Climate Audit

As a sort of Methadone replacement for Mann’s bristlecone PC1 (which uses Graybill’s strip bark chronologies), two Esper foxtail chronologies (Upper Wright Lakes and Boreal Plateau) have come into use by the Team. These two chronologies were used separately in the small Esper network (two of only 7 or so medieval chronologies) making a very distinct impact on the medieval-modern relationship. Subsequently, they were used as an average by Osborn and Briffa 2006 (which also used Mann’s PC1), thereby keeping the bristlecone-foxtail representation on the council at 2 of 12. They were also used as an average in Hegerl et al 2006, which , like Osborn and Briffa, also used Mann’s PC1, keeping the representation of these tees on the council at 2 (of 12, fewer in the MWP). They were used individually in the Juckes et al 2007 Union reconstruction, once again keeping representation on the council at 2 (of 12).

Graumlich herself did not publish a temperature reconstruction from these sites. The Team typically cites Lloyd and Graumlich 1997 as authority, but it doesn’t actually justify the use of these chronologies as a temperature proxy and doesn’t even present the chronologies now used. Lloyd and Graumlich 1997 discuss treeline changes, observing that the medieval treeline was higher than at present. Graumlich 1991 – which IPCC reviewers cited in their “justification” of continuing to use Mann’s PC1 – used nearby (but different) chronologies; she criticized the Lamarche-Graybill CO2 fertilization hypothesis observing that her chronologies did not show the 20th century growth pulse of the Graybill chronologies. After a number of years of effort on my part, some Graumlich foxtail measurement data was archived in May 2007 – a small bit of progress which I noted at the time, The archived information includes the measurements used by Esper (and subsequent Team articles relying on the Esper versions) , but not the earlier information used in Graumlich 1991 (where 20th century growth was said not to be anomalous).

After noticing the remarkable difference in core widths within one strip bark tree in Tree 31 from our drilling at Almagre, I examined the data from Upper Wright Lakes (and Boreal Plateau) and noticed similar divergence between individual cores in some trees where two cores were archived. I sent an email to Andrea Lloyd observing the similarity in pattern in some Upper Wright trees to strip bark Almagre trees and inquired whether the trees at Upper Wright Lakes with similar patterns were also strip barked. I got a very cordial reply in which she said that the hypothesis was “intriguing!”, that she recalled that some trees had strip bark and undertook to examine her notebooks. She has now sent some very interesting fresh information from her notebooks. Pay particular attention to UFA003 and UFA004 below.

On a personal note, and it’s funny how small the world is sometimes, the daughter of one of my best friends is at Middlebury and took classes from Andrea Lloyd as a freshman. She plays hockey for Middlebury BTW. The daughter had previously mentioned to Lloyd that a family friend was involved in hockey stick matters prompting a discussion, so I probably got a little more sympathetic hearing than I’ve received at Tucson. A Middlebury sophomore assisted with our day two coring at Almagre as well.

My original inquiry was as follows:

Dear Andrea,

if you have a few moments, I’d appreciate any thought that you could give to the plots below.

here is a plot of two cores from the same BCP from our sample this summer. This tree had been strip barked and was quite elliptical. Black is the core from the S opposite the bare trunk; red is core from the SW. The horizontal dashed line shows the average ring width over all our samples to date. It seems evident that the asymmetrical impact of bark stripping/increasing ellipticalness has resulted in a huge difference between the samples. BTW recent widths appear to be rather narrow in the few samples that I’ve got back so far.

In the Upper Wright Lakes data, there are not a lot of cores that go up to data that have duplicates. I’ve noticed the same sort of radial asymmetry in a quick look in trees UWL118 and UWL123 as shown below. Do you recall anything about these trees?

Tree UFA003 (which has a very different history than tree UFA004) has the type of growth pulse that seems to be associated with strip bark in the Almagre trees. Do you recall anything about it?

Thanks, Steve McIntyre

As noted above, Lloyd responded cordially and promptly to these inquiries. She said that she herself had used the tree-ring data from her sites for “purely ecological purposes (cross-dating dead wood for reconstructing treeline), and never did any climate reconstructions on them”. She described them as “challenging chronologies from the perspective of climate reconstructions, as they have a highly confounded response to temperature and precipitation, but they’re well-replicated and long”.

She thought that the above hypothesis was “intriguing” and observed in her first response that “there were definitely some strip-barked trees up there”.

In passing, the nomenclature for sites in Lloyd and Graumlich 1997 is inconsistent with the archived information and Lloyd provided a concordance, agreeing with the need for such concordance information. She said that Boreal Plateau = Upper Treeline 1 and Upper Wright Lakes = Upper Treeline 3, East Slope, and Meadow. Graumlich did not archive one site at Cirque Peak (= Upper Treeline 2) on the basis that that she didn’t do any climate reconstructions on that one. However, Lloyd agreed that the data should still be archived and undertook to locate and archive the data. Lloyd had no information on the status of data from some of Graumlich’s earlier sites (e.g. Tyndall), that were reported in Graumlich 1991.

She was able to find notes on all of the UF trees, the EF trees, and the TT or TN trees (all of which are encompassed in the Upper Wright Lakes file on the ITRDB), but couldn’t find her notes on the UWL trees, as they were trees that were cored outside of her permanent study plots, so they are not included in those notes. She undertook to look for the UWL notes in a few other locations. However, the two UFA trees, UFA003 and UFA004, stand in pretty stark contrast. Here are her comments:

Here’s what I can tell you, direct from my field notes. The trees were fairly close to one another, but were among the few highest elevation live trees— so they’re growing right at the limit of trees in that area. (Elev ~3500’, although I wouldn’t navigate based on that— I was using an old and cranky altimeter that summer.)

UFA003: Has a second trunk on the uphill side, which was dead and heavily charred. The trunk that we cored was missing a big strip of bark on the N (uphill) side. It has many branches, and although the crown is bent (i.e., flagged), it was dense & healthy. The tree was probably very old (DBH = 99.5 cm) but was very rotten in the center, thus we were unable to get a particularly long core. No record of where the core(s) were taken from, but when a tree was strip-barked like that, I generally took a core from the opposite side. (Often I’d take one from the dead cambium as well, but I apparently did not in this case.)

UFA004: Had some scarring on the west side of the tree. A bit smaller than UFA 003 (dbh=62.8 cm), but more solid. Most branches were on the S/SE side of the tree. Crown was healthy. Cores were taken not quite opposite each other. (core A at an azimuth of 204 deg, core b at an azimuth of 94 deg).

If we compare these field notes back to the plots shown above, UFA003 – the strip bark tree – was the one where 20th century growth reached 3-sigma levels in the 20th century; while UFA004, the healthy one, had improved 20th century growth, but not the “lens-y” growth that we observed in the strip bark Tree 31 at Almagre and, seemingly also at UFA003, also strip barked. My surmise had been that the growth at UFA003 had been lens-y and associated with strip bark – so this was very nice support for the hypothesis from “out of sample” data. In effect, I’d been able to predict which tree was strip bark and which was whole bakr from their measurement history.

There was one loose end in the core information, which gave me pause when I initially compared the plots to the notes. In her notes, Lloyd mentions two cores at UFA004 and one core at UFA003, whereas my plots showed one archived core at UFA004 and two cores at UFA003. Arrgh. However the measurement data had
what may be a surprising reconciliation of the discrepant information. The archived UFA004 core is UFA004C, suggesting that there were three cores at UFA004, of which only one was archived; while the archived UFA003 cores are UFA003B and UFA003C, suggesting that this also had three cores. Perhaps the A cores had problems with rot – rot being a problem seldom discussed in the literature, but far more prevalent at Almagre than we’d expected.

So the very strong 20th century uptick in Upper Wright Lakes UFA003 is associated with a strip bark tree – a type that the NAS panel said should be “avoided” for reasons that have become increasingly clear as we explore these strip bark trees. On the basis that UFA003 is strip barked, my guess is that other nearby trees with similar patterns may also be strip barked. Candidates include the following:

Update: Since this morning, I inquired about the above trees and got the following responses:

UFA001: “Circumferential bark. Crown full & dense, but limbs reach towards the east”. DBH =59.5 cm
UFA005: “1/3 dieback on S/SW side”. DBH=80.3 cm
UFA006: “Large live tree- highest in site. Crown dead, south branches dead. Bark in strip, ~75% missing.” DBH=126.3 cm
UFA007:”50% bark. Crown fairly healthy; better than UFA006. Has cones, lots of live branches.” DBH=80 cm.
UFA008: I wrote nothing down other than that its dbh was 106 cm; I suspect that means that it was a normal (i.e., not strip-barked) tree, but I wouldn’t necessarily bank on that.
EFA003: I have excellent notes for both EFA001 and EFA002, but apparently stopped writing things down at tree 3. I can see it on my map (it’s right next to EFA002), but I recorded nothing. I can tell you that EFA001 had full circumferential bark (and I put an exclamation point after that fact, which may indicate that it was an oddity), and EFA002 had <25% living bark. But that doesn’t help much with 003. Alas.

Here are plots for the trees mentioned in the reply that I didn’t post graphics up for this a.m.

So, just in case there was any doubt, the Upper Wright Lakes foxtail site has many strip barked trees – and virtually all of the trees with very loud (2-3 sigma) 20th century growth, for which we’ve been able to assess barking, have been strip barked.