The River That Vanished: How Scientists Traced the Colorado's Lost 5-Million-Year Journey Through an Ancient Mega-Lake

The Colorado River has been flowing through western Colorado for at least 11 million years. It first emerged from the western end of the Grand Canyon roughly 5.6 million years ago. Between those two facts lies a gap — approximately 5 million years during which the river essentially disappeared from the geological record, leaving no downstream sedimentary trace of its passage [1].

A team of geologists from UCLA, the U.S. Geological Survey, the Arizona Geological Survey, the University of Washington, and Paradise Valley Community College now claims to have solved the mystery. In a paper published April 16, 2026 in Science, lead author John He and corresponding author Ryan Crow present evidence that the river spent those missing millennia flowing into a vast lake in what is now northeastern Arizona — an ancient body of water the researchers informally call Bidahochi Lake, after a local geological formation on the Navajo Nation [1][2].

The Evidence: Zircon Crystals as Geological Fingerprints

The core methodology behind the discovery is a technique called detrital zircon geochronology. Zircons are microscopic crystals formed when magma cools. Because they incorporate uranium atoms that decay into lead at a known rate, each crystal carries a radiometric timestamp of its origin. By firing lasers or ion beams at hundreds of individual zircon grains extracted from a sandstone sample, geologists can read the uranium-to-lead isotope ratios and determine both the age and geochemical signature of each crystal [1][3].

"Zircons are some of the oldest fragments of our Earth," He said. "By looking at the age and geochemical signature of zircons, we can tell where sediment originated" [1].

The researchers collected sandstone samples from the Bidahochi Formation — a sequence of sedimentary layers in northeastern Arizona — and compared the detrital zircon signatures with those from known ancestral Colorado River deposits both upstream and downstream. The results showed that sediments deposited approximately 6.6 million years ago in the Bidahochi basin closely matched the zircon profiles of the Browns Park Formation in northern Utah and Colorado, a well-documented Colorado River deposit [1][2].

The match was not limited to isotopic chemistry. Field examination of rock layers from this time period revealed ripple patterns characteristic of a strong river flowing into standing water. Paleontologists also identified fossils of large fish species with enlarged fins and slender tail bases — anatomical features associated with fast-flowing river environments, not still lakes [1][3]. Together, these independent lines of evidence pointed to the same conclusion: the Colorado River was supplying water and sediment to the Bidahochi basin millions of years before it began flowing through the Grand Canyon.

The Lake Spillover Hypothesis

The picture that emerges from the data is one of a river that, between roughly 11 and 6 million years ago, flowed southwest from the Rockies and terminated in a large, internally drained basin east of the present-day Grand Canyon. Over hundreds of thousands of years, the lake grew — at its maximum extent, possibly spanning more than 90 miles (150 kilometers) across [1]. Eventually, the water level rose high enough to overtop the Kaibab uplift, a geological barrier that separates the Bidahochi basin from the western Colorado Plateau. When it did, the overflow established the river's current course through the Grand Canyon and onward to the Gulf of California, which it reached roughly 5 million years ago [2][4].

Source: He et al. 2026, Science

Data as of Apr 16, 2026CSV

"In some ways, you could really think of it as the birth of the Colorado River that we know today," He said [1].

The lake spillover hypothesis is not new. Geologists have debated some version of it for decades. But earlier iterations lacked strong evidence that the Colorado River itself — as opposed to smaller, local drainages — actually fed the Bidahochi basin [5]. The new zircon data, according to Ryan Crow, provides that missing link. "It's clear that this lake had to have played a role in the formation of the canyon," Crow said. He described the spillover mechanism as "perhaps an easier and much simpler and more likely mechanism" than competing alternatives [5].

Dating a River's Disappearance: Precision and Uncertainty

How do scientists determine that a river "disappeared" from the geological record for a specific interval? The answer lies in the absence of evidence at downstream locations where deposits should exist. The Muddy Creek Formation, which lies across the river's eventual downstream course in Nevada, contains no recognizable Colorado River sediment, indicating the river had not yet reached that area while the formation was accumulating [6]. The youngest Muddy Creek deposits date to approximately 6 million years ago, providing an upper bound on when the river arrived downstream.

The "5 million years" figure is an approximation derived from the difference between when the river was present in western Colorado (about 11 million years ago, based on upstream deposits) and when it first appeared at the Grand Canyon's western exit (about 5.6 million years ago, based on downstream sedimentary evidence) [1]. The margin of error in uranium-lead zircon dating is typically 1–2% for Precambrian zircons, though it can be wider for younger samples. The 6.6 million-year date for Bidahochi sediments, for instance, depends on which zircon populations are sampled and how maximum depositional ages are interpreted. The gap could plausibly be 4 to 6 million years rather than precisely 5 [1][2].

The Grand Canyon Age Debate

The new findings land in the middle of one of geology's most contentious disputes: how old is the Grand Canyon? Estimates have ranged from 5 to 70 million years, and the answer depends partly on what counts as "the Grand Canyon" [6][7].

The emerging consensus — if it can be called that — is that the canyon is not a single feature with a single age. Different segments formed at different times. The "Hurricane" segment in the west may have been carved 50 to 70 million years ago. The "Eastern Grand Canyon" was likely cut between 15 and 25 million years ago. The "Marble Canyon" and "Westernmost Grand Canyon" segments, by contrast, appear to date only to the last 5 to 6 million years [6].

The new study supports the view that the Colorado River's arrival through the Bidahochi spillover was the catalytic event that linked these older, disconnected canyon segments into the continuous gorge visible today. As John Douglass of Paradise Valley Community College put it, "Geologists have proposed over a dozen hypotheses for the canyon's formation and the Colorado River's path" [1]. The spillover model does not claim to explain all segments of the canyon — only the mechanism by which the modern river established its current through-flowing course.

Dissenting Voices: The Paleocanyon Alternative

Not all geologists are persuaded. Karl Karlstrom of the University of New Mexico, one of the most prominent researchers in Grand Canyon geology, has raised pointed objections. Karlstrom argues that "the key details of [the authors'] proposed lake spillover conclusion remain untested" [5].

His central counterargument is that an older "paleocanyon" had already cut a path across the Kaibab uplift well before the Colorado River arrived in the area. Data gathered by Karlstrom and his colleague Laura Crossey suggest that the Little Colorado River carved a notch through the Kaibab Arch approximately 10 million years before the main Colorado River reached the Bidahochi basin [5][8]. If such a pre-existing passage existed, water from the proto-Colorado would have drained through it rather than pooling into a lake deep enough to spill over.

William R. Dickinson, in a 2013 paper published in Geosphere, went further, arguing that the Bidahochi paleogeography indicates "Hopi Lake was a playa system that never achieved appreciable depth." Dickinson calculated that the maximum elevation of Bidahochi lakebeds was not high enough to overtop the Kaibab-Coconino Plateau — a lake surface at or below 2,000 meters could not have breached the barrier in its present configuration [9].

Additional alternative mechanisms have been proposed. One involves karst piracy — the dissolution of underground limestone cave networks by groundwater until roof collapse exposed an incipient river channel [8]. Another invokes stream capture, in which small drainage systems on the western side of the plateau eroded headward (upstream) until they intercepted and "captured" the Colorado River's flow [5].

Karlstrom's own preferred model proposes that the Grand Canyon was carved in multiple stages by different drainage systems over tens of millions of years, with the modern Colorado River arriving last — integrating older paleocanyons into a single through-flowing system approximately 5 to 6 million years ago. This model does not require a mega-lake or a catastrophic spillover event [8].

Gaps in the Geological Record: How Unusual Is This?

Multi-million-year gaps in river records are not unique to the Colorado. The geological record is, by nature, incomplete. Erosion, tectonic uplift, and basin subsidence routinely destroy or bury sedimentary evidence of ancient waterways.

The Platte River system, for example, has been reconstructed from "fossil rivers" spanning 2 to 33 million years, with significant gaps in the record [10]. At a broader scale, geologists have identified the "Great Unconformity" — a worldwide gap in the rock record spanning roughly a billion years — that appears on every continent [11]. Such unconformities are created when erosional forces (rivers, glaciers, tectonic uplift) strip away sediments faster than they accumulate, effectively tearing pages from the geological diary.

Still, a 5-million-year gap in the record of a major continental river is notable precisely because it left enough adjacent evidence — upstream and downstream deposits — for researchers to bracket the missing interval. Many ancient river systems simply vanish from the record entirely, with no downstream bookends to define the gap. The Colorado case is unusual not because the gap exists, but because enough surrounding evidence survived to make the gap identifiable and, eventually, solvable.

Source: OpenAlex

Data as of Jan 1, 2026CSV

From Deep Time to Present Crisis

Understanding where the Colorado River flowed millions of years ago might seem disconnected from the practical realities of a waterway that today serves 40 million people across seven states and Mexico [12]. But the connection is more direct than it first appears.

Ancient river courses map ancient aquifers. When a river flowed through a basin for millions of years, it deposited sediments that today form permeable layers capable of storing groundwater. Identifying the Bidahochi basin as a former Colorado River terminus raises questions about what hydrological legacy that ancient lake left behind — potential groundwater reserves in a region where the Navajo Nation already faces acute water scarcity [12][13].

The Colorado River's modern crisis is well documented. The river has been overallocated since the 1922 Colorado River Compact divided its waters among seven states based on flow estimates that proved optimistic. Average flows have declined nearly 20% since 2000, with roughly half of that decline attributable to rising temperatures [12]. The Bureau of Reclamation has estimated that basin states need to reduce consumptive use by up to 4 million acre-feet — about a quarter of the total allocated volume [12].

Source: Bureau of Reclamation

Data as of Dec 1, 2025CSV

Lake Powell and Lake Mead, the two massive reservoirs that store the river's water, have fallen to roughly 34% of capacity [12]. Temperatures in the basin are projected to rise another 2 to 5 degrees Fahrenheit by 2050, which could reduce flows by an additional 10 to 40% [12].

In this context, geological research serves two purposes. First, mapping ancient river paths and lake deposits can identify previously unknown aquifer systems — fossil water stored in sediments laid down by rivers that no longer flow there. Second, understanding how the Colorado River responded to past climatic and tectonic changes provides a longer baseline for modeling its future behavior under conditions of sustained warming and reduced snowpack.

What Remains Unresolved

The new study fills a significant gap in the Colorado River's biography, but several questions persist. The Bidahochi Lake deposits are largely eroded, making it difficult to determine the lake's original depth, extent, and duration [1]. Without better constraints on the lake's maximum surface elevation, the debate over whether it could have overtopped the Kaibab uplift will continue.

The precise mechanism of spillover — whether it was a gradual rise followed by steady overflow, or a more catastrophic breach event — is also unresolved. And the relationship between the spillover (if it occurred) and the pre-existing paleocanyons identified by Karlstrom and others remains to be integrated into a unified model [5][8].

What is not in dispute is that the Colorado River reached its current configuration relatively recently in geological terms, and that its path through the Grand Canyon represents the last chapter in a much longer, more complicated story. Whether that chapter began with a lake overflowing or a river finding an existing gap, the result was the same: a 1,450-mile waterway connecting the Rocky Mountains to the sea, carving one of the planet's most recognizable landscapes in the process.

As Douglass noted, the sheer number of competing hypotheses — more than a dozen — reflects the genuine complexity of the evidence. "This is a story with as many twists and turns as the Colorado River that carved it" [14].