Stable Isotopes in Dendroclimatology: Moving Beyond ‘Potential’ - Dendroclimatology - page 164

13 C. Low assimilation rates at the start of
the growth season (with high c i and discrimination) give relatively depleted

but more strongly defined cycles in δ

13 C
values, as A increases, to the height of the growing season, discrimination drops
and

δ

13 C rises, to a seasonal maximum. The pattern is then reversed to the end
of the growing season, closing the cycle. Poussart et al. ( 2004 ) also find an asso-
ciation between enrichment (as a result of moisture stress) with El Niño/Southern
Oscillation (ENSO) events, suggesting that chronology and signal may be derived
from

δ

13 C.
Whist tropical isotope dendroclimatology is still very much in development, and
is only one of a range of tree-ring methods that are being used to exploit the tropical
tree-ring archive (e.g., Worbes 2002 ; Poussart et al. 2006 ) , in areas where the lack of
a dry season prevents annual ring formation, isotope-derived hydrological proxies
might be used to improve the paleoclimate picture of tropical phenomena such as
ENSO (Evans and Schrag 2004 ) . To date, replication and an indication of common
signal strength has been lacking in most tropical isotope dendroclimatology stud-
ies. However, more rigorous analyses of signal strength, replication requirements,
and quantitative climate calibrations are beginning to appear in the tropical isotope
dendroclimatology literature (e.g., Anchukaitis et al. 2008b ) .

δ

13 C to Rising CO 2 Concentrations

6.5.4 Long-Term Response of δ

13 C series often show a decline, particularly over recent
decades, for which there is no evidence of a climatic cause (e.g., Treydte et al. 2001 ;
Gagen et al. 2007 ) . The decline is often site and species specific (e.g., Liu et al.
2007 ) . The atmospheric concentration of CO 2 ( c a ) is an important component of
the stable carbon isotope fractionation model;

Atmospheric corrected δ

13 C reflects the control of c i relative
to changes in c a , over longer timescales. Thus, rising c a has the potential to affect
tree-ring

δ

13 C, and an increasing body of evidence supports this conclusion. If the
effects of changing c a are becoming an additional source of noise in tree-ring

δ

13 C,

δ

this will become a problem for proxy climate calibration.
The simplest way to correct for rising c a is to simply add a correction factor based
on average tree response. However, it is not logical to assume that all trees, at all
sites, will respond in the same way to rising c a , and there is evidence that response
is highly individual (Waterhouse et al. 2004 ) . Simply selecting different correction
factors from the literature is ill-advised, as it seems to result in very different time
series. Loader et al. ( 2007 ) apply correction factors from Feng and Epstein ( 1995 ) ,
Kürschner ( 1996 ) , and Treydte et al. ( 2001 ) toa

13 C dataset of Pinus sylvestris
from northern Finland and derive significantly different corrected series as a result.
A reconstruction based on their different correction factors would produce either
cooling or warming, depending upon which factor was selected; this is clearly a sub-
jective and inappropriate method. It may be more appropriate to correct c i according
to how it changes through time, on a tree-by-tree basis, as c a rises.
Whilst changes in

δ

13 C derived from shifts in c a must be removed prior to cali-
bration, they do reveal important information about tree response to rising CO 2 that

δ