I more or less jumped into the middle of some technical issues pertaining to Esper et al , without properly describing the study. Esper et al.  is one of the multiproxy studies that is included in all the spaghetti diagrams. It supposedly shows more "variability" than Mann, which is the issue that the Hockey Team is keen on discussing. The issue which interests me is a different one: what are the "active ingredients" in determining the relative levels between the MWP and the 20th century.
Every time I look at these diagrams and try to figure out how they work, I keep hearing strains from a pop song from my childhood:
who put the bop in the bopsy-wopsy-wop?
who put the ram in the rama-rama-ding-dong?
We will look for the rama-rama-ding-dong in Esper et al .
Esper uses 14 sites, shown in the following graphic. You’ll see lots of familiar faces from other multiproxy studies: Polar Urals, Tornetrask, 2 foxtail pine sites. More on this later.
Figure 1 from Esper et al. 
Here’s what Esper’s reconstruction eventually looks like. It’s maximum is at 980, but the 11th and 12th centuries are below 20th century levels. In spaghetti graphs, the instrumental record is jammed on at the end. The bottom part of the graphic shows the number of chronologies used per year in the reconstruction – declining to 4 in the early part of the 11th century. The number of sites used increases through the 12th century.
As in all of these studies, not all of the sites go back to the MWP. Esper provides the following figure in his SI, showing which sites are contributing. If you look closely at the 11th century, you’ll see 3 sites contributing to the "nonlinear" chronology [nonlinear to be discussed, rather than explained below] – Polar Urals, Sol Dav (Mongolia), Tornetrask; and 3 "linear" sites – Boreal, Taimir and Upper Wright. Polar Urals and Tornetrask are old friends – both are Briffa RCS series and both have been discussed at length here. Esper only uses RW series and presumably re-calculated, so his specific series won’t be the same as Briffa’s reconstruction, but some of the issues will presumably carry over. I’m not 100% certain that the Polar Urals version used here is the same as Briffa, since it looks like it goes earlier than Briffa. There may be two different measurement series in play. Polar Urals and Tornetrask are 2 of 3 sites in the 11th century in Jones et al , so this is hardly "independent" of that study. Sol Dav is a Jacoby site and has quickly been incorporated into the Hockey Team multiproxy corpus, being used in MBH98, Mann and Jones , Jones and Mann .
Boreal and Upper Wright are both foxtail sites. (Foxtails are on the opposite side of the Bishop valley from the bristlecones and will interbreed with bristlecones.) So, for the issue of relative modern and MWP levels, this sure looks like the usual gruel of bristlecone/foxtails, Polar Urals, Tornetrask etc.
Esper does not provide any data citations. For reference, AGU has an excellent policy on data citation, unfortunately not applied in paleoclimate studies in AGU journals, and no comparable policy seems to exist for Science. Instead of providing data citations, Esper thanks a variety of authors and gives limited references in the following statement:
Here, we will not explicitly model the temperature signals of the individual tree-ring chronologies, because this has mostly been done already (11, 13–19).
The acknowledgement and references 11, 13-18 are summarized below, together with a candidate version at WDCP where I’ve been able to deduce one. Reference 19 is to the multiproxy study Jones et al, 1998, which, whatever its merits or lack thereof, cannot be construed as "explicitly modeling" the temperature signals of these tree ring chronologies. Not all of the references deal with modeling the "temperature signal" for their site. Briffa’s two articles obviously do. I’ve commented extensively on the Tornetrask and Polar Urals sites. D’Arrigo [Jacoby et al.] also do for their Mongolian site; I haven’t seen much information on the ecology of the Sol Dav, Mongolia site, not a strong concern of the Hockey Team, but I’d be astonished if precipitation were not a co-variable in a Mongolian site. Luckman’s "Athabaska" site seems to be called the "Jasper" site elsewhere and does discuss temperature – so that’s 4 out of 14. Lloyd and Graumlich  is a highly interesting article on treeline changes at California foxtail (bristlecone brother) sites, but does not provide a calibration to temperature for the Boreal and Upperwright sites. Mangazeja, Jaemdtland and Quebec do not extend into the 20th century; Mangazeja and Jaemdtland not even very far into the 19th Century, so it’s not clear how calibration was demonstrated for them or even why they are included out of all the possible sites in the world. I’ve tracked down some references to the non-Briffa/Jacoby sites and will post on them. Here’s a summary:
|Ath, Athabasca;||B. Luckman||Luckman et al., Holocene 1997||cana170w|
|Bor, Boreal;||Graumlich and A. Lloyd||Lloyd and Graumlich, Ecology 1997||NA|
|Mac, Mackenzie;||Szeicz, J.||Szeicz and MacDonald, Quat Res 1995||cana138|
|Que, Quebec;||S. Payette and L. Filion||NA||NA|
|Upp, Upperwright;||Graumlich and A. Lloyd||Lloyd and Graumlich, Ecology 1997||NA|
|Got, Gotland;||Bartholin and Karlen||NA||swed022|
|Jae, Jaemtland;||Bartholin and Karlen||NA||swed023|
|Tir, Tirol;||V. Siebenlist-Kerner||NA||swit177|
|Tor, Tornetraesk;||Bartholin and Karlen||Briffa et al, Clim Dyn 1992||swed019|
|Man, Mangazeja;||S. Shiyatov||NA||russ067|
|Mon, Mongolia;||G. Jacoby||D’Arrigo et al , GRL 2001||mong003|
|Pol, Polar Urals;||S. Shiyatov||Briffa et al., Nature 1995||russ021|
|Tai, Taimir||M. Naurzbaev||NA||NA|
|Zha, Zhaschiviersk.||F. Schweingruber||NA||russ053|
In a couple of cases (Graumlich’s 2 foxtail sites, Quebec), I can’t find any formal publication of the chronology, although I’ve found some interesting site discussions which I will review on a nother occasion.
Linear and Nonlinear
As noted before, Esper divides the cores into "linear" and "nonlinear", but does not provide an operational definitions of how to allocate cores to one or the other, and, the reference for this procedure (Esper et al, 2003) does not address the question. I’ve looked at a lot of core plots and cannot even begin to surmise an operational definition. Anyway, his SI provides the following illustration. Each squiggle will be a RCS diagram from a different site. I’ve been illustrating some age-standardization curves for some Esper sites, but without the division into linear and nonlinear ( copying the procedure for Gotland in Esper et al , the supposed reference for linear/nonlinear.)
Original Caption: Supplemental Figure 1. Arithmetic mean curves of individual ring width series from different sites after aligning by cambial age. Athabasca, Mangazeja and Polar Urals are represented in both data sets, because some of the series from these sites belong to the linear and others to the nonlinear data set.
Going from Site Chronologies to a NH Index
I’m pretty sure, but not 100%, sure of what is done here. Some sites had two RCS chronologies calculated and other sites had one RCS chronology. Then each of the linear and nonlinear were averaged to produce a NH linear and a NH nonlinear. Then the NH average was obtained by averaging the two. Esper et al :
Two smoothed RCs were estimated from the averaged biological age-aligned data in the linear and nonlinear groups [SM – presumably for each site]. Tree-ring departures from each RC were calculated as ratios for the linear and nonlinear data sets following standard procedures (6). The resulting tree-ring indices were then averaged into linear and nonlinear mean value functions to produce two nearly independent tree-ring chronologies covering the years 800–1990 (Fig. 2A).
The caption to Figure 2 adds:
D). The NH reconstruction was derived from the 14 site RCS chronologies after each was smoothed with a 20-year lowpass àÆà ⽬ter to emphasize multidecadal to multicentennial time scales. The two-tailed 95% confidence limits were estimated with the use of a bootstrap procedure (8).
In passing, the reference to the bootstrap procedure in the SI (8) is to Briffa et al. 1992, which does not contain any description of a bootstrap procedure. Here is the allocation of sites to linear and nonlinear, as far as I can tell:
|Pol, Polar Urals;||yes||yes (modern)|
Note: bold if in 11th century
The two chronologies are very similar over the past ~1200 years. Each shows evidence for inferred above average temperatures during the MWP (900–1300), although some differences in the timing of peak conditions are evident. This may reflect weakness of sample depth before the year 1200 (Fig. 2B), and the lack of uniform spatial representation as the number of sites decreases back in time (8). However, it could also reflect higher spatial variability of MWP temperatures in the NH extratropics (22). More tree-ring collections that span the MWP are needed that are long and well-replicated to determine which hypothesis is correct. After the year 1200, the linear and nonlinear RCS chronologies lock together remarkably well at multidecadal and centennial time scale.
I guess "very similar" is in the eye of the beholder. The two series do not look similar to me in the MWP. This non-similiarity is attributed by Crowley to regionalization – but then why is there so much similarity in later periods? We are really dealing with only 6 series in the MWP –
2 of which are foxtails. I’ll post up some more on the individual sites.
Overall, it’s hard to see how this study is "independent" of the other Hockey Team studies. Without access to every single data set, it’s impossible to do a sensitivity study to the impact of foxtails and/or the impact of screwed up crossdating at Polar Urals.