Stable Isotopes in Dendroclimatology: Moving Beyond ‘Potential’ - Dendroclimatology - page 155

6.3.2 Data Treatment of Stable Isotope Time Series

One of the great advantages of stable isotopes over the traditional growth proxies is
that there appear to be no long-term age-related trends in stable isotope time series.
The evidence for this conclusion is good for carbon (e.g., Gagen et al. 2007 ) , but
less certain for oxygen and hydrogen. If confirmed, this conclusion would mean
that statistical detrending is not required, and the attendant problems of loss of low-
frequency climate signals can be avoided. The 'segment length curse' (Cook et al.
1995 ) , therefore, would not apply.
Stable carbon isotope series may require some treatment, to remove a 'juve-
nile effect' and to deal with the effect of changing atmospheric composition since
industrialization.
The juvenile effect, seen as depleted but rising values for the first few decades
of a tree's life (Fig. 6.2 ) , varies with site and species, but magnitudes of ~2‰ are
common over 20-40 years (e.g., Bert et al. 1997 ; Duquesnay et al. 1998 ; McCarroll
and Pawellek 2001 ; Anderson et al. 2002 ; Arneth et al. 2002 ; Gagen et al. 2004 ;
Raffalli-Delerce et al. 2004 ) . In Fig. 6.2 , the averaged series reveal the dramatic
juvenile trend in

13 C after atmospheric correction), which lasts
for different periods of time at the two sites. After a few decades the series no longer
show sharply rising

13 C corr (tree-ring

δ δ

13 C corr values but do continue to show variability. However,
crucially, there is no long-term trend pattern after the juvenile period. Over very
long timescales, the presence or absence of long-term trends in

δ

13 C has not yet
been assessed, but there is good evidence for the lack of age-dependent trends in the
post-juvenile period.
Theories on the cause of the juvenile trend include recycling of respired CO 2 ,
changes in hydraulic conductivity with height gain, and changes in light levels
(Schleser and Jayasekera 1985 ; Heaton 1999 ; McDowell et al. 2002 ) . The effects
can simply be removed by linear detrending (e.g., Wilson and Grinsted 1977 ) or
by using methods such as regional curve standardization (Briffa et al. 1992 , 1996 ;
Gagen et al. 2008 ) , but the simplest solution is to remove the juvenile portion from
each tree prior to analysis (e.g., Tans and Mook 1980 ; Arneth et al. 2002 ) . It is
advisable, at any new site, to analyze a few trees to pith, in order to correctly gauge
the length of the juvenile period for a particular species and site.
Because fossil fuel-derived CO 2 is depleted in 13 C (Freyer and Belacy 1983 ) ,
there is a nonlinear decline in the

δ

13 C atm ) since about
AD 1850. Fractionation is additive (Farquhar et al. 1982 ) , so any change in the
isotopic ratio of CO 2 is incorporated into the isotopic ratio of the cellulose produced
within the tree. Thus, all tree-ring

13 C of atmospheric CO 2 (

δ δ

13 C series, whether or not they display any trend
over the industrial period, must be corrected for changes in the isotopic ratio of
atmospheric CO 2 . Since annual values of

δ

13 C atm are available (e.g., McCarroll and

δ

13 C values can be corrected to a preindustrial (AD 1850)

Loader 2004 ) , tree-ring δ

13 C cor ) by using:

standard value of − 6.4‰ ( δ

13 C cor = δ 13 C plant − 13 C atm +

δ ( δ 6.4) (6.4)