Stream-terrace morphostratigraphy and optically stimulated luminescence (OSL) geochronology indicate that storm-driven sedimentation has caused down-system decoupling of the uppermost reaches of McMillan Creek (southern California, USA) from the lower reaches of McMillan Creek since 1960 ± 190 yr B.P. This is significant because source-to-sink studies report high degrees of sediment transport connectivity over millennial time scales during periods of high fluvial discharge in sediment routing systems.

The most recent relatively large-magnitude episode of sedimentation emplaced a sediment slug in the ephemeral channel of McMillan Creek. The sediment slug is correlated to the “California Storm of January 1862” via OSL dating. In this paper, a conceptual model of sediment slug dynamics in an ephemeral stream over 16 decades is developed based on fluvial sedimentation events that in most instances included reworking slug-derived sediment. Due to the episodic nature of streamflow in ephemeral streams and the dearth of sediment transport between streamflow events, sediment slug coherency is sustained over longer periods of time in ephemeral streams than in perennial streams having steady or variable flow regimes.

The longevity of sediment-slug coherency in ephemeral streams leads to more prolonged down-system decoupling in sediment routing systems than down-system decoupling caused by ordinary fluvial sedimentation. In McMillan Creek, it is possible that up-system decoupling driven by sedimentation has been contemporaneous with down-system decoupling, but factors other than sedimentation may have a more significant role in up-system decoupling. Source-to-sink studies completed in areas having a Mediterranean climate cannot assume that sediment flux out of upland source areas includes the total amount of sediment available for transport.

Definitions

In this paper, we adopt the terminology of Harvey (2002) as specified in this subsection in italicized text. Reach-to-reach-scale decoupling influences both down-system and up-system stream-channel connectivity. Reach-to-reach-scale, down-system decoupling disrupts sediment transport. Within a drainage network affected by down-system decoupling, when one reach is decoupled from another reach, it implies that sediment is accumulating in upper reach and little or no sediment is being delivered to the lower reach. Reach-to-reach-scale, up-system decoupling disrupts upstream propagation of base-level changes. A stream network that is decoupled from base level is isolated from the effect of base level. Regional-scale decoupling is caused by tectonic uplift and/or stream capture. One mechanism that can cause regional-scale decoupling of a drainage basin from base level is the time required for base-level fall to propagate upstream from the lower part of the drainage basin to the upper part of the drainage basin.

Climate Forcing and Sedimentation in Erosional Landscapes

The character of sedimentation in terrain where erosion predominates has been proposed to be a fundamental influence on denudation rates (Gallen et al., 2015, and references therein). More specifically, a commonly cited challenge that erosional-landscape studies must address is understanding the influence of exogenic forcing mechanisms such as climate on stream-incision rates (Perron, 2017). Because erosional uplands are source areas in source-to-sink studies and erosion rates influence sediment flux, in the source-to-sink frame of reference, the character of sedimentation in source areas influences sediment transport out of source areas.

Sediment-transport rates impart a fundamental influence on climate-signal transfer times, which in turn are a fundamental component of source-to-sink analyses of stratigraphic sequences (Tofelde et al., 2021). The current level of understanding of grain mobility over greater than decadal time scales that is necessary for source-to-sink analyses is inadequate (Tofelde et al., 2021). This paper identifies processes and factors that have impeded sediment transport over centennial to millennial time scales in a setting regarded as a source area in a source-to-sink framework.

Both conceptual and quantitative models benefit from field-based results that show how climate forcing influences erosion rates and sediment transport rates. The morphostratigraphic character and the geochronology of sediments that compose the single terrace level in a 25.7 km2 drainage basin indicate that since ca. 2 ka, floods triggered by relatively large storms have led to net aggradation in the axial stream (McMillan Creek, southern California, USA) of the fluvial network draining the basin. Aggradation in the channel of McMillan Creek has led to reach-to-reach-scale decoupling of the upper reaches of McMillan Creek from the lower reaches of McMillan Creek since ca. 2 ka.

Climate Forcing and Sediment Slugs in Ephemeral Stream Channels

The most recent episode of aggradation preserved in the morphostratigraphy of McMillan Creek stream terraces is associated with a sediment slug that is correlated to the “California Storm of January 1862” of Engstrom (1996). Sediment slugs have also been referred to as “sediment pulses” (Harvey, 2002), “aggradation-degradation cycles” (Reneau et al., 2004), as well as “sediment waves” and “aggradation-degradation episodes” (James, 2010). Except for settings where human activity leads to extreme production and delivery of sediment to streams, sediment slugs are related to climate forcing because they are typically generated by landslides triggered by intense rainfall (e.g., Nelson and Dubé, 2016). Sediment slugs consist of sediments that, after abrupt delivery to a stream channel, are transported collectively downstream as wave-like, coherent masses (Gilbert, 1917; Miller and Benda, 2000; Lisle et al., 2001; Nelson and Dubé, 2016; Moody, 2017). Typically, for some period of time, sediment slugs remain coherent, and transport of slug-related sediment occurs collectively via wave-like downstream translation. Eventually transport of slug-related sediment becomes dispersive rather than wave-like (Lisle et al., 2001; Moody, 2017).

Despite a great deal of research on sediment slugs and advances in understanding of the processes by which relatively large amounts of sediments are transported collectively downstream, more work must be completed to understand factors that lead to either dispersal of the sediment that constitutes the slug or sediment transport as distinct wave-like pulses (Moody, 2017). This paper assesses sediment-slug coherence in an ephemeral stream over a relatively long time scale. Whereas most modern sediment-slug studies are based on slugs that are typically a few decades old, the geochronological correlation of the studied sediment slug to a historic storm makes it possible to evaluate sediment slug dynamics over ~16 decades.

The time scales over which sediment slugs remain coherent and identification of the factors that favor coherence or cause dispersal have been assessed mostly by studying streams or conducting experiments wherein steady-flow regimes predominate (Moody, 2017; e.g., Knighton, 1989; Miller and Benda, 2000; Lisle et al., 2001; Kasai et al., 2004; Bartley and Rutherfurd, 2005; Nelson and Dubé, 2016). Literature searches of geoscience databases indicate that very few studies of sediment slugs in ephemeral streams have been completed. This is significant because it has been concluded that the dynamics of sediment slugs in streams where a steady-flow regime predominates are different than the dynamics of sediment slugs in streams where the flow regime varies (Moody, 2017). It follows that sediment-slug dynamics in ephemeral streams are likely to be different than sediment-slug dynamics in perennial streams where steady or variable flow regimes predominate. This paper provides insight into sediment-slug dynamics in ephemeral streams and proposes a conceptual model for sediment slug dynamics over a relatively long time scale.

Geology of Gillis Canyon

McMillan Creek is the axial stream of Gillis Canyon (Figs. 1 and 2). Gillis Canyon is formed in unconsolidated to weakly consolidated units of the Plio-Pleistocene Paso Robles Formation (Dibblee, 1974). The Paso Robles Formation in Gillis Canyon is a sequence composed of (from bottom to top): sand and gravel (unit QTg); sand and clay (QTs); and clay (QTc; Dibblee, 1974). Distinctively red rocks of the Caliente Formation (unit Tc of Dibblee, 1974) are exposed within the McMillan Creek drainage basin but south of Gillis Canyon in an area known as the Red Hills.

Regional Tectonics and Physiography of Gillis Canyon

Gillis Canyon is within the topographic expression of the San Andreas fault zone near the northwestern end of the Temblor Range (Fig. 1; please note that the location of significant features within Gillis Canyon that are referred to throughout this paper are denoted in Fig. 2). The predominant altitude of the highest peaks and ridges of the upland is ~610 m. The typical maximum relief between interfluves and the adjacent erosional valleys is ~90 m.

McMillan Creek is a tributary of San Juan Creek and is part of the Salinas River drainage network. The McMillan Creek–San Juan Creek confluence is well over 100 km from where the Salinas River flows into the Pacific Ocean at Monterey Bay (Fig. 1). The lower 4.9 km of McMillan Creek is an alluvial stream, and the upper 6.6 km of McMillan Creek is a mixed alluvial and bedrock-eroding stream (Fig. 3).

The headwaters of McMillan Creek are 2.5 km southwest of an active strand of the San Andreas fault zone (Vedder and Wallace, 1970), and the confluence of McMillan Creek and San Juan Creek is 12.3 km southwest of this same active strand (Fig. 1). The upland into which Gillis Canyon is cut most likely formed via oblique convergence at a relatively low angle along a linear segment of the San Andreas fault zone (e.g., Hill et al., 1990; Page et al., 1998; Spotila et al., 2007). The Paso Robles Formation is “fanglomerate” (Dibblee, 1974) in that it was deposited by a stream network that is an unrelated predecessor to the current stream network. In the context of the proximity of Gillis Canyon to an active fault zone at a plate boundary, Pleistocene reorganization of drainage patterns and the presence of Pleistocene alluvium as dissected substrate in a topographic upland indicates relatively recent and ongoing relief-generating tectonism.

The Gabilan Mesa is the closest site to Gillis Canyon for which rock uplift rates have been established. The Gabilan Mesa is adjacent to the Temblor Range along orogenic strike (Fig. 1), and like the Temblor Range, the mesa is also adjacent to a linear segment of the San Andreas fault zone. The degree of Pleistocene drainage reorganization (Galehouse, 1967) and the degree of topographic development in the Gabilan Mesa are similar to the degree of Pleistocene drainage reorganization and the degree of topographic development in the Temblor Range. Ongoing rock uplift rates in Gabilan Mesa are between 0.32 and 0.42 mm/yr (Titus et al., 2011; García and Mahan, 2012). No other data exist regarding rock uplift rates in the Gillis Canyon area.

Despite the modest relief within the northwestern end of the Temblor Range, the substrate in which the upper hillslopes and ridgelines of Gillis Canyon are formed (unconsolidated sand as well as sand and clay; units QTs and QTc of Dibblee, 1974) ensures that hillslope denudation is dominated by mass movements. Earthflows that are many tens of square kilometers in extent and colluvial hollows evacuated to a highly variable degree are ubiquitous throughout the McMillan Creek drainage basin.

Climate in the Gillis Canyon Area

Summers in the Gillis Canyon area are relatively warm and dry, and winters are relatively cool and moist. The source of precipitation data nearest to Gillis Canyon is a California forestry station in the town of Shandon, which is 5.5 km northwest of the study site. Mean annual precipitation (MAP) at Shandon is 30.5 cm (Public Works, Water Resources, County of San Luis Obipso, 2023). Topographic relief between Shandon and Gillis Canyon is relatively small, and it is likely that orographic effect is minimal. Therefore, the best available estimate of MAP at Gillis Canyon is 30.5 cm. The average daily low temperature in Shandon during winter is 1.6 °C, and the average daily high temperature during summer is 33.8 °C (Western Regional Climate Center, 2023).

Field Mapping and Measurements

The surficial geology of a segment of Gillis Canyon was mapped in the field on the Camatta Canyon and Cholame quadrangles of the U.S. Geological Survey (USGS) 1:24,000-scale topographic map series. The maps were enlarged to a scale of ~1:14,500. Areas mapped as unit Qa1a and Qa1 terraces include the channel of McMillan Creek (Fig. 4). Stream terrace and alluvial fan sedimentary sequences were logged where exposures exist on stream banks. The heights above the stream channel of terrace treads, the cross-valley width of alluvial map units, and the along-channel extent of terraces and channel characteristics were measured using a measuring tape for measurements <10 m and using a laser range finder for measurements >10 m.

Observations and measurements were made in lower Gillis Canyon from downstream of the mapped area in Figure 4 to where McMillan Creek flows into San Juan Creek, but limited access precluded mapping that area.

Digital Elevation Models and Stream Longitudinal Profiles

The digital elevation model in Figure 2 was derived from a lidar point cloud. A point cloud of the study area in LAZ format was obtained from the USGS 3D Elevation Program (3DEP) portal on Open Topography; this point cloud was collected in 2019 and has a point density of 4.73 points per square meter (Open Topography, 2021). Quick Terrain Modeler software (https://appliedimagery.com/) was used to create a digital terrain model with 1 m grid sampling of the ground returns in the pre-classified lidar point cloud.

A longitudinal profile for McMillan Creek (Fig. 3) was created from the Figure 2 digital terrain model in ArcMap (ESRI ArcGIS version 10.6). No hydraulic conditioning was conducted on the profile, so road crossings and bridges across McMillan Creek are included in the profile. The human-made structures were identified in the field as well as on Google Earth images and denoted on the profiles. In one instance, a human-made structure was surveyed using a centimeter-accuracy, GPS-based surveying kit. Stream channel gradients for select reaches of McMillan Creek were determined using a Microsoft Excel spreadsheet to generate best-fit line functions to the stream profile for McMillan Creek that was created from the digital terrain model in ArcMap.

Sediment Sampling for OSL Dates

Samples collected from Gillis Canyon for optically stimulated luminescence (OSL) dating were obtained under moonlight between 1 and 4 a.m. on 22 June 2013. The sampled outcrops were first identified while completing surficial geologic mapping. Samples were collected in black, 35 mm photographic-film vials from within 10–20 cm deep excavations that were made in the outcrop surfaces.

OSL Lab Analysis Methodology

Depositional ages of the sediment are based on luminescence analyses of fine sand–sized (0.125–0.25 mm) quartz grains. Quartz grains were isolated first by chemical solutions such as hydrogen chloride and hydrogen peroxide; magnetic separation; then heavy liquids followed by etching with hydrofluoric acid; and lastly, broken feldspar grains were removed from quartz batches by re-sieving using standard protocols (Mahan et al., 2015; Nelson et al., 2015). All quartz-grain samples were measured using the single-aliquot regenerated dose (SAR) procedure (Wintle and Murray, 2006; Thomsen et al., 2008) with a blue-light stimulation preceded by an infrared stimulated luminescence (IRSL) “wash” of 100 s at 60 °C. Sample aliquot size was small for the SAR protocol (generally 200–250 grains), centered in the middle of the disc, and composed of grains in the 125–250 μm size range.

Previous studies have found that using the coarser-sized sand grains results in better bleached results (less “partial bleaching” of grains); explanations vary for this difference in bleaching (e.g., Truelsen and Wallinga, 2003; Alexanderson, 2007; Vandenberghe et al., 2007). Most of these hypotheses are related to the mode of transport (slower travel path for coarser material, more likely to be deposited on point bars and subaerially exposed; reference details in Rittenour, 2008). The central age model (CAM) as well as the minimum age model 3 (MAM-3), and radial plots were used to generate the most appropriate equivalent doses for the final dates; when two out of the three matched, the averaged equivalent dose was used. Additional details regarding the methodology of the dosimetry, preheats, and equivalent-dose distributions are described in Mahan et al. (2015). Finally, similar to radiocarbon results, all luminescence ages are presented in years before present (yr B.P., where 0 yr B.P. = A.D. 2000), and uncertainties are given at the 67% (1σ) confidence level.

Time Scale of Sediment-Slug Coherency in an Ephemeral Stream

On the basis of results from studies of perennial streams, it has been postulated that over decadal time scales, the relatively steep, downstream end of a sediment slug leads to increased stream-channel erosion rates and dispersal of the sediment that constitutes the slug (Nelson and Dubé, 2016). Results of our study indicate that the relatively steep slope of the downstream end of the sediment slug in the channel of McMillan Creek has persisted since A.D. 1862 and is the downstream end of a convexity in the channel of McMillan Creek.

We propose that in ephemeral streams, no degradation of the steep, downstream end of the sediment slug occurs between large-magnitude discharge events. Large-magnitude discharge events remobilize sediment that composes the slug, and this ample source of sediment ensures that stream-channel aggradation at the downstream end of the slug minimizes dispersion.

The climatic influence on sediment supply where ephemeral streams exist also influences the time scale over which sediment-slug coherency is preserved. Relatively high sediment yield occurs in areas having a “Mediterranean” climate (Langbein and Schumm, 1958; Gray et al., 2018). The prevalence of relatively high sediment yield in areas having a Mediterranean climate ensures that during climatic events that lead to only moderate discharge, delivery of relatively large sediment loads to stream channels from hillslopes occurs. Whereas sediment slugs in perennial streams are denuded and dispersed during climatic events that lead to moderate discharge (e.g., Nelson and Dubé, 2016), delivery of relatively large sediment loads to ephemeral stream channels during climatic events that lead to moderate discharge increase the mass of a sediment slug.

Influence of Sediment Slugs on Down-System Stream-Network Connectivity

Sediment slugs comprise transient in-channel depocenters (e.g., Knighton, 1989). Once emplaced, sediment reworked from the slug accumulates at the downstream end of the slug via channel aggradation that causes channel-widening anastomosis, as well as other morphological channel and floodplain changes (e.g., Knighton, 1989; Nelson and Dubé, 2016). The specific mechanism that causes channel aggradation is the relationship of critical stream power to stream power in reaches where sediment slugs exist (Harvey, 2002). Once a sediment slug is emplaced, if slug-sediment mobilization is purely or close to purely translational, sediment mobilization is episodic (e.g., Lisle et al., 2001). The results of our study (see the Results section and the Interpretation of Resuts section) show that episodes of downstream translation of a sediment slug in an ephemeral channel are relatively infrequent and that the sediment slug in McMillan Creek has translated ~1 km since 140 ± 20 yr B.P. The combination of (1) relatively long periods of time between episodes of sediment remobilization in ephemeral streams and (2) aggradation at the downstream end of the sediment slug constitute periods of disrupted sediment transport regarded as decoupling by Harvey (2002). Therefore, as noted by Harvey (2002), the presence of a sediment slug in a stream channel causes decoupling.

Influence of Sediment Slugs on Up-System Stream-Network Connectivity

Over time scales ranging from decades to tens of thousands of years, sedimentation in a discrete part of a drainage basin can impede the effect of base-level fall upstream of where sedimentation occurred (Harvey, 2002). After the sedimentation event, a stream or hillslope upstream of the depocenter is decoupled from the base level of the system (Harvey, 2002). The spatial scales over which up-system decoupling occurs ranges from an individual hillslope that is decoupled from an adjacent stream channel to segments of stream networks draining catchments ranging in size from a few square kilometers to thousands of square kilometers (Harvey, 2002). It is proposed that because sediment slugs migrate downstream and because of the dynamics of base-level fall propagation (see the following subsection), the time scale over which sediment-slug sedimentation disrupts up-system fluvial system connectivity is greater than the time scale over which spatially restricted sedimentation of comparable magnitude disrupts up-system fluvial-system connectivity.

Conceptual Model of the Influence of Sediment-Slug Dynamics on Tectonically Driven Base-Level Fall and Incision Waves

An incision wave is a temporal and spatial pattern of stream incision by which base-level fall propagates upstream (Safran, 1998). Incision in one reach lowers local base level for the adjacent upstream reach. The increase in channel slope caused by local base-level fall increases available stream power, and the upstream reach incises and adjusts to the new, lower local base level (Safran, 1998). Except in locations where orogenesis is caused by tilting, stream-network development in tectonically active areas most commonly occurs as pulses of tectonically driven uplift trigger upstream-translating incision waves (Safran, 1998; Zaprowski et al., 2001; García et al., 2004; e.g., García and Mahan, 2014).

The specific mechanism that drives upstream propagation of base-level fall is an increase in channel slope–controlled stream power due to local base-level fall (Safran, 1998; García et al., 2004). If an incision wave were to propagate into an in-channel depocenter such as a sediment slug, the increase in channel-slope controlled stream power could be insufficient to transport the sediment load and the incision wave would not propagate through the depocenter.

The tendency for sediment slugs to translate downstream more effectively inhibits incision-wave propagation than spatially restricted aggradation. In the combined frameworks of the models of Harvey (2002) and Safran (1998), the downstream migration of a sediment slug increases the length of the segment of the fluvial system that is decoupled and thereby buffered from the effect of base-level fall, which increases the time it takes for base-level fall to propagate through a drainage basin. The relatively slow incision rates that result from the suppressed effect of base-level fall and the relatively long temporal scale over which sediment slugs persist in ephemeral streams constitute newly recognized complexities in the relationship between climate forcing, sedimentation, rates of drainage basin denudation, and rates of sediment transport in the uplands that are source areas in the source-to-sink framework.

Mapping Results

Map Unit Qa1 and Qa1a Stream Terraces

A single terrace level exists in Gillis Canyon. In a map-view perspective, the terraces that constitute the upper 600 m of the single terrace level lack uniformly distinctive tread-forming sediments and are mapped as Qa1 (Fig. 4 and Table 1). Along the rest of McMillan Creek that is bordered by the single terrace level, terraces have treads formed by a distinctively gray, typically 1- to 1.5-m-thick bed of sand and gravel and are mapped as unit Qa1a (Fig. 4; Table 1). In a map-view perspective, this distinctively gray bed is the upper two-thirds of the sediment slug that was emplaced during the California Storm of January 1862 (Fig. 4). Data supporting the conclusion that this bed is part of a sediment slug is presented in the following subsection and the Interpretation of Results section. The sediment slug is identified here to facilitate a concise and clear narrative that establishes the significance of other landforms and sediments in Gillis Canyon.

The channel of McMillan Creek is composed of coarse gravel and sand in reaches where unit Qa1a is mapped. In these reaches, cobbles are abundant and boulders are commonly present. The bases of Qa1a and Qa1 sequences are not exposed. The treads of Qa1a and Qa1 terraces are typically inset into tributary fans along risers that vary in height from <1 m to 2 m (Figs. 4 and 5). Terrace treads are locally inset into valley side slopes formed in the Paso Robles Formation (Fig. 4).

Along the upper 500-m- to 600-m-long segment of the reach of McMillan Creek that is bordered by Qa1a terraces, the height of the terrace treads above the McMillan Creek channel is typically between 4 and 6 m. However, the height of the tread on both sides of the channel decreases consistently downstream, and along the southeast side of McMillan Creek, the Qa1a tread merges with the McMillan Creek channel (Fig. 4). At this same location, but on the northwestern side of the channel, the Qa1a tread terminates downstream along an inset contact with a tributary fan, and the height of the Qa1a tread is <0.5 m above the channel of McMillan Creek (Fig. 4).

The location where the unit Qa1a tread merges with the McMillan Creek channel will henceforth be referred to as “the lowermost extent of Qa1a terraces” and is interpreted as the downstream end of the sediment slug when it was first emplaced (e.g., Miller and Benda, 2000; Reneau et al., 2004). The channel of McMillan Creek is mapped as unit Qa2 in the reach where the Qa1a tread merges with the channel. Downstream of the lowermost extent of Qa1a terrace alluvium that constitutes the channel and floodplains of McMillan Creek is also mapped as Qa2 (Fig. 4).

Map Unit Qa2: Spatial and Sedimentological Characteristics

The reference frame for all measurements in this subsection is the uppermost McMillan Creek headwaters at the drainage divide, which will henceforth be referred to as “the drainage divide”. The lowermost location where McMillan Creek flows across a road (lowermost road crossing [LRC]) is also used as a reference frame. The spatial relationships of features described in this subsection to the sediment slug and to each other are specified in Table 2 as well as Figures 2 and 3 and are illustrated by a partial longitudinal profile of McMillan Creek and a Google Earth image in Figure 6.

Unit Qa2 is the part of the sediment slug that has translated beyond what was the downstream end of the slug when it was first emplaced. Qa2 is mostly sand and gravel that locally includes boulder-sized concrete blocks (Fig. 7). From 11,620 to 11,920 m downstream of the drainage divide (105 m upstream of the LRC to 195 m downstream of the LRC), the area mapped as Qa2 is composed of an 11–12-m-wide stream channel and adjacent gravelly floodplains (Table 2). In this 300-m-long reach, the gravelly Qa2 channel alluvium is similar to the channel alluvium within areas mapped as unit Qa1a, and the gravel in the floodplains consists predominantly of rounded cobbles. Rounded boulders having median diameters of as much as 70 cm are common on these floodplains. Farther downstream, between 11,920 and 12,048 m downstream of the drainage divide (195–323 m downstream of the LRC), Qa2 grades from a channel and adjacent floodplains to a multithread stream channel (Figs. 6 and 8). In this transitional reach, the channel is 11 m wide and the northwestern stream-channel bank is formed by a tributary fan. The southeastern bank is formed in alluvium composed of coarse gravel and sand (Fig. 8); the surface of this alluvial feature is as high as 1.3 m above the currently active channel and includes a 16-m-wide paleochannel that is backfilled at its downstream end (Fig. 8).

From 12,048 to 12,666 m downstream of the drainage divide (323–941 downstream of the LRC), unit Qa2 is a multithread channel (Fig. 9). The channel widens from 28 m at 12,048 m downstream of the drainage divide to 60 m at 12,666 m downstream of the drainage divide (Figs. 6 and 9). There is no floodplain between 12,048 and 12,666 m downstream of the drainage divide. Instead, Qa2 is inset into tributary fans (Fig. 4) and is composed of multiple channels and abundant, sparsely vegetated sand and coarse gravel bars that are elongate parallel to streamflow. The source of the gravelly sediment that composes the multithread channel is the reaches of McMillan Creek that are upstream of the lowermost extent of Qa1a terraces. The basis of this conclusion is the character of the alluvial fans that issue forth from McMillan Creek tributaries and form the banks of the McMillan Creek channel downstream of the lowermost extent of Qa1a terraces: These fans lack feeder channels that are incised below the fan surface (Figs. 4 and 8), therefore the tributary channels feeding the fans are decoupled from the channel of McMillen Creek (sensuHarvey, 2002).

Farther than 12,666 m and as far as 13,120 m downstream of the drainage divide (941–1395 m downstream of the LRC), the channel and floodplain alluvium mapped as unit Qa2 are the same as the channel alluvium within areas mapped as unit Qa1a but the gravel in the floodplains lack boulders; beyond 13,120 m downstream of the drainage divide (1395 m downstream of the LRC), Qa2 floodplains and in-channel gravel bars consist of pebbly sand with few cobbles.

Longitudinal Profile of McMillan Creek

The upper 8 km of the McMillan Creek longitudinal profile has conspicuous convexities and is unsmooth. Between ~1500 m to 7250 m downstream of the drainage divide, there are three distinct, relatively long-wavelength convexities (Fig. 3). One of these convexities is adjacent to an earthflow that extends downslope to the channel of McMillan Creek (Figs. 2 and 3).

The lower 6 km of the McMillan Creek longitudinal profile is mostly linear, locally gently convex up, and relatively smooth (Fig. 3). The most well-developed convexity extends from the base of the knickpoint formed where McMillan Creek crosses Gillis Canyon Road (the LRC referred to above) to 12,666 m downstream of the drainage divide (941 m downstream of the LRC). Note that the downstream end of the multithread reach described above is also at 12,666 m downstream of the drainage divide (Fig. 6; Table 2). The slope of the downstream half of the convexity is 0.016.

Unit Qa1a Stratigraphy

Please note that detailed logs of unit Qa1a stratigraphy are available in the Supplemental Material1. The most noteworthy stratigraphic unit in Gillis Canyon is the sediment slug that forms the tread of Qa1a. The sediment slug is distinctly and uniformly gray, which is consistent with relatively recent deposition. The thickness of the bed is typically 1–1.5 m, but it is locally as much as 2.3 m thick. The sediment slug is composed of lenses of planar- and cross-laminated gravel and sand, which indicates deposition in a fluvial environment. Gravel clasts in the sediment slug have median diameters of as much as 90 mm. The lenses of gravel and sand are meters wide and decimeters thick. Well preserved cross laminae formed in sand exist within 1 cm of the terrace tread (Fig. 10). The presence of these well-preserved cross laminae <1 cm below the terrace tread indicate little to no pedogenesis or bioturbation.

The 3.5–2.5-m-thick sequence that underlies the sediment slug typically consists of interlayered and interfingered, lenticular and tabular beds of sand, pebbly sand, and clast-supported gravel and sand. Locally, the unit Qa1a sediment that underlies the sediment slug is composed of massive, matrix-supported, poorly sorted mud and granule- to boulder-sized gravel.

Gillis Canyon Geochronology

OSL age estimates on unit Qa1a alluvium are consistent with stratigraphic relationships. Sample GC-2 was collected from 1.45 m above the channel of McMillan Creek and 3.3 m below the terrace tread, and it yielded an older date than sample GC-1, which was collected within the sediment slug and closer to the terrace tread. The OSL date on sample GC-2 is 1960 ± 190 yr B.P. (Table 3), therefore the upper age range of sample GC-2 correlates to the latter part of the highstand of Soda Lake (55 km southeast of Gillis Canyon; Figs. 1 and 11) that occurred at ca. 2250 cal yr B.P. (Eigenbrode, 1999). The upper age range is also close to the 2.2 ka maxima of a glacial advance in the Big Pine Creek drainage basin of Sierra Nevada (Bowerman and Clark, 2011; Figs. 1 and 11). The lower age range of sample GC-2 is within the OSL age range of lowermost inset terraces in Bitter Creek (Garcia et al., 2017), which is a stream having a catchment area of 45 km2 in the San Emigdio Mountains (115 km southeast of Gillis Canyon; Figs. 1 and 11). Records of large-magnitude floods have been identified in the stratigraphy of the marine Santa Barbara Channel, which is 155 km south-southeast of Gillis Canyon (Fig. 1). The floods occurred at 107 B.C. as well as A.D. 33, 53, and 184 (Hendy et al., 2013). In OSL chronology, 107 B.C. is 2123 yr ago and A.D. 184 is 1832 yr ago, therefore the OSL ages of Qa1a alluvium that underlies the sediment slug encompass the time when the floods identified by Hendy et al. (2013) occurred (Fig. 11).

Sample GC-1 was collected within the sediment slug, 0.53 m below the unit Qa1a terrace tread and ~3.5–3 m above the channel of McMillan Creek. The OSL date on sample GC-1 is 140 ± 20 yr B.P., therefore the range of the OSL date correlates to the calendar years between A.D. 1840 and 1880, with a median at 1860 (OSL ages are relative to A.D. 2000). The California Storm of January 1862 caused the largest-magnitude flood recorded in California history and consisted of 45 days of continuous rainfall starting on 24 December 1861 (Engstrom, 1996).

The effects of the California Storm of January 1862 were widespread. Total rainfalls in January 1862 in San Diego (450 km southeast of Gillis Canyon) and San Francisco (310 km northwest of Gillis Canyon), respectively, were 300% and 502% greater than “normal” (Engstrom, 1996, p. 1430). A lake-altitude curve for Mono Lake (Fig. 1) has a prominent spike at 1862, indicating a short-lived increase in the altitude of the lake surface (Stine, 1990). Another storm that caused floods in California having greater magnitude than any 20th- or 21st century floods occurred in December 1867 (Engstrom, 1996). In summary, the age range of sample GC-1 is centered within 1–2 yr of the largest flood recorded in the history of California and centered within 7–8 yr of one of the three largest floods recorded in the history of California (Engstrom, 1996).

The relative depositional age of unit Qa2 is established on the basis of map relationships and the presence of concrete blocks in coarse gravelly bars and floodplains. Qa2 is inset into and, therefore, younger than Qa1a. Widespread use of concrete for building structures was uncommon in California until ca. A.D. 1920 (R. Moss, 2023, personal commun.). It follows that the maximum age of Qa2 sedimentation postdates emplacement of the sediment slug by ~60 yr.

Observations Made on 12 January 2023

On 9 January 2023, the California Department of Forestry weather station in Shandon recorded 65 mm of rainfall within 24 h, which is slightly more than twice the MAP at this location. However, including the storm on 9 January 2023, comparable magnitudes of rainfall have occurred at this weather station four or five times over 17 yr (Table 4). Unfortunately, discharge data are not available for McMillan Creek. The observations included below were made on 12 January 2023.

Abundant and widespread shallow-level slope failures were observed throughout Gillis Canyon and the surrounding area. The slope failures delivered sand and mud to the valley floors of Gillis Canyon and of the San Juan Creek stream valley. However, there was no evidence for overbank flow along the channel of McMillan Creek from the San Juan Creek confluence to the uppermost culvert in the reaches where unit Qa1 terraces are mapped, which is ~8200 m downstream of the drainage divide.

Although evidence for erosion was observed locally, the channel of McMillan Creek from the San Juan Creek confluence to the upstream-most end of unit Qa1 terraces (Figs. 1 and 4) was almost entirely mantled by sand and mud. Laterally directed stream-bank scour occurred locally along the outside part of some meander loops. However, a channel within a meander loop that is cut into the sediment slug was buried in sand and mud (Fig. 12). Although vertical incision occurred locally, mostly near culverts and road crossings, within the sediment slug areas between longitudinal gravel bars backfilled with sand and mud and the gravel bars were partly buried in sand and mud (Fig. 9).

Sediment Slug

In addition to geochronological correlation to the California Storm of January 1862, morphostratigraphic characteristics indicate that the distinctive gray bed that forms the tread of unit Qa1a terraces was deposited as a sediment slug during a relatively recent flood. Fluvial aggradation occurring during extreme floods is accompanied by plan-view erosion of the valley-bottom alluvium that the flood-related sediments are ultimately inset into (Moody and Meade, 2008). This relationship is exemplified by the inset relationships of the Qa1a terrace treads to risers formed in fans (map unit Qf in Fig. 4) that issue forth from McMillan Creek tributaries (Fig. 5). The abundant cut- and-fill structures at the contact between the sediment slug and underlying sediment (Fig. 10) is also consistent with erosion of previously deposited alluvium during a flood.

The post-flood geomorphic expression of a relatively large-scale sediment slug includes a coarse bedload channel incised between paired terraces (Miller and Benda, 2000; Nelson and Dubé, 2016). The terraces are partly or entirely composed of sediment deposited as part of the slug. Syn- or post-flood incision into the sediment slug proceeds from upstream to downstream, so that the height of the terraces above the channel decreases downstream to as little as <1 m below the terrace tread (Miller and Benda, 2000; Reneau et al., 2004). The decreasing height of the unit Qa1a terrace tread above the channel of McMillan Creek and the merger of the terrace tread with the channel are typical of the geomorphology of a sediment slug (Figs. 4 and 13; Miller and Benda, 2000).

The large number of landslides in the McMillan Creek drainage basin are a plausible source of large-volume sedimentation associated with sediment slugs. On the basis of geomorphic features and OSL geochronology, it is concluded that the sediment slug was deposited during the California Storm of January 1862.

Interpretation of Unit Qa1a Sequences

Lithostratigraphic and chronostratigraphic correlation of unit Qa1a stream-terrace stratigraphy in the herein named the “southeast sequence” to beds in the herein named “northwest sequence” (Fig. 14) provide insight into sedimentation at 1960 ± 190 yr B.P. The southeast and northwest sequences are within 20 m of each other in an along-channel reference frame (Fig. 14). The sediment slug is present in both sequences and facilitates chronostratigraphic correlation of beds in the southeast sequence to beds in the northwest sequence.

The upper 1–2 m of the southeast sequence is composed of gray sand and gravel of the sediment slug. Beds of gray to tan sand and gravel as well as beds of red sand and gravel constitute the lower 3.5–4 m of the southeast sequence. The minimal presence of clasts smaller than sand and the lack of features indicative of pedogenesis within the red sand and gravel indicate that the red color is not due to pedogenesis or weathering after deposition as unit Qa1a alluvium. The relatively strong red color indicates that the sand and gravel are derived from the red conglomerate and fanglomerate of the Caliente Formation. The relatively coarse character of the lower 3.5–4.0 m of the southeast sequence (Fig. 14; Supplemental Material [see footnote 1]) as well as the absence of buried soils and floodplain facies in the lower 3.5–4.0 m of the southeast sequence indicate that this part of the terrace sequence was deposited from stream flow with little interruption.

In the northwest sequence, a 0.5-m-thick bed composed of red fluvial sand and gravel underlies the gray sand and gravel of the sediment slug (Fig. 14). On the basis of the color and clast composition this bed of red fluvial sand and gravel, it is lithostratigraphically correlated to beds in the lower 3.5–4.0 m of the southeast sequence that are derived from the Caliente Formation and also underlie the sediment slug (Fig. 14).

The upper part of the northwest sequence consists of the sediment slug and the 0.5-m-thick bed of Caliente Formation–derived fluvial sediment underlying an active tributary fan and overlying a 3.5-m-thick sequence of massive, matrix-supported mud through boulder-sized clasts (Fig. 14; Supplemental Material). Therefore, although the lower 3.5–4.0 m of the southeast sequence is composed of fluvial sediment, morphostratigraphy indicates that the lower 3.5 m of the northwest sequence was deposited by debris flow in a tributary fan. The stratigraphy of the northwest and southeast sequences as well as the OSL date on sample GC-2 indicate fluvial sedimentation in McMillan Creek and debris-fan sedimentation at valley margins at 1960 ± 190 yr B.P. The up-section change from debris-fan sedimentation to fluvial sedimentation within the northwest sequence indicates that during the final stages of deposition at 1960 ± 190 yr B.P., lateral erosion followed by fluvial sedimentation occurred where a tributary-fan toe had existed. These stratigraphic relationships are similar to inset relationships in the current landscape of Gillis Canyon where the sediment slug is inset into tributary fans (Fig. 5). These relationships are also consistent with the model of Moody and Meade (2008) of plan-view erosion and vertical accretion caused by large floods.

Lastly, the stratigraphic relationships and the character of the beds that underlie the sediment slug in both the northwest and southeast sequences indicate sedimentation during a large flood at ca. 1960 ± 190 yr B.P. Alternatively, the beds that underlie the sediment slug in both the northwest and southeast sequences indicate sedimentation as a consequence of more than one flood at ca. 1960 ± 190 yr B.P.

Stream-Channel Longitudinal Profile of the Upper 8 km of McMillan Creek

Due in part to an abundance of stream power resulting from base-level fall driven by rock uplift, over hundreds to thousands of years, streams in tectonically active areas commonly develop concave-up and smooth longitudinal profiles (e.g., Pazzaglia et al., 1998; Seybold et al., 2021). The time scale over which the unsmooth nature of and convexities in the upper 8 km of the McMillan Creek longitudinal profile (Fig. 3) have developed and persisted is unknown, which precludes regarding these characteristics as evidence for decoupling of the upper part of McMillan Creek from the effect of base-level fall. The unsmooth and locally convex profile is, however, consistent with a channel decoupled from tectonically driven base-level fall.

Stream-Channel Longitudinal Profile of the Lower 6 km of McMillan Creek

Gillis Canyon is formed in uniformly weak unconsolidated sediment along the lower 6 km of McMillan Creek. No evidence of faults (active or inactive) was identified in this part of Gillis Canyon, which is consistent with the map of Dibblee (1974). Therefore, the convexity that extends from lowermost location where the channel of McMillan Creek crosses Gillis Canyon Road (LRC referred to in Fig. 6) to 12,666 m from the drainage divide (Figs. 2 and 6; Table 2) cannot be attributed to the influence of geologic structures or lithology.

An assessment was completed to determine whether or not the road crossing identified as LRC (Fig. 6; Table 2) has imparted an influence on the longitudinal profile of McMillan Creek. The assessment is based on field observations and terrain measurements completed using centimeter-accuracy GPS surveying equipment. Although there is a knickpoint adjacent to and downstream of the road, field observations yielded no evidence for otherwise atypical erosion or aggradation within 100 m upstream and downstream of the road. A longitudinal profile based on the GPS survey data (Fig. 15) shows that the slope of McMillan Creek within 100 m upstream and downstream of the road crossing is very similar. The slopes of the best fit line to the data are 0.0157 (R2 value 0.9922) upstream of the knickpoint and 0.0147 (R2 value 0.9765) downstream of the knickpoint. It is concluded that the influence of the road crossing on development of the convexity in the reach of McMillan Creek that extends from the road to 12,666 m from the drainage divide (941 m from the LRC) is negligible.

The slope of the downstream half of the convexity that extends from the road crossing to 12,666 m from the drainage divide is 0.0161, which is slightly less but close to the range of the slopes (0.018–0.019) of a coherent, in-channel sediment slug reported in Kasai et al. (2004). The greater slope of the sediment slug studied by Kasai et al. (2004) may reflect the setting of the study, which is a bedrock-eroding channel reach that has a catchment of 5 km2 and is close to the headwaters of a mountainous drainage basin. The spatial relationship of the downstream extent of unit Qa1a terraces to the adjacent and downstream, convex-up reach of McMillan Creek indicates that the convexity is a topographic expression of the sediment slug. The relatively steep downstream half of the convexity (Fig. 6) indicates that the sediment slug remains coherent (Lisle et al., 2001; Nelson and Dubé, 2016). This conclusion is supported by the spatial patterns of unit Qa2 sedimentation, which are interpreted in the subsection that follows.

Transport and Deposition of Unit Qa2 Sediment

In this subsection, it is demonstrated that after emplacement during the California Storm of January 1862, the sediment slug subsequently translated ~1 km downstream during at least two episodes of sediment remobilization that caused aggradation at the downstream end of the slug.

The signature of a coherent, migrating sediment slug includes a downstream transition from a single-thread channel that is bordered by terraces to a multithread channel that is substantially wider than the single-thread channel (Miller and Benda, 2000; Nelson and Dubé, 2016). Sediment slugs oversteepen valley floors, which leads to the incision that produces the aforementioned terraces. The sediment that is remobilized by the terrace-producing incision causes aggradation that leads to the development of a multithread channel downstream of the terraces (Miller and Benda, 2000; Reneau et al., 2004; Nelson and Dubé, 2016). In the model of Miller and Benda (2000) and as documented in Reneau et al. (2004), the first multithread reach that forms downstream of the first suite of terraces is also eventually incised, and a second suite of terraces as well as a second aggrading, multithread channel forms. This pattern of aggradation and degradation repeats as the sediment slug migrates downstream.

In the reach between 11,920 and 12,048 m downstream of the drainage divide (195–323 m downstream of the LRC; Table 2), which transitions from a single channel to a multithread channel, the channel of McMillan Creek is adjacent to and incised 1.3–1.5 m below an alluvial surface (Figs. 6 and 8). The character of the alluvial surface is consistent with an abandoned gravelly, multithread channel transforming into a terrace tread. For example, at the downstream end of the alluvial surface, there is a backfilled paleochannel (Figs. 6 and 8). The paleochannel indicates relatively recent avulsion followed by ~1–1.5 m of vertical incision below the alluvial surface. This abandoned alluvial surface is the former multithread reach that formed during the first episode of terrace-forming incision into the slug that produced the unit Qa1a suite of paired terraces.

An active multithread reach of McMillan Creek exists downstream of and adjacent to the abandoned alluvial surface that was the multithread reach that formed during the first episode of terrace-forming incision into the slug. This active multithread reach extends from 12,048 to 12,666 m downstream of the drainage divide (323–941 m from the LRC; Table 2) and has multiple channels separated by longitudinal bars that are composed of cobbles and boulders in a matrix of pebbly sand (Fig. 9). Multithread channels and coarse gravel within a matrix of abundant, relatively fine-grained sediment indicate fluvial aggradation (Miller and Benda, 2000; Stokes et al., 2012; Nelson and Dubé, 2016). It is concluded that the multithread reach of McMillan Creek that extends from 12,048 and 12,666 m downstream of the drainage divide (323–941 m from the LRC) actively aggraded during a flood or floods.

As predicted in models of how sediment slugs migrate downstream (e.g., Reneau et al., 2004; Nelson and Dubé, 2016), the abandoned alluvial surface that was the multithread reach that formed during the first episode of terrace-forming incision into the slug is transforming into a terrace produced by a second episode of incision into the sediment slug. The corresponding multithread, aggrading reach produced by this second episode of incision and that extends from 12,048 and 12,666 m downstream of the drainage divide (323–941 m from the LRC) is the current downstream end of the sediment slug (Table 2). The convexity in the longitudinal profile of McMillan Creek that extends from the road crossing to 12,666 m from the drainage divide (941 m from the LRC; Fig. 6) coincides with the downstream end of the sediment slug and indicates that the slug is coherent (Miller and Benda, 2000; Lisle et al., 2001; Nelson and Dubé, 2016).

It is common for floods occurring after an event that emplaces a sediment slug to cause aggradation (Miller and Benda, 2000; Reneau et al., 2004; Nelson and Dubé, 2016). As documented in previous sediment-slug studies completed in steady-flow streams, the time scale of channel incision and resultant downstream aggradation followed by a second channel-incision and aggradation episode occurring farther downstream is typically less than one decade (Miller and Benda, 2000; Nelson and Dubé, 2016). In a stream where flow is unsteady, incision and remobilization of sediment-slug deposits were observed for 28 yr after initial deposition of the sediment slug (Moody, 2017). The presence of boulder-sized concrete blocks in the unit Qa2 sediment that composes the downstream end of the sediment slug indicate that relatively coarse slug-related sediments were remobilized and deposited after A.D. 1920 and more than 60 years after the California Storm of January 1862.

A reliable indicator of the downstream extent of a migrating sediment slug is evidence of aggradation (cf. Miller and Benda, 2000; Nelson and Dubé, 2016). The most recent slug-related aggradation in McMillan Creek occurred as far as 12,666 m from the drainage divide, which is 1046 m downstream of the lowermost extent of unit Qa1a terraces. Assuming the lowermost extent of Qa1a terraces was the downstream end of the sediment slug when it was first emplaced, the OSL age of the sediment slug (140 ± 20 yr B.P.) yields a downstream translation rate of 6.5 m/yr to 8.7 m/yr. Note that in the model proposed here, the sediment slug would be stationary for decades between storms having sufficient magnitude to generate streamflow that can remobilize slug-related sediment.

Process Response to the 9 January 2023 Storm

The mantle of sand and mud that developed along the entire studied reach of McMillan Creek during the 9 January 2023 storm has two pertinent implications. The first is that stream flow occurred along the entire studied reach of McMillan Creek. The second implication is that the moderate-magnitude stream flow (Table 4) led to aggradation throughout the reach of McMillan Creek that is within the sediment slug. Aggradation is attributed to: (1) the readily mobilized unconsolidated sediment that Gillis Canyon is formed in; and/or (2) the relatively high sediment yield from hillslopes that occurs in Mediterranean climatic settings (Langbein and Schumm, 1958; Gray et al., 2018) where ephemeral streams like McMillan Creek predominate.

Storm-Driven, Reach-to-Reach, Down-System Decoupling in McMillan Creek

The OSL ages of unit Qa1a terrace alluvium correlate with large-magnitude storms that caused floods in Gillis Canyon. The upper 1.5–2.3 m of the terrace alluvium is the sediment slug deposited at 140 ± 20 yr B.P. The OSL age of the lower part of the terrace alluvium is 1960 ± 190 yr B.P., which correlates with floods of sufficiently large magnitude to be detectable in the marine stratigraphy of the Santa Barbara Channel (Hendy et al., 2013; Figs. 1 and 11). The morphostratigraphy of unit Qa1a terraces also indicates flood-driven sedimentation. It is concluded that sedimentation in McMillan Creek since 1960 ± 190 yr B.P. occurred due to large storms that triggered floods.

Sedimentation in lower McMillan Creek has been outpacing valley-bottom denudation since ca. 1960 ± 190 yr B.P. The bed dated at 1960 ± 190 yr B.P. is 1.45 m above the channel of McMillan Creek and 3.3 m below the terrace tread. Therefore, the surface of the valley floor at the sample-collection site (Fig. 4) has been raised by ~3 m relative to the channel of McMillan Creek since 1960 ± 190 yr B.P. The channel of McMillan Creek has not incised below the base of unit Qa1a terrace alluvium, and the only evidence for channel degradation since 1960 ± 190 yr B.P. is localized incision caused by valley-floor steepening related to the emplacement of a sediment slug at 140 ± 20 yr B.P. The sediment mobilized by this localized incision was transported only ~1 km before causing aggradation in the channel of McMillan Creek where it was deposited at the downstream end of the sediment slug as map unit Qa2.

The conceptual model of Harvey (2002) hypothesizes that sediment slugs cause reach-to-reach-scale, down-system decoupling and resultant disruption of sediment transport. In addition to the prediction made in this model, the hypothesized disruption of sediment transport in McMillan Creek since 140 ± 20 yr B.P. is supported by deposition of unit Qa2 alluvium and observations made on 12 January 2023. The history of aggradation deduced from Gillis Canyon fluvial morphostratigraphy and the lack of evidence for degradation indicate down-system decoupling of the upper reaches of McMillan Creek from the lower reaches of McMillan Creek since at least 1960 ± 190 yr B.P. Specifically, little or no sediment delivered to the reaches of McMillan Creek that extend from the drainage divide to the downstream end of the sediment slug (Figs. 2 and 6) is transported beyond the downstream end of the sediment slug. Due to the relatively long periods of time that sediment slugs retain coherency in ephemeral channels, the disruption of sediment transport due to slug-induced, down-system decoupling occurs over longer time periods that ordinary fluvial aggradation.

Storm-Driven, Reach-to-Reach, Up-System Decoupling in McMillan Creek

Although reach-to-reach-scale decoupling causes both up-system and down-system decoupling, the specific role of up-system decoupling caused by sedimentation in connectivity to base level for McMillan Creek is uncertain. The source of the uncertainty is that the location where tectonism causes base-level fall for the McMillan Creek drainage basin is unknown.

Two scenarios are considered here. In the first scenario, base-level fall for McMillan Creek has occurred near enough to Gillis Canyon to influence stream incision in McMillan Creek since 2 ka. In the second scenario, base-level fall for McMillan Creek has occurred too far from Gillis Canyon to influence stream incision in McMillan Creek since 2 ka. In the first scenario, sedimentation-driven, reach-to-reach decoupling of upper McMillan Creek from lower McMillan Creek has influenced the effect of base-level fall propagation since 2 ka. In the second scenario, sedimentation driven, reach-to-reach decoupling of upper McMillan Creek from lower McMillan Creek had no influence on the effect of base-level fall propagation because regional-scale decoupling has isolated the entire McMillan Creek drainage basin from the effect of tectonically driven base-level fall since 2 ka. No data that could facilitate distinguishing the relative plausibility of these two scenarios have been identified.

Sediment Availability versus Sediment Flux out of Source Areas

Source-to-sink studies have shown that over millennial time scales, sediment delivery by drainage basins having catchment areas that are thousands of square kilometers to offshore depocenters is associated with large-magnitude and high-frequency rainfall events (Covault et al., 2010). Such high degree of sediment routing system connectivity was documented in the Santa Ana River drainage basin in southern California (Covault et al., 2010). Although a much greater variety of rock types exist in the Santa Ana River drainage basin than in the McMillan Creek drainage basin, both localities are tectonically active areas having high (Santa Ana River basin) to moderate (McMillan Creek basin) topographic relief and a Mediterranean climate. Therefore, it is likely that sediment yield from hillslopes in the Santa Ana River drainage basin is comparable to or greater than sediment yield from hillslopes in Gillis Canyon. This is noteworthy because in the 25.7 km2 drainage basin of McMillan Creek, which in the source-to-sink framework is a source area, the alluvium that constitutes axial-stream terraces in Gillis Canyon is composed entirely of sediment deposited during large storms with a return frequency of ~2000 years. Therefore, in source-to-sink studies encompassing long time scales and large spatial scales, it should be noted that sediment storage within source areas occurs during times of high fluvial discharge and apparent high degree of connectivity throughout sediment routing systems. It should also be noted that the relative low efficacy of sediment transport in small drainage basins may be related to reach-to-reach decoupling (sensuHarvey, 2002) and sediment slug dynamics as documented in Gillis Canyon.

The millennial time scale over which both up-system and down-system decoupling in McMillan Creek have occurred influence the overall sediment flux as would be analyzed in larger-spatial-scale studies. Quantifying the influence of millennial-time-scale, storm-driven decoupling on sediment flux out of Gillis Canyon is beyond the scope of this study. However, the results of this paper require that over millennial time scales, the amount of sediment delivered to offshore sinks from source areas is less than the amount of sediment available for transport in the source area during that time. Specific mechanisms that influence sediment flux rates include: (1) down-system decoupling caused by storm-driven sedimentation, which temporarily arrests down-system sediment transport (for at least ~2 k.y. in Gillis Canyon), and (2) slower incision rates into drainage-basin bedrock due to up-system decoupling and insulation from the effect of base-level fall (not necessarily driven by storm-related sedimentation), which lower the amount of sediment detached from the bed and made available for transport.

A sediment slug was deposited in McMillan Creek at 140 ± 20 yr B.P. during the California Storm of January 1862. The sediment slug has remained coherent for ~160 yr. In the ephemeral channel of McMillan Creek, minimal or no degradation of the downstream end of the sediment slug occurred between relatively large discharge events. Fluvial aggradation composed of sediment reworked from the sediment slug during relatively large discharge events has sustained the relatively steep, downstream end of the sediment slug.

The response of McMillan Creek to the storm of 9 January 2023 supports the proposed model of sediment-slug dynamics. Sediment slugs in perennial streams are degraded by moderate-magnitude streamflow (Nelson and Dubé, 2016). In the ephemeral channel of McMillan Creek, sediment deposition throughout the slug occurred during a moderate-magnitude storm (Table 4) and increased the volume of sediment composing the slug. The growth of the sediment slug as a consequence of a moderate climatic event underscores a fundamental difference between sediment-slug dynamics in ephemeral streams and sediment-slug dynamics in perennial streams having steady-flow regimes and variable-flow regimes. Lastly, source-to-sink studies encompassing millennial time scales during times of relatively high fluvial discharge in areas having Mediterranean climates should consider the impact of storm-driven sedimentation in source areas. The impact includes reach-to-reach-scale, down-system and up-system decoupling, which lead to sediment storage and in some instances relatively slow rates of erosion.

1Supplemental Material. Detailed description of unit Qa1a sediments and stratigraphy. Please visit https://doi.org/10.1130/GEOS.S.24596337 to access the supplemental material, and contact editing@geosociety.org with any questions.
Science Editor: Andrea Hampel
Associate Editor: Andrea Fildani

John A. Moody made significant contributions to this paper. Thanks to Stuart Lane for encouraging the first author to complete this work. We thank the two anonymous reviewers and Associate Editor Andrea Fildani for suggestions that greatly improved this manuscript. Thanks to Martin Stokes for a review of a previous version of this paper. Special thanks to Weijie Dong, Nate Onderdonk, and Greg Pasternack for help and support in a multitude of ways. Thanks to Emmi Sturckow for creating the digital elevation model of the northwestern Temblor Range. The financial support of the Ticho Foundation significantly helped this study. Thanks to Irv McMillan and Flint Spear for granting access to their ranchlands. Thanks to Dennis Pegelow for directing the first author to Gillis Canyon. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.

Gold Open Access: This paper is published under the terms of the CC-BY-NC license.