Millennial Quebec Tree Rings

In today’s post, I’m going to discuss an important new 1000-year chronology from northern treeline spruce in Quebec (Gennaretti et al 2014, PNAS here).  The chronology is interesting on multiple counts.  This is the first Quebec northern treeline chronology to include the medieval warm period.  Second, it provides a long overdue crosscheck against the Jacoby-D’Arrigo chronologies (including Gaspe) that have been embedded in a number of canonical reconstructions.  Its results are very different.  Third, the Quebec (and Labrador) northern treeline is the closest treeline to the Baffin Island ice core and varve thickness series.  I’ve observed on several occasions that interpretation of Baffin Island varve thickness series (Big Round Lake) is presently inconsistent with the interpretation of the similar Hvitarvatn series in Iceland and, in my opinion, there are serious questions about whether PAGES2K has oriented this series correctly.

Gennaretti et al Reconstruction/Chronology

The Gennaretti temperature reconstruction is based on a very large collection of living and subfossil black spruce from the latitudinal treeline. It extends back through the MWP to 918AD and continues forward to 2011.   Their reconstruction shows a very elevated MWP, with the most extreme cold occurring in the early 19th century.  It shows a relatively cool 20th century with a sharp increase in the 21st century.  The Gennaretti reconstruction is linear in the underlying tree ring chronology.  In the figure below, I’ve compared it to (arguably) the two most influential previous tree ring chronologies from Quebec –  two versions of the Gaspe chronology: as used in Jacoby and D’Arrigo (1989), Mann et al 1998 and Mann et al 2008 (red); and a variation used in D’Arrigo et al 2006 (green). Both chronologies showed much reduced growth in the early 19th century (with Gennaretti being even more intense). But the Gaspe chronology (in both variations) was at hockeystick peak in 1980 (dotted red line),whereas the Gennaretti chronology was still below its long-term mean in 1980, only recovering above its long-term mean in the 21st century (see dotted black line at 2000).

comparison_to_gaspeFigure 2. Quebec Tree Ring Chronologies.  All scaled to SD Units. Black – Gennaretti et al 2014 black spruce from northern treeline. Red- Gaspe chronology used in Jacoby and D’Arrigo 1989 NH reconstruction and in Mann et al 1998. Green – Gaspe chronology incorporating additional measurements from the mid-1980s as used in D’Arrigo et al 2006.

 

Comparison to Varve Series

Gennaretti et al compared their results to recent analyses by Gifford Miller, the lead author of the Hvitarvatn varve thickness series that has been much discussed at CA recently, a discussion that presumably led to PAGES2K’s remarkable decision to invert the orientation of the series in their reconstruction. Miller has also done important work on Baffin Island, the interpretation of which (especially the Big Round Lake varve thickness series) has become a bit of a battleground issue.  Gennaretti et al also compare their interpretation to some other Quebec data (e.g.  ice wedges on Hudson Strait) that I am not familiar with, but plan to examine.

Miller had interpreted Baffin Island to have had a medieval warm period to about 1250AD (“ice melt” in diagram below), with a subsequent period of relative cold (“continuous ice cover” in the diagram below) commencing about 1450 and continuing into the 20th century. Gennaretti’s reconstructed medieval warmth neatly matched Miller’s period of “ice melt” on Baffin Island.

strec_figure 2AFigure2. Gennaretti et al 2014. Northern Quebec tree ring temperature reconstruction (blue), compared to Gifford Miller’s Baffin Island interpretation.

The next graphic compares the Gennaretti reconstruction from Quebec (top panel) to Miller’s Hvitarvatn, Iceland (inverted varve thickness – third panel and  IRD (ice-rafted debris) bottom panel)  and the Big Round Lake, Baffin Island varve thickness series (also inverted here and thus opposite to orientation of PAGES2K and numerous multiproxy studies.

quebec_vs_varves

Figure 3. Top –  Quebec northern treeline temperature reconstruction; middle – Big Round Lake varve thickness (logarithm), shown in inverted orientation – opposite to the orientation used in Kaufman et al 2009, PAGES2K, Ljungqvist 2010 and numerous other studies; bottom – Hvitarvatn, Iceland varve thickness, shown in inverted orientation, following interpretation of Miller et al 2012, advocated at Climate Audit here and finally applied in PAGES2K -2014 revision.

Looking first at the 19th century, Miller interpreted the association of thick 19th century varves together with IRD (ice-rafted debris) levels not seen since the LGM (Last Glacial Maximum) and thus interpreted thick varves as evidence of glacier advance and cold, rather than warmth, whereas thick varves had been interpreted in Kaufman et al 2009 and PAGES2K as evidence of warmth. Miller interpreted the thin medieval Hvitarvatn varves as evidence of glacier recession and warmth.  Because thick varves require both that a glacier be nearby and that it be retreating, it appears that thin varves can be evidence of both warmth (glacier absence) and extreme cold and thus cannot be simply dumped into a multiproxy black box.

At Big Round Lake, Baffin Island, the period of above-average thick varves exactly match the periods of above-average thick varves at Hvitarvatn, Iceland.  It also corresponds to the period designated by Miller (on other evidence) as cold (as shown in Figure 2 here).  Interpretation of thick Baffin Island varves as evidence of warmth (as done in Kaufman et al 2009 and PAGES2K) requires either Miller’s interpretations at Iceland and Baffin Island be overturned or that some physical mechanism be provided for climate in Iceland and Baffin Island proceeding in symmetrically opposite directions over the course of the last millennium – a physics issue completely ignored by the PAGES2K authors.

The Quebec ring-width chronology shows lowest growth in the periods of relatively thick varves.  Its correlation to Hvitarvatn varve thickness (inverted) is 0.33  – a value that is (in my experience) unusually high between unlike proxies.  When one looks at the medieval period, one gets the impression that varve thickness series are “damped” in its upper register – logical enough since varve thickness cannot be less than zero.  When Hvitarvatn varves are compared over the Holocene to other measures (e.g. BSi), one gets a similar impression of “damping” in the upper register.R

Comparison to PAGES2k-2013 and Hanhijarvi

The next graphic overplots the PAGES2k-2013 Arctic reconstruction (blue – unrevised as used in AR5 – revised version not available digitally to date) and Hanhijarvi et al 2013 North Atlantic Arctic reconstruction (green) onto the Gennaretti reconstruction in Quebec.  The amplitudes are much greater. In addition, the 20th century levels in the Gennaretti reconstruction are substantially lower than medieval levels, even as compared to Hanhijarvi et al 2013.  To the extent that this expands the regional coverage of Hanhijarvi, it would not move it in the direction of PAGES2K-2013, but in the opposite direction.

quebec_vs_pages2k

 

Figure 4. Comparison of Gennaretti 2014 reconstruction (black) to PAGES2k-2013 (pre-revision in blue) and Hanhijarvi et al 2013 (green). 

Discussion

Gennaretti’s ring width chronology reinforces the impression from Hvitarvatn and other Gifford Miller analysis of the early 19th century as a period of extreme relative cold and of an actual medieval warm period. It further supports my previous argument that the Big Round Lake varve thickness series (used in numerous multiproxy studies, including PAGES2K) has been incorrectly oriented and needs to be inverted.    Re-examining the underlying article (Thomas et al 2009),  the original authors report that they “do not know whether greater changes in glacier position occurred from 1000 to 1900 AD that would have influenced the Big Round Lake varve record”.  Major changes in glacier position have been documented in Baffin Island during the Little Ice Age. Given that such changes are believed by Miller to be the driving force behind the very similar Hvitarvatn record,  it seems far more plausible that changes in Baffin Island glacier position have wrought similar effects at Big Round Lake than that such changes can be disregarded.

As Hodder et al. (2007) point out, factors other than temperature can influence varve thickness. Leonard (1997) demonstrated that changes in glacier activity and position relative to the lake influence varve thickness. The Big Round Lake varve record is probably not influenced by changes in glacier position, at least during the twentieth century. The maximum Holocene extent of the glacier was likely attained in the late nineteenth century (Briner et al. in press). The glacier remained at this maximum position until the mid-twentieth century (Fig. 1b) and has retreated *250 m since 1960 AD, a small distance compared to the size of the glacier (*2 km across) and the 5-km-long proglacial stream. We cannot determine pre-twentieth-century glacier position, however, and therefore do not know whether greater changes in glacier position occurred from 1000 to 1900 AD that would have influenced the Big Round Lake varve record. This adds uncertainty to our varve-inferred temperature reconstruction prior to the twentieth century. Furthermore, we cannot determine changes in subglacial sediment storage that would influence varve thickness. The sedimentation regime in Big Round Lake changed dramatically around 970 AD (58 cm depth in the long core). Neoglaciation initiated on Baffin Island *3 ka (Briner et al. in press), and the glacier likely was farther from the lake before this time. The distance between the glacier and Big Round Lake likely was changing around 970 AD, and may have caused the change in sedimentation at the lake.

We’ve already seen that inversion of Hvitarvatn had a major impact on the PAGES2K Arctic reconstruction. Inversion of Big Round Lake would likely have a similar impact.

The Gennaretti reconstruction also raises questions about the longstanding use of the Gaspe chronology without any apparent attempt at verification.  In McIntyre and McKitrick 2005,  we noted many issues about the validity of this series as a temperature proxy.  Gaspe is well south of the northern treeline, which had spruce chronologies rather than cedar. Prior to our 2005 studies, Ross and I met with two of the leading cedar specialists in Canada and were considerably influenced by their botanical view that cedar growth was best in cool, moist years. We also observed that other cedar chronologies, including a nearby cedar series in Maine, did not have a comparable shape.  We also learned that there had been additional measurement information from Gaspe, which Jacoby and D’Arrigo refused to provide.  We also sought exact information on where the Gaspe samples had been taken, with the information once again being refused.  The northern treeline reconstruction of Jacoby and D’Arrigo 1989 used only the 10 most “temperature-sensitive” sites of 35. Jacoby (recently deceased) refused to archive the information from the other 25 sites on the following grounds:

If we get a good climatic story from a chronology, we write a paper using it. That is our funded mission. It does not make sense to expend efforts on marginal or poor data and it is a waste of funding agency and taxpayer dollars. The rejected data are set aside and not archived.

At the NAS panel in 2006, D’Arrigo created quite a sensation when, as I reported at the time, she “put up a slide about “cherry picking” and then she explained to the panel that that’s what you have to do if you want to make cherry pie”.

The PAGES2K study used 13 tree ring chronologies: four Jacoby-D’Arrigo chronologies from North America and three Briffa chronologies from Asia (including the Yamal superstick of Briffa et al 2008). The HSI of these seven chronologies (1.56) is much higher than the HSI (0.56) of the other five authors. I, for one, would find their work more convincing if it were replicated by independent groups. That their Quebec were not replicated in the much more comprehensive study of Genarretti et al 2014 is hardly reassuring.


Postscript

Comparison to Wilson 2007 NQU

In comments below, Rob Wilson observed that Wilson et al (2007) contained a northern Quebec (NQU) chronology and suggested a comparison, provided below. Wilson’s NQU chronology is based on an old Fritts series (Fort Chimo) and still unarchived data for Pyramid and Lac Tesaliuk, both sites attributed to Payette (2007), though only the former site is mentioned in the article. Both chronologies are sort-of similar through from 1775-1915 or so, but the Wilson chronology has a much more pronounced pulse in the 20th century and more of a divergence problem.
gennaretti_vs_wilson

Examining Payette (2007), it appears that even nearby site chronologies can have very different patterns, depending on regeneration. It’s not evident to me that you can average such incommensurate series in any reasonable way. Payette 2007 has some interesting pictures showing subfossil trees above the present treeline, reminiscent of Polar Urals.

payette_2007_figure_8_pyramid

Comparison to D’Arrigo et al 2006 Labrador

Here is a similar comparison of the new Gennaretti chronology to the D’Arrigo et al 2006 Labrador RCS chronology. This is related to the Labrador STD chronology shown in their 2003 article, but isn’t the same. While there are points of similarity in the early 19th century, the D’Arrigo RCS version shows much weaker growth in the downticks around 1600 and 1700, and has a very large growth pulse in the mid-20th century that is not observed in the Gennaretti data.  One of the issues that concerns me about the D’Arrigo 2006 LAB chronology is that inhomogeneity between sites appears to be expressed in the chronology – a very difficult and perhaps insurmountable issue with RCS methods.
gennaretti_vs_darrigo-lab

Reference:

Gennaretti, F et al (2014. Volcano-induced regime shifts in millennial tree-ring chronologies from northeastern North America, PNAS. pdf NOAA data

Thomas et al 2009. Climate of the past millennium inferred from varved proglacial lake sediments on northeast Baffin Island, Arctic Canada pdf
Payette 2007. Contrasted dynamics of northern Labrador tree lines caused by climate change and migrational lag. pdf


42 Comments

  1. Posted Oct 13, 2014 at 1:59 PM | Permalink

    How do tree-ring (or lake sediment) chronologies generate full, annual data curves when there is a gap every winter? Is there enough winter growth (or sediment accumulation) to infer something about the seasonal temperature anomaly? Or is the data simply filled in by author-generated assumptions?

    • Craig Loehle
      Posted Oct 13, 2014 at 2:20 PM | Permalink

      Tree rings typically only record growing season temperatures. Usually this is spring/summer but in some climates growth occurs in winter if it is mild and wet then (like PNW USA).

      • Follow the Money
        Posted Oct 13, 2014 at 3:11 PM | Permalink

        “Tree rings typically only record growing season temperatures.”

        ? Outside of Australian Climate Science Ltd, I have never read anything arguing temperature anomalies are a detectable signal in tree rings except for possibly those from treeline specimens.

        • Craig Loehle
          Posted Oct 13, 2014 at 3:35 PM | Permalink

          You are right of course. Trees respond far more to rainfall/snow variation than to temperature. The dendros do try to get treeline data and assert that the rainfall effect is a random effect. However, studies have shown that major rainfall zones such as the ITCZ and monsoon in China can shift up and down for long periods. The Sahara was a grassland/shrubland 6000 years ago. So this assumption really is unproven if you go back more than a few hundred years (if that).

    • Steve McIntyre
      Posted Oct 13, 2014 at 2:35 PM | Permalink

      when specialists are being most careful with other specialists, they limit reconstructions to summer season and sometimes only a sub-season.

      To get to annual data, they assert that summer temperature fluctuations correlate to annual. This assumption is very not tenable on Holocene scales where there are major changes in summer insolation. Indeed, the changes in summer insolation over the past 2000 years are not negligible, a point made effectively in Esper et al 2012.

      • Posted Oct 13, 2014 at 3:31 PM | Permalink

        It seems that dendro graphs should have a flat line representing annual below-growth temperatures with data points shown only above the line (assuming it is presented rightside up!). Or perhaps they could use strongly lop-sided error bars that stretch to approximately infinity on the winter scale.

        It is truly a remarkable field of study if one can substitute so many questionable assumptions for actual data and still be published.

  2. Not Sure
    Posted Oct 13, 2014 at 2:04 PM | Permalink

    Does anyone else find the big dip in the middle of the MWP (around 1100 CE) interesting? Maybe the growth response to temperature becomes negative?

  3. Steven Mosher
    Posted Oct 13, 2014 at 2:09 PM | Permalink

    would there be a dip around 1813 and a peak around 1875ish?

    • Steve McIntyre
      Posted Oct 13, 2014 at 2:25 PM | Permalink

      yes. The sharp decline commences about 1809 and the peak is 1875-1880.

      19th_century

      • Steven Mosher
        Posted Oct 13, 2014 at 3:21 PM | Permalink

        I have to read the paper more closely.. CRUTS3 is an interesting choice

  4. tty
    Posted Oct 13, 2014 at 4:07 PM | Permalink

    It’s interesting that the two “low points” both coincide with “extinction events” in Greenland. The Norse colony was last mentioned in icelandic sources in 1410 and archeologists date the final demise to c. 1450. Inuits in East Greenland were last seen alive in 1824 and were extinct by 1873.

  5. Robert Way
    Posted Oct 13, 2014 at 4:31 PM | Permalink

    Very interesting paper. I work in the general area and am from the Labrador portion so it definitely fills an interesting knowledge gap. My MSc worked on glaciers in northern Labrador (Torngat Mountains) including estimating the timing of maximum LIA marginal positions. The D’Arrigo et al (2003) series that I use is very different from the D’Arrigo et al (2006) series you’re using with a much deeper cooling in the mid-1600s and a much smaller cooling in the late 1800s. I got that series as part of Kinnard et al’s 2011 release. My work on Torngat glaciers supports a maximum LIA marginal position being around ~1650 for a subset of glaciers which does not seemingly agreement with what one would expect from the series you’ve presented above. Undoubtedly regional variations play a role here but it might be worthwhile to mention the D’arrigo et al (2003) paper on Labrador tree rings.

    As an aside – tree line was higher than present in Labrador during the MWP and the holocene hypsithermal according to tree ring data was only ~3000-4000 years ago in the area because the region deglaciated very late. Undoubtedly these factors play a role.

    • Steve McIntyre
      Posted Oct 13, 2014 at 8:45 PM | Permalink

      Robert, I have some notes on Labrador tree ring series and will do a post. I used the Gaspe series as a comparandum because they’ve been used as “Quebec” series in multiproxy studies, while the Labrador series has been little used. In addition, Jacoby and D’Arrigo’s Labrador archive is incomplete. Plus, I think that it’s important to see what people other than Jacoby-D’Arrigo and Briffa come up with.

  6. Robert Way
    Posted Oct 13, 2014 at 4:36 PM | Permalink

    I should also mention that as we discussed in our recent paper (Way and Viau, 2014) linked below the response of Labrador air temperatures to volcanic events is profound but also related to the interaction between volcanism and the Arctic Oscillation which has a tendency towards positive values (correspondingly colder Labrador Air Temperatures) in the years following a high magnitude event.

    Way, R.G. and Viau, A.E. (In press). Natural and forced air temperature variability in the Labrador region of Canada during the past century. Applied and Theoretical Climatology.
    http://link.springer.com/article/10.1007/s00704-014-1248-2

    • Steven Mosher
      Posted Oct 13, 2014 at 11:14 PM | Permalink

      Congrats! glad to see that published!

  7. Jeff Norman
    Posted Oct 13, 2014 at 5:02 PM | Permalink

    Apropros of nothing, I find it interesting that the last glaciers final disappeared from northern Quebec a little over 6,000 years ago.

    • Robert Way
      Posted Oct 13, 2014 at 5:06 PM | Permalink

      Technically there are still a couple of glaciers on the Quebec side 😉
      http://www.igsoc.org/journal/60/223/t13j195.html

      • Jeff Norman
        Posted Oct 13, 2014 at 6:05 PM | Permalink

        Okay Robert that is very neat, thank you. Nice paper. I enjoyed reading it and learned. Are these ice masses considered to be the remnants of the Laurentide Glaciation?

        • Robert Way
          Posted Oct 14, 2014 at 10:52 AM | Permalink

          Well that’s one of the big research questions we have. They’re certainly not remnants of the ice sheet per say in that locally they have extended far beyond their small cirque basins (many many km) and we do not know whether they’re neoglacial of origin or if they survived their the hypsithermal and were individual glaciers that coalesced with the ice sheet when it overtopped the Torngats. There was a great deal of prior research on whether the Torngats were overtopped by ice sheets at the LGM or not but a couple papers (Marquette et al. 2004; Staiger et al. 2005) have provided compelling evidence based on cosmogenic exposure dating of erratics that the Torngats were overtopped by cold-based ice that did not erode the mountaintop block fields.

          So maybe the short answer is that eventhough they once were part of the laurentide ice sheet – they most likely are able to form on their own with climatic deterioration due to their unique setting on the North Atlantic.

          http://geochronology.earthsciences.dal.ca/gosse-downloads/Marquette%20CJES%202004.pdf
          http://geochronology.earthsciences.dal.ca/gosse-downloads/Staiger2005ESPL.pdf

  8. Jeff Norman
    Posted Oct 13, 2014 at 5:07 PM | Permalink

    Apropos to this post, this 1975 paper discusses the correlation between prehistoric climate in Northern Quebec and Baffin island.

    http://pubs.aina.ucalgary.ca/arctic/Arctic29-1-48.pdf

  9. DGH
    Posted Oct 13, 2014 at 7:14 PM | Permalink

    Figure 1 is labeled “Gennaretti 2013”. Should be 2014, no?

  10. MikeN
    Posted Oct 13, 2014 at 9:32 PM | Permalink

    That warm period in 1790, how far does that extend? First I’ve noticed it.

  11. Alexej Buergin
    Posted Oct 14, 2014 at 3:05 AM | Permalink

    Seems they did not have to “hide the decline”.

  12. Rob Wilson
    Posted Oct 14, 2014 at 3:41 AM | Permalink

    Hi Steve
    you’re quite busy lately – nice to see you back looking at data and methods again.
    The Gennaretti study is very important and is a welcome addition to the expanding NH network of longer temperature sensitive tree-ring records.

    You however forgot to include the post 1750 Quebec record I used for my JGR07 paper. I think you’ll see that they agree better than the Gaspe record but all these records are not located close together, so we need to keep in mind that there will be regional differences within Quebec. Having said that, these new data are a significant addition for the study of past climate for northern Quebec.

    NB, the data have been RCS processed – using a non-standard approach – not that there really is a standard approach. It would be good to see more experiments with different variants of the RCS method – i.e. without “corrections”, use of signal free etc – to assess how sensitive the final chronology “shape” is to different processing choices.

    My gut feeling is that the really sexy record that will come from these samples will be the MXD series but do not know if they have the funds to generate the relevant data.
    Rob

    Steve: Rob, I’ll take a look at your chronology, but with respect, the Gaspe chronology has been used in multiproxy studies in one guise or another. BTW despite all the articles presenting “RCS” results, there isn’t any technical article that presents or justifies the various ad hoc site decisions that specialists use in the development of such chronologies. Obviously specialists are aware of inter-site inhomogeneity – Esper describes this very clearly, but dealing with it on a practical basis is not at all easy and, whenever I look at actual RCS examples in detail, e.g. Briffa’s Yamal article in 2013, the issues seem to be ad hocked or ignored, with the result that effects arising from site inhomogeneity are presented as chronology. As we’ve discussed before, I view the problem as a complicated mixed effects problem and think that specialists are unwise to ignore this prior research. I also think that the tradition of “dimensionless” chronology units ought to be abandoned in favor of more conventional statistical approaches, but that’s another story.

    • Rob Wilson
      Posted Oct 14, 2014 at 10:40 AM | Permalink

      I agree – hence why it is best to try multiple approaches to get a feel of the sensitivity of the chronology outcome due to different “ad hoc” choices.

      • Steve McIntyre
        Posted Oct 14, 2014 at 11:46 AM | Permalink

        Rob writes:

        I agree – hence why it is best to try multiple approaches to get a feel of the sensitivity of the chronology outcome due to different “ad hoc” choices.

        From a statistical perspective, disputes arise when these various experiments are not reported and seemingly arbitrary choices are presented. Wagenmakers has sharply criticized such practices in social psychology as “data torture”. In less contentious terminology, any results that emerge from exercising such “researcher degrees of freedom” are exploratory, rather than confirmatory – thus requiring an out-of-sample fresh sample to confirm. Because Jacoby only archived a fraction of his data – the data that contained the “signal” that he was looking for – I don’t think that any of the Jacoby-D’Arrigo data meets objective standards for confirmatory analysis. Thus the importance of the Gennaretti results.

      • Tom T
        Posted Oct 14, 2014 at 2:34 PM | Permalink

        You mean like removing a highly questionable proxy from your data to see its effect. And upon seeing that your entire signal is dependent on that one series you chose to keep it in and call your reconstruction robust?

      • dfhunter
        Posted Oct 14, 2014 at 6:17 PM | Permalink

        Hi Rob,
        but was wondering how “SCOT2K:(2013-2016) Reconstructing 2000 years of Scottish climate from tree-rings” is going ?
        any early results available ?
        sorry bit OT Steve

    • Steve McIntyre
      Posted Oct 14, 2014 at 11:39 AM | Permalink

      Added the following postscript to the post:

      In comments below, Rob Wilson observed that Wilson et al (2007) contained a northern Quebec (NQU) chronology and suggested a comparison, provided below. Wilson’s NQU chronology is based on an old Fritts series (Fort Chimo) and still unarchived data for Pyramid and Lac Tesaliuk, both sites attributed to Payette (2007), though only the former site is mentioned in the article. Both chronologies are sort-of similar through from 1775-1915 or so, but the Wilson chronology has a much more pronounced pulse in the 20th century and more of a divergence problem.
      gennaretti_vs_wilson

      Examining Payette (2007), it appears that even nearby site chronologies can have very different patterns, depending on regeneration. It’s not evident to me that you can average such incommensurate series in any reasonable way. Payette 2007 has some interesting pictures showing subfossil trees above the present treeline, reminiscent of Polar Urals.

      payette_2007_figure_8_pyramid

      • Craig Loehle
        Posted Oct 14, 2014 at 11:58 AM | Permalink

        I have been following Payette’s work for years. I think he is an excellent ecologist and knows his trees. Not someone whose only contact with trees is the core he drills.

    • Genghis
      Posted Oct 18, 2014 at 9:15 AM | Permalink

      Could it be that inhomogeneity is the rule rather than the exception?

      Maybe that is the most important finding from all of this.

  13. scf
    Posted Oct 14, 2014 at 6:18 AM | Permalink

    I believe I can see in the Quebec tree ring chronologies the effect of the laki haze in the northern hemisphere from the laki volcanic eruption in 1783 in Iceland.

  14. JamesG
    Posted Oct 14, 2014 at 9:39 AM | Permalink

    Does it now correlate to the local thermometer record I wonder?

  15. Steve McIntyre
    Posted Oct 14, 2014 at 12:36 PM | Permalink

    Here is a similar comparison of the new Gennaretti chronology to the D’Arrigo et al 2006 Labrador RCS chronology. This is related to the Labrador STD chronology shown in their 2003 article, but isn’t the same. While there are points of similarity in the early 19th century, the D’Arrigo RCS version shows much weaker growth in the downticks around 1600 and 1700, and has a very large growth pulse in the mid-20th century that is not observed in the Gennaretti data.  One of the issues that concerns me about the D’Arrigo 2006 LAB chronology is that inhomogeneity between sites appears to be expressed in the chronology – a very difficult and perhaps insurmountable issue with RCS methods.
    gennaretti_vs_darrigo-lab

  16. Geoff Sherrington
    Posted Oct 14, 2014 at 10:29 PM | Permalink

    Am I missing something?
    In the calibration step one would usually construct a graph-type relation between tree response and applicable temperature. This calibration would give adequate information as to which is the right way up.
    (Unless a less clear method has been devised).

  17. Jimmy Haigh
    Posted Oct 15, 2014 at 3:40 AM | Permalink

    Wow. That is almost a coherent sentence from Mosh.

    • Steven Mosher
      Posted Oct 15, 2014 at 1:48 PM | Permalink

      thank heavy traffic

  18. Posted Oct 15, 2014 at 12:41 PM | Permalink

    Nick Stokes, why do you minimize the impact of not getting the work done correctly?

    Yes, errors can be made. But what is your point?

    A key problem is that erroneous results of scientists of an alarmist mindset are being used to interfere in people’s lives, as often urged by those scientists, there is a moral obligation to get it right, and in the event of error to correct the record.

    Stephen has pointed to a few individuals who correct some errors, how rare are they?

    • Posted Oct 15, 2014 at 12:44 PM | Permalink

      And note that inverting the sense of data shows failure to read through the work thinking all the while.

      That’s what “peer review” should do.

  19. DocMartyn
    Posted Oct 16, 2014 at 6:45 AM | Permalink

    Why tree-rings widths in North America may not give a simple correlation with climate could be due to the soil ecology. his is the opening of “Distribution and impacts of invasive earthworms in Canadian forest ecosystems”; J. A. Addison.

    “In Canada it is generally accepted that most indigenous earthworms did not survive glaciation, and that the majority of the earthworms now inhabiting Canadian soils are relatively recent introductions of European origin. Although these exotic earthworms are generally considered to be beneficial in agricultural soils, their effects can be less benign in forested ecosystems. Studies have shown that invading earthworms can significantly alter the forest floor, affecting the distribution of carbon, nitrogen and other chemicals, roots, microbes and other elements of the soil fauna, and even understory vegetation. This paper summarizes the current distribution of exotic and native earthworm species in Canadian forests and draws on the results of studies of invasion patterns and environmental impacts in northern forests in North America and Europe to discuss potential outcomes for forests in Canada. The potential for variables such as temperature, pH, litter palatability and dispersal, to limit or promote the invasion of exotic earthworms in Canadian forests is discussed, and areas for future research are proposed. The same earthworm species that are invading forests in northern Europe and the US are also invading Canadian forests. Several species of exotic earthworms are already established in a wide range of deciduous and coniferous forest types, including the boreal. Evidence is presented to suggest that further expansion into Canadian forests is likely.”

    Typically when earthworms invade a local there is first a slowing of growth, while the ecosystem is remodelled and then generally faster rates of growth.

    http://www.nrri.umn.edu/worms/downloads/publications/Larson%20et%20al%202009.pdf

    In any North American forest where earthworms are found one knows that these are an introduced species, and that somewhere in the record should be their footprint.

  20. Posted Oct 18, 2014 at 2:44 PM | Permalink

    If one wants to make a flat hockey stick handle, grab a bunch of data that contains far more noise than signal and mush it all together. The peaks will integrate and the result will be worse than we expected.

  21. Posted Nov 11, 2014 at 9:29 AM | Permalink

    Thank you very much for this interesting article comparing our results to other studies.

    Here below some considerations:

    1. The main theme of our paper is the influence of volcanic events on the climate and the forest growth over the central Quebec-Labrador Peninsula during the last millennium. We also think that the longer term climate trends (i.e. Medieval anomaly vs last century) of our regional climate reconstruction are robust (see Fig. S6 of our paper). However, some improvements can still be done to our chronologies, especially in the most recent part (i.e. improved sampling strategy of living trees).

    2. Similarities and dissimilarities with the Gaspe chronology: This is very intriguing because before the 20th century the chronologies are very similar and after that they diverge greatly. What I can say is that we tried to do our best to homogenize subfossil and living trees in our chronologies to avoid biased chronology ends (see our paper). However, as I said before, some minor changes can be expected with an improved sampling strategy of living trees.

    3. The choice of the standardization method: We tried the signal free RCS but this method would not change our results significantly.

    4. The comparison of our reconstruction with data from Labrador and Baffin Island: You have to keep in mind that the climate of the central Quebec-Labrador Peninsula at an annual time scale is very different from the climate of the Labrador coast or the Baffin Island. For example, our reconstruction has null and negative correlations with summer temperatures over the Labrador coast and the Baffin Island, respectively (see fig. 2 of our paper).

    5. The importance of tree growth forms on tree-ring data: You interpret the large difference between the two Pyramid chronologies in the Payette’s paper as regeneration-related. However, this difference is most likely the result of stronger winter abrasion of tree growth forms at the more exposed Pyramid 1 site, as compared to the Pyramid 2 site. These trees were eroded during the very cold early 19th century and could not recover when the climate improved at the end of the 19th century. As discussed in several additional Payette papers this shows the non-linear tree-growth-climate relationship in wind-exposed trees at the tree line. In our study region, trees are not eroded (as shown by their normal growth form), and all of our trees were sampled in wind-protected sites to avoid the influence of changing growth forms.

    6. The most important factor that could have disturbed tree growth in our sites is fire disturbance. To limit the influence of wildfire on our data we selected special sites that rarely burned (6 different sites), we sampled as many trees as possible within each sites to build 6 independent local chronologies with large sample replication, and we used a master chronology implementation method that reduces the influence of outliers in local chronologies. If you want to know more on the influence of wildfires at our sites, please have a look at the two papers that we published in Journal of Ecology.

    7. The amplitude of our reconstruction is greater than the others of your figure 4 because our reconstruction is based on tree-ring data from one small region and not on an assemblage of proxy data from several locations. This means that the spatial domain of our reconstruction is much smaller. However, I think that further studies on the sensitivity to the choice of the reconstruction method can be done to clarify this aspect.

    Thanks again for your interest in our data

    Fabio Gennaretti

    PS. point 5 was written by Dominique Arseneault

%d bloggers like this: