Lateral bedrock erosion and valley formation in a heterogeneously layered landscape, Northeast Kansas

In this study, we present direct field measurements of modern lateral and vertical bedrock erosion during a 2‐year study period, and optically stimulated luminescence (OSL) ages of fluvial material capping a flat bedrock surface at Kings Creek located in northeast Kansas, USA. These data provide insight into rates and mechanisms of bedrock erosion and valley‐widening in a heterogeneously layered limestone‐shale landscape. Lateral bedrock erosion outpaced vertical incision during our 2‐year study period. Modern erosion rates, measured at erosion pins in limestone and shale bedrock reveal that shale erosion rate is a function of wetting and drying cycles, while limestone erosion rate is controlled by discharge and fracture spacing. Variability in fracture spacing amongst field sites controls the size of limestone block collapse into the stream, which either allowed continued lateral erosion following rapid detachment and transport of limestone blocks, or inhibited lateral erosion due to limestone blocks that protected the valley wall from further erosion. The OSL ages of fluvial material sourced from the strath terrace were older than any material previously dated at our study site and indicate that Kings Creek was actively aggrading and incising throughout the late Pleistocene. Coupling field measurements and observations with ages of fluvial terraces can be useful to investigate the timing and processes linked to how bedrock rivers erode laterally over time to form wide bedrock valleys.


| INTRODUCTION
Bedrock rivers play a central role in landscape evolution by communicating signals of climate shifts, base level changes, and tectonics through landscapes (e.g., Hancock et al., 1998;Lavé & Avouac, 2000;Whipple, 2004;Whipple & Tucker, 1999). Vertical erosion and subsequent channel adjustment in bedrock systems has been a highly researched area for the past 30 years through landscape evolution models (e.g., Snyder et al., 2003;Tucker & Whipple, 2002;Whipple & Tucker, 1999), flume experiments (e.g., Attal et al., 2006;Sklar & Dietrich, 2001), and field studies (e.g., Brocard & Van Der Beek, 2006;Whipple et al., 2000). Despite the substantial advances on processes of vertical incision, comparatively few studies have attempted to resolve controls on the processes and rates of lateral bedrock erosion (Beer et al., 2017;Bufe et al., 2017;Collins et al., 2016;Fuller et al., 2016;Langston & Tucker, 2018;Li et al., 2020;Turowski, 2018). This presents a fundamental knowledge gap in understanding how bedrock rivers respond to climatic and tectonic changes by widening their valleys, and how channel geometry is maintained and changed over time.
Wide bedrock valleys indicate a period when lateral erosion outpaces vertical incision. Evidence of prolonged periods of lateral bedrock erosion in the past are left in the form of strath terraces, flat beveled bedrock surfaces that are capped by fluvial sediments. In order to better use strath terraces to interpret past periods of climate change (e.g., Hancock et al., 1999), the controls on valley widening, and the mechanisms of both vertical and lateral erosion, must be considered together to describe the formation of strath terraces. However, due to an incomplete understanding of the drivers of lateral erosion and controls on rates of bedrock valley widening, it is difficult to interpret how climate, and fluvial processes influenced by climate, drove vertical and lateral river erosion in the past and will continue to shape landscapes in the future.
While some studies suggest specific mechanisms or drivers of lateral bedrock erosion (e.g., Fuller et al., 2016;Johnson & Finnegan, 2015;Langston & Tucker, 2018;Li et al., 2020), few have measured the rate at which lateral bedrock erosion occurs in natural channels (Beer et al., 2017;Collins et al., 2016) nor explored how lateral bedrock erosion rates and other processes influence valley widening rates. Due to the insufficient understanding of these interacting processes, lateral migration and valley widening in bedrock channels is neglected in nearly all existing landscape evolution models (Langston & Tucker, 2018). Furthermore, landscapes with bedrock units of variable erodibility are ubiquitous around the world, yet are seldom systematically studied in landscape evolution models (Barnhart et al., 2020;Forte et al., 2016;Perne et al., 2017). Thus, considerable research is needed to achieve the same level of understanding of processes and mechanisms of lateral bedrock erosion compared to vertical incision.

| LITERATURE REVIEW
Fluvial incision into bedrock is achieved largely through abrasion and plucking (e.g., Whipple et al., 2000), and the dominant mechanism of erosion is often a function of lithological properties of the bedrock (Chatanantavet & Parker, 2009;Hartshorn et al., 2002;Langston & Temme, 2019b;Spotila et al., 2015;Whipple et al., 2000). Erosion via abrasion increases with increasing sediment due to greater number of sediment impacts (the "tools effect"; Attal & Lavé, 2009;Lamb et al., 2008;Sklar & Dietrich, 2004). Continuous sediment addition to the channel can lead to decreased erosion rates, as sediment covers the bed protecting it from sediment impacts (the "cover effect"; Johnson & Whipple, 2010;Sklar & Dietrich, 2001, which in turn may increase lateral erosion (e.g., Turowski et al., 2008). Plucking is the removal of bedrock blocks from the channel bed and occurs when hydraulic lift forces are sufficient to extract blocks from the channel bed (Whipple et al., 2000;Wilkinson et al., 2018). Flows capable of plucking tend to occur less frequently than flows capable of abrasion (Snyder et al., 2003). However, when plucking occurs, erosion tends to be more efficient than via abrasion and can lead to large amounts of erosion in a small amount of time, such as during flood events capable of entraining and transporting large blocks (Baynes et al., 2015;Lamb et al., 2015;Lamb & Fonstad, 2010).
Lateral erosion that outpaces vertical incision can be achieved when sediment covering the bed inhibits vertical incision (Hancock & Anderson, 2002;Johnson & Whipple, 2010;Langston et al., 2015;Turowski et al., 2008). Studies suggest that lateral erosion rates increase in high sediment supply settings, for example, when the channel bed is protected from vertical incision (the "cover effect"), and sediment impacts on the channel banks laterally erodes bedrock (Beer et al., 2017;Li et al., 2020). Lateral erosion through sediment impacts on bedrock walls can occur when sediment trajectories deviate from flow lines and impact the channel wall, from interaction with roughness elements in the channel bed (Fuller et al., 2016) or in channel bends (Beer et al., 2017;Cook et al., 2014;Wohl & Ikeda, 1998). This can happen independently of the sediment "cover effect". A high sediment supply environment also encourages channel mobility, thus driving lateral erosion and the development of wide bedrock valleys (Baynes et al., 2020;Bufe et al., 2016;Langston & Tucker, 2018;Tomkin et al., 2003;Wickert et al., 2013). Greater channel mobility as a result of high sediment load contributes to lateral erosion due to a higher frequency of contact between the river and channel banks (Brocard & Van Der Beek, 2006;Hancock & Anderson, 2002).
Wide bedrock valleys and strath terraces are often found in weak lithologies, such as mudstones or sandstone (Collins et al., 2016;Johnson & Finnegan, 2015;Montgomery, 2004;, suggesting higher lateral erosion rates in weak lithology. Mudstones may erode more rapidly even when not submerged due to cyclical wetting and drying cycles that decrease the tensile strength and cause crumbling upon drying (Montgomery, 2004), and can be easily swept away by the flow (Johnson & Finnegan, 2015). Rapid lateral erosion of weak lithologies often happens via plucking (Langston & Temme, 2019a), yet plucking has been noted as the dominant erosion mechanism even in massive or resistant lithologies, such as quartzite when the resistant lithology is highly fractured (Spotila et al., 2015).
The width of a bedrock valley is additionally influenced by valleywidening mechanisms (Langston & Temme, 2019b), as valleys with differing bedrock strengths or bedrock properties widen through different mechanisms (Spotila et al., 2015;Wohl & David, 2008). Valley widening is achieved when the base of the valley wall is undercut by the river and the overlying material collapses onto the valley floor (Shobe et al., 2016). In soft lithologies, collapsed valley wall material that collects at the base of the wall often consists of small grain sizes or easily weathered blocks that are readily transported away ("erodible mechanism"; Langston & Temme, 2019b). In more resistant lithologies, massive material made of large blocks often collapses into the valley bottom, covering the base of valley wall and halting further valley widening until the collapsed blocks are transported away ("resistant mechanism"; Langston & Temme, 2019b). The valley widening mechanism is also determined by the size of the material released from the valley wall and the transport competence of the river. In natural systems, valley widening may be limited by one of these factors or may occur through a combination of both mechanisms.
Many studies suggest that rivers spend long periods laterally carving wide bedrock valleys that are abandoned by punctuated intervals of rapid vertical bedrock incision (Dühnforth et al., 2012;Foster et al., 2017;Hancock & Anderson, 2002). However, recent studies show that lateral erosion rates can reach tens of meters per year (Collins et al., 2016;Cook et al., 2014) and that dramatic changes in the ratio of lateral to vertical erosion rate are needed to explain some instances of strath terrace formation (Bufe et al., 2017).
Here, we report direct field measurements of lateral and vertical bedrock erosion over a 2-year period in a landscape with heterogeneously layered lithology. We used optically stimulated luminescence (OSL) dating to determine the ages of fluvial material capping a beveled bedrock surface. We found that modern lateral erosion rates in both limestone and shale lithologies at our study site were much higher than originally anticipated owing to several high magnitude flow events during an exceptionally wet year. We also found that limestone and shale erosion rates are a strong function of discharge and wetting-drying cycles. We use these observations to explore the relationship between the modern lateral erosion rates and the OSL dated terrace material to better understand valley-widening mechanisms in this layered landscape.

| STUDY AREA
This study was conducted at Konza Prairie Biological Station and Long Term Ecological Research (LTER) site, here after referred to as Konza.
Konza is a tallgrass prairie ecosystem located in the Flint Hills region of northeast Kansas (Figure 1b). Konza has a mean annual precipitation of 835 mm/yr that primarily occurs between April and September in relatively brief and intense events (Costigan et al., 2015). The elevation of Konza ranges from 317 to 455 m above sea level in an incised landscape, with native tallgrass throughout and gallery forests along the stream channels. The underlying bedrock consists of alternating layers of fractured and jointed limestone and tightly bedded shale of Permian age (Oviatt, 1998). The limestone layers are 1-2 m thick and the shale layers are 2-4 m thick (Macpherson, 1996).
The landscape is largely a product of weathering and fluvial erosion of streams that are tributaries of the Kansas River over significant geologic time, and streams dissecting the landscape reveal the limestone and shale layers (Macpherson, 1996).

Kings
Creek is the main stream draining Konza (Figure 1a) and has a drainage area of $17 km 2. Kings Creek is an intermittent stream, with perennial portions on the main trunk. Kings Creek is a mixedalluvial bedrock river in most locations of the main trunk, with primarily bedrock channels in the upstream tributaries. It typically exhibits high variability in streamflow, with greatest amount of discharge often occurring in April, May, and July and lowest average flows in the late summer and winter resulting in a near-completely dry channel (Gray et al., 1998).

| Fluvial terraces
Several prominent fluvial terrace levels exist surrounding Kings Creek ( Figure 1a) that mark higher elevations formerly occupied by the stream and indicate numerous periods of aggradation and incision. Smith (1991) found that two fill terraces in the downstream most reach of Kings Creek ( Figure 1a) were Holocene in age (8920 AE 120 cal yr BP, upper terrace; 1770 AE 80 cal yr BP lower terrace) from radiocarbon dating. These radiocarbon ages correlate well with previous work establishing fluvial chronologies and stream behavior in the central Great Plains during the Holocene (e.g., Johnson & Martin, 1987;Mandel, 2008). Smith (1991) also made a comprehensive map and correlated different terrace elevations to each other based on sedimentary structures, lithologies, textures, and terraces heights above the channel ( Figure 1a). We identified other high-elevation terraces overlying bedrock in Konza with red, oxidized fluvial material (Supporting Information Figure SB2) similar to our Main Trunk site, as opposed to tan-colored fluvial material in the Holocene terraces described by Smith (1991). While previous work on Kings Creek terraces focused on aggregational terraces in the downstream-most reaches of the stream, strath-like terraces with fluvial deposits capping a bedrock surface also exist within Konza Prairie and the region.
F I G U R E 1 (a) Overview of a portion of Kings Creek, with each study site indicated with a black star. The terrace extent along Kings Creek is in blue, with elevations ranging from 0 to 15 m above the current channel. The black circle in the top left corner shows the location of the two terraces that were dated by Smith (1991)

| Field sites
Three reaches at Kings Creek were selected as study areas for this project ( Figure 1a): The North Fork site, the Nature Trail site, and the Main Trunk site. All sites are located at meander bends where the stream is actively eroding bedrock along both the banks and bed ( Figure SB1), rather than straight segments of the stream because no straight segments with both exposed bedrock bed and banks were found.

| North Fork site
The upstream most study site is located along the North

| Fieldwork
We installed erosion pins in the bedrock banks at North Fork and Nature Trail sites to determine modern erosion rates. At the Main Trunk site, we collected samples from exposed terrace material for OSL dating.
Moultrie Trail cameras were installed facing the stream bank at the Nature Trail and North Fork sites to monitor the water stage at each site, help determine mechanisms of bedrock bank erosion, and identify when limestone blocks collapsed and were transported. The trail cameras took photographs at 30-minute intervals for the duration of the study period. The trail camera photographs were paired with discharge data to determine the water stage during high flow events to help determine how long the study sites were submerged.
Fracture spacing is a key parameter in predicting bedrock erosion rates and block sizes in channels (Moore et al., 2009;Sklar et al., 2017). We measured fracture spacing in both horizontal and vertical directions in limestone layers at sites with erosion pin transects (two transects at the North Fork site and two transects at the Nature Trail site). We selected the measurement sites to coincide with the location of our erosion pins; the lithological sequences were similar along the channel bank between the erosion pin locations such that we considered the sampling area representative. Horizontal

| Erosion pins
Twenty-five erosion pins were installed horizontally in the bedrock channel banks and 11 were installed vertically in the bedrock beds in Spring 2018. At each transect, one or more erosion pins were placed in each of the limestone and shale layers in order to capture variability in erosion rate within lithology sequences and distance from the channel bed ( Figure 4). The erosion pins in the limestone layers were 80 mm-long anchored masonry nails, and in the shale layers erosion pins are 160 mm-long lag screws. Both limestone and shale pins are $6.5 mm in diameter.
Erosion pins were measured six times during the study period, and only five times at the upstream Nature Trail site. Erosion pins were measured after flow events whenever possible to capture the effects. Bed pins often could not be located due to sediment or moss cover; thus, two pins were measured only once each during the study period. Erosion pins were measured using Vernier calipers (accurate within 0.05 mm). The measured length of the exposed pin relative to the channel bank was the total erosion between each visit. One person measured the erosion pins during this study. Each pin was measured at the same location around the perimeter of the pin head to reduce measurement error.

| Luminescence dating
To determine the depositional age of fluvial deposits sourced from a strath terrace we used single grain OSL dating of quartz grains from sediment samples collected at the Main Trunk site. The fluvial deposits that cap bedrock at this site were well-suited for OSL dating, compared to the sediments that cap the bedrock at the North Fork and Nature Trail sites. OSL dating is a technique that estimates the time of burial, or time since quartz or feldspar grains were last exposed to sunlight, therefore yielding a depositional age.
Four samples for OSL dating were collected from the fluvial material exposed on the channel bank: three samples at an upstream location (Site 2; Figure 3) and one sample $20 m downstream (Site 1;  profile from the fluvial material overlying exposed bedrock on the channel bank ( Figure 3b). KNZ003 is the stratigraphically lowest sample, obtained $0.20 m above the bedrock from a massively bedded, fine-grained layer that was sandwiched between layers of imbricated gravel-sized sediments. KNZ004 was collected 1.05 m above the bedrock and was separated from KNZ003 by a thick layer of gravel and cobble-sized material and. KNZ005 was taken 1.55 m above the bedrock and was also separated from KNZ004 by a former channel deposit consisting of gravel-sized sediment. A final cobble layer sits above the KNZ005 sampling layer, indicating that the stream was active at a higher elevation after the deposition of KNZ005.
Samples were processed at the Desert Research Institute Luminescence Laboratory (DRILL) in Reno, Nevada, USA. We used the 90-125 μm fraction size of quartz grains for luminescence dating.
"Pseudo" single grain or "micro-hole" quartz dating (Berger, 2011) was used due to the small size of the quartz grains. Pseudo single grain differs from single grain dating such that in single grain dating, each grain is an aliquot, whereas with pseudo single grain dating, one hole on the measurement disc is considered an aliquot, which were counted to contain $3-10 grains. Measurement procedures are detailed in Supporting Information Tables SA1 and SA2; Figure SA1.

| Erosion rate measurements
Field observations indicate that plucking was the primary erosion mechanism in both limestone and shale lithologies. Plucking was likely the mechanism in the shale lithologies because of the slakey nature of shale ( Figure SB3), and in limestones based on observations of large flakes of rock that were removed from rock surfaces ( Figure Sb4). The highest lateral erosion rates occurred in the shale layers and range from 28 to 159 mm/yr ( Table 2). Most measured lateral erosion pin values were positive and correlate to bank retreat; however, a few negative readings occurred (a total decrease in the length of exposed pin) when small amounts of overlying bank material accumulated around the pin. Seven out of 13 shale erosion pins were removed from the bank by stream flow over the study period. Erosion rates from lost pins were calculated using the entire length of the pin under the assumption that a pin's length of erosion had occurred.
Limestone lateral erosion rates over the measurement period ranged from 0 to 65 mm/yr ( Figure 5; Table 2). Most limestone erosion pins tended to erode very little or eroded in large blocks that completely removed the pin. Six out of the 13 limestone pins on the upper parts of the banks had no measurable erosion and three out of 13 limestone pins were removed from the bank by stream flow during the study period. The greatest magnitude of lateral erosion in both lithologies occurred near the bottom of the banks due to more flow events capable of eroding bedrock, and generally decreased with distance away from the channel bed ( Figure 5; Table 2).

| Bedrock erosion rates versus discharge
Kings Creek exhibits variable stream flow, and there are often long periods in which the stream has zero measured flow (up to 356 days; Costigan et al., 2015). The summer 2019 season was exceptionally wet and was the third wettest summer on record in a 41 years of discharge data (US Geological Survey, 2020a). There were four high flow events in excess of 35 m 3 /s, which is the estimated size of the 5-year flood (US Geological Survey, 2020b) at Kings Creek during the study period ( Figure 6).
We observed that shale erodes more efficiently when it is dry rather than when it is submerged and wet. "Wet" versus "dry" determinations were made using the trail camera photographs and discharge data. Therefore, shale erosion is not only a function of discharge, but also a strong function of wetting and drying cycles.
Conversely, limestone erosion is a function of discharge such that a higher discharge generally results in a higher erosion rate. We summarize the connection between stream discharge and measured erosion during the study period in five time intervals between measurement dates. We report erosion rates in limestone and shale that were calculated over the individual time intervals and averaged over all pin transects. We report "interval-averaged erosion rates" in units of milli-  The D e value of all samples were calculated using the Minimum Age Model (Galbraith et al., 1999).
Note: S lithology indicates that the pin is in shale, LS indicates a pin installed in limestone, and LSB indicates a pin that is installed in an instream limestone boulder. •, pin spotted visually, but could not be measured because water was too deep and/or cold; ♦, pin not found visually due to sediment cover, turbid water or moss cover; ⊠, pin missing. a Erosion rates were calculated over the duration of the study period or over the time period the pin was present in the bedrock bank. If a pin was missing at the time of measurement, we used the entire length of the erosion pin and assumed that was the amount of erosion that occurred for calculations.
time in nearly a year, and the channel began to dry out. During this interval, peak shale lateral erosion rates again were high

| Vertical incision
Erosion pins installed in the channel beds were often impossible to locate due to sediment cover or water in the channel and thus were measured less frequently compared to the bank pins. Bed pins were never measured at the Nature Trail site because of the highwater stage during the study interval; after the end of the study period, three bed pins at the Nature Trail were located that showed 0 mm of erosion. Fork in excess of erosion measured at the single bed pin. Figure SB6 shows $15 cm of vertical incision in the limestone bed in a location $1 m away from the bed erosion pins. Our field observations suggest that all of this $15 cm of vertical incision occurred in T3 and T4. At this location near the bed pins in the North Fork channel, the studyaveraged vertical incision rate was estimated at $88 mm/yr.

| Lateral/vertical erosion ratio
Using measured rates of vertical and lateral erosion, we calculated a ratio of erosion rates (El/Ev) that reflects the relative speed of lateral erosion versus vertical incision. Vertical incision on the bed over the study period ranged from 0 mm at most erosion pins to $150 mm near the bed pins ( Figure Sb6)  lateral erosion rate is dependent on whether or not slaking has occurred between high flow events .

| Large block collapse
Shale erodes most rapidly when it undergoes cyclical wetting and drying cycles that decrease the tensile strength and cause crumbling upon drying (Montgomery, 2004). Once shale is slaked, it can be easily swept away by small flow events (Johnson & Finnegan, 2015;Small et al., 2015).
The highest amount of shale erosion (200 mm/yr, interval averaged across all shale erosion pins), occurred during the T2 time interval, an interval with generally low flow culminating with a large discharge event. The high erosion rates are likely due to a combination of a long, dry period in which weak, slaked shale could develop followed by a large discharge event. During the T3 time interval, shale erosion rate was only 30 mm/yr, despite two large discharge events in this interval. We conclude that this low erosion rate is the result of the stream eroding unslaked shale bedrock, which is more resistant because the shale was submerged and cohesive, and thus had not been slaked. The conclusion that shale erodes most rapidly with frequent wetting and drying is additionally supported by shale erosion rates during the T5 time interval. During T5, the channel at the North Fork site completely dried out and the final two remaining shale erosion pins were completely eroded from the bank, whereas shale erosion rates at the Nature Trail site were low during T5 because the pins were still largely submerged.
Our results point to the importance of event sequencing in bedrock incision and landscape evolution, a phenomenon that has rarely been recognized or documented (Baartman et al., 2013). Our data show that in shale bedrock, and other lithologies prone to slaking, erodes most rapidly not under conditions of high discharge, but under conditions of cycles of shale drying and slaking followed by a moderate discharge event. Conversely, the limestone bedrock in our study area showed no evidence that event sequencing is important in erosion. These results demonstrate the complexities of bedrock erosion processes in different lithologies and suggest that an understanding of antecedent conditions (here, cycles of shale wetting and drying) is important for modeling and predicting bedrock incision.
Limestone is a relatively resistant lithology (Sklar & Dietrich, 2001), and limestone erosion rates during the study period were higher than anticipated, exceeding 100 mm/yr. Higher than anticipated erosion rates were due to exceptional high flow events during the study period ( Figure 6) and bedrock erosion via plucking. It is possible that erosion rates may be higher on meandering sections of the creek, in comparison to straight sections lacking flow obstacles (Beer et al., 2017), given that bedrock erosion is often concentrated on curved channel sections, either due to bedload impacts (Cook et al., 2014) or increased shear stress (Johnson & Finnegan, 2015).
Many limestone layers at Konza are highly fractured, which favors erosion via plucking. In some places, limestone layers grade into shaley limestone that is also prone to erosion via plucking. Additionally, the limestone bedrock may be weakened over time through weathering or abrasion by impacts from bedload particles (Chatanantavet & Parker, 2009). Dissolution of limestone was not observed during the study period; however, it may also contribute to the weakening of the rock and make plucking more efficient (Krautblatter et al., 2012). Given our field measurements and observations, limestone erosion at Konza occurs as a threshold function of peak discharge (e.g., Figure SB8), consistent with erosion via plucking.
The rapid limestone erosion measured during this study period is likely not typical, given the exceptionally high flow events during the study period that were necessary to erode the limestone bed and banks.
Because both vertical and lateral erosion rates were highly spatially variable, there is a wide range of values for the El/Ev ratio that describes the relative speed of lateral versus vertical erosion and the likelihood that a stream can carve an incised canyon versus a wide bedrock valley or strath terraces (Bufe et al., 2017;Merritts et al., 1994;Pazzaglia, 2013). Vertical incision was highly episodic in the study locations due to periodic sediment cover on the bed and the general infrequency of flood events that are capable of plucking limestone. Many studies suggest that vertical incision is much faster than lateral erosion (Dühnforth et al., 2012;Foster et al., 2017;Hancock & Anderson, 2002). However, calculated ratios of El/Ev from our data indicate that lateral erosion rates can equal or outpace vertical incision rates, even in the same lithology, particularly where vertical incision is inhibited by sediment cover on the bed. Previous numerical modeling studies found that substantial lateral channel migration and development of wide bedrock valleys can occur when El/Ev ratios exceed 1.0 (Langston & Tucker, 2018); this data supports potentially higher, if short lived, ratios of lateral to vertical incision.
Our data show that El/Ev in shale banks is consistently equal or much greater than 1, approaching 20. Such high El/Ev ratios are necessary to explain rapid lateral erosion and terrace beveling reported by Bufe et al. (2017). Very high El/Ev ratios can occur in scenarios with weak bedrock banks and a resistant bedrock bed, for example in heterogeneous layered lithology. A potentially common scenario in landscapes with strong contrasts in rock strength among flat-lying lithologic units, is that rivers may be unable to incise into a resistant bed, allowing sufficient time to erode weaker banks laterally.
Our data suggest that even when streams incise into a typically erodible lithology, such as shale, vertical incision rates can be relatively low compared to lateral erosion rates if the lithology only becomes easily erodible via slaking following drying (Small et al., 2015). High El/Ev ratios can also potentially occur where resistant, coarse grained bed material protects a less resistant bedrock channel, for example at the transition from the crystalline core of the Front Range of the Rocky Mountains to the High Plains. In such cases, the resistant, coarse grained bedload protects the shale channel bed from incision, while bedrock banks are exposed to potentially rapid lateral erosion (Langston & Temme, 2019a).

| Valley widening in layered lithologies
The rapid lateral erosion rates measured during this study are likely not representative of long-term lateral erosion rates in these reaches.
The bedrock channel banks and bed in our study area are likely intermittently covered and uncovered by colluvium and fluvial deposits over decadal timescales, which may halt both lateral erosion and vertical incision (e.g., Lague et al., 2005). Bedrock erosion rates are zero not only when bedrock is shielded, but also when there is no flowing water such as during a drought or channel avulsion away from the bedrock valley wall (Brocard & Van Der Beek, 2006;Bufe et al., 2016;Hancock & Anderson, 2002).
Long-term rates of bedrock valley widening also depend on lateral bedrock erosion rate and how effectively a stream can transport collapsed valley wall material (Langston & Temme, 2019a, 2019b. In layered landscapes with a strong contrast in rock properties, the collapse of large or resistant overlying blocks into the stream can effectively shut down lateral erosion on the valley wall if the stream is unable to readily erode or transport the collapsed blocks, as observed at the North Fork downstream site ( Figure SB7). It is likely that continued rapid erosion of shale bedrock at the undisturbed North Fork upstream site will undercut the bank and cause limestone blocks collapse into the stream, shielding the bank from continued rapid erosion at this location as well.
Our data suggest that fracture spacing in the overlying limestone units at each site predicts the potential transport and removal of collapsed blocks from the channel bank (e.g., Gabet, 2020). but also on rock properties of overlying material as well (Brocard & Van Der Beek, 2006;Forteet al., 2016). Our data also show that in Konza and northeast Kansas, varying lateral erosion rates can influence valley widening rates, even in channel banks comprised of a single lithology. Variation in lateral erosion and valley widening rates in the same lithology on annual and decadal timescales can occur due to variable rock properties that make different limestone units more or less susceptible to abrasion or plucking, and may be influenced in part by fracture spacing, which is a first-order control on block sizes of collapsed overlying material.

| Fluvial terraces in Northeast Kansas
The OSL ages from this study are the oldest ages of fluvial material reported from Kings Creek at Konza and northeast Kansas. Most previous studies in northeast Kansas report fluvial material that is no older than Holocene in age (Johnson & Martin, 1987;Mandel, 2008), including fluvial terraces in Konza (Ross, 1995;Smith, 1991). The two other high-elevation terraces overlying bedrock in Konza with red, oxidized fluvial material similar to our Middle Trunk site suggests a longer time exposed to weathering and an older age than tan colored terraces in Konza (Birkeland et al., 1991;Foster et al., 2017). The similarity in color and elevation above stream level suggest that Pleistocene-aged surfaces and strath terraces may be more common than previously indicated by prior studies in northeast Kansas (Johnson & Martin, 1987;Mandel, 2008;Ross, 1995;Smith, 1991). Smith (1991) originally mapped terraces in Konza as one terrace level based on elevation above current stream level ranging from $1.5 m to 6 m. Our OSL ages of the Main Trunk terrace site, laying at $7-8 m above the modern channel, coupled with the varying terrace material demonstrate that terraces similar in elevation above the channel do not always share the same age (e.g., Foster et al., 2017), and in fact have the potential to be quite different in age (Merritts et al., 1994). The OSL ages of fluvial material from this study ($30 ka to 19 ka) and the carbon-14 ( 14 C) ages from previous studies ($9 ka, Smith, 1991;Ross, 1995) (Smith, 1991). Incision of the Main Trunk terrace may have occurred due to changes in mid-continent climate or vegetation during early to mid-Holocene (Knox, 1984;Mandel, 2008 (Dühnforth et al., 2012;Hancock et al., 1999;Molnar, 1994).
Northeast Kansas was generally cool and dry during glacial periods (Baker et al., 2009;Mandel et al., 2016), which could have induced stream aggradation by decreasing the transport capacity of the stream (Schildgen et al., 2016). Stream aggradation may also have been driven by increased sediment flux from hillslopes due frostcracking and freeze-thaw processes, which were more active than widening is likely substantially longer than the earlier mentioned minimum estimates, given that erosion rates in layered landscapes are complex and vary greatly in space and time as shown by our data and previous studies in layered landscapes (Perne et al., 2017). Additionally, complexity in spatio-temporal erosion rates could be due to the fact that our erosion rates were measured at meander bends along the stream, potential erosion hot spots where erosion rates may be higher than straight portions of the channel where it is possible that little or no erosion may have occurred. Furthermore, if long-term lateral bedrock erosion rates were the same as measured modern rates over the Creek, data from our study have the potential to aid in interpreting climate patterns of the late Pleistocene, and how a changing climate influences the timing of aggradation and incision cycles in Kings Creek and other streams on the Great Plains region. Identifying the drivers of present and past change is essential to predict how climate will continue to influence river behavior and shape landscapes in the future.

CONFLICT OF INTEREST
The authors declare that they have no known competing interests.

DATA AVAILABILITY STATEMENT
The data will be available upon request from the corresponding author. ORCID