Marine offset thresholds

[Chris Chinnock (Chris)]

Hi all, 

I have recently had a date back from a lab in the UK. One of the samples  has been calibrated with a mixed terrestrial and marine calculation. 

In the report it states that 'Human bone collagen with a δ¹³C value above -20‰, accompanied by a raised δ¹⁵N value, is taken to indicate a marine component in the diet'. The value for the sample in question is -19.5‰. However, in the Historic England guidance document, table 1 states that a typical δ¹³C value is -19‰, and that a modest component of marine or freshwater fish 'can be indicated by only slightly elevated δ¹³C values (>-19.0‰)'. By this measure, the sample in question would not require a mixed calibration. Is that right? Or is it the elevated δ¹⁵N value that is the key here? I'm assuming it is but I'm unsure how to apply my judgement.

Additionally, I figured I would take the opportunity to try and replicate the mixed calibration in OxCal. After much faffing around I got something to work, I think. However, it has given me an unmodelled and modelled result. The unmodelled result matches the one presented in the attached SUERC report but the modelled one is different. Have I done something wrong here. Should I be using the modelled result when presenting rather than the unmodelled (code used below)?

Hopefully those aren't dumb questions, I'm relatively new to dealing with mixed results and so was unsure on how to interpret this one.

Thanks

Chris 

 {

  Curve("IntCal20","intcal20.14c");

  Curve("Marine20","marine20.14c");

  Delta_R("LocalMarine4",-150,52);

  Mix_Curves("Mixed4","IntCal20","LocalMarine4",18,10);

  R_Date("SUERC-127734 (1960)",1887,25);

 };

[Christopher Ramsey]

Dear Chris

That code looks right - and should be what you get if you use the tool: [Tools > Curve > Mixed] to set up the model.

The unmodeled result uses classical statistics with normal distributions to generate the range. The modelled result uses an MCMC algorithm and takes account of some other effects. With a single date these are normally quite minor and I think in this case the ranges are very similar. The key differences are:

1. Although the mixture is given as a normal distribution - in practice you cannot have less than 0% marine or more than 100%.

2. The Delta_R and mixtures are not all equally likely given the nature of the calibration curve. The model assumes a uniform prior on age which will have an effect on the marginal posteriors of the other parameters.

3. If part of a larger model (not the case here) the mixtures and Delta_R are also influenced by other constraints.

Best wishes

Christopher

> --
> You received this message because you are subscribed to the Google Groups "OxCal" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to oxcal+un...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/oxcal/71a0055b-2841-485e-8932-f14f2067cccdn%40googlegroups.com.

[ineslopezl...@gmail.com]

Hello Chris

To answer the first part of your message: I don't think there are universal boundaries for what is aquatic and what is terrestrial. Ideally you would need to find a good baseline for your specific area and period, if there isn't one in the literature for your area you can try to establish your own by finding stable isotope data for purely terrestrial samples in the area. If you decide that your sample has an aquatic dietary component and you estimate the % (a decision which you should make as transparent as possible on your paper) you would also need to decide whether this is marine or freshwater (based for example on the location or other evidence fom the site, any fish bones or shellfish?) and find a local reservoir correction (again, try to find literature for the local area and period).

There is lots of literature on the topic; for example, this is a good starting point:

Cook, G.T., Ascough, P.L., Bonsall, C., Hamilton, W.D., Russell, N., Sayle,
K.L., Bownes, J.M., Scott, E.M., Best Practice Methodology for 14C Calibration of Marine and Mixed
Terrestrial/Marine Samples, Quaternary Geochronology (2015), doi: 10.1016/j.quageo.2015.02.024.

Hope this helps!

Ines

[Chris Chinnock (Chris)]

Thanks Chris and Ines, 

Both very helpful answers and much apreciated.

All the best

Chris

[ineslopezl...@gmail.com]

Apologies, just a small clarification, when you think about the provenance of your sample, you also need to think if it's a local or extralocal individual that just came to the area towards the end of his/her life - you may need to measure other stable isotopes to find out where your individual may have come from to ascertain which baseline is applicable. In specific periods you could have a cemetery with different individuals each with different isotopic baselines...

[MILLARD, ANDREW R.]

Hi Chris,

 

A cutoff of -19‰ seems a bit high to me for English samples, and I would think about using a marine correction above -20‰ (which I did for Medieval samples from Hull[1]). Looking at the lowest δ¹³C in other humans from the area may help in deciding an appropriate value.  But -19.5‰ giving an estimated 18% marine contribution seems too high to me. The 100% marine endpoint is about -12 to -14‰ (depending on which resources are taken as representative), and when combined with a -20‰ terrestrial endpoint this implies a marine contribution more like 8% than 18%.

 

Elevated δ¹⁵N is an indicator of aquatic food sources, but a small shift in δ¹³C caused by a small marine component will be accompanied by a small shift in δ¹⁵N. However, freshwater foods can cause elevated δ¹⁵N with little difference in δ¹³C from a terrestrial diet.

 

[1] Roberts CA, Millard AR, Nowell GM, Gröcke DR, Macpherson CG, Pearson DG, Evans DH. 2013. Isotopic tracing of the impact of mobility on infectious disease: The origin of people with treponematosis buried in Hull, England, in the late medieval period. American Journal of Physical Anthropology 150:273-285.

 

Best wishes 

Andrew 

-- 

Dr. Andrew Millard 

Associate Professor of Archaeology,

Durham University, UK 

Email: A.R.M...@durham.ac.uk  

Personal page: https://www.durham.ac.uk/staff/a-r-millard/ 

Dunbar 1650 MOOC: https://www.futurelearn.com/courses/battle-of-dunbar-1650 

 

From: ox...@googlegroups.com <ox...@googlegroups.com> On Behalf Of ineslopezl...@gmail.com
Sent: 15 October 2024 10:16
To: OxCal <ox...@googlegroups.com>
Subject: Re: Marine offset thresholds

 

[EXTERNAL EMAIL]

--

You received this message because you are subscribed to the Google Groups "OxCal" group.
To unsubscribe from this group and stop receiving emails from it, send an email to oxcal+un...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/oxcal/dd041579-230a-4bc9-915a-550029747e87n%40googlegroups.com.

[Chris Chinnock (Chris)]

Thanks Andrew, I'll take a look at that paper.

The lab report states that marine component 'was calculated using end-members of -21.0‰ (fully terrestrial) and -12.5‰ (fully marine) with an uncertainty of 10% applied'.

I've clearly got some more reading to do though. I think I had hoped there were some accepted cut offs that might help me know when and where to apply mixed calibration. I can see now that there are too many variables for that.

Cheers

Chris