The Next Steps in Near-Infrared Collagen Prediction for Archaeology

A recent study published in the Journal of Archaeological Science investigated how near-infrared (NIR) prescreening could serve as a better alternative technique for studying collagen in archaeological bone compared to percent nitrogen analysis and other destructive sampling techniques.^1,2

In this brief Q&A, Christina Ryder, who is a postdoctoral researcher at Texas A&M University and the lead author of this study, explains the remaining limitations in using NIR spectroscopy for collagen prediction and how it is currently changing workflows for archaeologists and paleontologists.

Because the technique is rapid and field-portable, how do you envision it changing workflows for archaeologists working in remote excavations or large museum repositories with limited access to laboratory facilities?

Because NIR spectroscopy is rapid, nondestructive, and affordable, it has the potential to fundamentally change how archeologists and paleontologists approach sampling decisions, particularly in these remote excavations or in museum collections with limited laboratory access. In the field context, one of the biggest challenges is the permitting process. Obtaining these permits is very difficult and extensive, and it can involve a lot of writing that doesn't contribute to your tenure. It can also take years, so using the NIR prescreening ensures that the samples selected for export are likely to yield results.

A similar issue exists in museum collection, so the destructive sampling requests often require extensive review from these curators and the governing bodies that process the permits. By screening the specimens, you're increasing the likelihood of success, while reducing the unnecessary damage to the curated collections. We're at this interesting time in archeological science where the number of destructive techniques seems to be steeply increasing, but the number of samples on the shelves are not, so the fossils, ivory, antler, and dentin are finite resources that seem to be coming off the shelf faster than they're going back on. Because techniques are advancing so rapidly, which is incredible, it often leaves collection managers and curators a little frustrated because they're inundated with many destructive sampling requests, so this approach supports a more responsible stewardship of these archeological collections.

I'm currently in a postdoctoral position at Texas A&M University, and we are trying to establish the chronology of megafauna extinctions across North America at the end of the Pleistocene. What I do almost every single day is I go into museum collections, and I'm prescreening these megafaunas for collagen preservation. Once I find the samples that are suitable for radiocarbon dating, I pursue a destructive screening request (DSR), and then I get a radiocarbon date on these samples.

What I also do is I give the museum all of the information. I give the museum collagen predictions and consolidant identifications from everything that I collected, so they have that stored within their database. Depending on the curator or collection manager, maybe they'll use that information or maybe they won't, but five years from now, they’ll get a destructive screening request for a sample that I prescreened that looks truly dead with no collagen. They can use that information, or they can email me, but they can use that information to decide on the preservation of the sample. And then finally, what's really cool about this technique is it's so fast. Each spectrum is collected in approximately 3–5 seconds, and that allows me to average about 50 scans per spectrum. It's so fast that I could quickly move through a large collection, so I can spend two to three days in a collection. That allows me to investigate collections that have been dismissed because of low collagen yields in the past.

So, for example, consider Florida. Florida is obviously a warm, humid place, which is not great for collagen preservation. So, paleontologists studying Late Pleistocene samples in Florida have found high failure rates when they tried to radiocarbon date something. And so, for a long time, museum curators just said that they are not doing any more destructive sampling, but I would be able to work through all the many Late Pleistocene samples stored at the Florida State Museum in Gainesville in about a week.

As a result, we were able to find specific sites and elements that do look like they'll have collagen yields, so we will be able to get dates on these specimens that have never been dated before that we have, you know, we know these taxa were present in Florida, but we have no idea when. So, this [NIR prescreening] allows us to find the hidden gems because it's little effort. It's not after developing the models. But after developing in the models, we can quickly go through and find the specimen that could change our understanding of megafauna extinctions in North America at the end of the Pleistocene.

What are the main limitations or sources of uncertainty that remain for NIR-based collagen prediction, and what steps are needed before the method can be widely adopted as a standard prescreening tool in bioarchaeology?

Like any predictive spectroscopic method, NIR-based collagen prescreening has several important limitations and sources of uncertainty that I still need to address. One consideration relates to the penetration depth and having a more comprehensive understanding of how deep NIR light is penetrating through this material. Although NIR light penetrates more deeply than many other spectroscopic techniques in taxa with relatively thick cortical bone, such as wooly mammoths or American mastodons, two of the groups that did go extinct at the end of Pleistocene have extremely thick cortical bone, which is an adaptation to their massive size, and this cortical layer can often exceed the effective penetration depth of the instrument.

In addition, surface processes, such as diagenetic mineralization, turn bones into fossils, so you can see mineralization happen on the exterior surface of the bone, meaning that the internal cortical bone, which is our thick structured bone, the internal portion can be well preserved, but that exterior portion is genetically altered.

To combat this, what I try to do is collect spectra from many locations on the bone. If there is an exposed cross section, I try to target the cross section when you have that really thick cortical bone. Something I want to do in the future is capture the penetration depth and assess its limitations as it goes through different materials as well because understanding the penetration depth and different materials is critical as well.

In terms of limitations, there are two. One is the instrument cost barrier, and then the second would be model transferability. I have found that if you're using the same instrument, which I use the lab spec for standard resolution, my mother model transfers well. So, I've been using these models on three different NIR spectrometers now, and the models all transfer very well, but there are obviously many NIR devices currently on the market. My colleagues that I collaborate with frequently often have access to these instruments via shared facilities or through different departments of their institution, meaning the hardware can vary considerably.

So, what I'm currently working on now is a collaboration with four different universities, developing models that can transfer models to different devices. This will allow you to not need to start from the ground up on a new device. So far, direct standardization is proving most reliable, but I'm still collecting spectra, and the transformation model will likely be dictated by the child instrument. And then, the other issue I mentioned is instrument cost. High quality, portable NIR spectrometers can be expensive.

I'm also developing models using a lower cost handheld spectrometer, which have become more common in the past 5–10 years. So, the spectrometer I'm working with costs about $2,500, and it pairs with a smartphone. I'm still working on the proof of concept for these models. I need a little bit more data before I publish these models, but I think this will make it much more accessible for people in the field, especially for people who don't have the ability to purchase a $60,000 instrument.

Demonstrating that reliable prescreening can be achieved with a more affordable instrument will make the technique far more accessible to researchers working in field settings or in large museum collections. The portability of having something in your hand is great because archeology and paleontology happen in caves. People are squeezing through caves, and these are often in really remote regions where you do not have access to electricity. Therefore, a handheld spectrometer could potentially really shift the field.

References

1. Ryder, C.; Celis, G.; Devièse, T. et al. Refining Near-infrared Spectroscopy for Collagen Quantification: A New Predictive Model for Archaeological Bone. J. Arch. Sci. 2026, 185, 106448. DOI: 10.1016/j.jas.2025.106448
2. Wetzel, W.; Spectroscopy Staff. Collagen Preservation in Archaeological Bone Using NIR Spectroscopy. Spectroscopy. Available at: https://www.spectroscopyonline.com/view/collagen-preservation-in-archaeological-bone-using-nir-spectroscopy (accessed 2026-03-31).