Stable Isotopes in Dendroclimatology: Moving Beyond ‘Potential’ - Dendroclimatology - page 152

tree (approximately 1-1.5‰; Leavitt and Long 1985 ) becomes relatively insignifi-
cant compared to between-tree variability (typically 2‰ or greater; McCarroll and
Pawellek 1998 ) , and there may be less to be gained from measuring multiple radii
(McCarroll and Loader 2004 ) . Any resulting 'spare' analytical capacity may be
better used to increase replication.
Where the aim is to obtain a record at annual resolution, cores can be cut into
either whole-ring or latewood increments. Investigation of the variability of isotope
ratios within rings suggests that, for some conifers, the whole ring may provide the
best signal (Helle and Schleser 2004 ) . For deciduous species it may be necessary
to separate the latewood, since earlywood is formed using reserves from previ-
ous years. After the rings have been cut, they can either be analyzed separately,
or pooled together to reduce the cost and effort involved in chemical treatment and
mass spectrometry. Where annual resolution is not required, groups of rings can
be pooled together from several trees. Pooling material prior to analysis is attrac-
tive because it is much easier and cheaper, but there is a high cost in terms of the
quality of the resulting data. A pooled record rapidly provides a single time series,
but if tree rings are analyzed independently, then the usual array of statistical tech-
niques can be used to test the signal strength, check that replication is adequate
throughout the record, and to produce confidence limits around reconstructed cli-
matic series. The common signal in isotope series is often much stronger than in
ring widths or densities from the same trees (e.g., Robertson et al. 1997a , b; Gagen
et al. 2004 ) , so as few as four or five trees can provide a representative mean value.
Given that high-frequency information is so readily available from ring width and
density measurements, there is an argument for using multiple-year pooling to focus
on capturing the lower-frequency signal that is so problematic to retain in traditional
tree-ring proxies.
In trees that do not form annual rings, sampling multiple points through a year
of growth allows identification of seasonal cycles in stable isotopes that can be used
for both chronology and climatic signal. Cores, or a radial cut with a ~1 cm 2 cross
section, are sampled at regular intervals (of a few tens of microns), by using a scalpel
or a microtome. A sampling strategy should be designed to capture the seasonal
variability needed for a chronology; 4-6 samples per annual cycle are suggested
and these should be combined with radiocarbon or 'known' dates (Poussart et al.
2004 ) . In order to detect interannual cycles in

18 O in Costa Rican cloud forest
species, Anchukaitis et al. ( 2008b ) subsampled cores, combining ten 20

δ

μ m slices

in each extraction, to give a sampling resolution of 200 m, which captured the

seasonal cycle. Trees that grow consistently throughout the year with relatively little
sensitivity in growth rate are most useful in such studies because changes in ring
width affect the capture of the annual signal if regular sampling increments are
used. Thus, in opposition to the rules of traditional dendroclimatology, a constant
growth rate is an advantage in tropical isotope dendroclimatology.
The wood component normally selected for isotopic analysis is

μ

-cellulose; how-
ever, there is increasing evidence that whole wood might be used for

α

18 O
analysis (Verheyden et al. 2005 ; Cullen and Grierson 2006 ) . In conifers, resins have

13 Cor

δ δ