JOURNAL OF QUATERNARY SCIENCE (2005) 20(4) 363–376

Copyright ß 2005 John Wiley & Sons, Ltd.

Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/jqs.924

Climate change in lowland Central America

during the late deglacial and early Holocene
MICHAEL B. HILLESHEIM,1* DAVID A. HODELL,1 BARBARA W. LEYDEN,2 MARK BRENNER,1 JASON H. CURTIS,1
FLAVIO S. ANSELMETTI,3 DANIEL ARIZTEGUI,4 DAVID G. BUCK,5 THOMAS P. GUILDERSON,6
MICHAEL F. ROSENMEIER1y and DOUGLAS W. SCHNURRENBERGER7
1
Department of Geological Sciences and Land Use and Environmental Change Institute (LUECI), University of Florida, Gainesville,
FL 32611, USA
2
Department of Geology, University of South Florida, Tampa, FL 33620, USA
3
Geological Institute, Swiss Federal Institute of Technology ETH, Zurich, Switzerland
4
Institut Forel and Department of Geology, University of Geneva, Geneva, Switzerland
5
School of Natural Resources and Environment and Land Use and Environmental Change Institute (LUECI), University of Florida,
Gainesville, FL 32611, USA
6
Center for Accelerator Mass Spectrometry, Lawrence Livermore National Laboratory, Livermore, CA 94551, USA
7
National Lacustrine Core Repository and Limnological Research Center, University of Minnesota, Minneapolis, MN 55455, USA

Hillesheim, M. B., Hodell, D. A., Leyden, B. W., Brenner, M., Curtis, J. H., Anselmetti, F. S., Ariztegui, D., Buck, D. G., Guilderson, T. P., Rosenmeier, M. F.
and Schnurrenberger, D. W. 2005. Climate change in lowland Central America during the late deglacial and early Holocene. J. Quaternary Sci., Vol. 20 pp.
363–376. ISSN 0267-8179.
Received 11 August 2004; Revised 8 February 2005; Accepted 10 February 2005

ABSTRACT: The transition from arid glacial to moist early Holocene conditions represented a pro-
found change in northern lowland Neotropical climate. Here we report a detailed record of changes
in moisture availability during the latter part of this transition (
11 250 to 7500 cal. yr BP) inferred
from sediment cores retrieved in Lake Petén Itzá, northern Guatemala. Pollen assemblages demon-
strate that a mesic forest had been largely established by
11 250 cal. yr BP, but sediment properties
indicate that lake level was more than 35 m below modern stage. From 11 250 to 10 350 cal. yr BP,
during the Preboreal period, lithologic changes in sediments from deep-water cores (>50 m below
modern water level) indicate several wet–dry cycles that suggest distinct changes in effective
moisture. Four dry events (designated PBE1-4) occurred centred at 11 200, 10 900, 10 700 and
10 400 cal. yr BP and correlate with similar variability observed in the Cariaco Basin titanium record
and glacial meltwater pulses into the Gulf of Mexico. After 10 350 cal. yr BP, multiple sediment
proxies suggest a shift to a more persistently moist early Holocene climate. Comparison of results
from Lake Petén Itzá with other records from the circum-Caribbean demonstrates a coherent climate
response during the entire span of our record. Furthermore, lowland Neotropical climate during the
late deglacial and early Holocene period appears to be tightly linked to climate change in the high-
latitude North Atlantic. We speculate that the observed changes in lowland Neotropical precipitation
were related to the intensity of the annual cycle and associated displacements in the mean latitudinal
position of the Intertropical Convergence Zone and Azores–Bermuda high-pressure system. This
mechanism operated on millennial-to-submillennial timescales and may have responded to changes
in solar radiation, glacial meltwater, North Atlantic sea ice, and the Atlantic meridional overturning
circulation (MOC). Copyright ß 2005 John Wiley & Sons, Ltd.
KEYWORDS: Central America; Neotropics; palaeoclimate; lake sediments; geochemistry; stable isotopes; pollen; Holocene; Preboreal; deglacia-
tion; Intertropical Convergence Zone; 8.2-kyr Event.

Introduction ern Mexico, Central America, and northern South America, as

well as the West Indies (Fig. 1(A)). During the last deglaciation,
this region experienced a pronounced increase in moisture
The northern lowland Neotropics encompasses the area of the availability. At the onset of deglaciation, all shallow lakes were
New World with elevations < 500 m and extends north from dry and only deep basins contained water. The Petén Lake Dis-
the equator to the Tropic of Cancer. It includes areas of south- trict of northern Guatemala (Fig. 1(B)) is one of the few regions
with basins deep enough to have retained water during the arid
* Correspondence to: Michael B. Hillesheim, Department of Geological Sciences late Pleistocene (Deevey et al., 1980). Cores recovered from
and Land Use and Environmental Change Institute (LUECI), University of Florida,
Gainesville, FL 32611, USA. E-mail: [email protected]
Lakes Quexil (zmax ¼ 32 m) and Salpetén (zmax ¼ 32 m) in
y
Department of Geology and Planetary Science, University of Pittsburgh, 1980 provided some of the first insights into climatic and
Pittsburgh, PA 15260, USA (current address). ecological change in the lowland Neotropics during the last
364 JOURNAL OF QUATERNARY SCIENCE

Figure 1 (A) Map of the northern lowland Neotropics and Gulf of Mexico showing site locations discussed in this study: 1, Petén Lake District,
northern Guatemala; 2, Cariaco Basin, offshore northern Venezuela; 3, Lake Valencia, Venezuela; 4, Lake Miragoane, Haiti; 5, Louisiana slope, north-
ern Gulf of Mexico. (B) Map of the Petén Lake District. (C) Bathymetric map of Lake Petén Itzá with the locations of cores collected along seismic line
11 in 2002. Black dots (*) indicate cores used in this study and the black line (— —) is the location of seismic line 11. Cores along other seismic lines
in the northern basin are also indicated (5A, 8A, 8B, 14A). 93 refers to the location of Core PI 6-VII-93 (Curtis et al., 1998) taken from Petén Itzá’s
southern basin east of the island of Flores

Ice Age and the transition to the Holocene (Deevey et al., same climate regime, such as the marine Cariaco Basin off
1983). Pollen and geochemical profiles demonstrated an Venezuela (Haug et al., 2001), as well as with the climate his-
abrupt shift from cool, arid conditions during the last glaciation tory of the high-latitude North Atlantic recorded in marine sedi-
to warmer, moister conditions during the early Holocene ments and Greenland ice cores (Bond et al., 2001; Stuiver et al.,
(Leyden et al., 1993, 1994; Brenner, 1994). Because of incom- 1995). We find a coherent response of climate throughout the
plete recovery, drilling disturbance and poor dating resolution circum-Caribbean region during the late deglaciation and early
of the 1980 cores, details of the timing, rate and structure of this Holocene with tight linkages to climate changes observed in
transition were uncertain. the high-latitude North Atlantic.
In 1999, we completed a seismic survey (3.5 kHz) of six
Petén lakes to identify potential targets for coring and drilling.
We produced the first detailed bathymetric map of Lake Petén
Itzá (Fig. 1(C)), the largest (100 km2) and deepest Study site
(zmax ¼ 160 m) lake in the region. Seismic profiles revealed a
palaeoshoreline feature at
58 m below modern lake level
that possibly represents the lake lowstand during the last glacial Lake Petén Itzá is located at
17.00  N, 89.50  W in the
period (Anselmetti et al., in review). Because of the great depth Department of Petén, northern Guatemala (Fig. 1(B)). The lake
of its northern basin, Lake Petén Itzá probably contains one of is comprised of two connected basins. The deeper, northern
the few continuous sediment records in Central America of the basin (zmax ¼ 160 m) occupies a large half-graben formed by a
last glaciation. Consequently, it is a prime target for future dril- series of east–west aligned en echelon faults (Vinson, 1962).
ling operations to obtain long, undisturbed sediment cores for The northern shore is marked by a steep karst ridge consisting
palaeoclimate studies. of lower Tertiary limestone that follows the strike of the fault sys-
In June 2002, we retrieved 13 Kullenberg-type piston cores tem, whereas the southern shore is gently sloping and, in places,
from 10 sites in Petén Itzá’s northern basin (Fig. 1(C)). Undis- rimmed by poorly drained seasonal swamps (bajos). The smaller
turbed glacial-age sediment was not recovered because of southern basin is much shallower, averaging
5 m water depth.
the thickness of cohesive late Holocene sediments, which lim- Lake Petén Itzá’s water is dilute (11.22 meq l  1) and domi-
ited penetration of the piston corer to  6 m. The oldest contin- nated by calcium and bicarbonate, with magnesium and sul-
uous sediment record recovered is dated to 11 250 cal. yr BP, fate following closely in concentration (Table 1). Lakewater
and captures only the terminal part of the last deglaciation. pH is high (
8.0) and is saturated for calcium carbonate,
Here we report on the palaeoclimatic history of the Petén low- and there are abundant shells of carbonate microfossils in the
lands during the late deglacial and early Holocene period lake sediments (Covich, 1976; Curtis et al., 1998). Lake Petén
(
11 250 to 7500 cal. yr BP). We compare our results from Itzá is a terminal basin fed by precipitation, subsurface ground-
Lake Petén Itzá with proxy records from other sites under the water inflow, and a small input stream in the southeast. The

Copyright ß 2005 John Wiley & Sons, Ltd. J. Quaternary Sci., Vol. 20(4) 363–376 (2005)
 CLIMATE CHANGE IN CENTRAL AMERICA 365

Table 1 Mean ion concentrations (n ¼ 24) and oxygen isotopic com- northward migration of the ITCZ. The rainy season is often
position (
18Olw) V-SMOW of Petén Itzá lakewater (n ¼ 34) from sam- interrupted by a slight rainfall decrease in July and August,
ples collected at various sites and water depths in June and August which is known as the canicula or ‘little dry’ (Magaña et al.,
2002. Oxygen isotopic composition of precipitation (
18Op) interpo- 1999). The rainy season usually ends by late October and is fol-
lated from circum-Caribbean stations in the GNIP database (IAEA,
lowed by a pronounced dry season during the winter (January
2001).
18Olw was measured on a VG Prism II mass spectrometer using
a modified method from Socki et al. (1992). Ion concentrations were
to May) as the ITCZ moves equatorward (i.e. southward),
measured on a Dionex Model DX 500 chromatograph. Bicarbonate allowing the Azores–Bermuda high to dominate the Gulf of
was calculated as the sum of the cation charges minus the sum of chlor- Mexico and Caribbean Sea region. The region is also marked
ide and sulfate charges by subsidence during winter related to the descending limb
of the Hadley Cell, centred at
20  N (Waliser et al., 1999).
Ion Concentration (meq l1)
18Olw (n ¼ 34; %)
18Op (%) Lake Petén Itzá’s volume is sensitive to precipitation changes
and has fluctuated markedly in the recent past. For example,
Ca2þ 3.15 þ2.9% 4.0% mean annual rainfall during the period from 1934 to 1942
Mg2þ 1.86 was relatively high (2055 mm yr1) and resulted in increased
Naþ 0.60 lake levels and flooding (Deevey et al., 1980). In contrast, the
Cl 0.26
early to mid-1970s were relatively dry (mean annual rain-
SO42 2.11
fall ¼ 1415 mm yr1) resulting in lower lake levels. During the
HCO3 3.24
late 1970s lake level rose again in response to increased preci-
Total 11.22 pitation continuing until the early 1990s at which time the
trend reversed. Petén Itzá’s level varies seasonally, by as much
as 80 cm, displaying a rise after the summer rainy season and a
basin is effectively closed, lacking any surface outlets, although fall during the dry season with a 1–2 month lag (Deevey et al.,
some downward leakage may occur. The oxygen isotopic com- 1980). On seasonal to decadal timescales, lake volume and
position of lakewater averages 2.9%, which is enriched by water chemistry are sensitive to changes in the balance

7% relative to regional groundwater (Table 1). This reflects between precipitation and evaporation.
the importance of evaporation on the lake’s water budget. Ther-
mal stratification is persistent year round with hypolimnetic
temperatures averaging
25.4  C, close to the mean annual
air temperature (Fig. 2(A)). Hypolimnetic oxygen depletion Methods
has been observed near the deepest point in the lake (Fig. 2(B)).
Petén’s climate is marked by uniformly high temperatures
(averaging 25  C), and variability is most notably expressed In June 2002, we obtained six piston cores (Table 2) along a
by temporal and spatial changes in rainfall. Precipitation in depth transect on seismic line 11 in water depths ranging from
Petén varies from 900 to 2500 mm yr1, with a regional mean 9.7 to 63.2 m (Fig. 3). Cores were recovered using a Kullenberg-
of
1600 mm yr1 (Deevey et al., 1980). Lake Petén Itzá is type piston corer triggered by a mud–water interface (MUCK)
situated in a climatically sensitive region where the amount corer. In the laboratory, each core section was measured in its
of rainfall is related to the seasonal migration of the Intertropi- original polycarbonate liner for magnetic susceptibility and
cal Convergence Zone (ITCZ) and Azores–Bermuda high- gamma-ray attenuation (GRA) bulk density at 0.5-cm incre-
pressure system (Hastenrath, 1984). Approximately 90% of ments using a GEOTEK Multi-Sensor Core Logger (MSCL). The
the precipitation falls during the summer months (June to instrument was calibrated at the start of each day using an alu-
October) driven by increased convection associated with the minium standard. After logging, cores were split and imaged
with a GEOTEK digital color linescan camera that was cali-
brated each day using a white ceramic calibration tile.
The primary focus of this study is on deep-water cores PI 8-
VI-02 11A (water depth ¼ 58.2 m), PI 5-VI-02 11B (water
depth ¼ 51.6 m), and shallow-water Core PI 6-VI-02 11D
(water depth ¼ 20.9 m), hereafter referred to as Cores 11A,
11B, and 11D. Core 11B was converted to the same sediment
depth scale as Core 11A by correlating variations in colour
(rgb) intensity (correlation coefficient, r ¼ 0.84) using Analy-
series 1.1 (Paillard et al., 1996). The working halves of Cores
11A and 11B were sampled at 1-cm and 5-cm intervals, respec-
tively. Small subsamples (
5 cc) were taken for elemental

Table 2 Length, water depth and location of cores retrieved along

seismic line 11 in Lake Petén Itzá in June 2002. Cores are labelled by
abbreviated lake name, day–month–year, and site

Core Length Depth Latitude Longitude

(cm) (m) (  N) (  W)

PI 8-VI-02 11A 558 58.2 17.000 89.779

Figure 2 Water column profiles of (A) temperature (  C), (B) dissolved PI 5-VI-02 11B 515 51.6 16.998 89.779
oxygen (mg l1), and (C) oxygen isotopic composition (
18Olw,%) PI 9-VI-02 11C 275 30.0 16.991 89.780
V-SMOW of Petén Itzá lakewater. The temperature profile and samples PI 6-VI-02 11D 210 20.9 16.988 89.779
for
18O analysis were collected on 13 August 2002 at 17.00  N, PI 8-VI-02 11E 255 9.7 16.984 89.779
89.85  W. Dissolved oxygen is plotted for May (solid line) and August PI 9-VI-02 11F 585 63.2 17.004 89.780
(dashed line) 1980 (unpublished data)

Copyright ß 2005 John Wiley & Sons, Ltd. J. Quaternary Sci., Vol. 20(4) 363–376 (2005)
366 JOURNAL OF QUATERNARY SCIENCE

Figure 3 (A) High-resolution seismic image of line 11 showing the location of piston Cores 11A-11F (see also Table 2). Note the Late Glacial/Holo-
cene (LG/H) sequence boundary (bold dashed–solid line) and the palaeoshoreline feature at
58 m palaeo water depth. (B) Images of piston cores
taken along a depth transect on seismic line 11. Water depths are relative to modern (2001) lake level. Dated horizons are indicated by arrows and
inter-core correlations are marked by solid lines. The LG/H boundary (bold line) is marked by dark palaeosols in shallow-water Cores 11C, 11D, and
11E and gypsum precipitation in deeper-water Cores 11A and 11B. This figure is available in colour online at www.interscience.wiley.com/journal/jqs

analysis, oven-dried, and ground to a fine powder with a mor- and was sampled at 5-cm intervals for elemental composition
tar and pestle. The remainder of the sample was wet-sieved at and pollen analysis.
63 mm to isolate microfossils and terrestrial organic material for Radiocarbon ages were determined at the Center for Accel-
isotopic and radiocarbon analysis. The three shallow-water erator Mass Spectrometry (CAMS) at Lawrence Livermore
cores are PI 9-VI-02 11C, PI 6-VI-02 11D, and PI 8-VI-02 11E National Laboratory. Eighteen samples of terrestrial organic
(Table 2). Core 11D was selected as representative of the sedi- matter (wood, leaf, or charcoal) and eight samples of aquatic
ment stratigraphy in shallow-water cores (<35 m water depth) gastropod (Pyrgophorus sp.) shell were dated. Samples of shell

Copyright ß 2005 John Wiley & Sons, Ltd. J. Quaternary Sci., Vol. 20(4) 363–376 (2005)
 CLIMATE CHANGE IN CENTRAL AMERICA 367

material were used to date intervals where terrestrial organic automated preparation system. Isotopic ratios of purified CO2
matter was scarce. Samples of terrestrial organic matter used gas were measured online with a ThermoFinnigan 252 mass
for radiocarbon dating were pretreated at the University of Flor- spectrometer. All isotopic values are reported in standard delta
ida using the acid–base–acid pretreatment method. Radiocar- (
) notation relative to the VPDB standard. Analytical precision
bon ages were determined after blank subtraction as was estimated by measuring eight samples of a powdered car-
determined on similarly pretreated and processed 14C-free bonate standard (NBS-19) with each sample run. Precision for
wood. Organic matter samples were also analysed for
13C
18O was estimated to be  0.10% (1; n ¼ 24 standards).
content and corrected to 25% prior to calibration. Five sam- Samples for pollen analysis were processed using 1 cm3 of
ples from Core 11A were too small for
13C analysis and were sediment taken from Core 11B at 1-cm to 10-cm intervals
assumed to have values of 25%. Calibrated ages, in calendar and Core 11D at 5-cm intervals. One Lycopodium tablet (batch
years before 1950 (cal. yr BP), were calculated using 124961; mean ¼ 12 542; SD ¼  931) was added to each sam-
INTCAL98, with a 100-yr moving average of the tree ring cali- ple to calculate pollen concentration. Samples were processed
bration data set (Stuiver and Reimer, 1993; Stuiver et al., 1998). using a standard sequence of HCL, KOH, HF and acetolysis.
Inorganic carbon (IC) was determined by coulometric titration The residue was stained with basic fuchsin and suspended in
using a UIC/Coulometrics 5011 coulometer coupled with a UIC tertiary butyl alcohol (TBA), and measured aliquots of the sus-
5240-TIC carbonate autosampler. Weight percent calcium pension were mounted in silicone oil on microscope slides.
carbonate (CaCO3, %) was calculated by multiplying IC by Pollen from Core 11D were counted at 400  magnification
8.33. Total weight percent carbon (TC) and sulphur (S, %) were to a minimum sum of 200, excluding pteridophytes, aquatic
measured using a Carlo Erba NA 1500 CNS elemental analyser algae and macrophytes, and unidentifiable grains. For Core
with autosampler. Weight percent organic carbon (OC) was 11B, Brosimum-type pollen was also excluded from the sum
estimated by subtracting IC from TC. Weight percent organic to enhance the representation of the other taxa. Concentrations
matter (OM, %) was estimated by multiplying OC by 2.2. were converted to accumulation rates (grains cm1 yr1) using
Oxygen isotope ratios in Core 11A were measured on adult the age model for Core 11B.
carapaces of the ostracod Limnocythere sp., which were iso-
lated from the >250 mm sediment fraction. Ostracod valves
were abundant from 558 to 500 cm, but gradually decreased Results
above this level until they became absent at 484 cm. Ostracods
were cleaned with 15% H2O2 to remove organic matter, and
then washed with de-ionised water, rinsed in methanol, and Radiocarbon
dried. Ostracod aggregates of 15–20 valves were measured
from each 1-cm sample. The samples were reacted in 100% All 13 radiocarbon samples from Core 11A yielded dates in
orthophosphoric acid at 70  C using a ThermoFinnigan Kiel III stratigraphic order (Table 3). An age–depth model was derived

Table 3 Accelerator mass spectrometry (AMS) radiocarbon dates and calibrated ages of terrestrial organic and aquatic shell material from Lake Petén
Itzá Cores 11A, 11B, 11C, 11D and 11E. All radiocarbon ages were corrected to
13C values of 25%, except leaf and wood samples (indicated with
#), that were too small for
13C analysis and are assumed to be 25%

Core Sample Depth CAMS
13C (%) Radiocarbon Calibrated age

type (cm) number age (1, 14C yr BP) (2, cal. yr BP)

PI 8-VI-02 11A Leaf 75.5 99102 23.31 1660  40 1570  60

Wood 190.5 99103 # 2095  35 2070  80
Wood 247.5 101228 # 2530  60 2600  150
Wood 297.5 99104 22.27 2570  45 2740  35
Wood 353.5 92837 30.30 3240  30 3435  55
Wood 369.5 99105 31.68 3525  25 3795  75
Wood 446.5 101229 23.63 6160  70 7065  180
Leaf 448.0 101230 # 6155  50 7025  135
Wood 463.5 99106 # 6585  35 7470  40
Wood 489.5 92836 30.80 7925  30 8720  90
Wood 504.5 101231 28.67 8425  30 9475  55
Wood 547.0 101232 # 9470  110 10 820  340
Wood 552.0 101233 22.70 9555  30 11 015  75
PI 5-VI-02 11B Shell 280.0 94597 2.69 3970  40 4440  90a
Charcoal 365.0 94606 27.12 5930  40 6730  70
Wood 433.0 92835 28.30 7330  40 8105  85
PI 9-VI-02 11C Charcoal 140.0 94607 15.67 9665  45 11 120  60
Shell 140.0 94598 0.47 13 810  45 16 580  480a
Shell 270.0 94599 0.56 32 420  290 37 635  315a,b
PI 6-VI-02 11D Charcoal 75.0 94608 17.92 9460  45 10 670  110
Shell 75.0 94600 1.50 11 025  40 13 020  150a
Shell 140.0 94601 1.85 19 320  70 22 950  730a
Shell 205.0 94602 1.59 22 110  90 26 035  105a,b
PI 8-VI-02 11E Charcoal 165.0 94609 12.71 9055  50 10 220  70
Shell 165.0 94603 3.68 10 280  40 12 070  310a
Shell 250.0 94604 3.50 17 250  60 20 540  650a

a
Ages are potentially subject to hard-water-lake error (Deevey and Stuiver, 1964).
b
Radiocarbon ages were too old for INTCAL98; therefore, ages were calibrated using the following equation from Bard et al. (1998): age
(cal. yr BP) ¼  3.0126  106 (14C age BP)2 þ 1.2896 (14C age BP)  1005.

Copyright ß 2005 John Wiley & Sons, Ltd. J. Quaternary Sci., Vol. 20(4) 363–376 (2005)
368 JOURNAL OF QUATERNARY SCIENCE

Lithology and geochemistry

Shallow-water cores

Sediment properties of Core 11D (Fig. 5) represent the general

pattern of lithologic and geochemical changes observed in the
three shallow-water cores (<35 m water depth). Pleistocene
age sediments consist of dense, carbonate-rich silty clays over-
lain by a palaeosol that is marked by increased magnetic sus-
ceptibility (Fig. 5(D)), the presence of terrestrial gastropods and
plant roots, and the absence of aquatic microfossils. The transi-
tion to lacustrine deposition, dated between 11 100 and
10 200 cal. yr BP (Fig. 3(B)), is marked by decreases in density
and magnetic susceptibility concurrent with increases in per-
centage carbonate and organic matter content (Fig. 5), as well
as the reappearance of lacustrine gastropods and ostracods.

Figure 4 Calibrated age versus depth for Lake Petén Itzá Core 11A. Deep-water cores
Core chronology was established by linear interpolation between 13
calibrated radiocarbon dates measured on terrestrial wood and leaf
The bottom 25 cm (
11 250 to 10 350 cal. yr BP) in both deep-
samples. Average sedimentation rates vary between
0.2 and

0.025 cm yr1, with a transition in sedimentation rate occurring at water cores (11A, 11B) is composed of dense gypsum sands

3000 cal. yr BP. Dating error (2) on all samples is smaller than the interbedded with silty clays. These lithologic changes are
plot symbols, except where indicated by error bar reflected in the calcium carbonate (CaCO3), organic matter
(OM), and sulphur (S) records (%, Fig. 6(A)–(C)). Gypsum-rich
by linear interpolation between dated horizons (Fig. 4). We intervals are indicated by high sulphur content centred at
estimated the basal age of Core 11A to be
11 250 cal. yr BP
10 400, 10 700, 10 900 and 11 200 cal. yr BP. The interven-
by extrapolating the linear sedimentation rate 6 cm beyond ing silty-clay layers contain higher percentages of carbonate
the lowermost radiocarbon date at 552 cm. Linear sedimenta- and organic matter and are centred at
10 500, 10 800 and
tion rates (Fig. 4) average
0.025 cm yr1 from the base of the 11 000 cal. yr BP. At 10 350 cal. yr BP, gypsum precipitation
core at 558 cm to 335 cm (11 250 to 3000 cal. yr BP). Between ceased and sediments became dominated by calcium carbo-
335 and 57 cm (3000 to
1400 cal. yr BP), sedimentation rates nate and organic matter. Between 10 350 and 7500 cal. yr BP,
increase to
0.2 cm yr1 associated with the onset of deposi- %CaCO3 decreased from 60% to 10% and %OM increased
tion of the Maya Clay unit. The temporal resolution averages slightly, averaging 14%, with the remainder of the sediment

40 yr for each 1-cm sample during the late deglacial and being composed mostly of inorganic clay. Superimposed upon
early Holocene part of the sequence (558 to 463 cm or these long-term trends are small, short-term variations. At
11 250 to 7500 cal. yr BP). Charcoal near the top of a soil hor- 9700 cal. yr BP, %CaCO3 decreased by 25% and %OM
izon found in the three shallow-water cores (Table 3; Fig. 3(B)) increased by
10%. From 9700 to 8800 cal. yr BP, %CaCO3
was dated to constrain the onset of lacustrine deposition (i.e. and %OM remained relatively constant averaging 25% and
flooding surface) associated with lake filling. Ages ranged from 14%, respectively. At 8800 cal. yr BP, %CaCO3 began to
11 120  60 cal. yr BP in Core 11C (31.8 m water depth) to decrease more rapidly, coincident with a distinct increase in
10 220  70 cal. yr BP in Core 11E (9.7 m water depth). In addi- %OM. This trend is interrupted at 8100 cal. yr BP by a
10%
tion, three shell samples from the base of each shallow-water decrease in organic matter and a
5% increase in carbonate,
core were dated, yielding ages between 20 540 and relative to the bulk sediment, that persisted until 7800 cal. yr BP
37 650 cal. yr BP (Table 3; Figure 3(B)). when both proxies returned to pre-excursion values.

Figure 5 Sediment variables measured in Lake Petén Itzá Core 11D. (A) Weight percent calcium carbonate (CaCO3, %), (B) weight percent organic
matter (OM, %), (C) GRA bulk density (g cc1), (D) magnetic susceptibility (cgs), (E) red colour intensity (nm). The dashed line denotes the dated
lithologic transition from palaeosol to lacustrine deposition. The approximate thickness of the palaeosol is highlighted in grey

Copyright ß 2005 John Wiley & Sons, Ltd. J. Quaternary Sci., Vol. 20(4) 363–376 (2005)
 CLIMATE CHANGE IN CENTRAL AMERICA 369

Figure 6 Sediment variables measured in Lake Petén Itzá Core 11A. (A) Weight percent calcium carbonate (CaCO3, %), (B) weight percent organic
matter (OM, %), (C) weight percent sulphur (S, %), (D) GRA bulk density (g cc1), (E) oxygen isotopic composition (
18O, %) of ostracods. Distinct
changes in sediment composition are highlighted in grey

GRA bulk density follows lithologic changes between 11 250 sition. Total pollen accumulation rates in Core 11B were low
and 10 350 cal. yr BP. High density is associated with gypsum prior to 10 250 cal. yr BP, less than one-third of the values cal-
beds and low density with clay-rich layers (Fig. 6(D)). At culated for later deposits. Subsequent accumulation rates fol-
10 350 cal. yr BP, density decreased abruptly followed by pro- low the percentage trends. Brosimum-type predominates in
gressively lower values until the end of the record at the basal zone from 11 250 to 10 250 cal. yr BP together with
7500 cal. yr BP. This trend was interrupted, however, by a small Bursera, Ulmus, Gymnanthes, pine and oak. Percentages for
(0.1 g cc1), short-lived (
100 yr) increase centred at Liquidambar (sweet gum), Typha (cattail) and Asteraceae
9400 cal. yr BP and a larger (0.2 g cc1), longer-lived are notable. From 10 250 to 8250 cal. yr BP Brosimum-type,
(
300 yr) increase centred at
8000 cal. yr BP. other Moraceae, Celtis and Melastomataceae/Combretaceae
Ostracods from the base of Core 11A record the greatest increase while Liquidambar, oak and, to a lesser extent, Gym-

18O values averaging 4.5% (Fig. 6(E)). Between
11 250 nanthes decline. The aquatic alga Botryococcus increases shar-
and 10 400 cal. yr BP,
18O generally decreased by
1.3%. ply by 8800 cal. yr BP. After 8300 cal. yr BP Brosimum-type and
Superimposed upon this trend are two distinct
18O minima Gymnanthes decrease and secondary forest taxa such as Bur-
centred at 11 000 and 10 500 cal. yr BP. At 10 400 cal. yr BP, sera and Cecropia increase, followed later by increases in

18O values increased slightly and then gradually decreased Byrsonima, grasses, and charcoal.
until
9600 cal. yr BP, at which point values begin to increase
again. At 9300 cal. yr BP, this trend reversed and
18O reached
its lowest values of
3.1% at 8800 cal. yr BP. Ostracods are
absent in the sediment profile after this time. Interpretation of proxies

The oxygen isotopic (
18O) composition of shell carbonate is

Pollen influenced by temperature,
18O of lake water, and vital effects
(i.e. taxon-specific fractionation) (Covich and Stuiver, 1974).
Pollen is poorly preserved prior to
11 500 cal. yr BP (86 cm) Vital effects were minimised by measuring a single species of
in Core 11D (Fig. 7(A)). Within the soil horizon (75 to adult ostracod (Limnocythere sp.) from the > 250 mm sediment
115 cm), Pinus (pine) predominates together with Quercus fraction. Temperature change in the region during the Prebor-
(oak), Asteraceae (composites) and Poaceae (grasses), but pol- eal and early Holocene was probably small given the minor
len concentrations are low. A small increase in aquatic taxa change in pollen assemblage relative to the last glacial, as well
occurs as lacustrine deposition begins at 10 670 cal. yr BP as the lack of cool indicator species such as Juniperus and
(75 cm). After which Brosimum-type becomes a dominant Liquidambar. We therefore suggest that ostracod
18O was
taxon, along with Gymnanthes, pine, oak, Bursera and grasses. controlled principally by changes in the
18O of lakewater,
The pollen percentage profile for Core 11B (Fig. 7(B)) can be which is related to the ratio of evaporation to precipitation
divided into three zones that correlate with the sedimentologi- (E/P) and the
18O of precipitation (Fontes and Gonfiantini,
cal and isotopic records, but typically lag behind them by 1967; Gasse et al., 1990). Between 11 500 and 8700 cal. yr BP,

100 yr based on the high-resolution sampling of pollen (i.e. the
18O of western Caribbean seawater decreased by
0.4%
1-cm intervals) over the late deglacial-to-early Holocene tran- (Schmidt et al., 2004), which should have decreased the
18O

Copyright ß 2005 John Wiley & Sons, Ltd. J. Quaternary Sci., Vol. 20(4) 363–376 (2005)
370 JOURNAL OF QUATERNARY SCIENCE

Figure 7 Pollen percentage diagrams from Lake Petén Itzá. (A) Shallow-water Core 11D and (B) deep-water Core 11B. For both cores, certain taxa
(i.e. ferns, algae, aquatic vegetation, and unidentifiable pollen) were excluded from the pollen sum used to calculate percentages. Aquatic taxa were
excluded so that the pollen sum reflects only terrestrial taxa. In addition, Brosimum-type was excluded from the pollen sum of Core 11B, in order to
enhance the changes in the other taxa. Excluded taxa shown in the diagrams have percentages calculated using the restricted sum as their percentages
have interpretative value. Grey curves in Core 11D represent a factor of five exaggeration of the original pollen percentages

of precipitation by the same amount because the Caribbean is the

primary source of water vapour to the Yucatan Peninsula. The

18O of precipitation in the lowland tropics is also dependent
on the amount of rainfall, with lower
18O values associated with
greater precipitation (Rozanski et al., 1993). Both the ‘amount
effect’ and changes in E/P influence lakewater
18O in the same
direction such that increased precipitation results in lower
18O
values of both rainfall and lakewater. We conclude that varia-
tions in Lake Petén Itzá ostracod
18O represent relative changes
in the ratio of precipitation to evaporation. Times of changing E/P
are represented by changes in the slope of
18O and periods of
unchanging
18O represent times of fairly constant E/P. We there-
fore compared the first derivative (i.e. changing slope) of the
18O
signal with other proxy records (Fig. 8).
Changes in precipitation affect delivery rates of detrital mate-
rial to the lake basin. Changing rainfall also alters the concen-
tration of dissolved salts by enriching or diluting ions in
lakewater. We used sediment lithology and chemical composi-
tion to infer past changes in detrital input and lakewater chem-
istry that are, in turn, controlled by climate. For example,
greater gypsum content in sediments and associated increases
in sediment density reflect lake volume reduction and higher
lakewater salinity during dry periods, similar to the interpreta-
tion of the record from Lake Chichancanab, northern Yucatan
Peninsula (Hodell et al., 1995, 2001, 2005). Scanning electron
micrographs (SEMs) of gypsum grains from Lake Petén Itzá
show they are euhedral and have no evidence of rounding or Figure 8 Comparison of the first-derivative of the detrended oxygen
abrasion that would indicate transport or redeposition. We sug- isotopic signal d(
18O)/dt of ostracods from Petén Itzá Core 11A
gest that gypsum formed authigenically in the littoral zone ~
( ) with (A) density (g cc1) and (B) percentage carbonate (CaCO3,

where the evaporation rate was higher than that in deeper, %). PBE1–4 designate dry events listed in Table 4 and are highlighted
open water. in grey

Copyright ß 2005 John Wiley & Sons, Ltd. J. Quaternary Sci., Vol. 20(4) 363–376 (2005)
 CLIMATE CHANGE IN CENTRAL AMERICA 371

Discussion depth), from subaerial exposure (palaeosol) to subaqueous

deposition (lacustrine sediments), marks a lake-level rise at this
time (Figs 3(B) and 5). In deep-water Core 11A, variations in
Palaeoclimatic history of Lake Petén Itzá oxygen isotopes, density and sediment composition records
(Fig. 6) show a number of abrupt changes that suggest varying
Seismic profiles suggest that lake level was
58 m below mod- effective moisture and lake-level instability. These changes
ern prior to 11 250 cal. yr BP (Fig. 3(A); Anselmetti et al., in represent a series of wet–dry cycles of
250 yr duration.
review), which required an
87% (4.8 km3) reduction in During the Preboreal, four distinct dry events (designated
volume relative to today (Fig. 9). Shallow-water cores PBE1-4) are identified by lithologic changes (i.e., increased
( < 35 m below modern lake level) contain a palaeosol indicat- gypsum content) centred at 11 200, 10 900, 10 700, and
ing subaerial exposure. Cores from nearby, shallower Lakes 10 400 cal. yr BP (Fig. 8). Changes in
18O, as expressed by
Quexil and Salpetén also indicated lake-level lowering of at the first derivative of the signal, are consistent with the litholo-
least 30–40 m during the last glacial period (Deevey et al., gic changes, typically leading them by
50 yr (Fig. 8).
1983), which probably caused them to become ephemeral. Low temporal sampling resolution (
50–200 yr) for pollen
Glacial-age vegetation consisted of sparse temperate thorn over this interval precludes detailed interpretation of vegetation
scrub and taxa such as Juniperus, indicating much cooler tem- changes during this period. Average climate conditions at the
peratures. One estimate suggests that temperatures were
6.5 beginning of the Preboreal were somewhat cooler and drier.
to 8  C lower than today (Leyden et al., 1993, 1994). But after
10 700 cal. yr BP, pine, oak, Liquidambar and Gym-
During the last deglaciation, Petén climate was marked by nanthes begin to decline while Acalypha, Trema and charcoal
an abrupt shift from arid to humid conditions, as inferred from increase (Fig. 7(B)). Liquidambar reflects cooler conditions and
multiple sediment proxies (Leyden et al., 1993, 1994; Brenner, Gymnanthes prefers drier, open conditions. Acalypha and
1994). Pollen assemblages in shallow-water Core 11D (Fig. Trema are common tropical secondary or successional taxa.
7(A)) changed from Pinus to Brosimum-type dominance conco- Although percentages for Brosimum do not increase, the accu-
mitant with the rise of other mesic taxa. This increase in mesic mulation rate is slightly higher. During this interval, pollen
taxa correlates with a similar change in the pollen profile from changes suggest moister, and perhaps warmer, climatic condi-
nearby Lake Quexil that occurred after
12 600 cal. yr BP tions, in general agreement with other proxy indicators.
(Leyden et al., 1993, 1994). An important change in Lake Petén Itzá’s hydrology
Deep-water cores from Lake Petén Itzá do not capture the occurred at
10 350 cal. yr BP, coinciding with the Prebor-
entire deglacial sequence. Instead, they begin during the latter eal/Boreal boundary observed in northern European pollen
part of deglaciation, at
11 250 cal. yr BP. Predominance of records. At this time, an increase in effective moisture is
Brosimum at the base of Core 11B indicates a mesic tropical inferred from sharp decreases in density and sulphur content
forest had largely been established in Petén by this time (Fig. that indicate cessation of gypsum precipitation (Fig. 6(C) and
7(B)). This is in agreement with
18O and pollen records from (D)) concomitant with increases in carbonate and organic mat-
Lake Quexil that indicate warmer and wetter conditions fol- ter content (Fig. 6(A) and (B)). A small increase in the abun-
lowing the glacial period (Leyden et al., 1993, 1994). Dated dance of Brosimum-type pollen lags this transition by
palaeosols in shallow-water Petén Itzá cores, however, indi-
100 yr (Fig. 7(B)). This represents the time at which Lake
cate lake level was at least 35 m lower than today (Fig. 3(B)). Petén Itzá’s water chemistry fell below gypsum saturation
This conclusion is supported by the presence of Typha pollen and sediment lithology and density became less variable.
in deep-water cores ( > 50 m) that indicates a nearby littoral After 10 350 cal. yr BP, reduced variability of sediment com-
zone with emergent vegetation (Fig. 7(B)). position and ostracod
18O values indicate a more persistent
From 11 250 to 10 350 cal. yr BP, during the Preboreal per- moist climate (Fig. 6). This is further supported by the resump-
iod, the
18O of Lake Petén Itzá ostracods in Core 11A tion of lacustrine deposition at sites shallower than 8 m water
decreased, indicating greater effective moisture (Fig. 6(E)). depth (Fig. 3(B); Curtis et al., 1998) indicating lake stage had
The lithologic change in shallow-water cores ( < 35 m water risen to near present levels by 10 200 cal. yr BP (Fig. 9). Centred
at 9400 cal. yr BP, small increases in density and
18O may sug-
gest a brief (
100 yr) return to slightly drier conditions
(Fig. 6(D) and (E)). At 8800 cal. yr BP, the abundance of aquatic
algae (Botryococcus) increases (Fig. 7(B)) concomitant with an
increase in organic matter content (Fig. 6(B)). This also marks
the time when ostracods in deep-water cores disappear, indi-
cating the onset of hypolimnetic anoxia, which may also
account for better preservation of organic matter. Between
8100 and 7800 cal. yr BP, changes in sediment composition
suggest a temporary return to drier conditions (Fig. 6(A)–(D)),
after which humid climate conditions prevailed again.
From 10 250 to 8300 cal. yr BP, the pollen record also indi-
cates a relatively moist, stable climate. The continued promi-
nence of Gymnanthes, however, suggests that parts of the
watershed were still edaphically dry or unstable, perhaps on
some of the steep basin slopes. Given the limited sampling
resolution (
200 yr) over this interval, the pollen record sug-
gests drier conditions between 8100 and 7800 cal. yr BP, con-
sistent with lithologic changes. This period also marks the
beginning of fundamental changes in the vegetation, that
include the introduction of other secondary taxa such as Cecro-
Figure 9 Hypsographic curve of cumulative lake volume versus water pia that reflect a more diverse, open forest and increased sea-
depth. Arrows indicate inferred lake levels based on flood surface ages sonality (Fig. 7(B)).

Copyright ß 2005 John Wiley & Sons, Ltd. J. Quaternary Sci., Vol. 20(4) 363–376 (2005)
372 JOURNAL OF QUATERNARY SCIENCE

Comparisons with other circum-Caribbean records from Lakes Valencia and Petén Itzá are the most reli-
and North Atlantic records ably dated because the chronologies are based exclusively
on radiocarbon analysis of discrete terrestrial macrofossils. In
Here we compare our results with other marine and lake contrast, the Lake Miragoane chronology is based on radiocar-
records from the circum-Caribbean and Gulf of Mexico, in bon analysis of carbonate microfossils with the assumption of a
order to assess the regional extent of palaeoclimatic changes constant hard-water lake error. Despite the dating uncertainty
observed in Lake Petén Itzá proxies. Previous studies of the last in Lake Miragoane the pattern of oxygen isotope variation is
glacial-to-interglacial cycle have reported similarities between similar to that from the other lakes, indicating several abrupt
climate change in the Caribbean and the high-latitude North shifts in moisture availability on an otherwise long-term trend
Atlantic on a variety of timescales (Hughen et al., 1996; Peter- toward wetter conditions in the early Holocene.
son et al., 2000). Consequently, we also compare the Petén Itzá The most detailed records of the last deglaciation and Holo-
record with proxy signals from the high latitudes of the northern cene in the Caribbean come from studies of laminated sedi-
hemisphere. When comparing Lake Petén Itzá proxies with ments in the Cariaco Basin (11  N) off the coast of Venezuela
other palaeoclimate records, each signal was plotted on its (Hughen et al., 1996, Peterson et al., 2000, Haug et al., 2001).
own independent timescale and no attempt was made to align In general, the density record from Lake Petén Itzá Core 11A
features in the records. and the titanium (Ti) record from Cariaco (Haug et al., 2001)
There are relatively few lacustrine records of the last degla- are remarkably similar (Fig. 11). Both sediment proxies are
ciation in the circum-Caribbean because most shallow lakes probably related indirectly to precipitation. In Cariaco, rainfall
remained dry well into the early Holocene. Lake Miragoane, controls the input of terrigenous material to the basin. Low Ti
Haiti, and Lake Valencia, Venezuela, are two lakes located values in Cariaco sediments have been interpreted to reflect
on the northern and southern rim of the Caribbean Basin, drier climate and vice versa. In Lake Petén Itzá, detrital input
respectively, that were deep enough to preserve records of and lakewater chemistry control sediment composition, which
the latter part of the last deglaciation (Bradbury et al., 1981; is related indirectly to precipitation.
Leyden, 1985; Hodell et al., 1991; Curtis and Hodell, 1993; The Petén Itzá record does not extend back to the time of the
Brenner et al., 1994; Higuera-Gundy et al., 1999; Curtis et al., Bølling–Allerød or Younger Dryas events that are prominent in
1999, 2001). Between 11 250 and 8800 cal. yr BP, oxygen iso- the Cariaco Ti signal, but rather begins in the Preboreal period
topic records from Lakes Miragoane, Valencia and Petén Itzá when Cariaco Ti values were generally increasing but lower
are similar, showing large-scale fluctuations superimposed on than those of the early Holocene. The late deglacial is marked
a long-term trend of increasing moisture availability (Fig. 10). by density variations in Core 11A that correspond to changes in
Several
18O maxima and minima can be correlated among Ti content of Cariaco Basin sediments. Within dating uncer-
the records within the uncertainty of the chronologies. The tainties, reductions in Cariaco Ti content correlate to intervals
of increased density and gypsum content in Lake Petén Itzá
(Fig. 11(B)), suggesting that the dry Preboreal events (PBE1-4)
observed in the Petén Itzá record may have been pan-Carib-
bean in extent.
At the end of the Preboreal period (
10 350 cal. yr BP), both
records show an abrupt shift that suggests increased precipita-
tion (Fig. 11(A)). This transition is more pronounced in the
Petén Itzá density record than in Cariaco Ti owing to the sud-
den cessation of gypsum deposition in the lake. During the
early-to-middle Holocene, the two records track one another
closely, and display two synchronous returns to inferred drier
conditions (Fig 11(A)). The first, centred at 9400 cal. yr BP
was small and brief (
100 yr). The second, centred at
8000 cal. yr BP, was greater in magnitude and longer in dura-
tion (
300 yr). The latter event is also recorded in a speleothem
from lowland Costa Rica, that suggests this was a time of pro-
tracted dry conditions associated with a weakening of the Cen-
tral American monsoon (Lachniet et al., 2004).
The Gulf of Mexico may have played an important role in
regional climate during the last deglaciation because of melt-
water input derived from the retreat of the Laurentide Ice Sheet.
Many studies have focused on the early part of the last degla-
ciation (i.e. Bølling–Allerød/Younger Dryas transition) when
meltwater was redirected from the Mississippi River to the St.
Figure 10 Oxygen isotopic (
18O) records of ostracods from three cir- Lawrence River into the North Atlantic (Clark et al., 2001). A
cum-Caribbean lakes. (A) Lake Miragoane, Haiti,
18O (%, 5-pt run- recent study, however, suggests that episodic meltwater floods
ning average) of Candona sp. (Curtis and Hodell, 1993), (B), Lake to the Gulf of Mexico continued after the Younger Dryas per-
Petén Itzá,
18O (%, 3-pt running average) of Limnocythere sp., and iod. Aharon (2003) identified four events during the Preboreal,
(C) Lake Valencia, Venezuela,
18O (%, 5-pt running average) of Het- when meltwater was rerouted south. These are centred at 9900,
erocypris sp. (Curtis et al., 1999). Radiocarbon dates (14C yr BP) from
9700, 9400 and 9100 14C yr BP. When converted to calendar
Lakes Miragoane are based on carbonate microfossils (ostracods) that
years, these events are dated at 11 000, 10 700, 10 400 and
are susceptible to hard-water-lake error (Deevey and Stuiver, 1964),
whereas the Lakes Valencia and Petén Itzá chronologies are based on 10 000 cal. yr BP (Table 4; Aharon, pers. comm.). Three of these
terrestrial organic matter. Dates from the Lakes Miragoane and Valen- discharge events correspond with the Preboreal events (PBE1-
cia cores were calibrated to calendar years by the authors, in order to 4) observed in the Petén Itzá density and
18O records (Fig. 8),
compare them with the Petén Itzá record. Each proxy is plotted on its as well as variations in Cariaco Ti (Fig. 11(B)). This may suggest
own respective timescale that episodic meltwater discharge into the Gulf of Mexico

Copyright ß 2005 John Wiley & Sons, Ltd. J. Quaternary Sci., Vol. 20(4) 363–376 (2005)
 CLIMATE CHANGE IN CENTRAL AMERICA 373

Figure 11 (A) Comparison of density in Petén Itzá Core 11A (black line; g cc1; inverted scale) and the 3-point running average of percentage tita-
nium (grey line; Ti, %) from the Cariaco Basin (Haug et al., 2001), with August minus February insolation (dashed line; W m2) at 15  N latitude
(Berger, 1978). Major cooling events observed in the North Atlantic such as the 8.2-kyr Event (Alley et al., 1997) and Preboreal Oscillation (PBO)
at
11 300 cal. yr BP (Björck et al., 1997) are highlighted. Major climate intervals are denoted as the Holocene (H), Preboreal (PB), Younger Dryas
(Y-D), and Bølling–Allerød (B–A). Each record is plotted on its own respective timescale. Comparisons were not made younger than 3000 cal. yr BP
because human disturbance of the Petén Itzá watershed began at this time. (B) Petén Itzá Core 11A density smoothed with a 5-pt running average and
Cariaco Ti smoothed with a 15-pt running average to mute short-term variability. The Cariaco Ti record was shifted 100 yr younger by correlating the
PB/H transition. This was done to better illustrate changes observed in the two records

during the Preboreal was sufficient to cause regional changes in responses to a common forcing, such as deglaciation in high
E/P. Alternatively, the correlation between Caribbean drying northern latitudes.
and meltwater pulses to the Gulf of Mexico may represent Finally, we compared the Lake Petén Itzá
18O and density
records with palaeoclimate changes in the high-latitude North
Atlantic. Bond et al. (2001) used fluctuations in haematite-
Table 4 Chronology of Preboreal events (PBE1–4) recorded in Lake stained grains (HSG) to infer increases in drift ice associated
Petén Itzá Core 11A compared to periods of increased meltwater deliv- with cool SST in the North Atlantic for the last glaciation and
ery (MWF-5 a,c,e,g) to the Gulf of Mexico (Aharon, 2003) and the Pre- Holocene. From 11 500 to 8000 cal. yr BP, they recognised four
boreal Oscillation (PBO; Björck et al., 1997) distinct peaks in HSG abundance centred at 11 300, 10 300,
9400 and 8400 cal. yr BP (Fig. 12). The three older events cor-
Preboreal Intervala GOM and Intervalb relate with inferred dry climate conditions observed in the
events (PBE) (cal. kyr BP) NA events (cal. kyr BP) Petén Itzá
18O record (Fig. 12). Furthermore, three dry events
that are also observed as density increases in our record coin-
MWF-5 g 9.75–10.07 (10.0) cide with abrupt decreases in
18O values measured in the
4 10.31–10.46 (10.4) MWF-5 e 10.23–10.43 (10.4)
GISP2 ice core (Fig. 13; Stuiver et al., 1995). The two most
3 10.66–10.80 (10.7) MWF-5 c 10.69–10.79 (10.7)
2 10.88–11.01 (10.9) MWF-5 a 10.97–11.07 (11.0)
pronounced events, the Preboreal Oscillation (PBO) at
1 11.15– > 11.25 (11.2) PBO 11.15–11.30
11 300 cal. yr BP (Björck et al., 1997), which corresponds
to PBE1 (Table 4), and the 8.2-kyr Event (Alley et al., 1997)
a
Numbers in parentheses indicate times of peak aridity in the Petén, have been interpreted as periods of significant cooling in the
inferred from
18O of ostracods. high-latitude North Atlantic. We, therefore, suggest that cool
b
Numbers in parentheses indicate maximum meltwater fluxes to the events in the high-latitude North Atlantic were associated with
Gulf of Mexico. dry climate conditions in the northern lowland Neotropics.

Copyright ß 2005 John Wiley & Sons, Ltd. J. Quaternary Sci., Vol. 20(4) 363–376 (2005)
374 JOURNAL OF QUATERNARY SCIENCE

Chiang et al., 2003; Chiang, 2004). In turn, the mean latitudinal

position of the ITCZ and Azores–Bermuda subtropical high
may be influenced by numerous factors including: tropical
Atlantic SST, seasonal insolation, glacial boundary conditions,
thermohaline circulation (via cross-equatorial heat transport
and sea-ice extent), solar irradiance and ocean–atmosphere
interactions such as NAO and ENSO (Hodell et al., 1991;
Chang et al., 1997; Peterson et al., 2000; Haug et al., 2001;
Poore et al., 2003; Chiang et al., 2003; Chiang, 2004).
Long-term (glacial-to-interglacial) variations in Neotropical
precipitation are affected by the seasonal distribution
(August–February) of insolation related to variations in Earth’s
orbit, especially precession (Hodell et al., 1991; Haug et al.,
2001). Precipitation in the northern Neotropics is enhanced
when the seasonal insolation difference is high, and reduced
when seasonality is low. Long-term changes in palaeoprecipi-
tation proxies (density and
18O) in Lake Petén Itzá cores gen-
erally follow the signal of seasonal insolation difference at
15  N (Fig. 11(A)), but orbital forcing cannot explain the abrupt
changes that occurred during the latter part of the last deglacia-
tion and early Holocene.
During the Preboreal, the general trend toward more
Figure 12 (A) 10Be Flux (105 atoms cm2 yr1, smoothed and humid conditions in the Petén was interrupted by four dry
detrended) from the GISP2 and GRIP ice core records, after Bond events (PBE1-4) centred at 11 200, 10 900, 10 700 and
et al. (2001), (B) stacked plot of percentage haematite stained grains 10 400 cal. yr BP (Fig. 8). The oldest event (PBE-1) coincides
from North Atlantic cores (HSG, %; Bond et al., 2001), and (C) oxygen with the Preboreal Oscillation (Björck et al., 1997; van der
isotopic composition (
18O, %) of Petén Itzá ostracods, which was Plicht et al., 2004). The three younger events appear to corre-
smoothed with a 3-pt running average to better illustrate long-term
spond with episodic diversions of glacial meltwater to the Gulf
variability. Each record is plotted on its own respective timescale
of Mexico (Table 4; Aharon, 2003) that were related to changes
in glacial boundary conditions (Clark et al., 2001). A possible
explanation for this phenomenon comes from modelling stu-
dies that impose a 6  C cooling of the Gulf of Mexico related
to the delivery of cold, fresh glacial waters to the basin (Over-
peck et al., 1989; Oglesby et al., 1989; Maasch and Oglesby,
1990). These model sensitivity tests imply that cooling Gulf of
Mexico SST results in a westward shift of the Azores–Bermuda
high to a position over the Intra-Americas Sea. This would, in
turn, suppress convection and reduce precipitation in the
region (Hastenrath, 1984). Meltwater discharge during the Pre-
boreal, however, was significantly less than that used to infer a
6  C cooling of the entire Gulf of Mexico. Instead, meltwater
fluxes were more comparable in magnitude to the largest of
modern Mississippi River floods (Aharon, 2003), which prob-
ably only influenced part of the Gulf of Mexico. Alternatively,
early Holocene deglacial events may have provided meltwater
input to both the Gulf of Mexico and high-latitude North Atlan-
Figure 13 Comparison of (A) oxygen isotopic composition (
18O, %)
tic, thereby causing cooling, an expansion of sea ice, and per-
of ice from the GISP2 ice core (Stuiver et al., 1995) with (B) density
(g cc1) from Lake Petén Itzá Core 11A. Major cooling events observed
haps a reduction in the Atlantic meridional overturning
in the North Atlantic such as the 8.2-kyr Event (Alley et al., 1997) and circulation (MOC).
Preboreal Oscillation (PBO) at
11 300 cal. yr BP (Björck et al., 1997) Two of the Preboreal events (PBE1 and PBE4) at 11 200 and
are highlighted. Each proxy is plotted on its own respective timescale 10 400 cal. yr BP, respectively, and an early Holocene event at
9400 cal. yr BP may correlate to increases in the
18O records
of Lake Miragoane and Lake Valencia (Fig. 10; Curtis and
Proposed mechanisms for climate change Hodell, 1993; Curtis et al., 1999). These dry events together
in the Preboreal and early Holocene with another early Holocene event centred at 8000 cal. yr BP,
correspond within chronological uncertainty, to cooler tem-
Precipitation proxies in the circum-Caribbean display a coher- peratures inferred from low
18O values in the GISP2 ice core
ent response during the latter part of the last deglaciation and (Fig. 13; Stuiver et al., 1995) and fluctuations in the abundance
early Holocene. This is expected because the northern Neotro- of haematite-stained grains (HSG) found in high-latitude North
pics are influenced by the same climate regime, where rainfall Atlantic cores (Fig. 12; Bond et al., 2001). Bond et al. (2001)
is related to the intensity of the annual cycle and associated dis- related fluctuations in HSG abundance with changes in the
placement of the Atlantic ITCZ and Azores–Bermuda high- production rates of 10Be and 14C, such that increases in HSG
pressure system (Hodell et al., 1991; Peterson et al., 2000; abundance occurred during times of decreased solar irradiance
Haug et al., 2001). This mechanism has been shown to and cold temperatures in the high-latitude North Atlantic.
operate on short (annual) to long (millennial) timescales Increased aridity in the circum-Caribbean may have accompa-
(Hodell et al., 1991; Chang et al., 1997; Peterson et al., nied these cooling events, and may have been caused by a
2000; Haug et al., 2001; Lea et al., 2003; Black et al., 2004; southward displacement of the ITCZ related to decreased solar

Copyright ß 2005 John Wiley & Sons, Ltd. J. Quaternary Sci., Vol. 20(4) 363–376 (2005)
 CLIMATE CHANGE IN CENTRAL AMERICA 375

luminosity, increased sea ice or decreased thermohaline circu- subtropical high. This mechanism operated on a variety
lation. For example, the 8.2-kyr Event, which was the most pro- of timescales and responded sensitively to changes in solar
longed and distinct cooling in the northern hemisphere since radiation, glacial meltwater, North Atlantic sea ice extent and
the Younger Dryas, may have been related to both reduced Atlantic MOC.
solar luminosity (Muscheler et al., 2004) and a reduction in
MOC from glacial meltwater discharge into the North
Atlantic via Hudson Strait (Clark et al., 2001). A similar Acknowledgements We thank Liseth Perez, Laura Rodriguez and
mechanism has been proposed for the Preboreal Oscillation Jacobo Blijdenstein for field assistance, Gerard Bond for HSG and
10
(PBO; Björck et al., 1997; van der Plicht et al., 2004). Be data, Paul Aharon for calibrating the LOUIS stacked-core age
model, and Mike Binford for assisting with limnological data collection
in 1980. We also thank Arturo Godoy and Roan McNab from the Wild-
life Conservation Society (WCS), Margaret and Michael Dix from the
Universidad del Valle, and the Consejo Nacional de Areas Protegidas
Summary and conclusions (CONAP) for providing logistical assistance in the field. The Limnolo-
gical Research Center (LRC) at the University of Minnesota supplied the
coring system. Radiocarbon analyses were performed under the aus-
Seismic data indicate that prior to the last deglaciation, the pices of the U.S. Department of Energy by the University of California
level of Lake Petén Itzá was about
58 m below modern stage. Lawrence Livermore National Laboratory (contract W-7405-ENG-48).
This observation is consistent with previous studies of Petén Paul Baker and Larry Peterson provided thoughtful reviews that
lakes that indicated arid climate conditions. The last deglacia- improved the manuscript. This work was supported by National
tion saw an abrupt shift from arid to humid conditions in the Science Foundation (NSF) grant ATM-0117148 and is a publication
region. The age of our oldest deep-water core from Petén Itzá of the University of Florida, Land Use and Environmental Change Insti-
tute (LUECI).
is 11 250 cal. yr BP, which captures only the terminal part of
this arid-to-humid transition. Pollen assemblages from the base
of the core demonstrate that a mesic forest had been largely
established by
11 250 cal. yr BP, but Petén Itzá’s level was
still > 35 m below modern. From 11 250 to 10 350 cal. yr BP References
during the Preboreal Period, cores from < 35 m below present
water depth contain soil horizons, indicating subaerial expo-
sure. During this period, deep-water cores from > 50 m below Aharon P. 2003. Meltwater flooding events in the Gulf of Mexico revis-
present water depth contain gypsum sands interbedded ited: implications for rapid climate changes during the last deglacia-
tion. Paleoceanography 18: DOI:10.1029/2002PA000840.
with carbonate-rich clays, indicating several dry–wet cycles.
Alley RB, Mayewski PA, Sowers T, Stuiver M, Taylor KC, Clark PU.
Oxygen isotopic ratios generally decrease during the Preboreal 1997. Holocene climate instability: a prominent, widespread event
Period indicating a trend toward more mesic conditions. 8,200 yr ago. Geology 25: 483–486.
This trend was interrupted, however, by four Preboreal dry Anselmetti FS, Ariztegui D, Hodell DA, Hillesheim MB, Brenner M,
events (PBE1-4) centred at 11 200, 10 900, 10 700 and Gilli A, McKenzie JA. In review. Late Quaternary climate-induced
10 400 cal. yr BP. The four events correlate to decreases in the lake level variations in Lake Petén Itzá, Guatemala, inferred from
Cariaco Ti record, suggesting they may have been widespread seismic stratigraphic analysis. Palaeogeography, Palaeoclimatology,
throughout the circum-Caribbean region. Three of the events Palaeoecology.
coincided with episodic diversions of glacial meltwater to the Bard E, Arnold M, Hamelin B, Tisnerat-Laborde N, Cabioch G. 1998.
Gulf of Mexico. Radiocarbon calibration by means of mass spectrometric 230Th/234U
and 14C ages of corals; an updated database including samples from
At 10 350 cal. yr BP, increased precipitation and lake-level
Barbados, Mururoa, and Tahiti. Radiocarbon 40: 1085–1092.
rise is documented in deep-water cores by cessation of gypsum Berger A. 1978. Long-term variations of daily insolation and Quaternary
deposition and the onset of lacustrine sedimentation at shal- climate change. Journal of the Atmospheric Sciences 35: 2362–2367.
low-water sites. Increased abundance of tropical mesic pollen Björck S, Rundgren M, Ingolfsson O, Funder S. 1997. The Preboreal
taxa lagged this transition by
100 yr. This event occurred at oscillation around the Nordic Seas: terrestrial and lacustrine
the Preboreal-to-Boreal transition and coincided with an responses. Journal of Quaternary Science 12: 455–465.
inferred increase in precipitation recorded throughout the cir- Black DE, Thunell RC, Kaplan A, Peterson LC, Tappa EJ. 2004. A 2000-
cum-Caribbean. The early Holocene was a time of generally year record of Caribbean and tropical North Atlantic hydrographic
reduced climate variability and steadily increasing moisture variability. Paleoceanography 19: DOI:10.1029/2003PA000982.
availability except for two returns to slightly drier conditions Bond G, Kromer B, Beer J, Muscheler R, Evans MN, Showers W, Hoff-
man S, Lotti-Bond R, Hajdas I, Bonani G. 2001. Persistent solar influ-
centred at 9400 and 8000 cal. yr BP. The prolonged dry event
ence on North Atlantic climate during the Holocene. Science 294:
at 8000 cal. yr BP correlates with inferred reductions in cir- 2130–2136.
cum-Caribbean precipitation as well as cooler temperatures Bradbury JP, Leyden BW, Salgado-Labouriau ML, Lewis WM, Schubert
in the high-latitude North Atlantic. C, Binford MW, Frey DG, Whitehead DR, Weibezahn FH. 1981.
Comparison of the Lake Petén Itzá record with other lacus- Late Quaternary environmental history of Lake Valencia, Venezuela.
trine and marine records from the circum-Caribbean and Gulf Science 214: 1299–1305.
of Mexico suggests that climate in this region responded coher- Brenner M. 1994. Lakes Salpetén and Quexil, Petén, Guatemala, Cen-
ently during the Preboreal and early Holocene periods. In addi- tral America. In Global Geological Record of Lake Basins, Vol. 1,
tion, lowland Neotropical climate during this time was tightly Kelts K, Gierlowski-Kordesch E (eds). Cambridge University Press:
linked to high-latitude climate change in the North Atlantic. Cambridge, UK; 377–380.
Brenner M, Curtis JH, Higuera-Gundy A, Hodell DA, Jones GA, Binford
Although the general trend in the northern lowland Neotropics
MW, Dorsey KT. 1994. Lake Miragoane, Haiti (Caribbean). In Global
was toward increasingly mesic conditions from
11 250 to Geological Record of Lake Basins, Vol. 1, Kelts K, Gierlowski-
7500 cal. yr BP, climate became drier at times of cooling in the Kordesch E (eds). Cambridge University Press: Cambridge, UK;
high-latitude North Atlantic. We speculate that the observed 403–405.
changes in lowland Neotropical precipitation were related to Chang P, Ji L, Li H. 1997. A decadal climate variation in the tropical
the intensity of the annual cycle and associated displacements Atlantic Ocean from thermodynamic air–sea interactions. Nature
in the mean position of the ITCZ and Azores–Bermuda 385: 518–518.

Copyright ß 2005 John Wiley & Sons, Ltd. J. Quaternary Sci., Vol. 20(4) 363–376 (2005)
376 JOURNAL OF QUATERNARY SCIENCE

Chiang JCH. 2004. Present-day climate variability in the tropical Atlan- IAEA. 2001. IAEA/WMO global network of isotopes in precipitation:
tic: a model for paleoclimate change? In The Hadley Circulation: the GNIP Database. http://isohis.iaea.org/ [1 July 2004].
Past, Present, and Future, Diaz HF, Bradley RS (eds). Cambridge Lachniet MS, Asmerom Y, Burns SJ, Patterson WP, Polyak VJ, Seltzer
University Press: Cambridge, UK; 465–488. GO. 2004. Tropical response to the 8200 yr B.P. cold event? Spe-
Chiang JCH, Biasutti M, Battisti DS. 2003. Sensitivity of the Atlantic leothem isotopes indicate a weakened early Holocene monsoon in
Intertropical Convergence Zone to Last Glacial Maximum boundary Costa Rica. Geology 32: 957–960.
conditions. Paleoceanography 18: DOI:10.1029/2003PA000916. Lea DW, Pak DK, Peterson LC, Hughen KA. 2003. Synchroneity of
Clark PU, Marshall SJ, Clarke GKC, Hostetler SW, Liccardi JM, Teller tropical and high-latitude Atlantic temperatures over the last glacial
JT. 2001. Freshwater forcing of abrupt climate change during the last termination. Science 302: 1361–1364.
glaciation. Science 193: 283–287. Leyden BW. 1985. Late Quaternary aridity and Holocene moisture
Covich AP, Stuiver M. 1974. Changes in oxygen 18 as a measure of fluctuations in Lake Valencia Basin, Venezuela. Ecology 66: 1279–
long-term fluctuations in tropical lake levels and molluscan popula- 1295.
tions. Limnology and Oceanography 19: 682–691. Leyden BW, Brenner M, Hodell DA, Curtis JH. 1993. Late Pleistocene
Covich AP. 1976. Recent changes in molluscan diversity of a large tro- climate in the Central American lowlands. In Climate Change in
pical lake (Lago de Petén, Guatemala). Limnology and Oceanogra- Continental Isotopic Records, Geophysical Monograph 78, Swart
phy 21: 51–59. PK, Lohmann KC, McKenzie J, Savin S (eds). American Geophysical
Curtis JH, Hodell DA. 1993. An isotopic and trace element study of Union: Washington, DC; 165–178.
ostracods from Lake Miragoane, Haiti: a 10.5 kyr record of paleosa- Leyden BW, Brenner M, Hodell DA, Curtis JH. 1994. Orbital and inter-
linity and paleotemperature changes in the Caribbean. In Climate nal forcing of climate on the Yucatan Peninsula for the past ca. 36 ka.
Change in Continental Isotopic Records, Geophysical Monograph Palaeogeography, Palaeoclimatology, Palaeoecology 109: 193–210.
78, Swart PK, Lohmann KC, McKenzie J, Savin S (eds). American Maasch KA, Oglesby RJ. 1990. Meltwater cooling of the Gulf of
Geophysical Union: Washington, DC; 135–152. Mexico: a GCM simulation of climatic conditions at 12 ka. Paleocea-
Curtis JH, Brenner M, Hodell DA, Balser RA, Islebe GA, Hoogheimstra nography 5: 977–996.
H. 1998. A multi-proxy study of Holocene environmental change in Magaña V, Amador JA, Medina S. 1999. The midsummer drought over
the Maya Lowlands of Petén, Guatemala. Journal of Paleolimnology Mexico and Central America. Journal of Climate 12: 1577–1588.
19: 139–159. Muscheler R, Beer J, Vonmoss M. 2004. Causes and timing of the 8200
Curtis JH, Brenner M, Hodell DA. 1999. Climate change in Lake Valen- yr BP event inferred from comparison of the GRIP 10Be and the tree
cia Basin, Venezuela,
12 600 yr BP to present. The Holocene 9: ring
14C record. Quaternary Science Reviews 23: 2101–2111.
609–619. Oglesby RJ, Maasch KA, Saltzman B. 1989. Glacial meltwater cooling
Curtis JH, Brenner M, Hodell DA. 2001. Climate change in the circum- of the Gulf of Mexico: GCM implications for Holocene and present-
Caribbean (late Pleistocene-to-Present) and implications for regional day climates. Climate Dynamics 3: 115–133.
biogeography. In Biogeography of the West Indies, Woods CA (ed.). Overpeck JT, Peterson LC, Kipp N, Imbrie J, Rind D. 1989. Climate
CRC Press: Boca Raton, FL; 35–54. change in the circum-North Atlantic region during the last deglacia-
Deevey ES, Stuiver M. 1964. Distribution of natural isotopes of carbon tion. Nature 338: 553–557.
in Linsley Pond and other New England lakes. Limnology and Ocea- Paillard D, Labeyrie L, Yiou P. 1996. Macintosh program performs
nography 9: 1–11. time-series analysis. AGU Eos Transactions 77: 379.
Deevey ES, Brenner M, Flannery MS, Yezdani GH. 1980. Lakes Yaxha Peterson LC, Haug GH, Hughen KA, Röhl U. 2000. Rapid changes in
and Sacnab, Petén, Guatemala: limnology and hydrology. Archiv für the hydrologic cycle of the tropical Atlantic during the Last Glacial.
Hydrobiologie 57: 419–460. Science 290: 1947–1951.
Deevey ES, Brenner M, Binford MW. 1983. Paleolimnology of the Poore RZ, Dowsett HJ, Verardo S, Quinn TM. 2003. Millennial- to
Petén Lake District, Guatemala III: Late Pleistocene and Gamblian century-scale variability in the Gulf of Mexico Holocene climate
environments of the Maya area. Hydrobiologia 103: 211–216. records. Paleoceanography 18: DOI:10.1029/2002PA000868.
Fontes JC, Gonfiantini R. 1967. Component isotopique au cours de Rozanski K, Araguá-Araguá L, Gonfiantini R. 1993. Isotopic patterns of
l’evaporation de deux basins sahariens. Earth and Planetary Science modern global precipitation. In Climate Change in Continental Iso-
Letters 3: 258–266. topic Records, Geophysical Monograph 78, Swart PK, Lohmann KC,
Gasse F, Techet R, Durand A, Gilbert E, Fontes JC. 1990. The arid– McKenzie J, Savin S (eds). American Geophysical Union: Washing-
humid transition in the Sahara and Sahel during the last deglaciation. ton, DC; 1–36.
Nature 346: 141–146. Schmidt MW, Spero HJ, Lea DW. 2004. Links between salinity varia-
Hastenrath S. 1984. Interannual variability and the annual cycle: tion in the Caribbean and North Atlantic thermohaline circulation.
mechanisms of circulation and climate in the tropical Atlantic sector. Nature 428: 160–163.
Monthly Weather Review 112: 1097–1107. Socki RA, Karlsson HR, Gibson EK. 1992. Extraction technique for
Haug GH, Hughen KA, Sigman DM, Peterson LC, Röhl U. 2001. South- determination of oxygen-18 in water using preevacuated glass vials.
ward migration of the Intertropical Convergence Zone through the Analytical Chemistry 64: 829–831.
Holocene. Science 293: 1304–1308. Stuiver M, Reimer PJ. 1993. Extended 14C data base and revised Calib
Higuera-Gundy A, Brenner M, Hodell DA, Curtis JH, Leyden BW, 3.0 14C age calibration program. Radiocarbon 35: 35–65.
Binford MW. 1999. A 10 300 14C-year record of climate and vegeta- Stuiver M, Grootes PM, Braziunas TF. 1995. The GISP2
18O climate
tion change from Haiti. Quaternary Research 52: 159–170. record of the past 16 500 years and the role of the sun, ocean, and
Hodell DA, Curtis JH, Jones GA, Higuera-Gundy A, Brenner M, Binford volcanoes. Quaternary Research 44: 341–354.
MW, Dorsey KT. 1991. Reconstruction of Caribbean climate change Stuiver M, Reimer PJ, Bard E, Beck JW, Burr GS, Hughen KA, Kromer B,
over the past 10 500 years. Nature 352: 790–793. McCormac G, van der Plicht J, Spurk M. 1998. INTCAL98 radiocar-
Hodell DA, Curtis JH, Brenner M. 1995. Possible role of climate change bon age calibration, 24 000–0 cal BP. Radiocarbon 40: 1041–1083.
in the collapse of the Maya civilization. Nature 375: 391–394. van der Plicht J, van Geel B, Bohncke SJP, Bos JAA, Blaauw M,
Hodell DA, Brenner M, Curtis JH, Guilderson T. 2001. Solar forcing of Speranza AOM, Muscheler R, Björck S. 2004. The Preboreal climate
the drought frequency in the Maya lowlands. Science 292: 1367– reversal and a subsequent solar-forced climate shift. Journal of Qua-
1369. ternary Science 19: 263–269.
Hodell DA, Brenner M, Curtis JH. 2005. Terminal Classic drought in Vinson GL. 1962. Upper Cretaceous and Tertiary stratigraphy of Gua-
the Maya lowlands inferred from multiple sediment cores in Lake temala. Bulletin of the American Association of Petroleum Geolo-
Chichancanab (Mexico). Quaternary Science Reviews. in press. gists 46: 425–456.
Hughen KA, Overpeck JT, Peterson LC, Trumbore S. 1996. Rapid cli- Waliser DE, Shi Z, Lanzante JR, Oort AH. 1999. The Hadley Cell circu-
mate changes in the tropical Atlantic region during the last deglacia- lation: assessing NCEP/NCAR reanalysis and sparse in-situ estimates.
tion. Nature 380: 51–54. Climate Dynamics 15: 719–735.

Copyright ß 2005 John Wiley & Sons, Ltd. J. Quaternary Sci., Vol. 20(4) 363–376 (2005)