Pooled versus separate measurements of tree-ring stable isotopes
Research Highlights
► Pooled δ13C and δ18O chronologies are expected to be similar to the mean. ► Empirical pooled chronologies δ13C and δ18O and the mean show a high synchronicity. ► Pooled chronologies differ especially when comparing δ18O for decadal variation. ► Imprecision in sample preparation as cellulose inhomogeneity may introduce a bias.
Introduction
Among the characteristics of wood that may be correlated with environmental parameters, stable isotopes in tree rings are valuable sources for studies on past climate and other proxy-based reconstructions (e.g., Danis et al., 2006, Gagen et al., 2007, Treydte et al., 2009).
Apart from changes in the source value (δ13C of atmospheric CO2, δ18O of soil water), the variability of isotope records from tree rings is closely dependent on the impact of environmental changes on plant physiological processes, mainly photosynthesis and transpiration (e.g., McCarroll and Loader, 2004). Furthermore, tree-ring isotope records are affected by the intra-annual growth dynamics, as well as post-photosynthetic processes related to seasonal changes in carbon allocation or oxygen isotope exchange between sugar and xylem/phloem water prior to cellulose formation (e.g., Helle and Schleser, 2004). Due to the continuous formation of wood during the vegetation period, information about the plant ecophysiological processes is integrated over time and stable isotope analyses become very attractive as a tool for ecophysiological studies, e.g., investigations on the dynamics of intrinsic water-use efficiency of trees (Saurer et al., 2004, Seibt et al., 2008).
Besides the positive attributes of tree-ring series such as the annual resolution, the accurate dating and the wide geographical distribution of trees; stable isotopes in tree rings offer additional advantages. One of the most relevant is that stable isotope chronologies, after a short “juvenile increase”, do not seem to contain any long-term age related trends and, therefore, do not require any statistical detrending (McCarroll et al., 2009, McCarroll and Loader, 2004), and therefore low-frequency fluctuations are fully preserved and the problem of the “segment length curse” is avoided (Cook et al., 1995, Gagen et al., 2007). This statement is nowadays being questioned (Esper et al., 2010) but no agreement has been reached yet. Compared to ring width or wood density, individual stable isotope series display a stronger common signal between trees which reduces the effective sample size needed to have a representative site signal. In general, the amount of samples needed to build a reliable isotope chronology is much lower than for any other tree-ring parameter and usually 4 to 5 trees are enough (Gagen et al., 2004, Gagen et al., 2007, Leavitt, 2010, Leavitt and Long, 1984, Treydte et al., 2001).
To date, annually resolved stable isotope chronologies are built in three ways. The first method follows the standard procedure of tree-ring width chronology development by measuring individual tree-rings and creating a mean site value (Gagen et al., 2006, Gagen et al., 2007, Treydte et al., 2001). The second strategy is to pool the tree rings of one year from a chosen number of trees prior to the chemical extraction of cellulose and isotope mass spectrometry determination, obtaining one single representative annual value (Leavitt and Long, 1984, McCarroll and Pawellek, 1998, Treydte et al., 2001, Treydte et al., 2006, Treydte et al., 2009). The third approach involves serial pooling of shifted tree-ring blocks for building of isotope chronologies which produces a quasi-annual time series (Boettger and Friedrich, 2009). Although annual and quasi-annual approaches may yield similar information about tree response to environmental dynamics, in this study we focus on comparing the first two methods, which rely on annual dissection of individual tree-rings.
The pooling of all material of a particular year to a composite sample before chemical treatment and isotopic analysis is normally done regardless of the mass contribution of each tree ring. This produces a mass-weighted mean chronology which is, theoretically, close to the ring-width weighted mean (Borella et al., 1998, Leavitt, 2008, Treydte et al., 2001). However, it may be different compared to averaging the δ13C and/or δ18O values from individual series. The pooling techniques always suffer from unequal mass contribution of the samples to the pool and therefore to the mean isotopic value. If the contribution to the pool would be of equal masses of wood or cellulose from each tree, the result should not be different to the average of the individual series (unweighted mean). Furthermore, in standard dendrochronological procedures, an equal contribution of every tree to the final chronology is ensured by indexing the individual series before averaging them (Cook and Kairiuskstis, 1990). When working with isotope series, high variability among the mean values of the individual series composing a chronology would bias the result towards the samples with the highest mean isotopic values.
Generally, studies using pooled samples instead of individual tree-ring analysis are more common because of savings in resources and time, especially when the number of samples is very high due to a long time-span or high replications. Cellulose extraction is relatively time-consuming and is still a non-automatized process that needs to be done individually for every sample. Pooling may produce representative series, which is an advantage when there are time and cost constraints. Among the disadvantages is the impossibility to identify any possible non-climatic (e.g., age-related) trends in the isotopic series and the loss of any information related to inter tree variability. Therefore, it prevents calculation of statistics that ensure the reliability and site representativeness of the resulting stable isotope chronology (e.g.: expressed population signal, EPS; Wigley et al., 1984) as well as to establish confidence intervals around the mean.
Only few studies have tested the representativeness of pooled isotope series in comparison to the weighted and unweighted mean of individual series and most of them were performed using δ13C. Borella et al. (1998) compared weighted and unweighted means of two sites and concluded that in general terms, there is no need of mixing the same amount of wood material, but care should be taken if a period with a high growth variation is detected in any tree. Treydte et al. (2001) showed that pooling of multiple cores display similar results to those obtained from individual series, despite significant differences in some years.
This paper goes one step further and compares pooled chronologies of two stable isotopes (δ13C and δ18O) from two sites at the Iberian Peninsula with the individual stable isotope tree-ring series of the same individuals that compose the pool. Our comparison is purely made on real values rather than on a combination of empirical and simulated data, and therefore, promises to shed more light on the crucial question to what extent pooling is comparable to the average of individual measurements. The test includes a comparison of the pooled chronologies (Pool) with the mass-calibrated mean chronologies (MassC), defined as the theoretical result of a pool if the isotopic value of every single tree and its contribution in mass to the pool is considered. It is also compared to the arithmetic average of the single series (Mean).
Section snippets
Site and sampling
During the European project ISONET, we extracted wood cores (5 mm diameter) of Pinus uncinata Ramond ex DC. in Lam. et DC at the Sierra de Cadí-Pedraforca in the Pyrenees and Pinus nigra Arn. spp. salzmanii (Dunal) at Sierra de Cazorla, Segura y las Villas during summer of 2003. The sites were selected because of their long-lived forests placed in climatic ecotones, where individuals are expected to be highly sensitive to climate. Specifically, Parque Natural de Sierra de Cazorla, Segura y las
Individual trees
The δ13C mean values of individual trees vary within a total range of 1.23 and 1.27‰ at NCZ and UPF respectively (Table 1). Stable oxygen isotopes series at UPF likewise differ from the mean by 1.28‰, while at NCZ differences among the means of the individual trees were higher (3.46‰). This is basically due to the fact that one tree, (NCZδ18O027) displays a considerable low mean value (31.12‰) in contrast to the means of the other three trees, varying within a range of 1.66‰ (32.92‰, 34.58‰,
Discussion
Similar to other tree-ring proxies, stable isotopes require adequate replication to ensure data reliability by leading to a representative site chronology with reduced genetic and site specific influences (Fritts, 1976). For this purpose, the first step is to assess and evaluate the common signal for all trees. In our case, most of the individual series means and SD show small differences between trees, especially in δ13C values. They also exhibit similar dispersion of the data, but some
Conclusions
This study showed that there is good agreement between mean master series created by averaging raw individual values (Mean) or generating a mass-calibrated mean (MassC). Coherence in the inter-annual up and downward variations are consistent among the different types of master series of the two sampling locations and isotope records. Moreover, variations in the magnitude that lead to some local incongruities in the inter-decadal variations are small in most of the cases.
Only UPFδ18OPool
Acknowledgments
We thank Marc Fillot and Peter Nyfeler for their contribution to this work, Christoph Küppers for isotope analysis and Björn Günther for performing the density measurements. This research was funded by EU projects ISONET (contract EV K2-2001-00237) and MILLENNIUM (017008–2).
References (50)
- et al.
Reconstruction of past precipitation δ18O using tree-ring cellulose δ18O and δ13C: a calibration study near Lac d'Annecy, France
Earth Planet Sci Lett
(2006) - et al.
Do tree ring δ13C series from Pinus sylvestris in northern Fennoscandia contain long-term non-climatic trends?
Chem Geol
(2008) Tree-ring isotopic pooling without regard to mass: no difference from averaging δ13C values of each tree
Chem Geol
(2008)Tree-ring C–H–O isotope variability and sampling
Sci Total Environ
(2010)- et al.
A modified method for the batch processing of small wholewood samples to α-cellulose
Chem Geol (Isot Geosci)
(1997) - et al.
Stable isotopes in tree rings
Quatern Sci Rev
(2004) - et al.
Correction of tree ring stable carbon isotope chronologies for changes in the carbon dioxide content of the atmosphere
Geochim Cosmochim Acta
(2009) Cross-dating of tree ring δ18O and δ13C time series
Chem Geol
(2008)- et al.
Impact of climate and CO2 on a millennium-long tree-ring carbon isotope record
Geochim Cosmochim Acta
(2009) Statistical evidence for exchange of oxygen isotopes in holocellulose during long-term storage
Chem Geol
(2008)
Climatic significance of tree-ring width and δ13C in a Spanish pine forest network
Tellus Ser B
The use of serial pooling method of shifted tree-ring blocks for the construction of long tree ring isotope chronologies with annual resolution
Isot Environ Health Stud
Wood cellulose preparation methods and mass spectrometric analyses of δ13C, δ18O and non-exchangeable δ2H values in cellulose, sugar and starch: an inter laboratory comparison
Anal Chem
Reducing uncertainties in δ13C analysis of tree rings: pooling, milling, and cellulose extraction
J Geophys Res
Methods of dendrochronology: applications in the environmental sciences
International Institute for Applied Systems Analysis
The ‘segment length curse’ in long tree-ring chronology development for palaeoclimatic studies
Holocene
Beitrag zur rationalisierung eines dendrochronologischen verfahrens und zur analyse seiner aussagesicherheit
Forstwiss. Cent.bl.
Stable isotope constraints on Holocene carbon cycle changes from an Antarctic ice core
Nature
Technical advances in the radiodensitometrical determination of wood density
Dendrochronologia
Tree rings and climate
The effect of site conditions on pine tree ring width, density and δ13C series
Arct Antarct Alpine Res
Combining tree ring width, density and stable carbon isotope series to enhance the climate signal in tree rings: an example from the French Alps
Clim Change
Exorcising the ‘segment length curse’: summer temperature reconstruction since AD 1640 using non-detrended stable carbon isotope ratios from pine trees in northern Finland
Holocene
Correlations between the 13C content of primary and secondary plant products in different cell compartments and that in decomposing basidiomycetes
Plant Physiol
Cited by (54)
Tree ring oxygen isotope in Asia
2024, Global and Planetary ChangeThe what, how, why, and when of dendrochemistry: (paleo)environmental information from the chemical analysis of tree rings
2021, Science of the Total Environment