I’ve talked recently about the phenomenon of cherry picking tree ring chronologies with upticks in the small-subset (10-20) compilations used in typical Hockey Team multiproxy studies (e.g. Jones et al 1998; Crowley and Lowery 2000, etc., most recently Osborn and Briffa, 2006; and to a slight lesser extent D’Arrigo et al, 2006 (where there was a discipline resulting from the need to report their own extensive fieldwork.) I’ve referred in passing to evidence that there has been a large-scale decline in density and ring width over hundreds of sites and it might be useful to do a quick survey of this evidence – which comes ironically mainly from Briffa himself. Given that there has been an overall decline in both measures for a population of over 300 sites supposedly chosen for temperature sensitivity, what are the chances of picking (by chance) the Yamal, Mongolia, bristlecone and foxtail series – all with strong growth. None.
Also see posts 529, 570 and 586.
First, here is an important figure from Briffa et al. [Proc Roy Soc 1998] showing the relative decline of MXD and RW relative to temperatures since 1960. (Briffa has published much more on MXD (density) series than on RW (ring width) series – so watch below for the difference. )
Briffa et al. 1998 Original Caption. Figure 6. Twenty-year smoothed plots of averaged ring-width (dashed) and tree-ring density (thin solid line), averaged across all sites in Figure 1, and shown as standardized anomalies from a common base (1881-1940), and compared with equivalent-area averages of mean April-September temperature anomalies (thick line). [SM – it looks to me like the labels in the caption are reversed between desnity and temperature]
In the same year, Briffa et al [Nature 391, 1998] reported that the phenomenon was regional, showing two graphics to illustrate the effect for 314 sites. Here is an interesting graphic showing the geographical distribution of the effect – and that the declines in both MXD and RW were in northerly sites, with increased growth characteristic of more southerly sites (which would presumably be more affected by precipitaiton although this is not discussed in the article.)
Briffa et al 1998 Caption: Figure 1 Spatial patterns of relative tree-growth decline. a, The location (circles) of tree-ring chronologies and the division (black lines) into regional averages. b, The locations (black lines) of the grid-box temperatures used for comparison with the tree growth series. The coloured contours show where the ring density (a) and ring width (b) are enhanced (positive) or suppressed (negative) relative to summer temperature during the period 1975-85 compared to the period 1935-45. Values are dimensionless, being the change in the difference of two normalized time series in each grid box. Definition of regions and number of sites: southwestern (SWNA, 53 sites), northwestern (NWNA, 30) and eastern (ENA, 34) North America; northern (NEUR, 46) and southern (SEUR, 72) Europe; western (WSIB, 42), central (CSIB, 31) and eastern (ESIB, 6) Siberia; all 125 sites in SWNA and SEUR form the composite region SOUTH, and all 189 sites in the six northern regions form the NORTH region; ALL is an average of all 314 sites.
Briffa et al [Nature 391 1998] summarized the situation as follows, showing a figure with a comparison by region:
During the second half of the twentieth century, the decadal-scale trends in wood density and summer temperatures have increasingly diverged as wood density has progressively fallen. The cause of this increasing insensitivity of wood density to temperature changes is not known, but if it is not taken into account in dendroclimatic reconstructions, past temperatures could be overestimated… In the areas where the growth data extend through to the warm late 1980s and early 1990s (NEUR, WSIB, CSIB, ESIB), the divergence is at a maximum in the most recent years. Over the hemisphere, the divergence between tree growth and mean summer temperatures began perhaps as early as the 1930s; became clearly recognisable, particularly in the north, after 1960; and has continued to increase up until the end of the common record at around 1990.
After these 1998 articles, you don’t see much further discussion by Briffa of the RW decline; but the MXD decline is illustrated a few more times and discussed up to the present. Briffa [QSR 2000] shows on an annual basis, the decline in average MXD up to the 1990s.
Briffa [2000[ Original Caption. Fig. 5. An indication of growing season temperature changes across the whole of the northern boreal forest. The histogram indicates yearly averages of maximum ring density at nearly 400 sites around the globe, with the upper curve highlighting multidecadal temperature changes. Extreme low density values frequently coincide with the occurrence of large explosive volcanic eruptions, i.e. large values of the Volcanic Explosivity Index (VEI) shown here as arrows (see Briffa et al., 1998a). The LFD curve indicates low-frequency density changes produced by processing the original data in a manner designed to preserve long-timescale temperature signals (Briffa et al., 1998c). Note the recent disparity in density and measured temperatures (¹) discussed in Briffa et al., 1998a, 1999b). Note that the right hand axis scale refers only to the high-frequency density data.
If you’re worried that the Hockey Team might have “moved on”, here’s a graphic from the more recent Briffa et al.  showing the relative MXD decline (but not this time, the RW decline).
The Briffa et al MXD reconstruction is canonical in Hockey Team spaghetti graphs and was used in IPCC TAR. I’ve pointed out that IPCC TAR used a version which was truncated in 1960, rather than showing the post-1960 decline. I posted on this in May here and here showing how the IPCC truncation was imperceptible without a blow-up.
Shortly after I posted on this, the matter was discussed at Wikipedia here, where they seemed to think that Briffa had adequately justified the truncation. However, IPCC cited Briffa [QSR] where there is no truncation. The first occurrence of the truncation was in Briffa et al [JGR 2001] (not cited by IPCC), where the truncation is done in Plate 3 (as can be prroven in the subsequently archived digital version). HOwever, I have been unable to find ANY mention or justification of the truncation in Briffa et al  itself.
Needless to say, Briffa has attempted to explain this widespread decline – a seemingly difficult project given the Hockey Team commitment to the existence of a linear relationship between tree ring indexes and temperature. The first articles in 1998 merely noted the problem and didn’t make a serious attempt to explain the problem, observing [Prox Roy Soc 1998] only:
It is salient to note that relative tree-ring width, and basal area increment, also show a relative decline and divergence from the temperature curve(s), arguing against the decline in density being a compensation reaction to increasing ring growth (as is seen in forestry soil fertilizing experiments). I would imagine that higher temperatures, and possibly some increasing sensitivity to lower summer soil moisture are involved, but some additional growth-limiting factor must also be implicated. Higher CO2 would be expected to increase basal area growth, so I consider it unlikely that this is the factor.
Briffa’s big worry was that the decline could cause prior reconstructions to have high values, arguing that the late decline should be excluded in calibration. They seemingly ignoring the elephant in the room: whether these proxies can pick up warm climates at all. Briffa[ et al 1998:
in the context of long-term (multicentury and above) dendroclimatic reconstruction, this partial non-climatic enhancement of twentieth-century tree growth, particularly if it acts in tandem with temperature forcing, will bias the coefficients in any regression-based equation estimating tree growth as a function of recent measured temperatures. Hence, the magnitude of modern warming might be overestimated in the context of earlier reconstructed variability.
Again in 1998 [Nature]:
The reason for this increasingly apparent and widespread phenomenon is not known but any one, or a combination, of several factors might be involved.The common use of least-squares regression for developing dendroclimatic transfer function equations to estimate past climates, imposes an equality of means in both the predictand and predictor time series over the fitting or “‘œcalibration’ period10. Any bias in mean tree growth will, therefore, be “‘corrected’ during calibration, with the consequence that the derived regression coefficients will be biased. Our results imply that this might increasingly result in systematic overestimation of past temperatures, particularly in regions where the loss of low-frequency temperature sensitivity in tree growth is greatest (eastern Siberia and eastern North America: see Fig. 1),
In JGR 2001, although there is no mention of the post-1960 truncation, once you realize that this was done, you can see a faint allusion here (although that wouldn’t by itself give you any clue as to the truncation):
Calibration of all reconstructions against the same target series and over the same calibration period (1881-1960) reduces the potential differences due to calibraiton issues, though we acknowledge that the selected region and season may not be optimal for all reconstructions.
Briffa et al  report as follows on the problem and give what, in my opinion, is one of the most bizarre explanations – even by Hockey Team standards. In fact, I’m pretty sure that it was after reading this, that I wrote to Mann in April 2003 asking for his data:
Briffa et al. (1998b) discuss various causes for this decline in tree growth parameters, and Vaganov et al. (1999) suggest a role for increasing winter snowfall. We have considered the latter mechanism in the earlier section on chronology climate signals, but it appears likely to be limited to a small part of northern Siberia. In the absence of a substantiated explanation for the decline, we make the assumption that it is likely to be a response to some kind of recent anthropogenic forcing. On the basis of this assumption, the pre-twentieth century part of the reconstructions can be considered to be free from similar events and thus accurately represent past temperature variability.
Talk about the ultimate deus ex machina. Again, have they moved on? Briffa et al 2004 reuminate that the recent decline might be caused by ozone, but far well short of providing any proof. They say:
The network was built over many years from trees selected to maximise their sensitivity to changing temperature…. However, in many tree-ring chronologies, we do not observe the expected rate of ring density increases that would be compatible with observed late 20th century warming. This changing climate sensitivity may be the result of other environmental factors that have, since the 1950s, increasingly acted to reduce tree-ring density below the level expected on the basis of summer temperature changes. This prevents us from claiming unprecedented hemispheric warming during recent decades on the basis of these tree-ring density data alone. Here we show very preliminary results of an investigation of the links between recent changes in MXD and ozone (the latter assumed to be associated with the incidence of UV radiation at the ground). For the time being, we circumvent this problem by restricting the calibration of the density data to the period before 1960.
This ostrich-in-the-sand policy continues right through to Osborn et al [submitted 2005] (available on the internet) which says:
A number of factors were taken into account when selecting the most appropriate periods for calibration and verification of the gridded density data against observed temperatures. The most important factor is the identification by Briffa et al. (1998a) of a recent downward trend in the high latitude tree-ring density data, relative to (and apparently unrelated to) warm-season temperature. This density decline becomes large enough to impair the calibration after about 1960. For this reason, both Briffa et al. (2001) and Briffa et al. (2002a) used only pre-1961 data for calibration of their subcontinental, regional temperature reconstructions. This is a reasonable choice, provided that it is explicitly stated that this approach assumes the apparent recent density decline is due to some anthropogenic factor and that similar behaviour is assumed, therefore, not to have occurred earlier in the reconstruction period – which would otherwise introduce bias in the reconstructed temperatures. At present, no satisfactory explanation of the relative MXD decline has been identified, and further work must dictate whether this assumption will be supported or rejected (Briffa et al., 1998a, 2003, and Vaganov et al., 1999, discuss and investigate possible causes).
Again in their summary, Osborn et al say:
The second key issue that arose during the calibration procedure is more specific to the treering density data set used here, because it relates to the decline (relative to that expected on the basis of observed summer temperatures) in density over recent decades at the high latitudes (Briffa et al., 1998). It is extremely important to try to identify the cause of this decline, though investigations are currently hampered by the early sampling of many of the sites and thus the lack of widespread data since the mid-1980s (Briffa et al., 2003). Without a satisfactory explanation, we make the untested assumption that the decline is due to an anthropogenic factor that did not occur earlier in the reconstruction period. Nevertheless, additional uncertainty must surely be associated with the reconstructions because of this assumption, particularly for earlier warm periods. The decline also complicates the reconstruction method. To prevent it from unduly influencing the calibration, the grid-box calibration was first undertaken using high-pass filtered data (the decline was removed by the filtering). The tree-ring density data were also adjusted, in an artificial way, to temporarily remove the decline; this made only a small difference in comparison with the filtered calibration, although it reduced the sensitivity of the reconstruction’s mean level to the choice of calibration period and thus proved a useful, if ad hoc, way of dealing with the decline during calibration.
While Hockey Team ruminations are always amusing, the statistical issue is different. The big population of Northern Hemisphere temperature-sensitive sites (well over 300) shows declining ring widths in the latter part of the 20th century. Osborn and Briffa, 2006 have nonetheless selected some sites with remarkable late-century growth pulses: Sheep Mountain bristlecone, foxtails, Yamal (as re-processed by Briffa), Mongolia and a couple with modest pulses (Grudd’s Tornetrask). Osborn and Briffa purport to justify their selection of sites because they are relying on prior selections by Mann and Jones ; Esper, Cook and Schweingruber  etc. But this simply exemplifies the problems of lack of independence of authors and lack of independence of proxies. If the population as a whole showed dramatic increases in ring width, then you could perhaps accept the selectivity. But when the selections have such different behavior than the population of temperature-sensitive sites selected by the same author, you have to wonder.
Briffa, K.R., Schweingruber, F.H., Jones, P.D., Osborn, T.J., Shiyatov, S.G. and Vaganov, E.A. 1998b: Reduced sensitivity of recent tree-growth to temperature at high northern latitudes. Nature 391, 678-82.
K. R. Briffa, F. H. Schweingruber, P. D. Jones, T. J. Osborn, I. C. Harris, S. G. Shiyatov, E. A. Vaganov and H. Grudd, 1998. Phil.Trans. R. Soc. Lond. B (1998) 353, 65-73
K.R. Briffa, 2000. Quaternary Science Reviews 19 (2000) 87-105
Briffa, K.R., Osborn, T.J., Schweingruber, F.H., Harris, I.C., Jones, P.D., Shiyatov, S.G., Vaganov, E.A., 2001. Low-frequency temperature variations from a northern tree ring density network. Journal of Geophysical Research 106, 2929- 2941.
Briffa, K.R., Osborn, T.J., Schweingruber, F.H., Jones, P.D., Shiyatov, S.G., Vaganov, E.A., 2002a. Tree-ring width and density around the Northern Hemisphere: Part 1. Local and regional climate signals. Holocene 12, 737-757.
K.R. Briffa,, T.J. Osborn, F.H. Schweingruber, 2004, Large-scale temperature inferences from tree rings: a review, Global and Planetary Change 40 (2004) 11 -26
Osborn., T.J., K. R. Briffa et al. , 2005, submitted to GPC http://www.cru.uea.ac.uk/~timo/papepages/osborn_summertemppatt_submit2gpc.pdf
Pstscript on Briffa  and Briffa