BTW, I thought that the Chronicle colloquy was pretty interesting. I don’t usually get to ask people questions directly, so I appreciated that the Chronicle allowed questions through without Gavin Schmidt/realclimate censoring.
Now that I think of it, the only other question that I’ve been able to ask a climate scientist directly was to Caspar Ammann at AGU last year – what was the verification r2 of Mann’s reconstruction? Ammann filibustered about why that was an irrelevant question, but didn’t answer it.
I asked North about bristlecones (and also about error bars), but here I’ll just talk about his answer to bristlecones.
In the sidebar to Richard Monastersky’s article, he said:
For the earliest part of the 1999 analysis, Mr. Mann’s group relied heavily on bristlecone pines from western North America. The original study noted that there were some difficulties in using such trees because of peculiarities in their recent growth, but Mr. Mann and his group attempted to quantify those problems and to work around them. The National Research Council suggested that researchers avoid using trees that are the most difficult to interpret. More-recent studies have avoided those trees and reached similar conclusions.
I emailed Monastersky about that comment saying:
The most recent studies – Osborn and Briffa 2006 and Hegerl et al 2006 – both use bristlecones/foxtails and both even use Mann’s PC1. In fact, they use separate series for bristlecones and foxtails – making 2 series of this type (out of only 12-14 total series).
He replied promptly:
With reference to the NRC report, its recommendations about the use of bristlecone pines, and recent reconstructions, I would refer you to Franco Biondi, a dendrochronologist, who participated in producing the report.
Well, there are no dendrochronological issues as to whether bristlecones were used in Osborn and Briffa 2006 – it’s a matter of fact. You could turn this question over to Price Waterhouse and get an answer.
Anyway I asked North about this in the Chronicle colloquy in a different form. Here I asked not about the properties of the bristlecones, but simply whether the panel had performed any due diligence on whether bristlecones were used in other studies (having determined that they should not be used.) Again this is not a dendrochronological question, but an auditing question.
Question from Stephen McIntyre:
The NRC Panel stated that strip-bark tree forms, such as found in bristlecones and foxtails, should be avoided in temperature reconstructions and that these proxies were used by Mann et al. Did the Panel carry out any due diligence to determine whether these proxies were used in any of the other studies illustrated in the NRC spaghetti graph?
North’s answer was as follows:
There was much discussion of this matter during our deliberations. We did not dissect each and every study in the report to see which trees were used. The tree ring people are well aware of the problem you bring up. I feel certain that the most recent studies by Cook, d’arrigo and others do take this into account. The strip-bark forms in the bristlecones do seem to be influenced by the recent rise in CO2 and are therefore not suitable for use in the reconstructions over the last 150 years. One reason we place much more reliance on our conclusions about the last 400 years is that we have several other proxies besides tree rings in this period.
It’s a typical academic answer – somewhat related to the question, but quickly going up some other trail. I didn’t ask him about CO2 fertilization; I didn’t ask him about the impact of bristlecones in the determination of uncertainties. I asked him whether the panel, having determined that bristlecones should not be used, then carried out any due diligence on any other studies to determine whether they used bristlecones – something that any engineer would have done if he determined that certain materials should not be used in a design.
Anyway, the salient part of the answer appears to be: We did not dissect each and every study in the report to see which trees were used. The next question would obviously be: Did you dissect any studies to see whether they used bristlecones, and, if so, which studies did you dissect? Of course, we know the answer to it. They didn’t dissect any studies. Getting into the data is apparently an activity fit only for "amateurs"; by contrast, the "professionals" "just sort of winged it".
North "felt certain" that the latest studies by "Cook, D’Arrigo and others" do take this "into account" – whatever that means. If "others" includes the studies actually illustrated in their study – and which North relied on in his presentation as an overlay comparison to MBH – then those other studies – Mann and Jones 2003; Esper et al 2002; Moberg et al 2005 and Hegerl et al 2006 do use bristlecones/foxtails and, in 3 of 4 cases, more than once in very small networks. Whlie a reader might think that it was outside the scope of the NAS panel to "dissect each and every study", a reasonable reader would expect them to dissect the studies that they illustrated and relied on to see if they used bristlecones. Obviously they didn’t.
What makes it worse is that, surprise surprise, they had been alerted to this very issue in our written presentation. We stated:
On an overall basis for large populations (387 sites) of temperature-sensitive tree ring series (both width and density), ring widths and densities have been declining in the latter half of the 20th century [Briffa et al, 1998; Briffa 2000]. The canonical multiproxy studies all use very small (<20, usually <10, series). Unlike the large population series, the small-sample averages tend to “support”‘? MBH98-99 through a hockey stick shape, at least suggesting the possibility of biased selection. In particular, we note that bristlecones (or inter-related foxtails), known to be a problematic proxy, but having a distinctive hockey stick shape, are repetitively selected into these small samples either directly or though Mann’s even more accentuated PC1, thus affecting, not only MBH99, but Crowley and Lowery  (two series), Esper et al.  (two series), Mann and Jones , Jones and Mann  and Osborn and Briffa  (two series).
The underlying lack of robustness of a typical such study is illustrated below where the series of Crowley and Lowery  are re-calculated without two problematic bristlecone series and the equally problematic Dunde àŽⳏ18 series. If these proxies are problematic and reflect biased selection, then conclusions as to the relative medieval-modern levels are not at all robust. Similar results apply in varying degrees for other multiproxy studies.
Or again in our Conclusions:
Even the principal components methodology, which has been vociferously criticized, has continued in use in prominent studies: Mann and Jones , Jones and Mann . One of the networks in Rutherford et al.  (coauthored by the MBH authors) was identical to the MBH98 network, including the identical PC series. Only a few weeks ago, Osborn and Briffa  used the North American PC1 as one of only 14 proxies in a Science article.
Given the particular attention of our articles to the problems with bristlecones as a temperature proxy, “moving on”‘? from MBH would clearly require renunciation of bristlecones “€œ something that has obviously not happened.
So it’s not as though they weren’t told about the use of bristlecones in other studies. So they had notice of the problem, but ignored the specific notice. I guess that’s what happens when you "read a lot of papers" and "wing" it.
By contrast, the Wegman report specifically drew attention to the repetitive use of bristlecones in other studies and, in his testimony, Wegman stated that this issue needed to be specifically examined.