Comment on Bennett et al.’s Evidence of humans in North America during the Last Glacial Maximum: Looking promising, and I have a couple of minor questions

There has been no shortage lately of claims of very early archaeological evidence from the Americas, from the stunningly early but very problematic Cerutti Mastodon at 130kya, to the slightly less stunning and still problematic Chiquihuite cave at 30kya, to the more sedately dated definite maybes at Cooper’s Ferry and the Gault site.

The latest is freshly released by Bennet et al. (2021) and combines a relatively early date (23kya) with quite robust support. If it holds up, this would be the earliest solid date for archaeological remains in the Americas at the Last Glacial Maximum LGM), and would strongly hint that there is archaeological material hiding out there dating back to the last interglacial. If we have material from the LGM, the people who left it here were probably here much earlier.

The Claim

Bennet et al. report human footprints recovered from 21-23ky old deposits at White Sands National Park in New Mexico. The deposits in which the footprints were found were dated using seeds of aquatic plants. They report that some of the footprints had seeds embedded in them.

They have definitely found footprints that are unambiguously human, which removes many of the problems we encounter at sites that only have claimed stoned tools, for example, which might or might not have been made by humans.

On my patented scale of “could be real” to “this is real” (scale goes: could, could, might, might, probably, probably, this is real), this one so far is a strong might be real, slightly better than might be real,  and provisionally just short of probably, but not yet probably real. Of course, “this is real” is best treated as a utopian notion in Archaeology, so I won’t discuss it here.

The Support

The good news here is that the carbon dates on the seeds are tightly clustered and accord well with the stratigraphy across the site. There is no doubt that the site is archaeological (because footprints), the dating is internally consistent, and the dates are well associated with the footprints, from what I can see in the article. If the dates are good, this is LGM archaeology in the Americas.

Some Remaining Questions

The biggest issue is that there could be a systematic bias in the dating throughout the site, making all carbon dates older than they really are. This could be caused, for example, by a reservoir effect (the authors speak of a hard water effect in this case). The authors are very aware of this and do a good job of trying to address it, but  a few questions remain.

To be clear, the bias in the dates would have to be very substantial to make this find unremarkable in any case (Thanks to Eva Hulse for that comment). Whether they are 17ky old, or even 15ky old, this is very significant.

To address the question of bias, Bennett et al. point to another study (Allen et al. 2009) that shows that there was no such effect on dates from plant remains for the period between 44kya and 25kya, and argue that “an improbable series of events would be required to introduce a large hard-water effect by ~23 ka when such effects were minimal for the previous ~20 thousand years.” True. Still, I would feel better about the argument if the comparative study could be extended to overlap fully with the dates of the footprint bearing layers.

They also argue that given the local geology and hydrology, it is “unlikely” that a hard water effect would develop. The heart of the argument here (found in the supplemental material) is that “…the shallow depth of the water and turbation by wind would have facilitated the exchange of carbon between the water and atmosphere, thereby negating any deficiencies introduced by the groundwater.” That makes a lot of sense, and is the sort of thing that cries out for some experiments or some comparative observations of some kind. I would feel better about the whole thing if this was solidified a bit.

The authors attempted to further constrain the dates of the footprint bearing layers by dating a gypsum rich layer (i.e. a block of gypsum) found near the top of their sequence, and therefore overlaying most of the footprints. They concede, however, that the uranium series date on the gypsum is not very good, because of the local conditions. The Uranium series dating is in the ball park of the carbon dates (about 25kya), but it isn’t the clincher that we might like to see here.

And then there are the seeds themselves. This is more of a niggling, back of the mind sort of concern, but I would like it addressed at some point. The article clearly states that some seeds were found embedded in the footprints, and that each of 14 samples dated consisted of up to 60 seeds taken from a variety of locations on the site (all very good), but unless I missed it, there is no map of where the dated samples were collected, and no statement that any of the dated seeds came from any of the footprints. 

It would be nice to at least have clarity on that last point. From everything I can see here, the dated seeds are clearly associated with the footprint bearing layers, and I don’t think it would be a fatal blow to the claim if the dated seeds were not the ones found actually in the footprints. but I feel this just needs cleaning up.

In short, this is promising, and it can be made stronger by addressing a few issues. The main one that really needs to be ruled out more fully is the possibility of a systematic bias in dates because of a reservoir (hard water) effect. I really hope the authors focus on that for the next paper. They almost do that in the current paper, which is still better than if they almost did it.

References

Allen BD, DW Love, RG Myers 2009. Evidence for late Pleistocene hydrologic and climatic change from Lake Otero, Tularosa Basin, south-central New Mexico. N. M. Geol. 31, 9–25 (2009).

Bennett et al. 2021. Evidence of humans in North America during the Last Glacial Maximum, Science 373:1528-1531.

https://www.science.org/doi/10.1126/science.abg7586

Supplmental materials: https://www.science.org/doi/suppl/10.1126/science.abg7586/suppl_file/science.abg7586_SM.pdf