm; nominal) glass fiber filter. Filters were wrapped in aluminum foil and frozen. Field replicates and blanks of each constituent were collected throughout the study to ensure data quality. Results are summarized in appendix table 1.1 and appendix figure 1.1.

Water samples were analyzed for SSC, TP, and OP, and samples from the mainstem Boise River sites were also analyzed for TN, nitrate plus nitrite, and E. coli. For SSC, unfiltered water was shipped to the USGS Cascade Volcano Observatory (Vancouver, Washington) and analyzed using methods described in Guy (1969). For TP and TN, unfiltered water was preserved with sulfuric acid, chilled at 4 degrees Celsius (°C), shipped to the USGS National Water Quality Monitoring Laboratory (NWQL; Denver, Colorado), and analyzed using methods described in Fishman (1993) and Patton and Kryskalla (2003). For E. coli, unfiltered water was chilled at 4 °C and analyzed within 24 hours at the Idaho Department of Health and Welfare Laboratory (Boise, Idaho) using standard method number 9223B (Eaton and others, 1999). Water samples for dissolved nutrients (OP and nitrate plus nitrite) were filtered through a 0.45-µm capsule filter, chilled at 4 °C, and analyzed at NWQL using methods described in Fishman (1993) and Patton and Kryskalla (2011). Periphyton chl-a samples were analyzed within 1 week of collection by the Bureau of Reclamation Pacific Northwest Regional Laboratory (Boise, Idaho) using standard spectrophotometric method number 10200H (Eaton and others, 1999). All sample results are published in Baldwin and King (2025).

Discharge

Mean daily discharge records and their sources are provided in Baldwin and King (2025). Briefly, discharge records were from USGS stream gages (table 1; U.S. Geological Survey, 2024) except at the following sites. Mill Slough discharge was calculated as the sum of discharges from the North and South Middleton Drains (USGS stream gages 13210842 and 13210831, respectively; U.S. Geological Survey, 2024). Boise River at Middleton discharge was provided by the Bureau of Reclamation (Bureau of Reclamation, 2024). Discharge for Boise River at Eckert was calculated from water rights accounting records provided by the Idaho Department of Water Resources by subtracting the discharges in the New York Canal, Penitentiary Canal, and Ridenbaugh Canal from the discharge from Lucky Peak Dam (Bureau of Reclamation, 2024; Idaho Department of Water Resources, 2024).

Discharge at Fivemile and Tenmile Creeks—tributaries to Fifteenmile Creek—were measured following USGS standard methods (Turnipseed and Sauer, 2010). For dates when discharge measurements were not collected at these sites, flow was approximated with linear regressions between discharge at these sites and discharge at the Fifteenmile Creek streamgage. There were 23 and 24 discrete discharge observations at Fivemile and Tenmile Creeks, respectively, and both regressions had adjusted R-squared values of 0.7 or higher. Diversions and return flows between these sites and the streamgage at Fifteenmile Creek are thought to be the primary sources of variability in these relations.

Indian Creek discharge records varied by season. During winter months (mid-October–mid-April), Indian Creek briefly flows through a section of the Riverside Canal, which is otherwise dry during winter, before flowing into the Boise River near Caldwell. During this period, the Indian Creek discharge to the Boise River was provided by an Idaho Power Company streamgage in Caldwell, upstream from the Indian Creek–Riverside Canal confluence (Idaho Power Company, 2022). During summer months (mid-April–mid-October), Riverside Canal is flowing and therefore mixes with Indian Creek at their confluence. To reduce nutrient loads to the Boise River during the irrigation season, most of Riverside Canal downstream from the Indian Creek confluence is conveyed to irrigated lands west and south of Caldwell, rather than discharging directly into the Boise River. Some of this water does eventually discharge to the Boise River, but further downstream, through Dixie Slough (Etheridge and others, 2014) and through shallow groundwater return flows in the winter (Etheridge, 2013). There were some cases in the mid-April–mid-October period when flow in Riverside Canal downstream from the Indian Creek confluence was spilled to the Boise River via Indian Creek. The proportion of spill from Riverside Canal to the Boise River in the summer that came from Indian Creek was assumed to be proportional to the contribution of Indian Creek to Riverside Canal downstream from their confluence. For example, if 20 percent of the discharge in Riverside Canal came from Indian Creek, then 20 percent of the spill from Riverside Canal to the Boise River was assumed to come from Indian Creek (Idaho Power, written commun., 2023).

Recent Spatial and Seasonal Patterns

Recent spatial and seasonal patterns in constituent concentrations and loads were evaluated using data from 29 synoptic sampling events during the 5-year period from water year 2019–2023. Measured concentrations and daily discharge data were used to estimate concentrations and loads for each day of the 5-year period using the composite method in the R package loadflex (Appling and others, 2015). The composite method uses measured values when available, and estimates values for days between measured values using a combination of regression and interpolation (Aulenbach and Hooper, 2006). The regression models used in the composite method were based on LOADEST model number 7, which includes predictor variables for streamflow and season (Runkel and others, 2004; Lorenz and others, 2015). Loads are reported as mass per day (SSC, TP, OP, TN, nitrate plus nitrite) or most probable number per day (E. coli). Loads were not calculated for periphyton chl-a, as it is generally not a constituent that is transported in streamflow. To help highlight the importance of individual sites relative to others, loads are also presented as percentages of the total inflow load, where the total inflow load is the sum of loads from all monitored sites upstream from the Boise River near Parma (in other words, the Boise River at Middleton, Fifteenmile Creek, Mill Slough, Mason Creek, Indian Creek, Conway Gulch, and Dixie Slough). Comparing this total inflow load to the load at Boise River near Parma (the outflow) provided insights into the presence of unmeasured sources (tributaries, drains, groundwater seeps) or sinks (deposition/storage, biological uptake, loss to groundwater) between Middleton and Parma.

The percentages of TP as OP, and TN as nitrate plus nitrite, were calculated from measured values to help understand the importance of different sources in different seasons. Orthophosphate and nitrate are naturally occurring but environmental concentrations increase with the use of agricultural chemical fertilizers, manure, and composted organic materials, as well as from leaking septic systems and inadequately-treated wastewater (U.S. Geological Survey, 1999; Wakida and Lerner, 2005; Kim and others, 2015; Kent and others, 2020). Orthophosphate and nitrate readily leach into and are transported by shallow groundwater, creating an important source pathway to surface waters (U.S. Geological Survey, 1999; Domagalski and Johnson, 2012). In contrast, particulate phosphorus and nitrogen are typically mobilized by erosion and transported in surface waters. The percentages of TP as OP, and TN as nitrate plus nitrite, can therefore indicate the relative contributions from groundwater versus surface water.

For sites and constituents with concentrations below the reporting limit (left-censored concentrations; table 3), mean concentrations were estimated using regression on order statistics using the ros function in the NADA package in R (Lee, 2020). At the Boise River at Eckert, where some concentrations were left-censored (table 3), loads were estimated by substituting left-censored concentrations with the reporting-limit concentration. These estimates therefore represent the maximum possible loads coming into the lower Boise River and provide a baseline for comparison to sites downstream. Because some concentrations at Eckert were slightly lower than the reporting limit, the actual loads were likely slightly lower than our load estimates, and therefore load increases from Eckert to Middleton are likely slightly greater than the estimates presented here.

Temporal Trends

Temporal trends in measured concentrations were evaluated for the 30-year period from 1994–2023 for SSC, TP, OP, TN, nitrate plus nitrite, and periphyton chl-a; and from 1999 to 2023 for E. coli. The starting years were selected based on the beginning of regular monitoring across all sites, with at least one sample collected in each season (irrigation and non-irrigation seasons). Trends in periphyton chl-a concentrations were limited to samples collected in October–November because 64 percent of all samples were collected during those 2 months, and data in other months were lacking, especially in the early part of the record. At mainstem Boise River sites, trends were also evaluated for measured concentrations during the 15-year period from 2009 to 2023. Many watershed improvements occurred during this more recent period, and the resulting water-quality trends were not always well-represented by a monotonic trend analysis over 30 years. Trends were not evaluated for 2009–2023 at tributary sites because of inadequate data during the first half of that period.

Temporal trends in concentrations were evaluated using a Mann-Kendall Theil-Sen approach following Helsel and others (2020). This approach accommodated several constraints of our data, including (1) left-censored concentrations for some sites and constituents, especially at the Boise River at Eckert (table 3); (2) irregular sampling frequencies, with some sites and constituents having periods of frequent sampling (in other words, multiple samples per season) intermixed with gaps of multiple years; (3) correlations between discharge and the concentrations of some constituents; and (4) discharge data limited to discrete measurements on sampling dates for some sites during some periods. The potential effect of discharge on concentrations was removed (Helsel and others, 2020) by fitting a shrinkage cubic regression spline to the relation between the natural log (ln) of discharges and the natural log of measured concentrations to get the ln(concentration) residuals using the centrend function in the NADA2 package (Julian and Helsel, 2021). To minimize the potential weighting of the higher-frequency sampling periods, we then simplified the dataset to the annual medians of the ln(concentration) residuals (Helsel and others, 2020). Years with no samples were left as gaps. To evaluate trends specifically within the irrigation and non-irrigation seasons, we also created datasets of the median ln(concentration) residuals for each season of each water year. Trends over time in the annual median ln(concentration) residuals were then evaluated using the non-parametric Mann-Kendall trend test and Theil-Sen slope estimator (Theil, 1950; Sen, 1968). The Mann-Kendall and Theil-Sen tests were done using the rkt function in the rkt package in R (Marchetto, 2024), which allows for gaps in time-series data. For a subset of sites and constituents, linear (untransformed) concentrations were used instead of ln(concentrations) for better model fit; these were TP at Fifteenmile Creek and Boise River near Parma, OP at Fifteenmile Creek, and TN at Boise River at Eckert. For trends in periphyton chl-a concentrations, a modified approach was used because periphyton chl-a was sampled from rock surfaces, and concentrations were assumed to be independent of discharge (although changes in water depth or velocity could affect periphyton growth by altering light penetration). The potential effect of discharge on concentrations was therefore not removed, and instead trends in periphyton chl-a concentrations were evaluated on the measured concentrations themselves. Trends were not evaluated for several constituents at Boise River at Eckert (TP, OP, nitrate plus nitrite, E. coli) because high rates of left-censored data at that site (greater than [>] 20 percent; table 3) yielded unreliable trend results.

Temporal trends in measured loads were evaluated using a similar approach, with some modifications. Because loads are determined by discharge, the potential effect of discharge on loads was not removed. Instead, trends were evaluated on the annual or seasonal medians of the natural logs of instantaneous loads using the Mann-Kendall and Akritas-Theil-Sen tests using the ATS function in the NADA2 package in R (Julian and Helsel, 2021). In cases where the annual median load was calculated using a left-censored concentration value, the load was considered to be left-censored and was input as such in the trend test. Trends in loads may be driven by changes in either concentration or discharge, or both. To help differentiate these drivers, we also evaluated temporal trends in discharge values on days when samples were collected. Trends in discharge were evaluated using the same approach as trends in loads. Results of this analysis are meant to help understand trends in loads, but because discharge data were limited to sampling dates, the reported trends in discharge may not represent actual changes in discharge over time.

The uncertainty of trend results is presented along a spectrum of likelihood rather than a single binary threshold (Hirsch and others, 2015; Wasserstein and others, 2019; Day, 2021). Trend results with p-values less than (<) 0.05 were considered “highly likely”; p-values 0.05 to <0.10 were considered “likely”; and p-values ≥0.10 were considered “uncertain.” For models using ln(concentration) and ln(load), the percent change in concentrations or loads per year was calculated using equation 1, and the percent change over the entire trend analysis period was calculated using equation 2:

Slopes were converted to linear units (for example, milligrams per liter per year) using equation 3 (Helsel and others, 2020):

For models using untransformed concentrations, the percent change in concentration per year was estimated using equation 4:

Five-year means were used to reduce the potential effect of an unusually high or low year on the percent change calculation.

Recent Spatial and Seasonal Patterns

Measured concentrations and corresponding discharges used in the evaluation of recent (2019–2023) spatial and seasonal patterns are plotted in figures 2 and 3, and shown relative to TMDL thresholds in figure 4. Concentration summaries in the following section are based on these measured values. Daily concentrations during the 2019–2023 period, estimated using the composite model, are shown relative to measured concentrations in figure 5. Daily estimated loads relative to measured loads are shown in figure 6.

Lower Boise River

The 2019–23 period was characterized by a range of flow conditions in the lower Boise River, including two high-flow years (2019 and 2023) and three low-flow years (2020–22; figs. 2, 5). Discharge was generally greatest in late spring–early summer, driven by snowmelt but moderated by upstream reservoirs. Mean discharges during the non-irrigation season were similar from Eckert to Middleton but increased 1.8-fold from Middleton to Parma with inputs from tributaries (table 4). In contrast, during the irrigation season, mean discharge was reduced by nearly one-half (0.52-fold) between Eckert and Middleton because of diversions, then increased 1.4-fold from Middleton to Parma with irrigation returns and tributary inputs (table 4). Overall, mean discharge decreased 25 percent (0.75-fold) from Eckert to Parma during the irrigation season.

Table 4.    

Mean estimated discharge and constituent loads in Boise River tributaries and Sand Hollow Creek, Idaho during the irrigation and non-irrigation seasons, water years 2019–23.

[Discharges are based on daily values. Loads are based on daily estimates for the entire period using the composite method. The percentage of total load (percent of total) is based on the sum of the means from the monitored Boise River tributaries and the Boise River at Middleton. Abbreviations: Cr, Creek; ft³/s, cubic feet per second; I, irrigation; NI, non-irrigation; %, percent; kg, kilograms; MPN, most probable number; x, not included in percent of total calculation; --, not evaluated]

Table 4.    Mean estimated discharge and constituent loads in Boise River tributaries and Sand Hollow Creek, Idaho during the irrigation and non-irrigation seasons, water years 2019–23.

Constituent	Season	Unit	Site
			Boise River at Eckert	Boise River at Middleton	Fifteenmile Creek	Mill Slough		Mason Creek	Indian Creek	Conway Gulch	Dixie Drain	Boise River near Parma	Sand Hollow
Discharge	I	ft3/s	1,720	897	93.3	100	127		7.90	32.7	178	1295	126
		% of total	x	62.5	6.50	6.98	8.82		0.55	2.27	12.4	x	x
	NI	ft3/s	541	540	40.6	53.1	54.3		96.3	15.7	103	976	68.9
		% of total	x	59.8	4.50	5.89	6.01		10.7	1.74	11.4	x	x
Suspended sediment	I	kg/day	22,540	26,511	17,055	4,185	38,279		1,110	17,492	11,524	120,802	50,135
		% of total	x	22.8	14.7	3.60	33.0		0.96	15.1	9.92	x	x
	NI	kg/day	7,012	16,010	3,074	2,898	7,132		8,754	2,091	7,735	60,207	8,701
		% of total	x	33.6	6.45	6.08	15.0		18.4	4.38	16.2	x	x
Total phosphorus	I	kg/day	98.2	115	45.0	37.6	85.8		3.60	25.0	50.9	494	99.0
		% of total	x	31.6	12.4	10.4	23.7		0.99	6.89	14.1	x	x
	NI	kg/day	22.3	81.6	25.7	23.3	27.8		47.7	6.93	32.1	385	36.8
		% of total	x	33.3	10.5	9.52	11.3		19.5	2.83	13.1	x	x
Orthophosphate as phosphorus	I	kg/day	30.3	31.1	28.8	27.3	48.8		2.11	9.64	44.2	239.0	45.2
		% of total	x	16.2	15.0	14.2	25.4		1.10	5.02	23.0	x	x
	NI	kg/day	13.1	43.9	20.9	19.4	22.3		40.0	5.65	25.4	265.0	25.6
		% of total	x	24.7	11.8	10.9	12.6		22.5	3.18	14.3	x	x
Total nitrogen	I	kg/day	598	1,097	--	--	--		--	--	--	4,576	--
	NI	kg/day	200	1,990	--	--	--		--	--	--	8,302	--
Nitrate plus nitrite as nitrogen	I	kg/day	288	593	--	--	--		--	--	--	3,179	--
	NI	kg/day	78.9	1,598	--	--	--		--	--	--	7,221	--
E. coli bacteria	I	MPN/day, in billions	362	1,610	--	--	--		--	--	--	9,980	--
	NI	MPN/day, in billions	97.9	364	--	--	--		--	--	--	2,220	--

Concentrations of most of the monitored constituents increased considerably along the mainstem Boise River from Eckert to Middleton to Parma during the irrigation and non-irrigation seasons in 2019–2023. The measured concentrations of TP, OP, TN, nitrate plus nitrite, and E. coli all increased at least tenfold from Eckert to Parma (fig. 4; tables 5, 6). The only constituent that did not follow this spatial trend was periphyton chl-a, which occurred in the greatest concentrations at Middleton based on the limited data available (fig. 4). At Parma, the highest measured concentrations of SSC, E. coli, and to a lesser extent, TP, occurred in late spring–early summer, following the seasonal pattern of discharge (fig. 5). This pattern was less apparent at Eckert and Middleton. The mean proportion of TP as OP was the same during the irrigation and non-irrigation seasons at Eckert, but it was lower during the irrigation season at Middleton and Parma (fig. 5; tables 5, 6), likely associated with higher flows diluting OP contributions from groundwater, wastewater treatment plants, and other sources. Concentrations of OP, TN, and nitrate plus nitrite followed a seasonal pattern opposite that of SSC, E. coli, and TP, with lower concentrations during late spring to mid-summer (fig. 5). The percentage of TN in the form nitrate plus nitrite decreased during the irrigation season at Middleton and Parma (fig. 5; tables 5, 6), consistent with the understanding that nitrate plus nitrite is groundwater sourced and is diluted by higher flows.

Table 5.    

Mean measured and composite model-estimated constituent concentrations in the Boise River and Sand Hollow Creek, Idaho, water years 2019–23.

[Abbreviations: NA, unable to estimate because of low detection frequency; --, not evaluated; SSC, suspended-sediment concentration; TP, total phosphorus; OP, orthophosphate as phosphorus; TN, total nitrogen; NO₃+NO₂, nitrate plus nitrite as nitrogen; E. coli, Escherichia coli bacteria; chl-a, chlorophyll a; mg/L, milligrams per liter; MPN, most probable number per 100 milliliters; mg/m², milligrams per square meter; %, percent; I, irrigation; NI, non-irrigation]

Table 5.    Mean measured and composite model-estimated constituent concentrations in the Boise River and Sand Hollow Creek, Idaho, water years 2019–23.

Constituent	Units	Season	Site
			Boise River at Eckert		Boise River at Middleton		Boise River near Parma		Sand Hollow Creek
			Measured	Modeledb	Measured	Measured	Measured	Modeled	Measured	Modeled
SSC	mg/L	I	4.38	4.30	12.7	9.60	34.8	34.4	158	161
SSC	mg/L	NI	3.80a	3.15	6.80	5.86	24.8	18.6	70.5	44.3
TP	mg/L	I	0.0132a	0.020	0.082	0.077	0.188	0.188	0.324	0.320
TP	mg/L	NI	0.0144a	0.014	0.084	0.079	0.169	0.172	0.245	0.213
OP	mg/L	I	0.00561a	0.0080	0.024	0.025	0.109	0.109	0.147	0.146
OP	mg/L	NI	0.00622a	0.0090	0.041	0.044	0.129	0.130	0.152	0.153
TP as OP	%	I	43	40	30	32	58	58	46	46
TP as OP	%	NI	43	64	49	56	76	76	62	72
TN	mg/L	I	0.120	0.133	0.917	0.968	1.93	1.96	--	--
TN	mg/L	NI	0.147	0.150	2.03	2.34	3.75	4.01	--	--
NO3+NO2	mg/L	I	0.0311a	0.061	0.506	0.575	1.42	1.46	--	--
NO3+NO2	mg/L	NI	NA	0.046	1.55	1.95	3.09	3.54	--	--
E.coli	MPN	I	7.30a	11.0	66.9	91.1	255	276	--	--
E.coli	MPN	NI	7.34a	8.14	41.0	27.5	136	94.9	--	--
Periphyton chl-a	mg/m2	I	NA	--	114a	--	113a	--	--	--
Periphyton chl-a	mg/m2	NI	98.9a	--	287a	--	78.0a	--	--	--

^a

Estimated using regression on order statistics to account for left-censored values.

^b

Modeled with left-censored values set to the reporting limit.

Table 6.    

Mean measured and composite model-estimated constituent concentrations in select tributaries to the Boise River, Idaho, water years 2019–23.

[Abbreviations: SSC, suspended-sediment concentration; TP, total phosphorus; OP, orthophosphate as phosphorus; mg/L, milligrams per liter; %, percent; I, irrigation; NI, non-irrigation]

Table 6.    Mean measured and composite model-estimated constituent concentrations in select tributaries to the Boise River, Idaho, water years 2019–23.

Constituent	Units	Season	Site
			Fifteenmile Creek		Mill Slough		Mason Creek		Indian Creek		Dixie Slough		Conway Gulch
			Measured	Modeled	Measured	Modeled	Measured	Modeled	Measured	Modeled	Measured	Modeled	Measured	Modeled
SSC	mg/L	I	69.3	66.9	21.3	16.2	111	115	55.0	51.9	22.2	25.4	218	226
SSC	mg/L	NI	27.0	20.5	24.8	22.9	50.7	47.6	39.6	32.6	27.1	30.9	59.5	42.5
TP	mg/L	I	0.200	0.199	0.152	0.155	0.274	0.275	0.194	0.183	0.108	0.114	0.326	0.325
TP	mg/L	NI	0.256	0.285	0.186	0.184	0.217	0.207	0.192	0.209	0.135	0.128	0.195	0.178
OP	mg/L	I	0.127	0.134	0.105	0.112	0.153	0.159	0.128	0.114	0.073	0.101	0.125	0.123
OP	mg/L	NI	0.204	0.245	0.147	0.152	0.170	0.171	0.153	0.171	0.071	0.104	0.151	0.158
TP as OP	%	I	64	67	69	72	56	58	66	62	67	89	38	38
TP as OP	%	NI	80	86	79	83	78	83	80	82	53	81	78	89

Measured concentrations above TMDL targets were frequent for some constituents at lower Boise River sites during 2019–2023 (fig. 4). Concentrations of TP were greater than the TMDL target of 0.07 mg/L in 33 percent of samples from Boise River at Middleton and 99 percent of samples at Boise River near Parma during the irrigation (and TMDL) season. Periphyton chl-a concentrations were greater than the mean monthly TMDL target of 150 mg/m² in 75 percent of year-round samples from Boise River at Middleton and in 38 percent of samples from Boise River near Parma. Among the months represented by recent samples at Boise River at Middleton, some of the highest concentrations occurred during February (fig. 7). E. coli concentrations were greater than the TMDL target of 406 MPN/100 mL in 14 percent of samples at Boise River near Parma; and were greater than the TMDL target of 126 MPN/100 mL in 41 percent of samples at Boise River near Parma and in 14 percent of samples at Boise River at Middleton. Concentrations of SSC above the TMDL targets were infrequent, with 3.4 percent of samples from Middleton and 6.6 percent of samples from Parma greater than the 50 mg/L threshold, and no samples greater than the 80 mg/L threshold. Notably, many of these TMDL targets (for example, periphyton chl-a; the E. coli threshold of 126 MPN/100 mL; and SSC) are intended to be evaluated using means of multiple samples over a given time period. Therefore, a single sample above a target concentration does not necessarily indicate a TMDL exceedance.

Despite a mean loss of discharge along the Boise River from Eckert to Parma during the irrigation seasons, and only a minor increase during the non-irrigation seasons, mean loads of SSC, TP, OP, TN, nitrate plus nitrite, and E. coli all increased considerably (5.0-fold to 91-fold; table 4; fig. 6). For all these constituents, most of the load increases occurred between Middleton and Parma. The seasonal patterns of loads generally reflected those of concentrations, with the highest loads of SSC, TP, and E. coli occurring during late spring–early summer, and the highest loads of TN and nitrate plus nitrite occurring during autumn and winter (fig. 6). The seasonal pattern of OP loads was less clear, with high values in both winter and early summer, but consistently low values in late summer–early autumn, potentially related to increased uptake of OP by algae during the growing season (Hoelscher and Myers, 2003; Wood and Etheridge, 2011).

Fifteenmile Creek

Measured concentrations of SSC, TP, and OP in Fifteenmile Creek were generally greater than those in the Boise River at Middleton but similar to those in most other tributaries (tables 5, 6; fig. 4). Suspended-sediment concentrations followed a seasonal pattern and were greatest during the irrigation season (mean 2.57-fold greater than in the non-irrigation season), during which time 89 percent of samples were greater than the site-specific TMDL threshold of 23 mg/L (figs. 4 and 8; Idaho Department of Environmental Quality, 2015a). Total phosphorus concentrations consistently exceeded the TMDL threshold of 0.07 mg/L and, apart from two samples with high concentrations during the non-irrigation season, did not show a clear seasonal pattern (fig. 8). Orthophosphate concentrations and the percentage of TP as OP were lowest during late spring–early summer (fig. 8) as high flows diluted OP from sources such as groundwater and wastewater treatment plants.

Discharges from Fifteenmile Creek averaged 6.50 and 4.50 percent of the total measured inflow to the Boise River during the irrigation and non-irrigation seasons, respectively (table 4). However, the relative contributions of loads of TP and OP from Fifteenmile Creek to the Boise River were disproportionately greater. Of the total measured loads, Fifteenmile Creek accounted for 12.4 percent of the TP and 15.0 percent of the OP in the irrigation season, and 10.5 percent of the TP and 11.8 percent of the OP in the non-irrigation season (table 4; fig. 9). For SSC, Fifteenmile Creek contributed 6.45 percent of the total measured inflow load during non-irrigation season, comparable to its discharge contribution, but 14.7 percent of the total inflow SSC during irrigation season (fig. 9).

Fivemile Creek and Tenmile Creek are the two primary tributaries to Fifteenmile Creek. During the non-irrigation season, Fivemile Creek contributed slightly more discharge than Tenmile Creek (55 percent versus 45 percent of the total, respectively) but considerably more SSC (76 percent), TP (74 percent), and OP (71 percent; fig. 10), potentially associated in part with the discharge of the City of Meridian’s wastewater effluent to Fivemile Creek. During the irrigation season, mean discharge increased 1.8-fold at both sites. At Fivemile Creek, loads of SSC, TP, and OP also increased during the irrigation season, but disproportionately less than the increase in discharge (1.03–1.5-fold increases over the non-irrigation season loads) indicating a dilution effect by higher flows in the irrigation season. The mean percentage of TP as OP was nearly unchanged (60 percent) in Fivemile Creek between the irrigation and non-irrigation seasons. In contrast, at Tenmile Creek, loads of SSC increased 4.0-fold, loads of TP increased 2.9-fold, and the percentage of TP as OP decreased to 53 percent during the irrigation season. These results suggest greater mobilization of particulates during the irrigation season in the Tenmile Creek watershed compared to Fivemile Creek.

Temporal Trends

Trends in concentrations and loads were evaluated using samples collected over a 30-year period (1994–2023) and, at mainstem sites, a more recent 15-year period (2009–2023; figs. 11, 12). Trends in suspended-sediment concentrations were highly likely decreasing during the irrigation season at Mill Slough and Dixie Slough over the 30-year period, and at Boise River near Parma over the recent 15-year period (table 7; fig. 13). During the non-irrigation season, trends in SSC were uncertain except at Fifteenmile Creek, where they were highly likely decreasing over the 30-year period. Elsewhere, trends in SSC were uncertain. The greatest reduction in SSC was at Dixie Slough, where concentrations during the irrigation season decreased 73 percent over the 30-year period, or 1.0 mg/L/year, on average (table 7; fig. 13). Loads of suspended sediment showed decreasing trends at most sites, especially during the irrigation season (table 8). At sites where trends in suspended-sediment concentrations were uncertain (Boise River at Middleton, Mason Creek, Indian Creek, Conway Gulch), the decreasing trends in suspended-sediment loads were likely driven by decreasing trends in discharge at the time of sampling (table 9).

Total phosphorus concentrations during the non-irrigation season were highly likely decreasing at all sites except Mason Creek (fig. 13; table 7). During the irrigation season, TP concentrations were highly likely decreasing at all sites except Boise River at Middleton, Indian Creek, and Conway Gulch. Over the 30-year period, TP concentrations at each site decreased 23–83 percent in 1 or more seasons, suggesting that the TP TMDL and related point and nonpoint source reduction efforts over the study period have been effective. Loads of TP were also likely or highly likely decreasing at nearly all sites and seasons (table 8), reflecting both decreasing concentrations (table 7) and discharges at the time of sample collection (table 9). King and Yoder (2025) estimated 60 percent decreases in TP concentrations and loads at Boise River near Parma during 2003–2021. Our estimates, using different methods and time periods, generally support their findings, with TP concentrations decreasing 44–48 percent and loads decreasing 57–70 percent.

Like TP, concentrations of OP were likely or highly likely decreasing at all sites during one or more seasons (fig. 13; table 7). The largest decreases in OP concentrations were observed in the Boise River at Middleton (77–92 percent decreases, depending on season and evaluation period), Fifteenmile Creek (51–58 percent decreases), and Boise River near Parma (54–68 percent decreases). Similarly, loads of OP were likely or highly likely decreasing at all sites during one or more seasons (table 8).

Trends in TN, nitrate plus nitrite, E.coli, and periphyton chl-a were only evaluated at mainstem lower Boise River sites because tributary sites lacked adequate data for trend analysis. In contrast to TP and OP, concentrations of TN were highly likely increasing at Middleton (68 percent increase, or 1.68-fold) and at Parma (45 percent increase, or 1.45-fold) during 1994–2023 (fig. 13; table 7). Loads of TN at Middleton were also highly likely increasing during 1994–2023 (table 8). In contrast, loads of TN were highly likely decreasing at Eckert during 1994–2023, driven by decreasing discharges during sample collections (table 9).

Nitrate plus nitrite concentrations were uncertain at Middleton but likely increasing at Parma during the irrigation season (21 percent increase, or 1.21-fold) and highly likely increasing annually (44 percent increase, or 1.44-fold) during 1994–2023 (fig. 13; table 7). Increasing trends in nitrate plus nitrite concentrations have also been observed in surface water and groundwater elsewhere in the Snake River Basin (Frans and others, 2012; Lentz and others, 2018; Bjorneberg and others, 2020; Naymik and others, 2023) and have been attributed to shifts in agricultural practices including increases in dairy herds, animal feed crops, and use of chemical and manure fertilizer (Lentz and others, 2018). Trends in loads of nitrate plus nitrite were generally uncertain, but likely decreasing during the irrigation season at Parma during 2009–2023 (table 8).

Trends in E. coli concentrations were uncertain except during the 2009–2023 irrigation season at Eckert, where they were likely decreasing, and the 2009–2023 non-irrigation season at Parma, where they were highly likely increasing (fig. 13; table 7). Trends in loads of E. coli were uncertain.

Periphyton chl-a concentrations were highly likely increasing at Eckert during the October-November period (1994–2023; fig. 13; table 7). At Middleton and Parma, trends in periphyton chl-a concentrations were uncertain. Loads were not estimated for periphyton chl-a.