Late Pleistocene and early Holocene sea-level history and glacial retreat interpreted from shell-bearing marine deposits of southeastern Alaska, USA Open Access

The northern Pacific margin of the United States and Canada has undergone rapid and spatially variable changes in sea level since the Last Glacial Maximum (Clark et al., 2009) owing to the complex interactions between climate and the solid earth (Shugar et al., 2014). These interactions are numerous, including isostatic adjustment caused by the advance and retreat of the Cordilleran Ice Sheet (CIS; e.g., Lesnek et al., 2020), eustatic sea-level change related to the decay of global glacier complexes (Lambeck et al., 2014), and tectonism along the Pacific–North American plate boundary (e.g., Barrie et al., 2021). Regionally resolved reconstructions of sea-level fluctuations can provide insight into the relative roles of these processes on landscape change.

Sea-level reconstructions also serve as a key component of investigations into early human occupation of the northwestern North American Pacific coast, including the initial migration into the Americas during the Late Pleistocene (Braje et al., 2020). Detailed sea-level histories and high-resolution topographic and bathymetric data have successfully guided archaeological surveys near the modern Pacific coastline (Baichtal and Carlson, 2010; Carlson, 2012; Carlson and Baichtal, 2015; McLaren et al., 2014, 2018, 2020), leading to the discovery of dozens of cultural sites across southeastern Alaska (USA) (Carlson, 2012; Carlson and Baichtal, 2015) and British Columbia (Canada) (Letham et al., 2016; Fedje et al., 2018; Mackie et al., 2018). Indeed, these techniques have uncovered the oldest evidence to date of human presence on the Pacific coast of North America: a set of footprints on Calvert Island, British Columbia, that date to ca. 13,200 calibrated yr B.P. (McLaren et al., 2014, 2018). These findings demonstrate that highly resolved sea-level reconstructions are a powerful tool in the search for ancient cultural sites along the northwestern North American Pacific coast.

Here, we present an updated postglacial relative sea-level history for coastal southeastern Alaska (Fig. 1). We begin by documenting the reported and known occurrences of shell-bearing raised marine sediments (hereafter referred to as shell-bearing strata) across southeastern Alaska, from Yakutat to the northern shores of Haida Gwaii, British Columbia (Fig. 2). We then present 45 new ¹⁴C ages on shells from these deposits. Combining this data set with previously published sea-level constraints, we develop new sea-level curves for 11 areas in southeastern Alaska, which give insight into the region's complex isostatic and eustatic sea-level history. We also report the presence of environmental (i.e., non-anthropogenic) charcoal and Pacific sardine (Sardinops sagax) in select deposits, which we use to infer the timing of the regional Holocene Climatic Optimum. Finally, we discuss the potential impact of these sea-level changes on early human occupation of coastal southeastern Alaska.

Our study area includes the lands within the boundaries of the Tongass National Forest and Glacier Bay National Park, specifically from Yakutat Bay to Dixon Entrance and the northern shores of Haida Gwaii, and from the Queen Charlotte fault into the transboundary river valleys of British Columbia (Fig. 2). The study area roughly includes all of the Alexander Archipelago, extending 850 km from the northwest to the southeast and 270 km west to east (Fig. 1). The Alexander Archipelago includes ∼1100 islands and 16,000 km of coastline. The landscape of the archipelago rises abruptly from the sea, commonly reaching elevations of 1000 m above sea level (asl) within 2 km of tidewater. The region receives nearly continuous storms, owing to a semi-permanent low-pressure system in the Gulf of Alaska. Precipitation amounts vary from 150 to 750 cm of precipitation annually. Bedrock geology in southeastern Alaska consists mainly of volcanic and sedimentary deposits representing a succession of magmatic arcs ranging in age from the Neoproterozoic to the present, intruded by plutons related to the formation, deformation, and rifting of the successive arcs (Wilson et al., 2015). The Neoproterozoic, Paleozoic, and Mesozoic arcs were long-lived oceanic arcs (S.M. Karl, 2020, personal commun.), which today exist on a complex assemblage of accreted terranes. The morphology of the mountains in the region, as well as that of the many straits, passages, and channels that separate the islands, broadly reflects the underlying geologic structure but has been greatly modified by Pleistocene glaciation.

During the Late Pleistocene, much of southeastern Alaska was covered by the CIS (Dyke, 2004). Recent investigations have provided chronological constraints on both the extent and retreat of the CIS in this region. Cosmogenic ¹⁰Be ages from the westernmost islands of southeastern Alaska and ¹⁴C-dated animal bones from an ice-overrun cave on northern Prince of Wales Island suggest the maximum extent of the CIS occurred between ca. 20,000 and ca. 17,000 cal. yr B.P. (Lesnek et al., 2018; ¹⁴C ages were recalibrated using the Marine20 calibration curve and an updated marine reservoir correction; see below). Additional ¹⁰Be ages from other islands around southeastern Alaska indicate that the fjords and straits were deglaciated by ca. 14,900 cal. yr B.P., after which the CIS margins transitioned to being primarily land-terminating (Lesnek et al., 2020). Small, independent, remnant ice caps may have persisted in high-elevation areas until the early Holocene (Lesnek et al., 2020), although the history of these glaciers remains poorly constrained.

Today, within the bounds of the study area, spanning the United States and Canadian Coast Range, the Tongass National Forest, and Glacier Bay National Park, glaciers cover >28,000 km² (RGI Consortium, 2017). Baranof Island, in the western portion of the archipelago, hosts small glaciers that are likely remnants of Little Ice Age (1250–1850 CE) glacial advance (Gaglioti et al., 2019). Many areas throughout southeastern Alaska are still rebounding as the result of post–Little Ice Age melting, with uplift rates varying from 10 to 32 mm yr⁻¹ (Larsen et al., 2004, 2005). Continued thinning of the existing tidewater glaciers is resulting in accelerated uplift, particularly in the vicinity of Juneau (Larsen et al., 2004, 2005, 2007; Trüssel et al., 2013).

Shell-bearing fossiliferous marine deposits have long been observed and recorded throughout southeastern Alaska (Buddington and Chapin, 1929). These deposits have been referred to as uplifted marine sediments and terraces, fossiliferous marine gravel, sand, and clay, and glaciomarine sediments. The first detailed description of such deposits on Douglas Island came from W.H. Dall on the 1899 Harriman Alaska Expedition. Dall (1904) recognized the similarity between these sediments in southeastern Alaska and Pleistocene marine deposits in eastern North America. Subsequent research focused on cataloging the occurrences of raised marine deposits around Juneau (e.g., Knopf, 1912; Twenhofel, 1952). Miller (1972, 1973a, 1973b, 1975) characterized these facies and named these shell-bearing strata the Gastineau Channel Formation. Others further documented elevated shell-bearing marine clays in their evaluations of earthquake and other geologic hazards to coastal communities of southeastern Alaska (Lemke and Yehle 1972a, 1972b; Lemke, 1974a, 1974b, 1975; Yehle and Lemke, 1972; Yehle, 1974, 1978, 1979). Sainsbury (1961) and Swanston (1969) described marine till with shell fragments and a raised marine beach on Prince of Wales Island. Similarly, Loney (1964) described fossil-bearing marine glacial clay in stream valleys at various elevations on Admiralty Island. On Gravina Island, Berg (1973) described uplifted glacial-marine deposits. These authors recognized these deposits as evidence of a postglacial marine transgression on isostatically depressed lands.

Mobley (1988) developed Holocene sea-level curves for Heceta and Prince of Wales Islands based on a limited data set and compared these deposits to those described by Clague et al. (1982a). Mobley concluded that the Heceta Island sea-level curve was comparable to that of Haida Gwaii. Mann and Hamilton (1995) recognized the marine transgression described by Mobley (1988) and a transgression of similar magnitude and duration to that of Haida Gwaii archipelago (formerly the Queen Charlotte Islands). Putnam and Fifield (1995) documented raised shell-bearing strata for many of the estuaries on Prince of Wales Island, plotting coastal emergence gradients. Mann and Streveler (2008) and Connor et al. (2009) characterized the occurrence of shell-bearing strata and the glacial and relative sea-level history of the Icy Strait region. The highest reported occurrence of glaciomarine sediments in southeastern Alaska is from Montana Creek near Juneau at 228.6 m asl. The highest documented occurrence of shell-bearing strata containing foraminifera is exposed in a landslide scarp on Admiralty Island noted by Miller (1973a) at an elevation of 211.8 m asl; no date has been obtained from this site. The highest occurrences of sampled and dated shell-bearing strata are at 191.0 m asl at Spaulding Meadows near Juneau. These two samples date to 13,950 ± 520 cal. yr B.P. and 14,160 ± 570 cal. yr B.P. U.S. Forest Service archaeologists have documented raised shell-bearing strata in southeastern Alaska since the mid-1980s (Carlson, 1984, 1991, 1992; Campbell, 1995; Davis, 1990; Davis et al., 1991; Putnam and Fifield, 1995). Hastings (2005) developed geologic and biologic models along three elevation breaks based on geologic evidence and the occurrences of coastal cutthroat trout (Oncorhynchus clarki clarki) and Dolly Varden (Salvelinus malma) believed to have been isolated by postglacial isostatic rebound.

In 2004, the Tongass National Forest geology program began a focused effort to document and date raised shell-bearing strata across the region. Subsets of these data were published in Hastings (2005), Carlson (2007, 2012), Shugar et al. (2014), and Schmuck et al. (2021). This manuscript is the first instance where the complete data set is compiled; the data set includes 723 samples, 436 ¹⁴C ages (45 of which are new), and 110 new sites. The sea-level curves presented here further define the complexity of postglacial relative sea-level fluctuations published by Shugar et al. (2014) through the addition of 110 new data points, more precise elevation controls, and updated, locally derived marine reservoir correction values (see below). We use this updated data set to refine the region's glacial and isostatic history (Lesnek et al., 2020). Similar glacially induced crustal displacement and relative sea-level changes have been well documented along the coastal margins of British Columbia (Clague et al., 1982; Clague, 1983; Luternauer et al., 1989; Josenhans et al., 1995, 1997; Fedje and Christensen, 1999; Fedje and Josenhans, 2000; Barrie and Conway, 2002; James et al., 2002; Hetherington et al., 2003, 2004; Hetherington and Barrie, 2004; McLaren, 2008; McLaren et al., 2014, 2018, 2020; Fedje et al., 2018).

Our literature search and material collection and processing methods have been described in Carlson (2007, 2012), Carlson and Baichtal (2015). and Lesnek et al. (2020). The multiple elevation datums used for this data set reflect what was reported by various authors over >120 years. For all of our newly acquired samples, elevation measurements were taken using handheld barometric digital altimeters accurate to ±1 m and that were zeroed as frequently as possible. Measured elevations are reported as meters above or below mean lower low water (AMLLW or BMLLW, respectively) or a 0.0 m tide. No attempt was made to change originally reported elevations or datums unless the site was revisited and the elevation measured to a 0.0 m tide. Many samples are reported simply as above sea level or above mean high tide (AMHT). For these samples, we do not know where the precise datum of the elevation measurement was taken. In practice, the measurement is commonly taken at the drift line or near the extreme high-tide line, which is not a precise datum. Across the study area, the elevation range in the modern high-tide line (HTL), the line of intersection of the land with the water's surface at the maximum height reached by a rising tide, varies substantially. The HTL ranges from 3.9 m AMLLW near Sitka, Alaska, to 6.4 m AMLLW near Haines, Alaska. For simplicity, all elevations not directly measured are reported here as above or below sea level as in the original publications. All directly measured elevations are reported as AMLLW or BMLLW. Lidar data collected in 2017 and 2018 for Prince of Wales, Annette, Gravina, Kuiu, and Kupreanof Islands and the islands west of Prince of Wales Island were delivered with a vertical datum of North American Vertical Datum of 1988 (NAVD88), geoid model GEOID12B, in meters (Alaska Division of Geologic and Geophysical Surveys elevation portal; https://elevation.alaska.gov/; accessed February 2021). Lidar data for the Haines vicinity collected in 2020 were also delivered with a vertical datum of NAVD88, GEOID12B, in meters (Daanen et al., 2021).

As stated in Lesnek et al. (2020), bulk samples of shell-bearing strata were collected in gallon-sized resealable plastic bags. Recovered samples were washed and screened through two graduated U.S.A. Standard Testing Sieves: #16, 1.18 mm (0.0469 inches); and #200, 0.075 mm (0.0029 inches). Care was taken to date charcoal or wood in addition to shells whenever possible. When sampling charcoal or wood, individual conifer needles or small limbs were selected when possible to minimize possible “old wood” errors (Carlson, 2012). When dating shell material, short-lived species were favored over longer-lived species. If a large shell was the only datable material, only the outermost growth was used. At one site, described below, fish bones were recovered and sent for species-level identification (Carlson, 2007). Over 230 samples of shell-bearing strata are currently stored at the U.S. Forest Service Tongass National Forest in Thorne Bay, Alaska; the Museum of the North at the University of Alaska Fairbanks is the intended repository of these materials.

Samples were prepared for accelerator mass spectrometry measurements at Beta Analytic, Miami, Florida, USA, or the U.S. Geological Survey ¹⁴C Laboratory, Reston, Virginia, USA, and ¹⁴C measurements were completed at Beta Analytic or Lawrence Livermore National Laboratory (Livermore, California, USA). The ages were calibrated in OxCal version 4.3 (Bronk Ramsey, 2009) using the IntCal20 (Reimer et al., 2020) and Marine20 (Heaton et al., 2020) calibration curves and locally derived marine reservoir corrections (Schmuck et al., 2021; see below). Calibrated ages are reported as the upper and lower limits of the entire calibrated age range at 2σ uncertainty.

Schmuck et al. (2021) aggregated shell-wood sample pairs from publications across the northwestern North American Pacific Coast to identify variation in the marine reservoir effect between samples of similar ages, with significant regional shifts in the reservoir effect through time. One of their most significant findings was the identification of a dramatic decrease in the reservoir age [R(t)] from the Bølling-Allerød interstade [R(t) = 1100 ± 170 yr, ΔR = 575 ± 165] to roughly the global average in the Younger Dryas stade [R(t) = ± 200 yr, ΔR = −55 ± 110]. A causal mechanism for this shift is provided by Praetorius et al. (2020), who argued for a persistent meltwater pulse (via the Columbia River) that freshened the North Pacific and lowered sea-surface temperature during the Younger Dryas. This freshening and stratification of surface waters corresponded with a shutdown of deep-water ventilation and decreased coastal upwelling, delivering less “old carbon” from the ocean basin to nearshore benthic communities during this period. The variability inherent in the marine reservoir effect, compounded by plateaus in the radiocarbon calibration curve during the Younger Dryas, unfortunately results in high uncertainty, but acknowledging this variability most faithfully represents confidence in our data set. With the return of coastal upwelling in the early Holocene, reservoir effects increased [R(t) = 700 ± 195 yr, ΔR = 245 ± 200] before stabilizing for much of the Holocene [between ca. 9000 and 2000 cal. yr B.P.; R(t) = 665 ± 135 yr, ΔR = 145 ± 165]. In southeastern Alaska, early and mid-Holocene values are slightly higher [R(t) = 700 ± 195 yr, ΔR = 265 ± 205; and R(t) = 750 ± 170 yr, ΔR = 225 ± 185, respectively] likely due to the “hard water effect” of local karst systems. By the late Holocene (between 2000 cal. yr B.P. and 200 cal. yr B.P.), the reservoir correction decreased further [R(t) = 630 ± 110, ΔR = 140 ± 100] (Schmuck et al., 2021).

A total of 436 ¹⁴C ages from 723 samples of shell-bearing strata across southeastern Alaska (Fig. 2) provide constraints on postglacial sea-level change (Table S1 in the Supplemental Material¹). The dated material comes from a variety of elevations, ranging from 430 m below sea level to 191 m above sea level. We use these ages from shell-bearing strata in conjunction with ¹⁴C ages from marine sediment cores and other terrestrial deposits to develop relative sea-level curves for 11 zones within southeastern Alaska (Figs. 3A–3K). Our description of these zones is generally structured to take the reader from the areas of least uplift to those of greatest uplift. We interpret the disparity in postglacial uplift between these zones as crustal response due to variable ice loading and unloading, with the greatest ice loading occurring at the Juneau mainland and northern Admiralty Island. We provide relative sea-level curves for each of these zones based on the chronology presented here and the elevations of beach ridges apparent on lidar bare-earth models (Figs. 3A–3K).

The oldest dated sites that overlap at 2σ in all 11 zones range in age from 14,870 ± 630 to 12,820 ± 340 cal. yr B.P. (Fig. 4). Our relative sea-level curves are further constrained by paleoshorelines that were mapped from lidar digital terrain models, which, in the absence of ¹⁴C-dated shell material, have in some cases been used to define the maximum elevation of postglacial transgression. Below, we describe the chronology from each of the 11 zones, including ¹⁴C ages from both shell and charcoal samples.

In the vicinity of Sitka Sound, shell-bearing strata can be found in inland stream banks and modern upper intertidal sediments. ¹⁴C ages from these sites range from 7930 ± 390 to 1940 ± 480 cal. yr B.P.; ages on the four oldest deposits overlap within 2σ uncertainty and range from 7930 ± 390 to 7630 ± 90 cal. yr B.P. In this zone, the highest shell occurrence documented to date lies at 4.0 m AMLLW (Fig. 4). Storm berm surge deposits exposed where power poles have been placed at an elevation of 9.0 m AMLLW record the maximum marine transgression in the Sitka area; however, these deposits are not dated at this time (Fig. 5D).

By far the highest density of shell-bearing strata we report is from Prince of Wales Island and the islands to the west. Numerous shell beds can be found in inland stream banks, beneath wetlands, and from upper intertidal sediments (Figs. 5 and 6). These deposits have been located by field reconnaissance, and exposed by road construction, quarry development, and excavation. These range in age from 11,995 ± 415 to 40 ± 40 cal. yr B.P. (Table S1 [^{footnote 1}]). The oldest dated deposits within 2σ uncertainty range in age from 11,000 ± 390 to 10,500 ± 420 cal. yr B.P. (17 samples) (Fig. 7). The highest shell occurrence is 18 m AMLLW. Lidar data acquired in 2017 and 2018 clearly show an average of six eroded shorelines, with the upper shoreline or terrace ranging from 19 to 25 m asl (Alaska Division of Geologic and Geophysical Surveys, elevation portal; https://elevation.alaska.gov/) (Fig. 8).

Numerous shell-bearing strata have been reported from Mitkof Island by Yehle (1978) and Viens (2001). Many of these original sites have been relocated, had their elevations carefully measured, and been resampled. These shell beds are exposed in stream and river banks, cut banks, and ditches associated with road construction, landslides, and excavations. Mitkof Island dates range from 14,300 ± 580 to 10,490 ± 1000 cal. yr B.P., and the sites range in elevation from 5.5 to 72.0 m asl. The highest reported elevation of shell-bearing strata lies beneath the Petersburg Solid Waste Baler Facility at ∼72 m asl. During construction, it was reported that shell-bearing diamicton was excavated into and the facility built over it. No sample of this material has been recovered for analysis. The oldest dated deposits within 2σ uncertainty range in age from 14,300 ± 580 to 12,490 ± 1090 cal. yr B.P. (eight samples).

Numerous shell-bearing strata have been reported from Wrangell and Etolin Islands by Lemke (1974). Many of these original sites have been relocated, had their elevations carefully measured, and been resampled. These shell beds are exposed in stream and river banks, cut banks, and ditches associated with road construction, landslides, and excavations. Wrangell and Etolin Islands dates range from 14,060 ± 570 to 10,160 ± 1080 cal. yr B.P., and the sites range in elevation from 10.6 to 40.0 m asl. The oldest dated deposits within 2σ uncertainty range in age from 14,060 ± 570 to 13,310 ± 550 cal. yr B.P. (three samples).

Most of the confirmed shell-bearing strata on Gravina Island came from Bostwick Inlet and Seal Cove on the central-southern shore of the island. The two oldest sites consist of very hard, compacted, shell-bearing diamicton that contain abundant diatoms. The sample contains well-preserved sea-ice marine diatoms and marginal sea-ice diatoms including Fragilariopsis cylindrus, Thalassiosira nordenskioeldii, Thalassiosira antarctica, Thalassiosira hyalina, and Porosira glacialis (J.A. Barron, 2017, personal commun.). Gravina Island dates range from 14,870 ± 630 to 9180 ± 560 cal. yr B.P., and the sites range in elevation from 5.5 to 9.0 m AMLLW (Fig. 5A). The oldest dated deposits within 2σ uncertainty range in age from 14,870 ± 630 to 14,030 ± 550 cal. yr B.P. (three samples). Geomorphic interpretation of lidar digital terrain models acquired in 2018 has identified eight wave-eroded shorelines from 6.0 m to 76.0 m asl (Alaska Division of Geologic and Geophysical Surveys elevation portal; https://elevation.alaska.gov/). This likely represents the highest stand of postglacial sea level (Fig. 9).

Many shell-bearing sites have been reported from Kupreanof Island. These shell beds are exposed in stream and river banks, cut banks, and ditches associated with road construction and excavations. Kupreanof Island dates range from 14,280 ± 570 to 9150 ± 560 cal. yr B.P., and the sites range in elevation from 0.6 to 80.0 m asl. The oldest dated deposits within 2σ uncertainty range in age from 14,280 ± 570 to 13,910 ± 510 cal. yr B.P. (five samples). In 2018, high-resolution lidar data were acquired for some of Kupreanof Island near Kake. Geomorphic interpretation of lidar data acquired in 2018 has identified 18 possible shorelines from 4.5 m to 102.5 m asl (Alaska Division of Geologic and Geophysical Surveys elevation portal; https://elevation.alaska.gov/) (Fig. 10). The highest known shell occurrence sits in deep glacial clays exposed during road construction beneath the highest interpreted shoreline.

Numerous shell-bearing strata have been reported from Revillagigedo Island by Stuckenrath (1971) and Lemke (1975). Many of these original sites have been relocated, had their elevations carefully measured, and been resampled. These shell beds are exposed in stream and river banks, cut banks, and ditches associated with road construction, landslides, and excavations. Revillagigedo Island dates range from 14,550 ± 1110 to 7270 ± 470 cal. yr B.P., and the sites range in elevation from 10.0 to 85.0 m AMLLW. Wave-cut notches and terraces can be found in several places from 80 to 85 m asl. The oldest dated deposits within 2σ uncertainty range in age from 14,550 ± 1110 to 13,510 ± 430 cal. yr B.P. (seven samples).

Many shell-bearing sites have been reported from Chichagof Island. These shell beds are exposed in stream and river banks, cut banks, and ditches associated with road construction and excavations. Chichagof Island shell-bearing strata dates range from 14,230 ± 610 to 11,060 ± 800 cal. yr B.P., and the sites range in elevation from −4.7 to 60.0 m asl. In 2014, high-resolution lidar data were acquired on northeastern Chichagof Island near Hoonah. Geomorphic interpretation has identified a terrace eroded into glacial features and landslide deposits at 107 m asl (Alaska Division of Geologic and Geophysical Surveys elevation portal; https://elevation.alaska.gov/) (Fig. 11). This likely represents the highest stand of postglacial sea level.

Several shell-bearing strata have been reported near Haines, on the Alaska mainland, by Lemke and Yehle (1972a). In 2020, the highest and oldest shell-bearing diamicton in the vicinity of Haines was discovered (J.P. Norton, 2020, personal commun.) at an elevation of 75.0 m asl. Shell-bearing strata in the vicinity of Haines range in age from 12,820 ± 340 to 12,110 ± 380 cal. yr B.P., and the sites range in elevation from 6.1 to 75.0 m asl. Lidar digital terrain models acquired in 2014 and 2020 show at least 26 storm berms and wave-eroded shorelines up to 107 m asl (Alaska Division of Geologic and Geophysical Surveys elevation portal; https://elevation.alaska.gov/) (Fig. 12). This likely represents the highest elevation of postglacial sea level.

The Juneau and Douglas Island vicinities are the type locality for the Gastineau Channel Formation, first described by R.D. Miller in 1973 (Miller, 1973b) and mapped in 1975 (Miller, 1975). The highest-elevation marine diamicton is recorded from Montana Creek (Miller 1973b, 1975) at 229 m asl. No shell was described from this site, and the dated material was an upper limiting date from basal peat above the diamicton. The highest record of shell-bearing diamicton is on northern Admiralty Island at 212 m asl (Miller, 1973a). No sample is available from this site. The highest dated sample comes from Spaulding Meadows above Auke Lake north of Juneau at 191 m asl (Fig. 6A). The oldest sample taken from this site dates to 14,160 ± 570 cal. yr B.P. The oldest dated sample was recovered from Douglas Island on Fish Creek at an elevation of 164 m AMLLW, with an age of 14,510 ± 750 cal. yr B.P. Both of these samples are overlain by the Mount Edgecumbe dacite tephra, which dates to 13,160 ± 90 cal. yr B.P. (Begét and Motyka, 1998).

The Yakutat forelands lie to the west of the Fairweather fault, an extension of the Queen Charlotte fault. Shell-bearing strata in the vicinity of Yakutat, on the Alaska mainland, range in age from 10,050 ± 600 to 1510 ± 620 cal. yr B.P., and the sites range in elevation from 26.0 to 50.0 m asl. Shell fragments taken from the push moraine enclosing Harlequin Lake at an elevation of 26 m asl date to 9520 ± 570 cal. yr B.P. Shell fragments taken from the area at an estimated elevation of 50 m in 1974 from similar moraine deposits were dated to 9640 ± 1080 cal. yr B.P. (Yehle, 1979). These deposits suggest that the Harlequin fjord persisted to at least 9640 ± 1080 cal. yr B.P., and likely much longer. At that time, the Yakutat Glacier may have been a tidewater glacier. The clams that now exist as shell fragments in the moraine lived in the intertidal portions of the fjord. It is believed that the late Holocene advance of the Yakutat Glacier began ca. 3000 yr ago, culminating at ca. 1850 CE at the southern end of Harlequin Lake (Barclay et al., 2001). Harlequin Lake as we know it began forming by 1903 (Trüssel et al., 2013). The Little Ice Age advance that created the moraine at the end of Harlequin Lake dug up the older marine sediments, incorporating them into the moraine. The Little Ice Age sediments from the Yakutat Glacier infilled the fjord, and the Dangerous River now flows on top of those sediments.

Charcoal was recovered from shell-bearing marine deposits from 40 sites and 99 samples ranging in age from 11,000 ± 390 to 3400 ± 110 cal. yr B.P. Charcoal dating to between ca. 11,000 and 7630 cal. yr B.P. is particularly abundant (Fig. 13). Though five sites younger than ca. 7630 cal. yr B.P. do contain charcoal, most sites younger than this do not contain charcoal. No charcoal was recovered from sites predating ca. 11,000 cal. yr B.P. At two sites, Yatuk Creek on Prince of Wales Island and Bostwick Creek on Gravina Island, Pacific sardine (Sardinops sagax) bones were found in charcoal deposits ranging in age from 10,780 ± 570 to 9870 ± 640 cal. yr B.P. (Figs. 13 and 14).

Previously published ¹⁰Be exposure ages from across the southern part of southeastern Alaska and coastal British Columbia provide direct constraints on the timing of CIS retreat between ca. 17,000 and 15,000 yr ago (Darvill et al., 2018; Lesnek et al., 2018, 2020). The dated materials from shell-bearing marine deposits throughout the study area provide minimum-limiting constraints on CIS deglaciation and the timing of the marine transgression as the result of forebulge collapse along the western margin of southeastern Alaska. When interpreted together, these two data sets reveal a consistent pattern of ice-margin change and sea-level fluctuations in the region.

Beginning possibly as early as 18,000 cal. yr B.P., but likely well under way by 17,000 cal. yr B.P., local sea levels rapidly rose, flooding fjords and straits and causing rapid calving of the CIS ice within them. To the west, a forebulge persisted as a result of the weight of the glaciers present to the east and north and isolated ice caps on islands. This forebulge is expected to have changed in breadth and magnitude as the glaciers receded and thinned. The Queen Charlotte fault became the western terminus of the forebulge (Barrie et al., 2021).

To the east, the land was isostatically depressed by the weight of the ice. As the glaciers receded and sedimentation along the fjord sides and the glacial fronts slowed, marine organisms became established. First, those organisms that could tolerate the dynamics and energy of the changing landscape took hold; later, intertidal and near-tidal communities somewhat similar to those of the modern day became established. By 14,870 ± 630 to 12,820 ± 340 cal. yr B.P., most of the fjords and straits were ice free, and early mollusk communities became established.

As the main body of the CIS retreated toward interior British Columbia at ca. 14,500 cal. yr B.P. (Lesnek et al., 2020), isolated or stranded ice caps likely persisted on many of the islands in southeastern Alaska. Valleys throughout the region were likely occupied by alpine glaciers (Lesnek, et al., 2020). During intervals of high sea level, seaward portions of modern coastal river systems such as the Taku, Stikine, and Unuk Rivers in Alaska and the Iskut and Nass Rivers in British Columbia would have been submerged, providing habitat for marine organisms (Buddington and Chapin, 1929; McConnell, 1913; Twenhofel, 1952). Glaciers occupying these valleys would have rapidly calved back to fjord heads in response to relative sea-level rise (Clague and James, 2002). Furthermore, many of the largest islands such as Admiralty, Kupreanof, Kuiu, and Mitkof would have consisted of several smaller islands at that time.

Though not dated, foraminifera have been documented from a glaciomarine diamicton at an elevation of 211 m on northern Admiralty Island (Miller, 1973a). In the vicinity of Mitkof, Wrangell, and Etolin Islands, shell-bearing strata can be found up to 72 m asl (Yehle, 1978). Near Revillagigedo Island, shell-bearing strata are found up to 80 m asl. In the vicinity of the Juneau mainland and Douglas and northern Admiralty Islands, shell-bearing strata can be found up to an elevation of 191 m asl (Fig. 3J). These deposits lie ∼0.30 m beneath what appears to be the Mount Edgecumbe dacite tephra (Fig. 6). Age constraints on the tephra range from 13,582–12,769 cal. yr B.P. (Riehle et al., 1992a, 1992b; Begét and Motyka, 1998; Dunning, 2011; Baichtal, 2014). The Mount Edgecumbe dacite tephra serves as a marker horizon across much of the central and northern portions of the study area (Riehle et al., 1992b) and has been found in both terrestrial and marine records (Addison et al., 2010; Dunning, 2011). Several tephras lie stratigraphically below the Mount Edgecumbe dacite tephra, some of which have been attributed to other Mount Edgecumbe volcanic field eruptions and the Wrangell volcanic field (Addison et al., 2010; Dunning, 2011). A precise age for these earlier eruptions has not been determined, but Ager (2019) reported the presence of two tephras with ages of 14,600 and 13,760 cal. yr B.P. in a lake on Baranof Island. We identified two sites near Juneau (sample IDs DOUGLAS13, DOUGLAS31, DOUGLAS32, JUNEAU23, and JUNEAU68; Table S1 [see ^{footnote 1}]) that appear to contain Mount Edgecumbe dacite tephra similar to that described near Montana Creek by Begét and Motyka (1998).

The variable marine limit between the oldest dated sites from the inner zones of the study area reflects the differential loading of glacial ice and the isostatic response of the landscape. The shell-bearing strata have not been analyzed to determine the likely depth of deposition based on the mollusk and foraminifera species composition. Though at widely varying elevations from current sea level to 191 m asl, as discussed above, the oldest dated sites from the inner zones of the study area that overlap at 2σ range in age from 14,870 ± 630 to 12,820 ± 340 cal. yr B.P. Shortly after ca. 14,000 cal. yr B.P., land surfaces in the Juneau–Douglas Island–northern Admiralty Island region rapidly rose. Most landscapes had rebounded to within 25 to 30 m of present sea level by the beginning of the Younger Dryas at 12,900 cal. yr B.P. (Figs. 3E, 3F, 3G, 3H, 3I, and 3J). This means in the Juneau, Douglas Island, and northern Admiralty Island vicinity the land rose a minimum of 180 m in 2800 yr. In approximately the same amount of time, the land rose 35 m near Mitkof and Wrangell Islands and 50 m near Ketchikan, Alaska.

West of the CIS margins, a forebulge had developed as the ice retreated. This forebulge is similar in breadth and timing to that described by Hetherington et al. (2003, 2004) and Barrie et al. (2014) along the northern coast of British Columbia. Though undated, a terrace 165 m below sea level (bsl) is visible on multibeam bathymetry throughout southeastern Alaska (Carlson, 2007; Shugar et al., 2014). Terraces at 180 m to 170 m bsl are visible to the west of Kruzof Island (Greene et al., 2007, 2011). These terraces are similar to those described by Barrie and Conway (2012) in Hecate Strait between Haida Gwaii and the British Columbia mainland. We have no core data near these terraces for approximating their age; however, these terraces very likely predate the beginning of the Younger Dryas and mark where the fetch of the sea came into equilibrium with the forebulge front sometime before the glaciers fully retreated from the fjords and straits at ca. 14,000 cal. yr B.P.

Carlson and Baichtal (2015) first reported evidence of a forebulge along the western shore Prince of Wales Island from many of the islands immediately to the west. Baichtal et al. (2017a, 2017b) further discussed and defined the forebulge and associated sea-level fluctuations. Our analysis has refined the timing and description of this forebulge. Additional evidence for the development of a forebulge has been reported from four submerged localities; three sets of cores retrieved off the western coast of southeastern Alaska recorded the sea breaching basin sills in what are now saltwater basins (Barron et al., 2009; Addison et al., 2010; Barrie et al., 2021).

In Sitka Sound, a core (EW0408-40JC) obtained west of Baranof Island and south of Kruzof Island constrains the collapse of the northern edge of the forebulge off Baranof Island, tracking the shift of Sitka Sound from an isolated lacustrine environment to marine (Addison et al., 2010). The core records the breaching of a sill at 113 m bsl. A radiocarbon-dated bivalve 0.39 m above the breach record provides a minimum-limiting age of the inundation at 11,730–12,670 cal. yr B.P. (median age of 12,200 ± 480 cal. yr B.P.). Given that eustatic sea level was 61 m lower than today at that time (Peltier and Fairbanks, 2006), a 52-m-high forebulge was required to create the basin sill.

In the Gulf of Esquibel, west of Prince of Wales Island and south of Heceta Island, core EW0408-11JC (Barron et al., 2009) records sedimentation in the basin defined by a sill at 80 m bsl. Here, a bivalve provides a minimum age of 12,260–11,350 cal. yr B.P. (median age of 11,805 ± 455 cal. yr B.P.) for the flooding of the gulf, though underlying sediments are rich in sea-ice and freshwater diatoms and/or silicoflagellates mirroring the Younger Dryas freshening event described by Praetorius et al. (2020). Below these sediments lies a tephra correlated to a tephra in Leech Lake and possibly a tephra in a core from the Bald Mountain bog, both on Heceta Island (T.A. Ager, 2021, personal commun.). Wilcox et al. (2019a) dated a correlative black basaltic tephra from a lake on Baker Island to 13,492 ± 237 cal. yr B.P. Given the absence of Bølling-Allerød signatures between the tephra and the apparently Younger Dryas sediments (and the otherwise surprisingly low sedimentation rate linking the two stratigraphic markers), core EW0408-11J likely experienced a similar erosive event by the transgression of the sea early in the Younger Dryas that removed any sedimentation that occurred during the preceding period. Given that eustatic sea level then was 58 m lower than today (Peltier and Fairbanks, 2006), there must have been a 22-m-high forebulge to create the basin sill at that time.

Three cores reported by Barrie et al. (2021) southwest of Suemez Island further illuminate the shifts surrounding the collapse of the outer forebulge and rapid transgression of the sea, an erosive process. Radiocarbon dates on bivalves in marine muds, presumably representing an age prior to exposure (under an erosional boundary), and ages of bivalves bracketing a layer of coarse gravels and sands immediately following the transgression overlap significantly, a function of natural variability in the marine reservoir effect and plateaus in the radiocarbon calibration curve producing multiple intercepts during the Younger Dryas. Despite the lack of precision, these overlapping ages constrain the shifts between a shallow-marine, potentially subaerially exposed, and resubmerged landscape between 12,650 and 12,050 cal. yr B.P. Unfortunately, the cores do not provide linkages between the timing of the forebulge collapse and forebulge elevations. Bathymetry for the area is incomplete, so at this time we cannot determine whether the core locations are representative of the forebulge or lie in isolated basins. Core 201504-41 from 138.5 m bsl provides an appropriate elevation minimum, producing a dated bivalve providing a minimum age for the collapse of the forebulge between 12,410 and 11,580 cal. yr B.P., with a median age of 11,995 ± 415 cal. yr B.P. The sea level at that time was 59 m bsl. For this basin to have flooded, a forebulge of 80 m must have existed.

As the forebulge migrated eastward and collapsed, it marked the uppermost elevation of the maximum sea-level transgression, which can be seen on lidar data acquired in 2017 and 2018. Glacial features and fluvial terraces are clearly truncated at 19 m to 25 m above AMHT (Fig. 8). The highest shorelines are found on northeastern and southern Prince of Wales Island at 25 m AMHT. Glacially carved troughs, submarine moraines, and ice-flow indicators such as striations and drumlins suggest that the thick ice streams occupying Sumner Strait to the north and Dixon Entrance to the south flowed over Prince of Wales Island (Lesnek et al., 2020). Thus, we attribute the differences in maximum shoreline elevation across this zone to variable CIS loading during the Late Pleistocene. Six distinct terraces can be seen below the erosion of the glacial and fluvial features. Carlson and Baichtal (2015) attempted to define the timing and extent of these terraces to the location of early Holocene archaeological sites using the occurrences of shell-bearing strata; >30 sites have been found using this methodology (Carlson, 2017; Schmuck, 2021). However, the oldest shell deposits predate the oldest known cultural deposits by ∼40–700 yr. We interpret the age of these oldest deposits as a minimum-limiting age for the transgression associated with the collapse of the forebulge. Seventeen (17) of the oldest dated deposits within 2σ uncertainty range from southwestern Kuiu, Kosciusko, Heceta, and western Prince of Wales Islands date to 11,000 ± 390 to 10,500 ± 420 cal. yr B.P. (Fig. 7). We suggest this range is a minimum-limiting age for the collapse of the forebulge. Once the forebulge had collapsed and the shoreline had reached equilibrium, intertidal communities rapidly developed. Therefore, these older shell deposits do not provide the precise timing of the end of the forebulge collapse, but date to a time shortly after when intertidal communities became re-established. During this interval, the land was rising due to ongoing isostatic and tectonic adjustments, and within 40–700 yr, the wave-eroded terraces and deltas began to re-emerge. The six distinct terraces identified on lidar imagery suggest periods where eustatic sea level and isostatic uplift came into equilibrium long enough to result in either (1) erosion of a terrace at the maximum height of the sea or (2) deposition of a delta at the mouth of a fluvial system delivering sediment to the ocean.

Near Sitka on Baranof Island, the uppermost elevation of the maximum sea-level transgression as the forebulge to the west collapsed is much lower in elevation than on western Prince of Wales Island. The highest, oldest shell bed dates to 7930 ± 390 cal. yr B.P. at an elevation of 4.0 m AMLLW on Krestof Island to the west of Baranof Island (Fig. 5C). The explosive dacite eruption of Mount Edgecumbe on Kruzof Island blanketed the Sitka vicinity with tephra as much as 1 m deep at ca. 13,165 ± 85 cal. yr B.P. (Riehle et al., 1992a, 1992b; Begét and Motyka, 1998; Baichtal, 2014). The upper limit of the maximum transgression in the Sitka region is marked by the contact between the Mount Edgecumbe tephra and underlying bedrock. At elevations of up to 9.0 m AMLLW, the tephra has been eroded by wave action, exposing Quaternary basalt flows (Riehle et al., 1992a, 1992b) and Sitka Graywacke (Karl et al., 2015). In a few outcrops and where power-pole holes have been drilled, a storm berm consisting of wave-rounded cobbles running parallel to the paleoshoreline has been exposed. Some 5 m below the highest storm berm deposit is the upper limit of the shell-bearing strata (Fig. 5D).

Charcoal is particularly abundant in samples dating from 11,000 ± 390 to 7630 ± 90 cal. yr B.P. during what is inferred to be the local Holocene Climatic Optimum (Fig. 13). This interpretation is supported by the presence of Pacific sardine (Sardinops sagax) bones in charcoal deposits ranging in age from 10,780 ± 570 to 9870 ± 640 cal. yr B.P. (Figs. 13 and 14). In modern times, the habitat range of this fish species does not typically extend to southeastern Alaska; however, Pacific sardine has been observed in the region following the exceptionally warm, strong El Niño events in 1930–1931 and 1997–1998 (Wing et al., 2000). In the broader northeastern Pacific region, northward expansions of Pacific sardine habitats were also documented during the 2013–2015 El Niño event (Cavole et al., 2016). Anomalously warm El Niño events involve increased temperatures in the North Pacific, which may explain the change in habitat range of Pacific sardine observed during these intervals (Wing et al., 2000). Therefore, occurrences of Pacific sardine between ca. 10,800 and ca. 9900 cal. yr B.P. in southeastern Alaskan sediments suggest the presence of strong El Niño events and/or a warmer regional climate during the early Holocene (Carlson, 2007).

Biological proxies in terrestrial and marine sediment cores also suggest a warm early Holocene in southeastern Alaska. Heusser et al. (1985) reported on data derived from fossil pollen contained in two sections of muskeg located near Munday Creek on the Gulf of Alaska, ∼15 km northwest of Icy Cape in the Malaspina Glacier district; one section extends to ca. 10,200 cal. yr B.P., whereas the other reaches to beyond ca. 12,800 cal. yr B.P. They reported mean July temperatures of ∼14 °C at the end of the Pleistocene (11,700 cal. yr B.P.) and in the early Holocene, reaching a maximum ∼16 °C at ca. 8900 cal. yr B.P., then decreasing to 12 °C by ca. 6000 cal. yr B.P. Annual precipitation, just under 1400 mm at the beginning of the record, reached a minimum ∼1200 mm coinciding with the temperature maximum at 8900 cal. yr B.P. Ager (2019) also described an early Holocene warming with a mid-Holocene temperature shift to cooler, wetter conditions. These findings correlate closely to the time period in which we find the most abundant charcoal in the shell-bearing sediments. Barron et al. (2016) described diatom and silicoflagellate assemblages documented in cores EW0408-47JC, EW0408-47MC, and EW0408-46MC from the outer portion of Slocum Arm from western Chichagof Island. The base of the core (sample depth 1.78 m below seafloor) dates to 10,176–9746 cal. yr B.P. (Barron et al., 2016). They suggested that the record from the early portion of the core (>ca. 6800 cal. yr B.P.) contains oceanic diatoms indicative of cold winters and warm summers. Praetorius et al. (2015) documented a 4–5 °C warming event in the Gulf of Alaska during the transition into the early Holocene (11,400–10,900 cal. yr B.P.). This interval corresponds with the timing of the most abundant charcoal and the presence of Pacific sardine from our samples.

The coincidence of abundant charcoal in southeastern Alaskan raised shell-bearing marine deposits with warmer climate in the early Holocene and rarity of charcoal in the cooler middle and late Holocene raised marine deposits suggest a linkage between forest-fire history and climate in southeastern Alaska, as reported in neighboring British Columbia (Brown and Hebda, 2002, 2003). It is tempting to speculate on a potential anthropogenic source for charcoal in southeastern Alaska marine deposits predating confirmed archaeological evidence for humans. In coastal British Columbia, evidence for late Holocene anthropogenic forest fires has been used to argue for charcoal as a possible proxy for human presence in the absence of archaeological evidence (Hoffman et al., 2017; Mathewes et al., 2019). Archaeological evidence suggests a constant human presence since ca. 10,500 cal. yr B.P. in southeastern Alaska (Carlson 2007, 2012; Carlson and Baichtal, 2015). If the charcoal in the raised shell-bearing strata was cultural in origin, one would expect it to be present throughout all of the Holocene. Charcoal is conspicuously absent in the late Holocene sediments (≥7630 ± 90 cal. yr B.P.). In southeastern Alaska, the charcoal occurring in the raised marine sediments is not likely to be anthropogenic and instead records periods of increased naturally triggered forest-fire activity.

If humans were present in southeastern Alaska in the immediate postglacial period, what environmental hazards may they have faced? Following retreat of the CIS, intertidal mollusk species traditionally used for food began to reappear across the Northwest Coast by around 14,200 cal. yr B.P. (Schmuck et al., 2021). However, the establishment of the willow-sedge and pine parklands on land (Ager, 2019) was tempered by intense volcanism; the Mount Edgecumbe volcanic field alone was responsible for 22 individual eruptions in central southeastern Alaska, 19 within a peak interval between 14,600 and 13,100 cal. yr B.P. (Praetorius et al., 2016). There is mounting evidence for further eruptions from other volcanic fields in the southern part of southeastern Alaska at the end of that interval (Wilcox et al., 2019a; T.A. Ager, 2021, personal commun.). Furthermore, if coastal fisher-gatherers lived along the outer coast of southeastern Alaska by this time, rapid sea-level rise accompanying the collapse of the forebulge would likely have been recognizable at a human scale (Mackie et al., 2014). In all, these factors would have contributed to a dynamic environment for human occupation during the Late Pleistocene.

We hope that this data set and discussion will be a catalyst for further research into the late Quaternary history of the study area. Here, we offer a few suggestions for further study. A logical next step is to integrate the chronology presented here with the elevations of beach ridges apparent on lidar bare-earth models in a geophysical model. This exercise would result in a range of plausible relative sea-level curves for southeastern Alaska that can be directly compared to those presented here to assess crustal response to ice loading, constrain the possible range of ice thicknesses, or gain a better understanding of the region's mantle rheology. Faunal analysis of shell-bearing sediments holds a wealth of potential for determining the probable depth of deposition, salinity, and water temperature of these deposits, which could help to further refine the relative sea-level curves presented here and add information about environmental conditions through time. Fully dating and characterizing the now-submerged pāhoehoe lava flows and eruptive centers off the western coast of southeastern Alaska is necessary for linking these geomorphic features with tephra observed in sediment cores across the region, which has implications for both the region's glacial history and volcanic hazard assessments. Future marine sediment cores targeting closed basins at different elevations to determine the rate of sea-level rise is paramount in defining the timing and persistence of the forebulge along the western margin of the study area. Coupled with the sampling of marine basins, targeted sampling of Younger-Dryas and Bølling-Allerød marine deposits for shell-wood pairs is necessary for further refining the marine-reservoir-effect calibrations used to date shellfish, our proxies for sea-level change, in these critical periods. At present, the calibration curve used here is a regional synthesis of pairs from across the Northwest Coast (Schmuck et al., 2021). Further sample pairs from southeastern Alaska would provide a means to account for regional variation driven by the “hard water effect” of karst groundwaters.

The intense erosional processes described by Barrie et al. (2021) leave little hope for Late Pleistocene archaeological site survival on the once-exposed western coastal plain or in potential refugia on the now-submerged outer forebulge. Unlike on the eastern edge of Haida Gwaii, where shallow bays and a broad coastal plain are sheltered from wave action (Mackie et al., 2014), the waterways of the Alexander Archipelago are deeply incised channels that offer scant protection for submerged sites. Nevertheless, exploratory surveys guided by relative sea-level reconstructions (e.g., Carlson and Baichtal, 2015; McLaren et al., 2014, 2018, 2020) have demonstrated great promise in locating ancient cultural sites on the Pacific coast. Targeting similar areas above modern sea level in southeastern Alaska, particularly those near “sea-level hinges” (McLaren et al., 2014) where relative sea level has not changed significantly since deglaciation, may reveal pre-Holocene archaeological sites.

Southeastern Alaska is the ancestral homeland of the Tlingit, who remember the complex relationship between their people and sea-level change in oral traditions (Worl, 2010). If this communal memory records human-environment interaction in the region as far back as the Younger Dryas or Bølling-Allerød, it suggests that older cultural sites in danger of disturbance by modern development may exist in areas that have not yet been recognized by archaeologists. Many of the early Holocene sites first identified in southeastern Alaska were discovered during the destructive building of roads and other infrastructure (Carlson and Baichtal, 2015). Articulating the timing and prior extents of relative sea-level change across the landscape are critical for identifying and protecting ancient cultural sites in the future. Finally, two-way exchange with scientific results like these being communicated back to the native communities should be common to all future work.

The Tongass National Forest Geology Program has supported and funded this research for 30 years, as has the University of Alaska Southeast, Juneau. Many people through the years have contributed to discovery and sampling of shell-bearing strata throughout southeastern Alaska, many more than can be named here. Cathy Connor, Greg Streveler, Daniel Monteith, Rachel Myron, Jacqueline de Montigny, John Norton, Terry Brock, Linda Slaght, Gina Esposito, Gene Primaky, Kris Larson, David Love, and Katherine Prussian have supported data collection over the years. Susan Karl with the U.S. Geological Survey in Anchorage has been instrumental in both data collecting and interpretation, providing invaluable assistance. A special thanks to Susan Crockford of Pacific Identifications Inc., Victoria, British Columbia, for fish species identification. We would like to thank James Vaughn Barrie, Geological Survey of Canada, and Daniel Shugar, Department of Geoscience, University of Calgary, for their reviews and thoughtful comments that improved this manuscript.