Naurzbaev et al  is a terrific article by Naurzbaev, Hughes (yes, that Hughes) and Vaganov about deducing climate information from tree-ring growth curves in Siberia. (I find Naurzbaev’s work consistently interesting.) They studied 34 larch sites in a meridional transect from 55 to 72 N (at a longitude of about 90-100E) and 23 larch sites along an altitudinal transect from 1120 to 2350 m around Tuva (~ 51N, 95E). So they’ve already started off with a consistent population – not some grab-bag like MBH98 or cherry-picking like Jacoby. The transect includes the Taimyr peninsula, chronologies from which occur in Esper et al  and Mann et al [EOS 2003].
Their abstract (which I’ll quote below) gives no clue as to the interesting facts in the article. At each site, they fit a negative exponential age curve to ring width RW of the form:
RW ~ Imin + I0* exp (-C* age)
and then keep track of the Regional Growth Curve (RGC) parameters I0, Imin and the the average ring width for trees 50-100 years old (I50_100) across the transects. Sounds sensible already.
In the latitudinal transect, they found strong negative correlations of all three parameters with latitude (r =-0.74, -0.67 and -0.77) and high positive correlations with mean annual and mean summer temperatures, which they interpret as showing a systematic gradient in RGC parameters along the transect. (They also found a strong positive correlation to precipitation in the northern part of the transect (64-72N) and a negative correlation in the southern part (52-64N). They note a weak negative correlation between maximum age and average growth – along the lines of the point originally made by Schulman  that slowly growing trees achieve greater longevity than fast growing trees.
In the altitudinal transect, they found found a gradient in the growth curve with elevation that was nearly twice as strong per unit temperature change, observing that I 50_100 changes by 0.09 mm/deg C on the elevational transect and only 0.05 mm/deg C on the meridional transect. This is not discussed unfortunately.
They used these relationships to estimate past conditions for subfossil trees, nicely quantifying for a few interesting cases the seemingly obvious point that treeline changes reflect past climate. But here they do much better than that point: they’ve studied and quantified the past growth and placed in a modern context using the information from their transects. I’ve excerpted their Table 2 below, which summarizes the major points.
They comment about the site in the second line of the table as follows:
Trees that lived at the upper (elevational) tree limit during the so-called Medieval Warm Epoch (from A.D. 900 to 1200) show annual and summer temperature warmer by 1.5 and 2.3 deg C, respectively, approximately one standard deviation of modern temperature. Note that these trees grew 150-200 m higher (1-1.2 deg C cooler) than those at low elevation but the same latitude, implying that this may be an underestimate of the actual temperature difference.
Their summary is very odd here – surely it’s more reasonable to allow for the altitude effect: this would yield an estimate that the MWP annual temperature was 2.58-2.86 deg C warmer than at present and the summer MWP temperature was 3.38-3.66 deg C warmer than at present. There’s nothing here that’s inconsistent with Shiyatov’s observations about the Polar Urals, which I’ve mentioned before. The third line shows a subfossil site at 72N that was dated to ~ 4140-2700 BC (6140-4700 BP), before the climatic decline of the late 3rd millennium BC. Here they report:
The RGC corresponds to a mean annual temperature of -6.6 deg C and a mean May-September temperature of 9.2 deg C, and also to modern conditions at approximately 65N. The estimated latitudinal difference is about 7N. This implies that at the time these trees grew, growth conditions in this area were equal to those in the modern north taiga region. Average differences between modern and ancient annual temperature were estimated as 6.7 deg C (minimum difference is about 2.3 deg C if the standard deviation is taken into account), with the same magnitude of difference for May”€œSeptember temperature.
The key comparison numbers are the difference between the latitude in column 1 and the latitude equivalent on column 6. The fourth site (line 4) has similar results and a latitudinal difference of 6 deg 40′ is estimated. They conclude:
the Holocene Climatic Optimum on the Taimyr Peninsula, the northern tree line was located about 450 km (minimally 250 km) further north than at present….
This is consistent with other observations on ancient tree lines in northern Eurasia, some parts of North America and the north of Scandinavia (Huntley and Prentice, 1988; Kullman and Kjallgren, 2000; MacDonald et al., 1993, 2000). For example, Kullman and Kjallgren (2000) reported that the greatest elevational extension of forest in Scandinavian mountains at the beginning of the Holocene was about 500 m above the modern upper tree line, which means temperatures were warmer by at least 3 deg C (assuming 0.6 deg C temperature decrease for every 100 m increase in elevation). There are several other regions in Northern Eurasia and northern North America where RGC parameters may be subject to relatively simple control by temperature, and where abundant relict wood exists covering several periods in the Holocene. Thus, the possibility exists of adding a new kind of record for such times and regions, even if it is not possible to develop well-replicated tree-ring chronologies known to conserve century-scale climate information for the whole period.
With all this interesting information, you’d think that they’d have a dynamite abstract. Unfortunately, their abstract gives little clue as to these striking findings:
Regional growth curves (RGCs) have been recently used to provide a new basis for removing nonclimatic trend from tree-ring data. Here we propose a different use for RGCs and explore their properties along two transects, one meridional and the other elevational. RGCs consisting of mean ring width plotted against cambial age were developed for larch samples from 34 sites along a meridional transect (55″€œ72 degN) in central Siberia, and for 24 sites on an elevational gradient (1120 and 2350 m a.s.l.) in Tuva and neighboring Mongolia at approximately 51 deg N. There are systematic gradients of the parameters of the RGCs, such as I0-maximum tree-ring width near pith, and Imin, the asymptotic value of tree-ring width in old trees. They are smaller at higher latitude and elevation. Annual mean temperature and mean May”€œSeptember temperature are highly correlated with latitude here, and hence RGC parameters are correlated with these climatic variables. Correlations with precipitation are more complex, and contradictory between meridional and elevational transects. The presence of a similar gradient in the elevational transect is consistent with temperature being the causal factor for both gradients, rather than, for example, latitude dependent patterns of seasonal photoperiod change. Taking ring measurements from collections of relict and subfossil wood, the RGC”€œlatitude and RGC”€œtemperature relationships are used to estimate paleo-temperatures on centennial time scales. These estimates are consistent with earlier btraditionalQ dendroclimatic approaches, and with independent information on the northern extent of forest growth in the early mid-Holocene. It may be possible to use this same approach to make estimates of century-scale paleo-temperatures in other regions where abundant relict wood is present.
Mukhtar M. Naurzbaev, Malcolm K. Hughes, Eugene A. Vaganov, 2004. Tree-ring growth curves as sources of climatic information, Quaternary Research 62, 126″€œ 133