Evaluation of the mechanism of variability of the delta18O in alpha-cellulose of tropical deciduous tree pinus kesiya: conventional models and new thinking

PDF

Download

Open document

Flip pages

Contact Us

Copy asset link

Request this asset

Transcript (if available)

Content EVALUATION OF THE MECHANISM OF VARIABILITY OF THE δ
18
O IN α
CELLULOSE OF TROPICAL DECIDUOUS TREE PINUS KESIYA:
CONVENTIONAL MODELS AND NEW THINKING
by
Andrés C. Martinez
A Thesis Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements of the Degree
MASTER OF SCIENCE
(EARTH SCIENCE)
December 2007
Copyright 2007 Andres C. Martinez
ii
Dedication
This Manuscript is dedicated to Ruben and Maria Martinez
iii
Acknowledgements
I would like to acknowledge the tremendous amount of mentorship, support and
friendship I have received from: Lowell D. Stott, Douglas Hammond, David Bottjer,
Brendan Buckley, Robert Douglas, Cynthia Waite, Angela Cornell, Reetta Saikku,
Maximus Berkelhammer, Vardui TerSermonian, and the Department of Earth Science.
Above all I would like to thank Miguel Rincon for 6 years of guidance and friendship.
iv
Table of Contents
Dedication ii
Acknowledgements iii
List of Tables vi
List of Figures vii
Abstract viii
Chapter 1: Introduction 1
1a: History and Interest 1
1b: Isotope Method 4
Chapter 2: Hypothesis and Theory 10
2a: Hypothesis 10
2b: Tree Cellulose Model 17
2c: A Tropical Tree Model 19
2d: Isotope Theory 20
Chapter 3: Methods 24
3a: Sampling: Chiang Dao, Northern Thailand 25
3b: Microsampling Technique 26
3c: Alphacellulose preparation 27
3d: Isotopic Analysis: Pyrolysis, Elemental Analysis and
Isotoperatio Mass Spectrometry 29
Chapter 4: Results 30
4a: Cellulose Extraction 30
4b: Pinus kesiya 32
4bi: Growth Model 1 36
4bii: Growth Model 2 40
4biii: Growth Model 3 41
4c: Lower Frequency Isotopic Variability 41
Chapter 5: Discussion 43
5a: Seasonal Features, Color vs. Cellulose δ
18
O 43
5b: Rainfall Amount and Cellulose δ
18
O 47
5c: Interannual Variability 57
5ci: Influence of ENSO 57
5cii: Pacific Decadal Oscillation 60
5ciii: Southern Oscillation Index 60
5civ: Indian Ocean Dipole 61
5cv: Asian Monsoon Strength 62
v
5d: Mechanistic Theory 64
5e: Previously Published Data 66
5f: Final Proxy Reconstruction 71
Chapter 6: Conclusion 71
References: 75
vi
List of Tables
Table 1: History of stable oxygen isotope dendroclimatology 5
Table 2: History of Methods 8
Table 3: Duplicate Analysis 33
vii
List of Figures
Figure 1: Seasonal precipitation vs. latitude 12
Figure 2: Qualitative isotopic schematic 14
Figure 3: Isotopic Circulation patterns 16
Figure 4: Chiang Dao, Thailand 26
Figure 5:
13
C NMR Spectra of cellulose 31
Figure 6: δ
18
O record through time 34
Figure 7: Annual (averaged) δ
18
O αcellulose vs. color; with example
of a random annual tree ring. 35
Figure 8: Environmental Data (annual averages) 38
Figure 9: Possible growth patterns (cartoon) 40
Figure 10: δ
18
O vs. time with a 5 point smooth 42
Figure 11: δ
18
O and amount of rain at Chiang Dao vs. Time (20001987) 44
Figure 12: Bangkok δ
18
O with rainfall amount through time 45
Figure 13: Average Amount of Rainfall and δ
18
O Over a Year 46
Figure 14: δ
18
O and rainfall at Chiang Dao vs. time (197060) 50
Figure 15: Variance of Isotopic Record at 10 Year Intervals 51
Figure 16: Annual Rainfall vs. Annual Isotope Values 52
Figure 17: Rainfall vs. δ
18
O regression 53
Figure 18: δ
18
O with Color Designation through time 56
Figure 19: Interannual Environmental Oscillations with δ
18
O from Chiang Dao 59
Figure 20: Schematic of the pressure gradient created by trees 69
Figure 21: Linear rainfall proxy reconstruction and comparison to the
Instrumental record 72
viii
ABSTRACT
The relationship between environmental variables and the oxygen isotope
composition of αcellulose in ringed tropical trees may not be constrained to simple
“amount effects.” Here I present a 55year continuous record of δ
18
O of αcellulose
from Pinus kesiya. By comparing the cellulose isotope record to previous data, model
theory, and rain amount data it is clear that both amount effects as well as isotopic
changes in atmospheric circulation regimes are incorporated in the oxygen isotopic ratio
of the alpha cellulose. Regression analysis show most of the variability recorded in the
δ
18
O of cellulose results from changes in total annual rainfall, and large anomalies arise
during La Nina and El Nino events. A proxy rainfall reconstruction exhibits poor
correlation with rainfall variability and will require further work in order to constrain
how best to extract a quantitative estimate of rainfall.
1
Chapter 1: Introduction
1a. History and Interest
Efforts to extend the history of tropical climate variability beyond the short
instrumental record have been hampered by the lack of sufficient high resolution
terrestrial climate archives (Evans and Schrag 2004; Poussart et al 2004; Poussart and
Schrag 2005). Marine sediment cores, corals, and ice cores fall short of providing a
continuous, annually resolved history of atmospheric temperatures and precipitation.
However, treering αcellulose may be able to provide this information for the last
several thousand years (McCaroll and Loader 2004; Libby et al 1976). Climate
variability has traditionally been reconstructed from tree ring chronologies. In some
trees annual ring widths and densities are sensitive to changes in temperature and
precipitation. Tree ring records have been developed throughout most of the Holocene
(McCaroll and Loader 2004). Recently, climate signals have also been detected in the
stable oxygen, carbon and hydrogen isotopes of the raw wood, holocellulose and alpha
cellulose (Libby 1976; DeNiro 1981; Yapp and Epstein 1982; Buhay 1985; Roden
1999; Evans and Schrag 2004). These observations are promising for several reasons:
tree ring cellulose is abundant, easy to obtain, simple to work with and, in many cases,
available continuously for several hundreds of years (McCaroll 2004). Long lived trees
(>100 years) inhabit every continent, excluding Antarctica. Furthermore, many cultures
have incorporated large old growth trees into their dwellings, thereby preserving the
wood for many generations. The stable isotopic fractionation preserved in the trees
wood (including cellulose) results from several factors: the condensation of the heavy
and light isotopes of oxygen in water due to atmospheric processes (e.g. precipitation,
2
humidity and temperature; Dansgaard; 1964) and the original source of the atmospheric
precipitation. Most commonly this fractionation is variable as a function of the amount
of rainfall. As atmospheric water vapor pressure reaches a saturation point, the heavy
isotopes will condense more readily than lighter isotopes and, therefore, the resulting
precipitation will have an enrichment of H
2
18
O relative to the vapor. This phenomenon
is termed the “amount effect” by stable isotope geochemists. Changes in the isotopic
ratio as a function of the source reservoir have been described in earlier studies
(Dansgaard 1964).
Climatic information is transferred to trees via the cambial tissue in the form of
cellulose, but the process is potentially complicated by physiologic factors.
Physiological factors also influence the δ
18
O of cellulose synthesized by trees and these
will be discussed below in the Cellulose Model Chapter. Nonetheless, with current
sampling and analytical techniques it is now possible to produce a radial oxygen isotope
stratigraphy of tree cellulose at subannual resolution (Evans and Schrag 2004; Poussart
et al 2004; Poussart and Schrag 2005; Brendel 2000; McCaroll and Loader 2004; Tsuji
et al 2006). On long lived trees this offers a possibility of obtaining a continuous or near
continuous climatology from the oxygen isotope variability recorded by the tree.
At the present time there are a few terrestrial tropical paleoclimate records
scattered throughout Southeast Asia, Indonesia, Eastern China and the tropical Pacific
(IAEAGNIP, NOAA database, www.ngdc.noaa.gov/paleo/). Only a few of these are
records of oxygen isotope values from tree wood and cellulose (McCaroll 2004).
Further more, less than seven records showing δ
18
O of αcellulose ratios at subannual
3
resolution have been published (Roden et al 2000; Evans and Schrag 2004; Poussart et
al 2004; Poussart and Schrag 2005; Tsuji et al 2006).
Traditionally, treebased climatologies focus on locations within groves where
the trees are under an environmental or ecological stress. The reason for this method is
that in a stressed environment (water stress, light stress, or temperature stress) the trees
will exhibit the greatest growthrelated response to environmental change. In order to
utilize the growth behavior expressed by trees in these environments it is essential to
understand the physiological differentiation that takes place during early and late wood
syntheses and how environmental factors (such as temperature, and humidity) influence
growth. Precise observational width measurements are fed into statistical models and
collated into a single site average (McCaroll 2004; Robertson 2006). Creating a robust
record therefore requires nearly a dozen “well behaved” trees, a well constrained age
model, a detailed knowledge of absent and false rings (McCaroll 2004; Robertson 2006)
and a strong understanding of the current regional climate regime.
The oxygen isotope chemistry of tree cellulose may provide complementary or
unique information that cannot be obtained from tree ring width and density analyses
alone. Several indigenous tropical species from locations where there is little
temperature variability but large changes in rainfall have been found to have nonannual
rings or completely absent ring structures, which makes traditional statistical tree ring
width analysis impossible. In several tropical trees which do possess annual ring
structures the visible changes in wood color that characterize the ring structure of many
extratropical species are not observed. In many annual rings (taken from tropical trees)
intraannual transition periods from early to the late wood in a given growing season are
4
ambiguous. It can therefore be difficult to use annual ring structures in tropical trees to
differentiate boundaries between early and late wood, and therefore to obtain
temperature and/or precipitation information from qualitative methods (McCaroll
2004). Further, recent work by Briffa et al (2004) suggests that ring widths are
becoming decreasingly sensitive to climate (temperature and precipitation) with
increasing CO
2
levels, due to stress and availability of CO
2
at the leaf/atmosphere
interface. Therefore there is not only a need for more physically constrained methods
but also methods that are insensitive to the properties which are causing problems with
previous ring width data methods.
1b. Isotope Method
From the time Dansgaard first identified the physical connection between stable
isotope fractionation and climatic parameters, there have been efforts to apply oxygen
and hydrogen isotope geochemistry to tree wood and cellulose (Dansgaard 1964; Libby
et al 1976). The first studies were focused almost entirely in northern Europe due to the
relatively high availability of long instrumental records and the previous knowledge of
the physiology of the indigenous tree species which produce “well behaved annual ring
structures” (Libby et al 1976; Thompson and Gray 1977). Initial processing methods
called upon crude chemical techniques which were originally developed by the paper
and pulp industry for mass refining raw wood to form suitable material for commercial
use; however, these methods were not developed for rigorous analytical precision and
were not sensitive to stable isotope exchanges (Green; 1963). They were unfinished,
tedious, used tremendous amounts of reagents per sample that were often unnecessarily
5
hazardous (Green; 1963). These factors resulted in relatively few isotopic studies and
very few conclusive results. The climatic information inferred from the isotope records
in these earlier studies was mostly attributed to temperature influence on the δ
18
O of
tree wood at locations in Switzerland and England. There was no effort to
systematically relate the tree δ
18
O of wood to precipitation or humidity.
Of the studies that have been carried out, few have been able to reach agreement
as to how climate information is expressed in the oxygen isotopes of wood. Nor is there
agreement about how to handle differences in procedure, species, ecology, materials
and growth models. Table 1 shows the oxygen isotope dendroclimatology literature
base: experiments with many different species, wood types, chemical methodologies
and many different observed results. Several patterns are obvious between latitude,
species, materials, and methods.
Table 1: History of stable oxygen isotope dendroclimatology.
Author Species Location Methods Factors affecting variability
Anderson
1998/2002
Abies alba Central
Switzerland
Online
IRMS,
Primary temperature signal,
secondary Precipitation and
Relative humidity
Buhay and
Edwards 1995
Quercus p. Ontario Canada Offline,
bulk
cellulose
Signal composed of
precipitation and Relative
humidity.
Burk and Stuiver
1981
Maple North America Bulk
cellulose
Temperature
Danis P.A, et al.
2006
Quercus robur France Alpha
cellulose
Precipitation
Evans and Schrag
2004
Pinus strobes,
prosopis sp.
Thailand Alpha
cellulose
Precipitation, relative
humidity
Gray and
Thompson 1976/7
Picea glauca Canada Bulk
Cellulose
Precipitation/ Groundwater
Hemming et al.
1998
Pinus sylv. England Alpha
Cellulose
Relative humidity.
Tsuji H, et al
2006
Quercus
crispula
Northern Japan Alpha
cellulose
(no ref.)
Precipitation amount and
Humidity.
Libby et al 1976 Cryyptomeria
japonica
German,
Northern Japan
Whole raw
wood
Temperature
Lipp et al. 1996 Tamarix
jordanis
Italy Wood
cellulose
Precipitation
6
Table 1: continued
Poussart, Evans
and Schrag 2003
Podocarpus Indonesia and
Thailand
Alpha
cellulose
Precipitation
Poussart and
Schrag 2005
Pinus Merkusii Thailand Alpha
cellulose
Precipitation, relative
humidity
Raffallidelerce
G. et al 2004
Quercus robur Western France Alpha
cellulose
Precipitation
Ramesh et al.
1985/1986
Abies pindrow India Raw wood
(no ref)
Precipitation and relative
humidity
Rebetez, Saurer,
Cherubini 2003
Abies Alba Central
Switzerland
Raw
latewood
Temperature
Robertson et al
2001
Qurecus
petraea
East England Late wood
alpha
cellulose
Precipitation> Relative
humidity> temperature.
Roden and
Ehleringer 2000
Populus
fremontii
(cottonwood)
Southwest
USA
Alpha
cellulose
Precipitation and relative
humidity.
Saurer et al.
1998/1997/2002
Picea glauca,
pinus
Central
Switzerland
Alpha
cellulose
Ground water
Switzer et al.
1994/1996
Quercus robur East England Early and
late wood
Temperature and relative
humidity.
Verheyden A., et
al. 2004
Rhizophora
mucronata
(mangrove)
Gazi bay,
Kenya
Alpha wood Annual cycle, possible
salinity
Waterhouse J.S.,
wt al 2002
Qurecus robur Norfolk, UK Latewood
alpha
cellulose.
Summer precipitation.
Weiguo, et al
2004
Pinus
tabulaeformis
(Chinese pine)
Northwest
China
Alpha
cellulose
Precipitation, monsoon
activity
Table 1: A brief history of the literature of stable oxygen isotope analysis on tree ring material. Species,
location, material analyzed and interpreted results are listed. Bulk cellulose indicates 14 (beta)glucan
polymers of nonregular distribution. Alpha cellulose refers to 14(beta)glucan of between 10,000 and
15,000 DOP (Green 1963).
It is clear that at the higher latitudes (>45˚N), temperature seems to be the dominant
variable recorded. At lower latitudes (< 40˚N), precipitation and relative humidity
appear to be most important (Table 1). However, there are several exceptional studies
from high latitude sites that show precipitation, not temperature, as the dominant
forcing (Robertson 2001, Waterhouse 2002, Gray and Thompson 1976). From the
results summarized in Table 1, one might also conclude that the isotope/climate
relationship depends as much on the material chosen as it does on the location. Raw
wood (unprocessed woody material with all lignins, resins, polymers, sugars and oils), it
7
seems, usually leads to temperature as the interpreted result (Verheyden A 2002;
Switzer et al 1996; Rebetez, Saurer, Cherubini 2003; Libby et al 1976). When regular
cellulose (cellulose polymers of indiscriminate length) and αcellulose (cellulose of
DOP 10,00015,000) is separated from raw wood and analyzed isotopically the results
are typically shown to correlate most strongly with precipitation, changes in ground
water source and relative humidity (Evans and Schrag 2004; Poussart, Schrag and
Evans 2005, Poussart and Schrag 2006, etc).
There does not seem to be any noticeable pattern or predictability between
results from hard woods and soft woods. Because various materials and chemical
preparatory techniques have been applied to these studies it is difficult to assess whether
there is a coherent relationship between the oxygen isotope composition of various
woods (and components of wood) and a particular climate variable. Further, it is not
clear from the available data whether the isotopic composition recorded by early wood
or late wood is more sensitive to a particular climate variable. Poussart and Schrag
(2005) reported on four tree cores (of different species, from Bali, Indonesia and Chiang
Mai, Thailand) where there was no statistical correlation to either instrumental
precipitation or temperature record. These authors provided no explanation, either from
a physiological or isotopic standpoint other than to claim it as real signal and not an
instrumental artifact.
It is also evident from the literature review that there is no consistency among
chemical techniques applied to stable oxygen isotope measurements and preparation of
cellulose. Cellulose in the current literature is simply defined as a glucose polymer
linked together by 14 carbon bonds. However, there is considerable variability within
8
the simple definition. Cellulose may differ in its polymer length or degree of
polymerization (D.O.P.), regularity and purity. Chemically these forms are
distinguished as “alpha, beta, gamma, and holo.” Several methods neglect to exercise
the rigorous chemical differentiation and refer to their products as either “holo or alpha”
regardless of its true chemical nature.
Table 2: Methods Summary
Method Products
Brendel O. et al
2000
Alphacellulose, and
holocellulose:
Acid/ethanol
Modified Brendel 1
(Evans and Schrag
2004)
Alphacellulose
Modified Brendel 2
(Gaudinski et al
2005)
Alphacellulose:
extra NaOH rinse
Watermodified
Brendel (Gaudinski
et al 2005)
Alphacellulose: 3
extra water rinses.
Green 1963 Review of several
methods.
JaymeWise 1946 Holo cellulose:
benzene/ethanol
Leavitt and Danzer Holo cellulose:
benzene/methanol
Stuiver 1984 Holo cellulose
Sternberg 1989
(modified Wise)
Holo cellulose:
benzene/methanol
Leavitt and Danzer
1993
Alpha cellulose:
toluene/ethanol
Table 2: Several different popular methods for
cellulose extraction and their products.
There are several methods in common use; each differs in chemical treatment and also
in their final product (Table 2). The majority of techniques in use today are slight
modifications of the JaymeWise method that was originally developed over half a
century ago for mass production of paper pulp (Green, 1963). Procedures such as the
JaymeWise method involve a separation of the organic and aqueous phases and often
9
require use of a highly carcinogenic benzene mixture. Several techniques have been
developed more recently for oxygen isotope analysis of cellulose which recognize the
more precise chemical definition for cellulose such as that proposed by Ward (Ward
1946; Robertson et al 2001; Evans and Schrag 2004; Poussart and Schrag 2005,
Poussart et. al 2004).
Of the techniques currently being used there are three categories. First, raw
wood is considered by some workers to be the best recorder of climatic information. In
studies utilizing raw wood, little preparation is applied to the wood prior to analysis,
typically on a prepyrolysis wash with deionized water to remove any extra woody
contamination collected prior to sampling (Anderson et al 2002). Studies that have
attempted to separate cellulose from other wood components have utilized a phase
separation technique and dissolution/recrystalization method (Green 1963; Gaudinski et
al 2004). The products consist of αcellulose, βcellulose, γcellulose and holocellulose
in a random mixture. Lastly, there is a community of workers who have chosen to
isolate regular cellulose polymers (D.O.P. 10,000 15,000) from other wood
components and to analyze this discrete part of the wood. The method most often
employed in these studies involves a strong nitric acid solution (according to Ward
1946) to separate the cellulose from extra cellulosicorganic materials followed by a
series of washes to neutralize and clean the product. There is little literature to show that
one of the methods produces cellulose product superior (in purity or analytical
precision) for stable oxygen isotopic analysis (Brendel 2000; Gaudinski 2000; McCaroll
2004). In the present study I apply a modified Brendel method (Brendel 2000, Evans
and Schrag 2004) for its simplicity, safety, and efficiency (to be discussed later).
10
Chapter 2: Hypothesis and Theory
2a. Hypothesis
The hypothesis I seek to test is as follows: the ratio of
18
O to
16
O fixed in the α
cellulose of tropical tree Pinus kesiya, taken from Chiang Dao Northern Thailand is
entirely dependent on fluctuations in amount of precipitation, both at seasonal as well as
at interannual time scales. The null hypothesis is that, the oxygen isotope ratio of α
cellulose in this tree is controlled by origin of precipitation due to competing monsoonal
circulation cells shown to have distinct isotopic signatures (Aggarwal 2004). The
variability expressed through δ
18
O αcellulose is a function of changes in the amount of
rainfall both from the beginning of the rainy season to its termination as well as the
interannual variability, rather than as a result of the recognizable signals documented in
monsoon circulation cells recently by stable isotopists studying fractionation in rainfall.
In order to test this hypothesis it will be necessary to analyze a long period of
continuous core material for its δ
18
O and compare it to known instrumental precipitation
data. The data will be worked up in three ways: (1) using color identification of samples
within the ring as an age model; (2), tuning that age model using the isotopic results the
most positive isotopic value will be accepted as the beginning of the rainy season; and
(3), the data will be further tuned using the peak of the rainy season to identify the most
depleted isotopic value (these models will be explained in more depth in the results
section). The isotopic record will also be compared to the instrumental records for total
monsoonal rainfall in order to observe interannual trends.
Unfortunately there is no literature documenting the precise movements of the
major circulation cells over the Southern Asian continent the oceanic monsoon trough
11
(OMT), trade wind trough (TWT) and the continental monsoon trough (CMT) through
time, therefore the null hypothesis will be compared to known atmospheric circulation
phenomena and anomalies (for which there is data) through time. Results will be
interpreted from observations of how the isotopic record varies according to what one
might expect from the knowledge of the atmospheric circulation. For example, under
the Poussart, Evans and Schrag theory (hereafter referred to as PESt), during strong El
Nino events the rainy season isotopic excursions should be highly damped or
completely absent. Further, any “amount effect” relationship would be altered
systematically but still present. Whereas under the Asian circulation monsoonal theory
(hereafter referred to as ACMt), one may expect El Nino events to shift the normal
positions of circulations systems, yielding a different source of precipitation and a
different isotopic behavior altogether.
It may be that changes in the pressure and temperature systems which define the
Pacific Decadal Oscillation (PDO), Southern Ocean Index (SOI), El Niño Index, and
Indian Ocean Dipole (IOD) may impose influences on the atmospheric dynamics across
Southeast Asia that can alter the source of precipitation that falls on Chiang Dao. This is
because it sits at the boundary of several largescale atmospheric cells (Araguas
Araguas 1998; Aggarwal 2004). Therefore, the premise of this study is to show how
much oxygen isotope variability, expressed in the αcellulose component of wood, can
be tied to “amount effects;” any residual variability may reflect changes in atmospheric
circulation patterns or previously overlooked biological factors.
In the tropical regions, summertime precipitation is associated with the highest
rainfall (Figure 1; Peterson and Vose 1997; Evans and Schrag 2004). There are
12
Figure 1: Seasonal precipitation vs. latitude
Amplitude of the seasonal precipitation cycle as a function of latitude. This data was
taken from stations having at least 30 years of monthly observations. Taken from
Peterson and Vose (1997); Evans and Schrag (2004). The largest amplitude signals
come from the terrestrial tropics.
13
relatively small temperature fluctuations between summer and winter (Evans and
Schrag 2004). This observation has been used to argue that the oxygen isotopic value of
the precipitation is dominated by changes in rainfall amount (the “amount effect”)
rather than changes in temperature (as seen in precipitation at higher latitudes) (Roden
et al 2000; Barbour et al 2004; Evans and Schrag 2004; Tsuji et al 2006). This is indeed
what Michael Evans, Pascale Poussart and Dan Schrag proposed in a recent series of
papers (Evans and Schrag, 2004; Poussart et al, 2004; Poussart and Schrag 2005; Evans
et al 2006). In several low latitude locations in both the northern and southern
hemisphere, and also midlatitude northern hemisphere, they interpreted δ
18
O values in
αcellulose as being an expression of the isotopic fractionation arising from H
2
18
O
condensation during precipitation. Their research sites consist of Peru, Costa Rica, the
continental United States, Indonesia and Thailand, implying that the same results may
be observed over a wide range of climates from tropical rainforest to warmtemperate
areas (Evans and Schrag 2004, Poussart et al 2004, Poussart and Schrag 2005). The
greater the amount of rain the greater proportion of light isotopic H
2
16
O condensed from
atmospheric vapor. Hence, the highest rainfall amounts are associated with the lowest
δ
18
O values in acellulose. Heavy amounts of rain in this region are usually due to
moisture from walker circulation and monsoonal rain caused by the thermal temperature
gradient between the ocean and the land.
In the rainy summer months most of the tree cellulose is laid down so that an
isotope stratigraphy from a tropical tree is weighted towards the wet season. There is
some isotopic enrichment at the leafair interface due to diffusion of isotopically light
water vapor, particularly in the dry winter months (Figure 2; from Evans and Schrag,
14
Figure 2: Qualitative Isotopic Schematic
A simple qualitative model created by Evans and Schrag (2004) from their isotope data
and the model proposed by Roden et al 2000. δ
18
O αcellulose values are hypothesized
to be a function of both evapotranspirative enrichments and the amount effect through
the course of a year.
15
2004). However, the authors do not address the possibility that their 24‰ variability
could also reflect the influence of changing regional water source patterns or monsoonal
circulation cells that can vary in strength and location interannually and intraannually.
The qualitative model also implies a large seasonal bias (expressed mathematically in
the numerical model to be discussed later).
Within the stable isotope dendroclimatology community the Evans and Schrag
model is gaining acceptance. However, stable isotopists looking at and modeling the
oxygen isotope composition of precipitation over Asia, have documented isotope values
that are unique to the source of Asian monsoonal circulation cells, not from a simple
physical condensation (“amount effect”) alone as expressed in the cellulose model
(Figure 3; Aggarwal 2004; AraguasAraguas 2004). This presents a significant
complication to the effort to use the cellulose δ
18
O values as direct measure of rainfall
amount variations. It is not clear to what extent the oxygen isotope ratios in αcellulose
will reflect changes due to source water variations and to what extent they will represent
an “amount effect”.
This study will address both possibilities. After preliminary observations are
made, data will be tested against each theory. The mechanism responsible for the
variability of oxygen isotopes in precipitation and subsequently preserved in cellulose is
the key to our ability and limitations to interpret cellulose results, monsoon intensity,
and to apply these interpretations to El Niño history and to further calibrate models.
16
Figure 3: Isotopic circulation patterns
The circulation patterns over Asia and the δ
18
O signatures of the precipitation in the
Oceanic monsoon trough (OMT), Continental monsoon trough (CMT) and the trade
wind trough (TWT). Taken from Aggarwal (2004).
17
2b. Tree Cellulose Model
The pathway by which cellulose is synthesized and the possible sources of
oxygen isotopic fractionation are complex, but can be simplified by separating a tree
into its subsystems. Although the oxygen in cellulose ultimately comes from
precipitation, the δ
18
O of cellulose (or wood) is very different from the δ
18
O of
precipitation, indicating that a tree does not passively collect and store oxygen, but
rather acts to fractionate or, exchange oxygen isotopes, likely responding to
environmental factors (DeNiro, 1979; McCaroll et al 2004). It has been shown that
while oxygen is also present in CO
2
and O
2,
it is only the oxygen in H
2
O that affects the
δ
18
O of the αcellulose (DeNiro 1979; Yakir 1981).
A deciduous tree (one that loses its leaves in the dry season) may be thought of
as a combination of three communicating systems; the roots, the cambium, and the
crown. Water is introduced to the tree via the roots system. In the case of many pine
species this comes by way of a system of lateral roots and one tap root. The source of
cambial water is mostly from the lateral roots system, while the tap root primarily
provides stability (Evans and Schrag 2004; McCaroll and Loader 2004). There is no
oxygen fractionation during water uptake (Wershaw et al 1966; DeNiro 1979; Libby
1976; McCaroll 2004; Evans and Schrag 2004). There is speculation as to whether
evapotranspiration takes place within the soil creating a fractionation which is
dependent on root depth and tree maturity. However, the work done to monitor this
effect has shown depth to play a minor factor only in trees which utilize a good portion
of water from deep lateral roots and in locations of low rainfall (Dawson 1993; Dawson
and Pate 1996; Buhay and Edwards 1995; Tang and Feng 2001). Therefore, as long as
18
the taproot does not draw water from a deep underground aquifer, the water absorbed is
from atmospheric precipitation (Evans and Schrag 2004). Studies assume percolation
time is negligible (Buhay and Edwards 1995). Water is carried up the cambium via
capillary action and a pressure gradient described by:
h
p
8
4
r P
Jv
D
= (Equation 1)
Where J
v,
is the velocity of water, P is the pressure, r is the radius of the capillary tube,
and η is the relative viscosity (taken from Hopkins and Huner, 2003). The gradient is
created in large part by the evaporation of water at the leaf surface, creating positive
pressure at the roots. Water makes its way up the cambium, eventually reaching the
crown and leafs. The water in the leaf is used along with carbon from CO
2
in the
atmosphere during photosynthesis to make sucrose (glucose is not made at this step;
were it made here it would be absorbed in the branches of the crown system before
reaching the cambium), a simple sugar (DeNiro 1979; Yakir and DeNiro, 1981; Hill et
al 1995; McCaroll and Loader 2004). The isotopic composition of the sucrose reflects
that of the leaf water, with 27‰ enrichment due to exchange between organic carbonyl
oxygen and water oxygen (Sternberg 1986; DeNiro 1979; McCaroll and Loader 2004).
As sugars are carried down the trunk they come into contact with the xylem water and
undergo an exchange (Roden et al 2000; Anderson et al 2002; Barbour et. al 2004). This
occurs when the sucrose is cleaved into its monomer glucan units. These form hexose
phosphates, from which 20% of the oxygen exchanges with xylem water. To further
complicate the picture, a small portion of the hexose phosphate does not undergo
immediate conversion into cellulose, but rather breaks down to an intermediate “triose
19
phosphate,” which remains available for further exchange (Hill et al 1995; Barbour and
Farquhar 2000; Roden et al 2000; Evans and Schrag 2004; McCaroll and Loader 2004).
The second source of fractionation is physical rather then biochemical and
occurs via evapotranspirative effects at the leafatmosphere boundary. Trees are capable
of transpiring tremendous amounts of water. A large tree may transpire over 200 liters
of water an hour (Hopkins and Huner, 2003). This amount is dependent on stomatal
conductance and relative humidity (Roden et al 2000; McCaroll 2004). As water
evaporates at the leafatmosphere interfacecreating the pressure gradient isotopically
light oxygen preferentially transfers into the vapor phase, enriching the residual liquid
phase with the heavy oxygen. This may cause up to 20‰ enrichment above the source
water (Saurer et al 1998 a & b). The isotopic ratio imprinted on the tree ring is a balance
between the enrichments due to sucrose synthesis and evapotranspiration, and the
dampening by way of exchange with the xylem water.
2c. A Tropical Tree Model
Annual treerings from the northern hemisphere midlatitudes are traditionally
defined as having two visibly distinct sections; earlywood and latewood. The distinction
is based on color and density. In softwood conifers, such as pines, the earlywood is
made up of less dense large cells of light color, whereas late wood is much darker and is
composed of smaller cells more tightly packed, creating a denser wood. The change in
physiology arises due to external stresses. During peak growing season, the tree freely
produces cellulose and cambial cells. As the rains stop and the tree becomes more
stressed, less wood is produced, possibly analogous to the tree rationing its
20
photosynthate. Classically a ring can be thought of as spanning a year. However, the
boundary may not be a sharp line since trees are known to store photosynthate
(McCaroll and Loader 2004). Thus, it is possible for a tree to grow with stored
photosynthate which does not represent current source water conditions, but rather the
source water of the previous year. This usually takes place at the earliest parts of the
growing season or when the tree is most stressed (McCaroll and Loader 2004). This
traditional ring identification system has been applied to pines for a long time.
However, tropical pine trees have always presented dendrologists with a
problem. Trees which experience high amounts of rainfall annually do not lay down
earlywood and latewood with a clearly defined boundary. Rather there is a “gradient”
between early and latewood, due to the relatively small amount of stress placed on the
tree annually. This makes tropical growth models based on cellulose color and density
difficult to construct. It is visually unclear how much of an annual ring was synthesized
during the peak growing season and how much formed during the terminal end of the
rainy season or during the dry season from stored photosynthate. Studies have shown
tropical tree growth to be highly variable, with some tropical trees experiencing 90% of
their growth during the peak two weeks of the rainy season, while others grow equally
through virtually the entire rainy season (Bullock et al 2003). The only way to get
precise measurement of cambial growth through time is through direct attachment of
dendrographs to the trees in question. For this reason, this study has adopted a new
method for describing a sample position in the ring, based on the relative color of the
sample. This method will allow for a consistent analysis for all rings regardless of
thickness or abnormal growth patterns.
21
2d. Isotope Theory
Due to the difficulty in precisely measuring the absolute isotopic composition of
16
O to
18
O in cellulose, ratios are presented in δ (delta), notation in relation to a standard
or, Δ, in relation to source material. All values here are reported in relation to Vienna
Standard Mean Oceanic Water (vSMOW). Isotopic compositions are expressed as:
1000 1
standard
× ÷
ø
ö
ç
è
æ
- =
R
Rsample
d (Equation 2)
Where “R
sample
” and “R
standard
” are the
18
O/
16
O ratios of the given material and the
standard respectively. Using this nomenclature, the first simple model connecting the
enrichment in an open water system to fractionation that takes place during diffusion
and the liquidvapor phase change was developed by Craig and Gordon (Craig and
Gordon 1965). Years later, the model was extrapolated to describe leaf boundary effects
and diffusion of H
2
O through stomata:
ú
û
ù
ê
ë
é
÷
ø
ö
ç
è
æ
+ ÷
ø
ö
ç
è
æ -
+ ÷
ø
ö
ç
è
æ -
=
i
a
a
i
a s
wx kb
i
e i
wx k
e
e
e
e e
e
e e
R R R * Rwl a a a (Equation 3)
where R
wl
is the
18
O/
16
O ratio in the leaf water, R
wx
is the
18
O/
16
O ratio in the xylem
water, R
a
is the
18
O/
16
O ratio in the atmosphere, α* is the liquidgas fractionation
factor, α
k
is the kinetic fractionation factor (associated with diffusion through air), α
kb
is
the fractionation associated with diffusion through the atmosphereleaf boundary layer,
e
i
is the intercellular vapor pressure, e
s
is the water vapor pressure at the leaf surface,
and e
a
is water vapor pressure in the atmosphere (Craig and Gordon 1965; Farquhar et
al 1989; Flanagan et al 1991; Roden et al 2000). This equation simply states that the
18
O/
16
O ratio in leaf water is a function of the sum of the products of involved pressures
22
with their respective fractionation factors, analogous to an isotopic mass balance.
However, this description of leaf water enrichment is confusing as written with respect
to an isotopic standard. Recently, Barbour et al (2004) and Evans et al (2004) have
expressed the same leaf water enrichment model, in terms of the source water ratio
(Barbour et al 2002; Barbour et al 2004; Evans and Schrag 2004, Evans et al 2006):
( ) ( )
÷
÷
ø
ö
ç
ç
è
æ -
þ
ý
ü
î
í
ì
-
ú
û
ù
ê
ë
é
- D + + + = D
-
r
a a a
r
e
e
e
i
s
k wva k wl
1
1 1 * 1 (Equation 4)
Where αk and α* are the temperature dependent kinetic and equilibrium fractionation
factors, and Δwva is the isotopic composition of atmospheric water vapor. This model
also accounts for the so called “backdiffusion” or mixing of unevaporated stem water
where, “ρ” is the dimensionless Peclet number defined as:
cD
lE
= r (Equation 5)
Where l is the path length, E is the evaporation rate (g/cm
2
sec), c is the density of
water, and D is the diffusivity of H
2
18
O in water (this model is available for download
from ftp://ecophys.biology.utah.edu/tree_ring/). An understanding of isotopic behavior
in the leaf and xylem has allowed calibration of models designed to evaluate the oxygen
isotope composition of αcellulose from regions with recorded climate data. The most
popular method to model oxygen isotopes in tree ring cellulose was developed by John
Roden (Roden 1999; Roden et al 2000; Evans and Schrag 2004):
δ
18
O
cx
= f
o
(δ
18
O
wx
+ ε
o
) + (1 – f
o
) (δ
18
O
wl
+ ε
o
) (Equation 6)
Where δ
18
O
cx
, is the oxygen isotope composition of the treering cellulose, δ
18
O
wx
is the
isotopic composition of the xylem water, δ
18
O
wl
is the isotopic composition of the leaf
23
water, ε
o
is the isotopic fractionation factor for the enzymatic assimilation of oxygen
into cellulose and f
o
is a dimensionless constant which estimates the proportion of
exchange between cellulose and leaf water. ε
o
has been estimated to be + 27‰
(Sternberg 1989; Yakir and DeNiro 1983; Roden 2000) and “f
o
” has been estimated to
be ~0.42 (Roden et al 2000). This model, however, is a balance between two terms; the
rainy season “amount effect” depletion and a winter enrichment driven by
evapotranspiration. The model is calibrated around data sets where the variability is
thought to follow the qualitative PESt model (Figure 2) which suggests a heavy
seasonal bias, and therefore, this seasonal bias is implied in the model. This
relationship, (equation 6) is expressed qualitatively in figure 2. In this conceptual model
there is a significant bias toward the amount of precipitation received through the
effects of precipitation and evaporation and it does not account for shifts in storm
source water. The δ
18
O of tropical precipitation is explained simply as the “amount
effect” whereby precipitation preferentially rains out the heavier isotope, leaving further
precipitation isotopically depleted (Gat, 1996; Evans and Schrag, 2004). Therefore as
the rainy season continues the δ
18
O values of water taken up by the roots becomes
increasingly enriched in
16
O. Along with high humidity (which reduces leaf water
evaporation), the “f
o
(δ
18
O
wx
+ ε
o
)” term of equation 6 becomes the dominant term. As
the wet season gives way to the dry season, the humidity decreases, increasing the rate
of evaporation at the leafair interface, and thereby increasing the preferential removal
of the light isotope, leaving the subsequent leaf water enriched in the heavy isotope and
allowing the second term in equation 6 to the be the dominant forcing. These
relationships have been used to empirically develop dendochronometers, as well as
24
indications of the strength of El Niño events and total precipitation in areas dominated
by the Asian monsoon. Although little has been done to calibrate the model against
records with high isotopic variability, these results rely on the assumption that δ
18
O of
precipitation is driven largely by seasonally controlled factors and is consistent between
species and sample sites which may be too broad an umbrella to work under.
Chapter 3: Methods
The record necessary to test my hypothesis must be sampled at subannual
resolution to observe intraannual trends and structure. This is an accomplishment
which has only recently become possible because of the ability to sample and
isotopically analyze small, subannual cellulose samples. It is now possible to precisely
analyze samples of <0.3mg of cellulose on an isotope ratio mass spectrometer.
Traditional wood core sampling via a hand razor often yields only a single sample per
ring; at best two samples per ring, one from earlywood and one from the late wood.
Using a microtome it is possible to obtain between 4 and 12 samples per tropical tree
ring (depending on the thickness of the ring and the thickness of the core taken).
Methods used in this study (modified Brendel method; Brendel et al. 2000; Evans et al.
2004) are identical to those used in other studies which report data at subannual
resolution; therefore results may be treated as equivalent, and conclusions may be
attributed solely to the mechanistic pathway of fractionation, rather than differences in
analytical or chemical pathways.
For this experiment it is was necessary to develop an effective method to sample
material in a way that would identify the subannual color characteristics of the tree
25
wood in order to analyze results in portions of trees which are typically avoided by
tropical dendroclimatoligists due to their unusual intraring density behavior.
3a. Sampling: Chiang Dao, Northern Thailand (19˚37’ N, 98˚58’ E).
Pinus kesiya (living) were identified and sampled from a 2 km
2
grove in Chiang
Dao (Figure 4) by Brendan Buckley of Lamont Doherty Tree Ring Laboratory in the
summer of 2001. Chiang Dao is located in the northern portion of Thailand
(approximately 60 miles north of Chiang Mai, a site of previous oxygen isotope
cellulose work by Poussart and Schrag 2004), and experiences a complex annual
weather cycle dominated by the Asian monsoons. The monsoon manifests itself in three
distinct seasons: a wet hot season from May to September, a cool dry winter from
October to February, and warm spring from February to May (Poussart, Evans, and
Schrag 2004). The monthly rainfall varies throughout the year from around 1800mm in
the peak of the summer rains to 0mm in the dry winter (January). Mean temperatures
also show a smooth transition between the seasons from around 23˚C in the winter to
~28˚C in the summer. Extracted core material extended through the entire life of the
tree 20001929 (known from observing tree pith from cores). Pinus Krempfii was
chosen in part because it lays down annual ring structures. All collected tree cores
showed clear annual ring structures. However, the color/density gradients within rings
make the late/early wood distinction (optically) problematic.
26
Figure 4: Chiang Dao, Thailand
Chiang Dao, Thailand (19.22N, 98.58E). Approximately 60 kilometers north the Chiang
Mai (Modified from: http://geography.about.com/library/blank/thailand.jpg).
3b. Microsampling technique
Samples were extracted from the tree via a 16 inch Hagloft 5mm increment
borer. Tree cores were glued into wood sample channels for ring age model analysis.
Brendan Buckley of Lamont Doherty Tree Ring Laboratory, using ring width data from
a suite of cores within the same grove, provided an age model spanning the entire length
of the core, accurate to one year. Core DCK 06a was selected for isotopic analysis due
to its well annually occurring ring structures; showing no evidence of false or missing
rings. Upon receiving the core sample it was removed from the wooden channel. The
raw core material was placed in an Optical American 820 Rotary Microtome positioned
under a compound microscope so that the level of the microtome blade was in focus.
Microtome slices were taken at consecutive 30 μm intervals, using Surgipath high
profile Teflon coated surgical microtome blades. Blades were repositioned every 60
27
samples to avoid problems from dull blades (dull blades do not yield uniform slices of
wood). Samples are labeled according to their relative “color.” Three categories are
defined: “W
o
” is the lightest color wood (the wood of least density) in the entire annual
ring, appearing immediately after the latewood from the previous growing season. “W
1

W
n
” is described as the gradient wood which is also light in color but becomes
progressively less “light” (and denser) as it approaches the late wood. This is necessary
as there is an unclear latewoodearlywood boundary. “D” is designated as the darkest
wood in the ring and the clear late wood (also the densest wood in the ring). The
boundary between the latewood and the beginning of the early wood for the following
season is usually clear and abrupt. However, due to the nonplanar tree ring boundary
between “D” and “W” there is a small amount of overlap between these samples. This
overlap is designated at “M” or a sample mixed with both dark wood from the previous
year and white wood from the following year. Every three slices are combined into a
single sample for processing. Interval size, dependent entirely on the minimum weight
required for analysis, is calculated by the inequality: %Y*TW≥0.25μg; (where
“TW”=total preextracted weight, “%Y” = expected yield), and the thickness of the
microtome slices (slices thinner then 30 μm did not produce consistent sample sizes).
3c. Alphacellulose preparation
Unfortunately the term “alpha cellulose” (αcellulose) is not well defined in the
stable isotope dendrology literature. In the present study αcellulose is considered to be
defined chemically as by Ward (1946): a polymer of glucan (glucose) monomer units
beta linked at the 14 oxygencarbon, usually with degree of polymerization (D.O.P.)
28
between 10 and 15 x10
3
. Extraction methods for extracting this glucan monomer
involve separating the αcellulose from other lignins, cyclic esters, epoxides, resins and
cellulose chains with a significantly lower D.O.P. (< 10x10
3
) all of which may have
potentially different biochemical synthesis factors influencing their δ
18
O fractionation
and exchange (McCaroll 2004, Evans 2004). There are several published methods for
purification. For this study we use a modified Brendel method (Brendel 2000, Evans
2004) which allows for rapid sample processing without using carcinogenic materials.
Raw wood material is placed into clear 2 ml microcentrifuge tubes. To the samples a
mixture of concentrated nitric acid and glacial acetic acid is added (1:10). The samples
are capped and heated in a sand bath at 130ºC for 30 minutes. Samples are then
vortexed for 3 minutes. 400μl of pure ethanol is added to quench the reaction. Solution
is removed and 200μl of DI water is added. Samples are centrifuged for 30 minutes.
Supernatant is removed, 400μl of pure ethanol is added, and samples are centrifuged for
15 minutes. Supernatant is removed and the sample is washed and centrifuged a final
time with acetone to remove any final organics. Acetone is removed and samples are
placed in a 70˚C oven for 12 hours. Final yields range from 45% to 65%. These values
are consistent both with the fraction of alphacellulose in whole wood and the results of
Brendel (2000) and Evans et al. (2004). The finished product is observed to be a white
fibrous, “cottonlike” mass with no odor. This product is also consistent with previous
experiments (Brendel et al, 2000; Evans and Schrag, 2004).
29
3d. Isotopic analysis: pyrolysis, elemental analysis and isotoperatio mass
spectrometry.
Once dry, 200 –250 µg of pure αcellulose was placed in Costech silver capsules
(5mm x 9mm) and compressed with midrange forceps to remove any atmosphere
(comparisons of oxygen fractionations to beam height show minimal analytical
fractionation between 69 na. However, samples which do not produce beam heights in
the optimum range tend to show enrichment). Silver capsules are loaded into an auto
sampler capable of 50 consecutive analyses per run. The samples are pyrolysied in a
Carlo Erba Elemental Analyzer (EA) at 1050˚C in a quartz reaction column packed with
a combination of nickel wool and carbonnickel wool over quartz turnings. The
pyrolysis is thought to proceed as described by Shafizadeh (Shafizadeh 1973). At
temperatures above 1000 ˚C, it is unlikely kinetic effects cause a significant change to
the chemistry. The major products of the pyrolysis are dihydrogen gas and carbon
monoxide. The carbon monoxide is measured on a continuous flow MicroMass Isotope
Ratio Mass Spectrometer (coupled to the EA). The system is constantly purged with
ultrahigh purity helium and samples are measured relative to an ultrahigh purity
carbon monoxide gas. Sample measurements are adjusted relative to an interlaboratory
standard of granular cellulose power synthesized by J.T. Baker and calibrated by Nick
Evans at the Arizona Tree Ring Laboratory and the, Schrag laboratory at Harvard, as
28.62 permil vSMOW (Evans and Schrag 2004). Reproducibility of the mass
spectrometer on the Baker standard is 0.3 permil. MassLynx version 3.6i software was
used to compute δ
18
O ratios.
30
Chapter 4: Results
4a. Cellulose extraction
Work carried out by Gaudinski (2002) showed several
13
C nuclear magnetic
resonance spectra (NMR) of αcellulose products produced from different chemical
processing methods (Figure 5): JaymeWise, modified JaymeWise, Brendel, modified
Brendel and “Watermodified” Brendel (identical to the “modified” Brendel Procedure
with an added terminal water wash) (Gaudinski et al 2002). The authors concluded, for
the purposes of δ
18
O analysis, the Brendel method may not remove all long chain
alkanes and possibly celluloseacetate byproducts. However, NMR results show no
noticeable “shifts” (peaks) where long chain or short chain alkanes (or alkenes) bearing
carbonyl, alcohol or ester groups would be expected; ~0.8ppm, ~5.0ppm (Pavia et al
2001; p 122). Further distinguishing between long chain polymers of similar chemical
composition is not a simple feat and is almost always accompanied with other analytical
methods. Their proposed solution was an extra sodium hydroxide wash (NaOH, prior to
the first water wash). Although, with no oxygen bearing long chain alkanes present,
there is nothing with which cellulose oxygen can exchange. Further, the authors
suggested the appearance of the product of the JaymeWise method to be a cottonous,
fibrous product. However, this may in fact be a premature conclusion; there is a bleach
step carried out in the first section of this method which may guarantee a white fibrous
product, regardless of chemical composition. The formation of celluloseacetate is most
likely negligible due the steric and torsional hindrance acetate groups would cause;
furthermore, such a product would be washed away in the terminal acetone wash.
31
Figure 5:
13
C NMR Specrtra of Cellulose Products
13
C NMR analysis of cellulose prepared under several different methods (No
18
O
analysis is available because of nuclear spin). Modified Brendel method appears nearly
identical to the cellulose standard. Taken from Gaudinski et al (2002).
32
Several individual samples were chosen at random intervals throughout the core
and divided into duplicate samples for a test of reproducibility of chemical and
analytical techniques. These duplicate samples were processed identically to all other
samples. Reproducibility of the duplicate pine samples is well within the internal
reproducibility of the interlaboratory standard and within the analytical error of the
mass spectrometer. Table 3 shows several duplicate samples as well as their respective
δ
18
O value and the difference between δ
18
O values of the duplicates. On average it
seems the duplicate samples are within 0.20‰ of each other.
4b. Pinus kesiya
δ
18
O analysis of the αcellulose from core DCK 06a yielded a 55 year
continuous record (from 1945 through 2000) showing both subannual and subdecadal
variability. Figure 6 shows the raw data though time using the color designated
sampling system and the age model developed by Brendan Buckley. Variations
throughout the record are between 27‰ and 16‰ (this magnitude is consistent with
other subannual studies, Evans and Schrag 2004; Poussart et al 2004; Poussart and
Schrag 2005; Tsuji et al 2006).
Although there is up to 10‰ variability over the span of the entire record, within
the same “color,” all samples exhibit similar standard deviation when averaged over the
entire record. The average subannual variability is on the order of 2‰ (Figure 7a).
Isotopic variability of this amount agrees with the variability
33
Table 3: Duplicate Analysis
Sample
Location
δ
18
O
αcellulose (‰)
Standard
Deviation
1946 W2 a) 22.51
b) 22.58
0.049
1947 W1 a) 19.82
b) 19.87
0.035
1947 W4 a) 25.62
b) 25.41 0.148
1948 D1 a) 25.28
b) 25.42
0.099
19491948 M1 a) 22.73
b) 22.85
0.084
1951 W2 a) 21.01
b) 21.15
0.099
1957 W3 a) 21.93
b) 21.87
0.042
1960 W3 a) 22.41
b) 22.66
0.17
1967 W5 a) 17.92
b) 18.01
0.063
1971 W1 a) 21.82
b) 22.01
0.134
19721971 M1 a) 23.96
b) 23.79
0.120
1974 W0 a) 22.39
b) 22.28
0.077
1978 W1 a) 22.62
b) 22.45
0.120
Table 3: Duplicate analysis, including location/color designation within the annual tree
ring, (DCK 06a) and the standard deviation.
34
Figure 6: δ
18
O record thru time
DCK 06a δ
18
O (vSMOW) of αcellulose from Thailand tree Pinus kesiya vs. time. Raw
data, unprocessed. The data is continuous from 1945 through 2000. Data is sampled at
subannual resolution with the number of samples per ring in proportion to the thickness
of the ring.
35
Figure 7: Annual (averaged) δ
18
O αcellulose vs. Color; with an example of a
random annual tree ring.
B)
a) δ
18
O (αcellulose) from Pinus kesiya data binned according to color and averaged
over the entire length of the core plotted through one annual ring. This plot shows the
seasonal signal through the tree’s growing season. (b) A picture of an annual ring from
a Pinus kesiya tree (picture is not from core DCK 06a and did not provide the data in
figure 6a) arrows indicate the position of the samples in an annual ring as well as typical
ring behavior, with an ambiguous early/latewood boundary.
observed intrannually in published records from tropical locations; Poussart et al.
(2004) and Evans and Schrag (2004).
36
Ring thicknesses are highly variable from year to year; two consecutive rings
rarely yield the same number of samples. For example, years with very thins rings yield
only 3 or four samples, while very thick rings yield as many as 16 samples. To collect
average δ
18
O values over the entire span of the record the data was binned into equal
numbers of samples per year in order to normalize this “ringthickness” bias. Each year
is described by seven “W” samples “W
0
” indicating the sample of lightest color, the
beginning of the year and onset of the rainy season; followed by six consecutive white
(these samples may appear darker then W
0
) samples, “W
1
W
6
” describing the growth of
the tree through the rainy season; a single “D” sample designated by the darkest densest
portion of the ring (due to the ambiguity of the early/late wood transition naming more
than a single dark sample per ring is problematic); and a single “M” sample indicating
the transition from one growing season to the next. Figure 7b, shows a common
example of a ring selected at random from a Pinus kesiya tree (not from core DCK 06a).
Lines are drawn from figure 7a indicating by example where in the stratigraphy of the
ring each sample represents.
4bi. Growth Model 1
Here, a traditional dendrocalendar definition of a year is adopted. A single
cycle from earlywood through the adjacent latewood is considered a single year or
growing season. The early wood is produced from the onset of the rainy season through
the summer months and into early fall. When rain amount decreases in the late fall and
winter months the late wood is synthesized as the tree becomes increasingly more
stressed.
37
The “M” sample represents the boundary between samples of both “dark” wood
and white wood. However, this does not make clear whether this “mixed” sample is
more closely associated with the previous year or the following rainy season. I define
the “M” sample as the final sample in a year. This suggestion is supported by the
observation that the average “M” value is very close to that of the nearest “dark” value
relative to the following “W
0
” value (Figure 7a), much closer then the differences
between consecutive gradient samples, the difference between gradient and dark and the
difference between mix and white samples. This suggests a tree may call upon stored
photosynthate created from water from the earlier part of the year to create the first
layers of cellulose for the next growth season (McCaroll and Loader 2004).
The distribution in figure 7a also shows the “W
0
” sample is most similar to the
next adjacent white samples (rather then the “mix” sample). Due to the presence of rain
(on average) throughout the year, results have been interpreted for complete years, with
little to no discontinuity resulting from absence of water for long periods of time
(normally not longer then one month).
The annual record of precipitation is used as a primary input to of the intra
annual growth model (Thailand meteorological society). The amount of precipitation at
Chiang Dao reaches a peak (on average) around July and August (Figure 8a). Figures
8b and 8c show the corresponding average annual humidity and average annual
temperature distributions. Precipitation exhibits far more variability than does humidity
or temperature. Both humidity and temperature exhibit seasonal cycles. The humidity
data, however, does not exhibit the same seasonal behavior as precipitation with its
lowest point in March and its highest point only a few months after in August. This
38
Figure 8: Environmental data (annual averages)
A)
B)
C)
a) Binned average of precipitation amount (in mm) as a function of month. b) Binned
average of relative humidity data (%) as a function of month. c) Binned temperature
data (˚C) as a function of the month for an entire year. (All data is from the Thailand
Meteorological Society)
39
seasonal behavior does not match the behavior of the isotopic annual average (figure
8a). Furthermore, the relationship between humidity and leaf water enrichment
described in Barbour (2004) suggests the surface area of the leaf plays a large role in the
affect humidity may have on the final δ
18
O composition of the resulting cellulose. Pinus
kesiya, however, has long slender “needles” with relatively small surface areas as its
leaves. This suggests that the evapotranspirative enrichment due to fluctuations in
humidity would not significantly alter the δ
18
O of the αcellulose.
There are three possible patterns of tree growth that would be anticipated for a
seasonal cycle in this location. In figure 9 the red, blue and green lines depict, how
growth may occur over the course of a calendar year. The red line indicates most of the
growth occurring during the middle of the year, the blue line shows a bias toward a
larger percent of the trees total growth taking place between spring and early fall when
Chiang Dao usually receives most of its rainfall. The green line shows the most growth
occurring near the end of the rainy season as the amount of rain decreases. A
comparison of annual average rainfall amount data and the annual average δ
18
O of
cellulose would suggest that the largest depletion are in the middle of the calendar year
associated with the peak of the rainy season, which suggests the majority of cambial
growth may take place during the rainy season. There are four “white” samples between
the most enriched “W
1
” and the most depleted “W
4
” sample (these white samples
represent a binned average of all data in order to normalize the number of samples per
year) indicating a slight shift toward the first half of the year. The precipitation data is
recorded monthly; continuously from the present through 1980 (precipitation amount
data was provided by Brendan Buckley and the Thailand meteorological society).
40
Figure 9: Possible Growth Patterns.
The three possible patterns which a tree can grow the red line shows the majority of
growth occurring at the earliest part of the rainy season. The Green line shows a linear
growth pattern and the Blue line shows most of the growth occurring at the terminal
portion of the rainy season.
Therefore the error in the intraannual age model is most likely <2 months. Without
species specific dendrometer measurements this interpretive schematic is the best that
can be done with this age model.
4bii. Data Treatment: Model 2
However, there are other treatments which can be applied to the raw age model
(referred to as “Model 1”) which apply different definitions to the growth throughout
the year in contrast to those used by classic dendrology from visual observations. For
example it may be appropriate to tune the cellulose δ
18
O record to the relationship one
41
would expect from an “amount effect” Using model 1 as a starting point the beginning
of each year is reassigned from its “W
0
” value to the nearest most isotopically enriched
value. This new “start of the year value” is on average ± 2 samples from the designated
“W
0
” sample. This creates an assumption; the beginning of a year may not always be
represented exactly as the onset of earlywood. This age model is designated the
“isotopic age model” or “model 2.”
4biii. Data Treatment: Model 3
Finally, this isotopic age model is further tuned to the instrumental rainfall data
available at Chiang Dao. Assuming an amount effect relationship such as that described
by PESt, the most depleted isotopic value in a year is assigned to the month with the
most rainfall (i.e. the peak of the raining season). This model is referred to as the
“rainfall model” or “model 3.” For each of these models the El Nino and La Nina years
are treated separately. Further each model is tuned within a single year (± 2 samples
from the “W
0
”) therefore the long term interannual behavior is unchanged and will be
shown as the color model from figure 6 rather then three separate data treatment
records.
4c. Lower frequency isotopic variability
A 5point smoothing function is applied to the raw data (Figure 6) and the
record is replotted through time (Figure 10) in order to better observe lower frequency
patterns. Lower frequency isotopic variability is evident at subdecadal time scales
(Figure 10). These isotopic fluctuations tend to range between 1‰ to 9‰. Several of
42
Figure 10: δ
18
O vs. time with a 5 point smooth.
δ
18
O αcellulose from DCK 06a presented with a 5point smooth filter from
KaleidaGraph.
43
these sustained isotopic ‘shifts’ are evident in the early part of the record (from 1945
1970). There seems to be periods of relatively higher and relatively lower isotopic
values lasting between 5 and 10 years. From 194649, 195462 the values are on more
enriched on average. Between 194952 and 196368 the values are relatively more
depleted. Further it seems that the interannual variability in the signal becomes
dampened at 19762000. Although the Chiang Dao cellulose isotope record exhibits
extended, subdecadal length enrichments and relative depletions, there is no clear
longerterm (i.e. late 20
th
century trend) evident (Figure 10). There are also two time
periods of instrumental rainfall data (amount of rainfall from; 20001994, and 1993
1987) that, when compared to the αcellulose data appear both in sync (20001994) and
out of sync (19931987) with the PESt intraannual/seasonal model (Figures11 a and b).
There are no δ
18
O values available for Chiang Dao precipitation. However, δ
18
O
of precipitation has been documented for Bangkok (13° 50' N., 100° 29' E). In this data
the δ
18
O of rainwater increases with decreasing rainfall (Figure 12) showing the
presence of the “amount effect” at a site south of Chiang Dao, which experiences
significantly less annual rainfall but is likely influenced by the same atmospheric
circulation sources. This observation suggests the amount effect is the dominant mode
of variability at a site analogous to Chiang Dao.
Chapter 5: Discussion
5a. Seasonal features, Color versus Cellulose δ
18
O
The environmental changes that take place throughout the year have an
influence on cellulose growth rates and hence, the density/color contrasts, are associated
44
Figure 11: δ
18
O and amount of rain at Chiang Dao Thailand vs. Time (20001987)
a)
b)
a) DCK O6a δ
18
O (αcellulose) with amount of precipitation at Chiang Dao from 2000
through 1994. b) DCK O6a δ
18
O (αcellulose) with amount of precipitation at Chiang
Dao from 1993 through 1987.
45
Figure 12: Bangkok δ
18
O with rainfall amount
Shows the δ
18
O of precipitation at Bangkok and the amount of precipitation at Bangkok
vs. time from 19682000 (From Lowell Stott unpublished; data from IAEAGNIP
database).
with a discernable shifts in the cellulosic δ
18
O values as well. This is most evident
when the samples are binned into their respective color groupings and plotted over a
year as in figure 7a. The growing season samples (W
0
W
6
are on average 1 to 1.5‰
lower in δ
18
O compared to the “Mixed” and “Dark” samples. And while the interannual
variance in δ
18
O within each of color groups is only on the order of ± 2‰ , the
correspondence between color and the δ
18
O of cellulose is quite clear and is taken to be
a reflection of the close relationship between growth behavior and the amount of
rainfall throughout the year. This relationship can be illustrated as in figure 13, which
46
Figure 13: Average amount of rainfall and δ
18
O against one year
This figure plots the annual average precipitation (amount mm) with the average δ
18
O
(αcellulose), binned annually by “color” vs. one year.
shows the binned average δ
18
O values (from figure 9a) over the course of a year plotted
with the average annual rainfall amount. The highest δ
18
O values correspond to the
months of lowest rainfall. This implies that the white samples, which consistently
exhibit the lowest δ
18
O values, are formed during the period of increasing rainfall
amount. I infer that rainfall amount is the primary influence on the seasonal growth
behavior of this tree species. The correspondence between the timing of the depletion in
47
the cellulose and the evolution of the rainy season make a strong case for PESt’s
“amount effect” argument during the summer months. The enrichment during the latter
half of the annual cycle corresponds to the supposed “evapotranspirative” condensation
effects as the rain subsides during the later part of the year. The total variability
attributable to the “amount effect” over the length of the record is on the order of 6‰.
However, it is possible that the “binning” method applied to figure 7a dampens some
critical portions of the entire signal or gives an incomplete picture of the individual ring
structure. Data was binned attempting to avoid biasing the δ
18
O average record toward
years with more samples versus years with only a few samples (according to their
respective widths).
5b. Rainfall Amount and Cellulose δ
18
O
An analysis of discrete time periods shows further details in the annual picture.
From 1987 to 2000 (Figure 11a and 11b) continuous monthly rainfall was recorded at
Chiang Dao. While the resolution of this rainfall data is not as high as from other time
periods, when the δ
18
O of cellulose is plotted on top of precipitation amount for 2000
1994 (figure 11a), there is an unmistakable anticorrelation; drier periods of the year are
represented by higher δ
18
O values whereas wetter time periods correspond to lower
cellulose δ
18
O values. The magnitude of the intraannual variability is on the order of
2‰. While the qualitative schematic representation looks reasonable, the magnitude of
the changes in precipitation amount between years does not match the magnitude of the
corresponding δ
18
O. It is unclear whether this is normal behavior, as previously
published data sets did not compare δ
18
O to subannual levels of rain on a year to year
48
basis (Evans and Schrag 2004; Poussart et al 2004; Poussart and Schrag 2005). For this
period of time 0mm rainfall seems to be represented by ~23.5‰.
On the other hand, depletion which occurs during the rainy season stays
somewhat consistent from year to year, regardless of how much total rain fall was
received. In other words, there appears to be a δ
18
O minimum for cellulose. The values
are never lower than ~22 to 21.5‰. Further, the 1996 year of relatively normal rainfall
yielded a shift almost twice that of any other year through this span. The year associated
with the most rainfall, 1995, shows an average δ
18
O value typical of the rainy season.
Further, the rainy season depletion becomes consecutively greater through the end of
the period, independent of rainfall amount. Through this time there is clearly a seasonal
signal, however, there is also a significant amount of variability that appears to result
from another mechanism.
The previous portion of the record, 19931987 (figure 11b) tells a completely
different story. There is still an oscillation which, qualitatively, appears to behave
seasonally. However, it is reversed with respect to precipitation compared to the
previous period (Figure 11a). In this picture is seems that the middle of the white
section corresponds to the dry season. The scale of variability is similar to the previous
period. Further more, in this period there is less correlation between enriched and
depleted δ
18
O values hovering around an average value. This behavior may be the result
of a deviation from the accepted growth model if, in fact, the onset of growth began
significantly earlier or later. In this case the δ
18
O would appear to have been shifted.
Also it is possible that the expected “amount effect” behavior here has been completely
overshadowed by a reversal in the source of the moisture. Unfortunately the monthly
49
precipitation data terminates in 1987. During the period from 19871980 the data still
seems to follow a seasonal pattern (possibly still in phase with rainfall) although several
years in this interval (19821984) produced relatively thin rings and therefore few
samples per ring. The inphase observation from the latest part of the twentyfirst
century suggests the model 2 or model 3 may be better suited for interpretation of the
results. Therefore it is possible that the there is a mode of isotopic variability related to
interannual growth anomalies.
The data for the years 19761980 follow a seasonal cycle (Figure 7). However
the beginning and end of each cycle does not precisely correspond to the “dark” and
“white” sample of every calendar year (although they are always within a sample before
or after). The cyclic pattern is interrupted in 1976 by a 7‰ enrichment carrying though
1973.
One of the most striking features of the data set is the shift in δ
18
O values
between 1962 and 1963; this can be observed in figure 13 which spans 19701960.
During this time period the “M” values from 1962 through the white samples of 1963
record an isotopic shift towards lower values (from ~27‰ to ~18‰); the depleted
values stabilize and continue throughout much of the following decade (Figure 13).
This isotopic anomaly occurs close to an anomalous rain event that occurred in October
of 1963 when Chiang Dao received 5092 mm of rainfall; the highest amount ever
recorded. However, the isotopic depletion precedes the increase in rain amount. Perhaps
this suggests a change in circulation from CMT to OMT prior to the rain event. This
would initially deplete δ
18
O values (regardless of the amount of rain received), as the
CMT is shows more enriched values (Figure 4; Aggarwal 2004).
50
Figure 14: δ
18
O and Rainfall amount at Chiang Dao vs. time (19701960)
δ
18
O (cellulose) against amount of precipitation at Chiang Dao from 19701960
Statistical analysis shows the remainder of the record exhibits annual behavior
with a high average variance. The raw αcellulose data was averaged at 10 year
intervals (i.e. 19501959, 19601969 etc), the variance of the data was calculated and
plotted through time (Figure 15). It is clear that the variance has decreased through
time, implying that perhaps the monsoon variability has also been lower. This could
possibly indicate the weakening of the Asian monsoon, which has been speculated
recently by several authors (Hu et al 2000; Wang and Lau 2003). This may be caused
by a weakening of the oceanic monsoon trough and the continental monsoon trough
51
Figure 15: Variance of isotopic record at 10 year intervals
Variance of δ
18
O (αcellulose) through time. 10 years of data was collected and variance
was taken.
which provide a source of moisture for Chiang Dao during a normal (nonEl Niño or La
Niña) year (AraguasAraguas 1998; Aggarwal 2004).
The hypothesis may also be tested by simply comparing the total amount of
rainfall occurring during a given year to the total isotopic variability (expressed as the
range between the lowest and highest δ
18
O values for a give year) experienced in that
year (Figure 16). Although the “amount effect” does not manifest itself in a linear way,
distillation effects should be obvious if that is what is controlling the isotopic
composition at Chiang Dao. There are no striking correlations between the records and
there is no regression relationship.
52
Figure 16: Annual Rainfall and Annual Isotope values vs. time
Total annual rainfall amount at Chiang Dao vs. Annual range of isotope values from
DCK 06a. The large gap (1975) indicates absence of data.
Finally, a regression of growth model 1, δ
18
O αcellulose versus rainfall amount
is shown in figure 17a. El Nino and La Nina years are shown as separate regressions on
the same plot. Although it has been shown that empirically the seasonal rainfall pattern
seems to resemble the annual isotopic pattern the regression reveals little relationship
between the amount of rainfall and the δ
18
O of the cellulose. The presence of a visual
trend in the regression is arguable but the correlation coefficient is only 0.32. Further it
shows equally large scatter in La Niña and El Niño years.
53
Figure 17: Rainfall vs. δ
18
O regression
a) b)
Regression of growth model 1 (a), growth model 2 (b) and growth model 3 (c). Red data
represents normal rain year. Green data shows La Nina years, and blue data show El
Niño years. Data shown is only for the times which instrumental data at Chiang Dao is
available.
54
To attempt to search the record for the variability associated with the amount
effect, growth models 2 and 3 were also regressed against rainfall amount (these models
are tuned to show the presence of the amount relationship intraannually and on
interannual timescales are virtually identical to model 1). The regression using growth
model 2 is shown in figure 17b again with El Niño years and La Niña years in separate
colors. The difference from the growth model 1 regression is striking. There is a
moderate connection between the amount of rainfall during a normal year (El Niño or
La Niña year) and the δ
18
O of cellulose (r
2
= ~0.60). Furthermore, La Niña years seem
to be described by a more depleted and less robust relationship. During El Niño years
there is little if any relationship between cellulose δ
18
O and the amount of precipitation.
Finally the growth model 3 is regressed against the amount of rainfall at Chiang
Dao (figure 17c). This plot shows a clear dominant amount effect during nonEl Niño
years. The r
2
value for normal rain years is 0.75. A clear correlation between amount
and δ
18
O for La Niña years now emerges, again more depleted then a normal year but
still with an r
2
value of ~0.6. El Niño years still remain anomalous showing at best an
extremely weak relationship. However, El Niño values are also more depleted than both
normal rain years and La Niña events. Figure 17c shows strong evidence that there is a
dominant mode of variability associated with the amount effect imbedded in the
cellulose record during normal rain years and that there is a slightly altered relationship
during la Niña years. Further, it shows there still may be second order variability
concealed in the signal as well significantly different mechanisms acting on the source
of precipitation during El Niño years.
55
While the growth model 3 regression shows the total amount of theoretical
variability attributable to an “amount effect,” to be significantly greater then 2‰, they
are influenced by a strong seasonal bias. Therefore, a different mechanism may explain
the remaining variability.
Previous studies conducted at lower resolution, have also suggested that early or
latewood is selectively coupled to climate (usually specifically to summer or winter
temperature or precipitation) (Libby et al 1976; Thompson 1977; Switzer). Observing
the annual difference between dark, white and gradient it is clear that on average, one
needs all three sections to capture the total variability; this is best demonstrated by
Figure 18 which shows the raw data with its “color” designation. It is clear both from
figure 9a and 18 how the behavior of the isotopes may fluctuate between “colors;” this
is further made clear through events such as the 19621963 depletion which is large
enough that it is highly unlikely to be an artifact of tree biology. The entire length of the
tree ring records a signal which may be linked to the environment rather than only the
earlywood or latewood portions. It is likely the variability during El Niño years may be
a consequence of changes atmospheric circulation resulting from changes in the source
precipitation at Chiang Dao between the CMT and TWT. Therefore, it is reasonable to
examine the climate phenomena and atmospheric circulation changes through time with
a focus on circulation regimes during El Niño and La Niña events.
56
Figure 18: δ
18
O DCK 06a with color designation through time
The raw data from DCK 06a core with markers indicating the values and positions of
the “colors” over the entire core. The red line represents the raw data. Blue diamonds
indicate the position of “white” samples, the green circles indicate the position of the
“dark” samples and the black squares represent the locations of the “mix” samples. The
space between every blue diamond and green circle represents the growing season (i.e.
W
1
W
n
).
57
5c. Interannual variability
The fivepoint smoothing of the raw data (Figure 10) shows four distinct
isotopic “regimes” or intervals. Working from the most recent back in time; 2000 to
1971 the record oscillates around 22.5‰. Within this time interval there are several
large enrichments and depletions, each persisting for a few years. During the 1960’s the
variance increases significantly and the average values become depleted to around
18‰. From 19641953 mean values are enriched again; exhibiting a decadal trend.
During 1952 through 1945 shows a distinct trend from depleted values to enriched
values occurs. However, values at the beginning of the record are similar to those at the
end implying no long term lowfrequency patterns.
5ci: Influence of ENSO
Chiang Dao lies directly on the boundary between the Oceanic and Continental
monsoon troughs generated by the Asian/Indian monsoon. It is also directly northwest
the trade wind monsoon trough (from figure 3). Each circulation cell has been shown to
have a unique isotopic composition (Aggarwal, 2004; AraguasAraguas 1998; figure 3).
The precipitation at Chiang Dao may be influenced by source moisture from several
different locations. Therefore, I propose that El Niño can influence the isotopic values
at Chiang Dao by shifting the general position of the monsoon cells, which bring a
distinctive isotopic value to the rain water. To support the hypothesis, it is critical to
find the mechanism for a shifting of the cells, and show how known oscillations in
atmospheric temperature and pressure fluctuate in unison or antiunison with the stable
isotopes from the αcellulose. There are several published data sets with which to
58
compare the DCK 06a pine isotope record; the Pacific Decadal Oscillation (PDO), El
Nino events, the Nino 3 index, Nino 3.4 index, the Southern Oscillation Index (SOI),
the Indian Ocean Dipole (IOD), and estimates of total Asian Monsoon strength (as a
function of total rainfall). These oscillations represent changes in sea surface
temperature and pressure systems which may be powerful enough to shift the source of
precipitation over Chiang Dao. Each of these records are plotted side by side in figure
19 for comparison to the DCK 06a record. This permits investigation of, phase and anti
phase relationships, as well as similarities in their anomalies.
Plotting the Nino 3 index and the Nino 3.4 index (the index’s correspond to the
normalized changes in sea surface temperature in the western and eastern portions of
the equatorial pacific, respectively) against the δ
18
O (αcellulose) shows no immediate
striking qualitative correlations (Figure 19a). There are no shifts in either of the Nino
indexes which correspond to any of the “regime” changes. There is also no relationship
revealed from a regression. Further, the variability in the Nino index seems to increase
towards the present (whereas the δ
18
O of αcellulose decreases as time moves forward).
The average frequency of the Nino index is on the order of a few years, while the
cellulose shows no dominant 25 year patterns. The few periods where an argument for
correlation can be made; 1999, 1984, 1979, 1968, and 1965 are completely random with
respect to the magnitudes of the changes. The largest peak in the Nino 3.4 index (1999)
corresponds to the small area of variability in the cellulose record. There are no
discernable patterns between the periods of known “El Nino” events (as defined by the
shadowed portions of Figure 19a).
59
Figure 19: Interannual environmental oscillations plotted with δ
18
O (Chiang Dao)
a) b)
c) d)
(a) DCK O6a δ
18
O (αcellulose) plotted with both Nino 3 and Nino 3.4 indices through
time. The red line is the smoothed αcellulose data from Chiang Dao (from figure 10).
The blue dashed line is the Nino 3.4 index and the green line is the Nino 3.0 index. The
green shaded areas indicate periods of El Nino events (as given
http://ggweather.com/enso/years.htm). Nino index’s from: Mann et al 2000. (b)
Smoothed DCK 06a data against PDO data. From Mantua et al 1997. (c) Smoothed
DCK 06a δ
18
O (αcellulose) data against SOI data. From: Allan et al 1991; Konnen et al
1988. (d) Smoothed DCK 06a δ
18
O (αcellulose) data from Chiang Dao against Asian
monsoon summer (integrated rainfall amounts from June through September at several
sites throughout India) total rainfall amount.
60
5cii Pacific Decadal Oscillation
The PDO shows more similarity to the cellulose isotopes than the Nino indices
however, there is not a strong correlation between data sets (Figure 19b). The first three
major peaks in the PDO (centered around 1997, 1994, and 1988) correspond to relative
peaks in the cellulose record. The PDO also shows a relative increase in strength in the
positive phase from 19501945, in concert with the cellulose enrichment through the
same period. However, there is only a weak correspondence between the PDO and the
major depletion in the cellulose record (1960’s). I speculate that this may indicate that
during the later part of the century the change in the pressure gradients in the north
Pacific may be influenced the position of the Oceanic monsoon trough relative to the
continental monsoon trough. However, this variability is small and not persistent
through any other portion of the record.
5ciii: Southern Oscillation Index
The SOI (which records a normalized difference in sea surface pressure between
Tahiti and Darwin) seems to act in counterpoint to the PDO record. Through the 1990’s,
the SOI shows an antiphase relationship with the cellulose (Figure 19c). The late
cellulose peaks (which corresponded to the PDO peaks) are recorded as troughs in the
SOI record. However, there is little correspondence between the magnitudes of the
changes and a linear regression (not shown) produces little correlation. The early part of
the record, shows large enrichments in cellulose values, correspond to negative values
in the southern oscillation index. Interestingly, is the large cellulose depletion event in
the 1960’s corresponds to a significant decrease in SOI values, and the subsequent δ
18
O
61
increase in 1970 matches the increase in the southern oscillation. This may reflect two
mechanisms. First changes in the southern oscillation index may be influencing the
position of the trade wind trough relative to the oceanic and continental troughs, which
could bring precipitation from the trade wind trough over Chiang Dao or simply a more
enriched or depleted source from within the oceanic or continental troughs. It may also
be possible that the southern Indian oscillation is acting in concert with the PDO. They
may both be influencing the source moisture though different time periods, either
constructively or destructively. Perhaps through the early 1940’s and late 1980’s and
1990’s the PDO has a more powerful influence on the dominant circulation pattern over
Chiang Dao. Throughout the 1960’s and 1970’s the SOI assumes the source of the
dominant variability. The possibility of the SOI and the PDO interfering either
constructively or destructively with the position of monsoonal cells could account for
the lack of obvious relationships between the magnitude of changes in the decadal
oscillations and that of the isotope record. This could also explain the failed regression
between each individual decadal oscillation. However, the relative importance of the
two patterns (PDO and SOI) through time is unclear, and may be difficult to extract
from the cellulose record which only extends back through 1945 covering only a few
decadal oscillation cycles.
5civ: Indian Ocean Dipole
The IOD may be expected to act in concert with the SOI due to its roots in the
normalized difference of Indian Ocean sea level pressure variability. The IOD shows
weak correspondence to the cellulose record at its best. Interestingly, the IOD shows
62
high levels of variability in the periods where the cellulose shows its “regime” shifts.
However, the values in the IOD return to near “0” quickly after the increases and
decreases whereas the cellulose values tend to stabilize at their peak enrichment or
depletion. This may indicate that either there is nothing linking the isotopes in the
cellulose to the IOD record, or that the IOD is being forced by a mechanism similar to
that driving the SOI. However, the IOD may be too far south of Chiang Dao to have a
significant forcing on circulation patterns governing the Asian Monsoon over the
Southeast Asian continent.
5cv: Asian Monsoon Strength
The final variable investigated shows the most correlation by far, the strength of
the Asian summer monsoon indicated by integration over several sites in India of the
amount of precipitation between the peak summer months of JulySeptember (Figure
19d). Upon first glance there are striking similarities between summer monsoon rain
and δ
18
O of αcellulose throughout the entirety of the core. The peaks summer monsoon
strength do not necessarily correspond to the cellulose peaks; however, the periods of
high rainfall are periods of enriched cellulose values, and the periods of lower rainfall
are periods of depleted cellulose values. This may require a revision of the original
hypothesis. If the amount of rain is anticorrelated with the δ
18
O enrichments in the
Chiang Dao isotope record, this immediately lends credence to the original PESt model.
The summer Asian monsoon qualitatively seems to follow the cellulose in both phase
and magnitude. However, without an accurate growth model (and the knowledge of
63
whether growth occurs linearly or more rapidly during the rainy season) it is difficult to
create a robust regression.
Comparing the two records there are periods of both in and out of phase time.
This observation can supports a dominant “amount effect” modulated by changes in
source precipitation controlling the isotopic signal imparted on the cellulose. However,
while the “amount effect” is most likely to be a primary influence there are still a few
periods, some with no instrumental data when circulation changes may still be
observed. A critical period for comparison is the period between 1953 and 1963. This
period contains the largest isotopic excursion within the data set and precipitation data
for this interval also exhibits a distinctive excursion (figure 14). The anomalously high
rain event through 1962/63 corresponds to a drop in the cellulose δ
18
O values; however
the shift in δ
18
O values appears to lead the shift in precipitation. If the “amount effect”
was responsible for the isotope signal it would have to have changed after or at the same
time as the precipitation signal, but not before. I believe this observation is further
support for interspersed anomalous circulation events that are associated sometimes
with anomalous rainfall. Of course the validity of this interpretation depends on the
accuracy of the age model. It is possible the circulation cells shift before a rainfall
amount effect is taking place; the timing of the amount of precipitation is somewhat
irrelevant as to when the isotope record becomes depleted. As the isotope record
becomes depleted by 6‰ it corresponds to the time when the precipitation also begins
to drop during the year. Therefore it is possible that any further isotopic depletion that
follows is due to an “amount effect.” This suggestion is justified on the basis of the
isotopic signatures of the different monsoon cells. The OMT cell is up to 6‰ more
64
depleted then the CMT cell. Further AraguasAraguas (1998) argued this is due to
influences from arctic and other northern air masses. Therefore one might expect rain
associated with the CMT to have less depleted values since the moisture source is
centered in the tropics itself (the Indian Ocean). Comparing a possible amount affect
between the CMT and OMT I would predict that the “amount effect” during a CMT
storm will have less depleted δ
18
O values.
The timing difference between the enrichment and increase in rain in 1953 and
the decrease in rain and the isotopic depletion in 1963 is less then one year. Therefore,
if in this ten year interval, there was a single false ring and a single missing ring the
δ
18
O (αcellulose) record would become staggered, and the isotopic depletion would
line up more precisely to the Asian monsoon intensity record. This would imply that the
amount of rainfall was so large that the amount effect dominated the isotopic
composition of the precipitation, rather then the moisture source.
5d. Mechanistic theory
Although there is no straightforward data documenting the shifting of the OMT,
relative to the CMT and TWT there has been a breadth of literature surrounding the
atmospheric behavior during El Niño events, specifically in India. Not only is the
behavior of the oxygen isotopes during an El Niño year anomalous, but so is the
sustained monsoon rainfall regime during the El Niño year.
Traditionally during an El Nino event, tropical Walker circulation will shift
eastward along the equator. The rising air and moisture, which usually occurs over the
western Pacific, moves towards the central and eastern Pacific and the complementary
65
subsidence moves over the western pacific as far westward as the Indian continent,
suppressing the convection and precipitation that normally occur during the summer
monsoon. However, after 1980 this inverse (strong El Niño = weak monsoon)
relationship has not been observed (Singhrattna et al 2005, Wang et al 2001, Kumar et
at 1999). There have been normal monsoon precipitation patterns during the most recent
El Niño events. This implies the presence of a strong landocean thermal gradient that is
persistent through El Niño events. This has been shown to partially arise from decreases
in Eurasian and Himalayan snow levels. The snow raises the albedoreflecting radiation.
Furthermore, solar radiation can be converted to latent heat to melt the ice. The melt
water increases soil moisture, which would further decease continental warming during
spring and summer months. Over recent decades there has been an increase in summer
land surface temperatures relative to the Indian ocean SST’s which has sustained
normal monsoon behavior in spite of strong El Niño events. Although the annual
rainfall patterns have not changed this does not mean there is no change in the source of
the precipitation.
The specific climatology of Thailand was not extensively studied until 2005
(Singhrattna et al 2005). Studies show the same behavior as with the Indian Monsoon
during an El Niño event. This may explain the shifting of the OMT/CMT. In a “normal”
situation, due to Walker convective circulation patterns, the CMT is the dominant
source of precipitation over Chiang Dao, which shows good correlation associated with
an “amount effect” (AraguasAraguas 1998). During an El Niño this cell is shifted
eastward and the complex interaction of Asian monsoonal cells is replaced by a simple
landocean monsoon relationship. During such an event the landocean
66
temperature/pressure gradient would transport moisture from the Trade Wind trough
(equatorial Indian Ocean) to Chiang Dao, and this moisture would be isotopically
depleted relative to the CMT (Aggarwal 2004). It may also be possible that single storm
events that originate in the OMT or TWT may sporadically influence Chiang Dao
enough to yield the secondary variability present in figure 17c. However, their effect
should influence both amount of rain fall and average δ18O.
Also, perhaps during a La Nina there is a shift from the CMT to the OMT as the
source of precipitation over Chiang Dao, opposite to that observed in an El Nino event.
Moving the atmospheric systems westward would change the isotopic signal of the
source while retaining a strong influence of the “amount effect” (R= 0.77). The OMT
moisture is, on average, more depleted isotopically then the CMT (figure 3), and this,
combined with the increase in Walker circulation, could create a more depleted isotopic
signal relative to a normal year. Possibly because these (El Niño and La Niña)
circulatory changes take place on inter and intraannual time scales, there is not enough
time for the anomalous circulation to stabilize (yielding a clear amount effect) which
may explain the larger scatter in the La Niña data (when the CMT and OMT undergo a
relatively small shift) and the very large scatter in the El Niño data (when the cells
undergo substantial shifts). This observation combined with the regression shown in 34c
may explain the entire range of variability through time in the isotopic record.
5e. Previously Published Data
Published data from Evans, Schrag and Poussart (2004, 2005) from several
locations around the world and from several different tree species, shows similar
67
seasonal patterns to that observed for DCK 06a through the last part of the 20
th
century.
Their conclusions support a prominent “amount effect” influence on the isotopic
composition of tree cellulose through time. Their data may also be influenced by
changes in monsoonal circulation and described here for Chiang Dao. Taking the data
on a site by site basis it may be possible to show that different sources of circulation are
capable of causing a significant isotopic shift. Two of the sites studied by Schrag and
Poussart (Schrag and Pousart 2005) in Indonesia and Thailand are susceptible to the
same circulation variability (interactions between the CMT, OMT and TWT) as those
that affect Chiang Dao. It seems that several samples taken from Indonesia and
Thailand (Chiang Mai; approximately 60 km south of Chiang Dao) show similar 26‰
shifts, with significantly more sub decadal variability then intraannual variability.
Spruce and Oak trees from Japan, which are also influenced by Asian monsoon
circulation also show similar shifts; slightly smaller, this suggests again, the dominance
of an “amount effect” to circulation changes. Possibly this fidelity is due to the fact that
the Japan sites are in the middle of a circulation cell, rather then on the boundary
between two, and therefore do not see as much isotopic variability. Even in continental
United States (Massachusetts) where there is an influence of monsoonal Atlantic and
artic moisture sources (presumably with different isotopic signatures) the annual
patterns are consistent with an amount effect (Evans and Schrag 2004; Schrag and
Poussart 2005). It seems that the Japan and U.S. locations under investigation all
undergo, to some extent, a seasonal monsoon which has the ability to alter origin of
source moisture; however, their variability is consistently dominated by an amount
effect.
68
There is another phenomenon which the PESt model does not address;
accounting for evapotranspiration in climates with small variations in relative humidity
and in an organism with no leaves. Figure 20 (taken from Tutzet et al, 2003) shows the
pressure gradient formed by leaves, roots and capillary action. A tree’s leaves play the
largest role in developing and sustaining the pressure gradient. Therefore, as a
deciduous tree loses its leaves seasonally there is a time in every year when very little
new water can be brought up from the roots. Furthermore, there would be no leaves to
make new sugars. It may also be possible to explain a small part of the enrichment
through stored photosynthate synthesized earlier in the rainy season. For every aliquot
of water taken up by the tree, a portion is used to make sugars and that will be stored.
The usage of those stored supplies as the rainy season ends would yield a signal that
mimics the previous rainy season. The possible explanation to this observation again
lies in the growth model. Using growth model 3 as a guide to cellulose synthesis as a
function of calendar month it is possible to estimate the most likely growth pattern. This
growth model agrees closely with that proposed by the red line in figure 9a. In an
average year most of the samples correspond to rainfall values between February and
August. This result suggests either discontinuous synthesis over the course of a year or
>90% of the cellulose synthesis during the raining season, while the tree still has its
foliage.
There is one final possible mechanism to explain the previous data via an
observation from the original αcellulose model proposed by Evans and Schrag (2004;
equation 6). The model incorporates a ratio, “f
o
” to account for the mixing and re
exchange between xylem water and exchangeable carbonbound oxygen (Roden et al
69
Figure 20: Schematic of the pressure gradient created by trees
Taken from Tuzet et al 2003. Pressure gradient in a tree as a function of water potential
for the leaf, root and soil systems through time.
2000, Poussart et al 2004). This ratio, in previously published literature, has been
considered to be a constant, yearround (Roden et al, 2000; Evans and Schrag, 2004).
This may be an inappropriate assumption. The model is sensitive to fluctuations in this
ratio. Fluctuations in “f
o
” result in several per mil variability in the resulting αcellulose.
“f
o
” is described by a percent of exchange between oxygen and water. Therefore this
ratio depends inherently on the amount of water being carried up through the cambium
as well as on the amount of sugars being carried down the cambium for cellulose
synthesis; changes in the amounts of these place constraints on how much exchange
may take place. In a deciduous tree there is no reason to believe this ratio will stay
constant between the rainy season and the dry season. To the contrary, the weakening of
the pressure gradient and the decrease of amount of source water along with the
decrease in synthesis of sucrose and photosynthate may drastically alter the ratio of
mixing and thus the δ
18
O which becomes incorporated in the αcellulose. Nonetheless,
this explanation for the observed patterns of δ
18
O is also remote. Assuming little water
is being carried up the cambial tissue as a result of the loss of foliage, little cellulose
would be synthesized. Therefore, fluctuations in “fo” may be large over the course of
70
the year; however, the mass balance of resulting cellulose may not be significantly
influenced since little cellulose is synthesized at these times
It may be possible to modify the model to account for the variability inherent
throughout the course of the year;. Exchanging the ratio term for an optimized function
which allows “f
o
” to respond to changes in rainfall amount may create the seasonal
variability on a more accurate level. A reduction in the amount of rainfall affecting the
“amount effect” decreases the amount of xylem water introduced for sugar to exchange
with and the resulting δ
18
O of the αcellulose becomes more enriched by several permil.
This theory could also be tested qualitatively by evaluating leaf size. The surface area of
a leaf plays a large role in how much water may be transpired and therefore the
difference between the water transported while the leaves are blooming and while the
tree is bare (Roden et al 2000; Evans and Schrag 2004; Barbour et al 2004). The trees
examined in the published papers cover many species with a wide variety of leaves. The
variability may be expected to be larger in species which grow large leaves than in
species with smaller leafs or needles. This indeed may help explain data in some
instances. Podocarpus neriifolius from Chiang Mai show incredibly large
fractionations, much larger then any other species, ranging from 44‰ enrichments to
24‰ depletions (Poussart et al 2004). This species of tree has large full leaves up to 15
cm long. Whereas Samanea saman (from the island of Bali) has small round leaves and
exhibits a seasonal variation on the order of 6‰. Upon this evidence it also seems
generalizations over all species are premature and may hold important data which could
help resolve much future work.
71
5f. Final Proxy Reconstruction
The previous sections have made clear the problems of relying solely on the
amount effect to reconstruct proxy rainfall, however, the final phase of this experiment
will use the δ
18
O (cellulose) vs. rainfall amount relationship from growth model 3 to
carry out a proxy reconstruction of rainfall amount through time. Figure 21a shows the
raw DCK 06a data and the corresponding reconstructed rainfall amount values. The
relationship captures the inverse correlation between δ
18
O and rainfall amount. The
reconstruction is then plotted with the instrumental data in figure 21b (reconstructions
with second and third order functions are also shown). All three reconstructions provide
overestimates. And the linear reconstruction produces several subzero values
(indicating a negative rainfall amount). The seasonal trend is still apparent. However, it
does not correspond well to the instrumental data. Unfortunately while the data does
show strong evidence for significant variability attributable to the “amount effect” it is
not robust enough to yield a quantitative proxy technique.
Chapter 6: Conclusions
Analysis of Pinus kesiya core DCK 06a from Chiang Dao northern Thailand
supports the initial hypothesis showing strong support for the amount effect as the
primary control on the composition of the αcellulose. However, the story is shown to
be more complex with atmospheric circulation playing a large role. During El Niño
years when atmospheric circulation influences moisture source and overshadows any
systematic amount effect.
72
Figure 21: Linear rainfall proxy reconstruction and comparison to the
instrumental record
a)
b)
a) Proxy reconstructed rainfall amount (using growth model 3 relationship; rainfall=
1179.7(δ
18
O) + 28246) plotted against δ
18
O of cellulose recorded at Chiang Dao.
b) Proxy rainfall reconstruction with the instrumental record. Blue line is a
reconstruction from the growth model 3 linear relationship. Green line is a
reconstruction from a second power relationship (also from growth model 3) between
rainfall amount and isotopic composition of the cellulose. Black line is the relationship
at the third power.
73
However, influential decadal oscillations in the area seem to have little
correlation to the δ
18
O of the cellulose. The records which show any correlation are the
rainfall amount records, although the age model for ring growth makes lining the
records up on an intraannual basis difficult. However, the study yielded many ideas as
to the mechanisms responsible for the observed isotopic variability. Data have shown
variability is likely to be caused by the thermodynamic and kinetic Rayleigh
fractionation effects with secondary variability caused by changes in atmospheric
circulation and possibly variability in the biological cycling of fluids within the tree. It
is possible that previously published data for other sites may be influenced by identical
mechanisms.
The importance of assigning a proper age model has been revealed. The model
which is developed to tune the isotopic record against may reveal more or less of an
amount effect depending on how one defines a “growth year.” The tuning of the age
model to the instrumental rainfall data (growth model 3) proved to be the best treatment
for the data. The linear age model created by dividing every tree ring into a systematic
grouping of “colors” (that isolating the earlywood or latewood for analysis) may
truncate the resulting variability and therefore interpretation. This has important
implications in the traditional dendroclimate field. Reanalysis combining both an
isotopic rainfall age model and adjusting the ring width/density data sets may yield
more truthful correlations to climate phenomena. However, the rainfall amount
reconstruction is not yet robust enough for quantitative proxy reconstructions.
While this study supports the conclusion that there is important environmental
information recorded in the oxygen isotopes of the αcellulose of wood from Pinus
74
kesiya there is still significant ground which needs to be explored before this novel
method can yield its full potential. δD and δ
13
C ratios of αcellulose need to be analyzed
for tropical trees along with δ
18
O data. δD ratios regressed against δ
18
O can be used to
show deviations from source water lines (Sternberg 1981). This could make clear the
interplay between the “amount effect” and circulation patterns and perhaps resolve
questions introduced by the lack of correlation between the reconstructed rainfall values
and the instrumental values. Further, coupled to δ
13
C (which is dependent on several
biologically controlled parameters; McCaroll and Loader 2004) empirical growth
models can be developed, perhaps without the use of dendrometers. However,
dendrometer data would provide the most accurate means for calibration of potential
growth models. Unfortunately dendrographs gather data in real time, and compiling a
statistically robust data set could take up to 10 years worth of measurements. A robust
growth model has the ability to constrain how different trees species respond to annual
patterns in precipitation.
It is also important for the stable isotope dendrological community to agree on a
single method of analysis so that there are no discrepancies originating from different
analytical and chemical methods. Further a single chemical definition should also be
enforced. More communication must also take place between those who study the stable
isotopes of precipitation and the dendroisotopists. The evidence presented here
suggests that a there are a number of controls on the isotopic composition of α
cellulose, which implies a great deal of information about several parameters may be
extracted once we posses the correct understanding.
75
References
Aggarwal, P. Frohlich, K. Kulkarni, and K.M. Gourcy, L.L. (2004). Stable isotope
evidence for moisture sources in the Asian summer monsoon under present and past
climate regimes. Geophysical Research Letters, Vol. 31, L08203.
Allan, R.J., Nicholls, N., Jones, P.D. and Butterworth, I.J. (1991) A further extension of
the TahitiDarwin SOI, early SOI results and Darwin pressure. J. Climate Vol. 4, p.
743749.
Anderson W.T., Bernasconi S.M., McKenzie J.A., Saurer M., and Schweingruber F.
(2002). Model Evaluation for reconstructing the oxygen isotopic compostion in
precipitation from tree ring cellulose over the last century. Chemical Geology. Vol. 182,
p 121137.
Anderson W.T., Bernasconi S.M., and McKenzie J.A. (1999). Oxygen and Carbon
Isotopic record of climatic variability in tree ring cellulose (Picea abies): An example
from central Switzerland, (19131995). Journal of Geophysical Research. Vol. 103 No.
D24, p 31,62531,636.
AraguasAraguas, L. and Froehlich, K. (1998). Stable isotope composition of
precipitation over Southeast Asia. Journal of Geophysical Research Vol. 103 p 28,721
28,742.
Barbour, M.M., Roden J.S., Farquhar G.D., and Ehleringer J.R (2004). Expressing leaf
water and cellulose oxygen isotope ratios as enrichment above source water reveals
evidence of a Peclet effect. Oecologia. Vol. 138 p 426435.
Brendel O., Innetta P.P.M., and Stewart D. (2000). A rapid and simple method to isolate
pure AlphaCellulose. Phytochemical Analysis. Vol. 11, p 710.
Briffa K.R., Schweingruber F.H., Jones P.D., Osborn T.J., Shiytov S.G., and Vaganov
E.A. (1998). Reduced sensitivity of recent treegrowth to temperature at high northern
latitudes. Nature. Vol. 391 p 678.
Buhay W.M., Edwards T.W.D., and Aravena R. (1996). Evaluating kinetic fractionation
factors used for ecologic and paleoclimatic reconstructions from oxygen and hydrogen
isotopic rations in plant water and cellulose. Geochimica et Cosmochimca Acta. Vol.
60. No. 12. p 22092218.
Buhay W.M., and Edwards T.W.D. (1995). Climate in Southwestern Ontairo, Canada,
between AD 1610 and 1885 inferred from Oxygen and Hydrogen Isotopic
Measurements of Wood Cellulose from Trees in Different Hydrologic settings.
Quaternary Research. Vol. 44, p 438446.
76
Buhay W.M., Edwards T.W.D., and Aravena R., (1996) Evaluating kinetic fractionation
factors used for ecological and paleoclimatic reconstructions from oxygen and hydrogen
isotope ratios in plant water and cellulose. Geochimica et Cosmochimica Acta. Vol. 60,
p 22092218.
Burk R.L., and Stuiver M. (1981). Oxygen Isotope Ratios in Trees Reflect Mean
Annual Temperature and Humidity. Science. Vol. 211, No. 4489, p 14171419
Craig H. and Gordon L. I. (1995). Deuterium and oxygen18 variations in the ocean and
the marine atmosphere. In Proceedings of a conference on stable isotopes in
oceanographic studies and paleotemperatures (ed. E. Tongiorgi). Spoleto, Italy. p 9130
Danis, P.A. MassonDelmontte, V. Stievenard, M. Guillemin M.T. Daux V. Naveau,
Ph. and Von Grafenstein, U. (2006). Reconstruction of past precipitation dO18 using
treering cellulose dO18 and dC13: a calibration study near Lac d’Annecy, France.
Earth and Planetary Science letters. Vol. 243 p 439448.
Dansgaard W. (1964). Stable Isotopes in Precipitation. Tellus. Vol. 4, p 436467.
Dawson T.E. (1993) Water sources of plants as determined from xylemwater isotopic
composition: perspectives on plant competition, distribution, and water relations. In
Stable Isotopes and Plant CarbonWater relations (ed. J.R. Ehleringer et al) p. 465496.
Dawson T.E. and Ehleringer J.R., (1993) Isotopic enrichment of water in the “woody”
tissues of plants: Implications for water source, water uptake, and other studies which
use the stable isotopic composition of cellulose. Geochimica et Cosmochimica Acta.
Vol. 57, p 34873492.
DeNiro M.J., and Epstein S. (1979). Relationship Between the Oxygen Isotope Ratios
of Terrestrial Plant Cellulose, Carbon Dioxide and Water. Science Vol. 204, No. 4388,
p 5153.
DeNiro M.J., and Cooper L.W. (1989). Postphotocynthetic modification of oxygen
isotope ratios of carbohydrates in the potato: Implications for paleoclimate
reconstruction based upon isotopic analysis of wood cellulose. Geochimica et
Cosmochimica Acta. Vol. 53, p 25732580.
DeNiro M.J., and Cooper L.W., (1990) Water is lost from leaves and trunks of trees by
fundamentally different mechanisms. Geochimica et Cosmochimica Acta. Vol. 54, p.
18451846.
Edwards T.W.D. and Fritz P. (1986) Assessing meteoric water composition and relative
humidity from 18O and 2H in wood cellulose: paleoclimatic implications for southern
Ontario, Canada. Applied Geochemistry. Vol. 1, p 715723.
77
Edwards T.W.D. (1990). New contributions to isotope dendroclimatology form studies
of plants. Geochimica et Cosmochimica Acta. Vol. 54, p 18431844.
Epstein S. and Yapp C.J. (1977). Isotope tree thermometers (comment on Libby et al.).
Nature. Vol.266. p 477478.
Epstein S., Thompson P. and Yapp C.J. (1977). Oxygen and Hydrogen Isotopic Ratios
in Plant Cellulose. Science. Vol. 198. No. 4323 p 12091214.
Evans M.N. and Schrag D.P. (2004). A stable isotopebased approach to tropical
dendroclimatology. Geochimica et Cosmochimica Acta, Vol. 68, No. 16. p 32953305.
Farquhar G.D., Barbour M.M., and Henry B.K. (1998). Interpretation of oxygen isotope
composition of leaf material. In Stable Isotopes (ed. H. Griffiths) p 2762. BIOS
Scientific Pub. Ltd.
Flanagan L.B. (1993) Environmental and biological influence on the stable oxygen and
hydrogen isotopic composition of leaf water. In Stable isotopes and Plant CarbonWater
relations (ed. J.R. Ehleringer et al) p 7190.
Flanagan L.B. Bain J.F. and Ehleringer J.R. (1991). Stable oxygen and hydrogen
isotope composition of leaf water in C
3
and C
4
plant species under field conditions.
Oecologia. Vol. 88, p 394400.
Flanagan L.B., Phillips S.L., Ehleringer J.R., Lloyd J., and Farquhar G.D. (1994) Effect
of changes in leaf water oxygen isotopic composition on discrimination against
C
18
O
16
O during photosynthetic gas exchange. Australian Journal of Plant Physiology.
Vol. 96, p 588596.
Gary J. and Thompson P. (1976). Climatic information from 18O/16O ratios of
cellulose in tree rings. Nature. Vol. 262 p 481483.
Gary J. and Thompson P. (1977). Climatic information from 18O/16O analysis of
cellulose, lignin, and whole wood from tree rings. Nature. Vol. 270, p708709.
Gat, J.R. (1996) Oxygen and hydrogen isotopes in the hydrologic cycle. Annual Review
of Earth and Planetary Science. Vol. 24, p 225262
Gaudinski J.B., Dawson T.E., Quideau S., Schuur E.A.G., Roden J.S., Trumbore S.E.,
Sandquist D.R., Oh S.W. and Wasylishen R.E. (2005). Analytical Chemistry, Vol. 77, p
72127224.
Green J.W. (1953) Methods in Carbohydrate Chemistry: Chapter 3 Wood Cellulose.
Vol. 3: Cellulose. Academic Press. New York and London.
78
Hill S.A., Waterhouse J.S., Field E.M., Switsur V.R., and Rees T. (1995). Rapid
recycling of triose phosphates in oak stem tissue. Plant, Cell, and Environment. Vol. 18,
p 931936.
Können, G.P., Jones, P.D., Kaltofen, M.H. and Allan, R.J. (1998) Pre1866 extensions
of the Southern Oscillation Index using early Indonesian and Tahitian meteorological
readings. J. Climate. Vol. 11, p 23252339.
Kripalani R.H. and Kulkarni A. (1998) The Relationship Between some Largescale
Atmospheric Parameters and Rainfall over Southeast Asia: A comparison with Features
over India. Theoretical Applied Climatology. Vol. 59, p 1.
Kumar K. K., Rajagopalan B. and Cane M.A. (1999) On the weakening Relationship
Between the Indian Monsoon and ENSO. Science. Vol. 284, p 2156.
Leavitt, S. W. and Danzer, S.R. (1993). Method for batch processing small wood
samples to holocellulose for stable carbon isotope analysis. Analytical Chemistry. Vol.
65, p 8789.
Libby M.L., Pandolfi L.J., Payton P.H., Marshall III, J., Becker B. and Giertz
Sienbenlist V. (1976). Isotopic tree thermometers. Nature. Vol. 267, p 284.
Lipp J., Trimborn P., Fritz P., Moser H., Becker B. and Franzel B. (1991) Stable
isotopes in tree ring cellulose and climatic change. Tellus. 43B, p 322330.
Lipp J., Trimborn P., Edwards T., Waisel Y. and Yakir D. (1996). Climatic effects on
the dO18 and d13C of cellulose in the desert tree Tamarix Jordanis. Geochimica et
Cosmochimica Acta. Vol. 60 No. 17 p 33053309.
Luo Y. H. and Sternberg L (1992). Hydrogen and oxygen isotope fractionation during
heterotrophic cellulose synthesis. Journal of experimental Botany. Vol. 43, p 4750.
Mann, M.E., E.P. Gille, R.S. Bradley, M.K. Hughes, J.T. Overpeck, F.T. Keimig and
W.S. Gross (2000) Global Temperature Patterns in Past Centuries: An Interactive
Presentation, Earth Interactions. Vol. 4, Paper 4.
Mantua, N.J. and S.R. Hare, Y. Zhang, J.M. Wallace, and R.C. Francis, (1997). A
Pacific interdecadal climate oscillation with impacts on salmon production. Bulletin of
the American Meteorological Society, Vol. 78, p. 10691079.
Peterson, T.C. and Vose R.S. (1997) An overview of the Global Historical Climatology
Network temperature data base. Bulletin of the American Meteorological Society. Vol.
78 p 28372849
79
Poussart P.F., Evans M.N., Schrag D.P. (2004). Resolving seasonality in tropical trees:
multidecade, high resolution oxygen and carbon isotope records from Indonesia and
Thailand. Earth and Planetary Science Letters. Vol. 218, p 301316.
Poussart P., Schrag D.P. (2005). Seasonally resolved stable isotope chronologies from
northern Thailand deciduous trees. Earth and Planetary Science Letters. Vol. 235, p
752765.
RaffalliDelerce G., MassonDelmotte V., Dupouey J.L., Steivenard M., Breda N. and
Moisselin J.M. (2004). Reconstruction of summer droughts using treering cellulose
isotopes: a calibration study with living oaks from Brittany (western France). Tellus.
Vol. 56B, p 160174.
Ramesh R., Bhattacharya S. K. and Gopalan K. (1986). Climatic correlation in the
stable isotope records of silver fir (Abies pindrow) trees from Kashmir, Inida. Earth and
Planetary Science Letters, Vol. 79 p 6674.
Rebetez M., Saurer M. and Cherubini P. (2003). To what extent can oxygen isotopes in
tree rings and precipitation be used to reconstruct past atmospheric temperature? A case
study. Climatic Change. Vol. 61, p 237248.
Robertson I., Waterhouse J.S., Barker A.C., Carter A.H.C. and Switsur V.R. (2001).
Oxygen Isotope ratios of oak in east England: implications for reconstructing the
isotopic composition of precipitation. Earth and Planetary Science Letters. Vol. 191, p
2131.
Roden J.S. and Ehleringer J.R. (1999). Observations of Hydrogen and Oxygen Isotopes
in Leaf Water Confirm the CraigGordon Model under WideRanging Environmental
Conditions. Plant physiology. Vol. 120. p 11651173.
Roden J.S. and Ehleringer J.R. (1999b). Hydrogen and oxygen isotope ratios of tree
ring cellulose for riparian trees grown longterm under hydroponic, controlled
environments. Oecologia. Vol. 121, p 467.
Roden J.S. and Ehleringer J.R. (1999c). Hydrogen and oxygen isotope ratios of leaf
water and treering cellulose for field grown riparian trees. Oecologia. Vol. 123, p 481.
Roden J.S., Lin G. and Ehleringer J.R. (2000). A mechanistic model for the
interpretation of hydrogen and oxygen isotope ratios in treering cellulose. Geochimica
et Cosmochimica Acta. Vol. 64 No. 1 p 2135.
Ropelewski, C.F. and Jones, P.D. (1987) An extension of the TahitiDarwin Southern
Oscillation Index. Monthly Weather Review. Vol. 115, p 21612165.
Saurer M., Borella S. and Leuenberger M (1997). δ
18
O of tree rings of beech (Fagus
silvatica) as a record of δ
18
O of growing season precipitation. Tellus. Vol. 49B. p 8092.
80
Saurer M., Robertson I., Siegwolf R. and Leuenberger M. (1998). Oxygen Isotope
Analysis of Cellulose: An Interlaboratory Comparison. Analytical Chemistry. Vol. 70, p
20742080.
Saurer M., Cherubini P. and Siegwolf R. (2000). Oxygen isotopes in tree rings Abies
alba: the climate significance of interdecadal variations. Journal of Geophysical
Research. Vol. 105, No. D10. p 12,46112,470.
Shafizadeh, F. and Fu, Y.L. (1973). Pyrolysis of cellulose. Carbohydrate research. Vol.
29, p 113122.
Singhrattna N., Rajagopalan B., Kumar K.K. and Clark M. (2005). Seasonal
Forecasting of Thailand summer monsoon rainfall. International Journal of
Climatology. Vol. 25, 649.
Singhrattna N., Rajagopalan B. and Kumar K.K. (2005). Interannual and Interdecadal
Variability of Thailand Summer Monsoon Season. Journal of Climate Vol 18, p 1697.
Sternberg L.S.L., Anderson W.T. and Morrison K. (1989) Separating soil and leaf water
18O isotopic signals in plant stem cellulose. Geochimica et Cosomochimica Acta. Vol
67, No. 14 p 25612566.
Sternberg L., Pinzon M.C., Anderson W.T. and Jahren A.H. (2006). Variation in
oxygen isotope fractionation during cellulose synthesis: intramolecular and biosynthetic
effects. Plant, Cell, and Environment. Vol. 29, p 1.
Switsur, V.R. Waterhouse, J.S. Field, E.M.F. Carter, A.H.C. Hall, M. Pollard, M.
Robertson, I. Pilcher, J.R. and Heaton, T.H.E. (1994). Stable isotope studies of oak
from the English Fenland and Northern Ireland. In; Funnell, B.M. Kay, R.L.F. (Eds.),
Palaeoclimate of the Last Glacial/Interglacial Cycle. Natural Environmental research
council special publication 94/2.
Switsur, V.R. Waterhouse, J.S. Field, E.M.F. and Carter, A.H.C. (1996). Climatic
signals from stable isotopes in oak treerings from East Anglia, Great Britain. In: Dean,
J.S. Meko, D.M. Swetnam, T.W. (Eds), Tree Rings, Environment and Humanity
Radiocarbon, p 637645.
Tsuji H., Nakatsuka T. and Takagi K (2006). dO18 of treering cellulose in two species
(spruce and oak) as proxies of precipitation amount and relative humidity in northern
Japan. Chemical Geology. Vol. 231 p 6776.
Tuzet, A. Perrier and R. Leuning (2003). A coupled model of stomatal conductance,
photosynthesis and transpiration. Plant Cell and Environment. Vol. 26, p. 1087–1096.
81
Verheyden A., Helle G., Scheleser G.H., Deharis F., H. Beeckman and Koedam N.
(2004). Annual cyclicity in highresolution stable carbon and oxygen isotope ratios in
the wood of the mangrove tree Rhizophora mucronata. Plant, Cell and Environment.
Vol. 27, p 15251536.
Waterhouse J.S., Switsur V.R., Barker A.C., Carter A.H.C. and Robertson I. (2002).
Oxygen and hydrogen isotop ratios in tree rings: how well do models predict observed
values? Earth and Planetary Science Letters. Vol. 201, p 421430.
Yakir D. and DeNiro M. J. (1990). Oxygen and Hydrogen Isotope Fractionation during
cellulose Metabolism in Lemna gibba L. Plant Physiology. Vol. 93, p 325332.
Yapp C.J. and Epstein S. (1982). A reexamination of cellulose carbon bound hydrogen
δD measurements and some factors affecting plantwater D/H relationships.
Geochimica et Cosmochimica Acta. Vol. 46, p 955965.
Zhang, Y., J.M. Wallace, and Battisti D.S. (1997). ENSOlike interdecadal variability:
190093. J. Climate. Vol. 10, p 10041020.

Abstract (if available)

Abstract The relationship between environmental variables and the oxygen isotope composition of alpha-cellulose in ringed tropical trees may not be constrained to simple "amount effects." Here I present a 55-year continuous record of delta18O of alpha-cellulose from Pinus kesiya. By comparing the cellulose isotope record to previous data, model theory, and rain amount data it is clear that both amount effects as well as isotopic changes in atmospheric circulation regimes are incorporated in the oxygen isotopic ratio of the alpha cellulose. Regression analysis show most of the variability recorded in the delta18O of cellulose results from changes in total annual rainfall, and large anomalies arise during La Nina and El Nino events. A proxy rainfall reconstruction exhibits poor correlation with rainfall variability and will require further work in order to constrain how best to extract a quantitative estimate of rainfall.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Recent variability in the hyrdological cycle of tropical Asia from oxygen isotopes of tree celulose

PDF

Exploring seasonal climate variability in East African trees: High-resolution oxygen and carbon isotope records of cellulose of Vitex payos from Chyulu Hills, Kenya

PDF

Abrupt climate variability during the last glacial and the late Holocene from the western tropical Pacific perspective