So far, I’ve received preliminary measurement data on 8 trees – all trees with Graybill tags. Today I’m providing some notes based on a first look at this data. As noted before, the archive identifications do not match the Graybill tags. Pete obtained some information which has enabled us to make a limited concordance.
We have learned that Graybill tagged 41 trees at Almagre (tags 84-37 through 84-77). Pete and Leslie located 17 of the 41 trees (before the lightning storm cut their trip short.) Pete may locate some more today or it may have to wait until next spring. Here’s something puzzling: of the 41 tagged trees, only 16 trees were archived (the Graybill 1984 collection are in the id range ALM02 to ALM30; measurements made by Lamarche in 1968 are in the ALM41 and higher range).
Of the 17 tagged trees that Pete located, 8 were drilled. Of these 8, only 3 were archived. However, as shown below, our results reconcile well on these 3 cross-checks.
Of the 41 Graybill tagged trees, we have learned that there is cross-dating information on 38 of them. So the non-archiving of these 22 trees isn’t because cores were lost and cross-dating did not take place. Of the three tagged trees which lack cross-dating information in Tucson, all three have been located and two were drilled (the three missing are 37, 50=PH28, 51=PH29).
Please refer to the tree gallery for pictures. And BTW none of the trees are showing record positive growth through teleconnnections to recent warn 1990s and 2000s – actually recent growth in these 8 trees (more will come soon) is at historically rather low levels.
Three Reconciled Trees
In a recent post, , we showed a cross-matched Graybill photo with one of Pete’s photos, showing that the Graybill photo was of his 84-55, our #30.) We have now determined that this tree reconciles to the archived ALM16. Here is a plot showing our measurements (red, black) versus Graybill’s (blue). This is an old tree. As you can see, one does not observe the explosive recent growth that characterizes some strip bark trees. There is some difference between Graybill and our values in the 15th century, but my guess is that this results only from elliptical-ness in the 15th century. Recent values have been declining.
The next tree cross-matched to archived data is our Tree #33 (Graybill 84-43), which is archived as ALM06. Again here is a plot showing our measurements in black-red and Graybill’s in blue. Once again, there is a good match between our measurements and Graybill’s (although our core was not as long as Graybill’s.) In this case there are no large discrepancies between the two cores – something that is not always the case, and which seems like a useful check on ellipticity.
In this case, as shown in the photo below, although the tree is strip-barked, the morphology doesn’t look complex and the “good” half of the tree has a relatively healthy appearance.
The third reconciled tree is our Tree #47 (Graybill 84-52; ALM13), which Pete observed to be relatively whole-barked. Here is a plot again showing our measurements (red,black) versus Graybill (blue). In this case, our cores were longer than Graybill’s, but again there is coherence in the overlap. In this case, one could plausibly imagine a negative exponential age effect (that is not plausibly detected in other series.)
In this case, as shown in the photo, this is a relatively whole bark tree.
Three Cross-dated Unarchived Trees
Next here are three trees that we cored, which have cross-dating information at Tucson, but which are unarchived.
The first is our Tree #1 (Graybill 84-56), which we’ve already discussed as an example of strip bark. Here’s a plot in slightly different format (left panel – smoothed with an 11year gaussian filter; right panel – 20th century.) In this case, ring widths peaked in 1915 and there has been a continuing decline through the 20th century right to the present. A picture of this tree is in the earlier post here. The discrepancy between cores appears to be entirely due to strip bark ellipticity.
The next cross-dated and unarchived tree is our Tree 37 (Graybill 84-41), which has almost the opposite history to the one above, as shown below. In this case, the tree experienced explosive growth in the 19th century reaching 3-sigma ring widths. This seems very characteristic of a strip bark pulse. In this case unfortunately we only have one core. Pete’s notes report that this tree was strip barked and was hollow.
The third tree in this class is our Tree 48 (Graybill 84-54.) The picture of this tree shows that it is strip barked. The plot below shows a noticeable discrepancy in the 19th century history between cores.
Two Other Tagged Trees
As noted above, there are three tagged trees for which our reconciliation shows no crossdated match, two of which we drilled.
The first is our Tree 28 (Graybill 84-50.) In this case, we have 3 cores shown below in black, red and green. Here the strip barking seems to have introduced a pretty complicated pattern: notice that the green core is lowest in the 17th century, highest in the 19th century and in the middle in the later 20th century, completing the permutation of cores. So it’s not that one aspect of the core is always the widest – but places can change. In the mid-20th century, there was a strong increase in the red core, accompanied by an equally strong decrease in the green core. Arrrgh.
The other such tree is our tree 29 (Graybill 84-51.) This history is relatively short, and had maximum values in the early part of the 20th century.
Some Thoughts on Sampling Strip Bark Trees
Based on our short introduction to strip bark sampling, one immediate conclusion is that, if one is going to bother sampling these trees at all, it is absolutely necessary to have multiple cores. The discrepancies between individual cores can be enormous – far more than any conceivable climate signal. The discrepancy may give some information on the timing of the strip bark occurrence. It’s tempting to think that an incipient discrepancy, such as in Tree #31, may give information on the timing of strip bark occurrence. But whether it does or not, it seems like the specialists should at least commence coming to grips with the impact of this phenomenon before including this information in samples.
It also seems clear that the “error” introduced by strip bark is not random in measured data. In the samples that we’ve seen (which are inevitably from the strip bark itself, the strip bark seems to introduce a strong positive pulse. There are no corresponding negative pulses to cancel out. So if strip bark pulses are not randomly distributed through a record, but concentrated in the later part of the record, you could easily get a “strip bark” bias. How many such trees would it take to affect results – one would need to do some calculations, but given the extreme size of the bias in some cases (3-4 sigmas), it wouldn’t take very many in the relatively small Graybill networks to distort results. As an objective statistical criterion for bristlecone and foxtail sites, it seems to me that any usable tree should have visual confirmation that it is not strip barked and that you have 2 or more cores that need to pass some sort of standard for non-ellipticity.
In addition, the failure of the Tucson dendroclimatologists to archive all of Graybill’s data is disquieting,particularly when there has been no previous information on the incompleteness of the archiving or the reasons for the incompleteness. The very incomplete nature of the Graybill archive at Almagre (only 16 out of 47 tagged trees) makes one wonder about what happened at other sites (Sheep Mountain, Campito Mountain). Graybill and Idso 1993 reported that they focused on strip bark, but this does not say anything about selective archiving:
These so-called’ strip-bark’ trees can have active cambium that is only a few centimeters in width. Foliage and cones are also accordingly limited. Trees of this nature were the primary focus of investigation whenever possible. They were most commonly found and sampled in stands of bristlecone and limber pine in the Great Basin and in stands of foxtail pine in the Sierra Nevada.
Did Graybill share Esper’s view that results can be withheld if the purpose is to “enhance” a signal? In the Almagre case, it’s not that I’m suggesting that the inclusion of the unarchived results would yield a “right” answer relative to what Graybill has reported. For the purposes of calculating a chronology, the NAS Panel recommended that strip bark trees be “avoided” and that seems like the only prudent course of action. But for the purposes of research, surely there is no excuse for selective archiving. Yeah, tree #31 is an extreme case and I’d agree that it shouldn’t be included in a chronology. But to give an idea of what strip barking can do, it’s sure informative. And by withholding this sort of information, perhaps readers become less attentive to the possibility of less extreme examples of the same effect. Regardless, all the results should be archived.
And, oh yes, none of the trees are showing record positive growth through teleconnections to the warm global 1990s and 2000s.