The Texas Creek rock avalanche is a prehistoric deposit in the Fraser River Canyon, 17 km south of Lillooet, southwestern British Columbia, Canada. Original mapping suggested that the debris consisted of two landslides: a 45 Mm3 event deposited after the Mazama tephra but before about 2 ka ago, and a 7.2 Mm3 event about 1.1 ka ago. The proposed timing of the younger landslide was correlated with a decline in the First Nations population and was proposed as an agent of cultural collapse driven by its impact on salmon returns vital to the population’s sustenance. We provide six surface exposure ages using 10Be from boulder tops, with three samples from each surface that were originally posited to be older and younger debris. The six samples yielded similar ages suggesting the landslide deposit represents a single event with an average age of 2.28 ± 0.19 (2σ external error) ka before 1950 AD. Evidently, the landslide played no role in the cultural collapse. Fraser River Holocene incision rates, estimated pre- and post-landslide are between 13 and 24 mm/yr, consistent with previous estimates for the mid-Fraser River region. Landslide timing is coincident with the explosive eruption of Mount Meager, 120 km to the northwest, and with a possible landslide at Mystery Creek 85 km to the west and 65 km south of Mount Meager. The landslide may have been seismically triggered, but attribution is speculative.

Historic and pre-historic rock slides and rock avalanches are common features of the Coast Mountains of British Columbia (Eisbacher 1979), and several very large events have directly affected the Fraser River (Ryder et al. 1990; Savigny and Clague 1992; Orwin et al. 2004). Landslides affecting the Fraser River may impact linear infrastructure, residential areas, and other natural resources. For example, in 1914, railway construction in the Fraser Canyon caused a landslide that introduced 100 000 m3 of debris at a constriction called Hell’s Gate 110 km south of Lillooet; the blockage impaired salmon returns for decades. More recently, in the summer of 2019, a rock slide occurred at Big Bar, 64 km north of Lillooet. About ∼85 000 m3 of rock debris blocked the river creating a 5 m tall waterfall that inhibited the salmon run that year. Costs for mitigative measures, including blasting the landslide deposit and building a permanent fishway, were estimated at $176 M (CBC 2020; DFO 2020).

Understanding landslide timing, impacts, and triggers is part of a larger project within Natural Resources Canada’s Public Safety Geoscience Program concerned with assessing the frequency–magnitude of rock avalanches and rock slides in the southern Coast Mountains of British Columbia. Here, we report on the age of the Texas Creek rock avalanche (Fig. 1), as it has been previously described (Ryder et al. 1990), with its age inferred from association with possible landslide-induced backwater sediments upstream at Lillooet (Ryder and Church 1986).

The purpose of this paper is to determine the timing of the Texas Creek landslide based on 10Be dating of surface rock fragments within the landslide deposit. Its age is of interest because it has bearing on academic debates in the archaeology of the mid-Fraser region (Hayden and Ryder 1991; Kuijt 2001), on landslide risk affecting human land uses (EGBC 2023), and on ongoing work on river incision in the Fraser Canyon (Venditti et al. 2014; Rennie et al. 2018; Gingerich 2021). The results will provide resolution of the controversy regarding the landslide’s impact on fish resources and cultural collapse, and new data to help understand the nature and timing of landslide interruption and incision rates on the Fraser River.

The study area

The Texas Creek rock avalanche is located 17 km downstream from Lillooet, British Columbia (Fig. 1). Lillooet is in the rain shadow of the Coast Mountains and has a continental climate. Mean annual temperature is 9.5 °C with extremes ranging from 42 °C in July and August to −30 °C from November to February. Mean annual precipitation is 350 mm with 323 mm as rain and 27 mm as snow (snow water equivalent), with a winter peak from November to January, and a second peak in July (Environment Canada 2022). Vegetation consists of Ponderosa Pine and Bunchgrass on the valley bottom from 180 to 300 m asl, Interior Douglas Fir on low- and mid-slopes to 1500 m asl, and Engelmann Spruce–Subalpine Fir up to the alpine at 2000 m asl (Meidinger and Pojar 1991).

In the Coast Mountains, summit elevations reach 2500–3000 m. Within the past 10 Ma, the Coast Mountains have experienced at least 2 km uplift, a rate of 5 mm/yr (Parrish 1983; Farley et al. 2001), with most attributed to isostatic response to Pleistocene glacial erosion (Ehlers et al. 2006), although the rate isostatic uplift is not well constrained. In the last 1 Ma, forced by glaciation, the Fraser River reversed its direction, from northerly to southerly, exploiting the Fraser Fault, and cutting south through the Coast Mountains to the Pacific Ocean at Vancouver (Mathewes and Rouse 1984; Tribe 2005; Andrews et al. 2012). The rate of canyon incision was estimated to be ∼0.2 mm/yr (Andrews et al. 2012). Hillslopes will adjust to river incision and valley deepening by landsliding (Burbank et al. 1996; Larsen and Montgomery 2012). Texas Creek provides an example from the Fraser Canyon.

At Texas Creek, the relief of slopes directly above the Fraser River varies from 800 to 1200 m with sidewall slopes of 50%–70%. Bedrock consists of fault-bounded slivers of sedimentary, meta-volcanic, meta-sedimentary rock, and granodioritic intrusions of Mesozoic age (Journeay and Monger 1994). The valley bottom is ∼1500 m wide and is filled with a 100–200 m thickness of Quaternary sediment recording advance, glacial maximum and retreat phases, and paraglacial fan formation (Church and Ryder 1972; Ryder 1976; Huntley and Broster 1994; Lian and Hicock 2001). The Fraser River is 200 m wide and is inset within the valley fill. Alluvial fan formation at the mouth of Texas Creek has pushed the Fraser River eastward, causing oversteepening of the eastern valley wall, forming the locally known “Big Slide,” and conditioning instability on that slope as indicated by tension cracks near the slope crest (Fig. 1b).

The Texas Creek landslide after Ryder et al. (1990) 

Landform description and geochronology

The landslide is composed of two main rock types, argillite and granodiorite, with lesser amounts of limestone and sandstone ranging in age from early to late Tertiary. The deposit onlaps glacial drift, alluvial fan, and fluvial terraces on either side of the mouth of Texas Creek on the west side of Fraser River (Fig. 1). Based on their detailed mapping and interpretation of the deposits, Ryder et al. (1990) concluded that the deposit consisted of two discrete landslide events, an older event with a volume of 45 Mm3 and a younger event with a volume of 7.2 Mm3.

As there is no Mazama tephra on the landslide deposit, Ryder et al. (1990) concluded it was younger than the Mazama eruption (7682–7584 cal year B.P., Egan et al. 2015). With no direct age dating control, they estimated the age of the landslide based on assumed rates of incision through Quaternary drift, from ∼310 m asl to an assumed river level of ∼190 m asl at the time of the landslide. Ryder and Church (1986) had estimated Fraser River incision rates at Lillooet to be between 10 and 24 mm/yr. Ryder et al. (1990) applied the maximum incision rate of 24 mm/yr to an incision depth of 122 m; accounting for the paraglacial period, they inferred incision started 7 ka ago, to estimate a minimum landslide age of ∼ 2 ka. Based on their observations, they suggested that the older event may have occurred sometime after the deposition of the Mazama tephra, but before about 2 ka ago.

According to Ryder and Church (1986), channel aggradation forced by downstream blockage caused the formation of the lowest two terrace levels at Lillooet, L3-T3 (230–240 m asl) and L4-T4 (210–220 m asl; Fig. 1a). Aggradation was evident in terrace cross-sections where sandy bar gravels were overridden by boulder channel gravels; while a backwater cause was evident from a thickening of the sediment package in the downstream direction. A First Nations fire hearth (archaeological site EeRl-12; Fig. 1a) buried by boulder gravel on a terrace at 220 m elevation, 20 m above river level, provided evidence for a short-lived phase of river aggradation after 905–1300 cal year B.P. (1180 ± 110 14C B.P., GSC-4101). The Texas Creek landslide was viewed as the best candidate for backwatering, and the posited second event was correlated with the short-lived phase of river aggradation after ∼1.1 ka at Lillooet.

Emplacement of the rock avalanche

The topography prior to the landslide consisted of an abandoned surface of the Texas Creek alluvial fan, which had developed on Quaternary valley fill (Fig. 1b). A major unit in the drift is a glaciolacustrine unit locally termed the “Lillooet Silt” (Ryder and Church 1986). In post-glacial time, Texas Creek incised these materials creating an inset terrace 400–500 m wide, before cutting down further to within 6 m of its present level.

The landslide ran across the Fraser River and onlapped the landforms at the mouth of Texas Creek. As intuited by Ryder et al. (1990), when the event overran the inset mid to late Holocene terrace along Texas Creek to gain the upper paraglacial terrace (area A1 on Figs. 1b and 1c, 300–320 m asl), a prominent trough was formed defining the north edge of a ridge-like feature in the debris (area A2 on Figs. 1b and 1c, 310–330 m asl). Subsequent to emplacement and incision of landslide debris by Texas Creek, a deep-seated rotational failure developed in the underlying glaciolacustrine unit, with the failure plane daylighted in the avalanche debris on the inset terrace to form the south boundary of the ridge-like feature (area A2 on Fig. 1b and 1c). The down-dropped surface forms area A3 (Fig. 1b and 1c, 280–310 m asl). Landslide areas B (210–250 m asl), C1 (220–250 m asl), C2 (250–305 m asl), and D (270–290 m asl) consist of lower terraces; B and C along Fraser River’s west bank; and D along Texas Creek. Terrace areas B and C were attributed to the younger landslide.

Area A is composed of both argillite and granodiorite, and the lithologic boundary cuts across morphologic subzones A1, A2, and A3 (Figs. 1b and 1c). The morphology was the product of emplacement, and the lithologic zonation is preserved from the bedrock source area, a common observation for rock avalanches. Areas D and B are composed of argillite, while area C is composed of granodiorite.

Evidence for two landslides

Two lines of evidence were used to infer that two rock avalanches had occurred. First, the existence of a 50–100 m tall river cut scarp between area A and area B was taken as evidence that incision of the river cut scarp followed area A emplacement and preceded deposition of area B. Second, along the Fraser River scarp at area C1, there are gravels found in section between avalanche debris interpreted as fluvial gravels separating the two landslide events (Ryder et al. 1990; their section 4).

Field methods

Argillite is a fine-grained sedimentary rock unsuited for cosmogenic surface age dating because it is friable. Argillite blocks in the landslide debris have broken down to form rubbly hummocks often with a fractured rock core (Fig. 2). Therefore, we targeted areas that contained 1–2 m sized granodiorite blocks selected from open settings with no forest cover (areas A2, C1; Fig. 1; Table 1). To date the posited older landslide, we selected three blocks for sampling on area A2; because area A1 lacks widespread granodiorite debris and area A3 showed evidence of post-depositional instability. To date the posited younger landslide, we selected three large blocks in area C1. Sampling was conducted using hammer and chisel (Fig. 3). Samples were collected from block top surfaces about 1 m above the surrounding ground, with surfaces selected to avoid areas with evident spalling. Each sample consisted of about 3 kg of rock chipped from the upper 3 cm. Most samples were from upward facing, ∼level facets, with two from sloping facets (TCQ5-1, TCQ5-2) with dip and dip direction recorded. The site location was recorded by handheld GPS. Topographic shielding was estimated by GIS with angle to horizon measured at 15° azimuth increments and with the inflection method (azimuths to positions on the skyline where slopes change significantly). The surface dip and horizon shielding data are used to compute the effect (shielding factor) on average cosmic ray flux to a sampled surface, using “The online calculators formerly known as the CRONUS-Earth online calculators” henceforth written as the Washington CRONUS Calculator (see Supplementary Data). The average dip and topographic shielding factors were 0.980 ± 0.004 (unitless) for the four boulders without dip, and 0.986 and 0.914 for the dipping surfaces (TCQ5-1 and TCQ5-2, respectively). The difference in the two topographic shielding methods was less than 1% for each boulder so we averaged their values for each sample.

Geochronology methods

Details of the laboratory, AMS, calculation, and uncertainty analysis procedures are provided in the Online Supplementary Data. Approximately 30 g of quartz was concentrated and purified from each sample, Be carrier added, the mixture digested, Be extracted, and targets of BeO for each sample and a process blank were loaded in stainless steel target holders at CRISDal Laboratory, Dalhousie University in August 2022. The targets were shipped to Center of AMS at Lawrence Livermore National Laboratory for the measurement of 10Be/9Be. Reduction of field, laboratory and AMS data, blank subtraction, adjustments to production rates, and error analyses were conducted at CRISDal Lab. The exposure age of each boulder surface was calculated using Washington CRONUS Calculator (version 3, see Table SD-3) and is interpreted to represent the timing of their deposition, and therefore the age of the underlying landslide deposits.

The reported ages are expressed as ka BP (1950 = 0 ka). We report relative to 1950 for comparison with calibrated 14C ages (IntCal20; Reimer et al. 2020) cited herein.

We calculate the exposure ages for three scenarios (Table 2): Scenario 1, the exposure age with zero erosion and zero snow cover; Scenario 2, the exposure age with 1 mm/ka erosion but zero snow cover; and Scenario 3, the exposure ages with 1 mm/ka and an estimated average snow cover (Table 2). The ages are provided with internal and external uncertainty (1σ), as per the Washington CRONUS Calculator. Note that the correction for erosion only added 4 to 6 days to each age, which is expected considering the short exposure duration. Likewise, the impact of snow cover was similarly low, adding a decade. We consider this as reasonable, based on average of 10 cm of recorded snow on the ground during December and January at Shalalth (Stn ID 1117215, 244 m asl, 1963–2004; Environment Canada 2022) and the height of the sampled boulders. Adjustment for thicker or longer snow cover could result in many decades, but not likely a century-level effect.

To compare the age of the potentially different landslide deposits, we will refer to the erosion and minimum snow-adjusted ages (Scenario 3; Table 2), and use the internal error (as the production rate systematic errors related to scaling, and temporal geomagnetic field effects essentially will be the same for all samples). If all six samples are considered to date one event, the mean age (corrected for erosion and snow) is 2.28 ± 0.13 1σ (0.19 2σ) ka with no outliers. We interpret this exposure age to be the age of the combined deposit. Treating the deposit as two separate landslide events, the most probable age of the three TCQ5 boulders is 2.26 ± 0.04 1σ (0.14 2σ) ka (no outlier). The mean age of the three TCQ1 boulders is 2.30 ± 0.18 1σ (0.55 2σ) ka, which overlaps the 1σ confidence age of range of TCQ5 boulders. As the posited younger surface TCQ1 cannot be older than the posited older surface TCQ5, and since they are statistically similar, we consider the deposits to represent a single event with an average age of 2.28 ± 0.13 1σ (0.19 2σ) ka.

If we have underestimated boulder erosion rate, snow cover, or the shielding effects of forest cover, or if the boulders have been exhumed from under the original deposit over centuries, this most probable age may be an underestimate for the age of the deposit. However, the consistency of the exposure ages on boulders suggests a close-bracketing minimum age.

One or two landslide events

The age results indicate that there was likely only a single landslide rather than two landslide events. We argue that the existence of the scarp between area A and area B (Fig. 1c) is not evidence that time was required between two landslide events to form the scarp; since the Fraser River had downcut to near its present position before the landslide took place. We interpret that the leading edge of the deposit simply ran up the scarp, depositing a sheet on the upper terrace (areas A, D; Figs. 1b and 1c), with the trailing debris (areas B, C; Figs. 1b and 1c) being confined by the Fraser River escarpment slope. Moreover, where the gravels are found interbedded with avalanche debris, the river cut section in area C1 could have developed by scarp collapse during the stream incision process, and need not have recorded multiple landslide events from the source area.

The Texas Creek landslide and cultural collapse

Starting as early as 4 ka ago, during the winter season, First Nations peoples in the interior of BC resided in semi-subterranean “pithouse” villages typically arranged from small clusters ranging up to large sites of 100–200 dwellings in areas with good access to anadromous fisheries; this has been named the Plateau Pithouse Tradition (Richards and Rousseau 1987). The number and size of pithouses reached an apogee during the Classic Lillooet Phase, just prior to ∼1.1 ka ago, and then declined significantly (Richards and Rousseau 1987; Chatters 1995; Pokotylo and Mitchell 1998).

These villages were highly dependent on salmon as a food resource. Hayden and Ryder (1991) proposed a catastrophist cause for the population decline, i.e., landslide-river interruption causing a decline in salmon returns. The inferred age for the posited younger landslide at Texas Creek, ca. 1.1 ka, was said to correlate with the sharp collapse in the size of pithouse villages upstream along the mid-Fraser River region (Fig. 1a). The landslide-induced cultural collapse hypothesis sparked a long-standing debate on the timing and causes of changes in winter village sizes and social inequality in the mid-Fraser region (Kuijt 2001; Hayden and Ryder 2003; Prentiss et al. 2003; Kuijt and Prentiss 2004; Hayden and Mathewes 2009; Prentiss et al. 2011).

Paleo-demographic studies use compilations of radiocarbon ages from samples collected from cultural sites as a proxy for population and often attribute external factors (earthquake, landslides, and climate changes) as drivers for changes in population density or continuity (e.g., Hutchinson and McMillan 1997). In the mid-Fraser region, Prentiss et al. (2003) and Prentiss et al. (2008, 2012) compiled village-scale demographies for Keatley Creek and Bridge River sites, respectively (Fig. 1a), and these revealed similar, but slightly different timing for the rise and fall of the Classic Lillooet Phase. Morin et al. (2009) provided a compilation from all known village sites in the mid-Fraser region (Fig. 1a). Their data indicated early occupation ∼2.35 ka ago, with population flourishing between 1.75 and 1.25 ka ago, followed by decline, and reoccupation from ∼0.5 ka ago to European contact.

Outside of the mid-Fraser region, other paleo-demographic studies have also identified population flourishing ∼2.0–1.2 ka ago with subsequent decline: in the lower Fraser River (Lepofsky et al. 2005; Ritchie et al. 2016) and in the Interior Plateau of BC and Washington (Prentiss et al. 2005; Ames 2012). This would suggest a regional driver. Hutchinson and Hall (2020) have recently made a strong case for climate-induced human population change on the upper Columbia River at Kettle Falls, Washington. They documented larger populations 1.7–1.3 ka ago and 0.8–0.5 ka ago, with an intervening lull, and they correlated the larger population size with increased salmon abundance, via cooler sea surface temperatures in the Pacific Ocean, and with episodes of glacial advance in the local mountains.

Our 10Be cosmogenic surface ages (Table 2) demonstrate that the landslide occurred 2.3 ka ago, before the rise of the Classic Lillooet Phase. In fact, the population decline ∼1 ka later was likely part of a larger regional trend affecting the lower Fraser River and the Fraser and Columbia River portions of the Interior Plateau (Morin et al. 2009; Ames 2012; Ritchie et al. 2016; Hutchinson and Hall 2020), and is likely related to climate variations, possibly by warming of sea-surface temperature subsequently reducing salmon runs (Hinch et al. 1995; Hutchinson and Hall 2020) and facilitated by other environmental/sociopolitical circumstances (Ames 2012; Prentiss et al. 2014).

River incision and landslide interruption

Clague et al. (2021) provided an age of ∼11 ka for a catastrophic glacial lake outburst flood affecting the Fraser River canyon. At Texas Creek, this flood scoured the glaciofluvial terraces standing at 305–350 m asl (FGt; Fig. 1c); and assuming a river level at the time of the landslide of ∼190 m asl, then 115 m of incision occurred between 11 and 2.3 ka, spanning ∼8.7 ka, yielding a long-term incision rate of 13 mm/yr. Or as done by Ryder and Church (1986), accounting for the paraglacial period (starting incision at 7 ka) gives 115 m incision spanning 4.6 ka, or a long-term incision rate of 24 mm/yr. These incision rates are identical to those estimated by Ryder and Church (1986), and, while the same order of magnitude, are 20%–50% less than the 55 mm/yr (based on OSL dating) reported by Gingerich (2021) for the Big Bar area 50 km north of Lillooet.

Post landslide incision rates are estimated from a maximum barrier height of 230 m asl and modern river level of 185 m asl, or 45 m depth over 2.3 ka, yielding an average incision rate of 20 mm/yr. This is within the range of the estimated early to mid-Holocene incision rates.

Hewitt (2006) defined the typical phases of a landslide–river interruption epicycle whereby a landslide forms a barrier, leading to upstream aggradation with possible downstream erosion or sedimentation, followed by incision and removal of the impoundment complex accompanied by downstream sedimentation, and finally the exhumation of buried valley fill and incision into the pre-landslide valley floor.

At Texas Creek, 2.3 ka ago, the landslide may have formed a persistent barrier. Ryder et al. (1990) noted that the surface of area A at ∼300 m elevation does not support channeling indicative of post-event erosion, but area B does, with the maximum barrier height of about 230 m or about 45 m above river level. Kuijt (2001) noted that archaeology site EeRl-171 (Fig. 1a) consisted of cultural materials at 230 m asl and dated to 4185 ± 205 14C year B.P. (5290-4220 cal year B.P.) and capped only by aeolian sediment; the ∼4.7 ka age is a maximum for the barrier elevation of 230 m asl for the Texas Creek landslide.

Ryder et al. (1990) estimated that the reservoir would fill within a day during freshet, or at most a week during winter low flows. Overtopping of the barrier would likely have resulted in dam break or rapid downcutting. However, downcutting may not have returned the channel to its former pre-slide elevation. The accumulation of lag debris may have formed a persistent, but lower elevation barrier. If a persistent barrier formed, then according to Ryder and Church (1986), the lake would have infilled rapidly within a few years and this would have created river flats extending upstream of the barrier. A review of Google Earth imagery between Texas Creek and Lillooet reveals a number of target terrace surface elevations (Fig. 1a) that may be legacy of infill and exhumation from the Texas Creek landslide.

Landslide barriers may play a significant role in mediating bedrock erosion (Korup 2006; Hewitt et al. 2011). However, while the Texas Creek landslide barrier may have persisted decades to centuries, after 2.3 ka it appears effectively removed. Incision rates of tens of mm/yr, over centuries to millennia, represent erosion through mostly sediment accumulations: drift in the pre-landslide era and colluvium in the post-landslide era. Over millions of years, incision rates are on the order of ∼0.1–1.0 mm/yr, reflecting erosion through bedrock. Thus, in the long-term scale of orogeny and bedrock canyon formation, the Texas Creek landslide may not be a significant impediment to erosion.


The bedrock lineaments above and extending north of the existing headscarp indicate the accumulation of gravitational stress (Ryder et al. 1990), and this may condition the slope for seismic triggering. We provide a most probable age of 2.3 ± 0.13 1σ (0.19 2σ) ka for the deposition of the landslide.

The 2.3 ± 0.19 ka age overlaps the timing of the Mount Meager Plinian-style volcanic eruption (Volcanic explosivity index 5; VEI5) that occurred 2360 cal year B.P. (2349–2704 cal year B.P.; Clague et al. 1995) and 120 km northwest of the Texas Creek landslide scar. The most probable upper bound age is four decades less than the lower bound of T6 Cascadia earthquake (2617–2506 cal year B.P.; Goldfinger et al. 2012); these events might be coincident if the shielding factors were underestimated for Texas Creek. We note that the Mystery Creek landslide near Whistler (Blais-Stevens et al. 2011), located 85 km west of Texas Creek and 65 km southeast of Mount Meager, dated using 36Cl, yielded three ages of 4.9 ± 1.3 ka, 4.5 ± 1.8 ka, and 2.4 ± 0.5 ka. While interpretation of the Mystery Creek samples allows for four landslide age scenarios (Blais-Stevens et al. 2011); in one scenario, the single 2.4 ka 36Cl sample may represent a landslide event also coincident with the Mount Meager volcanic eruption. The coincident timing of these three events is speculative as the inherent sample and laboratory errors, and the associated confidence limits span decades to several centuries.

This potential coincidence of two landslides and volcanic eruption raises the question, could a VEI5 eruption trigger large rockslope failures 65–120 km distant? To our knowledge, globally, this has not been previously reported in the literature. Could some, as yet un-documented crustal earthquake, or the T6 Cascadia earthquake, have triggered landslides at Mystery, Texas, and Meager creeks, unroofing the vent (Roberti et al. 2020), thereby triggering landslides and an eruption? These are questions for future research.

We sampled two discrete landslide surfaces originally thought to represent two separate landslide occurrences separated by centuries to millennia. However, the most probable exposure ages for each surface (TCQ5: 2.26 ± 0.04 1σ (0.14 2σ) ka and TCQ1: 2.30 ± 0.18 1σ (0.55 2σ) ka) overlap at 1σ internal error and have a mean (n = 6) exposure age of 2.28 ± 0.13 1σ (0.19 2σ) ka. We interpret this to represent a close-limiting minimum age for the deposit.

We conclude that the Texas Creek rock avalanche debris represents a single large landslide of 52 Mm3 volume, as previously estimated for the total deposit. The landslide may have formed a barrier that persisted for centuries, with an original crest of about 230 m asl, blocking the Fraser River long enough to produce “interruption epicycle” fluvial terraces at least 17 km upstream as far as Lillooet. The barrier, ∼fully eroded in 2.3 ka, did not persist long enough to significantly impede bedrock erosion over millions of years.

Occurring 2.3 ka, the Texas Creek landslide cannot be implicated as a factor for the demise of the prehistoric Classic Lillooet Phase culture that occurred shortly after 1.2 ka. Based on regional evidence for the population decline, both downstream in the Fraser Lowland and Salish Sea, and across the Interior plateaus of the Fraser River and Columbia River systems, it is suggested that other factors, likely climate-mediated salmon declines, are the most likely explanation.

With respect to potential triggers, the mean age correlates with the Mount Meager volcanic eruption, roughly 120 km northwest. There is also potential correlation with the Mystery Creek landslide, 85 km west. These correlations are tenuous because of the distances separating events and the error margins associated with the dating methods. While the Texas Creek landslide may have been seismically triggered, further research is required to explore these noted coincidences. Failure may be due to other factors (climate, time, etc.) forming the cause and trigger.

Considering future hazard and risk, occurrences of large landslides in the Fraser Canyon could have rapid, catastrophic, and persistent geomorphic impacts on the Fraser River. As well as direct impacts to linear infrastructure and residential areas, persistent barriers are impediments to salmon, and present backwater inundation and outburst flood hazards.

Future work on geomorphology of the mid-Fraser River would focus on identifying, mapping, and dating other landslides, understanding the stratigraphy and geochronology of potential interruption-epicycle fluvial terraces, and examining the effect of landslide barriers on the rate of bedrock incision. Results from dating other rock avalanches in the area may bring to light coincident events suggesting a seismic trigger.

We thank David Kallai for fieldwork support. At CRISDal Lab, Alexis Imperial completed the mineral separation and ICP-OES measurements, and Guang Yang the target chemistry. Alan Hidy supervised the AMS analyses at Lawrence Livermore National Lab. We thank G. Brooks, O. Lian, and two anonymous reviewers for providing comments and suggestions to the manuscript. Ariane Castagner helped in editing, literature review, and calibration of 14C ages.

Data generated or analyzed during this study are provided in full within the published article and its supplementary materials.

Conceptualization: PAF

Data curation: PAF

Formal analysis: PAF

Investigation: PAF, JCG

Methodology: PAF, JCG

Project administration: AB

Resources: AB, JCG

Supervision: AB

Writing – review & editing: PAF, AB, JCG

This work was funded by Natural Resources Canada’s Office of Energy Research and Development (GSC-19-103) and the Public Safety Geoscience Program within the Geological Survey of Canada. JCG acknowledges support from CFI-MSI No. 42447 and NSERC-DG-06785-19.

Supplementary data are available with the article at

This work is licensed under a Creative Commons Attribution 4.0 International License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author(s) and source are credited.

Supplementary data