Esper's Gotland, Part 2

One of the reasons that I’m going to parse through Esper’s Gotland series is the virtual absence of archived RCS chronologies, despite the fact that they are sweeping the field in multiproxy studies. Continuing on with Esper’s Gotland series

First (and this is not shown by Esper) is a simple histogram of RWM and MXD measurements. Obviously both distributions are skewed. It’s actually a somwhat interesting problem to think about what sort of distribution of temperatures and distribution of age responses would combine to yield final distributions like this.

Figure 1. Gotland Histograms – RW and MXD.

The top panel of Esper’s Figure 3 shows time series for RW and MXD as follows (the figure as it stands is hard to tell between living and dead):

Esper Figure 3 Top Panel.

Next is my emulation of Esper’s Figure 3 Top Panel. This is scaled a little differently, but it’s obviously that I’m working with the same data (as indicated before).

Esper Figure 3 Top Panel – Emulation.

Next is Esper’s plot of ring widths and MXD against age as shown below. Esper says of this graph:

Differences, such as slope and level, between living and dead trees, almost disappear after aligning the series by cambial age, indicating that living and dead tree data indeed belong to the same biological growth population (Figure 3B)

Esper Figure 3 Middle Panel.

Here is my replication of his Figure 3 Middle Panel. While it is very similar, there are some puzzling differences – most notably the upspike in the very young trees in Esper’s version, which I was unable to replicate.

Esper Figure 3 Middle Panel -Emulation

Next Esper makes a scatterplot for each core of the length of the core against the mean ring width, about which he says:

Another proof for the existence of a single Gotland biological growth population is the characteristically homogeneous decline of the average TRW with increasing series length [Fritts 1976]. A corresponding analysis of MXD, generally, does not show such a relationship.

Esper Figure 3 Bottom Panel

Here is my replication of this figure.

Esper Figure 3 Bottom Panel – Emulation

Is there a moral to this? I don’t know; Im still working through this. The skew in the distributions (which is very characteristic) is not mentioned by Esper and is worth keeping an eye on.


  1. TCO
    Posted Sep 11, 2005 at 5:27 PM | Permalink

    Does anyone use the RW and MXD more sophisticatedly than just pick one or the other? Would think that 4 choices of hi/lo RW/MXD would mean something. Maybe this can help to filter out precipitation for instance if one really understands the biology.

  2. Knut Knutsen
    Posted Sep 12, 2005 at 12:10 AM | Permalink

    Altitude effects, rise of treeline at Gotland???? The highest point on Gotland is about 100 m above sea level.

  3. Steve McIntyre
    Posted Sep 12, 2005 at 7:18 AM | Permalink

    Knut, you’re quite right about Gotland. My preamble was not very helpful. I think that there may be a salvageable point in the preamble, as the relatinoship between temperature and germination may apply at lower altitudes if they are temperature stressed as this one supposedly is. However, the post is about methodologies and the preamble is irrelevant so I’ve deleted it. I’ll return to the theme on another occasion.

    My reason for looking at this site is purely methodological. There is virtually no detailed published information on RCS methods; here, Esper has provided a more detailed account of his methodologies in connection with this site than elsewhere, so I’m parsing through his methods on this template.

    Thanks for the comment.

  4. Douglas Hoyt
    Posted Sep 13, 2005 at 6:15 AM | Permalink

    In Figure 1, if you took the cube root of the ring widths and then reversed the x-axis and plotted the numbers, would you get a distribution identical with the MXD distribution in Figure 1?

%d bloggers like this: