Tuesday, April 10, 2012


date: Wed, 07 Aug 2002 09:24:31 -0500
from: David Stahle <dstahleatXYZxyzk.edu>
subject: Holocene reviews
to: k.briffaatXYZxyz.ac.uk

Dear Keith:

Hope you are well and that I get to see you again soon.
Sorry for the delay on these reviews, but finally I have an opinion.
The Speer et al. paper is weak and unsuited for publication. The
Hunzicker and Camill paper is more problematic. Pretty strong
analyses, but the calibration and verification are awful, and I'm
afraid that can probably be traced back to the half ass chronology
development procedures. I wouldn't publish that one either, unless
they go all the way back and fix the chronology development, and then
substantially improve the calibration and verification. The reviews
are attached and reproduced below.

Sincerely, Dave

Review of J.H. Speer et al.: "Assessing the dendrochronological
potential of Pinus occidentalis Swartz in the cordillera Central of
the Dominican Republic" submitted to the Holocene.

This is an esoteric and highly speculative paper on annual
ring formation in West Indian pine, that draws conclusions only
marginally supported by the analysis. The evidence for annual rings
and "crossdating" is weak. They only report the "average interseries
correlation" in Table 3, which was just 0.44, and it is not entirely
clear how they arrive at this number (18 radii are mentioned on p.
12, ten trees are listed in Table 3). They do not show a plot with
all ring width time series to allow readers to judge for themsleves.
They also cite a "statistically significant correlation" of only 0.34
with annual precipitation at Jarabacoa as evidence that the
chronology does produce annual rings. This is a very weak
relationship, and alone hardly sufficient to prove the annual nature
of the growth rings. They do not display a time series comparison
between the tree-ring and climate data to allow the reader to judge.
The other three study sites did not crossdate and did not relate to
climate. So the evidence for annual rings and a useful climate
record is hardly sufficient to state (as they do in the abstract)
that: "P. occidentalis may allow dendrochronological reconstruction
of climate, fire history, and other environmental processes that
operate on the island of Hispanola." The paper is heavily weighted
toward a disription of ring types and ring quality, and six of the
nine figures are photos of rings. The paper might be better suited
for a journal of wood anatomy.

Review of David Hunzicker and Phil Camill: "Using a new 672-year
tree-ring drought reconstruction from westcentral Montana�" submitted
to the Holocene.

This is a well written, well executed paper that I would
unfortunately not recommend for publication in the Holocene. It's a
shame to read a paper like this. It is very well informed, well
referenced, places the work in a good scientific context, and
includes strong statistical analyses. However, the attention paid to
the analyses and interpretation of the reconstruction was evidently
not paid so carefully to the fundamental tree-ring chronology
development. They call it "crossdating," but the best I can tell
from the limited discussion it was simply computerized correlation
matching of measured time series, with a massive culling of the data
to pare down to those time series that produced straightforward
correlations in a COFECHA analysis. I was astounded to read that
their final chronology used only 61 out of the 152 trees sampled for
the study. The 60% of the trees not included apparently suffered
from "complacency, unresolvable sections of missing rings, or low
interseries correlation values." This appears to be the first
penalty for not applying rigorous dendrochronological methods to the
chronology development. I find it incredible that over half of the
Ponderosa pine samples would not be useful. I can't help but suspect
that by relying on COFECHA output, without any hard-nosed microscope
work and rigorous crossdating with the wood samples themselves, you
at best default to the simple, straightforward trees without missing
rings. That is, you default to a less climatically sensitive subset
of trees. This appears to be the second penalty for the seemingly
inexpert, quick and dirty chronology development.
These authors have obviously worked hard on this study and
bring excellent analytical skills and knowledge of the literature.
The paper itself is exceptionally well written (with a minor
complaint concerning the over use, and at times incorrect use of the
term "teleconnection"). But the calibration and validation reported
in the paper are clearly awful, and that surely ought not be the case
for Ponderosa pine on moisture-stressed sites in Montana. One hates
to be non-supportive of their work, so much of which is high quality,
but it seems to come down to fundamentals, and here the fundamental
dendrochronology and chronology development are in question. And I
also do not think it advisable to publish a reconstruction that
explains maybe 21% of the variance in the instrumental climate data,
when using an arid site conifer as the predictor (the persistence in
the standard chronology may be inflating even that figure). I just
can't believe the calibration could be so weak. It seems they need
to revisit their chronology development work, and dig deeper into the
climate response of their chronology. Then look very carefully at
climate data itself. These climate data are not guaranteed to be
homogeneous, especially in the mountain West during the early 20th
century. If all this could be done, and if the variance explained in
both the calibration and verification periods could be improved, then
publication in the Holocene would be well justified.


Attachment Converted: "c:\eudora\attach\Speer_Holocene8.02.doc"

Attachment Converted: "c:\eudora\attach\Hunzicker_Camill8.02.doc"
David W. Stahle
Dept. of Geosciences
Ozark Hall 113
University of Arkansas
Fayetteville, AR 72701 USA
479-575-3469 (FAX)


No comments:

Post a Comment