Membership has its privileges

A new paper in PNAS by Erik Trinkaus covers the mortality patterns of old versus young adults in Neandertals, early modern humans in the Levant and early Upper Paleolithic people of Europe Trinkaus:mortality:2011. The paper has gotten a lot of attention from the press, including the NY Times: "Life Span of Early Man Same as Neanderthals". Reporters worldwide (so far, 30 articles in Google News) were relying on a press release issued from Trinkaus' university.

I read this paper and got a sinking feeling. The results and methods looked to me very similar to those used by Rachel Caspari and Sang-Hee Lee in a series of papers from 2004 onward Caspari:2004 Caspari:Lee:invariant:2005 Caspari:Lee:Minichillo:2005 Caspari:Lee:culture:2006. Their initial paper attracted comments by Kristen Hawkes and James O'Connell Hawkes:OConnell:2005, and Tom Minichillo Minichillo:2005. This has been an active and highly cited research pathway with papers in PNAS, the American Journal of Physical Anthropology and the Journal of Human Evolution. I was extremely surprised that Trinkaus hadn't cited any of this work.

Anyone can compare these papers and draw their own conclusions. In my opinion, the results reported in the press are not new.

Caspari and Lee compared the proportions of old and young adults in Neandertals, early Upper Paleolithic and (in their 2006 paper) early modern humans. They ignored juvenile remains and focused only on the subject of adult mortality, only in those two age classes. They assigned the dental remains to categories based on wear criteria and tested the significance of sample differences. They showed that early modern humans and Neandertals have similar ratios of old to young skeletal remains. They tested whether burial versus non-burial archaeological contexts might influence the observed ratio of older to younger individuals. They demonstrated limits on the phylogenetic interpretation of the results and suggested behavioral and cultural factors that could account for them. I think their work was clever and simple, but more important it has been highly cited and presented at national meetings. These are not obscure sources.

I'm not the citation police -- I probably make more errors than anybody. But in my opinion Trinkaus' new paper uses substantially the same methods and finds the same results as Caspari and Lee without giving them credit. I find only one difference in method -- Trinkaus cut off his age categories at 20 and 40 years instead of 15 and 30. And one difference in result -- Trinkaus found fewer older adults in the early Upper Paleolithic compared to Caspari and Lee. If either difference is important, there's no citation to let us know why. It's like Caspari and Lee's papers have slipped down the memory hole.

NAS members can submit papers to PNAS directly, relying on reviews from peers that they select themselves. The editorial policy of the journal makes it very difficult to reply to these papers, and certainly no reply could gain the attention that this paper has already received.

Lucky for me, I just happen to have a blog for such occasions.

I wrote to Trinkaus to invite him to provide his comments for me to publish them along with my post. I knew he would probably have a different interpretation than me of the issues. He very kindly took the time to compose replies to my questions. I publish them here unedited along with my follow-up questions.

His comment on my initial request:

Thanks for this. I intentionally did not refer to those papers since they, at least in the original PNAS paper, completely ignored taphonomical and behavioral issues. My original 1995 paper, which I seem to remember they did not cite (I am in Spain on a slow internet connection), heavily emphasized those issues, and this paper does as well. The issue is not just longevity - it is its combination with all of the various biases in the samples, biases which are just as important as the presumed demographic ones.
Best, Erik

My first followup:

Thanks, Erik, I really appreciate your reply. I hope Spain is treating you better than Wisconsin is treating me this week!
Naturally I disagree but I am very glad to be able to include your comment.
I did check the citations of all the papers, thinking that you might have missed them for that reason. They did cite you. The 2004 PNAS paper discusses taphonomy extensively, including the quantitative comparison of burials versus non-burials. The exchange with Hawkes and O'Connell also discusses taphonomic issues. The behavioral issues are the subject of the 2006 paper, which was titled "Is human longevity the consequence of cultural change or modern biology?"

Trinkaus' reply:

Two comments.
I do not make it a point to quote everybody with whom I might disagree.
Your blog on the Zhirendong mandible could have benefitted from a reading of Weidenreich, Dobson and Trinkaus and/or Schwartz and Tattersall on what constitutes a modern human chin. Saint Cesaire and the Vindija mandibles do not have it, any more than ER 730 does, despite incipient trigones on them.

I wrote back:

Hi, Erik --
Thank you again for your responses. They have helped me to understand your position.
If PNAS has the same review standards as my blog, maybe I should just concede. It would take fifty undergraduates to find all my errors.
But I think we completely agree that a scholarly mandibular description should cite the sources you mention, both classic and recent.
What I don't understand is why you disagree in the present case. If you had ignored Rachel and Sang-Hee's papers because you thought they had nothing of value, why did you use such similar methods and come to such similar conclusions? Did you think those methods and conclusions were so obvious that they don't need citation? If so, then why did you issue a press release?

Trinkaus replied:

I originally found the Caspari and Lee PNAS paper unconvincing, and put it out of my mind. None of my peer-reviewers noted its absence, and it was not cited in recent papers relating to the topic (i.e., Smith et al. 2010). It did not occur to me to cite it, and nor did it occur to other people directly and actively involved in Neandertal/modern human life history. Hence it was not there.
I did not previously comment on the your blog on the Zhirendong mandible, since I almost never respond to such commentaries. I only did it since you raised issues relating to your blog. That paper appears to have set a record for misquotes (Dennell could not possibly have read it for his Nature commentary).
Best, Erik

I finished the exchange:

Hi, Erik --
Well, thank you again for taking the time to comment. I do appreciate it, particularly since you're out of the country.

I think our exchange was much more productive than a formal comment could be. Trinkaus wrote that his omission had been intentional, and I take him at his word that he "put it out of his mind". I am glad that I was able to bring his attention to the problem, though I surely wish that a better review had been done in the first place. I'm sure he still disagrees but I hope he will take the opportunity to engage with the current literature. Maybe someone can suggest some more studies to replicate.

I've pretty consistently criticized scientists who issue hyped-up press releases. They draw my attention. It is a frequent feature of press releases that they claim unmerited novelty and ignore prior work. This feature rarely creeps into the actual published paper in such an obvious way.

I'm also disturbed by the power imbalance this case demonstrates. Sitting here watching MythBusters with my daughter who wants to be a scientist someday, I hope we can start to do better.