Updated cosmogenic chronologies of Pleistocene mountain glaciation in the western United States and associated paleoclimate inferences

Highlights

• Cosmogenic chronologies of the last two Pleistocene mountain glaciations are recalculated using newer production rates and scaling models.

• Pleistocene glaciations in most mountains of the conterminous western U.S. were in phase with global glaciations.

• Oldest cosmogenic exposure ages of penultimate glaciation moraines have a mean of 138 ± 13 ka, corresponding to marine isotope stage 6.

• The structure of the last glaciation includes an earlier culmination at ca. 19.5 ± 2.3 ka and a later culmination at 17.0 ± 1.8 ka.

• Earlier glacier culmination reflects cooling during the LGM, later culmination reflects sustained cooling and/or increased precipitation.

Abstract

Surface exposure dating with terrestrial cosmogenic nuclides (TCNs) has become the primary method for determining numerical ages of Pleistocene mountain glacial deposits and landforms in the conterminous western United States (U.S.) and in numerous other glaciated settings worldwide. Recent updates to models of TCN production and scaling warrant a reconsideration of published exposure ages of moraines of the last two Pleistocene glaciations and associated paleoclimate inferences. Previously reported TCN exposure ages of moraines are recalculated here using newer production rates and scaling models for nuclides helium-3 (³He), beryllium-10 (¹⁰Be), aluminum-26 (²⁶Al), and chlorine-36 (³⁶Cl), in most cases yielding significant differences from originally reported ages. Recalculated TCN exposure ages of moraines of the penultimate glaciation display a high degree of variability for individual landforms, particularly toward the younger end of age distributions, suggesting that exposure history is affected by moraine denudation and that older age modes provide the best estimate of the depositional age of these moraines. Oldest exposure ages of penultimate glaciation moraines are well-aligned among mountain ranges across the western U.S. and yield a mean of 138 ± 13 ka, indicating that mountain glaciation occurred in step with global ice volume maxima during marine oxygen isotope stage 6. On average, terminal moraines of the last glaciation date to 19.5 ± 2.3 ka and correspond to the latter part of the global Last Glacial Maximum (LGM). Down-valley recessional moraines representing prolonged glacial stabilizations or readvances to ≥75% of maximum lengths have a mean exposure age of 17.0 ± 1.8 ka, suggesting that these moraine positions were last occupied during Heinrich Stadial 1. Evidence for multiple glacial culminations during the last glaciation is found in several mountain ranges and likely reflects at least two phases of Late Pleistocene climate: an earlier phase when glaciers attained their maximum length in response to cooling during the LGM, and a later phase when glaciers persisted at or readvanced to near-maximum lengths in response to sustained cold temperatures and/or increased precipitation.

Introduction

In the three decades since the first applications of terrestrial cosmogenic nuclide (TCN) exposure dating to Pleistocene moraines in the western United States (U.S.) (Phillips et al., 1990; Nishiizumi et al., 1993; Gosse et al., 1995a), these methods have become the most widely applied approach to numerical dating of alpine glacial features in the region (Fig. 1). TCN exposure dating techniques have revolutionized numerical dating of glacial deposits and landforms by overcoming some fundamental limitations of radiocarbon in these settings, such as the typical scarcity of preserved organic matter in glacigenic sediments and the general restriction of available radiocarbon ages to glaciolacustrine sediments and outwash that provide only minimum or maximum limiting ages of moraines. TCN dating methods have provided a more precise means of determining times of glacial maxima by enabling direct dating of moraine surfaces, which has led to a dramatic expansion of available chronologies of glacial landforms.

TCN exposure chronologies of moraines of the last glaciation now exist for most major glaciated mountain ranges in the conterminous western United States – the Sierra Nevada (e.g., Phillips et al., 1990, 1996; 2009; Nishiizumi et al., 1993; Rood et al., 2011), the Cascade Range and inland Pacific Northwest (e.g., Licciardi et al., 2004; Porter and Swanson, 2008; Speth et al., 2018), portions of the Great Basin/Basin and Range (e.g., Laabs et al., 2013; Wesnousky et al., 2016), the Colorado Plateau (Marchetti et al., 2005, 2007, 2011) and the Rocky Mountains (e.g., Gosse et al., 1995a, b; Phillips et al., 1997; Licciardi et al., 2001; Benson et al., 2005; Brugger, 2007; Licciardi and Pierce, 2008, 2018; Laabs et al., 2009; Ward et al., 2009; Young et al., 2011; Leonard et al., 2017a; Brugger et al., 2019a, 2019b) (Fig. 1). Many of these studies have focused on dating terminal moraines exclusively as a means of determining the timing of glacial maxima. In studies where multiple moraine crests were dated within single glaciated valleys or within a single mountain range, millennial-scale differences in the timing of glacial maxima and/or the timing of subsequent ice retreat have been found (e.g., Licciardi et al., 2004; Guido et al., 2007; Licciardi and Pierce, 2008; Laabs et al., 2009; Dühnforth and Anderson, 2011; Young et al., 2011; Leonard et al., 2017a; Marcott et al., 2019). These and similar studies have permitted detailed reconstructions of climate during and following the last glaciation in the western U.S. (e.g., Plummer and Phillips, 2003; Laabs et al., 2006; Refsnider et al., 2008; Leonard et al., 2017a; Quirk et al., 2018).

A smaller number of TCN exposure chronologies have been developed for moraines that predate the last glaciation; these include records in the Sierra Nevada (Rood et al., 2011; Wesnousky et al., 2016; Phillips et al., 2009), the Cascade Range (Porter and Swanson, 2008; Speth et al., 2018), the Great Basin (Laabs et al., 2013), the Colorado Plateau (Marchetti et al., 2011), and the Rocky Mountains (Phillips et al., 1997; Licciardi and Pierce, 2008, 2018; Dahms et al., 2018; Brugger et al., 2019a; Schweinsberg et al., 2020). These studies have correlated moraines of earlier glaciations to known glacial and stadial events of the Middle and Late Pleistocene while acknowledging potential issues regarding exposure history (e.g., burial of moraines by snow or sediment, moraine and boulder surface erosion, or intervals of prior exposure). Most of these studies have focused on the penultimate glaciation in mountains of the conterminous western U.S., finding that this episode corresponds to MIS 6 (191-130 ka; Lisiecki and Raymo, 2005) in numerous mountain ranges (Rood et al., 2011; Licciardi and Pierce, 2008, 2018; Brugger et al., 2019a; Schweinsberg et al., 2020). Other studies have not ruled out the possibility that ice advances occurred during subsequent cold intervals such as MIS 5d or 5b (123-109 ka and 96-87 ka, respectively; Lisiecki and Raymo, 2005) due to the broad range of exposure ages from these surfaces (Phillips et al., 1997; Porter and Swanson, 2008; Wesnousky et al., 2016). A small number of studies have identified signals of mountain glaciation in the conterminous western U.S. during MIS 4 or MIS 3, and these are spatially restricted to coastal settings in the Pacific Northwest and some glacial settings in the Rocky Mountains (e.g., Hall and Shroba, 1995; Thackray, 2001, 2008; Sharp et al., 2003; Pierce et al., 2011).

Most TCN exposure chronologies of glacial deposits are based on isotope production models that have been refined incrementally through numerous efforts to calibrate production rates for TCNs such as ³He, ¹⁰Be, ²⁶Al, and ³⁶Cl. Such efforts have involved measurements of the TCN inventory in independently dated surfaces suitable for exposure dating at a wide range of latitudes, altitudes, and ages, along with studies designed to improve models for scaling production over space and time (see Borchers et al. (2016) and Marrero et al. (2016a) and references therein for summaries of recent production calibrations). A key product of these efforts is a refined set of TCN production models that can be used to compute more accurate and precise TCN exposure ages of glacial deposits, thereby enabling more reliable comparisons of glacial chronologies to other independently dated geologic and paleoclimatic records. Recently developed production models have been incorporated into the most widely used online cosmogenic exposure age calculators (e.g., Balco et al., 2008; Marrero et al., 2016a; Martin et al., 2017). To date, however, few glacial chronologies for the western U.S. have been recalculated using the newer, more accurate production models implemented in these online calculators (e.g., Leonard et al., 2017a, 2017b; Quirk et al., 2018; Licciardi and Pierce, 2018; Marcott et al., 2019), and the variety of production models used in reports of TCN exposure ages of glacial deposits complicates comparisons of these chronologies. Some recent studies have determined that, at most sites, new production models yield ¹⁰Be exposure ages of a few percent to as much as ∼15% older than ages computed with earlier production models (e.g., Leonard et al., 2017b). The larger shifts in recalculated exposure ages are greater than isotope-measurement errors and suggest that previous correlations between moraine chronologies and independently dated climate changes should be reconsidered.

Mountain glaciers in the conterminous western U.S. are currently limited almost exclusively to cirques in the highest portions of coastal ranges and the Rocky Mountains. The distribution of glaciers across the region shown in Fig. 1 is governed by a modern climate characterized by strong precipitation gradients and contrasting atmospheric circulation patterns between summer and winter. North of ∼40°N latitude, winter (November–March) climate is characterized by westerly flow delivering moisture-laden air derived from the Pacific Ocean (Mitchell, 1976). The mountains of northern California, Oregon, and Washington intercept large amounts of winter precipitation while the ranges further east in northern Utah, northern Colorado, Idaho, Wyoming, and Montana receive less snowfall. South of 40°N latitude, winter airflow is predominantly from the southwest and moisture delivery to the mountains is generally less, due to persistent high atmospheric pressure in the southern Great Basin. Snowfall in the mountains of southern California, the Great Basin, Colorado Plateau, and portions of the Southern Rocky Mountains is more variable and dependent in large part on the El Niño Southern Oscillation (Sheppard et al., 2002; Steenburgh et al., 2013). During the summer (April–October) northwesterly flow is reduced, and much of the region receives less precipitation than during the winter. This is particularly true along the Pacific coast and in the Cascade Range, Sierra Nevada, the Middle and Northern Rocky Mountains, the Basin and Range (in Nevada) and the Columbia Plateau which receive summer moisture only from infrequent intrusions of southwesterly or northwesterly airflow. The southeastern part of the area shown in Fig. 1, including the Colorado Plateau and Southern Rocky Mountains, receives greater amounts of summer precipitation due to the influence of the North American monsoon (Sheppard et al., 2002), and many areas of the U.S. Southwest and Southern Rocky Mountains currently receive more precipitation during the summer than during the winter. Mountain temperatures throughout the region display a strong seasonal contrast as expected for middle latitudes, with the contrast increasing from coastal to inland ranges.

Inferences of paleoclimate in the western U.S. during the last glaciation and early deglaciation are numerous due to the widespread availability of climate proxy records from beyond mountain glacial limits and from numerical simulations of regional climate during the Last Glacial Maximum (LGM) (e.g., COHMAP Members, 1988; Thompson et al., 1993; Hostetler and Clark, 1997; Braconnot et al., 2012). Based on reconstructions of former mountain glacier extents, ice dynamics, and equilibrium-line altitudes (ELAs), the pattern of past glaciation suggests that regional airflow and moisture delivery to mountains was similar to the present, especially in areas far south of the Laurentide and Cordilleran Ice Sheet margins (Fig. 1). Westerly airflow delivered Pacific-derived moisture to the region, nourishing glaciers during the accumulation season and resulting in a pattern of low mountain glacier ELAs in the west and higher ELAs eastward (Porter et al., 1983; Zielinski and McCoy, 1987; Leonard, 1989). Northern mountains closer to ice sheet margins likely experienced larger temperature depressions compared to mountains farther south and were likely drier than modern due to anticyclonic, easterly airflow immediately south of the ice margins (Hostetler and Clark, 1997). Some northern mountains display evidence of muted glacier lengths during the LGM when moisture supply was limited and comparatively greater post-LGM glacier lengths when westerly airflow strengthened as the Laurentide Ice Sheet retreated (Thackray, 2001; Licciardi et al., 2001). Regionwide data-model comparisons support the idea of a drier LGM climate in the northwestern U.S. and a relatively wet southwestern U.S. (Oster et al., 2015a, b; Hudson et al., 2019). The comparatively wetter glacial climate in the southwestern U.S. has been attributed to a southward displacement of the polar jet stream and associated storm tracks (Antevs, 1955), which is a recurring feature in numerical simulations of LGM climate (Kutzbach and Wright, 1985; Thompson et al., 1993; Bartlein et al., 1998). After the LGM, when glaciers in some mountains were at or near their maximum lengths until as late as 15 ka (Licciardi et al., 2001; Thackray et al., 2004; Munroe et al., 2006; Young et al., 2011), Pacific-derived moisture likely increased throughout the region (Lyle et al., 2012; Munroe and Laabs, 2013; Ibarra et al., 2014) and the position of storm tracks varied across a broad range of latitudes (Lora and Ibarra, 2019; Hudson et al., 2019).

In this review, we present a large set of previously reported TCN exposure ages (herein referred to as “exposure ages”) augmented with some new ¹⁰Be exposure ages of Middle to Late Pleistocene (MIS 6 to MIS 2) moraines, all of which are recalculated using updated production rates and a selection of scaling models. The goals of this review are to: (1) synthesize TCN chronologies of the last two major Pleistocene glaciations for mountainous regions in the conterminous western U.S. using up-to-date, consistent, and substantiated production rates and scaling methods; and (2) reconsider existing interpretations of Pleistocene climate change in the region and discuss new scenarios based on recalculated cosmogenic ages of glacial deposits. We note that, although direct and indirect evidence of mountain glaciation between MIS 6 and MIS 2 exists in some locations in the conterminous western U.S. (e.g., Hall and Shroba, 1995; Thackray, 2001; Sharp et al., 2003), most pre-LGM moraines in the conterminous western U.S. have exposure ages corresponding to MIS 5 or 6. Therefore, the term “penultimate glaciation” refers to regional-wide glaciation during early MIS 5 or MIS 6 and not to later stadials that were generally less extensive than the MIS 2 glaciation. This review differs from other previous compilations of exposure-dated moraine ages (e.g., Shakun et al., 2015; Heyman et al., 2016; Palacios et al., 2020) by: (1) focusing only on the contiguous western U.S. and examining more TCN glacial chronologies in this region than any previous review, including new chronologies from penultimate-glaciation moraines in the Ruby Mountains (Basin and Range, Nevada) and from last-glaciation moraines in the northern Colorado Plateau (Utah) and the Bighorn Range (Middle Rocky Mountains, Wyoming); (2) considering exposure ages for the last two Pleistocene glaciations; (3) distinguishing TCN exposure ages of terminal and downvalley recessional moraines (representing near-maximum glacier lengths) of the last glaciation; and (4) examining differences between originally reported and recalculated exposure ages. Although this review includes recalculations of all exposure ages of terminal moraines in the western U.S. published as of this writing, the vast majority of these exposure ages are based on ¹⁰Be (Fig. 1). The following section summarizes refinements to the ¹⁰Be production rate and scaling models as an example of how production models have evolved through the history of TCN exposure dating of glacial deposits and landforms. These improvements have been accompanied by a more accurate determination of the ¹⁰Be half-life (Chmeleff et al., 2010), improved AMS standardizations for ¹⁰Be (Nishiizumi et al., 2007), and reduced ¹⁰Be detection limits with AMS, all of which have helped to improve the precision and accuracy of ¹⁰Be exposure dating. Other commonly used cosmogenic isotope dating methods including ³He and ³⁶Cl have also benefited from similar improvements and advances over the past three decades, but these isotope-specific developments are not reviewed here for brevity and can be found elsewhere (e.g., Goehring et al., 2010; Marrero et al., 2016b).