Eight Spear Points and a Kelp Highway

On April 14, 1952, the archaeologist Emil Haury arrived at a dry wash near Naco, Arizona, where the Navarrete family had already pulled two stone points out of a bone bed the year before [S7]. Over the next five days his crew uncovered six more — eight Clovis projectile points in total, all in direct association with a single Columbian mammoth [S7]. The dig wrapped on April 18 [S7]. For most of the seventy years that followed, that mammoth was the founding scene of the American story: a band of big-game hunters with Clovis fluted points sweeping down out of an interior corridor through Canada around 13,000 years ago, butchering megafauna as they went [S5][S7].

That story is now badly incomplete. The Clovis-first model — humans arrived about 13,000 years ago and nobody got here sooner — is effectively dead [S1].

A child walking on wet ground

In 2019, a Bournemouth University team began excavating preserved footprints on the exposed bed of a vanished lake at White Sands National Park in New Mexico, working under a National Park Service permit [S10]. Trackways turned up made by adults, juveniles, and at least one child — barefoot people walking on damp ground at the lake edge [S10]. In 2021, the team published radiocarbon dates from seeds of Ruppia cirrhosa, an aquatic plant, putting the prints at 21,000 to 23,000 years old [S10]. That was thousands of years older than Clovis, and it landed inside the Last Glacial Maximum — when much of North America was supposed to be empty.

Michael Waters of Texas A&M's Center for the Study of the First Americans pushed back hard. Ruppia is aquatic, he argued; it could be ingesting old dissolved carbon from groundwater and registering older than the sediment it sat in [S2][S10]. Press accounts called it the hard-water, or reservoir, effect [S10]. The underlying worry is the same one any first-year geochemistry student learns to worry about.

In 2023 the team came back with new dates from terrestrial conifer pollen — a different material with a different carbon source — and the ages matched [S10]. In June 2025, Science Advances published a third round: radiocarbon on mud cores from the trackway layers themselves [S1]. The team now has 55 dated samples — mud, seeds, pollen — pointing to between 20,700 and 22,400 years ago, closely matching the original 21,000–23,000 BP range [S1][S2]. Waters and a few others still want corroboration from other Last Glacial Maximum sites before they call the case fully closed [S10]. But the working consensus has flipped. People were in interior North America while the ice sheets were at their maximum extent — twice as long ago as the Naco mammoth.

How they didn't get here

For most of the twentieth century, the textbook map showed a clean diagonal arrow: Siberia, across the Bering land bridge, then south through an "ice-free corridor" between the Laurentide and Cordilleran ice sheets, down into Montana and the Great Plains [S5]. The corridor was the highway. Clovis was the arrival.

The problem is timing, and then it's biology. Cosmogenic nuclide exposure dating — measuring how long boulders along the corridor have been bathed in cosmic rays since the ice retreated — shows that the final opening of the corridor came well after pre-Clovis occupation [S5]. So the corridor cannot have been the route for the earliest Americans [S5]. And even after the ice was gone, sediment cores show the corridor floor was biologically barren for hundreds of years: no plants, no game, nothing to eat [S5]. You couldn't have walked it without starving.

That is the surprise that broke the model. The famous highway was, for centuries, a thousand-mile-long dead zone [S5].

The Kelp Highway

The replacement model is messier and wetter. It's called the coastal route — or, in the cinematic version, the Kelp Highway: people from northeastern Siberia following the rim of the Pacific by boat and along sea ice, foraging shellfish, sea mammals, and kelp-bed fauna all the way around the northern arc into the Americas [S5][S9].

Until recently the coastal route had one big problem: the outer coast was supposed to be glaciated. A 2024 Scientific Reports paper used radiocarbon, pollen, sedimentary ancient DNA, diatoms, and grain-size analysis from a lake-sediment core to argue that Topknot Lake on the west coast of northern Vancouver Island was not glaciated for the last 18,500 years [S9]. A refugium — exposed land, plants, animals — sat open along the Pacific rim while the interior was sealed [S9].

The technology here is the unsung star. Sedimentary ancient DNA, or sedaDNA, lets researchers read past ecosystems straight out of lake-bottom mud — no bones, no artifacts required [S9]. Cosmogenic nuclide dating tells them when a boulder first saw the sky after the ice released it [S5]. Marine sediment cores paired with climate-model simulations are being used to identify decadal "windows of opportunity" along the Pacific [S9]. These aren't better radiocarbon. They're new instruments answering questions radiocarbon couldn't reach.

A point from Hokkaido

In western Idaho, on a terrace above the lower Salmon River, sits the Nez Perce homeland site of Nipéhe — Cooper's Ferry, in the settler nomenclature [S3][S4]. Loren Davis of Oregon State has been excavating it for years. In 2019 he published in Science radiocarbon dates showing repeated human occupation starting between 16,560 and 15,280 calibrated years before the present [S3].

That is pre-Clovis by a comfortable margin, and old enough to predate the corridor opening — meaning whoever was at Cooper's Ferry didn't walk down through Canada [S3][S5]. They came around the coast.

But the artifacts are what made the field reread the map. The earliest Cooper's Ferry points are stemmed — narrow Christmas-tree silhouettes — and they closely resemble Upper Paleolithic points from Hokkaido and northern Honshu, Japan [S3][S4]. "The oldest artifacts uncovered at Cooper's Ferry also are very similar in form to older artifacts found in northeastern Asia, and particularly, Japan," Davis told his university's press office; he is now collaborating with Japanese researchers on direct comparisons of artifacts from Japan, Russia, and Cooper's Ferry [S4]. The argument isn't finished — a Siberian-interior origin for the earliest American toolkit remains a serious competing hypothesis among researchers — but the trans-Pacific stemmed-point parallel is now firmly on the table.

Down the funnel, into Arizona

Once people were south of the ice, where did they go? In the southwestern United States, the densest known Clovis-era cluster is not in Montana or New Mexico but along a single river in southeastern Arizona [S7]. The San Pedro River Valley runs north out of Sonora through the high desert toward the Gila, and within a fifty-mile radius along that valley sit nearly a dozen Clovis sites: Murray Springs, Lehner, Naco, Escapule, Leikem [S7]. They cluster the way a herd of mammoths clusters at a waterhole, because that is essentially what was happening.

Murray Springs, in what is now the San Pedro Riparian National Conservation Area, was excavated starting in the 1960s by C. Vance Haynes Jr. of the University of Arizona [S7]. It dates to roughly 12,900 calibrated years before present and is the richest Clovis-era butchery and tool-manufacture site in the US Southwest, with five buried animal kills and processing locations containing bones of mammoth, bison, horses, camels, canids, and rodents [S7].

A short drive away is Naco — the spot where Emil Haury's April 14–18, 1952 dig recovered six more Clovis points around the Navarrete family's earlier two, for eight points total embedded in a single mammoth carcass [S7]. The Navarretes had found the first two in 1951 [S7]. Haury's crew was on the ground five days [S7]. The San Pedro is not just a coincidence of preservation. It looks like a funnel — a watered corridor through the Sonoran and Chihuahuan borderlands where Clovis hunters could reliably find megafauna and the megafauna could reliably find water [S7]. If you wanted to draw a probable Pleistocene migration path through what is now Arizona, you would draw that valley.

The boy on the Anzick property

Anzick-1 is the only known Clovis-era human burial in the Americas — a young boy from the Shields River Valley in Montana, dated to roughly 12,990 to 12,840 years before present [S6]. In 2014, a team led by Eske Willerslev (Rasmussen et al.) published his genome in Nature [S6]. His mitochondrial DNA belongs to haplogroup D4h3a, a lineage still present in modern Native American populations [S6]. He is directly ancestral to many living Central and South American Native peoples and closely related to all Native American groups, though for some northern North American groups (Algonquian, Athabaskan) the relationship is sister-lineage rather than direct descent — a nuance often flattened in popular coverage [S6].

The other thing about Anzick-1 is what happened next. Sarah Anzick — who grew up on the property and went on to become a geneticist herself — coordinated with the Crow, Blackfeet, Northern Cheyenne, Sioux, and other tribes [S6]. On June 28, 2014, the boy was reburied on the original Montana hillside [S6]. The genome had been read; the modern descendants had effectively reclaimed him.

What ancient DNA keeps showing

The pure-Asia-to-America arrow keeps getting bent. A previously unknown lineage called the Ancient Beringians — represented by 11,500-year-old infants from the Upward Sun River site in Alaska — split from the American stem somewhere between roughly 22,000 and 18,000 calibrated years before present, before the Northern and Southern Native American clades split from each other between 17,500 and 14,600 years ago [S8]. Some Beringian populations stayed isolated in Alaska for thousands of years after others moved south; 9,000-year-old remains from Alaska's Seward Peninsula are one example [S8].

And movement was not one-way. Ancient DNA now shows that Native American ancestry flowed back across the Bering Sea into northeastern Asia thousands of years ago, contradicting the simple one-way-migration picture [S8]. The Beringian Standstill hypothesis — that ancestral Native Americans were genetically isolated in Beringia for thousands of years and became distinct from Siberians before any of them moved south — has gone from speculative to well-supported on the strength of those genomes [S8].

What's actually settled

Almost nothing is settled, and that's the honest version. The coastal route is now the dominant explanation for the first arrivals, but several researchers argue for a dual scenario in which the inland corridor was used by later waves — including ancestral Clovis populations — once it became habitable around 13,000 to 14,000 years ago [S5]. The Japanese stemmed-point connection is provocative but not yet resolved [S3][S4]. Even the White Sands dates, now triple-confirmed across seeds, pollen, and mud, still have skeptics asking for parallel Last Glacial Maximum sites [S10].

What has changed is the shape of the question. For decades the field argued about a single arrival date, a single route, a single people, a single technology. The current picture is plural: multiple lineages, in motion at different times, around a coast that was just barely open — leaving footprints on a New Mexico lakebed twenty-two thousand years ago and spear points in an Arizona mammoth nine thousand years after that [S1][S7]. A child's grave in Montana whose DNA links him to people now living thousands of miles south [S6]. A boy reburied by the descendants of his own descendants [S6].

The mammoth at Naco is still part of the story. It just isn't the start of it anymore.