Ice and ocean constraints on early human migrations into North America along the Pacific coast

Founding populations of the first Americans likely occupied parts of Beringia during the Last Glacial Maximum (LGM). The timing, pathways, and modes of their southward transit remain unknown, but blockage of the interior route by North American ice sheets between ~26 and 14 cal kyr BP (ka) favors a coastal route during this period. Using models and paleoceanographic data from the North Pacific, we identify climatically favorable intervals when humans could have plausibly traversed the Cordilleran coastal corridor during the terminal Pleistocene. Model simulations suggest that northward coastal currents strengthened during the LGM and at times of enhanced freshwater input, making southward transit by boat more difficult. Repeated Cordilleran glacial-calving events would have further challenged coastal transit on land and at sea. Following these events, ice-free coastal areas opened and seasonal sea ice was present along the Alaskan margin until at least 15 ka. Given evidence for humans south of the ice sheets by 16 ka and possibly earlier, we posit that early people may have taken advantage of winter sea ice that connected islands and coastal refugia. Marine ice-edge habitats offer a rich food supply and traversing coastal sea ice could have mitigated the difficulty of traveling southward in watercraft or on land over glaciers. We identify 24.5 to 22 ka and 16.4 to 14.8 ka as environmentally favorable time periods for coastal migration, when climate conditions provided both winter sea ice and ice-free summer conditions that facilitated year-round marine resource diversity and multiple modes of mobility along the North Pacific coast.

Human dispersal pathways from Beringia into North America continue to be debated. Prevailing ideas include a coastal route and an interior route via an ice-free corridor between the Laurentide and Cordilleran ice sheets (1–6). The Laurentide and Cordilleran ice sheets merged during the Last Glacial Maximum (LGM) (7), closing the ice-free inland corridor between ~26 ± 1 ka (Fig. 1 and ref. 8) and 13.8 ± 0.5 ka (ref. 9). Archaeological sites south of the ice sheets in North America during this time frame (10–15) thus require either a coastal route, or entry through the interior prior to the LGM. A pre-LGM migration scenario is at odds with apparent genetic divergence between Siberian and Beringian populations between about 25 to 24 ka (95% CI 21 to 28 ka; ref. 16) and an inferred “Beringian Standstill” in migration until 18 to 16 ka (16–19). Was this biogeographical pause due to favorable conditions in Beringia, glacial bottlenecks that prevented southward transit along the coast, or a combination of both? How did Beringians make the arduous journey along the Pacific Coast corridor – by land, sea, or ice? Was the coastal route effectively blocked throughout the LGM, or were there intervals when passage was more or less possible? Building on recent evidence for multiple intervals of Cordilleran ice retreat within the last ice age (20), we evaluate these scenarios and define relatively benign climatic intervals when human migration along the Cordilleran coast may have been most feasible.Open in a separate windowFig. 1.Map of coastlines and ice extent at various time periods A) 32.5 ka, B) 27.5 ka, C) 25 ka, D) 15 ka during the late Pleistocene, showing possible migration pathways at each stage. Relative sea level (RSL) and ice sheet topography are from (8) and are interpolated and applied to the ETOPO01 bathymetry grid (21). Post LGM glacial ice evolution is unknown for Siberia, though some ice sheets were likely present during these time periods. North American archaeological sites (black dots) are shown that have median dates for initial human occupation that fall within ± 1 ka of the various time slices shown (for a full list of archaeological site data and references, including those that fall outside of the time/space domains shown here, see SI Appendix, Table S1 and Dataset S1); sites with controversial evidence for human presence are denoted with question marks. White dashed line on panel (C) shows the estimated extent of winter sea ice during the LGM, based on (22). Seasonal sea ice was present along the Alaskan coastal corridor to varying degrees during all the periods shown, but the spatial extent is not as well defined for the other intervals. Sediment cores identified in panel (A) are for the various proxy datasets shown in Figs. 3 and ​and55.Despite evidence for older archaeological sites farther inland, thus far, there is no definitive evidence of human occupation along the Pacific Coast of North America prior to ~13.8 ka (23). The absence of earlier coastal sites may reflect submergence of former occupation sites by rising postglacial sea level, exacerbated locally by relaxation of a subsiding glacio-isostatic forebulge (3, 24). Other factors may also have limited the viability of a coastal transit at certain times. The most obvious obstacle is ice cover on land, with large outlet glaciers emanating from the Alaska Peninsula and Southeast Alaska terminating in the ocean. Heavily crevassed ice streams would have been difficult or impossible to cross on land and dangerous at sea, potentially preventing passage for migrating groups of people.The strength of the cyclonic Alaska Coastal Current (ACC) also may have partially impeded southward movement for seafarers, as this current flows northward against the direction of migration (25). The ACC is driven by wind and Coriolis forcing and strengthened by coastal freshwater inputs (26) (Fig. 2). Royer and Finney (25) hypothesized that southward migration was impeded by freshwater input and rapid sea-level rise that accelerated coastal currents during global Meltwater Pulse 1a (MWP1a: 14.65 to 14.30 ka; ref. 27), effectively assuming that local freshwater inputs tracked global-average sea-level rise. Testing this hypothesis requires reconstruction of regional ice retreat and the resulting reduction of coastal salinity from regional meltwater flux, along with quantitative modeling of coastal current strength, issues we address here.Open in a separate windowFig. 2.Simulations of ocean currents in the Northeast Pacific under different climate and sea level conditions: Modern climate state (A), LGM climate state, with sea level −120 m below modern (B), LGM boundary conditions with an increased freshwater flux (C), and intermediate sea level (−75 m), as would have occurred during the mid-deglacial period (D). Mean annual surface ocean velocity shows a strengthening of the cyclonic Alaska Current during the LGM relative to modern conditions, as well as a contraction of the shelf area on which the ACC flows. Boundary currents flow in a cyclonic (anticlockwise) direction.The extent of land ice, both along the coastal corridor and inland route, has been widely debated over many decades (4, 28–30). However, the assessment of ice in the marine environment—such as the extent of sea ice and icebergs, and their impact on human migration—has received less attention. Evidence from ice-rafted debris (IRD) in marine sediments shows that the seaward edge of the Cordilleran Ice Sheet (CIS) and its outlet glaciers was extremely variable and subject to repeated abrupt retreats onto land or into silled fjords during the late Pleistocene (referred to as “Siku Events”; ref. 20). Sea ice formed in the subarctic North Pacific through much of this interval (22, 31), which may have impacted boat transit and altered marine resource composition and availability during certain months of the year. Today, land-fast sea ice provides a relatively unobstructed and flat surface as a platform for travel between otherwise inaccessible high Arctic communities, typically in winter or spring (32, 33). In addition to ease of movement, sea ice facilitates hunting of marine mammals near the ice edge and sub-ice intertidal shellfishing; both are important food resources in the Arctic winter (32). With the seasonal melting of sea ice, kelp forest habitats can provide important marine resources in summer (34, 35). Reconstructions of North Pacific sea ice are essential to building a clearer picture of the conditions that coastal people in the North Pacific would have contended with during the glacial and deglacial periods.To help address these issues, we present records of sea-ice variations based on the %C_37:4 proxy (36) and synthesize previously published reconstructions of sea ice, sea-surface temperature (SST), salinity, and IRD from marine sediment cores in the North Pacific (Fig. 1). Together, these paleoenvironmental data help discern major climate and oceanographic changes that may have facilitated or impeded human migration during the terminal Pleistocene. We present model results from a high-resolution (1/6°) eddy-permitting general circulation model (MITgcm) and a lower resolution model (GENMOM) to evaluate changes in current velocity of the Alaska Current system between glacial and modern climate states, as well as in response to increases in regional freshwater discharges and intermediate sea level conditions. We compare paleo-SST reconstructions from the North Pacific with simulated SST from the transient deglacial simulation in iTRACE (37) for major climate intervals between the LGM and early Holocene. These paleoenvironmental reconstructions and models suggest possible time intervals when southward dispersal along the Northwest Coast was most feasible for people and provide insight into factors that may have influenced subsequent coastal habitability.