Mammoth Deathtrap: New Study Pushes South Dakota's Mammoth Trap Back Over 100,000 Years

Despite growing up in South Africa surrounded by African elephants, I have always had a soft spot for their extinct (often) fluffy relatives, the mammoths. I did not know this as a child, but there are many mammoth species: some larger, others smaller; some found only in Europe, others found only in North America. Among them are the Columbian mammoths (Mammuthus columbi), whose largest known concentrations have been found at Mammoth Site.

Mammoth Site is an ancient sinkhole located in Hot Springs, South Dakota. What makes it so special is that it is packed with over 60 Columbian mammoths, apparently all male and mostly young, the greatest concentration anywhere in North America. But figuring out when the sinkhole was filled can tell you a lot about ancient climates and the animals that roamed the Northern Great Plains.

The problem is that the site has multiple dates, and they don't agree with one another. According to earlier radiocarbon dates, the bones themselves date to around 42,000–23,000 years ago (42–23 ka); meanwhile, uranium-series dating and one thermoluminescence date suggested an age of roughly 150–101 ka. But which date (if any) is more reliable when they are both so different?

To figure out which date (if any) was more accurate, Shannon Mahan and her colleagues set out to redate the site, this time using luminescence dating. They concluded that the radiocarbon dates were probably a severe underestimation of the site's true age, with analyses returning ages of approximately 232 to 160 ka.

The Columbian Mammoths of the Black Hills

During the Pleistocene, the site lay within around 200 miles of the Missouri River, which marked the farthest advance of the great Laurentide Ice Sheet, a massive glacier that covered much of North America. The sinkhole likely formed within a large dome of uplifted rock made of soft shales, limestone, and gypsum-rich beds. As groundwater penetrated these layers, it dissolved some of them, probably causing the ground above to collapse and create a sinkhole. But it wasn’t just an easily avoidable hole in the ground; the warm water welled up and filled it, so a mammoth wading through the shallows may not notice the deeper deathtrap in the middle.

Mammoth Site was first discovered in the summer of 1974 when a heavy equipment operator, George Hanson, leveling a hill for houses, hit something unusual with his blade: a massive tusk and several giant bones. Local universities weren't interested, so Hanson showed his son, who had taken some anthropology and geology classes, and his son called his former professor, Larry Agenbroad. Since then, a mammoth graveyard has been uncovered alongside various other extinct creatures, including the fearsome short-faced bear and the extinct camel (yes, America had camels, weird right?)

What's more, this was no kill site where humans hunted and accumulated bones; rather, it was a natural sinkhole into which these ancient giants fell while still alive.

Fifty years later, the site remains an active dig, with much of the original bone-filled sediment in place, and doubles as a museum where people can observe the excavations.

Very Different Dates

Most early work at the site relied on six radiocarbon dates that clustered between 43 and 21 ka. However, directly dating the bones is difficult because the fossils lack collagen, the protein typically used to radiocarbon-date bones. The dates were mostly based on the bones' bioapatite (the mineral part) instead. But bioapatite can be problematic.

In wet ground like a sinkhole, the bone is subject to constant chemical reactions with the surrounding water. The absence of collagen was already a worrying sign that the bone's chemical composition had been altered.

"It is possible for bone apatite to be altered after burial and in most instances, it is highly likely to be altered," said Emma Krolczyk, a geologist with the U.S. Geological Survey.

"During the process of groundwater moving through the site, older carbonate in the bone could have been dissolved and recrystallized with younger, groundwater-derived carbon," she explained. This can reset the radiocarbon signal.

So the radiocarbon dates may not be meaningless; they may just be dating when the sinkhole dried out and hardened, not when the mammoths died.

A later attempt, using uranium-series dating, which measures the slow decay of uranium into thorium, was made on a mammoth tooth's enamel. One sample suggested an age of around 129 ka, another returned 150 ka. Much older than the radiocarbon ages but again with a caveat: for uranium-series to work, the dated material needs to be in a closed environment, aka, no new uranium moving in or out. But according to the study, the checks needed to determine whether this was the case were never reported, so the reliability of the dates remains uncertain.

The third method was thermoluminescence. The date was taken from the upper part of the site and never properly written up. All that survives is a date of around 101,000 years in the archives.

Emeritus researcher at the University of Arizona, Jim Mead, who helped dig the site in 1974 and returned as its science director in 2016, said he had always wanted to redate the site, especially since Agenbroad

"was uncertain about the reliability of the uranium series dates on the same bone apatite — it seemed like they were too old because they did not match the radiocarbon."

When he came back, redating was a priority.

Illuminating the Age of Mammoths

The researchers turned to infrared-stimulated luminescence dating, which works by determining the last time sediment grains were exposed to sunlight. As minerals like quartz or feldspar are buried, they soak up natural radiation and accumulate electrons at defects within the crystal structure. Sunlight wipes those signals clean, effectively resetting the "clock."

The authors collected ten samples, six from the bone bed, including material lying right beside mammoth remains. Four came from nearby river terraces and the modern Fall River channel for comparison.

What they found was illuminating (that was a bad pun, I know!).

The new feldspar dates ranged from around 232 ka near the base of the current excavation to around 159 ka near the top, with one out-of-place date of 131 ka (we will get back to that).

But how could the researchers be sure their results were trustworthy, when the other dating methods had their own problems?

The authors gave a number of reasons, among them that the feldspar grains passed the standard laboratory tests. Two separate laboratories analyzed splits of the same sample and returned similar results; all but one date is ordered correctly, with the oldest ages near the bottom and the youngest near the top.

These were also not completely unexpected ages; they broadly matched ages from earlier uranium series and thermoluminescence dating.

The team also checked the obvious problems. Warm water could have concentrated uranium in the sediment, throwing off the radiation calculation, but they found the effect faded fast with distance from the old spring vent and barely moved the dates. Sediment washing into a murky sinkhole might never have seen enough daylight to reset the clock, so they dated modern sand from the Fall River and found at least a third of the grains got enough. And a younger pond deposit nearby gave radiocarbon and luminescence ages that the team says broadly agreed, though not within error.

Another point came from the mammoths themselves: their characteristic form (basically, how they looked) was thought to be established by around 255–130 ka, which broadly aligns with the new window.

This does not make the chronology airtight, but it does make it plausible. Remember that weird 131 ka sample in the middle of the sequence; it's a genuine outlier. Layers should get younger as you move up, so a young date stranded in the middle is a puzzle. There are many things that might have caused it, from mammoths possibly churning up the pond as they walked through it to inaccurate dose rate calculations (aka an estimate of how much radiation the grains were soaking up), though the authors could not settle on a cause.

The other caveat is the material used: the study had to rely on feldspar instead of quartz because the quartz signal was saturated (basically, it was full and couldn't capture any more). This in and of itself suggests the site had to be over 75 ka, as saturated quartz can only give a minimum age, and here that minimum was 75 ka. But feldspar has its own problems; it can slowly leak some of its stored signal, which needs correcting for.

The team also had to estimate how wet the sediments were over time, since water soaks up some of the radiation the grains would otherwise absorb. Estimating how wet or dry something was over the past few thousand years, let alone hundreds of thousands of years, is difficult.

They also needed to figure out how quickly it filled to determine how long it remained a deadly little trap into which mammoths fell. But the answer depends on the fill rate you use: 12, 35, or 62 mm per year. They favored around 12 mm per year, as they thought it fit with the fine-sediment layers seen in the deposit and the long span implied by the dates.

If the new dates hold, the mammoths of Hot Springs died far earlier than previously thought. But why should that matter, if a mammoth died 40 ka or 150 ka ago? It could help us picture what that past environment looked like; radiocarbon dates place their deaths just before the Last Glacial Maximum (the peak of the last Ice Age), a completely different time when mammoths had only about 10,000 to 30,000 years left before the last of them died out.

Other well-dated deposits of this age are rare on the northern Great Plains, so the site offers a unique glimpse into a past we know comparatively very little else about. On top of that, the site's mammoths have long been compared to Columbian mammoths found elsewhere. But if these mammoths are around ten times older than previously assumed, that shifts the comparative benchmark. It's like comparing two humans for similarities and differences, then discovering one of them lived 150 ka ago. It wouldn't be a fair comparison.

The story is not over; the bottom third of the sinkhole has not been excavated, and the authors suspect it is even older, so we don't yet know how long this ancient death trap remained open or how many more mammoths may still be buried within.

Why do you think that one sample came back so much younger, and why do you think every mammoth found here so far appears to be male?

Let me know what you think, and if you'd like to support this blog and the person writing it, you can chip in once, monthly, or yearly. Only if you'd like to.

References

History - The Mammoth Site. (2022, May 31). Mammothsite.Org. https://mammothsite.org/history/

Mahan SA, Mead JI, Hanson PR, Krolczyk ET, Jass CN. Age reassignment using luminescence dating for The Mammoth Site, Black Hills, South Dakota, USA. Quaternary Research. Published online 2026:1-18. doi:10.1017/qua.2026.10102