University of New Hampshire University of New Hampshire Scholars' Repository
Doctoral Dissertations Student Scholarship
Fall 2006 Recent extreme events in a tropical stalagmite: Multi-proxy records and analysis of ecosystem delta carbon-13 value sensitivity to weak climate forcing Amy E Benoit Frappier University of New Hampshire, Durham
Follow this and additional works at: https://scholars.unh.edu/dissertation
Recommended Citation Frappier, Amy E Benoit, "Recent extreme events in a tropical stalagmite: Multi-proxy records and analysis of ecosystem delta carbon-13 value sensitivity to weak climate forcing" (2006). Doctoral Dissertations. 338. https://scholars.unh.edu/dissertation/338
This Dissertation is brought to you for free and open access by the Student Scholarship at University of New Hampshire Scholars' Repository. It has been accepted for inclusion in Doctoral Dissertations by an authorized administrator of University of New Hampshire Scholars' Repository. For more information, please contact [email protected]. RECENT EXTREME EVENTS IN A TROPICAL STALAGMITE:
MULTI-PROXY RECORDS AND
ANALYSIS OF ECOSYSTEM 513C VALUE SENSITIVITY TO WEAK CLIMATE FORCING
BY
AMY E. BENOIT FRAPPIER
B.S. University of Maine, 1999
M.S. University of New Hampshire, 2002
DISSERTATION
Submitted to the University of New Hampshire
in Partial Fulfillment of
the Requirements for the Degree of
Doctor of Philosophy
in
Earth and Environmental Sciences
September, 2006
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UMI Number: 3231351
Copyright 2006 by Frappier, Amy E. Benoit
All rights reserved.
INFORMATION TO USERS
The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleed-through, substandard margins, and improper alignment can adversely affect reproduction. In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion.
® UMI
UMI Microform 3231351 Copyright 2006 by ProQuest Information and Learning Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code.
ProQuest Information and Learning Company 300 North Zeeb Road P.O. Box 1346 Ann Arbor, Ml 48106-1346
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. This dissertation has been examined and approved.
Dissertatiojj^jirector, Dr. Dork L. Sahagian, Professor of Earth & Environmental Sciences, Lehigh University
c XOM*. ______Co-Director, Dr. Cameron P. Wake, Research Associate Professor of Earth Sciences and Earth, Oceans, and Space
Associate Professor of Geology
Dr. Serita D. Frey, Assistant Professor of Soil MicrobiaTEcology
4 A u . j Dr. Luis A. Gonzalez, / ) Associate Professor, The University of Kansas
Date
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ACKNOWLEDGEMENTS
I would like to thank all the many people who helped make this research possible. First,
I must thank my intrepid field assistants, Brian Frappier - who found ATM7 - and Todd Belanger,
and my dependable laboratory assistants Heather Foley and Joshua Blye for many long hours in
the service of science. I also thank Luis Gonzalez, Scott Carpenter, Serita Frey, Mel Knorr, Ruth
Varner, Jeff Miriam, Brian Frappier, Terry Plank, Louise Bolge, Jeff Donnelly, and Jess
Tierney for their hospitality, help, and interest during my visits to their laboratories. Special
thanks go to Jacqui Aitkenhead-Peterson for allowing me to work under her international soil
importation permit, and to Bob Eckert for his kindness in lending me bench space and use of his
laboratory. I appreciate the sustained and last-minute assistance and support I have received from
University of New Hampshire’s Department of Earth Sciences, as well as the Climate Change
Research Center’s outstanding staff: Belinda Camire, Suki Easter, Jennifer Boles, Linda Tibbetts.
I will always remember fondly the companionship and reality checks of my fellow
graduate students, and will continue to treasure the wonderful friendships that grew up along the
way. My parents Paul & Joanne Benoit always enthusiastically supported me throughout my
education. This endeavor would have been impossible without Brian Frappier, and I am ever thank
ful to him for his invaluable dedication, clear perspective, and uncountable actions of support.
My committee members each supported me through this project in many ways. I thank
Wally Bothner for both mentorship and teamwork in College Teaching. Cameron Wake has also
been a special mentor to me, and I appreciate the many opportunities he has sent my way over the
years. Over the years, my advisor Dork Sahagian struck the right balance between freedom and
guidance, allowing me to become a well-rounded scientist. His buoyant optimism, unflagging
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. enthusiasm, and high spirits made laughter the custom as we navigated the shoals of graduate
school. His mentorship is sure to continue long after I launch my career.
The research presented in this dissertation was funded primarily by NASA’s Earth
System Science Fellowship Program with additional support from the U.S. National Science
Foundation, and the Inter-American Institute for Global Change Research.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. TABLE OF CONTENTS
ACKNOWLEDGEMENTS ...... iii
LIST OF TABLES...... viii
LIST OF FIGURES...... ix
ABSTRACT...... xi
CHAPTER PAGE
INTRODUCTION...... 1
I. A THREE-STAGE SCREENING SYSTEM TO SELECT STORM-SENSITIVE
STALAGMITES...... 3
Abstract...... 3
Introduction ...... 3
A Three-stage Screening System to Select Storm-sensitive Stalagmites ...... 7
Potential Problems: Region ...... 8
Potential problems: Cave Configuration and History ...... 10
Potential Problems: Stalagmite ...... 14
Lab Selection Procedures ...... 16
Discussion ...... 18
II. A STALAGMITE STABLE ISOTOPE RECORD OF RECENT TROPICAL CYCLONE
EVENTS...... 24
A bstract...... 24
Introduction ...... 25
Hurricane Stable Isotope Values and Cave Depositional Settings ...... 26
v
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Methods 27
Identifying and Dating Isotopic Excursions ...... 28
Results and Discussion ...... 28
Investigating Potential Indicators of Storm Intensity ...... 30
Conclusions ...... 31
III. EMPIRICAL ORTHOGONAL FUNCTION ANALYSIS OF MULTIVARIATE
STALAGMITE TRACE ELEMENT DATA: DETECTING THE 1982 EL CHICHON
VOLCANIC ERUPTION AND OTHER ENVIRONMENTAL EXTREMES...... 38
Abstract...... 38
Introduction ...... 39
M ethods ...... 43
Data Analysis: LA-ICPMS ...... 45
Data Analysis: X RF ...... 46
R esults ...... 46
LA-ICPMS...... 46
Scanning XRF ...... 48
Discussion ...... 48
Conclusions ...... 53
IV. ROOTING OUT THE AMPLIFICATION MECHANISM LINKING A WEAK ENSO
TELECONNECTION SIGNAL AND A STRONG STALAGMITE CARBON ISOTOPE
RECORD ...... 63
Introduction ...... 63
M ethods ...... 68
R esults ...... 69
vi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Discussion ...... 71
Conclusions ...... 74
CONCLUSIONS...... 83
Future Directions ...... 86
APPENDICES...... I l l
APPENDIX A. CLIMATE, ECOLOGY, AND GEOLOGIC SETTING ...... 112
APPENDIX B. MICRO-MILLING AND STABLE ISOTOPE ANALYSIS...... 115
APPENDIX C. AGE MODEL...... 117
Radiometric Dating ...... 118
Layer Counting and Initial Age Model ...... 119
Age Model Changes ...... 120
Tests of the Updated Age Model ...... 122
APPENDIX D. PALEO-HURRICANE INTENSITY PROXY ANALYSIS ...... 125
APPENDIX E. TRACE ELEMENT DATA ...... 127
vii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LIST OF TABLES
Table 1-1. Summary of Screening System Steps ...... 21
Table 2-1. Historical intensity, precipitation, and best track data for Atlantic tropical cyclones
(NOAA Tropical Prediction Center). Intensity categories represent the maximum intensity at or
prior to landfall in Central America, such that 0 indicates tropical storms, and values 1-5 indicate
Saffir-Simpson hurricane intensity categories 1-5. Total rainfall during storm days is reported
from the Central Farm meteorological station, less than 15 km from the cave site ...... 33
Table 4-1. Data Sources ...... 76
Table 4-2. SOI Lag Cross-Correlation Coefficients ...... 77
Table A l. Results of a multiple linear regression to investigate storm and speleothem factors
related to the signal size of measured 8lsO value excursions ...... 126
viii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LIST OF FIGURES
Figure 1-1. Transmission of cyclogenic stable isotope signal in dual porosity system ...... 22
Figure 1-2. Conceptual model of the effect of epikarst thickness on speleothem sensitivity to
brief storm events ...... 23
Figure 2-1. Location map indicating the stalagmite collection site, cave Actun Tunichil Muknal
(star) and nearby tropical cyclone storm tracks (1978 - 2001) ...... 34
Figure 2-2. Photographs of stalagmite ATM7 showing depth of radiometric dating samples and
micro-milling track ...... 35
Figure 2-3. Recent tropical cyclones near Belize with stalagmite 5180 timeseries, 813C timeseries
with SOI, and storm proxy model results ...... 36
Figure 2-4. For locally active hurricane seasons, S18Ovalue detail across corresponding annual
couplets are shown to highlight cyclogenic excursions (A-K) and background isotopic
variability ...... 37
Figure 3-1. AVHRR satellite imagery showing atmospheric transport of El Chichon tephra
imaged on April 5, 1982 ...... 56
Figure 3-2. Photographs of stalagmite ATM7, and trace element sampling tracks ...... 57
Figure 3-3. Chart of LA-ICPMS EOF factor loadings for the first three EOF modes ...... 58
Figure 3-4. Timeseries plot of LA-ICPMS variance in the first three EOF modes ...... 59
IX
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 3-5. ATM7 Scanning XRF track and EOF results ...... 60
Figure 3-6. Monthly weather observations at Central Farm, Belize, showing water balance
(difference between precipitation and pan evaporation) in gray, and mean maximum temperature
in black. Two hydrological years are highlighted: a veiy wet year 1979 (A), and a slightly dry
year 1981 (B) ...... 61
Figure 3-7. LA-ICPMS EOF results for quiet latter period (1984 - 2001) ...... 62
Figure 4-1. ATM7 carbon isotope record and Southern Oscillation Index (SOI) ...... 81
Figure 4-2. Timeline of Data Sources. Abbreviations as in Table 1 ...... 82
Figure A l. Belize Climatology (Central Farm Meteorological Station, 1973-2005) ...... 113
Figure A2. Seasonality of Belize Hurricanes (1883 -2005) ...... 113
Figure A3. ATM7 137Cs activity profile ...... 119
Figure A4. ATM7 LA-ICPMS species concentrationsATM7 LA-ICPMS species concentrations
(counts per second normalized toCa) ...... 128
Figure A5. ATM7 LA-ICPMS data comparison with first three EOF modes (y-axis). EOF
modes are plotted as loading factors; P31, Na23, and Sr88 are plotted as counts per second
normalized to Ca ...... 133
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ABSTRACT
RECENT EXTREME EVENTS IN A TROPICAL STALAGMITE: MULTI-PROXY
RECORDS AND ANALYSIS OF ECOSYSTEM 513C VALUE SENSITIVITY TO WEAK
CLIMATE FORCING
by
Amy E. Benoit Frappier
University of New Hampshire, September, 2006
Speleothems are emerging as important and detailed chronological records of environmental
change. Integrating exploration of modem speleothem records of environmental variability,
related forcing factors, and proxy biogeochemical characteristics reveals gaps in current
understanding and suggests new potential proxies. This dissertation demonstrates the potential
for developing novel proxy records of past regional environmental extremes, such as tropical
cyclones, explosive volcanism, and enhanced seasonal contrasts from speleothems that are
sensitive to transient infiltration events through very high-resolution, multi-parameter analysis of
a rapidly growing, fracture-fed tropical stalagmite. A series of sample screening steps were
developed prior to stalagmite collection in the field in order to increase the likelihood of
collecting suitable samples sensitive to tropical cyclone precipitation events. Once the correct
annual dating was established for a recent period (2001-1978), a weekly-monthly record of
stalagmite stable isotope ratios showed low stable oxygen isotope excursions that corresponded to
the historical record of tropical storm strikes in the region. Furthermore, excursion amplitude was
related to the maximum intensity of storms prior to landfall. The same stalagmite contains a trace
element record of the El Chichon eruption, validated by both Laser Ablation Inductively Coupled
Plasma Mass Spectrometry (LA-ICPMS) and Scanning X-ray Fluorescence (S-XRF). A
xi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. dominant, broad spectrum trace element perturbation was recorded by both methods at 1982,
coincident with a major explosive eruption of the nearby trachy-andesite El Chichon volcano in
Chiapas, Mexico in April of that year. This result demonstrates the ability of stalagmites to
record proxy evidence of a major regional tephra fallout event. The LA-ICPMS technique
showed greater discriminating power between volcanic and extensive rainfall signals. A lag
correlation analysis of weak El Nino forcing in Belize using a large suite of meteorological and
satellite datasets required removal of seasonal variance in order to detect any El Nino response.
Individual variables responded independently to El Nino, and no coherent lag timescale was
evident. Analysis of biological factors including the belowground community will be required to
assess the non-linear processes linking small changes in temperature and moisture extremes to
large carbon isotope variations. Promising new proxies for tropical cyclone activity and low-
latitude explosive volcanism may emerge from this work.
xii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. INTRODUCTION
As Earth’s climate system comes into better focus, it is increasingly clear that solutions to
many outstanding scientific problems are predicated on a more complete understanding of
variability and feedback mechanisms within the Earth system. Of particular concern is the fact
that climate models that simulate the “correct” recent climate for incorrect reasons may yield an
inaccurate view of future behavior of the climate system. Potentially critical feedback effects are
not now included in global circulation models, such as the tropical cyclone effect on ocean
mixing and heat transport (Emanuel, 2002). Low-latitude volcanic forcing produces different
effects than high-latitude eruptions, yet tropical volcanic activity is not well constrained. Proxy
records that constrain such feedbacks, forcings, and variability are positioned to advance the state
of the science.
However, not all proxy records are equally well-positioned to improve current
understanding of climate system dynamics. When seen through the lens of paleoclimatology, our
view of past climate system behavior is inverted with respect to the modem climate system
viewed through the lens of surface- and satellite-based observational networks. Our view of pre
historic climate is dominated by annual to millennial-scale marine and polar perspectives, but our
view of modern climate is dominated by synoptic terrestrial meteorological observations from the
tropical regions and northern hemisphere mid-latitudes. Similarly, the understanding of current
climate variability has been shaped primarily by daily land-based observations; mobile shipboard
observations are relatively less common. In contrast, the current understanding of paleoclimate
variability has been shaped primarily by low-resolution marine sediment cores, plus a handful of
high-resolution ice core proxy records from the polar and high altitude regions. Only very
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. recently has global satellite remote sensing technology provided comprehensive coverage of
many environmental parameters of interest. Modem and paleo perspectives on the climate
system can be brought into better alignment by generating multi-proxy paleoclimate records with
annual or better temporal resolution from terrestrial lowlands, particularly in the tropical and sub
tropical regions (Briffa et al., 1995; Crowley, 2000; Jones et al., 2001; Mann et al., 1999).
Stalagmites are rapidly emerging as sources of high quality proxy records of low-latitude
terrestrial environmental change (Bertaux et al., 2002; Bums et al., 2002; Lachniet et al., 2004a;
Lachniet et al., 2004b; Lundblad and Holmgren, 2005; Wang et al., 2001; Yadava and Ramesh,
2005). A number of advantages make speleothems particularly sought after, including spatial
distribution, relative ease of precise radiometric dating, high temporal resolution, and duration of
record. Stalagmites are able to record seasonal and sub-seasonal variations over long periods of
time, like small paleo-weather stations (Bertaux et al., 2002; Brook et al., 1999; Fairchild et al.,
2001; Fairchild et al., 2000; Frappier et al., 2002a; Frisia et al., 2003; Hellstrom and McCulloch,
2000; Hou et al., 2002; Huang et al., 2001; Ihlenfeld et al., 2003; Kong et al., 2003; McMillan et
al., 2005; Pan, 1999; Qin et al., 2000; Roberts et al., 1998; Treble et al., 2003; Treble et al., 2005;
Treble et al., 2002). Many different signals can be measured in the same samples, providing a
rich view of climatic and biogeochemical events and trends over time. Stalagmites are thus
increasingly viewed as the “ice cores” of the tropical lowlands.
However, stalagmite proxy interpretation is still maturing. The complex biogeochemical
dynamics of caves are less well understood compared to many other more established, simpler
and more direct proxy systems. Despite relatively early recognition of their potential value
(Bogli, 1980; Broecker et al., 1960; Dreybrodt, 1980, 1981; Harmon et al., 1979; Hendy, 1971;
Pan, 1999), development of speleothem proxies was delayed in part by the complexity of the
spelean environment. Cave dripwater delivery rates often vary across a range of timescales, and
the chemical and isotopic composition of cave seepage water is spatially and temporally variable.
Extension rates are often slow, requiring very high-resolution analytical procedures and dating
1
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. techniques to obtain better-than-decadal sampling. As speleothem exploration and modem
calibration studies have expanded, understanding of speleothem depositional processes and links
to the epikarst and surface environment have also grown (Ayalon et al., 1998; Baker and
Brunsdon, 2003; Baskaran and Krishnamurthy, 1993; Dorale et al., 1992; Fairchild et al., 2000;
Frisia et al., 2000; Genty et al., 2001; Johnson and Ingram, 2004; Kaufmann, 2003; Kaufmann
and Dreybrodt, 2004; McMillan et al., 2005; Railsback et al., 1994; Sancho et al., 2004; Swart et
al., 1991; Tooth and Fairchild, 2003; Zhang et al., 2004). Interpretation of carbon isotope ratios
and trace element variations in speleothems remains especially problematic (Fairchild et al.,
2006; McDermott, 2004; Mickler et al., 2004a). In particular, higher-resolution studies show
great promise for uncovering indicators that enable investigators to distinguish between various
controls on proxy variations, delimiting cases in which various controls cannot be distinguished,
and constraining sensitivity to various forcing factors.
Integrated studies of speleothem stratigraphic records, environmental forcing and extreme
events, as well as forest-soil-cave biogeochemical dynamics, are poised to expand the range of
questions that speleothem records can address. Each of the following chapters addresses a
different aspect of an integrated investigation of the relations between sub-annual to decadal
proxy records from a tropical stalagmite and the involvement of forcing factors and cave-forest
biogeochemical system. Two studies describe the development of new high-resolution modem
calibration-based proxies of environmental extremes, and two studies investigate the importance
of environmental forcing and epikarst channel filter characteristics to the resulting stalagmite
proxy records. Together, this work points toward a number of new directions for ongoing
research on speleothem proxy systematics and paleoenvironmental variability. It is anticipated
that high-resolution speleothem records will contribute valuable new insights to paleoclimatology
and Earth systems science.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER I
A THREE-STAGE SCREENING SYSTEM TO SELECT STORM-SENSITIVE
STALAGMITES
Abstract
We describe the field and laboratory speleothem screening process applied by Frappier
and co-workers in a successful effort to select a stalagmite suitable for high resolution stable
isotope paleotempestology. Field and laboratory criteria were combined in series, to screen out
candidates whose characteristics indicated lower sensitivity. Together, this screening system is
expected to increase the likelihood that the selected stalagmite would contain large, measurable
stable isotope anomalies related to tropical cyclone precipitation. This particular screening
process is an important methodological consideration for researchers attempting to replicate the
original speleothem stable isotope paleotempestology study, or to apply the technique elsewhere.
Furthermore, the overall approach to stalagmite selection presented here can be adapted readily
by other investigators in support of different scientific goals.
Introduction
Paleo-environmental research is based upon connecting the impact of environmental factors of
interest with measurable properties of a particular proxy archive. Any proxy signal of interest to
paleo-environmental investigators is transmitted to the proxy record through a channel
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. consisting of a series of intervening processes (Brown, 1987). The conduit pathway serves as a
filter that can amplify, attenuate, interfere with, add noise to, eliminate, and/or simply the original
environmental signals. A scientific instrument (in this case, a proxy) can provide an unobstructed
view of the items acting on it despite the characteristics of the proxy archive itself and related
channel attributes (Dretske 1981, in Brown, 1987). The common process of instrument
calibration is a physical application of this concept. Proxy signals are most reliable where the
translating channel is simple and stable with a high signal to noise ratio. The resulting proxy
transfer functions are valid so long as the channel remains unchanged through time, facilitating
reliable paleo-environmental reconstruction. The assumption of system stationarity is central to
paleoclimate reconstruction in general. Replication studies of some proxy systems have
established broad applicability of proxy signal-channel transfer functions across a range of
locations and ages (Cuffey and Steig, 1998; Grootes et al., 1993; Mann, 2002). On the other
hand, rare but valuable quantitative studies of interruptions and distortions found in the proxy
record (Cuffey et al., 1994; Lohmann, 1987; Lohmann, 1988; Shackleton and Opdyke, 1973;
Spero et al., 1997) have illuminated the understanding of “ancient mechanisms of environmental
change, metabolism, or diagenesis encoded in proxy signals” (Weedon, 2003).
Multi-proxy paleo-environmental archives from secondary carbonate mineral deposits in
caves, or speleothems, are being developed rapidly (Fairchild et al., 2006; McDermott, 2004). In
cave depositional settings, the signal channel is in part, quite literally a conduit that transmits
seepage water along pathways from the surface to the cave. As a result of spatial heterogeneities
in conduit hydrology, bedrock, surface topography, soil properties, ecosystem structure, cave
geometry and microclimate, and seepage water chemical composition, each conduit has a unique
combination of vadose/kinetic processes that has the potential to induce considerable variations
between speleothem records (Dorale et al., 2002b). A replication test should result in essentially
identical proxy records only in the special case where differences in kinetic/vadose zone
processes, or “noise”, are negligible relative to the proxy signal of interest (Dorale et al., 2002b).
4
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Replication of proxy records thus lends confidence to interpretation as indicative of primary
environmental variables of interest (Wang et al., 2001).
Existing replication studies have found both congruence and idiosyncrasy in speleothem
records. Similar stalagmites cannot always be identified in every cave, and speleothem
replication has often been limited by practical, ethical, and resource considerations (Lundblad and
Holmgren, 2005). In the field, speleothem collection is typically subject to severe restrictions
from conservation ethics, consideration of cave aesthetics, and/or permits. The quality of internal
speleothem stratigraphy cannot generally be determined in the field, with the inevitable result that
some collected samples turn out to be useless for paleo-environmental studies. Furthermore, age
differences often play a role when the target of the study is ancient, inactive speleothems. It is
not unusual for the few promising samples remaining in a collection to differ widely in age,
making replication impossible without additional field sampling.
Even when fairly well-replicated records are present, contemporary stalagmite records
from the same cave also can reveal different proxy signals. Lack of convergence has been taken
by some as a sign that speleothems may be unreliable for paleo-environmental studies (e.g.
Betancourt et al., 2002). In this view, individual stalagmites may have rather idiosyncratic
sensitivities, and replication of paleo-environmental records is hindered by the very complexity of
the spelean depositional environment.
Yet, a growing number of studies have derived valuable insights into paleo-
environmental change from attributing differences between speleothem records to surface or
vadose zone effects. In some cases, proxy record differences can be attributed to field
relationships such different hill slope aspect above the chambers where stalagmites were collected
(Denniston et al., 1999). Denniston and colleagues contrasted two stalagmite records, utilizing
the water balance differences between stalagmites from cave chambers with different surface
aspect in order to distinguish different climatic regimes. Genty and co-workers found that soil
richness (or soil respiration rate) controlled 513C value amplitude in a suite of speleothems from
5
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. France (2003). Another group tracked the relations between precipitation and cave dripwater
stable isotope composition, finding that two types of vadose pathways feeding different cave
drips were sensitive to different aspects of weather variability (Ayalon et al., 1998). Other studies
have found dripwater variability related to land-use differences above a cave system, seasonal
cave ventilation shifts, soil composition, and distance from cave entrance (Baker, 2002; Baker
and Brunsdon, 2003; Musgrove and Banner, 2004; Spotl et al., 2005). In other cases, non
equilibrium factors have been implicated in the origin of idiosyncrasies in individual speleothem
records that diverge from signals common to other formations (Mickler et al., 2004a). Thus,
detailed observations of field relations and/or modern cave system process studies provide critical
aids to proxy interpretation.
As the biogeochemical factors driving stalagmite differences become better understood,
this diversity of sensitivity may turn out to be a great strength, when detailed field observations
are available. Failure of the replication test can thus indicate less desirable stalagmite types to
avoid in future analysis. Reports relating field conditions to speleothem characteristics provide a
means for distinguishing between stalagmite types that have conduits sensitive to different
aspects of paleoclimate variability (Ayalon et al., 1999; Ayalon et al., 1998; Tooth and Fairchild,
2003). Different kinds of speleothems thus provide different perspectives on the same history,
much the same as different species, lake types, and glaciers have been found to exhibit
sensitivities to particular environmental factors. In this view, the observed variation between
stalagmites presents an opportunity for more precise paleoclimatology - if each type can be
identified unambiguously.
Indeed, one expects formations that share specific signal properties to share certain
related channel characteristics. If strong links can be established between field characteristics and
proxy signal outcomes through cave monitoring and experimental studies, then more efficient
speleothem sample collection would advance the scientific value of speleothem proxy records.
Such information can be applied in the field in order to 1) avoid collecting unsuitable samples for
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. investigating the environmental variable of interest, and 2) enhance sensitivity of the target proxy
record to the phenomena of interest given the practical limitations of sampling and analytical
techniques. Similarly, using non-destructive or minimally destructive methods to target
sectioning and analytical foci can be particularly useful in the pre- and post-collection phase
(Mickler et al., 2004a; Mickler et al., 2004b). The following section describes a three-stage
stalagmite screening system that was used successfully to select a stalagmite sensitive to tropical
cyclone precipitation events.
A Three-stage Screening System to Select Storm-sensitive Stalagmites
We wished to develop a screening system to select, from the large number of stalagmites
available in the field, a small number of samples suitable for analysis as the basis for a paleo-
hurricane research study. Here, a case study is presented that outlines a three-stage screening
method used to select stalagmites used to develop a proxy record of individual tropical cyclone
rainfall events (Chapter 2). The stepped stalagmite screening approach presented here is a
variation on one commonly applied by paleoclimatologists to most expeditiously meet scientific
objectives, in which field sites are screened prior to and during fieldwork, and collected materials
are further screened in the laboratory. Prior to sample collection, a series of stalagmite screening
steps were designed to maximize the likelihood of recovering samples containing tropical cyclone
proxy signals that would be measurable using available analytical techniques. In developing the
sample screening system, related studies were used to link 1) the tropical cyclone precipitation
signal to seepage water, 2) the cave-epikarst system to seepage water chemistry, and 3) the cave-
epikarst system to speleothem proxy records. The result of this study (see Chapter 2) supports the
importance of the screening system described below for speleothem paleotempestology
applications.
7
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. On the basis of existing research linking field relations to cave hydrology and spelean
geochemistry, a series of factors were developed that should have increased the likelihood of
recovering samples containing the target signal, in this case the 8lsO anomaly of tropical cyclone
precipitation. The screening approach was chosen to filter out candidate stalagmites with lower
sensitivity. We screened candidates at three steps: karst regions, cave sites, and finally individual
stalagmites, both in the field and after initial analytical assessment. A sequential list of contra-
indicators was developed in order to reject less-qualified candidates, and candidates that passed
each stage were further subjected to the next step in the screening process. Using the
considerations described below as a guide, I selected a region, cave, and stalagmite that resulted
in a proxy record that in fact contained the target hurricane proxy records (Chapter 6).
A number of factors were recognized that could affect our ability to resolve the expected
tropical cyclone precipitation signal given the limitations of our selected analytical methods. For
each problem, we developed criteria to reject less promising candidates, described below. To
guide site and sample selection, the potential problems and criteria for each screening step was
organized in three practical stages: region, cave, and stalagmite (Table 1-1).
Potential Problems: Region
1. The most basic issue is that a speleothem proxy for tropical cyclone events can only be
found where caves and hurricanes co-exist. Thus, the first consideration was to reject karst
regions outside the hurricane belt. We favored karst regions in areas where tropical
cyclone strikes were high during recent decades, such as Florida, the Yucatan Peninsula,
Puerto Rico, Hispaniola, and Cuba.
2. We expected proxy signals from individual hurricane precipitation events to persist only
briefly. Earlier studies that analyzed speleothems at seasonal to annual resolution were
8
with permission of the copyright owner. Further reproduction prohibited without permission. unable to resolve individual storm signals (Malmquist, 1997; Schwehr, 1998). Tropical
cyclones can precipitate a week to a month’s equivalent rainfall. In order to meet our
target calcite micro-sampling rate of 50 samples per year with the available 20 micron
sampling technique, we required a relatively rapid average stalagmite extension rate of
~lmm yr'1. The relatively rapid chemical weathering rates in the tropics increased the
likelihood of collecting rapidly growing stalagmites from lower latitude caves as compared
to karst regions at higher latitude. Furthermore, snow has a low stable isotope composition
similar to tropical cyclone precipitation (Lawrence and Gedzelman, 1998). Avoiding
temperate zone karst regions also enabled us to avoid potential confusion between the
tropical cyclone and wintertime precipitation. On this basis, we rejected karst regions in
the temperate zone for our initial study.
3. We required stalagmites with annual layering in order to accurately date stable isotope
variations. A seasonal water deficit seems to contribute to annual calcite banding; thus we
rejected stalagmites from regions without a dry season.
4. Stormwater from within the rain shield of a tropical cyclone has a distinct isotopic
signature. A study of the isotopic composition of summer precipitation in Texas showed
that the precipitation stable isotope anomaly produced by tropical cyclones is large (~6 %o)
relative to the analytical precision of modem mass spectrometry (<0.1 %o) (Lawrence,
1998; Lawrence and Gedzelman, 1996; Lawrence et al., 2002; Lawrence et al., 1998). The
amount of SlsO value variation from tropical cyclones is comparable to typical seasonal
variability in tropical precipitation in Central America, where data has been collected
(Lachniet and Patterson, 2002). Using stoichiometric reasoning based upon speleothem
calcite precipitation equations (Hendy, 1971), the calcite 5lsO anomaly produced by a -6% o
9
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. precipitation anomaly would be about -2%o, still easily resolved by a Kiel carbonate device
and Finnigan MAT252 gas source stable isotope mass spectrometer. However,
transmission of tropical cyclone water to the cave could be affected by rainfall amount and
stormwater infiltration. David Malmquist (personal communication) indicated that tropical
cyclones typically produced very little precipitation on the low-lying island of Bermuda.
We determined that a record of tropical cyclone precipitation events was more likely in
regions where storm rainfall totals are large, i.e. areas of orographic relief (Roe, 2005).
5. Soil thickness is an important factor affecting the chemical and isotopic composition of
infiltrating groundwater. Meteoric water passing through thicker soils is more altered than
water that has passed through only thin soil cover. The soils of most limestone regions of
Florida, Central America, and the Caribbean are relatively thin; thus, these regions are
good candidates for high-resolution isotopic investigations (Ford and Williams, 1989).
We selected karst regions for fieldwork by combining the above criteria with the list of
areas for which we could obtain access and speleothem collection permits as well as export
permits. Using these criteria, we selected the Boundary Fault Karst region in central Belize for
our initial project. The region has a distinct dry season, is located in the foothills of the Maya
mountains, many large caves, and a historical tropical cyclone recurrence interval of about 3
years.
Potential problems: Cave Configuration and History
6. Tropical cyclone water infiltration and the soil buffer. For our purposes, where sufficient
precipitation is generated, stormwater must infiltrate in sufficient quantity to produce a
measurable calcite layer on speleothems. The clay-rich soils typical in karst regions often
10
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. transmit water slowly and reduce infiltration. However, soils are thinner in upland areas
with steep slopes (Brenner, 1978). We determined that a record of tropical cyclone
precipitation events was more likely in caves under thin soils, rejecting caves in low-lying
areas with low slope angles.
7. Cave entrance effects. Although deep cave environments are very stable, light levels,
relative humidity, air motion, and temperature may vary considerably near the cave
entrance (Mickler et al., 2004a; Spotl et al., 2005). Variations in relative humidity and
temperature in the cave can lead to kinetic non-equilibrium effects that interfere with the
speleothem record of stable isotopic composition of seepage water (Hendy, 1971). In
general, cave environments are stable deeper than 10m below the surface and greater than
50 m from entrances; however, this can vary with regional climate and cave morphology
(Hill and Forti, 1997). Large caves with interior rooms were thus preferred, and small
caves were rejected to avoid stalagmites sensitive to unstable cave environmental
conditions.
8. Flooding. Caves and cave rooms subject to flooding are generally avoided for speleothem
paleo-environmental studies (Hill and Forti, 1997). Abrasion and corrosion from
floodwaters can erode previously deposited calcite. Even when physical damage is
minimal, no dripwater deposition can take place while formations are submerged. After
floodwaters drain away, deposition may be further delayed until mud and debris are
washed from the speleothem surface. Floodwaters can also deliver residual clays with high
Th concentrations to stalagmite surfaces, making U-Th dating more difficult. Flood
evidence is often easily recognized in the field, including mud layers, ripple marks,
fragments of carbonate and organic debris draped over surfaces, horizontal bathtub-like
11
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. rings on the walls, and objects lodged into crevices by flowing water. We avoided any
caves and cave rooms with evidence of recent flooding.
9. Evaporation and wind. High humidity inhibits evaporation in cave chambers and promotes
isotopic equilibrium calcite deposition (Hendy, 1972). Internal cave rooms located far
from entrances, and with standing water represent favorable humidity conditions. A
hygrometer was used to quantitatively measure relative humidity in candidate caves. We
planned to reject cave rooms with relative humidity < 85-90%. However, this criterion
was of limited use in this case because such dry conditions were not encountered in any
candidate caves during this project. Similarly, air currents in caves can increase
evaporation and related kinetic effects in speleothem records (Hill and Forti, 1997).
Constricted passageways often inhibit air flow. Air currents may not be apparent to cave
visitors if they appear only seasonally, at a certain time of day, or associated with particular
weather patterns (Spotl et al., 2005). However, the presence of air currents that
substantially affect calcite deposition can be recognized often in the field (Hill and Forti,
1997). Although we felt no air movement in most caves, we avoided sampling in cave
rooms with speleothem evidence of air currents, particularly asymmetric forms (Hill and
Forti, 1997).
10. Vandalism. Although we did not have to reject any caves on this basis, we also planned to
avoid caves with substantial vandalism, as this may have reduced the number of available
stalagmites.
11. Epikarst conduit hydrology. We also wished to avoid two problems related to hydrologic
pathways through the epikarst. If stormwater travels a short distance through large
conduits and reaches the cave too quickly, it can be undersaturated with respect to calcite,
leaving behind no measurable stable isotope record. On the other hand, while stormwater
12
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. that takes a long and circuitous route through the bedrock’s primary pore network will
certainly deposit spelean calcite upon reaching the cave, the stormwater will be extensively
homogenized with other vadose groundwaters (Ayalon et ah, 1998). In this case, the
stormwater will leave behind a highly attenuated stable isotope record that lacks individual
storm events, but that nevertheless reflects long-term storm activity. Where feasible, it is
desirable to obtain long-term seepage water monitoring data from several candidate drip
sites prior to speleothem collection. However, in many instances, dripwater monitoring
data is not obtainable prior to collection because of funding limitations and the logistical
complications that arise from working in remote cave areas. Because long-term seepage
water monitoring data was neither available nor practical in this study, a field observations-
based approach (See #10 and 13, below) was designed to select stalagmites with seepage
pathways that that avoid both extremes.
12. Conduit hydrologic pathway: Calcite undersaturation. Following a significant storm event,
the increased height of the groundwater column is expected to cause cave drip rates to
increase within hours. A measurable tropical cyclone precipitation signal requires that the
storm water percolate slowly enough to reach supersaturation with respect to calcite before
reaching the cave. Drips that are undersaturated during periods of high flow (such as storm
events) can dissolve calcite in the top of previously-deposited stalagmites, forming a drill
hole morphology (Hill and Forti, 1997). While exploring a number of candidate caves in
the field, we noted that shallower caves with thin limestone roofs were more likely to
contain stalagmites with drill cups, whereas this morphology was absent in deeper caves.
The correlation between epikarst bedrock thickness and conduit infiltration time is not
absolute. Nevertheless, in general, as cave overburden thickness increases, seepage water
residence time and calcite saturation state both increase, and waters infiltrating from
13
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. different precipitation events and seasons becomes increasingly homogenized (Ayalon et
al., 1998). Note that the issue of epikarst thickness is separate from that of soil thickness,
described above. The presence of stalagmite drill cups indicated undesirable shallow caves
where dripwater was likely to be corrosive during periods of high flow, such as hurricane
events. However, it is important to distinguish between drill holes caused by
undersaturated drips and splash cups related to the height of the drop and rate of the drip
(Fig. 1-1).
We used the above criteria to select the cave Actun Tunichil Muknal for sample
collection. The air in the upper level cavern where speleothems were collected for this summer
had a temperature of 25°C and relative humidity of -90%. Stalagmites are ideally located far
from cave entrances in rooms with restricted air-flow to avoid associated complications in
mineral deposition and isotopic fractionation (Hill and Forti, 1997). An advantage is that this
cave site has been mapped, and has been visited regularly for the past several years, so there is
some local knowledge of environmental variability.
Potential Problems: Stalagmite
Only stalagmites were considered for this study. Speleothems appear in many forms, but
stalagmites are preferred for paleo-environmental studies because of their relatively simple
stratigraphy (Hill and Forti, 1997). Although a stalagmite’s internal growth form cannot
generally be assessed in the field, the pool of available stalagmites can be reduced somewhat
through observations of external characteristics.
14
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 13. Irregular growth form. A stable growth form tends to result from a stable seepage conduit.
Highly asymmetrical forms may indicate air motion or an unstable drip location that could
affect the proxy record. Stalagmites with irregular form are to be avoided.
14. Inactive stalagmites. Ancient stalagmites were not the target of this modem test-of-
concept study. Stalagmites without wet surfaces indicating active formation were excluded
due to interest in recovering modem material deposited during a period for which the
history of storm strikes is known.
15. Conduit hydrologic pathway: stormwater homogenization. A measurable tropical cyclone
precipitation signal also requires that the storm water reach the cave as a coherent slug of
isotopically distinct water. Limestone bedrock usually has “double porosity” hydrology,
meaning that some groundwater travels relatively rapidly through fractures, while the rest
moves slowly and circuitously through unfractured permeable bedrock (Ford and Williams,
1989). Different residence times and different dripwater conduit pathways produce
differences in the isotopic composition of the groundwater. Ayalon et al. (1998) found that
fast-drip water that entered a cave through fractures (after brief bedrock residence times)
retained isotopic identity related to the source precipitation. In contrast, slow-drip cave
water (with longer residence times) had a more homogenous isotopic composition, related
to the average accumulated seasonal stable isotope values. In an Israeli cave system,
mixing of water derived from these various sources within the top 40-50 m of the vadose
zone determines the isotopic composition of the groundwater reservoir (Ayalon et al.,
1998). Fracture-fed stalagmites are thus more likely to be sensitive to brief storm events
(Fig. 1-2). Fortunately, recognizing fracture-fed stalagmites is a relatively simple matter in
the field. Visible lineation traces on the ceiling and linear arrays of stalactites and/or
stalagmites indicated the presence of fractures. We rejected stalagmites without evidence
15
with permission of the copyright owner. Further reproduction prohibited without permission. of a fracture source for seepage water. Of the fracture-fed stalagmites available, those with
relatively slow drip rates were less desirable, indicating a less direct hydrologic connection
to the surface.
16. Dirty calcite. As discussed in the section on flooding, above, clays can contaminate calcite
with excess initial Th. So-called “dirty calcite” is more difficult to date precisely using U-
Th techniques. The presence of clays can often color speleothem calcite. Although many
other impurities can also affect calcite color, white stalagmites are thought to be less likely
to contain problematic clays.
17. Re-crystallization. Smaller crystals are more desirable than large crystals, which can
indicate re-crystallization or weak alteration of the speleothem (Kendall and Broughton,
1978).
For this study, six stalagmites were chosen on the basis of morphology, mineralogy, crystal
texture, drip water source, location in cave, soundness of structure, and effect of harvesting on
cave aesthetics. Of the six candidates that were collected, these were screened using simple
laboratory tests to select one candidate for micromilling and high-resolution stable isotope
analysis.
Lab Selection Procedures
18. Mineralogy. X-ray Diffraction (XRD) was used to detect the presence of aragonite, whose
stable isotope systematics differ from that of calcite. The chemical composition and
16
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. mineral structure of powdered stalagmite samples was determined by performing X-ray
diffraction using a Siemens D 5000 powder X-ray diffractometer. Powdered samples were
adhered to a quartz mounting plate using petroleum jelly. Each sample was rotated
through an appropriate 2-theta angle. The resulting XRD profiles were compared to
spectra from samples of known composition to test for mineralogical composition. In this
case, mineralogy was not very helpful in selecting formations because all speleothems
collected were composed entirely of calcite.
19. Stratigraphy. After stalagmites were halved and polished, the stratigraphy was described.
Speleothems with obvious hiatuses, or irregular growth form were not good candidates for
micro-milling.
20. Equilibrium test. Hendy tests were performed to screen for stalagmites deposited outside
of isotopic equilibrium with their source fluid (Hendy, 1971), and although the efficacy of
this test has been called into question (Mickler et al., 2004a), the test is likely valid in the
case of the ultimately selected stalagmite ATM-7 because of its wide bands, a result of
relatively rapid recent growth (see below).
21. Growth rate. Radiometric dating was used to measure the vertical extension rate of the
stalagmites. Because our samples were modem, we chose to analyze I37Cs and 210Pb.
Stalagmites with higher growth rates were preferred in order to maximize the temporal
sampling rate we could achieve with the available microsampling system.
17
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Stalagmite ATM7 was selected for high-resolution analysis. This stalagmite had the
highest drip rate in the field, and was located in a relatively inaccessible section of the upper
passage. ATM7 had smooth and relatively even layering, compact calcite with small crystals,
and a rapid growth rate in the most recent few centimeters of deposition, making it an excellent
candidate for our pilot study. Table 1-1 summarizes the screening system presented in this case
study.
Discussion
The first stalagmite selected using this screening system and analyzed at appropriate
temporal resolution successfully recorded recent tropical cyclone precipitation stable isotope
signals (Chapter 2). This initial success suggests that this screening approach is an important part
of the speleothem paleotempestology tool presented in Chapter 2. Not all stalagmites have the
appropriate characteristics to record tropical cyclone precipitation amount. Although the
stalagmite screening system presented here remains to be validated for different stalagmites and
cave sites, we strongly suggest that future efforts to replicate or apply this method for
paleotempestology should include the appropriate screening steps described above. In the future,
we hope to replicate this work using another stalagmite selected from the same cave site, Actun
Tunichil Muknal in central Belize.
More generally, while replication is a strong indicator of excellent proxy records (Dorale
et al., 2002b), it requires collecting a set of similar stalagmites. Some speleothem records with
different intrinsic sensitivities have the potential to contribute valuable new perspectives to paleo-
environmental studies. Applying explicit field and laboratory sample screening may improve the
odds of collecting stalagmites with sensitivity to the desired environmental variable(s), and
advance replication efforts. To achieve different scientific goals, other screening procedures
could be applied to target an appropriate subset of stalagmites.
18
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Furthermore, field observations can be critical to later interpretation of speleothem proxy
records, and it is difficult to over-emphasize the importance of field documentation. In some
cases, the overall physical configuration of the cave site, its history and climate may be relatively
easy to determine after the fact. However, careful attention should be paid in the field to
describing the exact speleothem location and micro-environment in detail and broad-brush,
because some observations (such as the rate of water flow on different sides of a stalagmite, any
water contributions from adjacent drips) cannot be easily re-constructed after removing the
formation. The scientific value of many speleothems in existing collections may be compromised
by incomplete descriptions and a lack of field provenance information. Where documentation is
lacking, any available information should be developed or compiled as soon as possible, in
cooperation with the individuals involved in fieldwork when possible, especially for speleothems
collected in past decades.
The reasoning behind field and laboratory stalagmite sample selection should be
evaluated in advance with respect to scientific goals. Although different investigators and groups
have developed rubrics and helpful observations used during field sample selection, these are
rarely published. Sharing selection approaches would advance speleothem paleo-environmental
studies in both a forward sense, with regard to sensitivity selection, as well as in an inverse sense,
with regard to record interpretation. Such information can also help to bring together parallel
efforts by paleo-environmental proxy interpretation experts and groups studying applicable
modem processes in cave, vadose hydrology, and soil systems.
Speleothems are a very promising source of high-resolution, multi-proxy continental
paleo-environmental records. Ultimately, information linking field relations to proxy records can
provide rich sources of new hypotheses for cave systems research. Sharing sample collection
approaches should advance ongoing stalagmite collection, replication, and interpretation efforts
by the broader scientific community while supporting cave conservation efforts. Attention to the
19
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. potential value of speleothems with different intrinsic sensitivities suggests a number of
promising avenues for future paleo-environmental research.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 1-1. Summary of Screening System Steps.
Step Factors Observations
Tropical karst region (bedrock maps) High TC landfall frequency NOAA National Hurricane Center Annual layering Climatology (Seasonal water deficit) High TC rainfall amount Orographic effect (Topographic maps)
Existing cave information Cave maps, reports, and other historical documents Vandalism Field observations and interviews Cave entrance effects depth of cave (cave maps, if available) RH hygrometer Flooding mud, debris, washing of formations Air currents formation deflection, constricted passageways Dual Porosity bedrock Field observations of fractures and decorations Epikarst thickness Field observations, map if available Soil thickness Field measurements
Actively growing Field observations (wet surfaces) Fracture-fed Field observations (fracture traces, linear arrays of stalactites) Shape, color, crystal size Field observations (broomstick or cone, white, small crystal size) Stratigraphy Polished cross-section. Mineralogy XRD Analysis 5180 and 513C analysis Dental Drill and Stable isotope ratio analysis 210Pb and 137Cs dating y-ray detector Hendy Tests Dental Drill and Stable isotope ratio analysis
21
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 1-1. Transmission of cyclogenic stable isotope signal in dual porosity system.
Blue indicates vadose water with background isotopic composition, red indicates tropical cyclone water. Purple indicates intermediate compositions.
Time 1 > 2 > 3 Background Storm Signal Storm Signal Signal Breakthrough Decay
22
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 1-2. Conceptual model of the effect of epikarst thickness (i.e. transmission rate or flow pathway) on speleothem sensitivity to brief storm events.
Bedrock
Fractu *es
Too thick * Bedrock Ceiling Thickness > Too thin Too slow < Infiltration time > Too fast Stormjii§rTcU M easurable Non-deposition [homogenized Storm Signal opdt^solution
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER II
A STALAGMITE STABLE ISOTOPE RECORD OF RECENT TROPICAL CYCLONE
EVENTS
Abstract
A 23-year tropical stalagmite record of recent stable isotope variations (1978 - 2001) at
monthly-weekly temporal resolution contains abrupt low excursions in the stable oxygen isotope
ratio (5lsO value) of calcite that correspond temporally with recent historical tropical cyclones in
the vicinity of the cave. A newly developed logistic discriminant model can reliably identify
tropical cyclone proxy signals using the measurable parameters 5180 value, 513C value, and single
point changes in S180 value. The logistic model correctly identified 80% of low 5,80 value
excursions as storm events and incorrectly classified only one of nearly 1200 non-storm sampling
points. In addition to enabling high-resolution tropical cyclone frequency reconstruction, this
geologic proxy also reveals information about the intensity of individual storm events. A multiple
regression predicted storm intensify (R2 = 0.48 p=0.030) using sampling frequency and excursion
amplitude. Consistent with previous low-resolution studies, we found that the decadal average
5180 value was lower during the 1990s when several tropical cyclones produced rainfall in the
area and conversely, was higher during the 1980s when only one storm struck. Longer,
accurately-dated, high-resolution speleothem stable isotope records thus may contribute a useful
new tool for paleotempestology to clarify associations between highly variable tropical cyclone
activity and the dynamic range of Quaternary climate.
24
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Introduction
Tropical cyclones, including hurricanes, cyclones, and typhoons, are among the most
deadly and costly natural hazards. Tropical cyclone intensity, frequency, spatial distribution, and
economic damages are highly variable and respond to an array of global and regional climatic
factors including the El Nino-Southern Oscillation System, or ENSO (Klotzbach and Gray, 2004)
and references therein, Landsea, 2000; Pielke et al., 1999; Tartaglione, et al. 2003). Conversely,
these storms may be an important yet poorly understood feedback in the climate system, possibly
controlling rates of marine poleward heat transport and thermohaline circulation (Emanuel,
2001). Despite the ongoing research effort to understand Earth’s climate system, associations
remain controversial between highly variable tropical cyclone activity and the dynamic range of
Quaternary climates (Emanuel, 2005; Giorgi et al., 2001; Elenderson-Sellers et al., 1998; Pielke et
al., 2005; Walsh, 2004; Walsh et al., 2004). Historical records from most tropical cyclone regions
contain few examples of the most catastrophic storms. Paleo-tempestological evidence of past
storms preserved in sedimentary and biogeochemical archives (Donnelly et al., 2001a; Donnelly
et al., 2001b; Liu and Feam, 1993, 2000, 2002; Nott and Hayne, 2001; Nott, 2003) has already
illuminated centennial-millennial variations in tropical cyclone activity (Eisner and Liu, 2003;
Eisner et al., 2000; Nott, 2004; Nott, 2003). Paleotempest proxies with annual-decadal resolution
could clarify the controversial problem of tropical cyclone hazard climatology and offer an
independent test of posited feedbacks among the climate system, thermohaline circulation, and
tropical cyclone activity (Emanuel, 2005). Toward that end, we present a new, high-resolution
tool for paleotempestology that uses the proxy record of past tropical cyclone rainfall preserved in
a calcite stalagmite, a type of speleothem or cave formation.
25
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Hurricane Stable Isotope Values and Cave Depositional Settings
The 5lsO value of average tropical cyclone precipitation is ~6%o lower than other summer
season rainwater (Gedzelman et al., 2003; Gedzelman and Lawrence, 1982; Lawrence, 1998;
Lawrence and Gedzelman, 1996; Lawrence et al., 2002), and perturbs the stable isotopic
composition of groundwater and streams in the storm’s wake (Lawrence, 1998). Recognition of
this natural isotopic tracer ‘spike’ by Lawrence (1998) led to exploration of potential paleo-
hurricane archives in corals (Cohen, 2001), fish otoliths (Patterson, 1998), tree rings (Miller et al.,
2003), and speleothems (Malmquist, 1997; Schwehr, 1998). Early attempts to develop a
speleothem isotope proxy for individual tropical cyclones were frustrated by low sampling
resolution (-annual). Nevertheless, these studies established significant decadal/multidecadal
anti-correlations between storm frequency and average speleothem calcite 5180 values, indicating
that intervals of enhanced tropical cyclone precipitation can depress speleothem isotopic
composition in affected regions on decadal timescales. A high-resolution speleothem isotope-
based paleotempest proxy would lend a number of complementary advantages to established
coastal paleotempest proxies, including relative ease of radiometric dating, high temporal
resolution, continuous deposition, storm intensity indicators, a detailed background climatic
record, and a stable depositional environment independent of Quaternary sea level variability.
Only tropical cyclones that produced significant rainfall could be recorded, and the proxy would
be limited to karst regions. Advances in micro-sampling technology for isotopic analysis of
carbonates (Carpenter, 1996; Frappier et al., 2002b) now enable us to examine the speleothem
record at sufficiently high resolution to detect proxy evidence of individual tropical cyclones.
A hurricane rainfall event over a cave results in a slug of low 5180 value water infiltrating
through soil and karst bedrock overburden; ultimately, an isotopic “spike” is recorded by growing
speleothems as a low §l80 value excursion. Inter-storm meteoric water with more typical, higher
§180 values provides the requisite isotopic contrast. To precipitate a measurable isotopic
26
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. excursion in speleothem calcite, the hydrologic pathway through the epikarst must 1) allow
infiltrating tropical cyclone water to reach supersaturation with respect to calcite, and 2) avoid
excessive homogenization of tropical cyclone water with other soil- and groundwaters. Even
absent excessive mixing, diffusion and dispersion during infiltration cause some tropical cyclone
water to be released slowly, affecting the isotopic composition of dripwater and calcite for some
time after the primary isotopic ‘spike’ passes through the conduit system. For individual storms,
the size of precipitation 5isO value anomalies increases with tropical cyclone intensity, and
decays with distance from the eye of the storm (Lawrence and Gedzelman, 1996; Lawrence et al.,
1998). The size of SlsO value anomalies at any site may reflect the “apparent local intensity” of
the storm, raising the possibility that the amplitude of individual stalagmite S180 excursions may
contain paleotempest intensity information. In contrast, tropical cyclones are unlikely to
substantially perturb dripwater S13C values. To investigate the suitability of speleothems as a
proxy archive for paleotempestology, we analyzed a stalagmite stable isotope record from Belize
over a recent 23-year period for which the history of nearby storm tracks and intensity is known
(Fig. 2-2-1, Table 2-1).
Methods
In January, 2001 we collected several actively growing stalagmites from the cave Actun
Tunichil Muknal in central Belize (additional site information is included in Appendix 1).
Stalagmite ATM7 was sectioned longitudinally, polished, and dated. Micro-sampling and stable
isotope analyses were performed at the Paul H. Nelson Stable Isotope Laboratory at University of
Iowa (Frappier et al., 2002, Appendix 2). Microsamples (approximately 0.02 to 0.05 mg of
CaCCL for each sample) were milled along the growth axis continuously at 20 pm resolution
(Fig. 2-2) using a computer-controlled microsampling device (Appendix 2). Analytical precision
27
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. on standards was better than ± 0 .1 %o for both S lsO and 8 13C values. All results are reported in per
mil (%o) relative to V-PDB.
Identifying and Dating Isotopic Excursions
We visually inspected the stable isotope record for the largest brief, low excursions in
5180 value not also associated with substantial declines in 813C value. It was necessary to account
for covariation between 8180 values and 813C values in order to distinguish the isotopic signature
of tropical cyclone precipitation from ENSO-related inter-annual climatic and ecological
variability (Frappier et al. 2002). We also updated and verified a previously developed age model
(Frappier et al., 2002b) to reflect new recognition of sub-annual layers (Appendix 3). We used the
resulting age model to identify the year in which each low 8lsO value excursion was deposited.
Results and Discussion
The ATM7 oxygen isotope record contains 11 large (> ~ 0.5 % o), short-lived, low
excursions in 8lsO values that are not coupled with similar decreases in S13C values (< 0.2 % o) and
are correlated with historical tropical cyclones near the cave (Fig. 2-3A, B). No similar SlsO value
excursions are evident during years lacking nearby tropical cyclones. These 8lsO value
excursions are associated with historical storms ranging in intensity from Tropical Storm to
Catastrophic Flurricane, and with storm tracks whose eyes passed the cave at distances of 40 to
370 km. The record also reflects inter-storm deposition, enabling accurate estimates of between-
storm intervals (Fig. 2-3 A, B). Multiple S180 value excursions occurred in years with multiple
storm strikes, as well as 1998 when Hurricane Mitch re-intensified as it passed the site a second
28
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. time (Fig. 2-2-1). In 1996, when Dolly and Kyle struck two months apart, the two excursions are
separated by several samples (Fig. 2-4). In 1995, when Opal and Roxanne struck two weeks
apart, the two excursions are separated by a single intervening point (Fig. 2-4). Hurricane Keith’s
stormwater signal was not recorded in 2000 as ATM7 was collected only three months after Keith
made landfall while the tropical cyclone-derived, or “cyclogenic” water was most likely still
moving down through the epikarst toward the cave. In agreement with previous studies
(Malmquist, 1997; Schwehr, 1998), decadal average 8lsO values in ATM7 were lower (higher) by
1.87%o ±0.19%o during the 1990’s (1980’s), a decade when many (few) tropical cyclones affected
the region (Fig. 2-3B).
To test the suitability of high-resolution stalagmite stable isotope records for detecting
unknown pre-historic tropical cyclones, we applied a simple numerical technique to distinguish
the proxy record of cyclogenic excursions from background variability in this dataset. Within the
11 historical excursions we identified as cyclogenic, we coded the sampling point with the lowest
5180 value as a storm event (1); all other sampling points were coded as non-storm (0). During
Hurricane Mitch’s long traverse across this region, the storm made landfall in both Honduras and
Yucatan; thus, we coded the two largest low S180 excursions in 1998 as storm events. A binary
logistic regression of sampling points identified as storm/non-storm with the predictor variables
oc* 18 Od (the difference in 8 18 O value between adjacent samples), 5 O 18 value, and 8 C value, 13
resulted in a probability function to quantify the likelihood that any sample represents a storm
event:
(1) Probability = -11.01 * 5,8Od- 1.85 * S180 ± 0.51 * SI3C - 12.86
Because of adjacent missing data due to a machine error on one side of the dual inlet
system, a cyclogenic excursion in 1993 was not included in this model as the 518Od could not be
29
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. calculated. Using equation (1), logistic probability values greater than 0.5 captured excursions
related to 8 of the 10 remaining isotopic excursions included in the model (Fig. 2-3D). Of over
1200 non-storm data points, the logistic probability model identified as a cyclogenic signal only
one sample that we coded as non-storm, with a logistic probability cut-off value greater than 0.5.
A third low SlsO excursion in 1998 identified by the model was also associated with Hurricane
Mitch. We can thus infer that multiple excursions can be produced by a single storm system that
produces multiple local precipitation events. Overall, this model reliably identified tropical
cyclone signals in an isotopic record with substantial background variability, and will serve as a
testable model for future speleothem paleotempestology records.
Investigating Potential Indicators of Storm Intensity
We examined relations between the proxy (low 5lsO value excursion amplitude) and the
characteristics of recorded storm events. We postulated that annual sampling frequency (S, #
stable isotope samples per year) could be an important secondary control on the amplitude of
measured excursions as a result of inter-annual growth rate fluctuations (Appendix 4). A standard
multiple regression using maximum storm intensity(I), proximal storm track distance (D), and S
as the independent variables explained 56.6% of variation in 5180 value excursion amplitude
(p<0.031) (Appendix, Table Al). The most important predictors of excursion amplitude were I
(semi-partial R2= 0.354) and S (semi-partial R2= 0.382), indicating that this application is limited
by microsampling technology. Surprisingly, excursion amplitude was not substantially related to
D (semi-partial R2= 0.025). We interpret these results to indicate that in this dataset, excursion
amplitude was primarily related to tropical cyclone intensity and was not substantially
confounded by storm track distance. This finding is surprising given observed radial stable
isotope gradients in tropical cyclones, but consistent with the observation that tropical cyclone
30
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. rainfall 5180 value is an integrated signal of the storm’s history (Lawrence et al., 1998). Storm
intensity was a better predictor of excursion amplitude than local precipitation amount (Appendix
4). Toward reconstructing tropical cyclone intensity from measurable predictor variables (S and
excursion amplitude), we performed a linear regression that explained 48.0% of variance in
maximum storm intensity (p<0.030). High-resolution pre-historic speleothem proxy datasets can
thus provide the basis for accurately reconstructing paleo-hurricane intensities. However, a robust
test of the models presented here for reconstructing paleotempest incidence and intensity was
precluded by the absence of an independent dataset of similar quality. A larger number of
recorded historical-era tropical cyclone events will be required to fully test the sensitivity of this
stable isotope proxy to various storm characteristics identified by Lawrence and co-workers
(1998) (e.g. various storm track distance and intensity measures, storm duration or short-term
intensity changes, storm radius, storm quadrant affecting the site).
Conclusions
The low 5lsO value excursions identified in this stalagmite record represent an accurate,
measurable high-resolution proxy for past tropical cyclone precipitation events, opening the door
to more detailed records of past hurricane frequency and intensity. The overall success of simple
statistical tools to distinguish cyclogenic b180 value excursions from a relatively noisy
background gives us confidence that individual tropical cyclone events can be reliably identified
in pre-historic speleothems. Additional independent stable isotope records and longer calibration-
period records are needed to test the proxies for tropical cyclone activity presented here. Close
agreement between this speleothem proxy archive and the historical record suggests that this tool
can bridge the gap between historical/meteorological hurricane observations (synoptic to multi-
decadal scale) and low-resolution (centennial to millennial scale) paleotempest proxy records.
31
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The speleothem proxy complements existing coastal paleotempestology proxies, lending
advantages related to ease of dating, independence from Holocene sea-level changes, and
potential applicability throughout the Quaternary. The proxy is limited by the availability of
storm-sensitive speleothems, stable isotope sampling frequency, and potential interference in
temperate regions from snow-melt infiltration(Dansgaard, 1964a). Only tropical cyclones that
produce significant rainfall in karst regions could be recorded. Between 1851 and 2004, this c
site alone was affected by 6% of major hurricanes and 8% of all named landfalling tropical
cyclones in the Atlantic Basin. Thus, several cave sites from karst regions across the Atlantic
Basin may may yield a representative record that reflects seasonal to centennial variation in
overall Atlantic paleotempest activity during Quaternary intervals of paleoclimatic interest. A
spatially and temporally distributed network of high-resolution paleotempest proxy records in
multiple tropical cyclone basins may ultimately contribute to coastal risk assessment and
detection/attribution programs aimed at investigating potential impacts of climate change on
tropical cyclone activity (Nott, 2004).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 2-1. Historical intensity, precipitation, and best track data for Atlantic tropical cyclones (NOAA Tropical Prediction Center). Intensity categories represent the maximum intensity at or prior to landfall in Central America, such that 0 indicates tropical storms, and values 1-5 indicate Saffir-Simpson hurricane intensity categories 1-5. Total rainfall during storm days is reported from the Central Farm meteorological station, less than 15 km from the cave site. * Combined rainfall from both Hurricane Mitch landfall events.
Landfall Distance between storm Storm Precipitation, Storm Date Intensity track and cave (km) Central Farm (mm)
Keith Oct. 2000 4 120 264
Katrina Oct. 1999 0 158 15
Mitch 2 Nov. 1998 0 369 *
Mitch Oct. 1998 5 282 246*
Kyle Oct. 1996 0 147 59
Dolly Aug. 1996 1 233 51
Roxanne Oct. 1995 3 311 43
Opal Sep.1995 0 248 68
Gert Sep.1993 0 86 24
Diana Aug. 1990 0 259 24
Hermine Sep. 1980 0 138 131
Greta Sep.1978 4 42 179
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 2-1. Location map indicating the stalagmite collection site, cave Actun Tunichil Muknal (star) and nearby tropical cyclone storm tracks (1978 - 2001). Black storm tracks indicate hurricanes; gray storm tracks indicate tropical storms (NOAA National Hurricane Center).
90 W 85° W ATLANTIC I !' Gulf of ...OCEAN I Mexico
20 N ' .Belfzer-f Caribbean Sea
A
i
34
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 2-2. Photographs of stalagmite ATM7 showing depth of radiometric dating samples, and micro-milling track. White circles show the depth of y-activity samples with positive l37Cs activity; gray circles indicate zero (undetectable) 137Cs activity. We interpret he onset of 137Cs y-activity to indicate local deposition of global fallout from atmospheric thermonuclear bomb testing after 1953. The polished cross section of the top portion of ATM7 shows the stable isotope micromilling track (blue outline), which was positioned to maintain alignment with the growth center and perpendicularity to growth banding throughout sampling (1978 -2001).
35
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 2-3. Recent tropical cyclones near Belize with stalagmite SlsO timeseries, S13C timeseries with SOI, and storm proxy model results. Horizontal axes for A, B, and D are identical (1978-2001); C is offset by a 1.5 year lead. Dating uncertainty for ATM7 is a few weeks to months; thus, events named in A and highlighted in B should not be expected to match exactly. A. Historical records of storm distance (black bars) from cave to nearby tropical cyclone storm tracks and local maximum intensity [red bars indicate Saflir- Simpson intensity categories TS (tropical storm) to H5 (Category 5 Hurricane)]. Blue circles denote storm rainfall totals from three meteorological stations near the cave site. B. ATM7 5180 and S13C values are shown in black and blue, respectively. Cyclogenic low 5lsO value excursions (A-K) are highlighted. Note the different vertical scales for §lsO values and S13C values. C. The Southern Oscillation Index (SOI) is shown inverted. Major El Nino events (vertical bars) during the period of record are associated with high 513C and 5lsO values in ATM7. D. Logistic regression model results to discriminate cyclogenic excursions from background isotopic variation and the amplitude of isotopic excursions. Solid circles denote excursions the model classified correctly as cyclogenic. Open circles (1980, 1995) denote excursions classified incorrectly as non-cyclogenic. One data point the model classified as cyclogenic (black) was associated with Mitch, but not a unique historical storm.
300 - Distance (km) Diana 2 0 0 - Gtsls P SfE KatrinaKeilh 100 - TS - H I - H2 - (-20D % H 3 - H 4 - "3 BS J Intensity 40D a
GF
b -3 - *0 -j, --9 m
1S60 19Q5 1995
-30
1960 19B5 51 1980 1 9B 5 199D 1 9 9 5 2000 0 ■3 s TJ~T M £ >
2 Cl.E
36
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1998.5 1999 •4 CM to 1991 1997.5 1990.5 •5 •6 •4 ■5 •4 to 1981 k 1997 1980.5 1996.5 ■5 •3 ■4 1996 1995.5 1978.5 1979 Q.
_ •3 •2 •4 -6 to >! ^ O CO Figure 2-4. Forlocally active hurricane seasons, 8lsO valuecyclogenic detail across excursions corresponding (A-K) annualand coupletsbackground are isotopicshown variability. to highlight U) ' j
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER III
EMPIRICAL ORTHOGONAL FUNCTION ANALYSIS OF MULTIVARIATE STALAGMITE
TRACE ELEMENT DATA: DETECTING THE 1982 EL CHICHON VOLCANIC ERUPTION
AND OTHER ENVIRONMENTAL EXTREMES
Abstract
We used Empirical Orthogonal Function (EOF) analysis to evaluate a recent Belize
stalagmite record of trace element concentrations obtained using Laser Ablation-Inductively
Coupled Plasma Mass Spectrometry (LA-ICPMS) and Scanning X-Ray Fluorescence (XRF). A
major perturbation in dripwater geochemistry occurred in 1982, coincident with local ash fallout
from Mexico’s El Chichon volcano. Tephra deposition had a marked effect on the trace element
geochemistry of this stalagmite that was not otherwise evident in visible stratigraphy or stable
isotope ratios. A different LA-ICPMS trace element fingerprint in 1979 is associated with a
distinct color change and an extended period of low stable oxygen isotope values, coincident with
extremely high rainfall during that year. The XRF series included some species not included in
the LA-ICPMS analysis, such as S, K, and Cl; likewise, LA-ICPMS series include species not
analyzed by XRF, such as Na. As a result, XRF was more sensitive to major volcanic events, and
LA-ICPMS showed greater discriminating power between volcanic events and years of high
precipitation. Applying Empirical Orthogonal Function (EOF) analysis to a smaller, more recent
window of LA-ICPMS data without extreme events (1984 - 2001) revealed an additional mode of
Sr:Mg anticorrelation that may reflect seasonal to interannual changes in drip-rate. The dominant
EOF modes extracted depend on the environmental controls that dominated during different
38
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. intervals; major environmental events can dominate brief trace element records, overshadowing
the weaker seasonal to interannual signals. EOF is a promising data analysis technique for
distinguishing between different modes of geochemical variability in long multivariate
speleothem records, and also providing a means to distinguish between environmental triggers.
High-resolution, multivariate trace element analysis revealed stalagmite sensitivity to various
environmental events, including major regional volcanic eruptions and extensive precipitation
anomalies. These observations suggest new potential speleothem proxies for extreme events,
including low-latitude tephrochronology.
Introduction
It has been long established that major volcanic eruptions can introduce large quantities
of sulfate into the stratosphere, with widespread climatic effects (Briffa et al., 1998; Lamb, 1970;
Robock, 2000; Stott et al., 2000). Stratospheric sulfuric acid aerosols derived from volcanic S02
backscatter incoming solar radiation and trap longwave radiation, cooling the troposphere and
warming the stratosphere for years after an eruption (Toon and Pollack, 1980). Global coverage
of the direct effects of aerosol loading began with satellite observations in 1979 (Bluth et al.,
1993; McCormick et al., 1993). Since then, the global effects of two major eruptions, Pinatubo
(1991) and El Chichon (1982) have been studied extensively. However, understanding the
climate system response to past variations in volcanic forcing is restricted by incomplete
knowledge of the history of past eruptions, as well as the sulfate loading produced by individual
volcanic episodes (Bradley and Jones, 1992).
Historical records of volcanic eruptions and related climatic anomalies have been
extended by development of volcanic eruptive records based on geologic evidence (e.g.
(Sigurdsson et al., 1984) and/or ice core records (Clausen et al., 1997; Delmas et al., 1992; Lyons
39
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. et al., 1990; Palmer, 2001; Thompson et al., 1986; Yalcin et al., 2003; Zielinski et al., 1996;
Zielinski et al., 1994). The vagaries of atmospheric transport ensure that multiple ice cores from
different regions are needed to provide a more complete record of major volcanic eruptions in the
past (Clausen and Hammer, 1988; Delmas et al., 1985; Zielinski et al., 1997). The atmospheric
transport and climatic effects differ considerably between tropical and high-latitude volcanoes
(Stendel et al., 2006), and comparison between Greenland and Antarctic records has been used to
differentiate between N. Hemisphere, S. Hemisphere, and equatorial volcanic eruptions.
Relatively few annually-resolved tropical records of past volcanism have been produced, and
dating tropical eruptions using eruption products has comparatively, low precision (103 yrs) (Bluth
et al., 1997; Thompson et al., 1986), leaving the essential quantities of the timing and latitudinal
resolution of volcanic aerosol pulses poorly known (Stendel et al., 2006).
The first major volcanic plumes observed by satellites were the 1982 trachy-andesite
eruptions of El Chichon (17°21"N, 93°13' W, 1150 m elevation), classified as V.E.I. 5 on the
Volcano Explosivity Index (Newhall and Self, 1982). Tephra deposits were produced in three
phases: a major Plinian eruption on March 28th (V.E.I. 4+) was followed by two larger eruptions
on April 4th (V.E.I. 5) (Tilling et al., 1984; Varekamp et al., 1984). The eruption column reached
stratospheric altitudes of at least 26 km, and within weeks the aerosols were transported
completely around the equatorial belt (Robock and Matson, 1983). El Chichon (VEI 5) is about
400 km west of the cave site. AVHRR and TOMS satellite imagery of the tephra and S02 clouds
shows that ash was transported over central Belize on April 3rd-6th (Figure 3-1) (Bluth et al.,
1997). Rampino and Self (1984) showed that El Chichon is a source of unusually large quantities
of sulfate aerosol. El Chichon is the source of with one of the largest and most climatically active
volcanic eruptions in the entire Holocene (Palais et al., 1992). The volcano’s history is also of
interest to anthropologists because earlier eruptions are associated with the decline of Maya
civilization, including an eruption -1250 A.D. that has been linked to a major volcanic horizon in
Greenland ice cores in 1259 A.D. (Gill, 2000; Oppenheimer, 2003). Although the hypothesized
40
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. impact of volcanoes on the El Nino-Southern Oscillation System (ENSO) has long been a source
of controversy (Robock and Free, 1995; Self et al., 1997), Adams and co-workers (Adams et al.,
2003) showed that large ash plumes from equatorial volcanoes such as El Chichon can double the
chances of an El Nino event the following winter.
Recent developments in speleothem research have led to a number of precisely dated,
high-resolution, multi-proxy speleothem records of environmental change (Genty et al., 2003;
Wang et al., 2001). Although ice core and tree-ring records are the primary annually-resolved
volcanic proxy records, there is some suggestion that speleothems (cave mineral deposits such as
stalagmites) may record volcanic signals. The spatial distribution of speleothem records is
different from ice cores and centered in the tropics. Establishing reliable speleothem proxy
records of regional volcanic episodes could lead to more complete records of their timing and
location, particularly in the low-latitude belt and in regions with few glaciers. A Scottish
stalagmite recorded a dramatic four-year cooling episode that has been tentatively linked to the
Icelandic Hekkla eruption in 1104 (Baker, 1999; Baker et al., 1995). Atmospheric sulfate signals
related to volcanism have recently been identified in stalagmites from northern Italy, yielding
sulfate deposition records comparable to ice cores (Frisia et al., 2005). However, direct tephra
fallout has not yet been identified in speleothems, and ice core work has shown that multiple
chemical species in ash must be analyzed in order to fingerprint the source volcano. We had
available a well-dated modem stalagmite of opportunity, ATM7, (Frappier et al., 2002a) from a
region that experienced fallout from the 1982 El Chichon eruption in Chiapas, Mexico. During
the 1982 El Chichon eruption, Terry McCloskey (Louisiana State University) was farming in the
village of Santa Marta in eastern Belize, not far from the Actun Tunichil Muknal cave site. Dust
fallout on crops was noticeable for some time after the eruption (McCloskey, 2006, personal
communication). McCloskey’s local experience provides ground truth for the expectation of
tephra fallout derived from remote sensing of the tephra cloud was actually deposited in central
Belize (Schneider et al., 1999). The goal of our investigation was thus to test the potential to
41
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. measure stalagmite tephrochronology signals by analyzing the multivariate trace element
“fingerprint” of the 1982 El Chichon eruption.
Unlike glaciers, where post-depositional chemical transformations of aerosols are
relatively small, atmospheric deposition in karst regions is filtered by the overlying ecosystem
and soil biogeochemical systems before any signal is transmitted by infiltrating water to growing
speleothem surfaces (Fairchild et al., 2006). Speleothem geochemistry may be perturbed in a
number of ways by volcanic eruptions, primarily controlled by soil zone processes. Wet
deposition of volcanic sulfate may alter soil water geochemistry, leading to rapid leaching toward
the underlying cave (Frisia et al., 2005). However, strong wind shear during the 1982 El Chichon
eruptions separated the S02 gas cloud from the silicate tephra cloud (Schneider et al., 1995). The
tephra cloud was carried eastward over Central Belize (Fig. 3-1), but the stratospheric sulfate
precursors traveled further to the west. Thus, we expect most of the geochemical effects of this
volcanic event to be driven primarily by tephra; most sulfate delivered to Central Belize in the
initial plume phase was probably adsorbed by tephra particles and deposited in dry form (Rose,
1977). The effects of direct sulfate deposition were probably secondary and related to slow
fallout of stratospheric sulfate months to years after the eruption (Bluth et al., 1997). Fresh tephra
is rapidly weathered on contact with water (Varekamp et al., 1984), and can be expected to have
multiple effects on soil geochemistry, with complex interactions with organic matter and clay
mineral phases present. Leaching products of tephra include metallic ions common in silicates
(Varekamp et al., 1984).
In general, investigations of inter-annual to sub-annual relationships between trace
element proxies and their controlling variables have been reported by a number of groups,
(Fairchild et al., 2001; McMillan et al., 2005; Treble et al., 2003). It is frequently the case that
multiple factors affect individual measurable trace element species and modem calibration is
required to establish any causal relationships between a speleothem record and volcanic
eruptions. The soil and epikarst filter makes it unlikely that tephra particles will be found in
42
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. stalagmites. Instead, we explored the stalagmite for evidence of geochemical perturbations
triggered by chemical weathering of tephra and associated secondary biogeochemical effects. To
distinguish the signature of event types which have affected the area (e.g. hurricanes, heavy
rainfall, tephra fallout), we analyzed a large suite of trace elements and applied principle
component analysis to derive common modes of variability. Here we report the results of our
exploration of a very high-resolution modem trace element record measured in stalagmite ATM7.
Methods
Actively growing stalagmites were collected from Actun Tunichil Muknal in January,
2001 (Frappier et al., 2002a). After sectioning stalagmite ATM7 vertically, a thick section was
cut from the upper portion and polished. The stalagmite was dated using layer counting after
testing for annual growth patterns using the onset of 137Cs activity (Chapter 2).
A smaller piece was cut to approximately 25 mm x 20 mm in order to fit the LA-ICPMS
sample cell mount. Laser ablation was carried out at the LA-ICPMS facility at the Department of
Geological Sciences at Boston University. We followed the methodology of Treble et al. (2003),
analyzing multiple trace element concentrations collected in rapid sequence during slow scans;
however, we made a few modifications. The instrument consists of a VG Plasma ExCell
Quadrupole ICPMS fitted with a VG Merchantek LUV MP II frequency-quintupled 213 nm Nd-
YAG laser ablation microprobe. The laser energy profile was flat at 101 J/cm2. The laser was
operated with a spot size of 40 pm and 50% output at 10Hz. Analyses were conducted while
scanning the sample mounted on a motorized stage moving at 3 mm/min to create a continuous
transect. The surface was pre-cleaned by scanned ablation using a larger laser spot size (300 pm
spot size and 30% power at 10 Hz) prior to data collection on each transect section in order to
remove surface impurities. We analyzed 11 parallel transects at 50 micron spacing (Fig. 3-2).
Microscopic observation showed that the width of each resulting ablation line was approximately
43
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 40 pm. A stream of Argon gas carried the laser-ablated sample material into the plasma source
for ionization. Nineteen species were measured, each with an integration time of 0.05 s (nB,
23Na, 26Mg, 27A1, 29Si, 31P, 43Ca, 47Ti, 49Ti, 55Mn, 57Fe, 85Rb, 88Sr, 89Y, 90Zr, 138Ba, 140Ce, 232Th, and
238U). This method produces approximately 100 data points per mm of transect, although
smoothing of the series is produced by partial overlapping of successive measurements.
Each transect was collected in three overlapping sections, which were spliced to create
each continuous transect. Background counts (carrier gas background with the laser off) were
collected before and after each section. Before and after each transect, a NIST612 glass standard
was also analyzed (Pearce et al., 1996). Although this silicate glass standard is not ideal for this
work, a widely accepted calcite standard with the range of species found in NIST 612 is not
available as an alternative. The raw data was processed by subtracting background counts, taking
the ratio of each species relative to the Calcium signal, and standardizing with respect to the
NIST612 analyses. Following Treble et al. (2003), all 11 parallel transects were stacked prior to
data analysis in order to assimilate between-track variability. We correlated the LA-ICPMS
transect to the adjacent, well-dated stable isotope transect from ATM7 (Chapter 2) in order to
determine the vertical extension rate along the LA-ICPMS transect (Fig. 3-2). Age model error is
on the order of a few months.
Scanning XRF analysis was carried out at Woods Hole Oceanographic Institute.
Scanning XRF is a rapid, non-destructive analytical technique that captures a series of broadband
emissions spectra from a sample surface after excitation with a focused x-ray beam as the sample
is translated under the beam. Spectral peaks can be analyzed to quantify the chemical
composition of the sample surface at relatively high resolution. We scanned 23 elements (Al, Si,
P, S, Cl, Ar, Ca, K, Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, Br, Rb, Sr, Ba, W, Pb) along a single
sampling track (Fig. 3-2). The scanning XRF system is designed to analyze sediment cores, and
the x-ray beam is a slit with a fixed width of 5mm and adjustable length (20 microns to 500
microns). A polished slab of ATM7 was mounted on the scanning track, and translated through
44
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the scanner at 30 s timesteps with the beam length set at 200 pm. This yields an average of 6
sampling points per year.
Data Analysis: LA-ICPMS
In order to determine the principle modes of variability present in the entire 18-species
trace element series, we performed Empirical Orthogonal Function (EOF) analysis, a method of
principle component analysis (Meeker et al., 1995; Peixoto and Oort, 1992). The EOF technique
partitions the total variance present in the data set into uncorrelated, orthogonal patterns, termed
empirical orthogonal functions, eigenvectors, or modes. The resulting modes describe the
uncorrelated linear combinations of observations that are most efficient in explaining the variance
in the data. The first mode explains the largest percentage of variance in the dataset. Each
successive mode explains decreasing amounts of variance, and any two modes are independent
and uncorrelated. Each eigenvector is “loaded” with a combination of variables, where the factor
loading for each species is the correlation coefficient between the mode and that variable. The
different modes thus presumably represent different kinds of variability, each driven by a
different process. The ice core community has applied this technique widely in order to
differentiate between sources and transport mechanisms through interpreting the relationships
between individual species described by each mode (Mayewski et al., 1994). Each mode thus
may reveal information about how the biogeochemical system of a stalagmite is controlled by a
different environmental parameter. EOF analysis was performed using L. D. Meeker’s Matlab
script toolbox (Meeker et al., 1995). Eigenvector factor loading and variance were recorded, and
EOF modes were plotted over time for comparison to the local historical record.
We also re-analyzed the EOF modes from 1983 - 2001 to explore smaller-amplitude
events during a more stable period of deposition. The late-period EOF modes showed that trace
elements were fairly stable throughout this interval, except for a few events deposited in 1996-
45
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1997. The character of the 1996 and 1997 events were chemically similar to the 1982 and 1981
events, although considerably smaller in amplitude.
Data Analysis: XRF
23 species were analyzed in order to determine whether this method was also able to
capture volcanic signals. In particular, we were interested in a few species that could not be
analyzed using LA-ICPMS, including S, K, and Cl. A few species of interest were only rarely
detectable using this method (P, Cu, Cl). The data were normalized to Ca. Reported XRF
concentrations are relative, because no calcite standard materials were available for analysis. We
hope to obtain reference materials in the future to report more quantitative values using this new
method. Nevertheless, this rapid, non-destructive trace element analytical method produced well-
behaved data with considerable structure as described below. As before, we performed EOF
analysis on the XRF series. Eigenvector factor loadings and variance were recorded, and EOF
modes were plotted for comparison to the local historical record and LA-ICPMS data.
Results
LA-ICPMS
All measured species show significant high-frequency variability (Fig. A-4). A few
species show long-term trends and inter-annual variability (Mg, Sr). Although additional
information about biogeochemical variability can be revealed through time-series analysis of
individual EOF modes, this application is limited by the short length of the ATM7 dataset.
46
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Without measured chlorine concentrations we were unable to correct for variations in cyclic salt
concentrations.
The first three EOF modes accounted for nearly 60% of the total variance in the trace
element dataset (Fig. 3-3, Fig. A-4). The remaining modes had low explanatory power (<~5%
variance), and are not reported here. Timeseries plots (Fig. 3-4) show how each mode, or multi
variate fingerprint, varied over the last few decades. The factor loadings and temporal variability
provide the basis for interpreting the most likely environmental factors that underlie each mode.
Variance was partitioned evenly across the first two EOF modes for Sr and Si; thus EOF is not a
particularly diagnostic tool for investigating either Si or Sr in this dataset.
EOF1 explains 37.1% of the total variance in the dataset, and is positively loaded with all
measured species. Several species are loaded at greater than 50% variance (Ti, Mn, Rb, Y, Ba,
Ce, Th, and U), while Ba, Na, and Fe were loaded at less than 25%. EOF1 variability is
concentrated in a single major event positive around 1982, with two lesser events centered at
1979 and 1981. EOF2 contains 14.5% of total variance, with a more complex chemical pattern.
The factor loadings of anticorrelated species have opposite sign, such that the concentrations of
species are high when those species loadings are negative, as in EOF2. The species B, Mg, Sr,
and Al are anticorrelated with Fe. EOF 2 is not loaded on Na, P, Ti49, Mn, Rb, Y, Zr, Ba, Ce, Th,
U. Figure 3-4 shows that EOF2 has a slight downward trend throughout the series, driven
primarily by Mg (Fig. 3-5). Positive events in EOF2 occurred -1976 and in 1979. In 1982, a
major negative event in EOF2 is concurrent with a positive event in EOF1. In 1981, EOF2 shows
an increase in variability without a clear signal or trend. EOF3 incorporates 7.9% of total
variance, primarily related to high-frequency P and Fe variations. EOF3 shows a single positive
spike in 1981 (Fig. 3-4; See also Fig. A-5).
47
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Scanning XRF
All measured species show significant high-frequency variability. A few species are
detectable in a minority of samples (K, Cu, Ar). The first EOF pattern explains 28.7% of the total
variance in the dataset, and is positively loaded with most measured species (Fig. 3-5). As for the
LA-ICPMS results, EOF1 variability is concentrated in a single major event around 1982.
Another large event occurred prior to the mid-1970s. The XRF EOF2 is somewhat different from
the second mode identified in LA-ICPMS. Here, EOF2 contains 15.7% of total variance, where
the transition metals (K, Ti, Fe, Cu, Zn) vary opposite to Al. The XRF EOF2 also shows major
events around 1982 and the early 1970s. The first two XRF modes are both dominated by two
events: 1982 and the early 1970s; although the modes are positively correlated for the early event
and anticorrelated during the latter event. These differences in loadings between modes indicate
that the two events are chemically different.
Discussion
The chemical and temporal fingerprint of each EOF pattern can be interpreted with
respect to the most likely controlling processes that could generate the observed mode. Tropical
cyclones striking the vicinity in 1978, 1980, 1990, 1993, 1995 (twice), 1996 (twice), 1998, 1999,
and 2001resulted in brief 5I80 value excursions (Chapter 2). No major chemical perturbations
related to tropical cyclones have been observed in this record using either LA-ICPMS or
Scanning XRF analytical techniques. This suggests that the high-resolution stable isotope proxy
48
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. demonstrated by Frappier and colleagues is thus far the only demonstrated speleothem proxy that
can distinguish tropical cyclones from background environmental variability.
For both LA-ICPMS data and XRF data, EOF1 represents a general flux of trace
elements into the groundwater system, likely related to extended periods of extreme wetness,
external input of silicate material, and/or changes in soil pH/redox conditions. The brief dataset
and lack of sea-salt correction in the LA-ICPMS data prevents us from distinguishing entirely
between these potential drivers. High La-ICPMS EOF2 values represent input of silicate
aluminosilicate minerals (clays) varying opposite to Fe, pointing to a physical flushing control
washing clays into the system. P and Fe are primarily loaded on LA-ICPMS EOF3. For the XRF
data, the additional loading from S in the XRF EOF1 mode suggests a stronger volcanic
signature, particularly during the 1982 event. The loadings of the second XRF mode strongly
suggest that it is also related to silicate minerals varying opposite to transition metals. If both
XRF EOFs are related to volcanic tephra, the second mode may represent differences in
composition between events. We now can interpret each trace element event (early 1970s -
2001).
Around 1978-1979, the stalagmite shows a striking increase in LA-ICPMS EOF2 and
EOF 1 that is coincident with color changes in stalagmite stratigraphy, stalagmite stable isotope
perturbations, and major historical rainfall events (Fig. 3-4C). Local meteorological records for
1979 show that rainfall exceeded three standard deviations above the climatological mean (Figure
3-6). At this time, the stalagmite shows a distinct rusty-colored layer in stalagmite stratigraphy
and an extended period of very low stable oxygen isotope ratio (low 5lsO values) (Chapter 2).
The positive EOF2 anomaly in 1979 reflects increased flushing of clays from the soil and into the
cave conduit system; this is consistent with that year’s extreme rainfall season. However, this
event does not seem to have affected EOF 3 significantly, indicating that P and Fe leaching were
not affected especially by this period of wetness. Because P is generally insoluble, the lack of a
response in EOF3 in 1979 is also consistent with control by an extended period of extreme
49
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. wetness. The XRF data shows no major event at this time, suggesting that species analyzed only
by LA-ICPMS may be the primary driver behind this geochemical event (e.g. Ba, Mg).
An explanation for the 1981 excursion in all LA-ICPMS EOFs is not immediately
apparent; however, the unique spike in EOF3 at the onset of this event may hold the key to
understanding the driver of this event (Fig. 3-4B). No environmental extreme or event was
reported that year in the region, although local meteorological observations indicate that rainfall
was somewhat lower than normal (Fig. 3-6). A rather speculative explanation is that the 1981
event represents the influence of fire. Although this cave site is located in the Tapir Mountain
Nature Preserve, at that time the preserve had only recently been established. Local subsistence
farmers living in the area practiced slash-and-bum agriculture in the riparian floodplain nearby,
traditionally setting fires just prior to the onset of the summer monsoon. The presence of large
trees at the cave site shows that no major forest clearing or major canopy fires took place in
recent decades. Understory biomass burning is one plausible explanation for the sudden leaching
of all species in 1981, particularly the release of biologically important species such as P and Fe.
The mineral form of P is particularly insoluble in the soil, while soluble P is tightly cycled within
the ecosystem (???). Fires combust organic matter, leaving soluble P behind in the ash,
contributing to short-lived post-burn P leaching (Hernandez et al., 1997; McColl and Grigal,
1975). Available P is quickly absorbed by primary producers, leached by infiltrating
groundwater, and converted to insoluble carbonates and phosphates (Ewel et al., 1981). Fe is also
limiting in many forest environments, and available Fe is quickly absorbed by organisms; thus the
Fe response is similar to P. Thus, the brief duration of the 1981 EOF2 event is consistent with a
fire source. If the 1981 stalagmite signal was pyrogenic, the XRF analysis should reveal increased
K and reduced S deposition at this time, because K is highly soluble and fire volatilizes S (Ewel
et al., 1981). An alternative interpretation is that the 1981 event in EOF1 and EOF2 is a volcanic
tephra signal, but a smaller event from a different volcanic center with a different geochemical
fingerprint from El Chichon’s signal, described below. If this represents a volcanic signal, we
50
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. would expect XRF to reveal an increase in S from tephra deposition (Frisia et al., 2005).
However, the XRF data shows no 1981 event, and no conclusive interpretation can be made for
this event given the available data.
The largest element perturbation in both the LA-ICPMS and XRF data also occurred
around 1982 (Fig. 3-4A). Although the ash cloud from El Chichon covered the cave site in 1982,
no visible changes in the stalagmite stratigraphy were observed. On the other hand, stable isotope
perturbations were evident in 1982: both S180 and Si3C values dropped sharply by about 0.2 per
mil (Chapter 2). This discontinuity in an otherwise remarkably continuous dataset points to a
period of non-deposition. The small stable isotope record offset and lack of stratigraphic
manifestation indicates that the hiatus was brief (on the order of weeks to months). For LA-
ICPMS, EOF1 and EOF2 show an inverse relation during the 1982 event, indicating greatly
increased leaching of all species from the soil (EOF1), but with enhanced Fe and P and without
flushing of clays (EOF2). We interpret broad leaching spectrum as a tephra weathering product
signal from El Chichon (Varekamp et al., 1984). The first two XRF modes suggest tephra
deposition with proportionally reduced (but still very concentrations high) of K and Fe and
enhanced proportion of Ar and Cl. This event is the most striking geochemical signal in the
entire period of record. The EOF fingerprint of this event points to the influence of tephra from
El Chichon.
The second largest XRF event occurred in the early 1970s, prior to the start of the LA-
ICPMS laser track. EOF1 shows that the early 1970s event was generally similar to, but
considerably smaller than, the 1982 event. This similarity suggests a possible volcanic origin for
the early 1970s event. In addition, this event is more heavily loaded on EOF2, an indicator of
alumino-silicate minerals with more Fe and transition metals. Thus, the early 1970s event can be
interpreted as a volcanic event with higher levels of Fe and other transition metals compared to El
Chichon. This pattern is consistent with ash deposition from the largest volcanic eruption in the
Neotropics during the 1970s, the late 1974 VEI 4 eruption of Fuego near Antigua, Guatemala.
51
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Fuego is a large stratovolcano volcano about 350 km from the cave site. Calc-alkaline basaltic
ash from the 1974 eruption produced volcanic sunsets across the northern hemisphere (Rose et
al., 1978). Thus, the greater Fe and transition metal content in the 1974 event compared to 1982
is consistent with the more mafic eruptive content of Fuego compared to El Chichon. No
similarly large geochemical perturbations are visible in ATM7 either around 1992-93, related to
the Pinatubo eruption in the Philippines, or around 1996-97, related to the Caribbean eruption of
Montserrat. No other major tropical eruptions occurred in Central America during the period of
record. The XRF series appears to be dominated by major volcanic eruptions in Central America.
Thus, the ATM7 stalagmite is highly sensitive to recording regional tephra fallout from VEI 4 or
greater eruptions.
Since 1983, the ATM7 trace element record has been relatively quiet. Results of a second
LA-ICPMS EOF analysis on the more recent period (1983 - 2001) show smaller events in 1996-
1997 with some similarities to both El Chichon ash fall and to the high rainfall period of 1979
(Fig. 3-7, EOF1 and EOF2). Local rainfall was in the normal range in 1997 and although the
previous two years had somewhat strong dry seasons, this is not particularly unusual (Fig. 3-5).
We can thus have little confidence in attributing the 1997 trace element event to unusual
seasonality. Another explanation for the geochemical character of the late 1990s trace element
events, biomass burning immediately over the cave, can be ruled out by reports from locals and
anthropologists working at the site during this period. Regional drought and fires in 1998 related
to El Nino may be a source of aerosol particles, however, the EOF analysis does not show a clear
biogenic signal. Alternatively, moderate Central American volcanic eruptions of at least VEI 3 in
the late 1990s could be related to the observed signal. Regional eruptions during this period of at
least VEI 3 include Popocatepetl, Mexico in June 1997 (-1000 km west of the cave site), and
Papaya, Guatemala in May 1998 (-300 km south of the cave site). Tephra aerosol trajectories for
these eruptions could rule out this explanation, although these are not readily available for the
listed eruptions. A definitive interpretation for the late 1990s trace element events remains
52
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. elusive, although these events are orders of magnitude smaller than the 1982 event. In order to
distinguish biomass burning events from tephra deposition, it may be helpful to analyze more
distinguishing species, such as K and S, in addition to considering a much larger number of
events of both types. Analyzing speleothem material from a site with a known history of bums
could contribute to resolving this problem. EOF 3 represents changes in the ratio of Sr + Ba:Mg,
which vaiy on inter-annual and annual time-scales. Anticorrelation between Mg and Sr indicates
control by crystal growth on the stalagmite surface, related to dripwater chemistry and/or drip rate
(Fairchild et al., 2006; Huang et al., 2001). To assess intervals not dominated by volcanic
influence, it is evident that controlling for sea-salt input will be important for assessing non
marine influences.
This analysis illustrates one way in which stalagmite geochemical records of regional
volcanic eruptions may be distinguished from other sources of geochemical change. Speleothem-
based records of explosive volcanism in the surrounding region would open the door to improved
understanding of tropical volcanic forcing and related climatic impacts. Furthermore, TIMS
dating error bars are large for speleothems with low Uranium content, such that identifiable
tephra events could provide absolutely dated horizons for additional age control. Although this
stalagmite was selected for its tendency to record brief infiltration events such as tropical
cyclones, ATM7 also recorded the geochemical signature of the 1982 El Chichon tephra cloud;
however, the brief hiatus associated with the eruption suggests that a longer infdtration time
might be more appropriate for the purpose of detecting volcanic events.
Conclusions
A rapidly growing, modem tropical stalagmite recorded the multivariate influence of a
number of environmental factors from 1978-2001. Applying EOF analysis to an exploratory LA-
53
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ICPMS multivariate stalagmite trace element dataset effectively separated the geochemical
signatures of several different geochemical factors, including the 1982 El Chichon eruption,
extremely wet conditions during 1979, and some smaller events that will require additional
analyses to interpret with confidence. No geochemical signal was linked specifically to historical
tropical cyclone events, indicating that stable isotope analysis remains the sole speleothem
tropical cyclone proxy identified to date. However, this interpretation was hampered by lack of
Cl analysis, which would enable a sea-salt signature to be identified. The geochemical fingerprint
of the 1982 El Chichon eruption was positively identified by both LA-ICPMS and XRF data
analysis, and we further suggest that a similar signature in the early 1970s reflects the 1974 Fuego
eruption. Furthermore, during periods when extreme events do not dominate the geochemical
signature of the stalagmite, seasonal to inter-annual fluctuations of Sr+Ba:Mg ratios appear to be
driven by growth rate as well as other dripwater chemistry changes. Additional species, such as S
and K, may prove diagnostic for distinguishing between potential drivers such as local biomass
burning and tephra fallout.
Tephra deposition over a cave site can result in distinctive and overwhelming
geochemical perturbations that persist for months after the fallout event, even when S is not
included in the analysis. The rapid, non-destructive Scanning XRF technique readily identified
volcanic signatures in ATM7, verifying the LA-ICPMS analysis. We find that Scanning XRF
seems to be a rapid and straightforward technique for identifying volcanic events in stalagmite
stratigraphy; however, several analyses per year are required to resolve the -annual-scale
volcanic signals. Scanning XRF has a number of critical advantages over LA-ICPMS, including
the ability to place entire stalagmites into the instrument and rapid, high-resolution multiple
species analytical capability. However, for speleothems with slow extension rates or growth
laminae that are asymmetrical, Scanning XRF is limited by the x-ray beam size and single-axis
scanning. Longer records with a larger number of volcanic signals will be required to further test
the sensitivity of a potential stalagmite proxy for tephra deposition, and to determine what species
54
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. are most useful for discerning volcanic signals from other geochemical event types. A larger
number of spatially distributed speleothem records around a volcanic region such as Central
America will enable the spatial sensitivity of speleothem tephrochronology records. Pursuing
potential speleothem records of volcanism could ultimately enhance the understanding of
volcanic forcing in the climate system, and contribute to global climate model validation studies.
In regions where speleothem uranium concentrations are low, tephra layer recognition may
provide additional marker horizons useful for dating. Elsewhere, it may be possible to use more
precisely-dated speleothem records of ashfall events to investigate the history of low-latitude
explosive volcanism.
EOF analysis of multiparameter stalagmite records through decomposition of dataset
variability is a powerful method for investigating the underlying mechanisms of biogeochemical
change in these sensitive geologic recorders. The EOF modes reveal the common responses of
the cave-epikarst biogeochemical system to various forcing factors, geochemical fingerprints
whose dynamics can be traced through time. Although annual-scale trace element variations have
been identified in several stalagmite studies (Fairchild et al., 2001; Fairchild et al., 2000; Huang
et al., 2001; Treble et al., 2003; Treble et al., 2002), we find that it may be inappropriate to use
the EOF technique to identify annual variations in very short records containing large volcanic
events that overwhelm background variability, such as the ATM7 record presented here. Longer
records may enable better separation of small annual geochemical variations from intermittent
extreme events. The EOF analysis approach to multivariate speleothem proxy data may prove
fruitful in the ongoing effort to understand interactions between the atmosphere, hydrosphere,
biosphere, and rock record preserved in caves.
55
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 3-1. AVHRR satellite imagery showing atmospheric transport of El Chichon tephra imaged on April 5, 1982. Colors of ash plume denote ash concentrations. Note eastward transport of large quantities of tephra from the volcano (©) toward Belize and the cave site (£2). The location of Fuego is also shown (®). Bar graph shows dates of ash and sulfate satellite observations (gray and yellow triangles) after the two eruptions on 4/4. Modified from Schneider et al., 1999.
Volcanic Ash Mass (kg/km2) 400 1,000 3,000 8,000 i i i i i 11______I___i__i i i i i i
56
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 3-2. Photographs of stalagmite ATM7. Polished upper cross-section shows the location of the laser ablation tracks (green) and XRF track (black) relative to the stable isotope micro-milling track (blue).
57
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 3-3. Chart of LA-ICPMS EOF factor loadings for the first three modes. The total variance explained by each mode of variance (EOF) is indicated. Positive loadings indicate positive correlation of a chemical species with mode; negative loadings indicate an inverse correlation of species with mode. Species with loadings of opposite sign on the same mode are anti-correlated. Species with small loadings, explaining less than 10% of variance, are not particularly important or well described by the mode in question.
EOF 1: 37.9 % variance 70 60 50 40 30 20 10
EOF 2: 11.1 % variance
B Mg Ti 00Mn 7 Na 27Al 31P Ti Fe ^ 88Sr 90Zr ^°Ce 238U
58
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 3-4. Timeseries plot of LA-ICPMS variance in the first three EOFs. Y-axes are standardized variance values for each mode. Events in 1979 (A), 1981 (B), and 1982 (C) are highlighted. Event A (EOFs 1 and 2) is coincident with the 1982 El Chichon eruption. Event B (EOF 1, 2, 3) marks a suspected local fire or smaller volcanic event in the region. Event C (EOF 1 and 2) is coincident with a rusty-colored layer, low 5lsO values, and extreme precipitation throughout most of 1979 when local rainfall was greater than three standard deviations above the norm. The first three EOF modes show relatively little variability from 1983 - 2001.
EOF 1 Time Series 0 .3
0.2
0.1
1 1 Ay ' l-.i-.L-.iiJ., . i ...
- 0.1
EOF 2 Time Series 0.1
0 .0 5
- 0 .0 5
- 0.1
EOF 3 Time Series
1 9 8 0 1985 1990 1995 2000
s_ v 01 u IBR
1—F f a .1—m < —1 H
59
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. without prohibited reproduction Further owner. copyright the of permission with Reproduced Figure 3-5. ATM7 Scanning XRF EOF results. Note major events A (1982), B (1979), and (1979), B (1982), A events major Note results. EOF XRF Scanning ATM7 3-5. Figure C (1974?). C Factor Loadinas -0.5 - 0.5 0 0.6 0.4 0.2 0.2
8 r r - 2000 2000 EOF1: 28.7% variance 28.7% EOF1: EOF2: 15.7 %variance 15.7 EOF2: 901980 1990 901980 1990 1970 1970 60
Temperature (C) Precipitation (mm) and pan evaporation) in gray, and mean maximum temperature in black. Two hydrological years are highlighted: a very wet Figure 3-6. Monthly weather observations at Central Farm, Belize, showing water balance (difference between precipitation year 1979 (A), and a slightly dry year 1981 (B).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 3-7. LA-ICPMS EOF results for quiet latter period (1984 - 2001). Factor Loadings and timeseries plots of the first three EOF modes. Note that in the absence of volcanic dominance, lower-amplitude trends are apparent (see EOF3). Here, EOF 3 represents changes in the ratio of Sr + Ba:Mg, which vary on inter-annual and annual time-scales. The event in 1996-1997 (EOF1 and 2) is similar in character to the volcanic signals in the overall record, but much smaller in amplitude.
EOF 1 : 24.2015% Total Var.
EOF 2 : 13.4247% Tota Var. (/} o> c T 3 CO O
O o 03 EOF 3: 10.1681% Total Var.
Mn Fe Rb Sr Y Zr Ba Ce Th U
EOF 1 Time Series
- 0.5 EOF 2 Time Series 0.2
0.1
- 0.1
- 0.2 EOF 3 Time Series 0.2
0.1
- 0.1
- 0.2 '1984 1 9 8 6 1 98 8 1990 1 9 9 2 1 9 9 4 1 9 9 6 1 9 9 8 2000
62
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER IV
ROOTING OUT THE AMPLIFICATION MECHANISM
LINKING A WEAK ENSO TELECONNECTION SIGNAL
AND A STRONG STALAGMITE CARBON ISOTOPE RECORD
Introduction
The El Nino-Southern Oscillation (ENSO) system is a primary driver of variations in
global weather conditions on interannual time scales (Philander, 1990). The ENSO system
oscillates chaotically between warm phases (El Nino events) and cold phases (La Nina events).
Regions with strong ENSO teleconnections typically experience major changes in average
rainfall and/or temperature regimes during strong El Nino events (Diaz and Kiladis, 1992). Some
regions, such as central Belize, show weak or instrumentally un-detectable ENSO teleconnection
patterns (Belize National Meteorological Service). We were thus surprised to find a strong record
of El Nino events in the carbon isotopic record of a Belize stalagmite (Frappier et al., 2002).
Stalagmites (mineral deposits typically composed of calcium carbonate) are formed in
caves by slowly dripping groundwater. The ultimate source of this water is precipitation;
however, the carbon isotopic ratios and dissolved ionic composition of rainwater is altered by
nteractions with the overlying biogeochemical system before reaching the cave. Stalagmite
carbon (C) comes from three primary sources: local bedrock, atmospheric carbon dioxide (C02),
and C 02 produced by soil respiration. Carbon dissolved in percolating drip water is the
iproximate carbon source to the growing stalagmite. Thus, if the local bedrock carbon source is
63
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. invariant, variations in stalagmite carbon isotopic ratios (513C values) reflect changes in the
overlying ecosystem’s carbon usage. However, large variations in groundwater carbon isotopic
ratios on inter-annual and shorter time scales had not been measured previously (Figure 4-1).
The ATM7 stalagmite 513C record shows a marked rise in 513C values corresponding to
recent El Nino events which occur with negative values on the Southern Oscillation Index (SOI),
a key indicator of ENSO phase (Troup, 1965). The amplitude of the recorded 513C excursions
varies between individual El Nino events, from about 5 to 10 per mil. Stalagmite stable carbon
isotopic ratio variations have been used as a proxy for ancient biome variations over caves (e.g.
C-3 forests vs. C-4 prairies) at century to millennial time scales (e.g. Denniston et al., 1999). In
this record, the variation occurs across a 6-month period, yet the amplitude range is similar to that
reported between forest and prairie biomes (Tieszen et al., 1989). Previous biome-based
interpretations of speleothem records are not called into question by the work of Frappier et al.,
(2002), because the isotopic changes occur over sufficiently long periods of time to allow soil
carbon turnover, and because they are consistent with lower-resolution regional pollen records.
However, on short, interannual timescales, there is no theoretical prediction of a relationship
between ENSO and carbon isotopes in terrestrial ecosystems, groundwater, or stalagmites - even
though biospheric carbon fluxes are closely tied to ENSO (e.g. Jones et al., 2001). The ATM7
record revealed the operation of some poorly constrained carbon cycle process(es) (Frappier et
al., 2002).
The discovery of a relationship between ENSO and carbon isotopic fractionation within
terrestrial ecosystems is particularly puzzling in light of the lack of significant local temperature
or precipitation anomalies during El Nino and La Nina intervals (Frappier et al., 2002). During
the past few decades, local meteorological variables had no significant variation in the interannual
(ENSO) band. A multivariate analysis showed that weather anomalies in Belize are also not
significantly related to ENSO. The ATM7 record indicates that the magnitude of El Nino
64
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. teleconnections at this site has apparently been amplified by the ecosystem carbon cycle with
respect to local weather variations. This result raises a series of important questions:
• How does ENSO affect ecosystem carbon fluxes in regions without
significant weather teleconnections?
• What is the relative importance of physical processes (e.g. barometric
pressure) compared to biological processes (e.g. soil respiration rates)?
• How does the carbon biogeochemical system amplify the apparent local
magnitude of ENSO forcing?
• What are the most appropriate variables to capture ENSO’s impact on
terrestrial ecosystems?
• What are the implications of these processes for our current understanding
of terrestrial carbon cycle dynamics?
In the absence of significant ENSO-related variations in daily meteorological
observations, the teleconnection between ENSO and Belize must be controlled by variables that
are not reported by the Belize Meteorological Service and/or distributed among multiple
environmental factors (Taylor et al., 2002). One way to explore the nature of the ENSO-Belize
teleconnection is through modeling. Another approach that is the subject of this paper is remote
sensing. NASA’s archive of remote sensing products enabled us to explore various
environmental factors for significant relationships to ENSO and Belize during recent decades.
Interannual variability in carbon fluxes from terrestrial ecosystems is more than twice
that of marine sources (Bousquet et al., 2000). ENSO-driven variability in terrestrial ecosystem
carbon cycle dynamics from regions without obvious teleconnections may constitute an important
contribution to global C 02 fluxes (Frappier et al., 2002). Identifying the specific carbon cycle
65
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. processes and their climatic controls is important for projecting future atmospheric greenhouse
gas concentrations. It is thus essential to determine how this stable carbon isotope record was
produced to understand the implications (if any) of this discovery for climate-carbon cycle
interactions.
Given the problem of identifying the local ENSO teleconnection driver(s), a key
observation to consider is the relative amplitude of the annual cycle compared to the inter-annual
signal. The annual cycle dominates local weather records; yet although the stalagmite contains
approximately annual layers, the ENSO signal clearly dominates the stable isotope record. We
expected to identify strong interannual variability in some local factors. Any proposed
mechanisms must be capable of producing the rapid and continuous response of groundwater
carbon isotopes to ENSO that are implied by the dynamical details of the SOI captured by the
ATM7 record. We expected the local teleconnection factor or factors involved to show a
relatively strong correlation or lag correlation with the SOI. Furthermore, the unknown
teleconnection factor or factors that drive this system may be dominated by physical (climatic)
processes or could be dominated by non-linear vegetation responses to physical forcing. For
example, forest canopy openness and soil respiration rates can affect the carbon isotopic
composition of soil gas (Amundson et al., 1998; Cerling et al., 1991; Sternberg et al., 1997).
Candidate drivers for the ENSO teleconnection must be characterized by temporal variations in
Belize with maximum power in the interannual band, correlate with ENSO phase, and be
consistent with local meteorological data. The behavior of these variables during ENSO phase
transitions may also be particularly critical (Behrenfeld et al., 2001).
In order to understand the links connecting El Nino events and ATM7 813C values, two
major issues must be addressed: 1) the teleconnection factor(s) by which ENSO affects this
ecosystem, and 2) the biogeochemical dynamics that transfer the ENSO signal to groundwater
8I3C values. This study focuses on former, which can be addressed by a analyzing a variety of
66
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. remotely sensed environmental parameters for Belize during past ENSO cycles. The objective of
this study was thus to explore the nature of the ENSO teleconnection in central Belize using a
wide array of available observations, including local ground-based and remotely sensed
measurements of physical and biological factors. A set of hypotheses have been generated to
address the relative importance of the most likely teleconnection factors linking ENSO and the
local carbon cycle that can be explored by analyzing patterns in local weather and remote sensing
datasets:
H I: ENSO affects weather variables not measured by local meteorological stations.
H2: ENSO has a coherent effect that is distributed across a set of atmospheric parameters.
H3: ENSO has a weak effect on local weather, amplified by ecosystem changes in the forest
canopy.
H4: ENSO has a weak effect on local weather, amplified by ecosystem changes below the forest
canopy.
This paper explores variations in atmospheric and forest canopy parameters in relation to
the SOI in order to examine what support there is for any of the forementioned hypotheses. We
explore variables readily available in the scientific literature that might provide indications
regarding the links between ENSO and the carbon cycle at the cave site in central Belize.
Parameters of interest include barometric pressure, surface temperature, atmospheric C 02
concentration, dust, cloud amount and optical thickness, potential evaporation, relative humidity,
precipitation, soil moisture, canopy openness, and leaf color indices (Graham et al., 2003;
Schaefer et al., 2002). The first and second hypotheses can be tested by incorporating a larger
number of remotely sensed parameters into a time-series lag correlation matrix using the SOI as
the predictor variable. The third hypothesis can be tested by comparing the SOI and local
weather teleconnections to remotely sensed forest canopy measures. The fourth hypothesis falls
67
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. outside the scope of this paper, but could be addressed by monitoring the micrometeorology and
carbon cycle dynamics at the forest-cave site across an ENSO cycle.
Methods
To investigate the relationships between the Southern Oscillation and the local weather
and forest ecosystem response, we collected several long-term remote sensing and meteorological
datasets describing weather, atmosphere, and vegetation near the cave site (Table 4-1). The
Trenberth Southern Oscillation Index (SOI) served as our primary ENSO descriptor (Trenberth,
1984). As SOI is a monthly index, we also sought descriptors of Belizean weather, atmosphere,
and vegetation at monthly temporal resolution. The Belize Meteorological Service supplied daily
weather observations of temperature, precipitation, and evaporation from the nearby Central Farm
station (17°11’ N lat., 89° W long., elevation 64 m). We compiled these daily observations into
monthly indices describing the maximum, minimum, mean, and in some cases totals within a
month, for each daily temperature, precipitation, sunshine, and pan evaporation (Table 4-1).
In all cases, the datasets containing the satellite-derived descriptors covered relatively
large areas measured in fractions of degrees. Thus, for all remote-sensing datasets, we used the
highest resolution grid datasets available (see Table 4-1) and queried the parameter value of the
pixel best centered on the cave site at 17.1167 N lat., 88.8833 W long., 100 m elevation. We
queried all MODIS, TOMS, and TRMM datasets using the GES-DISC Interactive Online
Visualization ANd aNalysis Infrastructure (Giovanni) web-based data archive query tool at
NASA's Goddard Earth Sciences (GES) Data and Information Services Center (DISC); these
datasets can be found at NASA’s permanent data archive center Distributed Active Archive
Center (DAAC, 2006). Our measure of vegetation amount over the cave was the Fourier-
Adjusted, Solar-zenith-angle corrected, Interpolated, Reconstructed (FASIR) Normalized
68
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Difference Vegetation Index (NDVI) available as monthly observations from 1983-1998 (Hall et
al., 2005). Again, we selected for each month the quarter degree pixel best centered over the cave
site.
Most of the variables showed strong seasonal trends that were much stronger than any
interannual variations. We removed seasonal signals from each parameter by calculating monthly
means for each parameter and subtracting the monthly means from each observation in their
respective months. We then calculated time-series cross-correlations between the SOI and the
monthly variables describing local surface weather, atmospheric conditions, and vegetation
(Table 4-1). Many satellite-derived parameters were only available for short time periods,
covering only one or two ENSO events (Fig. 4-1). We calculated correlation lags of between 0
and 24 months, with the SOI as the leading indicator, for the weather variables from the Belize
Meteorological service and the FASIR-NDVI. For all other remote-sensing derived variables
(primarily the satellite-derived atmospheric descriptors), we calculated only 12 month lags as the
data were available for only 5-6 years (Fig. 4-2); longer lags would have resulted in too small a
sample size to adequately calculate correlation coefficients. Confidence intervals were calculated
for all of the cross-correlations to determine the likelihood the correlation coefficients were noise.
Correlations beyond the 95% confidence intervals we considered likely to be real. All statistical
analyses were performed using R 2.0.1 (R et al., 2004).
Results
Most weather variables showed cross-correlations with the SOI beyond the 95%
confidence intervals at some lag between 0 and 24 months (Table 4-2). However, the maximum
correlation coefficient (R value) was 0.368 (Fatent heat at 4-5 km, Table 4-2) indicating that the
univariate cross-correlations between SOI and weather and atmospheric properties in Belize were
69
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. weak. This finding is consistent with results from a more simple lag correlation analysis using
only local weather data (Frappier et al., 2002a). In addition, there were no clear patterns in the
lag periods; variables seem to respond independently to ENSO forcing (Table 4-2).
Examining the variables separately, the strongest links to SOI were related to
atmospheric moisture. SOI seemed to be positively correlated (correlation coefficients between
0.270 and 0.368) with latent heat at between the surface and 6 km above the Earth’s surface 5
months later (Table 4-2). The SOI was negatively related to atmospheric column water vapor
content between lags of 0 and 6 months (correlation coefficients between 0.261 and 0.292; Table
4-2) . SOI was positively related to the maximum monthly evaporation with lags of 0 to 7 months
(R between 0.191 and 0.289; Table 4-2). However, we cannot infer that significant, longer lags
do not exist in the short-term variables. Because of the brief length of these series, lack of
evidence for long lags is not evidence of their absence.
A few variables showed weaker correlations that appeared much later; not surprisingly,
these longer lags were apparent in the longer observational series. El Nino did not affect average
temperature or precipitation amount, but increased the monthly extreme maximum daily
temperature by about +1.6 °C, about 13-14 months after SOI changes in the Pacific. The ENSO
effect is weak but significant in a multivariate 13-month regression, including maximum daily
temperature (both extreme daily max and min), maximum daily precipitation, hours of sunshine
(reflective of cloudiness), and evaporation.
To investigate the potential response of the forest canopy to ENSO, we also conducted a
lag correlation analysis on NDVI data. Like most other variables included in this analysis, the
seasonal NDVI variations are large compared to the interannual mode. Similar to the weather
observations and remotely sensed data described above, ENSO correlations were not significant
at the 95% confidence level unless seasonal variations were removed. NDVI analysis of a 0.25
degree grid cell containing the field site in Belize showed that while the ENSO effect is
significant, it is rather weak compared to seasonal effects. This NDVI effect size comparison is
70
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. consistent with the difference in the amplitude of climate forcing in Belize at the seasonal vs.
interannual scales. The NDVI lag was significant at 14-15 months. Interestingly, the longest
NDVI lag follows changes in local temperature extremes by only about 1 -2 months, but follows
the most widespread weather changes by about 6 months.
Discussion
We cast a wide net with the goal of capturing the teleconnection mechanism linking
ENSO variability in the Pacific with central Belize environmental conditions. Toward that end,
we combined data series of different length. This approach had the advantages of including more
variables and El Nino events, but also made it inherently more difficult to assess longer-period
lags. Thus, finding short-period ENSO lags in the brief satellite data series does not rule out the
possibility that atmospheric variables may also play a role over a longer term. This result is
consistent with previous work showing that the lag response of terrestrial ecosystem productivity
to relatively small ENSO-driven temperature change is biome-specific, and varies in strength and
even sign at different lag periods from zero to three years (Braswell et al., 1997). As the satellite
series grow over time, the links may become more clear between regional atmospheric conditions
over Central America and the Caribbean and local weather extremes in Belize. Some weather
variables not measured by local meteorological stations show an ENSO response (e.g. latent heat,
aerosols, cloudiness, and vegetation NDVI), lending some support to HI.
Each potential local forcing factor seems to respond independently. While no ENSO lag
correlations were found when seasonal variations were not removed (Frappier et al., 2002), the
present analysis of seasonally de-trended weather data found weak lags at 0-8 months and 13-14
months, distributed across several variables with respect to the SOI. The lack of a coherent lag
pattern distributed across several weather patterns refutes H2.
71
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Both Belize surface weather observations and remotely sensed parameters show the same
overall pattern: weak interannual variations superimposed upon very strong annual cycles. This
contrasts with the stalagmite stable isotope record which is dominated by ENSO, with a weaker
seasonal cycle. The present analysis confirms the presence of some integration or amplification
of the interannual signal and/or damping of the seasonal signal, operating in the gap between the
local teleconnection and the stalagmite surface. Some non-linear process(es) of carbon isotope
dynamics involved in amplifying the interannual effect of ENSO remain unexplained.
Nevertheless, the present analysis does point toward some interesting aspects of the local
manifestation of ENSO. Stronger lag correlations are associated with weak anomalies in daily
weather extremes (e.g. precipitation intensity, extreme daily temperatures), suggesting that daily
weather anomalies are involved in driving the stalagmite ENSO signal. Atmospheric moisture
also seems to be involved, though only short lags have been identified (5 months for latent heat in
the lower troposphere; 0-7 months for ground-based maximum daily evaporation). The
mechanism(s) through which these anomalies in moisture and daily extremes are translated into
carbon isotope dynamics is a critical part of this puzzle that requires further investigation using
different methods.
However, the lower-amplitude ENSO signal in the ATM7 stable oxygen isotope ratio
(5180 value) record provides some direction (Chapter 2, Fig. 3). This related proxy indicates that
local weather is affected by ENSO through one or more of the following mechanisms:
precipitation or evaporation amount, storm tracks, or the isotopic composition of the water vapor
source (organization of the tropical atmosphere) (Dansgaard, 1964b; Lawrence et al., 2004). In
particular, the extended period of increased evaporative extremes at lag periods of 0-7 months
could have important impacts on soil microbial activity. Simple changes in storm tracks or the
isotopic composition in local precipitation are unlikely to perturb 813C values. There is no
evidence for substantial changes in local precipitation amount related to ENSO, although El Nino
increases precipitation variability by contributing a slight increase to the intensity of daily rainfall
72
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. at 0-3 months. Thus, we infer that the local effect of ENSO is probably a temperature and
moisture signal delivered primarily through changes in evaporation and precipitation intensity.
Similar to the weather and remote sensing analysis, ENSO signals are not apparent in
NDVI data when seasonal patterns are not removed. ENSO has a small effect on the forest
canopy superimposed on a stronger seasonal cycle. The canopy responds weakly to a similarly
weak ENSO teleconnection distributed across an array of lags and a number of weather and
atmospheric variables. The seasonally de-trended NDVI lag analysis shows a long lag of 13-15
months, following the 11-13 month SOI lag in temperature extremes by 1-2 months. Research
supports variable canopy response timescales (Braswell et al., 1997). This suggests that the local
vegetation may be responding to either the ~l-year local temperature lag, or the shorter
atmospheric moisture/precipitation lag of 0-6 months found in the remote sensing series.
It is important to note that NDVI is a fairly coarse measure of vegetation cover, and may
not be the most appropriate measure for exploring ENSO-ecosystem links. Vegetation may
participate in the ENSO signal transmission through three primary factors: canopy openness (C 02
recycling), 513C values of respired C 0 2j and 8i3C values of photosynthetic products (Farquhar et
al., 1989). We suggest that a more conclusive study of canopy involvement could be conducted
using LANDSAT data. LANDSAT coverages are more spatially-resolved and contain a range of
spectral bands that can be queried in many combinations in order to focus on particular signals,
such as water stress, leaf area, and nutrient limitation (ERDAS, 1999; Lauer et al., 1997). A more
detailed analysis of LANDSAT data could provide a more quantitative assessment of the role
played by the forest canopy.
Ecological systems are notoriously non-linear, and can be extraordinarily sensitive to
small environmental changes at critical control points (Burkett et al., 2005). Belowground
communities in particular can respond very rapidly to changes in environmental extremes as well
as mean conditions (Wardle, 2002). Furthermore, related studies have not shown ENSO
variability in speleothem growth rate or trace element concentrations (Chapters 2 and 3). We
73
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. have not ruled out a physical controlling process. If the latent heat changes (5 month lag) and
barometric pressure variability (no lag) are related to changes in atmospheric turbulence, any soil
C 02 efflux response would change the 513Cvalue of soil C 02 (Malhi et al., 1998). However, two
lines of evidence in particular make it difficult to envision a purely physical driver for the carbon
isotopic manifestation of ENSO and point toward some biological amplification mechanism: 1)
the apparent weakness of the interannual local manifestation of ENSO compared to the seasonal
pattern, and 2) the confinement of the ENSO signal to the stable isotope ratios.
Conclusions
The ENSO teleconnection to Belize is only apparent after correcting for seasonal
variations. The effect of ENSO on Belize weather is weak, distributed amongst several weather
variables. The most important teleconnection factors linking ENSO to central Belize’s ENSO
appear to be anomalous moisture and temperature/precipitation extremes. While some remotely
sensed variables respond with a lag timescale of zero to six months, a more coherent set of
weather variables respond with a lag of about 13-14 months. The ENSO teleconnection in Belize
lags behind the equatorial Pacific by several months to a year, consistent with recent studies
showing a lag for the Caribbean region (Chiang and Sobel, 2002; Giannini et al., 2001; Giannini
et al., 2000).
A weak El Nino signal in the NDVI data was found to most closely follow the local
temperature lag. This pattern points toward a sub-canopy amplification mechanism; however,
canopy involvement cannot be ruled out on the basis of relatively coarse NDVI data alone. We
have also argued that the biogeochemical amplifying factors of the local ENSO teleconnection
are most likely non-linear biological processes sensitive to small changes in moisture and
temperature extremes.
74
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Future analyses could narrow the list of potential teleconnection drivers by exploring
forest canopy involvement using more sensitive LANDSAT coverages, and by assessing possible
longer ENSO lag signals distributed among remotely sensed atmospheric parameters. Remote
sensing techniques cannot detect sub-canopy or belowground linkages between ENSO forcing
and local carbon cycle. Rather, the transformation of a weak ENSO teleconnection signal into a
strong groundwater signal can be investigated most effectively through real-time monitoring of
local micrometeorology, ecosystem carbon processing, cave dripwater chemistry, and direct
measurements of local carbon reservoirs, carbon fluxes, and microbial activity. Monitoring
across an ENSO cycle will be vital to identifying the mysterious processes involved in the ATM7
record. The data obtained in such an effort may also contribute to the thorny disciplinary
problem of 513C value interpretation in speleothem records.
While this analysis has advanced our understanding of the nature of the ENSO
teleconnection in central Belize, many outstanding questions remain and the local manifestation
of ENSO remains incompletely understood. Any assessment of the potential implications of the
813C signal for the current understanding of the interactions between climate variability and
terrestrial carbon cycling, must be postponed until the proxy systematics that underpin the ATM7
record can be unraveled through higher resolution analyses and sub-canopy investigations. From
the perspective of paleoenvironmental research, it remains troubling that a sedimentary proxy
record should reflect local environmental factors in such a biased manner. The apparent Gordian
knot posed by the ATM7 record remains to be solved by future research.
75
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4-1. Data Sources. Original Spatial Resolution Data Parameter Temporal (Lat x Long) Sou ret Resolution Trenberth Southern Oscillation Index (SOI) Monthly Southern Pacific Basin 1 Daily MaximumTemperature, Monthly Maximum (deg) Daily Site 2 Daily Maximum Temperature, Monthly Minimum (deg) Daily Site 2 Maximum Temperature, Monthly Mean (deg) Daily Site 2 Daily Minimum Temperature, Monthly Maximum (deg) Daily Site 2 Daily Minimum Temperature, Monthly Minimum (deg) Daily Site 2 Minimum Temperature, Monthly Mean (deg) Daily Site 2 Daily Precipitation, Monthly Maximum (mm) Daily Site 2 Daily Precipitation, Monthly Minimum (mm) Daily Site 2 Total Precipitation, Monthly (mm) Daily Site 2 Mean Precipitation, Monthly (mm) Daily Site 2 Rain Days per Month (days) Daily Site 2 Mean Monthly Precipitation on Rain Days (mm) Daily Site 2 Daily Sunshine, Monthly Maximum (hrs) Daily Site 2 Daily Sunshine, Monthly Minimum (hrs) Daily Site 2 Total Sunshine, Monthly (hrs) Daily Site 2 Mean Sunshine, Monthly (hrs/day) Daily Site 2 Daily Evaporation, Maximum Monthly (mm) Daily Site 2 Daily Evaporation, Minimum Monthly (mm) Daily Site 2 Total Evaporation, Monthly (mm) Daily Site 2 Mean Evaporation, Monthly (mm/day) Daily Site 2 Aerosol Optical Depth at 0.55 Micron Monthly l°x 1° 3 Aerosol Fine Mode Fraction ' Monthly l°x 1° 3 Cirrus Fraction NIR Method Monthly l°x 1° 3 Cirrus Reflectance Monthly l°x 1° 3 Cloud Fraction Daytime IR Method Monthly l°x 1° 3 Cloud Effective Radius Combined Phase (microns) Monthly l°x 1° 3 Cloud Effective Radius Ice Phase (microns) Monthly l°x 1° 3 Cloud Effective Radius Water Phase (microns) Monthly 10 x 1° 3 Cloud Optical Thickness Combined Phase Monthly 1° x 1° 3 Cloud Optical Thickness Ice Phase Monthly l°x 1° 3 Cloud Optical Thickness Water Phase Monthly l°x 1° 3 Cloud Top Pressure (hPa) Monthly l°x 1° 3 Cloud Top Temperature (K) Monthly l°x 1° 3 Water Vapor Clear Sky NIR Method (cm) Monthly l°x 1° 3 Water Vapor Above Cloud NIR Method (cm) Monthly l°x 1° 3 Water Vapor Column IR Method (cm) Monthly l°x 1° 3 Accumulated Rain (mm) Monthly 0.5° x 0.5° 4 Convective Rainrate (mm/hr) Monthly 0.5° x 0.5° 4 Stratiform Rainrate (mm/hr) Monthly 0.5° x 0.5° 4 Surface RainRate (mm/hr) Monthly 0.5° x 0.5° 4 Latent heat surface-1 km (deg/hr) Monthly 0.5° x 0.5° 4 Latent heat surface-1-2 km (deg/hr) Monthly 0.5° x 0.5° 4 Latent heat surface-2-3 km (deg/hr) Monthly 0.5° x 0.5° 4 Latent heat surface-3-4 km (deg/hr) Monthly 0.5° x 0.5° 4 Latent heat surface-4-5 km (deg/hr) Monthly 0.5° x 0.5° 4 Latent heat surface-5-6 km (deg/hr) Monthly 0.5° x 0.5° 4 Latent heat surface-6-7 km (deg/hr) Monthly 0.5° x 0.5° 4 Latent heat surface-7-8 km (deg/hr) Monthly 0.5° x 0.5° 4 Latent heat surface-8-9 km (deg/hr) Monthly 0.5° x 0.5° 4 Latent heat surface-9-10 km (deg/hr) Monthly 0.5° x 0.5° 4 Latent heat surface 11-12 km (deg/hr) Monthly 0.5° x 0.5° 4 Latent heat surface 12-14 km (deg/hr) Monthly 0.5° x 0.5° 4 Latent heat surface 14-16 km (deg/hr) Monthly 0.5° x 0.5° 4 Latent heat surface 16-18 km (deg/hr) Monthly 0.5° x 0.5° 4 FASIR Normalized Difference Vegetation Index Monthly 0.25° x 0.25° 5 1 University Corporation for Atmospheric Research, National Center for Atmospheric Research (Trenberth, 1984). 2 Belize Meteorological Service (BMS) Daily Weather Observations, Central Farm Station (17° 1T N lat., 89° W long., 64 m elevation). 3 MODIS/Terra Atmosphere Monthly Global Product (MOD08 M3), NASA GES DISC (DAAC, 2006). 4 TRMM V6 Monthly TMI rain, latent heat, cloud liquid water profiles (3A12 V6), NASA GES DISC (DAAC, 2006). 5 Monthly FASIR-NDVI; The International Satellite Land-Surface Climatology Project, Initiative II Data Archive (ISLSCP II).
76
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. without prohibited reproduction Further owner. copyright the of permission with Reproduced
Table 4-2. SOI Lag Cross-Correlation Coefficients. Items in bold were significant at the 95% confidence interval. 2 § CU H o £ 3 o o s o- CQ o Po t- o t- I & •I 2 Q c2 | # o l § Q .§ *S s < ? GA cd GO5 J3 § C a - ^ g *►? 0 0 2 r r GO 6 s 3 s > CD cd p ® Q u x h I * •I *1H /"} — /—J / —H - < 2^ 2^ £ H * r —Hr o 3 H < z i a q J Q 1 .8. .1 2 & b a> vt-i cd GG
vs so NCN CN d d d d d d d d d d d d r-H GO 00 © CN xt d OS CN d d d o p os Xt O O o sr- vs p xt r-H © o «—«o r-H so d o d CN r- d Os O vs r-H f SO d V) rH d o H 1 ». ■ i i i i CN os vs 1—4 00 00 o o o © so vs OC OC CO CO SO CO CO d Tfr r-H o Nr CN r- CN SO CO d O d Os d r-H vs - r-H 1-H d fr T r-H o SO d FH CN ,i i i ■ i 1 cNcoxfrvssor-ooON CO CO CN so Ss so so OS o o OS o o o vs O CN CN CN o o O d d n Os d CN o co © r-H d r-H f—H d p d o O o o d Os so ©\ o o o NC CN o CN CN d o o < ■ i i 1 i ( n d d d o d d tj CN CO NCO CN - 1—4 H i r-H CN © p o p t-- CN d CO txt xt d CN vs OO d N f N so © 1-H o o d V) o r CN d 1-H « H i i i • « h i - r- O r—4 p o o d d d xj- d o o O O 0Os 00 1-H CN d O d o o r-H © VS © so CN O vs VS o o d O OS o o 1—1 d 1-H vs 00 ,-H r i i 1 i i n n o CN d 1-H CO CO t- vs CN O SO © © OCO CO d CO o o o d d d OS 1-H o so CN d xt o O o r— 1—4 O d • t i 1 n 00 Xf Os CN 1—4 1-H so Os o Oot- o CO r-H So OS o d 0ooo o o 00 d o d —Hr vs d p O d p O sOs Os d 1—4 o d o o 1-H 1-H 1-H sSO Os © d I/O • i 1 i ■ CN o V sr OS r- vs r- - 1-H 1-H xr r- i d d © © so xt r- d vs d r-H o O os CO d d d CN CN o 00 WH 0o 00 o o © d d - h • i i i CN d 1-H CN o NCN CN o d CN ^H O d d d d d r- © tOS xt d o O O p p o o xt d d P o r- Os 00 CN p o p VO o 1—H o —
Table 4-2. Continued. s . IS T*-< 5 w 5 9 . GO 2 S cd . o r> o c c O i_ cd X cd o o S 3 o c >
'3
T3 £ ,-o *'$b S t *G 2 *5 ^5—^ ✓~N /- /V -- Cu s cd £ G O C cn in £ c 3 >? h h \ . ' NDVI ooooooooooooo ooooorsooooooo */~>or-r^^i-t-HOONONVot‘-<'OTf“ OO VO © ON CO > I VO os ’—1 © 1 t ■ CN O O O O O O CN O O O O O O 1—1CO c o c o CO o c o c COCO © o © © Od"VO " COd © © © © © © © C NCN CN CO © © © © ’ © © © © © t r s - r os ^ r oo - d - d C-- © © © © © © © © © © © © © © © © © ©’ © © © © © © 1 I 1 1 1 i i i i © © © © © © © © n i o c i-H h - H 1 1 C~- C CN CN © © © ©
© © © © © © © © © n i © © © © © VO r*** vo co r***vo VO O© CO ©©© © © © © © © © © 1-H © © © © 9 © © CN n i N© CN i/O © © ■ i i t t 1 1 © © © © © © © - 00 r-H - r © r-H ©© © © © © Os© s 1—H O * d vo - r © 00 © © © © © © © © © © © © © © © © © © © © © © © - d © 00 i i i i 1 1 N - - N co CN co ©© O ■ VO VO © © © © © © r-HO CN ©© © © 1—HO < CN n © rN i - h
CN co CO - r © © © ' © © © © © O © oo oo © © CO © CN 1-H VO 01 O — VO OO — © 1 00 ,—4 1-H 1 I
t OOoONO^'4 ^r ^—'04m O N ooO O 'O ^tm 0 . 1 4 6 O© V ■'d' VO t CO© N COCN d C O © CNVO ’d- N C n t © © © © T-H Td- OO r-H VO rH © © © © © © © rH ©© © n c ’ '*CO 'd* ’— cN - N vn OS 00 CN uo VO uo CN 00 OS © © © © © © © © © © © © © © © ’d © O - d © co - d cn i— 'd- t o co CN CN vo vo CN CN co o t 'd- i— © © © © © © © © d oc oo *H * oo co co co ’d O - d © © © O H - r © 0 1—H© CO00 © vo © © * © © © © © © © © © © © © © © © O - V *—< CO r-H VO o COOO rH o o o o o s o - ^4 ' d 'd "d- ^ - If) VO ^ t o o © t © •11 1 1 1 1 1 • © • t • © © © © © © © h © © ■^d-co os vo © © © © i n r -i d -i r n i © i © © 1 - —H r-H r-H h t ■ N CO CN © © © © CO © i i © ©
CO © © © © © © - r © © 1-H N C © © © © © © © n i © O CN VO 1 -H C ) © ) C -H 1 1 1 1 o © © © © © o co co © CO © © ©© © © © o © o cn © CO r - co - r CO © VOVO o o o o 1—CO 1—H - COr-H t t - r n i 0 0
©© © © vo 00 © o O 1-H O n TT VO TT Os vo Os © r-H © •d- CN 'd" © o © T-Hi-H j ^H rj" © © © CO H © © © o o o o © o © s w © d o c C"-COVO *-h © © © n i n i i1 i i i i t t t i os n
o © CN 1—H © © CO CO © © N" o r-H r— vo © 1-H i 1 1
78 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. without prohibited reproduction Further owner. copyright the of permission with Reproduced
Table 4-2. Continued. H3 U XJ v O !H £ H o 3 O O o- 0> u "O 0 T3 -a W O 0 -O 3 U u U H tC § o o cd O o c/i s_ cd o G OI 3 > v 3 o\ h h
*o o u a- p cd ' o l C
cd » e Q rC p O h XJ £ £ r H s o Cd g o a O ih o
OPj O O ^ O jO P fO O ^ O O ^ - r o CN so o ® p ® CN NCN CN 0*0 © © * o ® O CN Os p o -'O’ O O O ’- ^ O ' ’- — ^ O O O - r © 0*000000000 Os r-H r-H ncn cn oo o - r soTfO i> Os f T o os ® so cn ® cn Hr-H —H r r-H 1-H cn © ® o o o 00 o © o o *— ■ It I I I I I I t I I 1 • © cn © n c ON O ® cn ® Os o © O H * o CN o o cn Os cn CN r- ® 0 0 © © OO p © 0 0 9 ® i t 79 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. without prohibited reproduction Further owner. copyright the of permission with Reproduced Table 4-2. Continued. 33 13) J3 "S c3 Date of ATM7 calcite deposition 2000 1995 1990 1985 1980 --6 -40 5 13C (per mil) V-PDB -20 ~-10 o - ■--14 2000 1995 1990 1985 1980 SOI date 81 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 4-2. Timeline of Data Sources. Abbreviations as in Table 4-1. 1 - SOI 2 - BMS Observations 3 - FASIR-NDVI 4 - MODIS/Terra Atmosphere (MOD08_M3)| 5 - TRMM V6 Monthly TMI (3A12 V6); ISLSCP II ______ I I I i I I I I 1970 1975 1980 1985 1990 1995 2000 2005 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CONCLUSIONS In the preceding chapters, I have demonstrated the sensitivity of a rapidly-growing, fracture-fed tropical stalagmite record to regional tropical cyclone activity, major regional volcanic tephra deposition events, and weak El Nino teleconnections. The implications of this research are exciting, as this work has generated far more questions than it has answered. In particular, this work forms the basis for new approaches to paleotempestology, and contributes to an emerging effort to generate speleothem-based records of explosive volcanic activity in the past. The field of speleothem paleoclimatology is in its infancy, and there is great potential for scientific discoveries in this area of inquiry. The results discussed herein form a foundation upon which to develop new tools for paleo-environmental research that will enable the broader research community to address outstanding fundamental questions in new ways, for example: Do field sample selection methods actually result in more sensitive stalagmite proxy records? What conditions must be satisfied for speleothem proxy records to be reproducible? How common are stalagmites that are sensitive recorders of tropical cyclone rainfall? What is the return period for catastrophic tropical cyclone strikes in regions without extensive historical records? What was hurricane activity like during intervals of paleoclimatic interest, such as the Last Glacial Maximum and the Eemian interglacial? What are the relations between tropical cyclone landfalls and large-scale climate boundary conditions? 83 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. How does weak hurricane activity during pre-historic “hyperactive” tropical cyclone periods compare to the 20th century (e.g. Liu, 2000)? Can stalagmite trace element spectra distinguish between tephra from different volcanoes? How sensitive are stalagmites to regional and pan-tropical volcanic fallout? Do major volcanic eruptions generate marker horizons recognizable in stalagmites across a region? Can stalagmites provide annually resolved records of tropical volcanic forcing akin to polar ice core records of high-latitude volcanism? By what mechanism(s) are weak local teleconnections amplified into strong carbon isotope proxy signals? The fidelity and sensitivity with which different speleothems record proxy evidence of various environmental factors depends fundamentally on the epikarst channel that acts as a signal filter between the aboveground environment and an active stalagmite surface. The amplitude of a proxy may not reflect the amplitude of the local forcing mechanism (Chapter 2, Chapter 4). Stalagmites that are ideally sensitive to one factor (such as hurricane infiltration events) may be less faithful recorders of other factors of interest. Trade-offs abound; for example, stalagmites selected for their responsiveness to brief, flashy infiltration may also be more likely to contain visible layering related to sub-annual events that must be carefully distinguished from annual layers (Chapter 2). The internal cave environment is often very stable; yet, stochastic events can cause singularities that must be distinguished from environmental controls. Coherence between multiple records is strong evidence of reliable proxy signals (Dorale et al., 2002a). Although analyzing multiple stalagmites was beyond the scope of the present work, replication of stalagmite records is a necessary next step in the development of paleo-tempest and paleo- 84 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. volcanism proxies. For the potential of speleothem approaches to Earth System Science research to be realized, existing singular records must be replicated. Developing new, multi-proxy records at the highest possible resolution is one avenue that leads effectively toward greater insight into the mechanisms and history implied by the stalagmite records. Comparing the behavior of different parameters can provide greater insight into system dynamics. Consider the findings that no ENSO events are evident in the trace element record, and that no volcanic signals are found in the stable isotope record. Even when the mechanisms are incompletely known, speleothem record exploration can reveal processes of interest occurring in the cave, ecosystem, or regional environment (Chapters 1-4). It is critical to conduct modem calibration of stalagmite proxy signals at all study sites as soon as it is feasible to deploy the monitoring equipment, particularly in cases where the exact mechanism of signal emplacement is unknown. Thus, a second line of inquiry that is likely to revolutionize our understanding of the possibilities and pitfalls of speleothem records involves modem process studies. This approach can integrate basic observational, experimental, and modeling approaches, and has been used successfully by the ice core community (Dibb, 1989; Wake et al., 1993; Wake et al., 2002). Few tropical cave sites are subject to systematic monitoring (Mickler et al., 2004a; Sondag et al., 2003). At this time, advancing in the state of the science particularly requires systematic, real-time studies of the links between modem cave systems, the overlying above- and below-ground ecosystem that acts as a biogeochemical signal processor, and external forcing factors from precipitation, climatic variability, aerosol deposition. Field relations indicating epikarst pathway characteristics and likely sensitivities as well as sensitivity trade-offs should be included in field sample collection and laboratory selection procedures in order to focus analytical resources on samples (e.g. Chapter 1), 85 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Future Directions My current research track has two rails: terrestrial carbon cycling and paleo- tempestology, the emerging science of past hurricanes. Through the former, I hope to shed light on the relation between climate variability and the natural/perturbed carbon cycle. Through the latter, I hope to quantify the relations between climate regimes and hurricane activity, a matter only beginning to come to the attention of the general public. Going forward, I am dedicated to contributing key datasets that could be used to address fundamental scientific questions with societal relevance, such as: • Beyond the instrumental record, how is hurricane activity related to climate factors such as ocean temperature and ENSO? • What links exist between hurricane activity and thermohaline circulation? • How important are the tropics in controlling climate change? • What triggers rapid changes in tropical moisture (e.g. Mega-droughts)? • How extensively do weak climatic variations perturb the terrestrial carbon cycle? • How have seasonal climate variability/weather extremes affected ecosystems/human culture? For example, in Central America, the relative importance of socioeconomic vs. environmental factors in precipitating cultural change will remain unresolved until the history of climatic and ecological changes are better constrained. The initial stalagmite tropical cyclone proxy record was validated against the historical record (Frappier et al., 2002; Chapter 2, Chapter 3). Based on our initial stalagmite record, we developed and calibrated a quantitative proxy model to reconstruct tropical cyclone frequency and intensity. However, to further validate the approach, it will be necessary to obtain additional 86 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. speleothem records on which to apply the model to determine its universal applicability so that it can be subsequently applied to times prior to observational records. Once the proposed paleo- hurricane proxy technique has been fully tested, it should be possible for paleotempestologists to quantify variations in tropical cyclone landfall frequency and intensity in any tropical cyclone basin during intervals of paleoclimatic interest throughout the Holocene and in the more distant past. Toward that end, I am pursuing a follow-up project to replicate and further validate the new speleothem paleo-Hurricane proxy. I have proposed to develop and test the proxy by producing three century-long high-resolution stalagmite stable isotope records and then apply it to an ancient century-long time interval for which we have no instrumental record. In collaboration with my colleagues from a variety of institutions, I will continue to pursue paleotempestology and the modem and ancient expressions of terrestrial biogeochemical change. Despite their devastating impact on societies throughout history, we know very little about the interactions between tropical cyclones and the climate system. I plan to apply my newly-developed tool to investigate hurricane frequency and intensity in the past during different climate regimes, which will provide new and valuable perspectives to inform modeling efforts and advance understanding of the consequences of global climate change for hurricane activity. Developing a spatially distributed network of cave sites is central to addressing the questions posed above. I recently helped establish a new international Collaborative Research Network for Paleotempestology. The Inter-American Institute for Global Change Research (IAI) has just awarded our group a major five-year grant to develop a pan-Caribbean network of paleo- hurricane records and to translate our findings into a regional risk GIS database for policy-makers and other stakeholders. My role in the network will be to lead the effort to sharpen the new speleothem hurricane proxy into a broadly applicable scientific tool. I am also developing relationships with ecologists in order to investigate the impact of the area’s weak El Nino teleconnections on carbon fluxes and ecosystem function in more detail, from a systems perspective. I hope to advance this research in cooperation with local Belizean 87 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. scientists by establishing a regular sampling program for water and soil carbon, and outfitting the site with an array of datalogging sensors. This will lay the groundwork for future terrestrial paleoecological and carbon cycling studies. This dissertation research has demonstrated that speleothems are a very promising source of high-resolution, multi-proxy paleo-environmental data. Speleothems are indeed emerging as the ‘ice cores of the lowland tropics’. In particular, I am optimistic about the potential for tropical speleothem records to revolutionize the current understanding of key climate system interactions, principally volcanic climate forcing, tropical cyclone-climate interactions, and the links between land-use history and terrestrial carbon cycling. Our understanding of past, present, and future interactions within the Earth System will grow as speleothem-based paleoclimatology matures. I anticipate contributing to this effort by building on the lines of inquiry initiated during this dissertation research. 88 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LIST OF REFERENCES Adams, J.B., Mann, M.E., and Ammann, C.M., 2003, Proxy evidence for an El Nino-like response to volcanic forcing: Nature, v. 426, p. 274-278. Amundson, R., Stem, L., Baisden, T., and Wang, Y., 1998, The isotopic composition of soil and soil-respired C02: Geoderma, v. 82, p. 83-114. Asmerom, Y., and Polyak, V.J., 2004, A test of annual resolution in stalagmites using tree rings: Quaternary Research, v. 61, p. 119-121. Ayalon, A., Bar-Matthews, M., and Kaufman, A., 1999, Petrography, strontium, barium and uranium concentrations, and strontium and uranium isotope ratios in speleothems as palaeoclimatic proxies: Soreq Cave, Israel: Holocene, v. 9, p. 715- 722. Ayalon, A., Bar-Matthews, M., and Sass, E., 1998, Rainfall-rechage relationships within a karstic terrain in the Eastern Mediterranean semi-arid region, Israel: d l8 0 and dD characteristics.: Journal of Hydrology, v. 207, p. 18-31. Baker, A., 1999, A Scottish speleothem record of the H-3 eruption or human impart? A reply to Dugmore, Coles and Buckland: Holocene, v. 9, p. 503-503. —, 2002, Drip-water hydrology, geochemistry and fluorescence at Stump Cross Caverns, Yorkshire, England, North-Central Section (36th) and Southeastern Section (51st), GSA Joint Annual Meeting: Lexington, Kentucky, p. Paper No. 24-0. Baker, A., and Brunsdon, C., 2003, Non-linearities in drip water hydrology: an example from Stump Cross Caverns, Yorkshire: Journal of Hydrology, v. 277, p. 151-163. Baker, A., Smart, P.L., Barnes, W.L., Edwards, R.L., and Farrant, A., 1995, The Hekla-3 Volcanic-Eruption Recorded in a Scottish Speleothem: Holocene, v. 5, p. 336-342. Baker, A., Smart, P.L., Edwards, R.L., and Richards, D.A., 1993, Annual Banding in a Cave Stalagmite: Nature, v. 364, p. 518-520. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Baskaran, M., and Krishnamnrthy, R.V., 1993, The dl3C in Cave deposits: a proxy indicator of the dl3C of atmospheric C02.: Geophysical Research Letters, v. 20, p. 2905-2908. Behrenfeld, M., Randerson, J., Mclain, C., Feldman, C., Los, S., and others, 2001, Biospheric primary production during an ENSO transition: Science, v. 291, p. 2594. Bertaux, J., Sondag, F., Santos, R., Soubies, F., Causse, C., Plagnes, V., Le Comec, F., and Seidel, A., 2002, Paleoclimatic record of speleothems in a tropical region: study of laminated sequences from a Holocene stalagmite in Central-West Brazil: Quaternary International, v. 89, p. 3-16. Bluth, G.J.S., Rose, W.I., Sprod, I.E., and Krueger, A.J., 1997, Stratospheric loading from explosive volcanic eruptions: Journal of Geology, v. 105, p. 671-683. Bluth, G.J.S., Schnetzler, C.C., Krueger, A.J., and Walter, L.S., 1993, The Contribution of Explosive Volcanism to Global Atmospheric Sulfur-Dioxide Concentrations: Nature, v. 366, p. 327-329. Bogli, A., 1980, Karst Hydrology and Physical Speleology: New York, Springer-Verlag, 284 p. Bousquet, P., Peylin, P., Ciais, P., Quere, C.L., Friedlingstein, P., and Tans, P.P., 2000, Regional changes in carbon dioxide fluxes of land and oceans since 1980: Science, v. 290, p. 1342-1346. Bradley, R.S., and Jones, P.D., 1992, Records of explosive volcanic eruptions over the last 500 years, in Bradley, R.S., and Jones, P.D., eds., Climate Since A. D. 1500: New York, Routledge, p. 606-622. Braswell, B.H., Schimel, D.S., Linder, E., and Moore, B., 1997, The response of global terrestrial ecosystems to interannual temperature variability: Science, v. 278, p. 870-872. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Briffa, K.R., Jones, P.D., Schweingruber, F.H., and Osborn, T.J., 1998, Influence of volcanic eruptions on Northern Hemisphere summer temperature over the past 600 years: Nature, v. 393, p. 450-455. Briffa, K.R., Jones, P.D., Schweingruber, F.H., Shiyatov, S.G., and Cook, E.R., 1995, Unusual twentieth-century summer warmth in a 1,000-year temperature record from Siberia: Nature, v. 376, p. 156-159. Broecker, W.S., Olson, E.A., and Orr, P.C., 1960, Radiocarbon measurements and annual rings in cave formations: Nature, v. 185, p. 93-94. Brook, G.A., Rafter, M.A., Railsback, L.B., Sheen, S.W., and Lundberg, J., 1999, A high-resolution proxy record of rainfall and ENSO since AD 1550 from layering in stalagmites from Anjohibe Cave, Madagascar: Holocene, v. 9, p. 695-705. Brown, H.I., 1987, Observation and Objectivity: New York, Oxford University Press, 255 p. Burkett, Y.R., Wilcox, D.A., Stottlemyer, R., Barrow, W., Fagre, D., Baron, J., Price, J., Nielsen, J.L., Allen, C.D., Peterson, D.L., Ruggerone, G., and Doyle, T., 2005, Nonlinear dynamics in ecosystem response to climatic change: Case studies and policy implications: Ecological Complexity, v. 2, p. 357-394. Bums, S.J., Fleitmann, D., Mudelsee, M., Neff, U., Matter, A., and Mangini, A., 2002, A 780-year annually resolved record of Indian Ocean monsoon precipitation from a speleothem from south Oman: Journal of Geophysical Research-Atmospheres, v. 107. Carpenter, S.J., 1996, A new microsampling device for extracting high-resolution isotope data from geologic and biologic materials.: Geological Society of America Abstracts with Programs, p. 360. Cerling, T., Solomon, D., Quade, J., and Bowman, J., 1991, On the isotopic composition of carbon in soil carbon dioxide: Geochim. Cosmochim. Acta, v. 55, p. 3403-3405. 91 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Chiang, J.C.H., and Sobel, A.H., 2002, Tropical tropospheric temperature variations caused by ENSO and their influence on the remote tropical climate: Journal of Climate, v. 15, p. 2616-2631. Clausen, H.B., and Hammer, C.U., 1988, The Laki and Tambora eruptions as revealed in Greenland ice cores from 11 locations: Annals of Glaciology, v. 10, p. 16-22. Clausen, H.B., Hammer, C.U., Hvidberg, C.S., DahlJensen, D., Steffensen, J.P., Kipfstuhl, J., and Legrand, M., 1997, A comparison of the volcanic records over the past 4000 years from the Greenland Ice Core Project and Dye 3 Greenland Ice Cores: Journal of Geophysical Research-Oceans, v. 102, p. 26707-26723. Cohen, A., 2001, Hurricane Tracking with Chemistry: Exploring the Coral Archive, Workshop on Atlantic Basin Paleohurricane Reconstructions from High Resolution Records: University of South Carolina, Columbia, South Carolina, U.S.A. Crowley, T.J., 2000, Causes of Climate Change Over the Past 1000 Years: Science, v. 289, p. 270-277. Cuffey, K.M., Alley, R.B., Grootes, P.M., Bolzan, J.F., and Anandakrishnan, S., 1994, Calibration of the deltal80 isotopic paleothermometer for central Greenland, using borehole temperatures: Journal of Glaciology, v. 40, p. 341-349. Cuffey, K.M., and Steig, E.J., 1998, Isotopic diffusion in polar firm Implications for interpretation of seasonal climate parameters in ice core records, with emphasis on central Greenland: Journal of Glaciology, v. 44, p. 273-284. DAAC, 2006, The data used in this study were acquired as part of the NASA's Earth-Sun System Division and archived and distributed by the Goddard Earth Sciences (GES) Data and Information Services Center (DISC) Distributed Active Archive Center (DAAC). http://daac.gsfc.nasa.gov/. Volume 2006. Dansgaard, W., 1964a, Stable isotopes in precipitation: Tellus, v. 16, p. 436-468. —, 1964b, Stable isotopes in precipitation.: Tellus, v. 16. 92 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Delmas, R.J., Kirchner, S., Palais, J.M., and Petit, J.R., 1992, 1000 Years of Explosive Volcanism Recorded at the South-Pole: Tellus Series B-Chemical and Physical Meteorology, v. 44, p. 335-350. Delmas, R.J., Legrand, M., Aristarain, A.J., and Zanolini, F., 1985, Volcanic deposits in Antarctic snow and ice: Journal of Geophysical Research, v. 90, p. 12,901-12,920. Denniston, R.F., Gonzalez, L.A., Baker, R.G., Asmerom, Y., Reagan, M.K., Edwards, R.L., and Alexander, C.E., 1999, Speleothem evidence for Holocene fluctuations of the prairie-forest ecotone, north-central USA: Holocene, v. 9, p. 671-676. Diaz, H., and Kiladis, G., 1992, Atmospheric Teleconnections Associated with the Extreme Phase of the Southern Oscillation, in Diaz, H., and Markgraf, V., eds., El Nino: Cambridge, Cambridge Univ. Press, p. 7-28. Dibb, J., 1989, The Chernobyl reference horizon in the Greenland ice sheet: Geophysical Research Letters, v. 16, p. 987-990. Donnelly, J.P., Bryant, S.S., Butler, J., Dowling, J., and others., a., 2001a, 700 yr sedimentary record of intense hurricane landfalls in southern New England.: Geological Society of America Bulletin, v. 113, p. 714-727. Donnelly, J.P., Roll, S., Wengren, M., Butler, J., Lederer, R., and Webb III, T., 2001b, Sedimentary evidence of intense hurricane strikes from New Jersey: Geology, v. 29, p. 615-618. Dorale, J.A., Edwards, R.L., and Onac, B.P., 2002a, Stable isotopes as environmental indicators in speleothems, in Yuan, S., ed., IGCP379, Volume Karst Processes and the Carbon Cycle. Final Report of IGCP 379: Beijing, China, Geological Publishing House. —, 2002b, Stable isotopes as environmental indicators in speleothems, in Yuan, S., ed., Karst Processes and the Carbon Cycle: Final Report of IGCP379: Beijing, Geological Publishing House, p. 107-120. 93 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Dorale, J.A., Gonzalez, L.A., Reagan, M.K., Pickett, D.A., and et al., 1992, A high- resolution record of Holocene climate change in speleothem calcite from Cold Water Cave, Northeast Iowa: Science, v. 258, p. 1626-1630. Dorr, H., and Miinnich, K.O., 1989, Downward movement of soil organic matter and its influence on trace-element transport (210Pb, 137Cs) in the soil: Radiocarbon, v. 31, p. 655-663. Dreybrodt, W., 1980, Deposition of Calcite from Thin-Films of Natural Calcareous Solutions and the Growth of Speleothems: Chemical Geology, v. 29, p. 89-105. —, 1981, The Kinetics of Calcite Precipitation from Thin-Films of Calcareous Solutions and the Growth of Speleothems - Revisited: Chemical Geology, v. 32, p. 237-245. Eisner, J.B., and Liu, K.B., 2003, Examining the ENSO-typhoon hypothesis: Climate Research, v. 25, p. 43-54. Eisner, J.B., Liu, K.-B., and B.Kocher, 2000, Spatial variations in major U.S. hurricane activity: Statistics and a physical mechanism.: Journal of Climate, v. 13, p. 2293- 2305. Emanuel, K., 2001, Contribution of tropical cyclones to meridional heat transport by the oceans.: Journal of Geophysical Research-Atmospheres, v. 106, p. 14771-14781. —, 2005, Increasing destructiveness of tropical cyclones over the past 30 years: Nature, v. 436, p. 686-688. ERDAS, 1999, Earth Resources Data Analysis System (ERDAS) Field Guide , ERDAS Inc. Ewel, J., Berish, C., Brown, B., Price, N., and Raich, J., 1981, Slash and Bum Impacts on a Costa Rican Wet Forest Site: Ecology, v. 62, p. 816-829. Fairchild, I.J., Baker, A., Borsato, A., Frisia, S., Hinton, R.W., McDermott, F., and Tooth, A.F., 2001, Annual to sub-annual resolution of multiple trace-element trends in speleothems: Journal of the Geological Society, v. 158, p. 831-841. 94 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Fairchild, I.J., Borsato, A., Tooth, A.F., Frisia, S., Hawkesworth, C.J., Huang, Y., McDermott, F., and Spiro, B., 2000, Controls on trace element (Sr-Mg) compositions of carbonate cave waters: implications for speleothem climatic records: Chemical Geology, v. 166, p. 255-269. Fairchild, I.J., Smith, C.L., Baker, A., Fuller, L., Spotl, C., Mattey, D., McDermott, F., and E.I.M.F., 2006, Modification and preservation of environmental signals in speleothems, Earth Science Reviews, Volume 75: ISOtopes in PALaeoenvironmental reconstruction (ISOPAL), p. 105-153. Farquhar, G.D., Ehleringer, J.R., and Hubick, K.T., 1989, Carbon isotope discrimination and photosynthesis: Annu. Rev. Plant Physiol. Plant Mol. Biol., v. 40, p. 503-537. Ford, D.C., and Williams, P.W., 1989, Karst geomorphology and hydrology: London, Unwin Hyman. Frappier, A., Sahagian, D., Gonzales, L., and Carpenter, S., 2002a, El Nino events recorded by stalagmite carbon isotopes: Science, v. 298, p. 565. Frappier, A., Sahagian, D., Gonzalez, L.A., and Carpenter, S.J., 2002b, El Nino events recorded by stalagmite carbon isotopes: Science, v. 298, p. 565-565. Frisia, S., Borsato, A., Fairchild, I.J., and McDermott, F., 2000, Calcite fabrics, growth mechanisms, and environments of formation in speleothems from the Italian Alps and southwestern Ireland: Journal of Sedimentary Research, v. 70, p. 1183-1196. Frisia, S., Borsato, A., Fairchild, I.J., and Susini, J., 2005, Variations in atmospheric sulphate recorded in stalagmites by synchrotron micro-XU and XANES analyses: Earth and Planetary Science Letters, v. 235, p. 729-740. Frisia, S., Borsato, A., Preto, N., and McDermott, F., 2003, Late Holocene annual growth in three Alpine stalagmites records the influence of solar activity and the North Atlantic Oscillation on winter climate: Earth and Planetary Science Letters, v. 216, p. 411-424. Frutos, R., 2006, The Climate of Belize, http://www.hydromet.gov.bz/: Belmopan, Belize. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Gedzelman, S., Lawrence, J., Gamache, J., Black, M., Hindman, E., Black, R., Dunion, J., Willoughby, H., and Zhang, X.P., 2003, Probing hurricanes with stable isotopes of rain and water vapor: Monthly Weather Review, v. 131, p. 1112-1127. Gedzelman, S.D., and Lawrence, J.R., 1982, The Isotopic Composition of Cyclonic Precipitation: Journal of Applied Meteorology, v. 21, p. 1385-1404. Genty, D., Baker, A., Massault, M., Proctor, C., Gilmour, M., Pons-Branchu, E., and Hamelin, B., 2001, Dead carbon in stalagmites: Carbonate bedrock paleodissolution vs. ageing of soil organic matter. Implications for C-13 variations in speleothems: Geochimica Et Cosmochimica Acta, v. 65, p. 3443-3457. Genty, D., Blamart, D., Ouahdi, R., Gilmour, M., Baker, A., Jouzel, J., and Van-Exter, S., 2003, Precise dating of Dansgaard-Oeschger climate oscillations in western Europe from stalagmite data: Nature, v. 421, p. 833-837. Giannini, A., Cane, M.A., and Kushnir, Y., 2001, Interdecadal changes in the ENSO teleconnection to the Caribbean region and the North Atlantic oscillation: Journal of Climate, v. 14, p. 2867-2879. Giannini, A., Kushnir, Y., and Cane, M.A., 2000, Interannual variability of Caribbean rainfall, ENSO, and the Atlantic Ocean: Journal of Climate, v. 13, p. 297-311. Gill, R.B., 2000, The Great Maya Droughts: Water, Life, and Death: Albuquerque, University of New Mexico Press, 464 p. Giorgi, F., Hewitson, B., Christensen, J., Hulme, M., Storch, H.V., Whetton, P., Jones, R., Mearns, L., and Fu, C., 2001, Regional climate change information—evaluation and projections., in Houghton, J.T., Ding, Y., Griggs, D.J., Noguer, M., van der Linden, P.J., and Xiaosu, D., eds., Climate Change 2001 - The Scientific Basis.: Cambridge and New York, Cambridge University Press, p. 583-638. Graham, E., Mulkey, S., Kitajima, K., and others, 2003, Cloud cover limits net C02 uptake and growth of a rainforest tree during tropical rainy seasons: Proc. Natl. Acad. Sci., v. 100, p. 572. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Grootes, P.M., Stuiver, M., White, J.W.C., Johnsen, S., and Jouzel, J., 1993, Comparison of oxygen-isotope records from the GISP2 and GRIP Greenland ice cores: Nature, v. 366, p. 552-554. Hall, F.G., Collatz, G.J., Los, S.O., Brown de Colstoun, E., and Landis, D., 2005, ISLSCP Intiative II http://islscp2.sesda.com/, DYD/CD-ROM, International Satellite Land Surface Climatology Project (ISLSCP): Washington D.C., NASA. Harmon, R.S., Schwarcz, H.P., and Oneil, J.R., 1979, D-H Ratios in Speleothem Fluid Inclusions - Guide to Variations in the Isotopic Composition of Meteoric Precipitation: Earth and Planetary Science Letters, v. 42, p. 254-266. Hellstrom, J.C., and McCulloch, M.T., 2000, Multi-proxy constraints on the climatic significance of trace element records from a New Zealand speleothem: Earth and Planetary Science Letters, v. 179, p. 287-297. Henderson-Sellers, A., Zhang, H., Berz, G., Emanuel, K., Gray, W., Landsea, C., Holland, G„ Lighthill, J., Shieh, S.L., Webster, P., and McGuffie, K„ 1998, Tropical cyclones and global climate change: A post-IPCC assessment: Bulletin of the American Meteorological Society, v. 79, p. 19-38. Hendy, C.H., 1971, The isotopic geochemistry of speleothems-I. The calculation of the effects of different modes of formation on the isotopic composition of speleothems and their applicability as palaeoclimatic indicators: Geochimica et Cosmochimica Acta, v. 35, p. 801-824. Hernandez, T., Garcia, C., and Reinhardt, I., 1997, Short-term effect of wildfire on the chemical, biochemical and microbiological properties of Mediterranean pine forest soils: Biology of Fertile Soils, v. 25, p. 109-116. Hill, C.A., and Forti, P., 1997, Cave Minerals of the World: Huntsville, Alabama, USA, National Speleological Society. Hou, J.Z., Tan, M., and Liu, D.S., 2002, Counting chronology and climate records with about 1000 annual layers of a Holocene stalagmite from the Water Cave in Liaoning Province, China: Science in China Series D-Earth Sciences, v. 45, p. 385- 391. 97 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Huang, H.M., Fairchild, I.J., Borsato, A., Frisia, S., Cassidy, N.J., McDermott, F., and Hawkesworth, C.J., 2001, Seasonal variations in Sr, Mg and P in modem speleothems (Grotta di Ernesto, Italy): Chemical Geology, v. 175, p. 429-448. Ihlenfeld, C., Norman, M.D., Gagan, M.K., Drysdale, R.N., Maas, R., and Webb, J., 2003, Climatic significance of seasonal trace element and stable isotope variations in a modem freshwater tufa: Geochimica Et Cosmochimica Acta, v. 67, p. 2341- 2357. Johnson, K.R., and Ingram, B.L., 2004, Spatial and temporal variability in the stable isotope systematics of modem precipitation in China: implications for paleoclimate reconstmctions: Earth and Planetary Science Letters, v. 220, p. 365-377. Jones, P.D., Osborn, T.J., and Briffa, K.R., 2001, The Evolution of Climate Over the Last Millennium: Science, v. 292, p. 662-667. Kaufmann, G., 2003, Stalagmite growth and palaeo-climate: the numerical perspective: Earth and Planetary Science Letters, v. 214, p. 251-266. Kaufmann, G., and Dreybrodt, W., 2004, Stalagmite growth and palaeo-climate: an inverse approach: Earth and Planetary Science Letters, v. 224, p. 529-545. Kendall, A.C., and Broughton, P.L., 1978, Origin of fabrics in speleothems composed of columnar calcite crystals: Journal of Sedimentary Petrology, v. 48, p. 519-538. King, D.N., and Jr., 1996, Cayo District Cretaceous-Tertiary boundary stratigraphy, Belize: Planetary Report, v. 16, p. 10-11, http:/7www.auburn.eduA kingdat/kt boundary page.htm#two. Klotzbach, P.J., and Gray, W.M., 2004, Updated 6-11-month prediction of Atlantic basin seasonal hurricane activity: Weather and Forecasting, v. 19, p. 917-934. Kong, X.G., Wang, Y.J., Wu, J.Y., and Cheng, H., 2003, A continuous 3000-year precipitation record of ENSO variability during LGM from a stalagmite in Nanjing: Chinese Science Bulletin, v. 48, p. 480-484. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Lachniet, M.S., Asmerom, Y., Bums, S.J., Patterson, W.P., Polyak, Y.J., and Seltzer, G.O., 2004a, Tropical response to the 8200 yr BP cold event? Speleothem isotopes indicate a weakened early Holocene monsoon in Costa Rica: Geology, v. 32, p. 957-960. Lachniet, M.S., Bums, S.J., Pipemo, D.R., Asmerom, Y., Polyak, V.J., Moy, C.M., and Christenson, K., 2004b, A 1500-year El Nino/Southem Oscillation and rainfall history for the Isthmus of Panama from speleothem calcite: Journal of Geophysical Research-Atmospheres, v. 109. Lachniet, M.S., and Patterson, W.P., 2002, Stable isotope values of Costa Rican surface waters: Journal of Hydrology, v. 260, p. 135-150. Lamb, H.H., 1970, Volcanic dust in the atmosphere; with a chronology and assessment of its meteorological significance: Philosophical Transactions of the Royal Society of London A, v. 266, p. 425-553. Lauer, D.T., Morain, S.A., and Salomonson, V.V., 1997, The Landsat Program: It’s Origins, Evolutions, and Impacts: Photogrammetric Engineering and Remote Sensing, v. 63, p. 831- 38. Lawrence, J.R., 1998, Isotopic spikes from tropical cyclones in surface waters: Opportunities in hydrology and paleoclimatology: Chemical Geology, v. 144, p. 153-160. Lawrence, J.R., and Gedzelman, S.D., 1996, Low stable isotope ratios of tropical cyclone rains: Geophysical Research Letters, v. 23, p. 527-530. Lawrence, J.R., Gedzelman, S.D., Dexheimer, D., Cho, H.K., Carrie, G.D., Gasparini, R., Anderson, C.R., Bowman, K.P., and Biggerstaff, M.I., 2004, Stable isotopic composition of water vapor in the tropics: Journal of Geophysical Research- Atmospheres, v. 109, p. D01:10.1029/2003JD004046. Lawrence, J.R., Gedzelman, S.D., Gamache, J., and Black, M., 2002, Stable isotope ratios: Hurricane Olivia: Journal of Atmospheric Chemistry, v. 41, p. 67-82. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Lawrence, J.R., Gedzelman, S.D., Zhang, X.P., and Arnold, R., 1998, Stable isotope ratios of rain and vapor in 1995 hurricanes: Journal of Geophysical Research- Atmospheres, v. 103, p. 11381-11400. Liu, K.B., and Feam, M.L., 1993, Lake-Sediment Record of Late Holocene Hurricane Activities from Coastal Alabama: Geology, v. 21, p. 793-796. —, 2000, Reconstruction of prehistoric landfall frequencies of catastrophic hurricanes in northwestern Florida from lake sediment records: Quaternary Research, v. 54, p. 238-245. —, 2002, Lake sediment evidence of coastal geologic evolution and hurricane history from Western Lake, Florida: Reply to Otvos: Quaternary Research, v. 57, p. 429- 431. Lohmann, K.C., 1987, Geochemical patterns of meteroric diagenetic systems and their application to studies of paleokarst: in Paleokarst,eds.N.P.James, and P.W. Choquette: Society of Economic Paleontologists and Mineralogists, p. p. 58-80. (125). —, 1988, Geochemical patterns of meteoric diagenetic systems and their application to studies of paleokarst, in James, N.P., and Choquette, P.W., eds., Paleokarst: Berlin, Springer-Verlag, p. 50-80. Lundblad, K., and Holmgren, K., 2005, Palaeoclimatological survey of stalagmites from coastal areas in Tanzania: Geografiska Annaler Series a-Physical Geography, v. 87A, p. 125-140. Lyons, W.B., Mayewski, P.A., Spencer, M.J., Twickler, M.S., and Graedel, T.E., 1990, A northern hemisphere volcanic chemistry record (1869-1984) and climatic implications using a South Greenland ice core: Annals of Glaciology, v. 14, p. 176- 182. Malhi, Y., Nobre, A.D., Grace, J., Kruijt, B., Pereira, M.G.P., Culf, A., and Scott, S., 1998, Carbon dioxide transfer over a Central Amazonian rain forest: Journal of Geophysical Research-Atmospheres, v. 103, p. 31593-31612. 100 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Malmquist, D.L., 1997, Oxygen isotopes in cave stalagmites as a proxy record of past tropical cyclone activity., 22nd Conference on Hurricanes and Tropical Meteorology, Volume 22: Proceedings of the 22nd Conference on Hurricanes and Tropical Meteorology: Fort Collins, American Meteorological Society, p. 393-394. Mann, M.E., 2002, Climate Reconstruction: The Value of Multiple Proxies: Science, v. 297, p. 1481-1482. Mann, M.E., Bradley, R.S., and Hughes, M.K., 1999, Northern hemisphere temperatures during the past millennium: Inferences, uncertainties, and limitations: Geophysical Research Letters, v. 26, p. 759-762. Mayewski, P.A., Meeker, L.D., Whitlow, S., Twickler, M.S., Morrison, M.C., Grootes, P.M., Bond, G.C., Alley, R.B., Meese, D.A., and Gow, T., 1994, Changes in atmospheric circulation and ocean ice cover over the North Atlantic region over the last 41,000 years: Science, v. 263. McColl, J.G., and Grigal, D.F., 1975, Forest fire: effects on phosphorus movement to lakes: Science, v. 188, p. 1109-1111. McCormick, P.M., Wang, P.H., and Poole, L.R., 1993, Stratospheric aerosols and clouds, in Hobbs, P.V., ed., Aerosol-Cloud-Climate Interactions: San Diego, Calif., Academic Press, p. 205- 220. McDermott, F., 2004, Palaeo-climate reconstruction from stable isotope variations in speleothems: a review: Quaternary Science Reviews, v. 23, p. 901-918. McMillan, E.A., Fairchild, I.J., Frisia, S., Borsato, A., and McDermott, F., 2005, Annual trace element cycles in calcite-aragonite speleothems: evidence of drought in the western Mediterranean 1200-1100 yr BP: Journal of Quaternary Science, v. 20, p. 423-433. Meeker, L.D., Mayewski, P.A., and others, 1995, A New Approach to Glaciochemical Time Series Analysis: NATO ASI Series, v. 130, p. 383-400. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Mickler, P.J., Banner, J.L., Stem, L., Asmerom, Y., Edwards, R.L., and Ito, E., 2004a, Stable isotope variations in modem tropical speleothems: Evaluating equilibrium vs. kinetic isotope effects: Geochimica et Cosmochimica Acta, v. 68, p. 4381-4393. Mickler, P.J., Ketcham, R.A., Colbert, M.W., and Banner, J.L., 2004b, Application of high-resolution X-ray computed tomography in determining the suitability of speleothems for use in paleoclimatic, paleohydrologic reconstructions: Journal of Cave and Karst Studies, v. 66, p. 4-8. Miller, D.L., Mora, C.I., Grissino-Mayer, H.D., Mock, C.J., and Uhle, M.E., Tropical Cyclone Activity and Climate Fluctuations Captured by Oxygen Isotopes in Tree- Ring Cellulose From the Southeastern US, Eos Trans. AGU, 84(46), Fall Meet. Suppl., Abstract B11C-0712, 2003. Miller, T.E., 1990, Belize: Tunichil Muknal: National Speleological Society News, v. 48, p. 32-35. —, 1996, Geologic and Hydrologic Controls on Karst and Cave Development in Belize: Journal of Cave and Karst Studies, v. 58, p. 100-120. Musgrove, M., and Banner, J.L., 2004, Controls on the spatial and temporal variability of vadose dripwater geochemistry: Edwards Aquifer, central Texas: Geochimica et Cosmochimica Acta, v. 68, p. 1007-1020. Newhall, G.C., and Self, S., 1982, The Volcanic Explosivity Index (VEI): an estimate of explosive magnitude for historical volcanism: Journal of Geophysical Research, v. 87, p. 1231-1238. Nott, J., 2004, Palaeotempestology: the study of and implications Review article prehistoric tropical cyclones - a review for hazard assessment: Environment International, v. 30, p. 433-447. Nott, J., and Hayne, M., 2001, High frequency of'super-cyclones' along the Great Barrier Reef over the past 5,000 years: Nature, v. 413, p. 508-512. Nott, J.F., 2003, Intensity of prehistoric tropical cyclones: Journal of Geophysical Research-Atmospheres, v. 108. 102 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Oppenheimer, C., 2003, Ice core and palaeoclimatic evidence for the timing and nature of the great mid-13th century volcanic eruption: International Journal of Climatology, v. 23, p. 417-426. Palais, J.M., Germani, M.S., and Zielinski, G.A., 1992, Interhemispheric transport of volcanic ash from the 1259 A.D. volcanic eruption to the Greenland and Antarctic Ice Sheets: Geophysical Research Letters, v. 19, p. 801-804. Palmer, A.S., 2001, High-precision dating of volcanic events - (AD 1301-1995) using ice cores from Law Dome, Antarctica: Journal of Geophysical Research-Atmospheres, v. 106, p. 28089. Pan, G., 1999, New progress in the study of annual lamination of stalagmites: Chinese Science Bulletin, v. 44, p. 1537-1537. Patterson, W.P., 1998, North American continental seasonality during the last millennium: high-resolution analysis of sagittal otoliths: Palaeogeography Palaeoclimatology Palaeoecology, v. 138, p. 271-303. Pearce, N.J.G., Perkins, W.T., Westgate, J.A., Gorton, M.P., Jackson, S.E., Neal, C.R., and Chenery, S.P., 1996, A compilation of new and published major and trace element data for NIST SRM 610 and NIST SRM 612 glass reference materials: Geostand. Newsl., v. 21, p. 115-144. Peixoto, J.P., and Oort, A.H., 1992, Physics of Climate: New York, Am. Inst, of Phys. Philander, S.G., 1990, El Nino, La Nina and the Southern Oscillation: Acdemic Press,San Diego, p. 293pp. Pielke, R.A., Jr., Landsea, C., Emanuel, K., Mayfield, M., Laver, J., and Pasch, R., 2005, Hurricanes and global warming: Bulletin of the American Meteorological Society, v. 86, p. 1571-1575. Pope, K.O., and Ocampo, A.C., 2000, Chicxulub high-altitude ballistic ejecta from central Belize: Lunar and Planetary Science, v. 31. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Proctor, C.J., Baker, A., Barnes, W.L., and Gilmour, R.A., 2000, A thousand year speleothem proxy record of North Atlantic climate from Scotland: Climate Dynamics, v. 16, p. 815-820. Qin, X.G., Tan, M., Liu, D.S., Wang, X.F., Li, T.Y., and Lu, J.P., 2000, Characteristics of annual laminae gray level variations in a stalagmite from Shihua Cave, Beijing and its climatic significance (II): Science in China Series D-Earth Sciences, v. 43, p. 521-533. R Core Development Team, 2004, R, http://cran.r-proiect.org/. Railsback, L.B., Brook, G.A., Chen, J., Kalin, R., and et ah, 1994, Environmental controls on the petrology of a late Holocene speleothem from Botswana with annual layers of aragonite and calcite: Journal of Sedimentary Research- A, v. 64, p. 147-155. Rampino, M.R., and Self, S., 1984, Sulphur-rich volcanic eruptions and stratospheric aerosols: Nature, v. 310, p. 677-679. Ritchie, and McHenry, 1990, Application of radioactive fallout cesium-137 for measuring soil erosion and sediment accumulation rates and patterns: A review: Journal of Environmental Quality, v. 19, p. 215-233. Robbins, J.A., Holmes, C., Hailey, R., Bothner, M., Shinn, E., Graney, J., Keeler, G., tenBrink, M., Orlandini, K.A., and Rudnick, D., 2000, Time-averaged fluxes of lead and fallout radionuclides to sediments in Florida Bay: Journal of Geophysical Research, v. 105, p. 28805-28821.Roberts, M.S., Smart, P.L., and Baker, A., 1998, Annual trace element variations in a Holocene speleothem: Earth and Planetary Science Letters, v. 154, p. 237-246. Robison, W.L., Conrado, C.L., and Hamilton, T.F., 1997, A comparative study on Cs 137 transfer from soil to vegetation in the Marshall Islands, International Meeting on Influence of Climatic Characteristics upon Behavior of Radioactive Elements: Aomori, Japan. October 14-16, 1997. Robock, A., 2000, Volcanic eruptions and climate: Reviews of Geophysics, v. 38, p. 191- 219. 104 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Robock, A., and Free, M.P., 1995, Ice cores as an index of global volcanism from 1850 to the present: Journal of Geophysical Research, v. 100, p. 11,549-11,567. Robock, A., and Matson, M., 1983, Circumglobal Transport of the El Chichon Volcanic Dust Cloud: Science, v. 221, p. 195-197. Roe, G.FL, 2005, Orographic precipitation: Annual Review of Earth and Planetary Sciences, v. 33, p. 645-671. Rose, W.I., 1977, Scavenging of Volcanic Aerosol by Ash - Atmospheric and Volcanologic Implications: Geology, v. 5, p. 621-624. Rose, W.I., Anderson, A.T., Jr., Woodruff, L.G., and Bonis, S.B., 1978, The October 1974 Basaltic Tephra from Fuego Volcano: Description and Flistory of the Magma Body: Journal of Volcanology and Geothermal Research, v. 4, p. 3-53. Rozanski, K., Johnsen, S.J., Schotterer, U., and Thompson, L.G., 1997, Reconstruction of past climates from stable isotope records of palaeo-precipitation preserved in continental archives: Hydrological Sci. Journal, v. 42, p. 725-745. Sancho, C., Pena, J.L., Mikkan, R., Osacar, C., and Quinif, Y., 2004, Morphological and speleothemic development in Brujas Cave (Southern Andean range, Argentine): palaeoenvironmental significance: Geomorphology, v. 57, p. 367-384. Schaefer, K., Denning, A.S., Suits, N., Kaduk, J., and others, 2002, Effect of climate on interannual variability of terrestrial C 02 fluxes: Global Biogeochem. Cycles, v. 16, p. 1102. Schneider, D.J., Rose, W.I., Bluth, G.J.S., Krueger, A.J., and Sprod, I., 1995, Observations of the separation of volcanic ash and sulfur dioxide in the April 4-6, 1982 El Chichon volcanic cloud using AVHRR and TOMS, Eos Transactions AGU, Volume 76, American Geophysical Union, p. F659. Schneider, D.J., Rose, W.I., Coke, L.R., Bluth, G.J.S., Sprod, I.E., and Kmeger, A.J., 1999, Early evolution of a stratospheric volcanic eruption cloud as observed with TOMS and AVHRR: Journal of Geophysical Research-Atmospheres, v. 104, p. 4037-4050. 105 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Schwehr, K.A., 1998, Oxygen isotopic variations of soda straw cave deposits from the Yucatan Peninsula: A test of their use as a paleoprecipitation tool [M.S. thesis]: Houston, University of Houston. Self, S., Rampino, M.R., Zhao, J., and Katz, M.G., 1997, Volcanic aerosol perturbations and strong El-Nino events - No general correlation: Geophysical Research Letters, v. 24, p. 1247- 1250. Shackleton, N.J., and Opdyke, N.D., 1973, Oxygen isotope and paleomagnetic stratigraphy of an equatorial Pacific core V28-238: Oxygen isotope temperatures and ice volumes on a 105 and 106 year time scale: Quaternary Research, v. 3, p. 39-55. Sigurdsson, H., Carey, S.N., and Espindola, J.M., 1984, The 1982 Eruptions of El- Chichon Volcano, Mexico - Stratigraphy of Pyroclastic Deposits: Journal of Volcanology and Geothermal Research, v. 23, p. 11-37. Sondag, F., van Ruymbeke, M., Soubies, F., Santos, R., Somerhausen, A., Seidel, A., and Boggiani, P., 2003, Monitoring present day climatic conditions in tropical caves using an Environmental Data Acquisition System (EDAS): Journal of Hydrology, v. 273, p. 103-118. Spero, H.J., Bijma, J., Lea, D.W., and Bemis, B.E., 1997, Effect of seawater carbonate concentration on foraminiferal carbon and oxygen isotopes: Nature, v. 390, p. 497- 500. Spotl, C., Fairchild, I.J., and Tooth, A.F., 2005, Cave air control on dripwater geochemistry, Obir Caves (Austria): Implications for speleothem deposition in dynamically ventilated caves: Geochimica et Cosmochimica Acta, v. 69, p. 2451- 2468. Stendel, M., Mogensen, I.A., and Christensen, J.H., 2006, Influence of various forcings on global climate in historical times using a coupled atmosphere-ocean general circulation model: Climate Dynamics, v. 26, p. 1-15. Sternberg, L.D.S., Moreira, M.Z., Martinelli, L.A., Victoria, R.L., Barbosa, E.M., Bonates, L.C.M., and Nepstad, D.C., 1997, Carbon dioxide recycling in two 106 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Amazonian tropical forests: Agricultural and Forest Meteorology, v. 88, p. 259- 268. Stott, P.A., Tett, S.F.B., Jones, G.S., Allen, M.R., Mitchell, J.F.B., and Jenkins, G.J., 2000, External Control of 20th Century Temperature by Natural and Anthropogenic Forcings: Science, v. 290, p. 2133 - 2137. Swart, P.K., Bums, S.J., and Leder, J.J., 1991, Fractionation of the stable isotopes of oxygen and carbon in carbon dioxide during the reaction of calcite with phosphoric acid as a function of temperature and technique: Chemical Geology, v. 86, p. 89- 96. Taylor, A., Allen, J., and Clark, P., 2002, Extraction of a weak climatic signal by an ecosystem: Nature, v. 416, p. 629-632. Thompson, L.G., Mosleythompson, E., Dansgaard, W., and Grootes, P.M., 1986, The Little Ice-Age as Recorded in the Stratigraphy of the Tropical Quelccaya Ice Cap: Science, v. 234, p. 361-364. Tieszen, L.L., Boutton, T.W., Ottichilo, W.K., Nelson, D.E., and others, 1989, An assessment of long-term food habits of Tsavo elephants based on stable carbon and nitrogen isotope ratios of bone collagen: African J. of Ecology, v. 27, p. 219-226. Tilling, R.I., Rubin, M., Sigurdsson, H., Carey, S., Duffield, W.A., and Rose, W.I., 1984, Holocene Eruptive Activity of El Chichon Volcano, Chiapas, Mexico: Science, v. 224, p. 747-749. Toon, O.B., and Pollack, J.B., 1980, Atmospheric Aerosols and Climate: American Scientist, v. 68, p. 268-278. Tooth, A.F., and Fairchild, I.J., 2003, Soil and karst aquifer hydrological controls on the geochemical evolution of speleothem-forming drip waters, Crag Cave, southwest Ireland: Journal of Hydrology, v. 273, p. 51-68. Tourre, Y.M., and White, W.B., 1995, ENSO Signals In Global Upper-Ocean Temperature Source: Journal of Physical Oceanography, v. 25, p. 1317 -1332. 107 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. —, 2005, Evolution of the ENSO signal over the tropical Pacific-Atlantic domain: Geophysical Research Letters, v. 32, p. L07605 doi:10.1029/2004GL022128. Treble, P., Shelley, J.M.G., and Chappell, J., 2003, Comparison of high resolution sub annual records of trace elements in a modem (1911-1992) speleothem with instrumental climate data from southwest Australia: Earth and Planetary Science Letters, v. 216, p. 141-153. Treble, P.C., Chappell, J., Gagan, M.K., McKeegan, K.D., and Harrison, T.M., 2005, In situ measurement of seasonal delta 0-18 variations and analysis of isotopic trends in a modem speleothem from southwest Australia: Earth and Planetary Science Letters, v. 233, p. 17-32. Treble, P.C., Harrisoin, T.M., Shelley, J.M.G., McKeegan, K., Grove, M., and McCulloch, M.T., 2002, High resolution trace element and oxygen isotope analyses of a modern speleothem: Geochimica et Cosmochimica Acta, v. 66, p. A784-A784. Trenberth, K.E., 1984, Signal versus noise in the Southern Oscillation: Monthly Weather Review, v. 112, p. 326-332. Troup, A.J., 1965, The Southern Oscillation: Quart. J. Roy. Meteor. Soc., v. 91, p. 490- 506. Varekamp, J.C., Luhr, J.F., and Prestegaard, K.L., 1984, The 1982 Eruptions of El- Chichon Volcano (Chiapas, Mexico) - Character of the Eruptions, Ash-Fall Deposits, and Gas-Phase: Journal of Volcanology and Geothermal Research, v. 23, p. 39-68. Vreugdenhil, D., Meerman, J., Meyrat, A., Gomez, L.D., and Graham, D.J., 2002, Map of the Ecosystems of Central America, Final Report: Washington, D.C., World Bank. Wake, C.P., Mayewski, P.A., Xie, Z., Wang, P., and Li, Z., 1993, Regional variation of monsoon and desert dust signals recorded in Asian glaciers: Geophysical Research Letters, v. 20, p. 1411-1414. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Wake, C.P., Yalcin, K., and Gundestrup, N.S., 2002, The climate signal recorded in the oxygen-isotope, accumulation and major-ion time series from the Eclipse ice core, Yukon Territory, Canada: Annals of Glaciology, v. 35, p. 416-422. Walker, R., Gessel, S., and Held, E., 1997, The ecosystem study on Rongelap Atoll: Health Physics, v. 73, p. 223-233. Walsh, K., 2004, Tropical cyclones and climate change: unresolved issues: Climate Research, v. 27, p. 77-83. Walsh, K.J.E., Nguyen, K.C., and McGregor, J.L., 2004, Fine-resolution regional climate model simulations of the impact of climate change on tropical cyclones near Australia: Climate Dynamics, v. 22, p. 47-56. Wang, Y.J., Cheng, H., Edwards, R.L., An, Z.S., Wu, J.Y., Shen, C.-C., and Dorale, J.A., 2001, A High-Resolution Absolute-Dated Late Pleistocene Monsoon Record from Hulu Cave, China: Science, v. 294, p. 2345. Wardle, D.A., 2002, Communities and Ecosystems: Linking the Aboveground and Belowground Components: Princeton, New Jersey, Princeton University Press, 400 p. Weedon, G.P., 2003, Time-series analysis and cyclostratigraphy: Examining stratigraphic records of environmental cycles, Cambridge University Press, 259 p. Yadava, M.G., and Ramesh, R., 2005, Monsoon reconstruction from radiocarbon dated tropical Indian speleothems: Holocene, v. 15, p. 48-59. Yalcin, K., Wake, C.P., and Germani, M.S., 2003, A 100-year record of North Pacific volcanism in an ice core from Eclipse Icefield, Yukon Territory, Canada: Journal of Geophysical Research-Atmospheres, v. 108. Zhang, P.Z., Johnson, K.R., Chen, Y.M., Chen, F.H., Ingram, L., Zhang, X.L., Zhang, C.J., Wang, S.M., Pang, F.S., and Long, L.D., 2004, Modem systematics and environmental significance of stable isotopic variations in Wanxiang Cave, Wudu, Gansu, China: Chinese Science Bulletin, v. 49, p. 1649-1652. 109 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Zielinski, G.A., Dibb, J.E., Yang, Q.Z., Mayewski, P.A., Whitlow, S., Twickler, M.S., and Germani, M.S., 1997, Assessment of the record of the 1982 El Chichon eruption as preserved in Greenland snow: Journal of Geophysical Research- Atmospheres, v. 102, p. 30031-30045. Zielinski, G.A., Mayewski, P.A., Meeker, L.D., Whitlow, S., and Twickler, M.S., 1996, A 110,000-yr record of explosive volcanism from the GISP2 (Greenland) ice core: Quaternary Research, v. 45, p. 109-118. Zielinski, G.A., Mayewski, P.A., Meeker, L.D., Whitlow, S., Twickler, M.S., Morrison, M., Meese, D.A., Gow, A.J., and Alley, R.B., 1994, Record of volcanism since 7000-BC from the GISP2 Greenland ice core and implications for the volcano- climate system: Science, v. 264, p. 948-952. 110 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. APPENDICES Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. APPENDIX A CLIMATE, ECOLOGY, AND GEOLOGIC SETTING Climate in Belize is tropical to subtropical. The average annual temperature in Central Belize is -25 °C with about 1600 mm of annual rainfall (Frutos, 2006). Central Belize has a seasonal water deficit during the dry season (early March through late May), followed by summer monsoon rainfall (Fig. A l). The wet season extends from June through January, transitioning to cold front precipitation by November. Precipitation is moderate in August during the so-called “little dry” period. Tropical cyclones have made landfall in Belize from June through November, although September is the most active month (Fig. A2) (Frutos, 2006). Tropical cyclone events are bracketed by other frontal and convective precipitation. Tropical cyclones bring heavy rainfall and flooding to Belize with a recurrence interval o f-2.5 years, contributing approximately 2.5% of annual precipitation (calculated for 1978 - 2001). Climate data is courtesy of the Belize Meteorological Service’s Central Farm meteorological station, located less than 15 km from the field site. Located between true rainforest of Southern Belize and the Yucatan tropical scrub to the north, the dominant forest type at the site is mature subtropical moist semi-evergreen broadleaf forest (Vreugdenhil et al., 2002). The cave is located in the Tapir Mountain Nature Reserve, which was donated to the Belizean government in 1975 and formally designated as a preserve in 1986. Common tree species include Swietenia macrophilla (mahogany), Manilkara zapota (chicle), Brosimum alicastrum (Ramon breadnut), Orbigyna cohune (cohune palm), Cryosophila stauracantha (“Give-and-Take Palm ”), Agonandra .s/?.(guinweo vine), Ficus crassiuscula (strangling fig), Cedrela odorata (Spanish “cedar”). 112 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. without prohibited reproduction Further owner. copyright the of permission with Reproduced Monthly Precipitation and - 9 ^ E CO 12 - 15 - 18 ae blne peiiain vprto)s eaie from negative evaporation)is - (precipitation Meteorological balance Farm Water (Central Climatology Belize l. A Figure February through May. through February vprto (ahd ie, n ma tmeaue bl rd line). red (bold temperature mean and line), (dashed evaporation Station, 1973-2005), including monthly precipitation (thin blue line), blue (thin precipitation monthly including 1973-2005), Station, iue 2 Saoaiy f eie urcns 18 - 20) After 2005). - (1883 season. Hurricanes wet the of middle the in occur Belize cyclones Tropical of 2006. Frutos, Seasonality A2. Figure 3 - 0 a Fb a Ar a Jn u Ag e Ot v Dec ov N Oct Sep Aug Jul Jun May Apr Mar Feb Jan en ep (C) Temp Mean 113 Limestones and dolomites of Late Cretaceous age and Paleozoic limestones make up the bulk of local bedrock (Miller, 1996). The cave site used in this study, Actun Tunichil Muknal (ATM), is developed in a massive pink limestone breccia. This bedrock unit correlates with a locally described rock unit in Central Belize known as the Albion formation or “Teakettle Diamictite,” which has been identified as the K-T boundary in central Belize (King and Jr., 1996; Pope and Ocampo, 2000). This breccia unit, common in the Boundary Fault Karst region of Belize, has low primary porosity with major speleogenesis occurring in conjunction with fractures (Miller, 1996). ATM is an active, multi-level phreatic cave with an underground river flowing through low passages (Miller, 1990). The river enters ATM through a sink in a different drainage basin, flowing underground for 5 km and ultimately emptying into Roaring Creek about 100 m from the resurgence. The keyhole-shaped lower cave entrance shows clear geomorphic evidence of the downcutting origin of multiple upper-level passages. Elevated cave passages in ATM are heavily decorated with speleothems, and were used for ceremonial purposes by the Maya civilization c. 400-900 CE (Miller, 1990). Some speleothems are actively forming, while others appear desiccated. An upper level passage, about 500 m from the cave entrance, where speleothems were collected for this study has a stable temperature of 25°C and relative humidity of >90%. Bedrock overburden above this portion of the cave is at least 10 m in thickness, but may extend a few tens of meters. Soil cores taken on the hill above the cave yielded thin, clay-rich soils (5-30 cm depth) with a minimal organic horizon and a thin covering of leaf litter. Carbonate bedrock outcrops also attest to the thin upland soils over ATM. 114 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. APPENDIX B MICRO-MILLING AND STABLE ISOTOPE ANALYSIS Micro-samples were milled continuously from the surface of ATM7 the polished stalagmite slab at 20 pm intervals (Pig. A2) with a CM-1 micro-milling system custom-built by Scott J. Carpenter. The system combines a fixed drill (Brasseler UP 200 controller and a UG 12 handpiece fitted with a 0.3 mm tungsten carbide dental bur), computer-controlled stage, and observation under a Nikon SMZU microscope. A rotational stage allowed us to rotate the sample relative to the drilling axis as milling progressed in order to maintain alignment with the growth axis (Fig. A2). The width of the continuously-milled micromilling track is shown as a blue outline in Fig. A2. The depth of all micro- samples from the surface of the slab was approximately 0.5 mm. The first 37 samples (two annual layers in the uppermost 1.48 mm) were milled at 40 micron intervals, and 5-6 mm in width. It was clear at this point that a smaller drilling track would generate sufficient material for stable isotope analysis. The remaining micro-samples were milled at 20 micron intervals, where the track was approximately 2.5 mm across. Individual calcite powder samples were transferred using a stainless steel scalpel from the milled surface to individual stainless steel-sample containers. Static electricity enables the calcite powder to cling to the scalpel so that samples can be recovered reliably. Between samples, the drill bit and milling surface were cleaned with a stream of compressed air. Each 20 micron micro-sample represents one to several weeks of deposition. Some commercially available micro-milling equipment can be used to the same effect. 115 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Stable Isotope Analysis was performed at the Paul H. Nelson Stable Isotope Laboratory at University of Iowa. Calcite powders were roasted in vacuo at 380°C for one hour to remove volatile contaminants and then desiccated. We analyzed -1300 samples using a Finnigan-MAT 252 IRMS with a Kiel III automated carbonate device. Powdered calcite samples (approximately 0.02 to 0.05 mg of CaC03 for each sample) were reacted with 2 drops of anhydrous phosphoric acid at 75°C. Daily analysis ofNIST powdered carbonate standards (NBS-18, 19, 20) and several in-house standards were conducted. Analytical precision on these standards was better than ±0.1 %o for both S180 and S13C values. All results are reported in per mil (%o) relative to V-PDB. 116 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. APPENDIX C AGE M ODEL We have updated the age model initially published for this stalagmite record to correct counting errors in a previously published dataset (Frappier et al, 2002). We show that these counting errors are related to the direct hydrological effects of heavy tropical cyclone precipitation. It is important to note that while the low 5180 excursions were important in identifying counting errors, the stable isotope excursions themselves are not used as a dating tool. We avoid a circular argument for dating the low S180 excursions by using independent lines of evidence to test the updated age model, outlined below. Thus verified, the updated age model can be used to determine the year in which low 5lsO excursions were deposited and compare those dates with the historical record of tropical cyclone activity in the area. In this section, we describe the initial age model development and justify the changes in the present work. Major groundwater flushing events can result in double bands or couplets of speleothem calcite, representing a single annual period (e.g. Asmerom and Polyak, 2004) and references therein; (Baker et al., 1993; Kaufmann and Dreybrodt, 2004; Proctor et al., 2000). The oscillation between seasons with positive and negative water balance triggers seasonal shifts in dripwater volume and isotopic composition, color, luminescence, trace element concentrations, and/or crystal fabric. Central Belize experiences an annual water deficit from March through May followed by onset of the summer monsoon (Fig. A l), conditions amenable to generating annual variations in dripwater chemistry and calcite morphology. The initial and updated age models are based on our interpretation of visible band pairs as annual deposits. After describing age model development and changes below, we show that both age models are supported by 137Cs dating, 117 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. and that the updated age model is further supported by independent stratigraphic, isotopic, and trace element evidence. Radiometric Dating The 137Cs activity depth profile was used to test whether layer counting was consistent with an annual pattern of calcite deposition (Fig. A3). Although no classic radiometric decay curve is evident in the data, the onset of 137Cs activity is clearly demarcated. It is important to note that a classical decay curve is not necessarily predicted in this depositional setting. Cesium substitutes strongly for the macronutrient Potassium; as a result, in tropical ecosystems 137Cs is not simply transmitted to the cave but is tightly cycled within the overlying ecosystem and soil (Dorr and Miinnich, 1989; Ritchie and McHenry, 1990; Robison et al., 1997; Walker et al., 1997). However, the depth of 137Cs activity onset provides sufficient age control to test the presence or absence of annual banding in this stalagmite. Using this approach, the error in 137Cs dating is primarily related to the spatial resolution of l37Cs samples and the assumptions used to derive ages from the depth of 137Cs activity onset. The onset of 137Cs activity at a depth between 48 mm and 39 mm marks the start of global atmospheric fallout from thermonuclear weapons testing (Robinson et al., 2002) (Fig. A3). We bracketed growth rates for the upper portion of ATM7 with three different assumptions: 1. The highest stratigraphic level at which zero 137Cs activity was measured (-48 mm) reflects pre-atmospheric weapons testing conditions that prevailed in 1953 or earlier. This assumption yields a maximum growth rate of 1.02 mm-yr'1. 2. The depth of 137Cs activity onset (-39 mm) may represent the onset of atmospheric fallout as early as 1954. This assumption yields a minimum growth rate of 0.85 mm-yr"1. 118 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 0 0 -I 80 'a- CO E 40 (0 O 20 O 0 1 2 3 4 5 6 Depth (cm) Figure A3. ATM7 137Cs activity profile. Error bars are 1 standard deviation. Note that the activity of 137Cs exceeds zero above 4.8 mm. 3. An early maximum of 137Cs activity (-39 mm) may also represent the peak of thermonuclear fallout in 1963. This assumption yields a growth rate of 1.05 mm-yr"1. The average of these three radiometrically-constrained growth rates was 1.03 mm-yr"1 ± 0.08. Laver Counting and Initial Age Model Paired visible couplets of clear and opaque calcite laminae in ATM7 are thought to correspond to the dry and wet seasons during the March - February hydrological year (Frappier et al., 2002a). The average radiometrically constrained growth rate estimate for the upper 60 mm (1.03 mm-yr"1 ± 0.08) was indistinguishable from the layer-counting growth rate calculated for the 119 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. same section (1.04 mm-yr '). However, we observed that layer thickness was greater in the upper -tw o dozen couplets, the portion of ATM7 subjected to high-resolution stable isotope analysis (Fig. 2). We surmised that the growth rate was higher than average in the uppermost portion of ATM7. The layer-counting-based growth rate in the micro-sampled portion was slightly higher, 1.15 mm-yr1. The radiometric age model and the more detailed annual layer-counting-based age model are consistent with one another. Having established consistency between ATM7 stratigraphic patterns and !37Cs dating, we concluded that opaque-clear couplets represent annual depositional cycles. We assigned a linear growth function to each annual couplet to generate a refined age model with a March date for the base of each hydrological year (Frappier et al. 2002). In the absence of counting errors, the simplifying assumption of linear growth within each hydrological year results in variable dating error for individual stable isotope samples on the order of a few weeks to a few months. Age Model Changes We identified counting errors in the original age model through recognition of stratigraphic associations between a few visible layers and large tropical cyclone-related 5lsO value excursions. In addition to annual banding generated by strong seasonal water balance variations, smaller rainfall events could likewise perturb drip rates and calcite morphology. In fact, we observed many minor fluctuations in opacity within the clear portions of annual band couplets. Annual layer counting was not affected by most lesser rainfall events, which apparently do not perturb speleothem growth extensively enough to cause confusion with annual layers. However, infiltration from particularly intense rain events (e.g. Hurricane Keith produced over 26 cm of rain) would rapidly raise the hydraulic head of the drip water source conduit. The resulting sudden change in drip rate could temporarily perturb speleothem deposition to a greater degree 120 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. than more frequent but smaller storms. It is important to note that in the case of a major precipitation event, drip rate in the cave would be affected immediately upon stormwater infiltration, but water isotopic composition at the speleothem surface would not reflect the storm event until after enough time had passed to allow the isotopically-distinct stormwater to infiltrate from the surface to the cave site.Thus, we expected any physical changes in crystal opacity resulting from tropical cyclone events to occur stratigraphically below any measured stable isotopic excursions. In ATM7, one visible couplet that we originally classified as annual was deposited a few months before stalagmite collection, about the time that major Hurricane Keith struck Belize. We originally classified three additional visible couplets as annual deposits that we later found to be located stratigraphically below associated low 5lsO value excursions. On further examination, compared to other band pairs that we classified as annual, these four low 5180 value excursion- related layers, or “storm couplets” were more akin to the minor sub-annual opacity variations than to other annual band pairs: storm couplets were typically much thinner, less distinct, and morphologically rougher and more angular. As a result of our present analysis, we now recognize these storm couplets as sub-annual events generated by tropical cyclone rainfall rather than complete annual periods. Although weak fluctuations in opacity were also associated with other low S180 value excursions, apparently, most tropical cyclone precipitation events and other major storms did not affect speleothem stratigraphy enough to affect the initial layer-counting process. Importantly, the high-resolution stable isotope record enabled us to distinguish clearly between the few cyclogenic storm couplets and annual couplets within the visible banding pattern. This discovery enabled us to refine the annual layer counting in the age model used in our previous analysis of this stalagmite (Frappier et al., 2002). After accounting for the storm couplets, we re-assigned a linear growth function to each annual couplet as before to generate an updated age model with a March date for the base of each hydrological year (Fig 2-2). The storm couplet at the top of the stalagmite (presumably generated 121 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. by stormwater from Hurricane Keith) shifted the entire record forward by several months, and three older layers associated with low 5lsO value excursions (1990, 1980, and 1978) further shifted the lower part of the record. Tests of the Updated Age Model When plotted using the updated age model, ATM7 stable isotope, stratigraphic, and trace element variations are very consistent with regional climatology and local meteorological observations. Three lines of evidence validate the updated age model: 1. The ENSO teleconnection correlation previously identified in ATM7 (Frappier et al. 2002) remains robust and stable in the updated age model (Fig. 3). Previously, we showed a total correlation offset (or proxy record lag) between the ATM7 stable isotope record and the SOI of approximately 1.5 years (Frappier et al. 2002). This lag has two components, comprised of the teleconnection time (time between changes in the core ENSO region of the equatorial Pacific and subsequent changes in Belize), plus an infiltration or recharge component (time for meteoric water to percolate through the epikarst to the stalagmite surface). A total lag of about 1.5 years is also inferred from the updated age model. We now infer that the teleconnection lag component is approximately 1 year, based on studies showing that Caribbean weather responds to ENSO forcing after a lag of approximately one year (Tourre and White, 1995, 2005). The remainder of the correlation offset between the SOI and ATM7 stable isotope record (~0.5 years) also enables us to estimate the percolation portion of the total time lag at approximately 6 months. While we have not yet been able to conduct a long-term tracer test to quantify the exact timescale of recharge, additional information enables us to bracket the infiltration time to 3-6 months. The updated age model gives dates for each low 5lsO value 122 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. excursion in the ATM7 record that match years of local tropical cyclone activity (Figure 2-3). Another constraint comes from the lack of observation of any excursion from Hurricane Keith, which struck Belize just three months before the stalagmite was collected in January 2001. Given the excellent fidelity of the proxy record to earlier tropical cyclone events, we surmise that cyclogenic water from Hurricane Keith was still percolating down through the epikarst at the time of collection, and the infiltration time must be at least 3 months. An infiltration time of 3-6 months is reasonable for this site, and is consistent with the duration of the stratigraphic offset between storm couplets and associated low 8lsO value excursions. The consistency between the ATM7 record, 1-year regional teleconnection lag, and infiltration time estimate, combined with the stability and clarity of the ENSO correlation together provide strong evidence that the updated age model is correct. 2. The extended low 5180 value interval near the base of the ATM7 record (located between K and J in Fig. 3) now dates to 1979, a year when local precipitation exceeded 2856 mm, more than three standard deviations (lcr= 323 mm-yr'1) above the climatological average (1618 mm-yr"1). This extended period of low 5lsO values is thus an expression of the “amount effect” and is unrelated to tropical cyclone precipitation events (Rozanski et al., 1997). Visible stratigraphy also reflects the extreme wetness of 1979. A distinct, rusty-colored layer apparent in the polished cross-section (Fig. 2) and trace element event (described below) is embedded within this extended period of low S180 values. This isotopic, trace element, and stratigraphic horizon constitutes an event horizon that is best explained by the combined effects of the extremely wet weather conditions that prevailed in central Belize during 1979. 3. Two different, major trace element perturbations in this stalagmite are dated to 1979 and 1982 using the updated age model (Chapter 2, see also Figure A-4 and A-5). The 1979 trace element event corresponds to the wet year and associated stratigraphic and isotopic changes in the ATM7 123 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. record described above. The 1982 trace element event represents an even greater disruption of the site’s biogeochemistry, and is unique in the record (2001 - 1978). The 1982 trace element event is significant because it occurred in a year when clouds of tephra from the eruption of the El Chichon volcano in nearby Chiapas, Mexico were observed to cover the study area (Chapter 2). This 1982 trace element event thus constitutes a marker horizon of known age that confirms the validity of the ATM7 age model update presented here. Together, these three independent and consistent lines of evidence provide a rigorous test of the updated age model. Isotopic, stratigraphic, and trace element evidence consistently support the updated age model, enabling us to avoid the circular use of the low 8lsO value excursions to date the stalagmite itself. The excellent temporal match observed between the history of storm events in this area and low 8lsO value excursions in ATM7 is thus an application of the updated age model, and not a support for that age model. However, this is discussed in further detail above. The remaining age model error is related to the assumption of linear deposition within each annual couplet, indicating dating uncertainty for individual stable isotope samples of a few weeks to a few months. 124 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. APPENDIX D PALEO-HURRICANE INTENSITY PROXY ANALYSIS To investigate the relations between signal amplitude and storm characteristics, we performed a standard multiple linear regression for the eleven storm signals we identified as cyclogenic. The four independent variables were storm maximum intensity at or prior to landfall (integer Saffir-Simpson intensity categories, Table 1) (I), minimum distance between storm track and cave site (km) ( D), mean storm precipitation recorded at three nearby meteorological stations (mm) (P), and sampling frequency (number of micro-samples per year) (5). Sampling frequency could exert a strong control on the ability of this technique to resolve the isotopic signature from very brief individual storm infiltration events. Given a constant sampling interval of 20 (im, inter-annual and sub-annual variations in stalagmite growth rate would control extent of contamination in storm micro-samples by adjacent, isotopically “normal” background calcite. Variations in the relative thickness of micro-samples compared to annual layers may modulate the resolving power of the stable isotope record for detecting storm events and/or intensity (Chapter 2). Although seasonal growth rate variations may be large, the tropical cyclone events in this dataset all occur within the rainy season (Fig. A2). Sampling rate is thus likely to be relatively stable for different storm events occurring within the same season. For multiple tropical cyclones in the same year, the measured amplitude of excursions is more likely to be controlled by differences in precipitation 5180 values. In contrast, growth rate differences between years has a relatively large effect on the amount of time represented by each sample, and thus could substantially affect the measured amplitude of different cyclogenic excursions. The 125 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. parameter sampling frequency varies with annual growth rate, but remains constant for multiple storm strikes during a single year, in keeping with our understanding of the relations among sampling, growth rate, and measured excursion amplitude. For example, within the multiple strike years 1995 and 1996, sampling frequency was stable and more intense storms were associated with larger excursions. The overall regression model results (Adj. R2= 0.653, p=0.031) are explained in Results, and tabulated below (Table Al). Distance to storm track was not a significant contributor to the overall model, and was not correlated with any other independent variables. Not surprisingly, local rainfall was highly correlated with storm maximum intensity. The correlation between storm maximum intensity and local precipitation means that reconstructions of past storm intensity also closely reflect the amount of local precipitation generated by those storm events. Interestingly, local storm precipitation is not significantly correlated with 5180 signal size, suggesting that given sufficient sampling resolution §180 signal size is controlled substantially by the maximum intensity reached by the storm while producing rainfall at the cave site, and is not a direct expression of the local “amount effect”. Table A l. Results of a multiple linear regression to investigate storm and speleothem factors related to the signal size of measured 5lsO value excursions. Coefficients correlations toleranc Factor unstandard Standard zero- semi- sig. e ized ized order partial Constant 0.364 ---- 0.168 Maximum Intensity 0.241 1.275 0.582 0.595 0.218 0.019 Resolution 0.007 0.662 0.381 0.618 0.872 0.016 (samples/ yr) Distance to Storm 0.001 0.169 0.247 0.158 0.877 0.428 Track (km) Local Precipitation -0.003 -0.639 0.383 -0.303 0.225 0.155 (mm) 126 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. APPENDIX E TRACE ELEMENT DATA Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure A-4. ATM7 LA-ICPMS species concentrations (counts per second normalized to Ca and NIST 612). 10-1 B 11 1 - 0.1 0.01 1980 1985 1990 1995 2000 1000 -I Na 23 100 - 10 100 Mg 26 10 10000 1 Al 27 1000 100 t 10 1 1980 1985 1990 1995 2000 198 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Si 29 1980 1985 1990 1995 2000 1000 P 31 100 10 1 0.1 100 Ti 47 10 - 1 - 0.1 - 0.01 - 0.001 Ti 49 0.01 0.001 1980 1985 1990 1995 2000 190 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 100 1 Mn 55 10 • U ■ • 0.1 1980 1985 1990 1995 2000 60 Fe 57 50 40 30 20 10 0 100 ! Rb 85 10 1 -\ 0.1 0.01 -\ 0.001 100 i Sr 88 10 - 1980 1985 1990 1995 2000 nn Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Y 89 1 H 0.1 •I .•* 0.01 - 0.001 1980 1985 1990 1995 2000 1 0 -1 Zr 90 0.1 - 0.01 100 n Ce 140 10 - 1 - 0.1 100 Ba 138 10 1 0.1 0.01 0.001 0.0001 1980 1985 1990 1995 2000 131 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Th 232 0.001 0.0004- 1980 1985 1990 1995 2000 U 238 0.001 0.0001 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure A-5. ATM7 LA-ICPMS data comparison with first three EOF modes (y-axis). EOF modes are plotted as loading factors; P31, Na23, and Sr88 are plotted as counts per second normalized to Ca. EOF 1 Time Series 0 .3 0.2 0.1 B 0 lit— ...... 1 .. v .. »i.'i >HL - 0.1 H O F 2 T i m e S e r i e s 0.1 0 .0 5 - 0 .0 5 - 0.1 EOF 3 Time Series 0.2 0.1 - 0.1 - 0.2 1 9 8 0 1 9 8 5 1 9 9 0 1 9 9 5 2000 10 0 0 P 31 1 0 0 1* i 1 1 0 - I 1 0.1 10 0 0 Na 23 Sr 88 133 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.