Purpose and Scope

The primary purpose of this study is to quantify fine sediment infiltration into Chinook salmon spawning gravels in the Sauk River during the incubation and pre-emergence periods. The Sauk River was selected because of its relatively pristine watershed conditions as an unregulated, forested mountain river that has robust populations of Chinook salmon along Sauk River main stem. The Sauk River has naturally high fine sediment loads because of basin geology and headwaters draining Glacier Peak, a heavily glaciated stratovolcano. A secondary purpose of the study is to determine if fine sediment infiltration of salmon redds can be qualitatively assessed by more easily measured factors, such as subsurface sediment samples within the channel at salmon spawning sites or on nearby exposed bars.

Study Area Description

The study was conducted in the main stem Sauk River (drainage area 1,896 square kilometers [km²]), which is free flowing and drains a relatively undeveloped landscape in the northern Cascade Range of Washington State (fig. 1). Principal tributaries are the White Chuck River (drainage area 222 km²) and the Suiattle River (drainage area 890 km²), which both drain Glacier Peak.

Basin geology is characterized by a north-south-trending fault that separates different geologic units (Tabor and others, 2002). The western half of the basin, which comprises the Sauk River main stem and downstream portions of the White Chuck and Suiattle Rivers, is composed of a diverse combination of sedimentary, intrusive, volcanic, and metasedimentary and metavolcanic units. The eastern half of the basin, which underlies most the White Chuck and Suiattle River Basins, is composed of intrusive and high-grade metamorphic units. The heavily glaciated Glacier Peak is composed of younger volcanic rocks relative to the surrounding area and lahars (volcanic debris flows). The eastern flank of Glacier Peak, feeding the headwaters of the Suiattle River, is particularly geomorphically active (Slaughter, 2004) and is a dominant source of sediment transported through the Suiattle River.

The Sauk River valley upstream from the White Chuck River confluence is unconsolidated glacial drift. Downstream from the White Chuck River, the river valley is mostly unconsolidated alluvium, landslide deposits, and glacial drift. The Sauk River is a coarse-grained, gravel-bedded alluvial mountain river that is single-thread meandering with pool-riffle morphology (Keck and others, 2022) with multi-threaded reaches that correspond to widened valley sections and extensive gravel bars downstream from the Suiattle River confluence, attributed in part to the high sediment loads delivered by the Suiattle River. The Suiattle River, which represents approximately 47 percent of the Sauk River Basin area, accounts for approximately 80 percent of the suspended-sediment load for Sauk River (Jaeger and others, 2017). Suspended-sediment load is approximately equal between the White Chuck River Basin and the upper Sauk River, despite headwaters of the White Chuck River draining Glacier Peak (Jaeger and others, 2017).

Watershed land cover is forested and primarily under U.S. Forest Service jurisdiction, with private ownership in the lower river. The Sauk River Basin climate is temperate, with cold, wet winters and warm, dry summers. Mean annual precipitation ranges from approximately 1,500 millimeters per year in river valleys to 3,300 millimeters per year at higher elevation ridges (Livneh and others, 2013). The streamflow regime has peak flows occurring during the autumn–winter (November–February) precipitation, additional high flows during spring snowmelt (April–June), and low summertime-based flows (July–September) that include glacial meltwater.

The three study sites are distributed along the main stem river to capture (1) the high elevation, mountain river conditions of the upper Sauk River upstream from the White Chuck River confluence (river kilometer [RKM] 51), (2) the middle alluvial reach upstream from the Suiattle River confluence (RKM 31), which does not have the high sediment loads from that river, and (3) the sediment rich lower river that has higher contributions of sand and finer sediments from the Suiattle River (RKM 15) relative to the Sauk River main stem. All three study sites have an adjacent gravel bar and correspond to the top of a riffle where Chinook salmon are known to spawn (Brett Barkdull, Washington Department of Fish and Wildlife, oral communication, 2018). The upper and middle Sauk River study sites are at the downstream end of gravel bars in a single-threaded reach of the river. The lower Sauk River study site is at the upstream end of a gravel bar of a side channel from the main stem. The channels are gravel and cobble bedded in a matrix of coarse sand with lower Sauk River characterized by a sandier matrix (fig. 1).

Installation, Recovery, and Data Collection

To avoid disturbance of spawning activity, which typically begins in autumn, installation of the infiltration boxes, infiltration bags, and dissolved oxygen sensors occurred in August 2018, 2019, and 2020 when streamflow was minimal (table 2). The August timing of installation was constrained by permitting guidelines that require activities in the channel to occur outside of important life history windows for Chinook salmon. At the time of installation, the geographic location of all data collection equipment buried in the channel bed were mapped either by recording the distances of two intersecting field tapes from two different locations along the bank or were recorded using real-time kinematic positioning.

Recovery of monitoring features occurred at the three study sites when water temperatures reached 900–980 accumulated thermal units (ATUs; measured in degrees Celsius), which approximately corresponds to the period of at least 50 percent emergence of Chinook salmon alevins from the subsurface (Geist and others, 2006; Beacham and Murray, 1989). This pre-emergence period includes the time to hatch from embryo to alevin, which occurs at approximately 525 ATUs (Beacham and Murray, 1989), after which, the alevins remain in the subsurface for further development. Accumulated thermal units are calculated by summing daily mean water temperature forward in time from the estimated mid-point of the spawning period at each study site. Dates demarcating the mid-point of the spawning season were informed by Washington Department of Fish and Wildlife expertise of Chinook salmon spawning in the Sauk River (Brett Barkdull, Washington Department of Fish and Wildlife, written commun., 2018–20). For study year 1, spawning mid-point dates were September 15 for upper Sauk River study site, October 1 for middle Sauk River study site, and October 15 for lower Sauk River study site. For study years 2 and 3, spawning mid-point dates were September 20 for upper Sauk River study site, September 25 for middle Sauk River study site, and October 1 for lower Sauk River study site. Later spawning dates for the downstream study sites is driven by higher downstream water temperatures such that spawning activity is delayed until water temperatures have cooled sufficiently (Baisez and others, 2011). The period between installation and the assigned mid-point spawning period date was not included in the ATU calculation because it would have resulted in an earlier recovery date that did not correspond with the approximate emergence period. Accumulated thermal units for upper Sauk River study site were based on daily temperature summarized from 15-minute temperature data at streamflow station Sauk River AB White Chuck River NR Darrington, WA (12186000, U.S. Geological Survey, 2022a), which was less than 1 kilometer (km) downstream from upper Sauk River study site. Accumulated thermal units for the middle and lower Sauk River study sites were based on daily temperature summarized from 15-minute temperature data at streamflow station Sauk River near Sauk, WA (12189500, U.S. Geological Survey, 2022b), which was approximately 22 and 6 km, downstream from the middle and lower Sauk River study sites, respectively. Recovery was generally in late February at upper Sauk River study site and early March to mid-April at middle and lower Sauk River study sites and occurred during winter low flow conditions.

Sediment Infiltration Boxes

Fine sediment infiltration during the incubation/pre-emergence period was measured using modified Whitlock-Vibert (WV) boxes, which were buried in constructed artificial redds in known Chinook salmon spawning locations. WV boxes were 13.5 × 6.0 × 8.7 (length × width × height) centimeters (cm) perforated polypropylene boxes used for fish egg planting in streams (Garrett and Bennett, 1996), but have been used for measuring fine sediment infiltration in streams (Wesche and others, 1989; Johnson and others, 2012). The top portion of the WV box included an egg tray with a protected egg chamber in which salmon embryos were placed for incubation (fig. 3). Alevins then drop down into the lower portion of the box in a protected nursery chamber, before exiting the box as fry during the emergence period. Use of the WV boxes followed methods similar to other studies (Johnson and others, 2012; Wesche and others, 1989) in which the egg tray was removed and the bottom of the box was lined with duct tape to prevent finer sediment exiting the box from the bottom. Each box contained a capped vial containing a passive integrated transponder (PIT tag, Biomark 23 mm) (fig. 1B–D). Leashed to the outside of the box was an approximately 20-cm long nylon string to which a second vial contained a second PIT tag; one to two additional strings with neon-colored floats (0.5-inch Worden Lil Corky) attached to the box to aid in retrieval (fig. 3C).

Each box was (1) pre-filled with sediment with a target total mass of 1050 grams and (2) approximately matched the grain size distribution of each study site with grain sizes less than 8 mm removed. The sediment grade scale that includes grain sizes between 8 and 4 mm correspond to fine gravel (Julien, 2010). Grain sizes 4.8 mm or less, which are on the small end of this grade class of fine gravel, have been included in fine sediment egg-to-fry survival curves (Jensen and others, 2009). Grain size distributions for each of the three study sites were derived from bulk sediment samples on exposed gravels bars at each study site conducted in the summer of study year 1, prior to the first installation. Grain size distribution specific to each of the three study study sites (hereafter, “pre-installed” grain size distributions) was used for all three study years.

WV boxes were buried in artificial redds that mimic Chinook salmon redds (Johnson and others, 2012; Buxton and others, 2015, Julien and Bergeron, 2006). Artificial redds were constructed by excavating approximately 30 cm into the riverbed and depositing displaced sediment directly downstream (fig. 4A). WV boxes were installed at the base of the excavated hole. The hole was then carefully backfilled with part of the displaced sediment placed downstream from the excavated hole with fines removed by the streamflow current to create the tailspill zone that is consistent with salmon redd construction (fig. 4B). Sediment immediately upstream from the filled hole containing the buried WV box was excavated to a depth of approximately 15 cm to create the pot zone (fig. 4B). In study year 1, a subset of WV boxes was installed at approximately 10 cm depth to evaluate if fine sediment infiltration was limited to the higher portion of the redd. In study years 2 and 3, all WV boxes were buried at 30 cm depth. WV boxes were installed in an approximately rectangular array within the study area, with spacing sufficient that installation and subsequent retrieval were not expected to affect adjacent study sites of buried monitoring features.

WV boxes were recovered by first identifying their approximate location using either triangulation with the taped distances or using the real-time kinematic positioning in combination with a handheld PIT tag detector. Some careful shoveling was usually necessary in combination with the PIT tag detector until the visual markers of the WV boxes were identified. A four-sided plywood shield was positioned immediately upstream from the exposed egg box to create a localized area of no flow around the WV box. A sealing plastic approximately 3.8-liter (1-gallon) bag (study year 1) or an 1,893 cubic centimeter (cm³) plastic lidded container (study years 2 and 3) was submerged immediately downstream from the exposed WV box. The WV box was then gently lifted, and the plastic bag or rigid container brought under the WV box to enclose the box. The bag or rigid container was then carefully brought to the surface and carried to the bank. Excess water was poured into additional labeled containers to be included in the analysis for that individual WV box and the bag or rigid container was labeled.

Sediment Infiltration Bags

There was concern that the extraction method of the WV box, which required excavation down to the WV box, may have inadvertently contributed additional fine sediment to the box that would not represent fine sediment infiltration that the study was intending to capture. Therefore, in study years 2 and 3 of the study, infiltration bags were installed at the study sites following methods described by Lisle and Eads (1991) and Anderson (2007) as an additional measure of fine sediment infiltration into the Chinook salmon spawning gravels (fig 4C–E). The extraction method of the infiltration bag eliminated the potential for additional fine sediment because the sediment column above the bag and extending to the surface was included in the sample. Infiltration bags were added for comparison to WV boxes.

Infiltration bags used in Anderson (2007) were installed in study year 2 (table 1). The approximate dimensions of the bags were 30.5-cm diameter and 46-cm depth. Additional bags were made following the design of Anderson (2007) to expand the sample size in study year 3 and replace unrecovered bags from study year 2 (table 1). Each infiltration bag had four stainless steel cables that were looped at the end extending from the top rim of the bag. PIT tags enclosed in capped vials were taped to the base and top of each of the four cables.

Infiltration bags were installed similarly to the WV boxes; however, a larger excavated area was required to accommodate the bags’ larger footprint. Excavated substrate was agitated in water to remove finer clast sizes and temporarily placed in a 19-liter (5-gallon) bucket. The collapsed bag was placed at the base of the pit at a depth of approximately 30 cm and held down with a bottomless 19-liter bucket with four stainless steel cables extending out (fig. 4C–E). The bottomless bucket and pit surrounding it was backfilled with substrate temporarily stored in the 19-liter buckets. The stainless steel cables extended through the backfilled substrate to lay on the surface of the riverbed.

Infiltration bags were recovered using the same location methods for locating the WV boxes until all four cables of an individual bag were located. Infiltration bags were extracted from the riverbed using a chain hoist (Kill Shot Tripod Game Hoist) positioned over the center of the bag (fig. 5). The four-sided plywood shield was placed immediately upstream from the hoist and infiltration bag location to create a localized area of no flow. Field assistants held down the plywood shield, which was buoyant when flow depths were higher than approximately 0.5 m. Water inside of the sediment infiltration bags was decanted into wide-mouth, 3.8-liter plastic jars; the substrate contents of the infiltration bags were poured into 19-liter lidded buckets, and the bags were placed in 19-liter lidded buckets to be included as part of the samples because of residual sediment adhering to the bags.

Subsurface Bulk Sediment Sampling

Grain size distribution of bulk sediment (subsurface) is more representative of substrate conditions for salmon during redd construction than surface sediments alone (Kondolf, 2000). Bulk sediment sampling occurred within the wetted channel at the three study sites in identified spawning areas. The spawning areas were co-located with the artificial redds that house monitoring features (for example, WV boxes, infiltration bags, and dissolved oxygen [DO] sensors). Sampling occurred during low flow conditions prior to spawning activity upon August installation of in-channel monitoring features and again during recovery of those features to capture potential changes in grain size distribution at study sites (table 1). Subsurface samples collected at the three study sites include (1) the submerged channel, which were referred to as “wet sediment samples,” and (2) adjacent gravel bars, which were referred to as “dry sediment samples.” Sediment sampling and analysis followed established methods for sampling in submerged and dry areas (Bunte and Abt, 2001; Peters and others, 2017). Subsurface samples in submerged areas were collected following methods described in Peters and others (2017) and used a four-sided plywood shield to create a local area of no-flow for the sediment sampling (fig. 6). Subsurface samples in the submerged in-channel locations and dry gravel bars both included removing surface grains at a depth that was approximately equivalent to the largest clast on the surface (Bunte and Abt, 2001). Submerged samples were collected at approximately 30 cm, which is consistent with the installation depths of the WV boxes and infiltration bags. Gravel bar samples were collected at depths that approached 50–100 cm. Submerged samples were approximately 27–45 kilograms (60–100 pounds, approximately equivalent to a single, filled 19-liter bucket); gravel bar samples were approximately 91–113 kilograms (200–250 pounds). Clast sizes larger than 16 mm were sieved at half-phi intervals (logarithm to the base two of the particle diameter) and weighed in the field (fig. 7). The remaining portions of the samples were dried, subsetted, and sieved at full-phi intervals to 0.063 mm by the USGS Cascades Volcano Observatory (CVO) sediment lab. Turbid water associated with the submerged samples, and some dry gravel bar samples, was included as part of the sample analyzed by CVO sediment lab.

A single water grab sample was collected using a wide-mouthed one-liter container in the water column defined by bottomless bucket (29.2-cm maximum depth) that corresponded to the water column immediately above each bulk subsurface sample area (fig. 6B). The grab sample was processed for suspended-sediment concentration (SSC) analysis to account for fine sediments within the riverbed that had gone into suspension during the sampling process following Peters and others (2017). The suspended-sediment volume was derived by taking the product of the SSC and the volume of the bucket. One additional grab sample was collected adjacent to the sampling zone at a similar depth that did not exceed approximately 29 cm at each of the upper, middle, and lower Sauk River study sites for a representative background level suspended-sediment concentration. The single, background SSC was differenced from each of the grab samples from the bulk subsurface samples. The resulting SSC and volume of the 5-gallon bucket was used to compute mass of suspended sediment associated with the bulk subsurface sampling.

Turbidity Monitoring

Turbidity monitoring occurred at Sauk River near Sauk, WA (12189500), and data are available from the USGS National Water Information System (NWIS, U.S. Geological Survey 2022b) for the three-year study period. Additionally, turbidity monitoring occurred at Sauk River AB White Chuck River NR Darrington, WA (12186000), during most of the three-year study (NWIS, U.S. Geological Survey 2022c). Turbidity monitoring at 12188380 was conducted by the Sauk-Suiattle Indian Tribe and the data were provided to USGS as a furnished record. Furnished records are provided by cooperating agencies for review to ensure the data have an accuracy comparable to data collected by USGS. Turbidity monitoring sites were both equipped with a Forest Technology Systems DTS-12 turbidimeter and operated according to standard USGS protocols (Wagner and others, 2006). Turbidity monitoring serves as a proxy for fine sediment in transit through the Sauk River, and qualitatively provides context on relative sediment availability on a seasonal and year-to-year timescale. Because the Sauk River near Sauk, WA gage is downstream from the Suiattle River, turbidity monitoring in the Suiattle River provides the contribution of this tributary allowing for deriving turbidity conditions in the Sauk River main stem that include the upper and middle Sauk River study sites by simple loading comparisons.

Instream Dissolved Oxygen Monitoring

DO sensors (Onset DO Data Loggers U26-001) were installed at study sites in study years 2 and 3 to monitor if and by how much dissolved oxygen concentration in the subsurface changed during the subsurface pre-emergence and presumed fine sediment infiltration period. DO concentration was collected hourly in units of milligrams per liter (mg/L). The DO sensors require replaceable caps that expire seven months after initialization; data collection of DO continues after the expiration date but may be considered unreliable. In this study, the pre-emergence period for upper Sauk River study site fit within the seven-month window of the DO sensors; the pre-emergence periods for middle and lower Sauk River study sites extending beyond March were outside of the seven-month expiration window.

Prior to field installation, DO sensors were calibrated in the lab according to the manufacturer’s specifications and following U.S. Geological Survey (2020), which included installation of a new sensor cap and calibrating of the sensor in saturated air and in a zero dissolved oxygen solution. Calibration and initialization occurred immediately prior to field deployment to maximize the seven-month period before the sensor caps expire. Initialized DO sensors were placed together in a 19-liter non-aerated bucket of water for approximately 18 hours prior to installation and following recovery to detect potential drift in individual sensors. A one-time field check of the DO sensors for each year of deployment was conducted using an In Situ Aqua Troll 600(C) multi-parameter water-quality sonde. The sonde was calibrated in the lab and attached to a peristaltic pump via a flow-through chamber and clean, flexible plastic tubing. A stainless-steel probe was inserted near the deployed DO sensors and interstitial waters pumped, at very low flow, to the flow-through chamber. Three field measurements of DO were collected at each DO sensor location after allowing the sonde to equilibrate. Field checks of subsurface DO occurred at depths between 0.25 and 0.5 m below the riverbed surface. Field checks were conducted on 25 February 2020 for the upper Sauk River study site and 19 April 2020 for the lower Sauk River study site for study year 2 and on 21 January 2021 for all three study sites in study year 3. No field checks were conducted in study year 2 at the middle Sauk River study site because DO sensors were not recovered at this study site due to excessive burial.

DO sensors were installed in the riverbed in a horizontal (prone) position in artificial redds (similar to the WV boxes and infiltration bags) at approximately 30-cm depth. DO sensors were housed in 50-cm long sections of slotted polyvinyl chloride (PVC) well screen capped at both ends and tethered to a 2.54-cm diameter, 0.46-meter-long rebar with a 0.5-meter-long stainless steel cable for an anchor, similar to Gendaszek and Sheibley (2021). The rebar piece was either (1) driven vertically into the riverbed when possible (middle and lower Sauk River study sites) or (2) shallowly buried horizontal in the subsurface and parallel to the flow direction to minimize being displaced by scour during high flows. Placement of the rebar anchor was immediately upstream from the DO sensor, based on the length of the tethering cable. PIT tags were attached to the cables (similar to the WV boxes and infiltration bags) to aid in recovery from the field.

Analysis

Instream Conditions During the Study—Discharge, Turbidity, and Water Temperature

Study year 1 had relatively small peak flows compared to study years 2 and 3, with no flows exceeding the 0.5 annual exceedance probability (fig. 8A). Study year 2 had four smaller flow events that did not exceed the 0.5 annual exceedance probability but had a flow exceeding the 0.2 annual exceedance probability occurred in February (fig. 8A). Substantial bank erosion, which occurred upstream from the middle Sauk River study site and the study site showed channel bed aggradation. Study year 3 had a moderately active peak-flow season, with one flow exceeding the 0.5 annual exceedance probability.

The turbidity time series reflects the typical seasonal patterns of elevated turbidity during the late summer associated with glacial melt from Glacier Peak and again during high flow events during autumn and winter seasons (fig. 8B; Jaeger and others, 2017). Late summer turbidity values in the Suiattle River are higher than the main stem Sauk River, revealing the dominating influence of Glacier Peak on glacial melt in the Suiattle River; turbidity is lower in the lower Sauk River study site because of mixing with the Sauk River main stem, which is less turbid although it does receive glacial melt from the White Chuck River. Turbidity values are still high in the late summer (July, August, and early-September) during low streamflow, which may create opportunity for fine sediment deposition and infiltration in-channel bed gravels during this pre-spawning period. However, high turbidity during late summer is largely characterized by glacial fines (silts and clays, <0.063 mm) that remain in suspension, even during low flows and are not expected to show fine sediment mobility and deposition during this low flow period.

Based on existing SSC-Turbidity and SSC-discharge relations for the Sauk River near Sauk, WA, we use turbidity and discharge to estimate suspended-sediment loads for the lower Sauk River study site for Water years 2019–2021 that approximately correspond to study years 1, 2, and 3 (table 3). Comparing these estimates to the five-year mean annual load, based on water years 2012–2016 (Jaeger and others, 2017), these estimated loads are approximately 0.36, 1.20, and 0.85 of the mean load for study years 1, 2, and 3, respectively and approximately correspond to the relative activity of peak-flow hydrology over the three-year study period.

Water temperatures were similar at the Sauk River near Sauk, WA and Sauk River AB White Chuck River NR Darrington, WA (fig. 8C; table 4). Maximum water temperatures were warmest in study year 2 at both study sites (19.5 °C) and more than a degree cooler in study year 3 (18.4 °C). Minimum temperature was also warmer in study year 2 at both study sites compared to study year 1, although daily minimum water temperature Sauk River near Sauk, WA remained similar to that of study year 2 (2.2 °C, table 3). Minimum temperature was the same for study years 1 and 2 at the lower Sauk River study site (0.3 °C) but was warmer (1.4 °C) in study year 3. In addition, mean daily temperatures cooled to 10 °C earlier in study year 3 (September 28 in study year 3 compared to October 12 in study year 1 and October 11 in study year 2) at upper Sauk Rive study site, but remained relatively similar in timing at lower Sauk River study site (October in study year 3 compared to October 12 in study year 1 and October 11 in study year 2).

Sediment Infiltration into WV Boxes and Infiltration Bags

All WV boxes and infiltration bags recovered across all study sites for all three study years show substantial fine sediment infiltration into the boxes and sediment infiltration bags (fig. 9). The grain size distribution of WV boxes post-recovery was markedly finer than pre-filled grain size distributions with more than 0.1 fraction of grain size <8 mm compared to the pre-filled grain size distribution of these boxes (fig. 9). Visual observations of recovered WV boxes suggest that large void spaces of prefilled WV boxes were filled with smaller size gravel, sand, silt, and finer (fig. 10). Grain size distributions of Whitlock-Vibert, infiltration bags, and subsurface sediment samples can be found in Jaeger and others (2024).

Silt clast sizes (<0.063 mm) account for less than 0.01 fraction finer (less than one percent) of the sediment grains in the WV boxes at middle and upper Sauk River study sites, but account for 0.03– 0.05 fraction finer of grains (three to five percent) in WV boxes at lower Sauk River study site across the three study years (table 5). Fraction finer of grains when including sand or smaller (size classes <2 mm) range from 0.12 to 0.23 across the three study sites and years, with the highest fraction finer corresponding to lower Sauk River study site (fig. 11; table 5). Higher fraction finer in lower Sauk River study site agrees with increasing suspended-sediment yields in the downstream direction (Jaeger and others, 2017), including fine gravel grain sizes in (<4 mm or smaller) increased fraction finer that range from 0.14 to 0.28 across all study sites and years. Fraction finer of fine sediment was higher for study years 2 and 3 at lower and upper Sauk River study sites, whether considering classes of sand or smaller (<2 mm) or classes of fine gravel or smaller (<4 mm) (table 5). Fraction finer was also higher at middle Sauk River for study year 2 (0.22 and 0.28 in study year 2 compared to 0.13 and 0.16 in study year 1 for <2 and <4 mm classes, respectively). However, average values for middle Sauk River study site are based on only three recovered WV boxes compared to an average from nine to 13 boxes recovered from upper and lower Sauk River study sites (table 1). Fraction finer values from middle Sauk River study site may be biased toward higher fraction finer because of the substantial excavation that was required to recover the WV boxes, which may have erroneously introduced additional fine-grained sediment into the WV box.

Comparisons in grain size distributions between infiltration bags and WV do not yield consistent differences for study sites or the two study years that infiltration bags were used and there is substantial variability in the grains size distribution of the infiltration bags and WV boxes at certain study sites and years (figs. 10–11). Fraction finer values for small clast sizes are similar between WV boxes and infiltration bags for upper Sauk River study site for study years 2 and 3, when both sediment infiltration features were installed. Fraction finer values are also similar at lower Sauk River study site for study year 3, but infiltration bags have higher fraction finer values for a representative small grain size in study year 2 relative to WV boxes. Additionally, the variability in fraction finer is wider across the 4 infiltration bags retrieved in study years 2 and 3 at the lower Sauk River study site compared to the WV boxes (very fine sand) (fig. 11; table 5). Limited recovery of boxes and bags at middle Sauk River study site for study years 2 and 3 limits comparison between WV boxes and infiltration bags at this study site. Additionally, the grain size distributions for the infiltration bags are based on a small sample size, which is approximately one-third that of the WV boxes. Therefore, given the variability in grain size distributions among bags within a single study site, grains size distributions should likely be interpreted as rough approximations with high uncertainty.

During the study year 2 recovery of WV boxes and infiltration bags at upper Sauk River study site in late February 2020, salmon alevin that had migrated into some WV boxes at this study site post installation were identified. The presence of alevin in the WV boxes indicates that despite high fine sediments at this study site, salmon were living in the gravels. Live salmon alevins were also in substrate captured by the infiltration bags at upper Sauk River study site during the extraction process from the riverbed.

Grain size distributions were derived for all bulk subsurface samples (fig. 12). We use the same representative grain sizes ranging between 0.063 and 4 mm for consistent comparison with the WV boxes and infiltration bags and for identification of potential differences in grain size distribution between (1) in-channel wet samples versus dry gravel bar samples, (2) seasons within a year (for example, late summer baseflow conditions versus late winter/early spring conditions), (3) across study years, and (4) across study sites (figs. 13–14).

There seems to be some differences between wet and dry samples depending on the season and study site. We had expected that wet in-channel bulk subsurface samples would be finer than subsurface samples from dry gravel bars adjacent to the wetted channel. This expectation is because the sampling area, constrained by the diameter of a bottomless 19-liter bucket in the in-channel samples, could limit extracting larger size clasts compared to the dry gravel bar samples, which were not similarly constrained. Conversely, there is potential that wet samples may be coarser than dry gravel bar samples if fine sediments were lost in the in-channel sampling process, despite the involved field sampling set up (fig. 6) to mitigate this bias. Wet samples were not consistently finer for all study sites and at all seasons as had been expected. Wet samples were notably finer than dry samples at lower Sauk River study site, and most markedly so during the late summer installation period in study years 2 and 3 (figs. 12–13). This late summer period corresponds with low flows and high turbidity from glacial melt. In contrast to lower Sauk River study site, there does not seem to be a meaningful difference between wet and dry samples at upper Sauk River study site (figs. 12–13; table 5). Wet and dry samples at middle Sauk River study site were similar throughout study years 1 and 2 installation, but the grain size distributions for the wet samples were noticeably finer compared to the dry observations in the study year 2 spring recovery and study year 3 installation periods (figs. 12–13; table 5). Study year 2 had more occurrences of larger magnitude flows and middle Sauk River study site showed substantial channel change and aggradation during this period, which likely is the cause of increased fining at this study site. There were no year 3 recovery period dry samples to compare with recovery period wet samples. Despite careful sampling, field methods for dry and wet bulk subsurface samples are both coarse and likely include measurable error. Nevertheless, the samples are an approximations of grain size distributions at the study sites. Further, we interpret finer grains size distributions in lower and middle Sauk River study site to reflect geomorphic differences between channel and adjacent gravel bar rather than differences in sampling methods between the channel and adjacent gravel bar.

Seasonal differences between grain size differences were apparent at lower Sauk River study site for the wet and dry samples (figs. 12–13; table 5). Wet and dry samples both had higher values of fraction finer for representative small grains during the installation (late summer) periods compared to the spring recovery periods following the winter storm season and prior to the spring runoff season for study years 1 and 2 (figs. 12–13; table 5). There did not seem to be seasonal differences in grain size distributions or fraction finer of representative small grain sizes at either upper or middle Sauk River study sites.

In terms of differences in grain size distributions between the wet and dry bulk surface samples across study years, wet samples are finer than dry samples in study year 2 recovery period and extending through study year 3 installation period at middle and lower Sauk River study site. Differences in dry samples are limited to study years 1 and 2 because of the lack of study year 3 samples. However, there does not seem to be differences in fraction finer of smaller grains in the dry gravel sites beginning at study year 2 recovery period that had been observed in the wet samples (figs. 12–13; table 5). Dry samples at upper Sauk River study sites were slightly finer in study year 2 with a higher fraction of fines for representative small size clasts, but the grain size seems to have coarsened by study year 3.

For differences in grain size distributions across study sites, lower Sauk River study site is notably finer than upper and middle Sauk River study sites across all study years, although these differences are most notable during the installation period than during the recovery period (fig. 13; table 5).

Subsurface Sediment Conditions Within the Wetted Channel and Adjacent Gravel Bars

Due to the onset of Covid19 in late winter 2020, field efforts for the study year 2 recovery were delayed at middle and lower Sauk River study site. An in-channel bulk sediment sample at lower Sauk River study site was taken in March 2020 and compared to in-channel subsurface samples in April during the recovery campaign of WV boxes and infiltration bags. There do not seem to be substantial differences between the grain size distribution of in March compared to those of the April samples (fig. 14); therefore, the April samples are considered an equivalent substitution for bed conditions during March 2020, when Chinook salmon were expected to still reside in the subsurface gravels at lower Sauk River study site.

Comparison between WV Boxes, Infiltration bags, and Bulk Subsurface Samples

We compare grain size distributions of the bulk subsurface samples to the WV boxes and infiltration bags to evaluate if bulk grain size distributions can be a proxy for the potential of sediment infiltration into spawning gravels, which is represented by the WV boxes and infiltration bags (fig. 15). Wet and dry bulk samples both have a higher fraction finer for the representative fine grain size (2 mm) than the WV boxes (fig. 15) and are generally finer for other small grain size classes (tables 5–7). One-sample t-test confirmed statistical differences between WV boxes and bulk subsurface samples, but only identified significant differences in the finest class of <0.063 between WV boxes and infiltration bags (table 7). Absence of statistical differences between infiltration bags and subsurface samples is attributed to low sample size of infiltration bags (no bags in study year 1 and recovery of only a single bag at middle Sauk River study site in study years 2 and 3) (table 6). When looking at individual study sites, it seems that fraction finer values for WV boxes and infiltration bags are similar to each other but are generally coarser than either wet or dry subsurface samples regardless of seasonal sampling.

Dissolved Oxygen Concentrations in Chinook Salmon Spawning Gravels

All recovered DO sensors remained operational except for one sensor (Sensor 3) at upper Sauk River study site in study year 2 when the sensor seems to have fouled on February 1, 2020, during which the DO record had a sharp decline. The field check indicated an average DO concentration of 11.7 mg/L and the sensor had a strong odor when recovered, suggesting fouling. DO values were removed from the record that are presumed to be during sensor fouling. All calibrated DO sensors were within 0.2 mg/L of the calibrated sonde and are within the USGS specifications for continuous DO measurements (plus or minus 0.3 mg/L; Wagner and others, 2006). Dissolved oxygen data can be found in Jaeger and others (2024).

Dissolved oxygen levels remained above 4 mg/L for most sensors for most of the study years 2 and 3 monitoring periods (figs. 16–17). Dissolved oxygen standards are considered “Fair” above 4 mg/L (Washington Department of Ecology, 2002). Dissolved oxygen levels increased with seasonal decreases in water temperature during the autumn and early winter, but individual sensors fluctuated substantially following this seasonal increase that corresponded to cooling water temperatures.

Aside from the fouled sensor at upper Sauk River study site in study year 2, 1 sensor in the upper Sauk River study site record DO was at or below 4 mg/L 5 times for periods ranging between 1 and 30 hours (table 8). One of the four sensors in lower Sauk River study site in study year 2 also recorded DO at or below 4 mg/L for 3 periods ranging from 3 to 79 hours. No sensors were recovered from the middle Sauk River study site in study year 2. In study year 3, the single sensor recovered from middle Sauk River study site recorded 4 periods at or below 4 mg/L ranging from 1 to 13 hours (table 8). In study year 2, periods of low DO occurred in the latter half of the pre-emergence period (ATU greater than 500) and presumably, embryos had hatched into alevins. Although hatching has high oxygen demands, oxygen demands decline post hatching (Rombough, 1988, cited in Greig and others, 2007) so increasing distance from this 500 ATU midpoint represent periods of less vulnerability for mobile alevin, presuming that there are other areas of higher DO to migrate to. The incubation period is more vulnerable to low DO conditions since embryos are not mobile, whereas post hatching, alevins can move in the subsurface away from areas of low DO (Greig and others, 2007).

Abrupt increases and decreases in DO correspond to high flow events, although sharp changes also occur even in the absence of events, which are presumed to reflect local changes in DO conditions associated with riverbed perturbations (figs. 18–19). The location of individual dissolved oxygen sensors may be unstable based on the large fluctuations and therefore may be more sensitive to shifts caused by high flows or local shifts in the bed conditions.

Considerations of the Level of Effort Among Different Sampling Methods

We had presumed that subsurface methods would be more easily measured and therefore a desirable alternative to the presumed logistically more complex methods involving WV boxes and infiltration bags. To evaluate if subsurface sampling methods were a physically less intensive and logistically less complex alternative to the WV boxes or infiltration bags, we developed a qualitative ranking based on factors we identified as representing physical labor and logistical complexity of all four methods (table 9). We ranked each factor subjectively from one to five corresponding to a range of relatively easy (1) to hard (5) such that a low overall sum of ranks assigned to each factor represent methods that are less physically laborious or logistically complex. This ranking may help when considering applying subsurface, WV box, and infiltration bag methods in future sediment infiltration field studies. Based on the assigned rankings derived from our personal experience in the study, the WV box method and dry gravel bar subsurface sample method had similarly low overall ranks, with all factors evaluated as either easy or moderate although differences existed between rankings of individual factors. Wet in-channel subsurface methods had rankings that were slightly higher because of moderate-to-hard and hard rankings for field logistical complexity and post field processing logistical complexity, respectively. The infiltration bag ranked higher, reflecting the challenges associated with the physical recovery of the bag and subsequent logistical challenges of managing the relatively large sample mass and volumes, which included many containers of associated river water. Based on this subjective, qualitative assessment of the different field method, fine sediment distributions of the river, and particularly the adjacent gravel bars, may be logistically easier alternative to using infiltration bags and potentially for WV boxes, depending on if the factors are ranked or weighted differently.