This item is only available as the following downloads:
CLIMATIC VARIATION IN THE CIRCUM-CARIBBEAN
DURING THE HOLOCENE
JASON H. CURTIS
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN
PARTIAL FULFILLMENT OF THE REQUIREMENTS
FOR THE DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
I wish to thank several people for their help in this project without whom I would
have been at a loss. First, I am extremely grateful to Dr. David Hodell for his constant
support and encouragement and his helpful prodding of me in the right direction. I am also
extremely grateful to Dr. Mark Brenner for all the support, encouragement and criticism
that he has given me over the past eight years. I would like to thank Dr. Claire Schelske
for serving on my committee, for his financial support, and for access to his gamma
counters for 210pb dating. I thank Dr. Jon Martin and Dr. Doug Jones for serving on my
committee and for all their support. I thank Dr. Rick Forester, Dr. Jonathan Holmes, and
Dr. Fred Thompson for faunal identification. I am very thankful to my parents for all the
support they have given me throughout my life and especially in academic endeavors. I am
especially thankful to Kathy for everything.
I am indebted to numerous friends and colleagues who facilitated fieldwork in
Mexico, Guatemala and Venezuela. I thank Drs. Alfredo Barrera Rubio and Tomas
Gallareta of the Instituto Nacional de Antropologia e Historia (INAH), in Merida, Yucatan,
Mexico. Work in Guatemala was made possible due to the efforts of Lic. Erick Manuel
Ponciano and Sr. Carmen Obbet Galvez Mis of the Instituto de Antropologia e Historia
(IDAEH). I thank Profesora Haymara Alvarez (Universidad Simon Bolivar) and Ingeniera
Martha Perdomo (Ministerio del Ambiente y de los Recursos Naturales Renovables Caso
Estudios Venezolanos sobre Cambios Climaticos) for their efforts on my behalf in
Venezuela. The following individuals provided good company and assistance in the field:
Roger Medina Gonzalez, Mary Hart, Mark Mulligan, Christina Ullmann, Luis Monserrat,
Dave Hodell, Mark Brenner, Joy Curtis, Bruce Curtis, and Kathy Venz Curtis. This work
was funded by NOAA grant NA36GP0304 to Mark Brenner and David Hodell.
TABLE OF CONTENTS
ACKNOWLEDGEMENTS .................................. ii
ABSTRACT ............................................ v
INTRODUCTION ........................................ 1
Statement of Objectives ................................ 1
Isotopic Fractionation .................................. 4
Oxygen Isotopic Systematics ............................ 5
Oxygen Isotopic Fractionation in Source Waters ............... 7
Oxygen Isotopic Composition of Lakes ..................... 9
Tropical Closed-Basin Lakes ............................ 11
Calcium Carbonate in Closed-Basin Lakes ................... 11
Previous Oxygen Isotopic Studies in Circum-Caribbean Region ..... 13
VALENCIA ............................................ 16
Introduction ....................................... 16
Study A rea ........................................ 18
M ethods .......................................... 20
Chronology and Sedimentation Rates ................. 23
W ater Chemistry ............................... 26
Elemental Geochemistry .......................... 26
Stable Isotope Geochemistry ....................... 28
Discussion ........................................ 31
Latest Pleistocene (-12,600 to -10,000 14C yr BP)........ 31
Earliest Holocene (-10,000 to -8,200 14C yr BP)......... 34
Early to Middle Holocene (-8,200 to -3,000 14C yr BP) .... 37
Late Holocene (-3,000 14C yr BP to the present) ......... 40
Conclusions ....................................... 41
PETEN-ITZA ........................................... 43
Introduction ....................................... 43
Study A rea ........................................ 47
M ethods.......................................... 49
R esults........................................... 53
W ater Chemistry ............................... 53
Chronology .................................. 54
Pollen ...................................... 59
Elemental Geochemistry .......................... 59
Carbonate Fossils and Stable Isotope Geochemistry ....... 62
M agnetic Susceptibility .......................... 66
D discussion ........................................ 66
Proxies of Environmental Change ................... 66
Earliest Holocene (>9,000 14C yr BP) ................ 67
Early Holocene (-9,000-7,300 14C yr BP) ............. 68
Mid-Holocene (7,300-4,800 14C yr BP)............... 72
Late Holocene (4,800 14C yr BP to the present).......... 76
Conclusions ....................................... 80
PUNTA LAGUNA ........................................ 82
Introduction ........................................ 82
Study A rea ........................................ 83
M ethods .......................................... 85
W ater Chemistry ............................... 90
Oxygen Isotopes ............................... 91
D discussion ........................................ 93
Oxygen Isotopes .................... ... .... . ... . 93
Paleoclimatic Interpretation ........................ 95
Archaeological Implications ........................ 99
Conclusions ....................................... 104
SYNTHESIS OF HOLOCENE CLIMATE CHANGE ................ 106
Introduction ........................................ 106
Chronological Control ................................. 107
Synthesis of Circum-Caribbean and Neotropical Climate Change .... 112
Late Pleistocene Aridity .......................... 112
Timing of Lake Filling ........................... 114
Earliest Holocene (-10,500 to -8,500 14C yr BP)......... 116
Early to Middle Holocene (-8,500 to -3,000 14C yr BP)..... 118
Late Holocene (-3,000 14C yr BP to the present).......... 120
Summary of Circum-Caribbean Climate ................ 123
Long-Term Climatic Controls ...................... 123
Short-Term Climatic Controls ...................... 127
Non-Climatic Controls ........................... 129
Conclusions ....................................... 130
Future W ork ...................................... 133
REFERENCES .......................................... 134
BIOGRAPHICAL SKETCH ................................. 148
Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
CLIMATIC VARIATION IN THE CIRCUM-CARIBBEAN
DURING THE HOLOCENE
Jason H. Curtis
Chairman: David A. Hodell
Major Department: Geology
Climate variability has been reconstructed in the circum-Caribbean region on the
basis of oxygen isotopic ratios in fossil shells of ostracods and gastropods from six lakes
including Lakes Punta Laguna, Chichancanab, and Coba, Yucatan Peninsula, Mexico;
Lake Peten-Itza, Peten, Guatemala; Lake Valencia, Venezuela; and Lake Miragoane, Haiti.
By using these records, changes in evaporation to precipitation ratios for the region during
the Holocene were reconstructed. Following arid conditions during the last Ice Age,
climate in the Neotropics became wetter and lake basins filled between -10,500 and -7,600
14C years BP. Holocene oxygen isotopic records for the six lakes, interpreted as a record
of evaporation to precipitation changes, are broadly similar but regional differences do
In the majority of the lakes, the overall climatic pattern indicates that conditions
were dry but becoming wetter during the earliest Holocene (-10,500 to -8,500 14C years
BP), followed by maximum moisture availability during the early to middle Holocene
(-8,500 to -3,000 14C years BP), and a return to drier conditions during the latest
Holocene (-3,000 14C years BP to present). This pattern may be explained by
precessionally driven changes in the seasonal distribution of solar energy that controls the
intensity of the annual cycle and rainfall abundances. Differences between records include
variability in the timing and rates of initial lake filling and the occurrence of centurial to
decadal climatic events (wet and dry periods). For example, the late Holocene history of
the Yucatan Peninsula was marked by several periods of drought (centered on 585, 862,
and 1391 AD) that coincided with major cultural discontinuities in the Classic Maya
civilization. Some of the decadal- to centurial-scale differences in isotopic records are
probably the result of local differences in a lake's response to climate forcing, such as lake
volume, altitude, orography, basin morphology, and rates of filling. Abrupt climatic
changes observed in the isotopic records can not be explained by orbitally driven forcing
and must have roots in other mechanisms, such as solar variability, volcanism, ocean-
atmosphere interactions, and natural unforced variability.
Statement of Objectives
Sediment cores from closed-basin lakes in the lowland tropics provide some of the
best terrestrial paleoclimatic records because of the sensitivity of their simple hydrologic
balances to changes in moisture availability. Climatic history is recorded in closed-basin
lakes by changes in water level, oxygen isotopic composition of water, and dissolved
solute composition that are driven by changes in the evaporation/precipitation (E/P) ratio.
Changes in E/P are, in turn, driven by seasonal-to-millennial scale climatic variability.
Previous studies have shown that oxygen isotopic records of ostracod calcite and gastropod
aragonite can be used to determine the history of E/P in closed-basin lakes (e.g. Covich &
Stuiver 1974; Hodell et al. 1991, 1995; Lister et al. 1991; Curtis 1992; Curtis & Hodell
1993; Palacios-Fest et al. 1993; Holmes 1996 and references therein). The purpose of this
study was to reconstruct climatic variability in the circum-Caribbean during the Holocene
from isotopic analysis of lake sediments from Mexico, Guatemala, and Venezuela.
Knowledge of the role climate plays in the tropics is important for understanding
the global climate system (Hastenrath 1991). The global hydrologic cycle, atmospheric
circulation, and oceanic circulation are intimately related to tropical climate. Compared to
mid- and high-latitude regions, the role of the tropics in climate change is not well known.
The classical view is that tropical biology, climate, ecology, and hydrology were nearly
invariant over long periods (Broecker 1995). This view of tropical constancy also included
the tropical oceans where sea surface temperatures were thought to have varied by less
than 2C between the last Glacial maximum and the present (CLIMAP 1976; Broecker
1995). In the past decade, however, newly obtained terrestrial and marine records show
that climate in the tropics was not as static as once believed. For instance, pollen analysis,
Sr/Ca ratios of corals, ice core records, and noble gas studies all suggest significant cooling
(>5C) of the tropics during the last glaciation (Beck et al. 1992; Leyden et al. 1993;
Guilderson et al. 1994, Thompson et al. 1995).
At the beginning of the Holocene, at -10,000 years BP, increased moisture
availability filled many dry lake basins in the circum-Caribbean region. Some lakes lying
close to sea level probably filled because of combined effects of increased moisture and
rising of the fresh water aquifer in response to sea level rise. Other, higher-elevation lakes
probably filled solely as a result of increased water availability. With the inception of
lacustrine sedimentation, preservation of ostracod carapaces and gastropod shells, along
with pollen and other fossils began. Oxygen isotopic evidence from ostracod carapaces
and gastropod shells, in addition to information from the study of pollen, diatoms,
sediment magnetic character, and sediment geochemistry of lake cores from throughout the
region, allows reconstruction of past regional climate and vegetation.
This study synthesizes lacustrine oxygen isotopic records from the circum-
Caribbean region, which lies completely within the northern hemisphere tropics (Figure 1-
1). The western portion of the region includes the countries of Central America, from
Panama through Belize and extends northward to Mexico's Yucatan Peninsula. The
northern extent of the region is the islands of Cuba, Hispaniola (Haiti and the Dominican
Republic), and Puerto Rico. To the east, the region is bordered by the Leeward and
Windward Islands. The southern edge of the region is bounded by Venezuela and
Specifically, this study incorporates findings from five lowland, Neotropical lakes,
including Coba, Punta Laguna, and Chichancanab (Mexico); Peten-Itza (Guatemala); and
Valencia (Venezuela) (Figure 1-1). In addition, results will be compared with previous
findings from Lake Miragoane (Haiti) (Figure 1-1) (Hodell et al. 1991; Curtis 1992; Curtis
& Hodell 1993). These records are used to test the hypothesis that millennial scale climate
10N Pacific 4
100W 90W 80W 70W
Figure 1-1 Map of the circum-Caribbean region indicating the location of lakes used in
this study. The numbers on the map indicate the position of the following lakes: 1.
Lakes Punta Laguna and Coba. 2. Lake Chichancanab. 3. Lake Peten-Itza. 4. Lake
Valencia. 5. Lake Miragoane.
change during the Holocene in the circum-Caribbean was controlled principally by orbitally
driven changes in the intensity of the annual cycle. These records will also be used to
examine the timing of rapid climate changes that cannot be explained by direct orbital
This study is significant in four major respects. First, the records presented here
provide a proxy for the natural amplitude of climatic variability at several sites in the
circum-Caribbean, an area with few paleoclimatic records. These records will be
incorporated into the growing data base of global climatic variability and are of sufficient
resolution for comparison with other paleoclimatic records developed from lake sediments,
ice cores, and marine cores with high accumulation rates. Second, the oxygen isotopic
records from these sites will be used to study the spatial and temporal relationships of
circum-Caribbean climatic variability. The geographic placement of these circum-
Caribbean sites will allow an assessment of whether past climatic variability can be
correlated locally or regionally. The long-term, continuous nature of these records will also
allow temporal evaluation of climatic variation. Correlation of millennial and shorter-term
climatic variability will be examined. Third, and finally, these records provide a record of
natural Holocene climatic variability that can be used as a baseline for interpreting future
climatic and environmental changes that may result from anthropogenic influences.
The primary method employed in this study for paleoclimatic reconstruction is
examination of past changes in the isotopic composition of carbonate microfossils that are
preserved in the fossil record. Isotopes of the same element behave differently in chemical,
physical or biological processes as a result of kinetic and equilibrium effects. Fractionation
occurs because some thermodynamic properties of molecules (primarily differences in
vibrational energy of the atoms in that molecule but also translational, rotational, and
electronic energy differences) are a function of the masses of atoms in that molecule. The
frequency of molecular vibration is inversely proportional to the square root of the
molecular mass. Therefore, a molecule that contains the lighter isotope of an atom will
have higher vibrational frequency, and thus higher energy, than the same molecule
containing the heavier isotope. Molecules with lighter isotopes have higher energy bonds
that are more easily broken and thus are more reactive than molecules with heavier isotopes
Differences in the behavior of isotopes during chemical, physical or biological
processes result in fractionation and are known as isotope effects. Equilibrium isotope
effects occur when one isotopic species substitutes into a phase differently than the other
isotopic species as a result of differences in thermodynamic properties of the isotopes.
Kinetic (nonequilibrium) isotopic effects happen when rates of transfer of one isotopic
species from one phase to another are different than rates of transfer for other isotopic
The fractionation factor (aX) indicates the extent of isotopic fractionation between
two phases (e.g. a liquid and a gas) and can be expressed as follows (Faure 1986):
(X= RA/RB (1-1)
where RA equals the ratios of heavy to light isotopes in phase A, and RB equals the heavy
to light ratio in phase B. Fractionation factors are dependent on temperature, with greater
fractionation occurring at lower temperature. Increasing energy availability (compared with
vibrational energy) with increasing temperature causes decreasing fractionation until no
fractionation occurs above some high temperature threshold.
Oxygen Isotopic Systematics
Oxygen is an element with stable isotopes that fractionate in natural processes,
including evaporation and precipitation. The natural abundance of the three stable isotopes
of oxygen (160, 170, and 180) are 99.763%, 0.0375%, and 0.1995%, respectively. For
simplicity 170 is not discussed because its fractionation (compared with 180) is exactly half
that of 160, so that no additional information is gained from its study.
Oxygen isotopic compositions of water and calcium carbonate are reported in
standard delta notation in units of permil (%,o) relative to a standard according to the
180 = [ (180/160Osample 180/16Ostandard) / 180/16Ostandard ] X 1000 (1-2)
The standard for water is Standard Mean Ocean Water (VSMOW), whereas the standard
for carbonates is PeeDee Belemnite (VPDB). Positive values of 6180 indicate that the
180/160 ratio of the sample is greater than the 180/160 ratio of the standard while negative
values indicate the opposite.
Fractionation of oxygen isotopes in evaporating water occurs as a result of a
combination of equilibrium and kinetic effects. The first, known as the equilibrium
isotope effect, results because the lighter isotopic species of water (H2160) more readily
makes the transition from liquid water to water vapor than the heavier species of water
(H2180) (Gat 1995). The second, known as the kinetic isotope effect, results from
differences in the diffusion rate of the various isotopic species of water vapor through the
air-boundary layer (Gat 1995). The sum of these two processes is that isotopically lighter
water has a higher vapor pressure than heavier water and thus tends to evaporate more
readily. The temperature at which evaporation occurs is important because it influences the
fractionation factors of oxygen isotopes. With increasing temperature a approaches unity
(less fractionation), whereas a increases (greater fractionation) at lower temperature.
Greater fractionation at lower temperature of evaporation produces water vapor more
depleted in 180 relative to the vapor formed at higher temperature. Regardless of
temperature, evaporation always leaves residual liquid enriched in 180 while any water
vapor formed is depleted in 180 relative to the original source water.
Relative humidity also affects the fractionation factor of oxygen isotopes during
evaporation. Lower humidity results in greater fractionation of oxygen isotopes in water.
Water vapor formed from evaporation under low humidity conditions is more depleted in
180 relative to water vapor formed under higher humidity conditions (Gat & Bowser
During the formation of precipitation in clouds (condensation), the heavier isotope
is preferentially incorporated in the condensate, leaving the vapor depleted in 180 compared
to the condensate. As with evaporation, fractionation during condensation is temperature-
dependent, with greater fractionation occurring at colder temperature. Condensation at
lower temperatures results in a liquid that is more enriched in 180 relative to residual water
The amount of fractionation that occurs during the condensation of rain drops from
a cloud also is influenced by relative humidity. Lower humidity results in greater
fractionation (larger values for a); thus precipitation formed at lower humidity will be more
enriched in 180 relative to precipitation formed at higher humidity (Gat & Bowser 1991).
Oxygen Isotopic Fractionation In Source Waters
Rain that supplies lakes in coastal areas is largely derived from evaporated
seawater. This water is isotopically fractionated during evaporation from the ocean
surface, during transportation in clouds, during condensation into rain drops and while it is
falling to the ground. These fractionations are combinations of those processes discussed
As seawater evaporates, three factors influence the 8180 of water vapor that is
formed: 1) the 5180 of sea water, 2) the temperature at which evaporation occurs, and 3)
the relative humidity of air above the ocean. The influence of the 8180 of source water is
straightforward in that the 8180 of water vapor will be affected by the 5180 of sea water.
At lower temperature, greater fractionation will produce isotopically lighter water vapor.
Likewise, greater fractionation at lower humidity will produce isotopically lighter water
During transport of an air mass, the amount of water lost from the air mass
influences the 56180 of water vapor remaining in the cloud. This process, Rayleigh
Distillation, can be described by the following equation:
R / Ro = f ( -1) (1-3)
where R is the isotopic 180/160 ratio of the water vapor remaining in the air mass, Ro is the
180/160 ratio of the original air mass, f is the fraction of original water vapor remaining,
and a is the isotope fractionation factor. As f (the amount of original water vapor)
decreases, the vapor becomes progressively more depleted with respect to the heavy
isotope. The temperature of condensation affects ax in Equation 1-3. Rayleigh distillation
is more effective (i.e. remaining water vapor more depleted in 180) at lower temperatures
than higher temperatures because of increased fractionation effect during condensation.
The oxygen isotopic composition of rain is influenced by the temperature of
condensation. As previously discussed, greater fractionation occurs at lower temperature,
so rain condensing in a cooler atmosphere is more enriched in 180 compared with rain
formed in a warmer atmosphere. The second factor influencing the 5180 of raindrops is
the amount of evaporation that occurs as raindrops fall to the earth's surface. Greater
evaporation causes an enrichment in H2180 in a raindrop as H2160 preferentially
evaporates. Additionally, relative humidity and temperature influence the extent of isotopic
fractionation that occurs as a result of evaporation as raindrops fall.
Oxygen Isotopic Composition of Lakes
The isotopic composition of a lake's input waters (both precipitation and in-flow)
strongly affects the 6180 of lake water, but processes that occur once that water reaches the
lake are also important in determining lake water 8180. In order to demonstrate the
controls of lake water oxygen isotopic composition, I will present an equation that relates
changes in 6180 of lake water to (1) fluxes to and from the lake and (2) the isotopic
compositions of these fluxes. First, the equation for the water balance of a lake is
AV= I In- XOut-E (1-4)
where AV is the change in volume of the lake, In is the sum of precipitation and inflow to
the lake, Out is outflow from the lake, and E is the evaporation flux from the lake. Adding
the oxygen isotopic compositions of these fluxes to Equation 1-4 results in an equation for
the rate of change of oxygen isotopic composition of a lake:
d 818Oiake / dt = [ ( 518Oin In ) (81801ake Out) ( 8180evap E) ] /A V (1-5)
where 180lake, 6180Oin, 618Oevap are the oxygen isotopic compositions of lake water, of
the inputs to the lake (precipitation and inflow), and of evaporating moisture from the lake,
respectively. In Equation 1-5, precipitation and inflow are combined and assumed to have
the same oxygen isotopic composition (Gat 1995). It can be seen in Equation 1-5 that lake
volume influences the rate of change of the isotopic composition of lake water. Larger
volume lakes with long residence times come to equilibrium more slowly than smaller
volume lakes with shorter residence times (Lister et al. 1991).
Lakes can be categorized into two main types: open-basin and closed-basin. Open-
basin lakes lose water through evaporation, surface outflow, and ground water seepage.
Closed-basin lakes, also known as terminal lakes, lose water solely though evaporation.
Because closed-basin lakes lose all their water through evaporation Equation 1-5 can be
d 6180iake /dt = [ ( 618Oin In ) ( 518Oevap E ) ] / A V (1-6)
All terms in Equations 1-5 and 1-6 are relatively simple and can be fairly easily
measured with the exception of 818Oevap (Gat 1995). The oxygen isotopic composition of
the evaporating flux is difficult to measure but it can be approximated by the following
equation (Gat 1995):
6180evap = ( 618O0ake h 6180atm ) / ( 1 h) (1-7)
where h is the relative humidity, 618Oatm is the oxygen isotopic composition of the
atmosphere over the lake, and s is equal to the term (1 ca). The main process for
modifying the oxygen isotopic composition of closed-basin lake water is evaporation.
Thus Equation 1-7 is useful for determining exactly how evaporation affects the 8180 of
lake water. As with evaporation from the ocean's surface, fractionation of water vapor
during lake water evaporation is controlled by both atmospheric relative humidity and
temperature at which evaporation takes place. Temperature also affects the humidity term
through its effect on the saturated vapor pressure at the water's surface (Gat 1995). As the
atmosphere's relative humidity decreases, the fractionation factor increases, leading to
water vapor that is more depleted in 180 and residual water that is more enriched in 180.
Similarly, as the temperature at which evaporation occurs decreases, fractionation
increases, again leading to lighter water vapor and heavier residual lake water.
Tropical Closed-Basin Lakes
In tropical, closed-basin lakes with seasonally dry climate, the 6180 of lake water is
controlled mainly by E/P (Fontes & Gonfiantini 1967; Covich & Stuiver 1974; Gasse et al.
1990; Talbot 1990; Lister et al. 1991). Times of greater evaporation or less precipitation
(high EIP) will result in lowered lake levels and higher 8180 of water. Alternatively,
during times of low E/P ratios, lakes will have higher water levels and lower 8180 values
of water. Atmospheric humidity helps determine how much evaporation and precipitation
occur in a region. When atmospheric humidity is high, regional precipitation tends to be
greater while evaporation tends to be less (Lister et al. 1991). Higher humidity results in
lower E/P while lower humidity results in higher E/P. The importance of evaporation on
the 6180 of modem lake water can be evaluated by comparing the 6180 of lake water with
5180 of input (ground and rain waters). The extent of enrichment of 180 in the modem
lake compared to ground and rain waters can be used to evaluate the extent of evaporative
Calcium Carbonate in Closed-Basin Lakes
Fresh water ostracods and gastropods secrete their calcium carbonate shells near
oxygen isotopic equilibrium with the water in which they live (Durazzi 1977; Abell 1985).
Shell material is incorporated stratigraphically into a lake's sediment and can later be
recovered by coring. In closed-basin lakes the 6180 of shell calcium carbonate
(818Ocaco3) depends upon three factors. First, the main control of 618Ocaco3 is the 8180
of ambient lake water (Lister et al. 1991). Second, 518OcaCO3 depends on water
temperature as illustrated by the paleotemperature equations for calcite (Craig 1965):
TC = 16.9 4.2 (818Ocalcite-8180Owater) + 0.13 (8180Ocalcite-6180Owater)2
and aragonite (Grossman & Ku, 1981):
TC = 19 3.52 (818Oaragonite-818Owater) + 0.03 (618Oaragonite-18Owater)2 (1-9)
Previous work in a tropical, closed-basin lake has shown that changes in the 618OcaCO3
due to temperature variation during the past 10,000 years have been small because
Holocene temperatures have been relatively constant (Curtis & Hodell 1993). The
temperature range between seasons can, however, cause variations in 618OcaCO3. Oxygen
isotopic measurement of multiple individuals of either ostracods or gastropods from each
stratigraphic level reduces the variance generated from seasonal changes. Third,
618OcaCO3 depends upon vital effects that cause non-equilibrium precipitation of calcite by
shell-bearing organisms when they produce shells. Use of a single species of adult
ostracods or gastropods, however, reduces variation caused by vital effects because the
fractionation effect for a single species is assumed to remain constant through time. Thus,
stratigraphic changes in 818OcaCO3 can be attributed to changes in E/P.
In summary, this dissertation applies the principles of the oxygen isotopic method
discussed previously to reconstruct the paleoclimatic history of the circum-Caribbean
region during the Holocene. The application of this method is important for several
reasons. First, oxygen isotopes in shell material are not affected by human disturbance.
This contrasts with palynological reconstructions of climate that are often ambiguous as a
result of human disturbance. This is because changes in vegetation due to human impact
often resemble changes in vegetation due to climatic change. Second, the oxygen isotopic
records presented in this dissertation will increase knowledge of past climate in a region
with few existing records. Third, oxygen isotopic analysis of shell material in tropical lake
cores is one of the best methods of reconstructing past changes in E/P. Fourth, shell
carbonate is preserved in the sediment of many lakes, thus allowing this method to be used
in a wide variety of locations. Fifth and finally, in providing qualitative reconstructions of
past changes in E/P this study should contribute to the development of an isotopic-based
method of quantitative estimation of changing E/P.
Previous Oxygen Isotopic Studies in Circum-Caribbean Region
Covich and Stuiver (1974) produced the first oxygen isotopic record from the
circum-Caribbean region by analysis of shell material from Lake Chichancanab, Yucatan.
Oxygen isotopes were measured in shells of the gastropod Pyrgophorus coronatus from a
12-m sediment core. A paleoclimate record was inferred, assuming that the major control
on 6180 of lake water (and shell carbonate) was the balance between evaporation and
precipitation. The record indicated that from -22,000 until ~8,000 years BP, Lake
Chichancanab was greatly reduced in size or possibly completely desiccated. A shallow
lake formed at about -8,000 years BP. From -5,500 to -1,500 years BP, lake level was
stable and deep. At -1,500 years BP, lake level decreased to near current level where it has
remained until present. Broad sample spacing (380 yr/sample) and poor age control in this
pioneering study yielded a paleoclimatic record of low temporal resolution. Hodell et al.
(1991) continued paleoclimatological study of sediments from Lake Chichancanab. Based
on evidence for oxygen isotopes of shell material and concentration of calcite and gypsum
in the sediments they reported that following a dry period the lake filled -8,200 14C years
BP and relatively wet conditions lasted until -3,000 14C years BP when a drying trend
began. The interval between -1,300 and -1,100 14C years BP was the driest periods of
the last 8,000 years and coincides with the collapse of the Classic Maya civilization
between AD 750 and 900.
Sediment cores from Lakes Coba, San Jose Chulchaca, and Sayaucil, also on
Mexico's Yucatan peninsula, were investigated for pollen, diatoms, and stable isotopes
(Leyden et al. 1996; Whitmore et al. 1996), but carbonate microfossils in sediment cores
from San Jose Chulchaca and Sayaucil are discontinuous, yielding incomplete paleoclimatic
results based on oxygen isotopic analysis. Analysis of sediment from Lake Coba indicated
that the lake first filled -8000 14C years BP and was initially shallow and saline (Whitmore
et al. 1996). Lake Coba's level was high with freshest waters at -2600 14C years BP, after
which water level has been decreasing and salinity increasing to the present.
Analysis of oxygen isotopes and trace metal (Sr/Ca, Mg/Ca) ratios in ostracod
shells from sediments from Lake Miragoane, Haiti, indicated that for the 500 years
following lake filling at -10,500 14C years BP, climate was drier and cooler than today.
From -10,000 until -7,000 14C years BP, temperature increased and salinity decreased as
lake level rose. From -7,000 to -3,500 14C years BP, the lowest salinities of the entire
record were recorded, coinciding with the early to middle Holocene moist period. From
-3,500 to -1,500 14C years BP, drier conditions prevailed, with an exceptionally dry
period from -2,500 to -1,500 14C years BP. Wetter conditions then returned and lasted
until -1,000 14C years BP. The last millennium has been characterized by a return to drier
Leyden et al. (1993) used oxygen isotopes of bulk sediment from Lake Quexil,
Peten, Guatemala, in a multiproxy study of late Pleistocene climate. Bulk sediment was
used because Lake Quexil's sediments lacked carbonate microfossils. Isotopic studies of
bulk carbonate are difficult to interpret because sedimented carbonate can come from
several sources, including detrital inwash, bio-induced precipitation, or autochthonous
microfossils. Leyden et al. (1993) showed that the isotopic composition of bulk lacustrine
carbonate was substantially different from that of the carbonate-rich soils that surround the
lake. Lacustrine carbonates were therefore predominately authigenic and contained little, if
any, detrital CaCO3. On the basis of isotopic, sedimentological, and pollen evidence,
Leyden et al. (1993) divided the last -36,000 years BP into three stages. From -36,000 to
24,000 years BP, lake level was lower than today with high E/P conditions. During the
last Ice Age, from 24,000 to 12,500 years BP, the lake was extremely low as a result of
extreme aridity. Coincident with deglaciation at about 12,500 years BP, conditions became
wetter as E/P ratios dropped. After -10,500 years BP, Lake Quexil filled rapidly as
conditions continued to moisten.
Street-Perrott et al. (1993) used oxygen isotopes of fine bulk sediment as part of a
multiproxy study of Wallywash Great Pond, Jamaica, for the last 120,000 years.
Following a lacustrine period between 106,000 and 93,000 years BP, sedimentation in
Wallywash Great Pond became intermittent. Thereafter, generally cool and dry conditions
were inferred through the end of the Pleistocene. The lake refilled and lacustrine
sedimentation resumed with the beginning of the Holocene. Street-Perrott et al. (1993)
report three transgressive-regressive cycles during the Holocene. Unfortunately, these
cycles were not dated accurately because of the effect of hard-water-lake error on 14C
dates. Inaccurate dating resulted in broad climate reconstructions.
Lake Valencia, Venezuela, is the largest freshwater body in northern South
America, and its Holocene sediments are a rich source of paleoenvironmental information.
Micropaleontological and geochemical studies of latest Pleistocene and Holocene sediments
of Lake Valencia have included pollen (Salgado-Labouriau 1980; Leyden 1985), diatoms
(Bradbury 1979; Bradbury et al. 1981), sediment chemistry (Lewis & Weibezahn 1981;
Binford 1982), and animal remains (Binford 1982). These studies have shown that the
water level of Lake Valencia fluctuated significantly during the latest Pleistocene and
Holocene (Bradbury et al. 1981; Lewis & Weibezahn 1981; Binford 1982) probably in
response to long-term changes in moisture availability.
Lake Valencia serves as the representative southern Caribbean site in a larger project
to reconstruct circum-Caribbean region paleoclimate (Hodell et al. 1991, 1995; Curtis and
Hodell 1993; Curtis et al. 1996; Curtis et al. in press). Lake Valencia was chosen for this
study because of its location, size, and accessibility. Furthermore, previous study (Binford
1982) showed that the lake sediments preserved plentiful ostracod and gastropod shell
material. Lake Valencia is also located near the marine Cariaco Basin just offshore of
Venezuela, that has yielded high-resolution paleoclimatic records (Overpeck et al. 1989;
Peterson et al. 1991; Hughen et al. 1996a, 1996b) providing the possibility of correlating
marine and terrestrial paleoclimate records.
In this study I present the first oxygen isotopic records of carbonate microfossils
preserved in the sediments of Lake Valencia. I measured oxygen isotopes of monospecific
ostracods and gastropods and present bulk elemental chemistry of sediments in a 568-cm
core. These data are the basis for reconstructing changes in lake level and climate of
northern South America during the latest Pleistocene and Holocene.
Today, the rainy season in the Caribbean occurs during the northern hemisphere
summer half year when the Intertropical Convergence Zone (ITCZ) moves north and
displaces the North Atlantic subtropical high and weakens the easterly trade winds
(Hastenrath 1976, 1984). The dry season occurs during the northern hemisphere half year
when the ITCZ moves south. Meteorological studies of interannual variability of rainfall in
the Atlantic region sector during the twentieth century show a strong correlation between
precipitation anomalies and intensity of the annual cycle (Hastenrath, 1984). Anomalously
rainy years coincide with an enhancement of the annual cycle when the ITCZ moves far
north during the summer (rainy season) and far south during the winter (dry season). In
contrast, reduced rainfall is associated with a reduction in the annual cycle (Hastenrath
Hodell et al. (1991) suggested that long-term changes in the ratio of evaporation to
precipitation (E/P) in the circum-Caribbean region are controlled by changes in the intensity
of the annual cycle that in turn were controlled by long-term insolation changes forced by
orbital mechanics. Times of intense seasonality should result in a far northerly position of
the ITCZ during the northern hemisphere summer, bringing increased rainfall to the
American tropics. The intensity of the annual cycle in the northern tropics reached a
maximum during the early Holocene and declined toward the present as a result of changes
in seasonal insolation forced by the Earth's processional cycle. I test the hypothesis of
Hodell et al. (1991) that E/P in the circum-Caribbean region responded to insolation forcing
by decreasing during the early Holocene when the intensity of the annual cycle was greatest
and increasing during the mid to late Holocene when the intensity of the annual cycle
diminished. If this hypothesis is correct, the level of Lake Valencia should be high during
the early and middle Holocene followed by lower levels during the late Holocene.
The Lake Valencia sediment record is ideal for testing this hypothesis because the
sediments contain abundant shell material throughout the Holocene section and accelerator
mass spectrometry (AMS) 14C dates provide good chronological control. Furthermore,
Lake Valencia is located at about the current northernmost position (10N) occupied by the
intertropical convergence zone (ITCZ) in its yearly march north and south, making it
sensitive to past changes in the position of the ITCZ.
Lake Valencia, Venezuela, lies at 1010'N, 6745'W, about 400 m above sea level
(asl) (Figure 2-1). The lake has a present maximum depth of -37 m and an area of -350
km2. Lake Valencia is located in the Aragua Valley, a tectonic depression between two
east-west running mountain ranges, the Cordillera de la Costa to the north and the Serranfa
del Interior to the south. Bedrock underlying the basin is composed mainly of Mesozoic
metamorphic rocks formed from Jurassic marine sediments (Bell 1971). Because the
Aragua Valley is densely inhabited (670 people/kin2, total of 2 x 106 people) and
industrialized, human impact on the basin is pronounced (Mogoll6n & Bifano 1994).
During the last several decades, lake level declined and lake water salinity increased
because input waters were diverted and ground waters were pumped for agricultural
irrigation and other uses. The lake is warm, monomictic and hypereutrophic (Lewis 1983;
Jaffd et al. 1993). Discharge of raw sewage causes high nutrient concentrations in the lake
and high rates of primary productivity. Chemical analyses of recent sediments in Lake
Valencia were used to investigate the basin's pollution history (Jaff6 et al. 1995).
Lake Valencia is located in an area with seasonally dry tropical climate.
Temperature data for a station in nearby Maracay (Figure 2-1) show a mean annual
temperature of 24.5C with maximum temperature in April and May (26.0C) and minimum
temperature in January (23.3C) (Snow 1976). Precipitation is highly seasonal with a wet
season lasting from May to October. Mean annual precipitation is 1,000-1,500 mm in the
l100 s"O 30 Lake Valencia \
Gul of Atlantic
Gulf of ),
20N a -
100W 90W 80W 70W
Figure 2-1A Map of Lake Valencia, Venezuela, showing the approximate location of
coring site Valencia 16-VII-94 near the northern shoreline. Contour isopleths in meters.
Arrow in 2-1 B indicates the location of Lake Valencia in northern Venezuela.
valley with greater precipitation at higher elevations in the catchment (Mogoll6n & Bifano
1994). Evaporation is also seasonally variable with a maximum in the dry season.
Hydrologic inputs include precipitation falling directly on the lake, intermittent flow
from a number of streams entering the lake, runoff and subsurface inseepage. Loss of
water from the lake occurs mainly through evaporation but possibly also through
subsurface outseepage. At times of significantly higher lake level, water can also be lost
via outflow over a sill at 427 m asl in the southwestern part of the lake.
The Lake Valencia basin possesses a substantial sediment record. Seismic survey
revealed sediment accumulations >100 m (Schubert & Laredo 1979; Schubert 1980). On
the basis of seismic stratigraphy, sediments were divided into three units separated by
prominent reflectors. Sedimentological and palynological evidence indicates that the
uppermost unit is a Holocene lacustrine deposit. Sediment from the lower two units have
never been sampled but probably represent older lacustrine sequences (Schubert 1980). It
has been suggested that the bottommost sediments may be as old as late Tertiary (Peeters
1970). Acoustic study also detected evidence in the Holocene sediments of two zones of
faulting running east-west across the lake (Schubert & Laredo 1979).
On 16 July 1994 a continuous 568-cm sediment core was taken in 9.4 m of water
from the northern side of Lake Valencia (Figure 2-1). The profile is hereafter referred to as
Val 16-VII-94. The uppermost 88 cm of sediment were taken using a piston corer
designed especially to retrieve undisturbed sediment/water interface profiles (Fisher et al.
1992). Deeper core sections (50 568 cm) were taken in 1-m sections using a square-rod
piston corer (Wright et al. 1984). Coring terminated in stiff deposits -5.7 m below the
sediment surface. The mud water interface (MWI) core was transported to the lake shore,
extruded, and sampled in 1-cm intervals in a vertical position to maintain the stratigraphy of
unconsolidated uppermost sediments. Samples were placed in labelled Whirl-pakT bags
for transportation to the laboratory. Deeper, consolidated sediments were extruded
horizontally in the boat, photographed, sealed in plastic wrap and aluminum foil, and
labelled. Core sections were placed in lengths of PVC tubing for transport.
In the laboratory, core sections taken with the square-rod piston corer were split
lengthwise, photographed again, and sampled. Half of the core was cut into 1-cm intervals
that were subsampled for geochemical analyses. Subsamples were stored in labelled 20-cc
scintillation vials. The remaining half core was archived.
Surface lake water samples were collected in QorpakT bottles at the coring site.
Rain water was collected over the lake and also in nearby Maracay (Figure 2-1). Lake
water and rain water samples were returned to the lab and stored at 4C prior to analysis.
Radiocarbon dating of samples from Val 16-VII-94 was accomplished by
Accelerator Mass Spectrometry at the National Ocean Sciences Facility (NOSAMS) at
Woods Hole Oceanographic Institution. Nine samples of terrestrial organic carbon (wood
or charcoal) were dated. Four samples of ostracod shell calcite were also dated to evaluate
the magnitude of hard-water-lake error (Deevey & Stuiver 1964). Radiocarbon ages were
converted to calendar ages using CALIB (Stuiver & Reimer 1993), with a 200-year moving
average of the tree ring calibration data set and a 500-year moving average of the
atmospheric spline of the coral calibration data set (Linick et al. 1986; Bard et al. 1993;
Kromer & Becker 1993; Pearson & Stuiver 1993; Pearson et al. 1993; Stuiver & Pearson
Subsamples for carbonate and organic matter content and elemental analyses were
oven-dried at 60C and ground into a fine powder with mortar and pestle. Organic matter
concentration in sediments was determined at 1-cm intervals in the MWI core and -10-cm
intervals thereafter by weight loss on ignition at 550C for two hours in a muffle furnace
(Hf.kanson & Jansson 1983). The percentage of carbonate in samples taken at 1-cm
intervals was measured by coulometric titration (Engleman et al. 1985) with a UIC Model
5011 coulometer with a System 140 preparation line that used 2N HC1 to evolve CO2.
Cations (Ca, Mg, Fe, K, Na) and total phosphorus in sediment samples were measured at
1-cm intervals in the MWI core and -10-cm intervals thereafter by inductively coupled
plasma spectroscopy (Jarrell-Ash ICP 9000) following combustion at 550C for two hours
and digestion in 1N HC1 for one hour (Anderson 1976). Organic matter, carbonate (as
CaCO3), and elemental concentrations in sediments are expressed as percent dry weight.
For stable isotope analysis, part of each 1-cm subsample was disaggregated by
soaking in 3% H202 and subsequently sieved at 250 itm to remove silts and clays. Coarse
material was collected on filter paper, dried at 60C and transferred to glass scintillation
vials. Ostracods and gastropods were separated from the matrix using a dissecting
microscope at lOx magnification. Adult specimens of the ostracod Cytheridella boldi were
picked from the sieved 425 500 |.tm fraction, Heterocypris communis were picked from
the 351 500 [tm fraction, and Cypria obtusa were picked from the 250 300 iim fraction.
Specimens of the gastropod Pyrgophorus sp. were separated from the sieved >1000 pIm
fraction. Ostracod valves were cleaned of organic material using 15% H202, then washed
in deionized water, rinsed in methanol, and dried. Gastropods were cleaned in the same
manner, except that specimens were cleaned further by sonication. Samples consisting of
many monospecific ostracod carapaces were loaded into a stainless steel carrying boat,
crushed with a glass rod dipped in methanol and allowed to dry overnight. Numerous
monospecific ostracod valves were used to constitute a sample to reduce variance that might
result from running single, short-lived individuals whose isotopic signatures might reflect
short-term (i.e. seasonal) water-column temperature and salinity conditions (Heaton et al.
1995). Multiple (-5) gastropod shells were ground to a fine powder with a mortar and
pestle and a portion of the pulverized aragonite was loaded into stainless steel carrying
boats for isotopic analysis. Multiple individuals, each living approximately one year, were
used to average short-term (i.e. inter-annual) changes in lake conditions.
All carbonate samples for stable isotopic analysis were reacted in a common acid
bath of 100% ortho-phosphoric acid (specific gravity = 1.92) at 90C using a VG Isocarb
preparation system. Isotopic ratios of purified CO2 gas were measured on-line by a triple-
collector VG Prism Series II mass spectrometer. All carbonate isotopic results are reported
in standard delta notation relative to the VPDB standard. Analytical precision was
estimated by routinely measuring a powdered carbonate standard (Carrara Marble-UF)
along with samples from Val 16-VII-94. Precision (1 SD) was -0.10%o for 6180 (n=101
standards). All isotopic data were smoothed using a 5-pt (i.e. 5 cm) running mean to
illustrate long-term trends.
Lake water and rain water samples for isotopic analysis were equilibrated with pure
CO2 gas in evacuated VacutainerTM tubes (Socki et al. 1992). Equilibrated CO2 was
distilled from water and non-condensable gases off-line. Next, the purified CO2 gas was
sealed in glass breakseal tubes and transferred to the VG Prism II mass spectrometer for
analysis. Oxygen isotopic results for waters are reported in delta notation relative to
Standard Mean Ocean Water (VSMOW). Analytical precision (1 SD) was +0.09%o for
8180 based on routine measurement of an internal standard (UF-alpha).
Chronology and Sedimentation Rates
All but two of the wood/charcoal samples submitted for AMS dating yielded dates
in stratigraphic order (Table 2-1, Figure 2-2). Dates on wood samples at 486 cm and 511
cm are reversed, but similar (10,200 55 and 9,960 70 radiocarbon years BP,
respectively). Ages for the samples at 486 and 511 cm were averaged, yielding a date of
10,080 radiocarbon years BP at 498.5 cm. The age/depth model for the Val 16-VII-94 was
developed using only dates on terrestrial organic matter. Dates on ostracod shells are
stratigraphically ordered and consistent with the age/depth model derived using terrestrial
organic matter (Figure 2-2, Table 2-1). Lack of offset between dates on terrestrial organic
matter and ostracod calcite indicates that hard-water-lake error does not affect dates in
Table 2-1. AMS radiocarbon dates, calibrated ages and mean linear sedimentation rates for
samples from Lake Valencia Core 16-VII-94. Radiocarbon ages are adjusted to a 513C
value of -25%o. Sedimentation rates between wood/charcoal dates were computed using
the calibrated ages and represent the mean rate between the level for which they are
indicated and the next dated level higher in the core. Ages for samples at 486 and 511 cm
were averaged for a date of 10,080 radiocarbon years BP at 498.5 cm and this depth/age
was used for calculations of sedimentation rate.
Sample Type Depth Accession Radiocarbon Age Calibrated Age Sed.
(cm) Number (yr BP) (AD/BC) (cm yr1)
Ostracod shells 31.5 OS-8855 48530 1426 AD
Ostracod shells 69.5 OS-8857 181035 220 AD
Ostracod shells 253 OS-8860 799045 6945 BC
Ostracod shells 378 OS-8861 937080 8390 BC
Terrestrial Wood 31 OS-8854 18540 1715 AD 0.13
Terrestrial Wood 69.5 OS-8856 173035 329 AD 0.03
Terrestrial Wood 139 OS-8858 331035 1575 BC 0.04
Terrestrial Wood 171 OS-8859 483040 3630 BC 0.02
Terrestrial Wood 231 OS-10011 767040 6470 BC 0.02
Charcoal 249 OS-10012 833085 7390 BC 0.02
Terrestrial Wood 486 OS-8862 10200+55 10000 BC
Terrestrial Wood 511 OS-8863 996070 9140 BC 0.11
Terrestrial Wood 560 OS-8864 1240060 12570 BC 0.02
* see explanation in table legend
14000 , , i I I I I , , . .
[3 Ostracods -
0 100 200 300 400 500 600
Depth in sediment (cm)
Figure 2-2 Radiocarbon ages versus depth for terrestrial wood/charcoal and ostracod shell
samples from Valencia core 16-VII-94. The core chronology was obtained by linear
interpolation between AMS 14C dates from nine terrestrial wood/charcoal samples (see
text). The mean of the two dates at 486 and 511 cm was calculated and the average was
used in the core chronology (see text). Dating error (la) on all samples ranged from 30 to
85 years and is within the plot symbols.
Holocene sediments from Lake Valencia (Deevey & Stuiver 1964). Linear sedimentation
rates vary over the length of the core between 0.02 and 0.13 cm yr1 (Table 1).
Three rain water samples collected on 16 July 1994 had 8180 values of -4.37%o,
-5.03%o, and -5.94%o (mean = -5.1%o). My mean value for rain water 8180 is -1%o more
negative than the long-term weighted mean for regional precipitation (-4.01%o) reported by
Rozanski et al. (1993). The discrepancy may be due to my collecting rain water on only
one summer day. Oxygen isotopic values of +3.27%o and +3.15%o (mean = +3.2%o) were
measured from surface lake water samples collected at the coring site. Mean lake water is
evidently enriched in 180 by evaporation and is 7.2%o more positive than the long-term
weighted mean for rain water.
The core terminated in stiff, clay-rich sediments that were dated by extrapolation at
-12,660 14(2 yr BP. Clay is indicated by relatively high concentrations of K (Figure 2-3).
Between -12,500 and 10,500 14C yr BP, the contribution of CaCO3 gradually decreased
as the concentration of K and Fe in the sediments increased. From -~ 10,500 to -10,000
14C yr BP, carbonate concentration dropped even more rapidly and the concentration of K
and total phosphorus in the sediments increased rapidly. In the clay-rich basal sediments,
organic matter concentrations were low. Inorganic carbon and Ca concentrations displayed
high positive correlation throughout the profile, suggesting that they both reflect CaCO3 in
the sediment matrix.
Elemental chemistry in the interval between -10,000 and -8,200 14C yr BP exhibits
the highest variability of the entire record (Figure 2-3). This interval of high variability
coincides with the only section of laminated sediment in the core. Within this interval,
concentrations of Ca and IC are inversely correlated with concentrations of Fe and K,
suggesting alternating deposition of non-carbonate plastics and carbonate. Organic matter
gradually increases from -10% of sediment dry mass at -10,000 14C yr BP to -20% of the
01 I 0
0 0I I0 0 0 0 o 0 0 0
01 0 0 0 0\ 0 0
(dal 8.IOX uoqj( ooipL) &V
Figure 2-3 Calcium carbonate content (%CaC03), organic matter content (%LOI), and
selected elemental concentrations (mg g-1 Ca, Mg, Fe, K, Na, and P) versus radiocarbon
age in Valencia core 16-VII-94.
sediment mass by -8,200 14C yr BP. Concentration of Mg in the sediments increases
dramatically after -10,000 14C yr BP, peaks at -9,000 14C yr BP and then decreases by
-8,400 14C yr BP. Carbonate content was low from -7,500 14C yr BP to -6,700 14C yr
BP (Figure 2-3). Over the same interval, organic matter constituted a much greater portion
of the sediment matrix and Fe and total phosphorus concentrations were both elevated.
Carbonate concentrations increased abruptly at -6700 14C yr BP and continued to
increase gradually until -4,000 14C yr BP (Figure 2-3). Accompanying this increase in
carbonate was a decrease in concentration of both organic matter and total phosphorus.
From -4,000 to -3,000 14C yr BP elemental chemistry showed little change except for a
very brief but dramatic drop in carbonate at -3,300 14C yr BP. This episode of low
carbonate was detected by 1-cm sampling for IC but was not observed in the other
elemental chemistries because the event fell between levels sampled at 10-cm intervals.
Visual examination of the samples at -3,300 14C yr BP revealed large amounts of organic
carbon (wood pieces) and very little shell material. One organic carbon date (3,31035 yr
BP) came from this layer.
At -3,000 14C yr BP carbonate began a very gradual decrease that lasted until the
present (Figure 2-3). During this same interval Fe, K, Mg, and Na all increased gradually.
During this interval clay was accumulating at an increased rate relative to carbonates. Total
phosphorus showed little change from -3,000 to -1,000 14C yr BP when it began a slight
increase. Organic matter concentration also changed little in the past -3,000 years.
Stable Isotope Geochemistry
The oxygen stable isotopic record from Lake Valencia consists of signals from three
ostracod and one gastropod species (Figure 2-4). Changes in species composition and
abundance precluded obtaining a single monospecific record for the entire core. Data for
H. communis were obtained from the deeper section of the core (-10,000 to -7,700 14C yr
BP) and from the uppermost section of the core (-2,020 14C yr BP to present). Specimens
of C. obtusa were abundant enough for isotopic analysis only in the bottom section of the
81 80 (%c, PDB)
Figure 2-4 Oxygen isotopic composition in ostracods (Cytheridella boldi, Heterocypris
communis, and Cypria obtusa) and gastropods (Pyrgophorus sp.) versus radiocarbon age
in Valencia core 16-VII-94. It was necessary to measure multiple species because changes
in species presence and abundance precluded obtaining a single monospecific record for the
entire core. Circles represent measurements of H. communis, plus symbols represent C.
boldi, x symbols represent C. obtusa and squares represent Pyrgophorus sp. All data have
been smoothed with a 5-pt running mean to better illustrate long-term trends. Carbonate
isotopic data are available from the National Geophysical Data Center, Boulder, Colorado,
core between -10,000 and -8,770 14C yr BP. C. boldi were analyzed in sediments
deposited between -8435 14C yr BP and present, and in two very short intervals deeper in
the core (-9,000 to -8,950 and -9,750 to -9,670 14C yr BP). The gastropod
Pyrgophorus sp. was found and analyzed only in the topmost 79 cm of the record (i.e.
sediments deposited from -1,960 14C yr BP to the present).
The oxygen isotopic records of H. communis and C. obtusa displayed high
variability in the older portion of the core from -10,000 to -8,700 14C yr BP (Figure 2-4).
Oxygen isotopic values of H. communis ranged from -0.5 to -3.5%o during this period
while values of C. obtusa range from -1 to -3%o. The 6180 records of these two
ostracods are highly correlated (r=--0.86) in samples that contained both species. At -8,770
14C yr BP when C. obtusa disappeared, the 6180 record of H. communis became less
variable. At that time, 5180 values of H. communis began to increase gradually, reaching a
maximum of -3.3%o at -8,600 14C yr BP. Thereafter, values decreased steadily to about
1.8%o by -8,100 14C yr BP. During the next 500 years, 6180 values ofH. communis
remained between 1.8%o and 2. 1%o and H. communis became rare after -7,600 14C yr BP.
Oxygen isotopic variability of C. boldi was low relative to that measured in H.
commnunis and C. obtusa (Figure 2-4). The persistent occurrence of C. boldi began at
-8,435 14C yr BP. The 6180 values of C. boldi in the older portion of the record were
generally greater than values later in the record. Notable features of the 6180 record of C.
boldi include; high 6180 at -7,000 14C yr BP, low 5180 at -4,900 14C yr BP, followed
by a lengthy period (-4,850 to 4,400 14C yr BP) with sparse C. boldi, another period of
sparse C. boldi from -3,100 to 2,750 14C yr BP, and a rapid shift in isotopic values from
low to high during the period -2,200 to 2,000 14C yr BP (Figure 2-4).
Beginning at -2,140 14C yr BP, isotopic results ofH. communis show high
variability (Figure 2-4) similar to that observed in the older portion of the record. Between
-2,140 and-500 14C yr BP, 6180 values ranged from -0.9%o to 1.5%c. Beginning at
-500 14C yr BP, 6180 values of H. communis increased, eventually reaching values of
-2.5%o at -100 14C yr BP. Oxygen isotopic values then decreased to about 1.2%o at
present. This excursion was not observed in the isotopic record of C. boldi.
From -1,960 14C yr BP to the present, the 6180 record of Pyrgophorus sp. was
generally similar to the 6180 record of C. bold, but unlike the signal ofH. communis
(Figure 2-4). The magnitude of variability in the Pyrgophorus sp. and C. boldi isotopic
records was comparable.
Lake Valencia is currently closed and its water level is at 402 m asl, well below the
outflow sill at 427 m asl. At different times in the past, Lake Valencia has varied between a
closed and open hydrologic system. This affected the oxygen isotopic balance of the lake
water and shell material that was precipitated in the lake. When the lake was open (i.e. lake
level was high and water was lost by outflow), the ratio of 180 to 160 of lake water was
low and relatively constant because the isotopic composition was controlled mainly by the
6180 of input waters. When the basin was closed (i.e. lake level was low and water was
lost only by evaporation), the lake water was isotopically enriched in 180 due to
preferential evaporative loss of 160. At low stage, the lake was not only isotopically
enriched in 180, but the isotopic composition of the lake was more variable because of the
high sensitivity of a small lake volume to changes in the ratio of evaporation to
Latest Pleistocene (-12,600 to -10,000 14C yr BP)
During the latest Pleistocene, between -12,600 and -10,000 14C yr BP, evidence
from sediment core Val 16-VII-94 suggests that conditions were dry with the coring site
intermittently covered by water. The core terminated in stiff sediments with relatively high
concentrations of Fe and K and moderate calcium carbonate content (Figure 2-3). This
basal material is similar to the calcareous and gypsiferous kaolinite and illite that Bradbury
et al. (1981) describe from the base of a section retrieved from the deepest part of the lake.
This clay unit is probably the uppermost, prominent seismic reflector that underlies the
entire lake (Schubert 1980; Lewis & Weibezahn 1981). Well-preserved diatoms were
found throughout the clay-rich basal sediments of core Val 16-VII-94, indicating that the
coring site was lacustrine at least intermittently between -12,600 and -10,000 14C yr BP.
Diatom assemblages found in the basal clay sediments suggest variable lacustrine
conditions (T. Whitmore, personal communication, 1997). Diatoms in most samples are
indicative of saline, shallow water except for a short event of highly eutrophic, fresh
conditions at -10,900 14C yr BP. Pollen analysis also suggests that the coring site was
covered by water intermittently prior to -10,000 14C yr BP (B. Leyden, personal
communication, 1997). Lacustrine carbonate microfossils ostracodss or gastropods),
however, are absent in sediments older than -10,025 14C yr BP.
Intermittent wetting of the coring site during the latest Pleistocene is consistent with
conclusions from previous studies in the Valencia Basin (Figure 2-5). Studies of Lake
Valencia by Salgado-Labouriau (1980), Bradbury et al. (1981), Lewis & Weibezahn
(1981), Binford (1982), and Leyden (1985) are based on a 7.5-m sediment core, hereafter
referred to as Val 14-1-77, taken from the deepest portion of the lake (water depth 37 m).
In addition to Val 14-1-77, Leyden (1985) and Binford (1982) used two shallow water
cores in their studies. On the basis of pollen analysis from a suite of three cores, Leyden
(1985) concluded that climate was arid during the late Pleistocene and that the Valencia
basin was occupied by a saline marsh surrounded by savanna. Similarly, Salgado-
Labouriau (1980) inferred a swamp or intermittent lake and semi-arid regional vegetation
based on pollen analysis. Lewis & Weibezahn (1981) interpreted geochemical data from
Val 14-1-77 to indicate a seasonal lake or dry basin. Combining pollen, diatoms,
geochemistry, and animal microfossil data, Bradbury et al. (1981) reported that the
Valencia basin contained a marshlike environment with fluctuating water levels during the
late Pleistocene. Dry conditions during the late Glacial have been proposed for many other
sites in the circum-Caribbean region including Florida (Watts 1975), Peten, Guatemala
Binford Bradbury Lewis and
1982 et al. 1981 Weibezahn
- Y I.
PAo iotfCl iirn
Figure 2-5 Comparison of lake level interpretations from Valencia core 16-VII-94 with
lake level, salinity, and vegetation interpretations of previous studies (Salgado-Labouriau
1980; Bradbury et al. 1981; Lewis & Weibezahn 1981; Binford 1982; and Leyden 1985).
LLL=Brief low lake level, C=Lake was closed-basin, O=Lake was open-basin,
stable lake S
(Deevey et al. 1983; Leyden 1984; Leyden et al. 1993, 1994; Brenner 1994), Panama
(Bush & Colinvaux 1990; Pipemo et al. 1990; Bush et al. 1992), Haiti (Hodell et al. 1991,
1995; Higuera-Gundy 1991; Curtis & Hodell 1993), and Jamaica (Street-Perrott et al.
1993; Holmes et al. 1995).
Diatoms in the basal section of core Val 16-VII-94, taken from 9.4 m water depth
was surprising considering that no diatoms were reported at depths > 4.92 m (-10,500 14C
yr BP) in Val 14-1-77, taken from 37 m water depth (Bradbury 1979; Bradbury et al.
1981). Bradbury et al. (1981) suggested that diagenetic removal of aquatic fossils may
have occurred when the lake was intermittent. Ostracods were, however, found in
sediments as old as -11,300 BP in Val 14-1-77 (Bradbury 1979; Bradbury et al. 1981;
Binford 1982). Ostracods appeared in sediments from the Val 14-1-77 site about 1,300
years before they appeared at the shallower site of Val 16-VII-94. Apparently, deep water
sediments in the basin did not preserve diatoms as well as deposits from the shallower, Val
16-VII-94 coring site. Conversely, deep water sediments preserved ostracods, whereas
they were not preserved at the site of Val 16-VII-94.
Earliest Holocene (-10,000 to -8,200 14C yr BP)
The Val 16-VII-94 coring site was covered permanently with water after -10,000
l4 yr BP as indicated by several lines of evidence. At -10,000 14C yr BP sediment
composition changed dramatically from grey clay with high concentrations of K and Fe to
organic marls and silts with higher carbonate and organic matter concentrations, indicating
permanent submergence of the coring site (Figure 2-3). Changes in sediment lithology,
,,, ether with the appearance of lacustrine ostracods at -10,025 14C yr BP, indicate the
inception of continuous lacustrine sedimentation. Pollen analysis in Val 16-VII-94 also
suggests that lake level rose -10,000 14C yr BP (B. Leyden, personal communication,
This estimated timing of permanent filling of Lake Valencia is consistent with
previous studies (Figure 2-5). Evidence from Val 14-1-77 revealed that the lake began to
fill permanently by 10,500 14C yr BP (Bradbury et al. 1981; Lewis & Weibezahn 1981;
Binford 1982; Leyden 1985). By -10,000 14C yr BP, lake level had risen above the
present 10-m depth contour (Binford 1982).
Increased rainfall may have initiated the permanent filling of Lake Valencia detected
at -10,500 14C yr BP in the deepest portion of the basin. Laminated marine sediments
from the nearby Cariaco Basin suggest that runoff from the northern Venezuela coast was
very high from -10,800 to -9,800 14C yr BP (Hughen et al. 1996a), coinciding with the
Younger Dryas (11,000 to 10,000 14C yr BP). Prior to permanent inundation of the Val
16-VII-94 coring site, there may have been increased runoff and plastic transport to the site,
as indicated by increased concentrations of clay-bound Fe and K in the core between
-10,500 and -10,000 14C yr BP. Additionally, pollen analysis revealed that effective
moisture in the Valencia basin increased between 10,500 and 9,800 14C yr BP (Leyden
1985). Pollen evidence from Panama also suggests that precipitation increased during the
Younger Dryas (Bush et al. 1992).
From -10,000 14C yr BP to -8,200 14C yr BP, all geochemical and oxygen
isotopic data indicate variable conditions in Lake Valencia (Figures 2-3 & 2-4). Although
the geochemical and isotopic evidence suggest fluctuating lake levels, the Val 16-VII-94
coring site remained water-covered throughout this entire period. Concentrations of
CaCO3 and clay-bound Fe and K are inversely correlated and may indicate fluctuating lake
levels (Figure 2-3). During low stands, detrital plastics provided the dominant input
whereas during higher lake stands, autochthonous carbonate predominated.
Laminations are visible in the earliest Holocene part of the Val 16-VII-94 core,
especially between -9,500 and -9,300 1l4C yr BP. The carbonate portion of these
laminations is composed of aragonite in Val 14-1-77. Laminae couplets have been
interpreted as supra-annual events, rather than annual varves (Lewis & Weibezahn 1981).
The highest Mg concentrations in the Val 16-VII-94 record occurred at this time, which is
to be expected because aragonite formation requires high Mg:Ca ratios (Lewis &
Weibezahn 1981). Lewis & Weibezahn (1981) reported high early Holocene Mg:Ca ratios
from Val 14-1-77.
Changes in the ostracod assemblage are interpreted as evidence of variable lake
levels and salinities during the early Holocene. The presence of H. communis, a halophilic
species that tolerates salinities between 2 and 40%o (Binford 1982), suggests relatively high
lake water salinity during this period. Alternatively, the appearance of the ostracod C. boldi
for two brief periods during the early Holocene suggests short-duration events of relatively
fresh conditions. This ostracod probably prefers low-salinity conditions on the basis of the
preference of its congener C. ilosvayi for freshwater lakes on the Yucatan Peninsula of
Mexico and in Guatemala (Curtis et al. 1996; Whitmore et al. 1996; Curtis et al. in press).
Isotopic records of H. communis and C. obtusa indicate considerable variation,
reflecting lake level and salinity fluctuations. These isotopic records correlate where they
overlap between -8,375 and -7,575 14C yr BP (Figure 2-4), even though one taxa is
planktonic or epiphytic (C. obtusa) and the other taxa is a benthic burrower that generally
lives offshore (H. communis) (Binford 1982). Despite these habitat differences, both taxa
recorded similar changes in lake water 6180.
The highly variable lake conditions observed in the earliest Holocene record of Lake
Valencia are probably caused, in part, by the lake being "closed" with reduced volume
during this period. Closed-basin lakes are highly sensitive to changes in E/P because water
is lost from the system only by evaporation. Furthermore, small-volume lakes are more
sensitive to changes in the ratio of evaporation to precipitation than are large-volume lakes.
The mechanism that generated highly variable E/P ratios that, in turn, caused large
changes in salinity and lake level in the earliest Holocene remains unclear. Because
limnologic variability has not been observed in early-Holocene records from other circum-
Caribbean lakes, the causal mechanism behind such variability is probably local
hydrological control rather than regional or global control. High variability in the earliest
Holocene cannot be explained by insolation-driven changes in the annual cycle's intensity
because rates of change detected in the sediment record are much too fast and are nonlinear.
Early to Middle Holocene (-8,200 to -3,000 14C yr BP)
During much of the early to middle Holocene (-8,200 to -3,000 14C yr BP), lake
levels were high in the Valencia basin, and water was lost by flow over the sill (427 m asl).
Following fluctuating conditions and lake levels of the early Holocene, long-term lake-level
rise is indicated by decreasing 6180 values of H. communis that began at -8,500 14C yr
BP and continued until -8,200 14C yr BP (Figure 2-4). H. communis occurs infrequently
after -7,500 14C yr BP and its absence indicates lower salinities. As discussed previously,
C. boldi is thought to live in lakes with low salinities, so its appearance at -8,400 14C yr
BP and persistence thereafter suggests freshening. During most of the early to middle
Holocene, the 5180 record of C. boldi shows little variation. This probably indicates that
E/P was not the major factor controlling the oxygen isotopic composition of lake water
when the lake was overflowing.
The isotope-based inference for high lake levels in Valencia during the early to
middle Holocene is consistent with most previous studies (Figure 2-5). Combining pollen,
diatoms, geochemistry, and animal microfossil data from Val 14-1-77, Bradbury et al.
(1981) reported high lake levels from 8,700 to 7,900 14C yr BP and again from 6,000 to
3,000 14C yr BP while Lewis & Weibezahn (1981) concluded that lake levels were high at
8,000 14C yr BP and again from 6,000 to 2,800 14C yr BP based on geochemical data also
from Val 14-1-77. Binford's (1982) interpretation is slightly different in that he found the
highest lake levels before 8,000 14C yr BP followed by declining water levels until 2,500
l4C yr BP. Bradbury (1979) reported one major period of increased freshening based on
diatom stratigraphy of Val 14-1-77 during the early to middle Holocene, from 8,000 to
5,200 14C yr BP. This was followed by a slightly more saline lake from 5,200 to 1,800
14C yr BP (Bradbury 1979). Bradbury et al. (1981) modified this chronology and reported
high lake level from 8700 to 7900 14(2 yr BP, then a period of low levels from 7900 to
6000 14C yr BP, followed once again by high levels with outflow from 6000 to 3000 14C
yr BP. High lake levels in Valencia occur during the early to middle Holocene moist period
that has been observed elsewhere in the Neotropics including the Yucatan Peninsula,
Mexico (Covich & Stuiver 1974; Hodell et al. 1991, 1995; Curtis & Hodell 1993), Peten,
Guatemala (Deevey et al. 1983; Leyden 1984; Islebe et al. 1996b), Panama (Pipemo et al.
1990) and Costa Rica (Islebe et al. 1996a).
High and overflowing water levels in Lake Valencia during the early to middle
Holocene can be attributed to low E/P driven by greater intensity of the annual cycle.
During this period, perihelion occurred during the northern hemisphere summer half year
resulting in warmer summers while aphelion fell during northern hemisphere winter half
year causing colder winters. This precessional forcing created large differences between
summer and winter insolation in the northern hemisphere tropics and lead to the high
seasonality which caused the ITCZ to travel far north and south in its annual march,
causing wet conditions at Lake Valencia and many other low-elevation sites in the northern
The record from Val 16-VII-94 suggests two intervals of lower water level during
the early to middle Holocene. The first occurred between -7,600 and -6,700 14( yr BP.
A stromatolite at 421 m asl with a date of 7,130 300 14C yr BP lies well below the
outflow at 427 m asl and indicates the lake was closed at that time (Binford 1982).
Presence of this stromatolite demonstrates that the water level remained at 418 m asl long
enough for stromatolite formation, but water level may have declined below 418 m asl.
Once lake stage fell below the outflow sill, water level would have been influenced again
by moisture availability and the 5180 of lake water would have been controlled again by
changes in E/P. The highest 6180 values of C. boldi (-2%o) in the entire record occur at
-7,000 14C yr BP, suggesting that E/P was high at that time, even higher than at present
(Figure 2-4). Despite lower lake level and higher E/P, the salt tolerant ostracods (H.
communis or C. obtusa) did not become more abundant in Val 16-VII-94 during this period
of lake level drop.
Lower lake level and higher salinity during the early Holocene have been reported
for Lake Valencia (Figure 2-5). Using Val 14-1-77, Bradbury et al. (1981) inferred lower
lake levels from 7,000 6,000 14C yr BP while Lewis & Weibezahn (1981) placed the
saline episode between 7,400 and 6,000 14C yr BP. Binford (1982) interpreted much of
the early to middle Holocene as a time of lowered lake levels.
By -6,600 14C yr BP 8180 of C. boldi returned to more negative values (Figure 2-
4). This suggests that water level increased, and probably reached the overflow at 427 m,
and that the oxygen isotopic composition of lake water was no longer controlled by E/P.
Binford (1982) reported lower lake levels at -5,480 300 14C yr BP based upon a
stromatolite at 418 m asl. A water level decline of this magnitude would have caused the
lake to become closed once again. Oxygen isotopic and geochemical records from Val 16-
VII-94, however, indicate no lake level drop. Ostracods are sparse from -5,150 to -4,150
14C yr BP and the poor sampling resolution of C. boldi for 6180 yields a poorly resolved
picture of lake level changes in this interval.
The second period of lowered lake level during the middle to late Holocene
occurred over a brief, 150-year interval centered on -3,300 14C yr BP. Lake level drop is
inferred from low carbonate concentrations (Figure 2-3) and large quantities of wood in
sediment. This event was brief enough that geochemical sampling at 10-cm intervals failed
to reveal it. It was, however, resolved by sampling at 1-cm intervals for calcium carbonate
analysis. Pieces of wood are very common in a 5-cm section of the core (-3,250 to -3,400
14C yr BP). Ostracods are almost entirely absent from this interval and, as a result, no
isotopic record exists for this event. The drop in lake level and recovery to higher levels
occurred quickly, over a period of one or two centuries.
Late Holocene (-3,000 14C yr BP to the present)
Geochemical evidence of increased Na concentrations in sediments since -3,000
14C yr BP suggests recent desiccation of Lake Valencia. Once the lake became
hydrologically closed, solutes in lake water began to increase.
Changes in carbonate microfossil taxa also suggest lowered lake levels and higher
salinity by -2,140 14C yr BP. At -2,140 14C yr BP the saline-tolerant ostracod H.
communis returned to the lake, suggesting increased salinity. Soon afterwards (-1,960
14C yr BP), gastropods of the genus Pyrgophorus appeared in the core, probably
indicating low lake levels. Binford (1982) reported that gastropods were absent in offshore
surficial sediments and suggested that their presence in a core indicated proximity of their
littoral habitat to the coring site at the time of deposition.
Both measured values and trends in 8180 records of C. boldi and Pyrgophorus sp.
are similar during the late Holocene. Neither record displays large changes (Figure 2-4).
Oxygen isotopic results of H. communis, however, reveal a positive excursion centered on
-500 14C yr BP (Figure 2-4). The difference between oxygen isotopic records from these
two species remains unexplained.
Previous reconstructions of latest Holocene water level in Lake Valencia are
contradictory (Figure 2-5). Bradbury et al. (1981) reported maximum freshness and lake
level at -3,000 14C yr BP, followed by declining lake level and increasing salinity to the
present_ Lewis & Weibezahn (1981) also infer freshest conditions at -3,000 14C yr BP,
with a salinity increase beginning at 2,800 14C yr BP and continuing to the present. They
report, however, overflow conditions at 1727 AD based on historical accounts (Humboldt
& Bonpland 1819). Binford (1982) reports lake level decline until 2,500 14C yr BP
followed by high lake levels and some overflow until 500 14C yr BP, and overflow at 1727
AD, followed by desiccation.
Data from Val 16-VII-94 do not support high lake levels or overflow during the
latest Holocene. Reports of lake overflow at 1727 AD (Humboldt & Bonpland 1819) have
been used as evidence for high lake levels in the recent past (e.g. Lewis & Weibezahn
1981; Binford 1982). Recent overflow and high lake levels have been dismissed by some
authors on the basis of archaeological, topographic, and stratigraphic evidence (Jahn 1940;
Cruxent & Rouse 1958).
The trend toward drier conditions during the past 3,000 years can be explained by a
reduction in seasonality caused by an orbitally driven decrease in the annual cycle's
intensity. The annual cycle's intensity decreased in the late Holocene as perihelion shifted
to the northern hemisphere winters, resulting in warmer winters. Aphelion shifted to
northern hemisphere summers, causing cooler summers. This resulted in a reduction of the
northerly and southerly movement of the ITCZ, in turn causing less precipitation in
northern South America and the circum-Caribbean region.
Climate variability and lake level change were inferred from several lines of isotopic
and sedimentological evidence from Lake Valencia, Venezuela. Evidence from a 568-cm
core (Val 16-VII-94) taken in 9.4 m of water suggests that during the latest Pleistocene
(- 12,600 to -10,000 14C yr BP) the Valencia basin was relatively dry with the coring site
covered intermittently with water. Beginning at -10,000 14C yr BP, the coring site was
covered with water permanently. During the earliest Holocene (10,000 to 8,200 14C yr
BP) fluctuations in lake level and salinity were pronounced. During this period, Lake
Valencia was a hydrologically closed basin, so that oxygen isotopic composition of lake
water and ostracods were controlled by E/P. The cause of variable lake conditions in the
earliest Holocene is unclear, but was probably local control of Lake Valencia's hydrologic
In contrast to the earliest Holocene, during much of the early to middle Holocene
(8,200 to 3,000 14C yr BP) lake levels were high enough so that water was lost by
overflow and the lake was hydrologically open. Thus, 6180 of the lake water was not
controlled solely by E/P. High lake levels during the early to middle Holocene were
probably a result of increased intensity of the annual cycle caused by large differences in
seasonal insolation driven by orbital mechanics. Two periods of lowered lake level, during
which time the lake basin was closed, are recorded during the early to middle Holocene.
During the first lake level drop, centered on -7,000 14C yr BP and lasting -900 years,
oxygen isotopes suggest high E/P. The second drop in lake level, at -3,300 14C yr BP,
was briefer (- 150 years) but sufficiently pronounced so that the littoral zone was at or near
the Val 16-VII-94 coring site. Since -3,000 14C yr BP, Lake Valencia has been
desiccating and its level dropping. Drying in northern South America during the past 3,000
years can be explained by a reduction in the intensity of the annual cycle driven by orbital
The Department of Peten, in the northern Guatemala lowlands, encompasses
35,854 km2 and constitutes much of the area in which ancient Maya civilization arose more
than 3000 years ago, flourished for nearly two millennia, and then collapsed mysteriously
in the ninth century AD. Postclassic Maya populations continued to occupy the region until
the European conquest, but never attained the numbers or achieved the cultural heights of
their ancestors. The Maya lowlands have captured the imagination of archaeologists and
natural scientists alike because the Maya were one of only a few prehistoric high cultures to
develop and persist in the context of a tropical lowland dry forest. Furthermore, the cause
of the cultural collapse that occurred in the ninth century AD remains a much debated
mystery. Today, relatively low human population density in the region bears testimony to
the inhospitable nature of the low-elevation tropical environment. Nevertheless, the local
population is now expanding in response to increasing land pressures in the Guatemalan
Post-Columbian population in Peten contrasts sharply with archaeologically-
inferred population estimates for the late Classic Maya period (ca. 800 AD). The
population of Peten is estimated to have been 3,027 in 1714 and increased to only 10,000
almost two centuries later (Schwartz 1990). By the mid-1960s the population stood at
25,207, but rose sharply thereafter, to about 300,000 by the mid-1980s, largely because of
in-migration. Contemporary population figures, nevertheless, pale by comparison to late
Classic estimates. Delimiting Tikal's areal extent to about 163 km2, Haviland (1969, 1972)
suggested that this urban site alone supported 40,000-49,000 individuals during the late
Classic. Archaeological work in several Peten drainage basins suggests that late Classic
Maya population densities ranged between 200 and 300 individuals per km2 (Rice & Rice
1990). Overall, pre-Columbian Maya population in Peten may have exceeded 106
inhabitants (Schwartz 1990).
Recent population growth in Peten is already having environmental consequences.
Schwartz (1990) suggests that 40-50% of the Peten landscape was deforested or degraded
by the mid-1980s, and as much as 60% of the vegetation may have been disturbed by
1989. Both logging and land clearance for agriculture have contributed to forest loss.
Upland sites in the karst, hilly terrain are preferred localities for swidden agriculture, but
possess soils that are most prone to erosion. Slash-and-bumrn practices on steep slopes
accelerate soil loss via colluviation.
Ample evidence indicates that Peten previously experienced a protracted episode of
vegetation removal and soil erosion. Several multidisciplinary studies of small drainage
basins have assessed ancient Maya impact on karst Peten watersheds by combining
archaeologically-determined reconstructions of human population densities with
paleolimnological data. Palynological data from lacustrine sediment cores show that people
had begun to remove regional vegetation by Middle Preclassic times, ca. 1000 BC (Cowgill
et al. 1966, Deevey 1978, Deevey et al. 1979, Leyden 1987, Vaughan et al. 1985). As
population densities increased in the Classic period (250-850 AD), widespread
Geochemical and sedimentological data demonstrate that deforestation was
accompanied by rapid soil loss and nutrient sequestering on lake bottoms (Deevey & Rice
1980, Deevey et al. 1980, Brenner 1983, 1994, Rice et al. 1985, Binford et al. 1987).
Scholars have speculated that the Classic Maya collapse was, in part, precipitated by
Paleoenvironmental studies in the tropical Peten lowlands have been informative
concerning long-term human influences on terrestrial and aquatic ecosystems as well as
Pleistocene/Holocene climate change (Deevey et al. 1983, Leyden 1984, 1987, Leyden et
al. 1993, 1994). Past studies involved the smaller basins of the central region, including
Lakes Yaxha, Sacnab, Quexil, Salpeten, Macanche, and Petenxil (Figure 3-1) as well as
savanna Lakes Chimaj, Chilonche and Oquevix farther to the south. The largest lake
investigated prior to this study was Yaxha (A = 7.4 km2), which possesses the ruins of an
ancient urban site on its north shore.
There were several rationales for focusing on smaller basins in the past. First,
small lakes or waterbodies with high watershed:lake ratios are highly sensitive to
anthropogenic disturbances in their drainage basins. Evidence of soil erosion is clearly
seen as a section of clay-rich colluvium in the sediments of small lakes surrounded by
extensive, steep slopes. Second, small, relatively deep lakes such as Quexil (A = 2.1 km2,
Zmax = 32 m), Salpeten (A = 2.6 km2, Zmax = 32 m), and Macanche (A = 2.2 km2, Zmax
= 57.5 m) (Deevey et al. 1980) were thought to be the most promising prospects for
retrieving Pleistocene-age lacustrine deposits that could be used to study climatic conditions
at low latitudes in the Neotropics during the last Ice Age. Deep basins were sought because
water level was believed to have been much lower during the relatively dry late Pleistocene.
This is the case in Florida, where shallow basins were shown to have filled with water only
about 8 kyr BP (Watts 1969), but some deep sinkhole lakes were found to contain
Pleistocene-age lake sediments (Watts 1975, Watts & Stuiver 1980, Watts & Hansen 1988,
1994, Watts et al. 1992, Grimm et al. 1993). Recent studies in lakes of the northern,
Mexican portion of the Yucatan Peninsula demonstrate that lacustrine deposition began in
those basins only after ca. 8000 years ago (Hodell et al. 1995, Whitmore et al. 1996
Leyden et al. 1996), probably in response to rising sea level and greater available moisture.
Focus on small, deep lakes was rewarded in 1980, when Pleistocene-age deposits were
retrieved from Peten Lakes Quexil and Salpeten (Deevey et al. 1983, Leyden 1984, Leyden
et al. 1993, 1994). Finally, smaller lakes were targeted in earlier studies because they are
logistically easier to core.
Gulf of \ Atlantic
Figure 3-1. Lake Peten-Itza showing the approximate location of coring site in the
southern basin. Inset B shows the Central Peten Lake District and locations of previously
studied lakes mentioned in the text. Arrow in inset C shows location of the Central Peten
Lake District in Central America.
In 1993 I undertook paleolimnological study of the largest basin in the district, Lake
Peten-Itza. My objective was to use multiple sediment variables, including elemental
composition, stable isotope geochemistry, pollen, and magnetic susceptibility of samples
from the same core to explore Holocene climate changes and human influences on the
regional environment. I selected Peten-Itza for several reasons. First, previous
paleolimnological study showed that Peten-Itza sediments preserved a continuous record of
changes in molluscan species diversity, and some taxa persisted through the entire
Holocene (Covich 1976). The preservation of carbonate microfossils was important
because I sought to analyze shells for oxygen isotope ratios (6180) which can be used to
infer past climate changes, i.e. shifts in the amount of evaporation relative to precipitation.
Second, stratigraphic changes in pollen spectra from cores collected in small Peten lakes,
although assumed to reflect regional vegetation changes, might in fact indicate shifts in
local, riparian plant communities. Pollen data from a core taken in a large lake may better
reflect regional pollen production. Paleoclimatic inferences could be compared with data
from previously examined circum-Caribbean sites, including Haiti (Hodell et al. 1991,
Curtis & Hodell, 1993, Brenner et al. 1994), Yucatan (Hodell et al. 1995, Curtis et al.
1996), Jamaica (Street-Perrott et al. 1993, Holmes et al. 1995), Belize (Hansen 1990), and
northern Venezuela (Bradbury et al. 1981, Leyden 1985). Third, I focused on Lake Peten-
Itza to determine whether Maya-induced colluvium ("Maya clay"), sometimes seen as a
several-meters-thick stratigraphic unit in small Peten lakes, would be apparent in sediment
cores from this large water body.
Lake Peten-Itza lies at 16055' N, 89'50' W, and is 110 m above mean sea level
(Figure 3-1). The lake has a surface area of 99.6 km2 and a measured maximum depth of
32 m (Brezonik & Fox 1974). There is no bathymetric map for the basin, but reports
suggest that the deepest part of the lake may exceed 60 m (Deevey et al. 1980). Peten-Itza
occupies a large depression in a series of east-west aligned en echelon faults (Deevey et al.
1979), and is divided into two major basins. The large northern basin, which is >20 km
long and some 3 to 4 km wide, supports the towns of San Jose and San Andres on its
north shore. The coring site for this study was located in the smaller southern basin, which
extends east-west for about 14 km and has a maximum width of about 1.5 km (Figure 3-
1). The southern basin today supports the bulk of riparian population in the densely settled
towns of Santa Elena, San Benito and Flores (Figure 3-1). Flores is an island that was
connected to the mainland by a causeway in the 1960s and serves as Peten's political hub.
The best meteorological data for the region come from Flores. Deevey et al. (1980)
presented mean monthly temperature and precipitation records for the periods 1934-1942
and 1971-1976. During the discontinuous 15-year period for which data are available,
annual temperature averaged 24.9C and mean annual rainfall was 1799 mm. Precipitation
is seasonally variable, with a pronounced dry season generally extending from January to
May. The Peten climate is also highly variable over decadal time scales. For instance, the
period from 1934 to 1942 was relatively wet (mean annual precipitation = 2055 mm yr1)
compared with the period of the early to mid-1970s (mean annual precipitation = 1415 mm
yr-1). High lake level and consequent flooding in Flores in 1938 (Penados 1980) were
probably due to plentiful rainfall. Such high lake stands were not recorded again until 1980
and persisted into the 1990s.
Limestone bedrock in Peten was deposited during the Cretaceous and Tertiary
periods (Vinson 1962) and local karst terrain varies in elevation from about 100 to 300 m
afoove mean sea level. The water table in the Peten is deep below the ground surface and
San ace waters are perched. Soils consist principally of mollisols (calcimorphic rendzinas)
,-at cover approximately 90% of the Peten landscape (Simmons et al. 1959). These
mineral soils give rise to or have the capacity to support tropical, lowland dry forest
(Lundell 1937, Holdridge 1947). Members of the Moraceae, Meliaceae, Sapotaceae,
I..-uminosae. and Lauraceae are well represented. Elsewhere, particularly south of Lake
Peten-Itza, hydromorphic, clay-rich soils support grasslands and savanna trees (e.g.
Byrsonima crassifolia, Acrocomia mexicana, and Curatella americana).
On 6 July 1993, a 545-cm sediment core from 7.6 m of water was retrieved in the
southern basin of Lake Peten-Itza (Figure 3-1). The coring site was located several
hundred meters ESE of Flores, near the site where Covich (1976) took a core to study
historical changes in the aquatic mollusc thanatocoenosis. The coring platform consisted of
two, attached wooden boats of the type used to ferry passengers on the lake. The
uppermost 83 cm of the sediment core were collected with a piston corer designed to
retrieve undisturbed sediment/water interface profiles (Fisher et al. 1992). Subsequent
sections, beginning 50 cm below the sediment surface, were collected in approximately 1-
m increments using a square-rod piston corer (Wright et al. 1984). Coring terminated in
stiff, indurated sediments ~5.5 m below the sediment surface.
The unconsolidated mud/water interface core was carried in a vertical position to the
lake shore where it was sectioned at 1-cm intervals by upward extrusion into a sampling
tray fitted to the top of the core barrel. Samples were transferred to labeled Whirl-pakTM
bags for transport to the laboratory. Consolidated deposits from deeper sections in the
profile were extruded in the boat, photographed, and sealed in plastic wrap and aluminum
foil. Labelled core sections were stored in lengths of PVC pipe for transport to the
Two epilimnetic water samples and two deepwater samples were collected in
QorpakTM bottles at the coring site. The latter were collected by free diving and filling the
bottles near the sediment surface. Two near-surface water samples were also taken near El
Remate, at the extreme east end of the northern basin of Lake Peten-Itza. Additionally,
rainwater was collected in Flores on several dates in July 1993 and samples were stored in
labelled QorpakTM bottles. Lakewater and rainwater samples were returned to the lab and
stored at 4C prior to analysis.
In the laboratory, core sections retrieved with the square-rod corer were split
lengthwise, described, and photographed again. Half the core was sliced into 1-cm
intervals and further subdivided for various analyses. Subsamples were stored in labelled
20-ml scintillation vials. Subsamples for 210pb dating, elemental analyses, organic matter
and carbonate content were oven-dried and ground to a fine powder with a mortar and
pestle. The remaining half core was used to measure magnetic susceptibility.
For 210Pb dating, adjacent dried samples from 2-cm intervals in the sediment/water
interface core were combined proportional to their densities. Each composite sample was
loaded into a plastic Sarstedt tube to a measured height of ~30 mm, weighed and sealed
with epoxy glue. Prior to radiometric counting, samples were allowed to sit for >3 weeks
to establish equilibrium between in situ 226Ra and measured 214Bi. Thereafter, isotopic
activities (210Pb, 214Bi, and 137Cs) in all 2-cm samples from the sediment surface to a
depth of 50 cm in the core were analyzed by direct gamma counting (Appleby et al. 1986).
Activities were measured with an ORTEC Intrinsic Germanium Detector connected to a
4096-channel, multichannel analyzer (Schelske et al. 1994). 214Bi was measured as a
proxy for supported 210Pb activity and is reported as 226Ra activity. Unsupported 210Pb
activity was calculated by subtracting supported 210Pb activity from total 210Pb activity on
a level-by-level basis. Age/depth relations in the top 32 cm of the core were calculated
from the stratigraphic distribution of unsupported 210Pb activity, using the CRS (Constant
Rate of Supply) model (Appleby & Oldfield 1978).
Radiocarbon dating of samples from the Peten-Itza core was done by Accelerator
Mass Spectrometry (AMS) at the National Ocean Sciences AMS Facility at Woods Hole
Oceanographic Institution. Eight samples of terrestrial organic matter (wood or charcoal)
were dated. Additionally, three samples of mollusc shells were submitted for 14(2 dating to
evaluate the magnitude of hard-water-lake error (Deevey & Stuiver 1964). Radiocarbon
dates were converted to calendar ages using CALIB, with a 100-year moving average of
the tree ring calibration data set (Stuiver & Becker 1993, Stuiver & Reimer 1993).
Organic matter concentration in sediments was measured by weight loss on ignition
at 550C for two hours in a muffle furnace (Hakanson & Jansson 1983). Inorganic
(carbonate) carbon (IC) in the sediments was measured coulometrically (Engleman et al.
1985) with a UIC/Coulometrics Model 5011 coulometer and a System 140 preparation line
that used 2N HC1 to evolve CO2. Cations (Ca, Mg, Fe, K, Na) and total phosphorus in
sediments were measured by inductively coupled plasma spectroscopy (Jarrell-Ash Model
9000) following combustion at 550C for 2 hours and digestion in 1N HCL for 1 hour
(Andersen 1976). Cations (Ca, Mg, Fe, K, Na) and chloride concentrations in lake water
were measured on the ICP. Sulfate concentration in lake water was measured
turbidimetrically (APHA 1992) on a Perkin Elmer Lambda 2 spectrometer.
Samples for pollen analysis were collected at 10-cm intervals and processed for
counting using acetolysis (Faegri & Iversen 1975). Samples were counted routinely at
400x magnification and at 1000lx with oil immersion when higher magnification was
necessary. Between 100 and 200 grains were counted at most levels. Pollen types were
assigned to four categories (high forest, pine, montane, and disturbance) to assess general
changes in vegetation through time. Deposits below 510 cm contained no pollen.
Gastropod and ostracod samples for stable isotope analysis were isolated from 1-
cm sediment subsamples by disaggregation with 3% H202, followed by sieving at 250 [tm
to remove clay and silt. Samples were filtered and then dried at 60C. Microfossils were
separated from the matrix by using a dissecting microscope at lOx magnification. Three
stable isotopic records were constructed, including two based on gastropods (Pyrgophorus
sp. and Cochliopina sp.), and one that used results from two ostracod species
(Cytheridella ilosvayi and Candona sp.). The two ostracod records were spliced together
at 433 cm because of a taxonomic change in the ostracod assemblage. Single gastropods
and recognizable fragments were picked from the >1000 Ltm fraction. Ostracod valves of
adult specimens of C. ilosvayi were picked from the 425 to 500 ptm fraction while adult
specimens of Candona sp. were picked from the 300 to 351 mrn fraction. Ostracod valves
were cleaned of organic material using 15% H202, then washed with deionized water,
rinsed in methanol, and dried. Gastropods were cleaned in the same manner, except that
specimens were additionally cleaned by sonication. Multiple ostracod specimens from each
1-cm sample were loaded into a stainless steel carrying boat, crushed with a glass rod
dipped in methanol and allowed to dry overnight. Numerous ostracod carapaces were used
to constitute a sample in order to reduce variance as a consequence of running single, short-
lived individuals whose isotopic signature might reflect short-term (i.e. seasonal) water-
column temperature and salinity conditions (Heaton et al. 1995). Multiple (-5) gastropod
shells were ground to a fine powder with a mortar and pestle and a portion of the
pulverized aragonite was loaded into stainless steel carrying boats for isotopic analysis.
Multiple individuals, each living approximately one year, were used to average short-term
(i.e. yearly) changes in lake conditions.
All carbonate samples for isotopic analysis were reacted in a common acid bath of
100% ortho-phosphoric acid (specific gravity = 1.92) at 90C using a VG Isocarb
preparation system. Isotopic ratios of purified CO2 gas were measured on-line by a triple-
collector VG Prism II mass spectrometer. All carbonate isotopic results are reported in
standard delta notation relative to the VPDB standard. Analytical precision was estimated
by routinely measuring a powdered carbonate standard (Carrara Marble-UF) along with
samples from Peten-Itza. Precision (1 SD) was 0.10%c for 6180 and 0.06 for 613C (n
= 355 standards). All isotopic data were smoothed by using a 5-pt running mean to better
illustrate long-term variation.
Lakewater and rainwater samples for isotopic analysis were equilibrated with pure
CO2 gas in evacuated VacutainerT tubes (Socki et al. 1992). Equilibrated CO2 was
distilled from water and non-condensable gases off-line. Next, the purified CO2 gas was
sealed in glass breakseal tubes and transferred to the VG Prism II mass spectrometer for
analysis. Oxygen isotopic results for waters are reported in delta notation relative to
Standard Mean Ocean Water (VSMOW). Analytical precision (1 SD) was -0.09%o based
on routine measurement of an internal standard (UF-alpha).
Samples for magnetic susceptibility and other magnetic parameters were taken at
contiguous intervals in the core below 50 cm using 8-cm3 plastic sampling boxes
paleomagneticic cubes") that removed a -2.5-cm-long section of sediment. Magnetic
susceptibility (k) was measured with a Bartington Instruments magnetic susceptibility meter
(Thompson & Oldfield 1986, Balser 1995).
In 1994, calcium and sulfate were the dominant ions in Lake Peten-Itza, but were
followed closely in concentration by magnesium and bicarbonate (Table 3-1). Lake Peten-
Itza waters had a total ionic content of 12.22 meq 1-1 in 1994. Samples from a 1969
expedition showed that magnesium slightly exceeded calcium in Peten-Itza waters and that
the lake was fresher than at present, with a total ionic concentration of 8.86 meq 1-1
(Brezonik & Fox 1974). Deevey et al. (1980) characterized Peten-Itza as a calcium sulfate
lake and reported a total ionic concentration of 10.45 meq 1-1 based on analysis of a sample
collected in May 1976.
The 8180 values of five rainwater samples collected between 2 and 6 July ranged
from -2.03 to +0.09%o (mean = -0.98%o) (Table 3-1). Oxygen isotopic ratios of +2.84%o,
+2.54%9, and +2.47%c (mean = +2.62%o) were measured on surface water from El
Remate, surface water from the core site, and deep water from the core site, respectively
(Table 3-1). Mean lake water was enriched in 180 by evaporation and was 3.6%o greater
than mean rainwater. With the exception of Lake Yaxha (+2.33%0), waters of the other six
Peten lakes studied in 1993 had greater 5180 values than Peten-Itza (Petenxil +2.71%o,
Quexil +3.45%o, Paxcaman +3.21%o, Macanche +3.08%o, Salpeten +3.52%o, and Sacnab
Table 3-1. Mean ionic concentrations (n = 6) and oxygen isotopic composition (6180) of
Lake Peten-Itza water (n = 3) and Flores rainwater (n = 5). Two lake water samples were
analyzed for sulfate. Coefficient of variation for major ions Ca2+, Mg2+, and S042- was
<3%. Bicarbonate was calculated as the sum of cation charges minus the sum of chloride
and sulfate charges. 8180 determined for a surface water sample from El Remate and a
surface water and a deep water sample from the core site. All samples collected in July
Ca2+ 3.68 meq 1-1
Mg2+ 1.88 meq 1-1
K+ 0.10 meq 1-1
Na+ 0.45 meq 1-1
Cl- 0.32 meq 1-1
S042- 3.05 meq 1-1
HC03- 2.74 meq 1-1
Total 12.22 meq 1-1
6180 +2.62%c (VSMOW)
6180 -0.98%o (VSMOW)
Total 210Pb activity in the Peten-Itza core declined from a high at 2-4 cm
(9.28+0.66 dpm g-1) to supported levels at 34 cm depth in the profile (Figure 3-2a).
Radium-226 activity was consistently low over the length of the core (<1 dpm g-1),
indicating that nearly all 210Pb activity was unsupported. The oldest 210Pb-datable
sediments in the core were from 32 cm depth (Figure 3-2b) and were deposited about 1854
AD (66 yr). The total residual 210Pb activity was only 13.6 dpm cm-2, from which a
Isotope activity (dpm/g)
0 2 4 6 8 10
Age (years before 1993)
0 50 100 150 200 250
Figure 3-2A Total 210Pb, 226Ra, and 137Cs activities (dpm g-1) in the Lake Peten-Itza
core. 3-2B Age/depth plot for the Peten-Itza core based on application of the CRS dating
model. Error bars indicate 1 standard deviation about the mean.
relatively low 210Pb fallout rate of 0.42 dpm cm-2 yr-1 was calculated. 137Cs activity was
very low (<0.16 dpm g-l) below 20 cm depth (i.e., prior to 1952). Thereafter, cesium
activity increased slightly, but remained between 0.37 and 1.00 dpm g-1 between 1952 and
1993. Within the 20th century, mass sediment accumulation rate at the coring site ranged
from 17 to 52 mg cm-2 yr-1, and there was no evidence for a hiatus in sedimentation.
All eight wood/charcoal samples submitted for AMS dating yielded radiocarbon
ages that were in stratigraphic order (Table 3-2, Figure 3-3). Likewise, dates on mollusc
shells were ordered, but were consistently older than dates on terrestrial organic matter or
ages determined by 210Pb dating. According to 21Pb dates, the 14C age of material at 28-
30 cm depth should have been about 70 radiocarbon years BP (14C yr BP), but shell
fragments yielded a date of 98530 14C yr BP, more than 900 years older. The shell date
at 28-30 cm was even older than a date on charcoal taken about 45 cm deeper in the core
(Table 3-2). Similarly, the AMS radiocarbon date on molluscs from 506 cm (1025050
14C yr BP) was 1770 years older than a wood sample at 505 cm depth (848055 14C yr
BP). Because of the apparent effect of hard-water-lake error (Deevey & Stuiver 1964) on
carbonate material, I relied solely on dates from terrestrial organic matter to develop the
chronology for the Peten-Itza core.
210Pb data from the Peten-Itza core indicated that sediments at 30 cm depth in the
core were deposited about 1875 AD (equivalent to about 75 yr BP) (Figure 3-2). I
eliminated from consideration the 14C date on wood from 57 cm (7525 14C yr BP)
because it appeared too young when compared with results of 210Pb dating. The large
piece of wood at 57 cm may have been displaced downward through the unconsolidated
uppermost sediments during the coring process.
The age/depth relation for the Lake Peten-Itza core was derived by interpolating
between AMS 14C dates on wood samples, assuming linear sedimentation rates between
dated horizons (Table 3-2, Figure 3-3). Ages above the dated sample at 73-76 cm were
determined by interpolation between the surface (1994, equivalent to -44 l4C yr BP) and
Table 3-2. AMS radiocarbon dates, calibrated ages and mean linear sedimentation rates for
samples from Lake Peten-Itza Core 6-VII-93. All radiocarbon ages are adjusted to a 813C
value of -25%o. Sedimentation rates between wood/charcoal dates were computed using
the calibrated ages and represent the mean rate between the level for which they are
indicated and the next dated level higher in the core. For sedimentation rate calculations,
the wood date at 57 cm was not used (see text) and the most recent value represents the
mean rate between 73-76 cm and the sediment surface.
0 100 200 300 400 500 600
Depth in sediment (cm)
Figure 3-3. Radiocarbon ages versus depth for terrestrial wood/charcoal and aquatic shell
samples in the Lake Peten-Itza core. The core chronology was established by linear
interpolation between AMS 14C dates from seven terrestrial wood/charcoal samples (see
text). Hard-water-lake error (Deevey & Stuiver 1964) makes dates on gastropod shells
systematically older than terrestrial organic matter from the same depth. Dating error (l)
on all samples ranged from 25 to 80 years and is within the plot symbols.
815 14C yr BP. The age at the base of the core was estimated to be -9120 14C yr BP based
on extrapolation of the linear sedimentation rate between the two bottommost dated
horizons. Linear sedimentation rates varied over the length of the core between 0.016 and
0.111 cm yr-1 (Table 3-2).
Pollen grains were not found in clay-rich sediments older than 8600 14C yr BP
(Figures 3-4a & 3-4b). At -8600 14C yr BP, the pollen spectrum indicates the presence of
well-established high forest. High forest species dominated the pollen record, accounting
for >74% of the grains from 8600 to 5500 14C yr BP (Islebe et al. 1996b). There was a
hint of forest removal as early as -5780 14C yr BP, but disturbance-type pollen grains
declined in abundance during the following millennium. Beginning about 4200 years ago,
the relative percentage of disturbance taxa and pine increased. Trees that were the source of
pine pollen may have been located in the same sites where they exist today, in the Maya
Mountains of Belize or near Poptun, southeast of Flores.
High forest pollen types represented <21% of the total grains from 2300 to 1000
14C yr BP, roughly corresponding to the period from the late Preclassic through the
Classic. Disturbance-type taxa reached a maximum at 1186 14C yr BP (-880 AD), about
the time of the Classic Maya collapse. Forest recovery was evident in the record (by -1000
14C yr BP; -1025 AD) from a decrease in disturbance pollen types and an increase in high-
forest grains. Nevertheless, during the last millennium of the record, high forest pollen
types were not as prevalent as in the early Holocene part of the record.
The core terminated on stiff, indurated, non-carbonate plastics, ca. 9120 14C yr BP.
Sediments older than -8700 14C yr BP (below -520 cm) are likely composed of
montmorillonite clays, the principal product of local bedrock weathering. These plastics are
indicated by relatively low concentrations of IC and Ca, and relatively high concentrations
of Fe and K (Figures 3-4b & 3-5). By -8700 14C yr BP, carbonates dominated the
I~U~U~ It!S U Iipuu~ e e. 11esipxee
% % % % % %u% e
x% % % %
%% % %% % %% % %% % % 19 19%
pu/% % % I jap/y /%qu%% Iv 13/P
%% %%% % %%uf%%%2 s s\ \ \ \ sv;
%%%/% % % % % % % % / ///// ///// %%%% f ///s
%%%%%%%%%%%%%% / // %/ % %%/%%%%% // % %% /
%%%%%%%%%%%h%%\% \\ \\\ \\ i
%% % % % % % % % % % \ \\%\\ % \\\%\ % %%'^^^
e I % xI %x% xe/I
ex.,% % xx% eexe% ..
%%%%%% I%%eee%% eex
% % % % % % % % % %
::.%x x% x % x%x%e%x 6%
0 O 0 0
C0 0 0 0\
(dfl sVaSC uoqjevoupBi) a3V
Figure 3-4A Relative abundance of various pollen types (high forest, montane, pine, and
disturbance) versus radiocarbon age in the Peten-Itza sediment core. 3-4B Sediment
composition as organic matter (LOI), calcium carbonate (figured as 2.5 x Ca
concentration), Fe2O3 (computed from iron content), and aluminum-silicates (figured as the
balance of the sediment matrix). 3-4C Magnetic susceptibility of sediment versus
radiocarbon age in the Peten-Itza core.
0 0 0S 0 0a 0 0P 0 0
r-4 r M IVt kn 110 t- aM C
(d9f sjvaS uoqxeaoipLH) a2V
Figure 3-5. Organic matter content (% LOI), inorganic (carbonate) carbon content (% IC),
and selected elemental concentrations (mg g-1 Ca, Mg, Fe, K, Na, and P) versus
radiocarbon age in the Peten-Itza sediment core.
sediment matrix, as expressed by a rise in Ca and IC concentrations. IC and Ca
concentrations are correlated throughout the profile, suggesting that both reflect CaCO3 in
the sediment matrix. CaCO3 constitutes >50% of the sediment mass over most of the core.
Mg also appears to be correlated with IC and Ca from -8800 to -7000 14C yr BP,
suggesting that some Mg may be carbonate-bound. Nevertheless, the presence of Mg in
carbonate-free basal deposits indicates that some Mg may be associated with clays. Mg
generally constitutes <1% of the sediment mass.
Carbonate content decreased by -6500 14C yr BP, as organic matter and clays (Fe
and K) began to constitute greater proportions of the sediment matrix. After -5500 14C yr
BP, carbonates again dominated the sediment matrix, with drops in both organic matter and
clays. Between -3000 and -1000 14C yr BP, the period of archaeologically-documented
human disturbance and palynologically-identified vegetation removal, inorganic clays again
constituted a larger fraction of the sediment matrix. Sediments that accumulated during the
last millennium are rich in carbonate, and very recent deposits, having not yet undergone
diagenesis, displayed relatively high (>35%) organic matter content.
Carbonate Fossils and Stable Isotope Geochemistry
Six ostracod species were found in appreciable numbers in the Peten-Itza core.
They included Candona sp., Cypridopsis vidua, Physocypria sp., Heterocypris punctata,
Limuocythere sp., and Cytheridella ilosvayi. With the exception of C. ilosvayi, which
dominated in nearly all sediments deposited after -7210 14C yr BP, other ostracod taxa
were found primarily in early Holocene deposits. Two gastropod taxa, Pyrgophorus sp.
and Cochliopina sp., were found at nearly all depths in the profile.
All three oxygen isotopic records show the same pattern of long-term change
(Figure 3-6). From -9000 to -6750 14C yr BP, 5180 of Cochliopina sp. averaged 2.0%o.
Over the same period Pyrgophorus sp. also averaged 2.0%o. From -8300 to 7210 14C yr
BP, Candona sp. were abundant, with 6180 values that averaged 2.6%o. Before -7210
14C yr BP, C. ilosvayi were absent, but they averaged 2.1%o from -7210 to -6750 14C yr
6180 (%o, PDB)
Cochliopinasp. Pyrgophorus sp.
0 1 2 0 1 2 3 0 1 2 3
Figure 3-6. Oxygen isotopic composition in snail shells (Cochliopina sp. and Pyrgophorus
sp.) and ostracod valves (Cytheridella ilosvayi and Candona sp.) versus radiocarbon age in
the Peten-Itza core. Because of a taxonomic change it was necessary to measure two
ostracod species. C. ilosvayi is represented by a line without plot symbols while Candona
sp. is represented by circles. All data have been smoothed with a 5-pt running mean to
better illustrate long-term trends. Carbonate isotopic data are available from the National
Geophysical Data Center, Boulder, Colorado, at email@example.com.
BP. At -6750 14C yr BP all three records show an initial marked decrease in 6180 values
followed by a more gradual decrease that continued until -5000 14C yr BP. From -5000
14C yr BP until the present, oxygen isotopic records for all three taxa show no major
changes. Over the last five millennia, Pyrgophorus sp. averaged 0.38%o, Cochliopina sp.
averaged 0.36%o, and C. ilosvayi averaged 0.86%o, with standard deviations of +0.28%o,
0.23%o, and +0.27%o, respectively.
Calculated 8180 values for aragonite and calcite, precipitated in equilibrium with
modem lake water (-18Owater = +2.6%o) at a mean annual temperature of 25C, yield
values of 0.92%o and 0.78%o, respectively (Craig 1965, Grossman & Ku 1981). These
theoretical values are close to mean values (0.85%o, 0.47%o, and 0.39%o) for recent
samples of Cochliopina sp., Pyrgophorus sp., and C. ilosvayi, respectively. This
suggests that all three taxa precipitate carbonate near equilibrium with the water in which
The carbon isotopic records of gastropods Cochliopina sp. and Pyrgophorus sp.
are very similar to one another (Figure 3-7). At -9000 14C yr BP, carbon isotopic values
of gastropods were low. From -9000 until -8300 14C yr BP, carbon isotopic values
gradually increased and then remained relatively constant until -7500 14C yr BP.
Thereafter they began to gradually decrease and reached the lowest values observed in each
record at -6500 14C yr BP. From -6500 to -5600 14C yr BP, carbon isotopic values
increased -3%o. From -5600 14C yr BP to the present, 613C values remained rather high
with modest, short-term decreases centered on -2500 and -1400 14C yr BP. Carbon
isotopic values in topmost sediments are more negative than at any time since ca. 6000 14C
The carbon isotopic record of Candona sp. is different from the gastropod records.
Candona sp. first appeared at -8300 14C yr BP with relatively low values of --6%o.
Values decreased to --10%o at -7875 14C yr BP and then increased again to --5%o at -7310
14C yr BP. The 613C record of C ilosvayi begins -7210 14C yr BP and shows trends
513C (%o, PDB)
Figure 3-7. Carbon isotopic composition in snail shells (Cochliopina sp. and Pyrgophorus
sp.) and ostracod valves (Cytheridella ilosvayi and Candona sp.) versus radiocarbon age in
the Peten-Itza core. Because of a taxonomic change it was necessary to measure two
ostracod species. C. ilosvayi is represented by a line without plot symbols while Candona
sp. is represented by circles. All data have been smoothed with a 5-pt running mean to
better illustrate long-term variations in 513C.
-5 -4 -3 -2 -4 -3 -2 -1
-8 -6 -4
I I I
similar to the gastropod records. After -6500 14(2 yr BP, 813C values gradually increased
until -5600 14C yr BP, after which they remained fairly constant to the present.
Magnetic susceptibility (k) of Peten-Itza sediments is weak, with low and
occasionally negative values (Figure 3-4c). Stiff, basal deposits yielded the highest values
for k, implying relatively high concentrations of magnetic minerals. Following high
susceptibility values at the bottom of the core, k was relatively low until -6800 14C yr BP.
From -6800 to -4700 14C yr BP, magnetic susceptibility was moderately high. Between
-4700 and -2800 14C yr BP magnetic susceptibility was again low. High values for k
occurred during the period between -2800 and -1100 14C yr BP. Magnetic susceptibility
returned to low levels at about -1100 14C yr BP and remained low to the present.
Proxies of environmental change
Because we cannot directly measure past climatic variables and assess human
impact on the environment, we rely on natural materials (proxies) that record information
about historical processes to infer past environmental change. Each proxy, including those
used in this study, is controlled by a host of environmental factors that must be understood
before past conditions can be reconstructed. For example, pollen preserved in lake
sediments is controlled by regional and local vegetation, that in turn, is influenced by both
climatic and non-climatic factors, including temperature, rainfall, seasonality, edaphic (soil)
conditions, nutrient availability and human impact. Lake sediment geochemistry is
controlled by processes including lake productivity, transport of materials into the lake, and
post-depositional diagenesis. Oxygen isotopic ratios of shell carbonate are a function of
several factors, including temperature, the oxygen isotopic composition of lake water, and
vital effects. In turn, the 5180 of lake water is dependent upon the ratio of evaporation to
precipitation (E/P), the 6180 of precipitation, and the effect of lake basin morphology on
the hydrologic budget (Fontes & Gonfiantini 1967, Covich & Stuiver 1974, Gasse et al.
1990, Talbot 1990, Lister et al. 1991). Previous work in Neotropical, closed-basin lakes
has shown that, during the Holocene, changes in the 6180 of shell material were controlled
primarily by the ratio of E/P (Hodell et al. 1991, Hodell et al. 1995, Curtis et al. 1996)
because of relatively stable temperatures and lack of large variations in 6180 of source
water (Curtis & Hodell 1993).
Carbon isotopes in biogenic carbonate reflect the 813C ratio of lakewater DIC,
which is controlled by several factors including lake primary productivity, 813C of
atmospheric CO2, methanogenesis, and 813C of dissolved bicarbonate from the watershed.
The latter can be affected by changes in the relative abundance of C-3 (--25%o) and C-4 (--
12%o) plants in the catchment. Lastly, magnetic susceptibility of lacustrine sediments can
be influenced by the type of material being eroded from the watershed, its rate of erosion,
and post-depositional diagenetic processes (Hawthorne & McKenzie 1993).
Different proxies may be controlled by common environmental variables, whereas
other variables may not be shared. If all proxies display consistent patterns of change, then
it is likely that a common environmental variable is involved. If proxies indicate different
patterns of change in the past, then it is probable that the proxies were responding to
different variables, or that one proxy failed to respond as expected to a "common" variable.
In such cases, inferring past environmental change becomes more challenging.
Earliest Holocene (>9000 14C yr BP)
During the earliest Holocene, prior to -9000 14C yr BP, all proxies in the Lake
Peten-Itza record suggest relatively dry conditions. The Peten-Itza core terminated at 545
cm in very stiff sediments with high concentrations of Fe and K, and low CaCO3 content,
indicative of a clay-rich paleosol (Figure 3-5). Basal deposits show relatively high
magnetic susceptibility (Figure 3-4c). No lacustrine microfossils or pollen were found in
these bottom sediments suggesting they are non-lacustrine. Many lowland, Neotropical
and subtropical lakes, including Peten lakes Salpeten and Quexil, also provide evidence for
lowered water tables and aridity during the latest Pleistocene and earliest Holocene (Watts
1975, Bradbury et al. 1981, Deevey et al. 1983, Leyden 1984, 1985, 1987, Markgraf
1989, Bush & Colinvaux 1990, Pipemo et al. 1990, Hodell et al. 1991, Bush et al. 1992,
Watts et al. 1992, Curtis & Hodell 1993, Leyden et al. 1993, 1994, Watts & Hansen 1994,
Islebe et al. 1995).
Although the coring site in the southern basin of Lake Peten-Itza was apparently dry
when the basal paleosol formed, the northern basin probably contained water owing to its
greater depth. This hypothesis is supported by the fact that nearby Lake Quexil held water
in its deep basin (presently >30 m) even during the arid late Pleistocene (Leyden 1984,
Leyden et al. 1993, 1994). Because Peten-Itza's main basin may exceed 60 m depth
(Deevey et al. 1983), deepwater sites in Peten-Itza may contain long records extending well
into the Pleistocene.
Early Holocene (-9000-7300 14C yr BP)
The Peten-Itza coring site was covered with water by -9000 14C yr BP. Aquatic
gastropods first appeared in the record with the inception of lacustrine sedimentation at
-9000 14C yr BP, and ostracods first occurred in the record at -8920 14C yr BP.
Additionally, the alga Botryococcus is first observed at -8600 14C years BP. Sediments
were dominated by CaCO3 and possessed low clay concentrations, suggesting that the
coring site was under water by -9000 14C yr BP.
The rise in lake water level by -9000 14C yr BP in Peten-Itza is consistent with the
rise of water in Lake Quexil, where a core taken in -6 m of water had a bottommost date on
wood of 8410180 14C yr BP (Ogden & Hart 1977, Vaughan et al. 1985). The filling of
lakes on the Yucatan Peninsula, Mexico, occurred about the same time as the lake level rise
in the Peten. Lake Chichancanab first filled at -8200 14C yr BP (Hodell et al. 1995) and
Lake Coba filled before -7600 14C yr BP (Whitmore et al. 1996). San Jose Chulchaca
filled prior to -7230 14C yr BP (Leyden et al. 1996). Filling of previously dry lake basins
throughout the lowland Neotropics and subtropics is attributed to increased moisture
availability and rising sea level that raised the level of freshwater aquifers in karst regions
(Watts 1969, Fairbanks 1989, Watts & Hansen 1994).
Once water covered the Peten-Itza coring site, lacustrine sediments preserved both
pollen grains and aquatic microfossils. Pollen data and 6180 analysis on shell material
from the Peten-Itza core yield apparently contradictory paleoclimatic inferences for the early
Holocene. The pollen spectrum in the early Holocene is dominated by grains of the
Moraceae-Urticaceae group, indicating that widespread tropical forest had been established
by -8560 14C yr BP (Figure 3-4a). The abundance of high forest taxa suggests moist early
Holocene conditions (Islebe et al. 1996b). In contrast, the relatively high oxygen isotope
values for both ostracods and gastropods (Figure 3-6) suggest that climate was relatively
dry from -9000 to -6800 14C yr BP. This interpretation assumes that 6180 is controlled
mainly by E/P, which may not be valid, as discussed below.
Early Holocene pollen and 8180 records from other lowland Neotropical lakes
reveal trends similar to those from the Peten-Itza core, suggesting that neither the pollen
record nor the 6180 record from Peten-Itza is anomalous. For instance, Leyden et al.
(1993) found that following the arid late Glacial episode, lowland forest was first
established in the Peten -9000 14C yr BP. Vaughan et al. (1985) found lowland forest
pollen in the oldest samples of the Quexil shallow-water core dated at 8410180 14C yr
BP. In northern Venezuela, the Lake Valencia watershed first became forested at -9000
14C yr BP, following the period of late Pleistocene aridity (Bradbury et al. 1981). Oxygen
isotopic values from Lake Valencia were high until -8400 14C yr BP and then declined
(Curtis et al. in prep). Oxygen isotope values from Lake Chichancanab, Yucatan (Mexico),
were high during the initial filling phase at -7800, which lasted until -7200 14C yr BP
(Hodell et al. 1995).
During the early Holocene, conditions in the Peten were moist, as inferred from the
presence of palynologically-documented, lowland tropical forest around Peten-Itza and
other Peten lakes (Vaughan et al. 1985, Leyden et al. 1993). Further support for moist
conditions in Peten during the early Holocene comes from a 6180 record on bulk sediment
from a deep-water core from Lake Quexil (Leyden et al. 1993). The Lake Quexil record
preserves a continuous record of 6180 variation from marine oxygen isotope stage 3 to
stage 1. The lowest 6180 values of the entire record occur near the Pleistocene/Holocene
boundary and may indicate increased moisture availability in the Peten at the end of the last
Ice Age (Leyden et al. 1993).
Several possible explanations for the high 6180 values of shell carbonate from
Peten-Itza during the early Holocene moist period include: 1) 8180 values of precipitation
were high in the early Holocene; 2) temperatures were low in the early Holocene; and 3) the
hydrologic budget of the southern basin of Peten-Itza exerted greater control on lake water
8180 than did regional E/P.
Historical changes in the 6180 of precipitation affect the 8180 of lake water and
lacustrine shell material. The 6180 of rainfall is determined by the water source and the
history of water vapor transport (distance from source, elevation, etc.). There is no reason
to suspect that any of these factors would have been dramatically different during the early
Holocene as compared to today. Lacking evidence for past shifts in the 8180 of
precipitation, I cannot evaluate its impact on the 6180 of lake water during the early
I calculated the degree of cooling, relative to present, that would be required to
account for early Holocene 8180 values, assuming that the early Holocene isotope increase
was solely attributable to low temperatures. I used the paleotemperature equations for
aragonite and calcite (Grossman & Ku 1981, Craig 1965), with early Holocene mean 8180
for shell carbonate aragonitee = 2.01%c, calcite = 2.60%o), and modern 8180water (2.6%o)
as independent variables. The resulting paleotemperature estimates suggest that the early
Holocene was -7C cooler than today. Such cool temperatures during the early Holocene
are highly unlikely because they are equivalent to temperature depression estimates for the
last Ice Age (i.e., 6.5 to 8.0C below present) based on palynological evidence (Leyden et
al. 1993). Slightly cooler temperatures may account, in part, for higher 6180 values of
shell carbonate during the early Holocene, but pollen results argue against substantial
cooling. For instance, had early Holocene temperatures been much cooler, one might
expect to see greater abundance of montane taxa such as juniper and oak.
The hydrology and morphology of lake basins can also influence the 6180 of lake
water. For example, based on the summer 1993 survey, inter-lake, surface-water 6180
values for Peten waterbodies vary by 1.5%o even though regional climate is similar and
basins presumably receive direct rainfall and runoff with the same 6180. Oxygen isotopic
differences among lakes probably reflect varying fractions of the lake volumes that are lost
to evaporation, the degree to which the various lakes are hydrologically "closed," and the
watershed/lake ratios of the basins. Within a single basin, change through time in the
relative amount of lake volume lost to evaporation is probably the most important factor
influencing lake water 6180. As lake water level rises or falls, the hypsometry and
hydrologic budget of the basin change. For instance, shallow, flat-bottomed lakes have
higher surface area to volume ratios than do deep, conically-shaped basins. In the early
stages of basin filling, the 8180 of lake water may be high because the waterbody has a
high surface area/volume ratio and loses a greater proportion of its volume to evaporation
each year than when the lake is full.
Shallow-water conditions at the Peten-Itza coring site are inferred for the early
Holocene based on an ostracod assemblage consisting of Candona sp., Cypridopsis vidua,
Physocypria sp., Heterocypris punctata, and Limnocythere sp. These species dominate
from -8900 to -7210 14C yr BP and suggest a wetland with standing or slow-flowing
water (R. Forester, pers. commun.). This shallow wetland, with a high surface
area/volume ratio, would have lost a great proportion of its annual water budget to
evaporation. As a result, 6180 values of water and ostracods would have been high during
the early stages of filling. Low lake levels and associated high surface area/volume ratios
are also inferred for Lakes Chichancanab (Hodell et al. 1995) and Valencia (Curtis et al. in
prep) where high 6180 values are observed in early Holocene shell remains.
During the early Holocene, climate conditions in Peten appear to have been wet
enough to support a Moraceae-Urticaceae lowland forest based on the pollen assemblage,
but the hydrologic balance of the smaller, early Holocene water body may have resulted in
the relatively high 6180 values observed in the Peten-Itza core.
Carbon isotopic ratios of microfossil carbonate were probably governed by several
factors during the early Holocene, making it difficult to interpret the 813C record. For
instance, as lake level rose, changing hydrologic conditions would have impacted lacustrine
productivity. Additionally, changing morphometry of the basin may have influenced the
relative proportion of dissolved inorganic carbon (DIC) coming from dissolution of local
limestone versus equilibration with atmospheric CO2. The rapid rise in 613C values of
gastropods at the inception of the record may reflect an increasing contribution of bedrock-
derived DIC as lake stage increased. Although both snail species display nearly identical
trends during the first -2 millennia of the record, the ostracod (Candona sp.) 813C record
differs markedly. This discrepancy may reflect the fact that the infaunal ostracods and
littoral gastropods have different habitat preferences, and therefore utilize different DIC
pools for shell construction. Specifically, the ostracods may have sampled 13C-depleted
DIC produced from oxidation of organic carbon in sedimentary pore waters.
Mid-Holocene (7300-4800 14C yr BP)
During the middle Holocene, several geochemical (Fe, K; Figures 3-4b & 3-5) and
magnetic susceptibility (Figure 3-4c) proxies gradually increase to moderately high values
and then decrease to previous levels. Simultaneously, CaCO3 percentages decrease before
returning to former high values (Figures 3-4b & 3-5). These changes began between
-7300 and -6800 14C yr BP with CaCO3 concentrations increasing until -6000 14C yr BP,
and then declining again by -4800 14C yr BP. The most plausible explanation may have
been increased and then decreased delivery of eroded soils to the lake. Rapid erosion of
soil particles to the lake both increased the concentrations of clay-bound Fe and K. Rapid
input of allochthonous clays increased bulk sedimentation rate, thereby diluting CaCO3
concentration. These erosional deposits display relatively high magnetic susceptibility.
Mean linear sediment rate reached its maximum value during the middle Holocene (0.111
cm yr-1; Table 3-2).
Lacustrine production probably increased from -7300 to -4800 14C yr BP as
illustrated by higher %LOI (Figure 3-5). Increased erosion likely caused higher rates of
nutrient loading. Higher primary production is also indicated by greater concentrations of
Botryococcus starting -6600 14C yr BP (Islebe et al. 1996b). Carbon isotopic values of
both gastropod and ostracod shells increased beginning -6500 14C yr BP (Figure 3-7).
Enhanced primary productivity could have led to higher 513C values in lake water DIC. As
primary production increased, a greater amount of 12C would have been preferentially
removed from the photic zone by photosynthesis and buried as organic matter in the
sediments, thereby leaving the DIC pool enriched with 13C.
Beginning -5780 14C yr BP, the pollen record reveals a shift from predominantly
lowland forest tree taxa (i.e. C-3 plants, 613C --25%0) to more open vegetation, including
grasses (i.e. C-4 plants, 813C -12%o). This vegetation change would shift the 513C of
organic matter in the lake catchment toward more positive values. Oxidation of terrestrial
organic matter generates CO2 that possesses an isotopic signal reflecting the source
material, some of which can dissolve into downward percolating ground water. Ground
water that ultimately reaches the lake influences the isotopic ratio of lake water DIC.
Furthermore, oxidation of allochthonous organic matter within the lake produces CO2 with
an isotopic signature that reflects the source material, and this CO2 in turn influences the
isotope ratio of the DIC pool.
The oxygen isotopic signals show a rather steady and irreversible decrease
beginning -6800 14C yr BP (Figure 3-6). The 8180 decrease follows a faunal shift from
an assemblage dominated by Candona sp. to one dominated by C. ilosvayi, suggesting a
change from a wetland environment with low water level to a deeper lake. By -5000 14C
yr BP, 8180 values of shell carbonate had stabilized at lower values. Mid-Holocene
decreases in 8180 have been observed in other lowland, Neotropical lake records. Oxygen
isotopic values decreased abruptly as Lake Chichancanab, Mexico, filled between ~7800
and -7200 14C yr BP (Hodell et al. 1995). In Lake Valencia, Venezuela, 8180 values
declined rapidly beginning at -8500 14C yr BP and continued to decrease gradually for the
next 2000 years (Curtis et al. in prep).
In Lake Peten-Itza, the decrease in 6180 values occurred during the early to middle
Holocene moist period that has been observed elsewhere in the Neotropics (Covich &
Stuiver 1974, Bradbury et al. 1981, Deevey et al. 1983, Leyden 1984, 1985, Piperno et al.
1990, Hodell et al. 1991, 1995, Islebe et al. 1996a) and Africa (Street & Grove 1976,
1979, Street-Perrott & Harrision 1985, Lezine 1989). Hodell et al. (1991) suggested that
increased precipitation in the Neotropics during the early to middle Holocene was a
consequence of greater intensity of the annual cycle, which was driven by changes in
seasonal insolation forced by the Earth's processional cycle. Increased precipitation and
runoff around Peten-Itza during the middle Holocene is inferred from geochemical and
magnetic susceptibility evidence that suggests increased erosion.
Hydrologic connection of the northern and southern basins of Lake Peten-Itza also
may have affected 8180 values of lake water at the coring site during the middle Holocene
(Figure 3-1). At some time during the filling of Peten-Itza, water level in the large,
northern basin exceeded the sill depth of the small, southern basin, thereby connecting the
two portions of the lake. Because the surface area/volume ratio of the northern basin was
much smaller than that for the shallower, southern basin, it is reasonable to assume that
5180 values in the northern side were more negative. Joining the two previously separate
waterbodies would have effectively reduced the 5180 of the southern basin. Furthermore,
the effective volume of the combined basins would have been much greater than that of the
southern basin alone. Thus, the proportion of water lost to evaporation each year from this
large volume of water would have been less than the fraction lost from the previously-
isolated southern basin alone. The large volume of the waterbody serves as a buffer
against evaporative enrichment of 180 in lake waters.
The time required for isotopic equilibrium to be reached following a shift in E/P is a
function of several factors, including lake morphology and hydrology. The decrease in
5180 values took longer in Peten-Itza (-6800 until -5000 14C yr BP) and Valencia (-8500
until -6500 14C yr BP; Curtis et al. in prep.) than in Chichancanab (-7800 until -7200 14C
yr BP; Hodell et al. 1995), probably because the former lakes are significantly larger. The
rate of 6180 change in these three lakes is probably related to the time required for the lakes
to fill. As a lake fills, the 6180 of lake water decreases because the ratio of water lost to
evaporation relative to total volume decreases. Lakes with large volumes (e.g. Valencia
and Peten-Itza) probably required longer to fill and thus achieved isotopic equilibrium more
slowly than small lakes that filled more rapidly (e.g. Chichancanab) (Lister et al. 1991).
In the case of Lake Chichancanab, early Holocene sea level rise may have directly
affected the lake water level by raising the phreatic water table, thereby increasing the lake
volume rapidly. It is, however, unlikely that sea level directly affected Peten-Itza's water
level because the local water table is deep, far below the lake bottom. Likewise, it is
improbable that Lake Valencia's hydrology was influenced directly by sea level rise
because the basin is located -400 m above msl.
A gradual decline of lowland forest beginning as early as -5780 14(2 yr BP is
documented by a reduction in Moraceae-Urticaceae pollen and an increase in open
vegetation taxa (including melastomes, Poaceae, and Byrsonima) (Islebe et al. 1996b).
From pollen data alone, it is not possible to determine whether initial forest loss was caused
by climatic or human-mediated processes. Forest decline as a consequence of climatic
drying is incompatible with an inferred reduction in EFP inferred from decreasing 6180
values. This increase in disturbance taxa during the middle Holocene may indicate early
human impact, which may explain the discrepancy between pollen and oxygen isotope
The gradual decline of high forest appears earlier in the Peten-Itza record compared
with previous studies in Peten that suggested Maya forest clearance began -3000 years ago
(Deevey et al. 1979, Vaughan et al. 1985, Leyden 1987). Dating of initial forest clearance
in previous studies, however, may have been inaccurate because sediment chronologies
relied largely on correlating palynological changes in cores with archaeologically-defmined
time periods. Evidence of earlier human disturbance has been reported from Panama
(Piperno et al. 1990, Bush & Colinvaux 1990) and Costa Rica (Islebe et al. 1996a).
The middle Holocene increase in 613C of shell carbonate preceded the decline in
lowland forest recorded in the pollen record (Figure 3-7). Assuming early vegetation
changes were anthropogenic, it is probable that riparian deforestation would have had a
greater effect on the lake's carbon budget than on the regional pollen rain. Earliest settlers
in the region likely occupied areas near the water, where their swidden agricultural practices
would have immediately affected the watershed carbon pool. Nevertheless, these localized
populations probably did not alter the regional pollen rain for a considerable period because
the rate of human population increase was slow (Deevey et al. 1979).
Late Holocene (4800 14C yr BP to the present)
From -4800 until -2800 14C yr BP the pollen record documents continued gradual
increase of disturbance taxa and decrease of lowland forest taxa (Figure 3-4). During this
same interval, sediment geochemistry and magnetic susceptibility show very little change
(Figures 3-4 & 3-5). Oxygen and carbon isotopes show no long-term changes after -4800
14C yr BP (Figure 3-6 & 3-7).
The pollen record illustrates that -2800 14C yr BP, lowland forest taxa declined
rapidly (Figure 3-4a). This episode of dramatic deforestation around Peten-Itza coincides
with documented forest removal around other Peten lakes, including Petenxil (Cowgill et
al. 1966), Quexil (Deevey 1978, Wiseman 1978, Vaughan et al. 1985), Salpeten (Leyden
1987), Sacnab (Deevey et al. 1979) and Macanche (Vaughan et al. 1985), and with
deforestation in Belize (Hansen 1990).
Forest clearance accelerated soil erosion in the watershed and massive colluviation
produced thick, clay-rich deposits on many Peten lake bottoms. This "Maya clay"
accumulated over a period of about 2600 years (3000-400 BP) and is almost 7 m thick at
deepwater sites in some Peten lakes (Binford et al. 1987). Curiously, the "Maya clay" is
not observed as a distinct stratigraphic unit in the Peten-Itza core. Nevertheless, several
sediment variables indicate the presence of fine, eroded detrital material in sediments
deposited after -2800 14C yr BP. For instance, increases in K and Fe at this time suggest a
greater presence of clays (Figure 3-5). Large increases in magnetic susceptibility are
consistent with more erosion of clastic material into the lake (Figure 3-4c).
The "Maya clay" unit may be less obvious in Lake Peten-Itza for several reasons.
Lake Peten-Itza (A=99.6 km2) is substantially larger than all other lakes in the region.
Yaxha, the next largest basin studied, has a surface area of only 7.4 km2. Even the
southern basin of Peten-Itza is large relative to many other regional lakes, and
allochthonous detrital inputs would be expected to be distributed over a large area.
Erosional inputs appear most clearly in small lakes with large watershed:lake ratios.
Another reason that the "Maya clay" in Peten-Itza is relatively cryptic is that the southern
basin is generally flat-bottomed, so that detrital inorganic sediments were probably
deposited fairly uniformly over much of the basin and diluted by autochthonous sediments.
Thickest colluvial sequences have been measured in cores collected from deep-water sites
in small lakes (Brenner 1983, 1994, Vaughan et al. 1985, Binford et al. 1987). Deep areas
of these cone-shaped basins received large amounts of focused, fine-grained material
(Deevey et al. 1977, Brenner 1983). A third factor that may account for inability to
visually discern the "Maya clay" in the Peten-Itza sequence is that erosional loading around
the lake shore may have been relatively low. Hillslopes along the southern shore are
gentle, and the coring site lies nearly a kilometer from the steeper shores of the Tayasal
Peninsula to the north. Lastly, Peten-Itza lacks overland inflows that might transport fine-
grained suspended loads for considerable distances into the lake.
Erosional inputs to Lake Peten-Itza began to decline about -1100 14(2 yr BP. Clay
constituents K and Fe diminish in concentration while CaCO3 content increases (Figures 3-
4b & 3-5). Likewise, magnetic susceptibility decline is consistent with reduction in the
quantity of eroded, magnetic minerals in the sediment (Figure 3-4c). At -1000 14C yr BP
(- 1025 AD) there is evidence for forest regeneration, as lowland forest taxa increase and
disturbance indicators decline in relative abundance (Figure 3-4a). Changes in
geochemical, magnetic and palynological variables in the Peten-Itza core postdate the
archaeologically-documented collapse of Classic Maya Civilization, ca. 800-900 AD (Lowe
1985). Cultural collapse and associated population decline were most pronounced in the
southern lowlands, and probably led to reduced environmental impact in the Peten region.
The reforestation postdating the Classic Maya collapse may have been a
consequence of reduced pressures on regional vegetation. The pollen record from Peten-
Itza suggests that forest recovery commenced about -1000 14C yr BP (-1025 AD) (Islebe
et al. 1996b). Most previous palynological studies in Peten failed to date the timing of
forest regeneration accurately because chronologies were based largely on correlation with
archaeological time periods (Vaughan et al. 1985). However, a high-resolution pollen
sequence from savanna Lake Chilonche yielded a 14C date on wood, suggesting that
reforestation began after -1600 AD (-350 14C yr BP) following European arrival in Peten
(Brenner et al. 1990). More palynological study of well-dated cores from the region will
be required to resolve this discrepancy.
Paleolimnological study of cores from Lakes Chichancanab and Punta Laguna in
the north-central Yucatan Peninsula (Mexico) provided isotopic (6180) evidence for a
protracted drought that coincided with the Classic Maya collapse (Hodell et al. 1995, Curtis
et al. 1996). The oxygen isotopic records from the Peten-Itza core display no evidence for
a Terminal Classic drought in the southern lowlands (Figure 3-6). Several factors or a
combination of factors may account for apparent climatic disparities between the two areas.
First, it is possible that the drought was local in extent and affected only the more northerly
portion of the Yucatan Peninsula, but did not extend into the Peten lowlands. This scenario
is unlikely given the emerging evidence for dry conditions at the end of the first millennium
A.D. in the central highlands of Mexico (Metcalfe et al.1994, Metcalfe 1995) and Costa
Rica (Horn & Sanford 1992). Abrupt drying -1100 A.D. is also reported from the
Bolivian/Peruvian Andes (Thompson et al. 1985, Ortloff & Kolata 1993, Chepstow-Lusty
et al. 1996, Abbott et al. in press, Binford et al. in press) suggesting teleconnective
linkages between tropical sites north and south of the equator. Second, sampling
resolution of the Peten-Itza core may have been insufficient to record the drought event.
This, however, is unlikely because the mean sample spacing in the Peten-Itza core for the
period from -1500 to 500 14C yr BP is about 15 years. Although lower than the sampling
resolution at Punta Laguna (-7 yr/sample) during the same period, the mean sampling
resolution exceeds that at Chichancanab (-19 yr/sample) where the drought was clearly
evident and was recorded in many samples. The most plausible explanation for the lack of
a drought signal in the Peten-Itza core is that the lake is simply too large (A=99 km2) and
deep (Zmax>60 m) to record climatic changes that persist for less than several centuries. In
comparison, Lakes Punta Laguna (0.9 km2, Zmax >12m) and Chichancanab (10 km2,
Zmax=12.5 m) have sufficiently small volumes that their lakewater 8180 responds quickly
to changes in E/P. Peten-Itza's large volume, and consequent long residence time, make it
relatively insensitive to all but the most dramatic, long-term shifts in E/P. To establish
whether the Terminal Classic drought affected the southern lowlands, where the collapse
was most pronounced, it will be necessary to conduct isotopic studies in smaller Peten
lakes with high sedimentation rates and continuous records of preserved carbonate
The negative shift in the 813C over the last few centuries may have been caused by
forest regrowth following human depopulation of the Peten (Figure 3-7). During this time
lowland forest taxa returned to dominance while grasses declined. Lower abundance of
isotopically heavy grasses around the lake may have caused a decrease in the 813C of
oxidized catchment organic matter. Additionally, a recent decline in primary production
could have contributed to the drop in 813C. Lower production at the top of the core is
suggested by low counts of Botryococcus (Islebe et al. 1996b). These lower rates of
production may have been a consequence of reduced nutrient loading to the lake, which is
supported by evidence of lowered rates of erosion (i.e. low Fe and K in recent deposits).
Overall, the 613C record of gastropods correlates closely with the total carbonate content of
the sediment (Figures 3-4 & 3-7), with high 613C values occurring at depths with high
carbonate concentration. The relationship may suggest that the concentration and 813C of
lakewater DIC is controlled by the amount of bicarbonate reaching the lake that is derived
from dissolution of catchment limestone.
I inferred climatic change and human impact on the environment of Peten,
Guatemala, based on multiple lines of evidence from sediments in Lake Peten-Itza. All
proxies suggest relatively dry conditions during the earliest Holocene (prior to -9000 14C
yr BP). Interpretation of early Holocene (-9000 until -7300 14C yr BP) oxygen isotopic
results in terms of E/P is at odds with the strong evidence for moist conditions based on the
presence of extensive lowland forest. The discrepancy between early Holocene pollen and
8180 results in the Peten-Itza core can be reconciled if lake water 6180 is assumed to have
been controlled by both regional climate and the changing surface area/volume ratio of the
filling lake. During the middle Holocene (-6800 until -4800 14C yr BP), 6180 values
indicate that conditions in Peten became wetter. At about the same time, decreasing
lowland forest and increasing disturbance taxa suggest early human impact on regional
vegetation. Accelerated forest removal by -2800 14C yr BP is attributable to increasing
Maya impact on the environment. The erosional "Maya clay" unit, apparent in many
smaller Peten basins, is not visible in the Peten-Itza core, but is discernible in the
geochemical and magnetic susceptibility records. Following the collapse of Classic Maya
civilization human impact on the environment declined. Multiple proxies indicate that forest
recovery and soil stabilization began -1100 to 1000 14C yr BP.
The Yucatan peninsula consists of low-lying Tertiary limestones that are covered by
thin soils and characterized by solution features including depressions, caverns, lakes, and
water-filled cenotes (sinkholes) (Wilson 1984). The sediment record from these
waterbodies constitutes a rich source of paleoclimatic information, but few paleoclimatic
studies have been undertaken in the region. Late Holocene studies of pollen from Yucatan
lakes produced equivocal climatic interpretations because the region has had a long history
of human settlement and man-induced vegetation changes can mimic climate-induced floral
shifts (Bradbury 1982). Several studies successfully employed geochemical methods
(e.g., 5180 and Sr/Ca) for paleoclimatic reconstructions, which are less affected by human
activities than is the pollen record (Covich & Stuiver 1974; Metcalfe et al. 1994; Hodell et
Covich and Stuiver (1974) produced the first oxygen isotopic record from Yucatan
by isotopic analysis of gastropods in a sediment core from Lake Chichancanab. Broad
sample spacing (380 yr/sample) and poor age control in this pioneering study yielded a
paleoclimatic record of low temporal resolution. A more recent oxygen isotopic record was
obtained from Lake Chichancanab and spans the last 9200 years, with a temporal resolution
of -20 yr/sample (Hodell et al. 1995). Sediment cores from Lakes San Jose Chulchaca,
Sayaucil and Coba have also been investigated for pollen and stable isotopes (Whitmore et
al. 1996; Leyden et al. 1996), but carbonate microfossils in sediment cores from these
lakes were discontinuous, yielding incomplete paleoclimatic records.
Here I report continuous oxygen isotopic records of monospecific ostracods and
gastropods in a 6.3-m sediment core from Lake Punta Laguna, Quintana Roo, Mexico.
This lake is an exceptional site for paleoclimatic research for several reasons. First, the
lake loses most of its hydrologic budget via evaporation and changes in the 180/160 ratio of
lake water are therefore controlled mainly by E/P. Second, the lake sediments accumulated
rapidly (mean rate = 0.18 cm/yr) providing records of high temporal resolution (i.e.,
decadal). Third, Punta Laguna's sediments are composed almost entirely (90-95%) of
calcium carbonate and contain a continuous record of well-preserved ostracod and
gastropod remains. Lastly, Punta Laguna is located near Coba, an important and well-
studied Maya archaeological site (Folan et al. 1983).
The goal of this study was to reconstruct climate variability on the Yucatan
Peninsula during the late Holocene (-last 3,310 14C yrs) by measurement of oxygen
isotopes in ostracod and gastropod shells from sediments of Lake Punta Laguna. The
paleoclimate of the region is important because the Maya Civilization developed in
Mesoamerica about 3000 years ago and after flourishing during the Classic period,
mysteriously collapsed around 800 to 900 A.D. (Lowe 1985). The climatic context in
which Maya civilization evolved, flourished, and ultimately collapsed is not well known,
although climate change may have played a role in its evolution (Gunn & Adams 1981;
Dahlin 1983; Folan et al. 1983; Folan & Hyde 1985; Messenger 1990; Gill 1995). For
example, paleoclimatic data from Lake Chichancanab (Yucatan, Mexico) yielded evidence
of a drought lasting two centuries (800 to 1000 A.D.) that coincided with the Classic Maya
collapse (Hodell et al. 1995). The present study tests this finding at a geographic location
further north on the Yucatan peninsula, and the higher sedimentation rates in Lake Punta
Laguna permit climate reconstruction at higher temporal resolution than was possible in
Lake Punta Laguna (20 38'N, 87 37'W, elevation -14 m asl) is located in the state
of Quintana Roo, northeastern Yucatan Peninsula, Mexico, about 20 km north-northeast of
Coba (Figure 4-1). The lake has an area of -90 ha and is comprised of two interconnected
basins, each with a maximum depth >10 m. The area around Punta Laguna is relatively
Figure 4-1. Map showing location of Lake Punta Laguna in Quintana Roo, Mexico. Inset
shows the area of study on the northern part of the Yucatan Peninsula.
unstudied archaeologically, but appears to have been sparsely populated during the Classic
Period (Castillo & Peraza 1991).
Punta Laguna is located in a region characterized by seasonally dry tropical climate.
Seventeen years of climatological data are available from Nuevo X-can (site 23-016),
located ~25 km north of Punta Laguna. Mean annual temperature at Nuevo X-can is 25.7
C with maximum temperature in August (27.8 C) and minimum temperature in January
(23.0 C). Mean annual precipitation is 1519 mm with maximum rainfall occurring during
the months of June (216 mm) and September (239 mm) and minimum rainfall during
February (41 mm). The rainy season is May through November with precipitation surplus
especially pronounced from June to September. Evaporation exceeds precipitation during
the dry half year from December to May, and the annual rainfall deficit in the region is
approximately 100-200 mm (INEGI 1981, 1983).
Hydrologic inputs to Lake Punta Laguna include precipitation falling directly on the
lake, runoff, and subsurface inseepage. Hydrologic losses include evaporation and
possibly outseepage. Because evaporation rates are seasonally high, evaporation accounts
for the greatest proportion of water lost from the basin.
In 1993, a continuous 6.3-m sediment core was taken in 6.3 m of water from the
western basin of Lake Punta Laguna. Coring terminated on very stiff deposits.
Uppermost sediments (0-89 cm) were recovered using a piston corer with a 125-cm long,
7-cm diameter clear polycarbonate core barrel (Fisher et al. 1992). Deeper sections (50-
630 cm) were retrieved in approximately 1-m intervals using a square-rod piston corer
(Wright et al. 1984). The first three sections of core were overlapped by 50 cm to
guarantee full recovery. A composite section was later developed by matching oxygen
isotopic signals in overlapping sections. All core sections were sampled at 1-cm intervals
in the field and samples were transferred to labelled plastic bags for transport. The mud
water interface (MWI) core was extruded and sampled in a vertical position to maintain
stratigraphy of the uppermost unconsolidated sediments. Deeper, consolidated sediments
were extruded and sampled horizontally.
The core was dated by measurement of 14C in five small terrestrial wood samples
(range -1 to -10 mg) by Accelerator Mass Spectrometry (AMS). Four radiocarbon
measurements were also made on shell material from Punta Laguna to estimate hard-water-
lake error (Deevey & Stuiver 1964). Radiocarbon samples were measured at the National
Ocean Science AMS Facility at Woods Hole Oceanographic Institution. Dates were
converted to calendar ages using the CALIB computer program with a 100-year moving
average of the treering calibration dataset (Stuiver & Becker 1993; Stuiver & Reimer 1993).
Samples of lake water from both basins of Punta Laguna, ground water, and rain
water were collected in June, 1993. Water samples for 5180 analysis were processed by
equilibrating water with pure CO2 gas in evacuated VacutainerM tubes (Socki et al. 1992).
Equilibrated C02 was distilled from water and non-condensable gases off-line and the
purified CO2 was sealed in 6 mm glass breakseal tubes for transfer to a VG Prism II mass
spectrometer where oxygen isotopes were measured. All 8180 results of water are
reported in standard delta notation relative to VSMOW. Cations and chloride in the waters
were measured by inductively coupled plasma spectroscopy (Jarrell-Ash ICP 9000) and
sulfates were determined turbidimetrically (APHA 1992) using a Perkin Elmer Lambda 2
Sediments were processed by splitting each 1-cm sample in half. One half was
preserved for archival purposes and the other half was divided for several geochemical
analyses. Sediment for carbonate analysis was dried in an oven at 60 C and ground to a
fine powder with a mortar and pestle. Percent carbonate in samples taken at 10-cm
intervals was measured by coulometric titration (Engleman et al. 1985).
For 8180 analysis, part of each 1-cm sediment sample was disaggregated using a
3% H202 solution. Samples were then washed through a 250-jim sieve and coarse
material (>250 jim) was collected on filter paper and dried at 60 C. When dry, the coarse
material was transferred to glass scintillation vials. Adult specimens of the ostracod
Cytheridella ilosvayi were picked from the sieved 425-500 mrn fraction in each 1-cm
sample using a binocular microscope at 10X magnification. Prior to isotopic analysis,
ostracod specimens were cleaned using 15% H202 to remove organic material and rinsed in
methanol before drying. Ostracods were loaded into stainless steel carrying boats and
crushed with a glass rod dipped in methanol. Crushed samples were dried overnight and
loaded into the sample carousel attached to the mass spectrometer. Approximately 15-20
individual ostracod carapaces, weighing a total of -500 p.g, were used for most samples.
Multiple ostracod specimens were measured from each stratigraphic level to reduce the
variance (noise) generated by running single short-lived individuals (Heaton et al. 1995).
Individual gastropod shells were also picked from every 1-cm interval for stable
isotopic analysis. Single specimens of Pyrgophorus coronatus were separated from the
sieved >1400 i.tm fraction under 10X magnification. Gastropod shells were soaked in
15% H202 to remove organic, cleaned and sonicated in de-ionized water, and ultimately
rinsed with methanol before drying. Individual gastropod shells were ground to a fine
powder using a mortar and pestle. A fraction of the ground carbonate was loaded into
stainless steel carrying boats for isotopic analysis.
All carbonate samples for isotopic analysis were reacted in a common acid bath of
100% ortho-phosphoric acid (specific gravity = 1.92) at 90 C using a VG Isocarb
preparation system. Isotopic ratios of purified CO2 gas were measured on-line by a triple-
collector VG Prism II mass spectrometer. All carbonate isotopic results are reported in
standard delta notation relative to the VPDB standard. Analytical precision was estimated
by routinely measuring a powdered carbonate standard (Carrara Marble-UF) along with
samples from Punta Laguna. Precision (1 SD) was 0.09%o for 8180 (n = 374
standards). Oxygen isotopic data are available from the National Geophysical Data Center,
Boulder, Colorado, at paleo @ ngdc.noaa.gov.
Both terrestrial wood and aquatic gastropod samples are properly ordered with
radiocarbon ages increasing downcore; however, dates on gastropods are older than the
wood samples from corresponding depths (Table 4-1, Figure 4-2). This discrepancy is due
to hard-water-lake error caused by dissolution of limestone and input of old, 14C-deficient
carbon into the dissolved inorganic carbon of lake water (Deevey & Stuiver 1964). Hard-
water-lake error in Punta Laguna is estimated to be about 1200 to 1300 years.
The age/depth relation for the Lake Punta Laguna core was derived by interpolating
between AMS 14C dates on wood samples, assuming linear sedimentation rates between
dated horizons (Table 4-1, Figure 4-2). Ages above the first dated sample at 81 cm were
determined by interpolation between the surface (1993, equivalent to -43 radiocarbon
years) and 610 14C yr BP. The age at the base of the core is estimated to be -3310 14C yr
based on extrapolation of the linear sedimentation rate between the two bottommost dated
horizons. Sedimentation rates vary over the length of the core (Table 4-1). The 1-cm
sampling interval used in this study yields an overall average sample spacing of -5 years
for stable isotopic analyses.
Table 4-1. Radiocarbon dates and sedimentation rates for the Lake Punta Laguna core
Sample Type Depth Accession Radiocarbon Calibrated Age sed rate
(cm) Number Age (yr BP) (AD/BC) (cm/year)
Terrestrial Wood: 81 OS-6550 610 50 1368 AD 0.14
145 OS-10009 965 25 1030 AD 0.19
197 OS-10010 1530 50 569AD 0.11
380 OS-6553 2440 45 455 BC 0.18
494 OS-6554 2840 30 976 BC 0.22
Aquatic Gastropods: 25 OS-6549 1320 25 683 AD
83 OS-6551 1930 70 60 AD
246 OS-6552 3160 30 1415 BC
600 OS-5760 3720 40 2092 BC
- Bfl E Terrestrial Wood I-
0 Aquatic Gastropod
I I I I I I I I I
0 100 200 300 400 500 600
Figure 4-2. Age-depth relationship used to establish chronology for the Lake Punta
Laguna core. Ages of each sample in the core were determined by linear interpolation
between AMS 14C dates from the five terrestrial wood samples (see text). Radiocarbon
ages of aquatic gastropod shells are systematically older than terrestrial wood dates because
of hard-water-lake error (Deevey & Stuiver 1964). Plot symbols encompass error bars.
Concentrations of major cations in water samples from Punta Laguna are ordered as
follows: Na+>Ca2+>Mg2+>K+. Chloride and bicarbonate are the dominant anions with
minor amounts of sulfate (Table 4-2). Oxygen isotopic analysis of lake water, ground
water, and rain gave mean values of 0.93%o, -3.92%o, and -3.91%o, respectively (Table 4-
3). The mean value of -3.91%o for rain water collected near Punta Laguna during June
1993 is similar to the regional weighted average value of -4%o reported by Rozanski et al.
(1993). Local ground water (-3.92%o) had a mean 8180 value similar to that of rainfall.
Oxygen isotopic values from deep waters (mean--0.93%o) of Punta Laguna were similar to
surface water values (mean=--0.66%o). Overall, lake water samples collected in summer
over a six year period (1990-1995) had a mean 6180 of 0.93%o, which is almost 5%o
heavier than mean values for rain or ground water.
Table 4-2. Lake Punta Laguna water chemistry from samples collected June 1993 (n=5).
Conductivity measured in the field in July 1992. Bicarbonate was calculated as the sum of
cation charges minus the sum of chloride and sulfate charges.
Mg++ 3.50 meq L-1
K+ 0.09 meq U1
Na+ 6.27 meq L1
C- 6.09 meq L-1
S04-- 0.56 meq L-1
HCO3- 5.36 meq L-1
Conductivity 1375 ptS cm"1
Table 4-3. Oxygen isotopic composition of waters relative to VSMOW
Sample 6180 (%o) n
Rainwater (1993) -3.91 4
Groundwater (1993) -3.92 3
Punta Laguna deepwater (1993) 0.93 2
Punta Laguna surface water (1993) 0.66 3
Punta Laguna mean (1990-95) 0.93 10
Oxygen isotopic data were smoothed using a 5-point running mean to emphasize
longer-term trends (Figure 4-3). The oxygen isotopic records of the ostracod C. ilosvayi
and the gastropod P. coronatus are remarkably similar to each other, suggesting that both
taxa faithfully recorded changes in environmental conditions. Both records are
characterized by considerable variability with distinct maxima and minima lasting several
decades. This multi-decadal variability is superimposed upon longer-term, millennial-scale
trends. The long-term record was divided into three intervals based on changes in the mean
of the 8180 signal (Figure 4-3). From 3,310 to 1,785 142 yr BP (Period I), mean oxygen
isotopic values of ostracods and gastropods were relatively low. The mean value for C.
ilosvayi was -0.56%o, whereas P. coronatus averaged -0.77%o. At about 1,785 14C yr
BP, the mean 8180 value increased by over 0.6%o and ostracod 8180 values averaged
0.07%o and gastropod 8180 values averaged -0.07%o between 1,785 and 930 14C yr BP
This Period (II) is marked by some of the highest oxygen isotopic values of the entire
record including distinct maxima at 1,510, 1,171, 1,019, and 943 14C yr BP. During this
interval, the minimum 6180 values for ostracods (-0.65%o) and gastropods (-1.48%o) are
close to the mean values for Period I and Period III (910 14C yr BP present). The interval
of enriched 6180 values (Period II) terminates abruptly with a decrease in 8180 values at
930 14C yr BP During Period III (930 14C yr BP to the surface), mean oxygen isotopic
values are similar to Period I with mean 8180 values of -0.47%o and -1.21%o for ostracods
8'80 (%o, PDB)
Cytheridella ilosvayi Pyrgophorus coronatus
-2 -1 0 1 -3 -2 -1 0 1
Figure 4-3. Oxygen isotopic composition of the ostracod Cytheridella ilosvayi and the
gastropod Pyrgophorus coronatus in the sediment core from Lake Punta Laguna, Mexico.
The oxygen isotopic data were smoothed using a 5-point running mean to emphasize
longer-term trends. Record is divided into three periods on the basis of shifts in mean
5180 values: Period I (-3,310 to -1,785 14C yrs BP), Period II (-1,785 to -930 14C yrs
BP), and Period I1I (-930 14C yrs BP to the present).
and gastropods, respectively. During period II, a 56180 maximum is centered at 559 14C
yr BP (equivalent to 1391 A.D.).
Oxygen isotopic records from ostracod valves in lake sediments are used routinely
for paleoclimatic reconstructions (Chivas et al.1993; Holmes et al. 1995; Hodell et al.
1991, 1995; Lister et al. 1991; Curtis & Hodell 1993) and oxygen isotopic records of
gastropod shells have been used in a similar manner (Covich & Stuiver 1974; Abell 1985,
Hodell et al. 1995; Leyden et al. 1996). The oxygen isotopic composition of shell material
is dependent upon several variables including temperature, the oxygen isotopic ratio of
water (818Ow), and vital effects. Although several factors influence the oxygen isotopic
signal in sedimented shells, 518Ow is the most important factor in Lake Punta Laguna.
Temperature is a potentially important variable affecting the oxygen isotopic
composition of carbonate as illustrated by the paleotemperature equations for calcite (Craig
T C = 16.9 4.2 (618Ocalcite-.18Owater) + 0.13 (818Ocalcite- 18Owater)2 (4-1)
and aragonite (Grossman & Ku 1981):
T C = 19 3.52 (618Oaragonite-818Owater) + 0.03 (518Oaragonite-818Owater)2 (4-2)
Substituting the average 56180 of Punta Laguna lake water (0.93%o) and an annual water
temperature range of 23 to 27 C (for nearby Lake Chichancanab; Covich & Stuiver,
1974), the calculated annual range of 18Ocalcite is -0.9%o (from -1.32%o to -0.46%o), and
of 18Oaragonite is -l. 1%o (from -1.30%o to -0.20%o). Water temperatures from Lake
Chichancanab were used because thermal data from Lake Punta Laguna were unavailable.
The average annual air temperature range at stations near the lakes (Nuevo X-can, site 23-
016, -25 km north of Punta Laguna and La Presumida, site 023-021, -2 km west of
Chichancanab; INEGI 1981) is nearly identical and maximum water depths are similar.
Seasonal variations in water temperature from the two lakes should likewise be similar.
The calculated effect of temperature on 6180 of ostracod calcite is almost certainly
overestimated because hypolimnetic water temperature probably remains nearly constant
year round due to stable thermal stratification. Additionally, analysis of multiple ostracod
valves reduces seasonal temperature effects because each sample contains individuals that
grew during all seasons and different years. Likewise, measurement of the entire
gastropod shell reduces the seasonal temperature effect because individuals grow
throughout the year. However, the temperature range in the littoral environment inhabited
by gastropods may be greater than the range in deep-water, limnetic areas. I assume that
temperature changes in the tropics during the late Holocene have been relatively minor and,
therefore, changes in 18Ocarbonate due to temperature are small relative to changes in the
6180 of lake water.
In closed-basin tropical to subtropical lakes with seasonally dry climate, the 6180 of
lake water is controlled mainly by the ratio of evaporation to precipitation (Fontes &
Gonfiantini 1967; Covich & Stuiver 1974; Gasse et al. 1990; Talbot 1990; Lister et al.
1991). This is demonstrated for Lake Punta Laguna by comparing the measured 6180
values of lake water, rainfall, and ground water. Local ground water and rain water have
depleted 6180 values of -3.91%c and -3.92%o, respectively, whereas the lake water is
significantly enriched, averaging 0.93%o (Table 4-3). The great difference in oxygen
isotopic values (4.85%c) between input waters and lake water indicates that Punta Laguna
loses a significant fraction of its hydrologic budget to evaporation. During evaporation of
lake water, H2160 evaporates faster than H2180 due to its higher vapor pressure, thereby
enriching the lake in the heavier isotope. The oxygen isotopic composition of evaporating
water is also dependent upon the relative humidity, temperature, and wind speed at the
air/water interface at which evaporation occurs (Craig & Gordon 1965). I assume that
during times of higher E/P (dry times), the ratio of 180 to 160 in the lake water, and in
carbonate precipitated in equilibrium with lake water, will increase as the lake volume is